Rug01-000813602 2010 0001 Ac

Mind-reading in adolescents and adults
with a pervasive developmental disorder
Koen Ponnet
Promotor: Prof. Dr. Herbert Roeyers

Copromotor: Prof. Dr. Ann Buysse
Proefschrift ingediend tot het behalen van de academische graad

van Doctor in de Psychologische Wetenschappen
2004
Gutta cavat lapidem non vi, sed saepe cadendo.
Contents
Chapter 1
Introduction
Chapter 2
Advanced mind-reading in adults with Asperger
1
17
syndrome
Chapter 3
Empathic accuracy in adults with a pervasive
43
developmental disorder during an unstructured

conversation with a typically developing stranger
Chapter 4
Empathic accuracy in young adults with PDD:
81
Does structure matter ?

Chapter 5
Perceived and performed mind-reading in adults
113
with a pervasive developmental disorder

Chapter 6
Schematic knowledge and levels of processing
137
in young adults with PDD

Chapter 7
Alternative explanations to the empathic accuracy
163
findings
Chapter 8
General discussion
191
Acknowledgements
It was a fascinating and absorbing experience to realize this thesis. Obviously,
the process of creating this work was greatly facilitated by the help and support of
many people. It is my pleasure to acknowledge each of them.
First of all, I would like to thank all adolescents and adults with PDD, and their
parents, who were willing to participate in the studies and have taught me so
much.
My special gratitude goes to my supervisors Prof. Dr. Herbert Roeyers and
Prof. Dr. Ann Buysse, who accompanied me from the very beginning of this
journey, and all the way through to its completion. They were always willing to
give me suggestions, comments, opinions and recommendations, and -as suchhave helped steer the course of my research into exciting directions. In addition,
their perspective was invaluable in helping me figure out how to separate the vital
wheat from the expendable chaff. Thank you Herbert and Ann.
I am very indebted to Prof. Dr. Armand De Clercq, thanks to whom the
empathic accuracy design was digitised. Of course, I also appreciate the various
distracting chats in his office. Furthermore, I would like to express my true
appreciation to Prof. Dr. Ivan Mervielde and Prof. Dr. Geert Van Hove, for their
advice and comments on my research work, and to Prof. Dr. Paulette Van Oost,
for being -in a way- the shepherd of our department.
I also want to express my appreciation to all colleagues, especially the members
of Herberts team, for their interest, enthusiasm and collegiality. Special thanks go
out to Yolande D., Annick V. and Wouter B., for their sense of humour. In
addition, I want to thank Ruth R., for her most appreciated skills as graphic
designer; Kim D. and Olivier P., for the several discussions we had; and Stefaan
V., for the proofreading and his down-to-earthness that really made me more
relaxed. As a coffee-addict, I want to apologize to my Einde Were colleagues for
the distress they suffered from the good (but sometimes too strong) coffee that I
made.
Over the years, many authors and clinicians have enriched my own thinking
and learned me more about autism, empathy, male/female relationships, normal
and so-called abnormal thinking, and so on. I would like to thank them all. I am
also indebted to several undergraduate students for their help in data collection.
I have a few words to say about some people who helped keep me reasonably
sane during the writing of this thesis. First and foremost, I want to thank my
mother and father, brother and sisters, and -of course- their partners, for their
continuous support and encouragement. Second, I want to thank my
experimental friends for the glorious moments on the badminton court and the
several amusing incidents out of it. Third, I want to thank all my friends,
especially Frederik D. and Katelijne V., who never fail to give support and offer
full scope to me to broaden my horizon.
Finally, I want to thank Vronique. She endured my hours at the office and my
-from time to time dreadful- restlessness. She helped me with the final editing, and
saw me through an exhilarating but demanding period of my life.
Koen Ponnet
January 2004
CHAPTER
Introduction
Chapter 1
PERVASIVE DEVELOPMENTAL DISORDERS

People with a pervasive developmental disorder (PDD) experience profound
impairments in reciprocal social interaction, communication and imagination, and
show restricted, repetitive and stereotyped patterns of behaviour, interests and
activities (APA, 1994, 2000; Wing, 1988). According to the DSM-IV-TR (APA,
2000), Pervasive Developmental Disorders encompasses Autistic Disorder, Retts
Disorder, Childhood Disintegrative Disorder, Aspergers Disorder, and Pervasive
Developmental Disorder Not Otherwise Specified (PDD-NOS), including atypical
autism.
Besides this categorical division, there is also an assumption that autism is just
one part of a spectrum of disorders that encompasses other variants such as atypical
autism, PDD-NOS and Asperger syndrome (Baron-Cohen, Wheelwright, Skinner,
Martin, & Clubley, 2001; Bolton et al., 1994; Szatmari, 2003). According to this
view, the spectrum ranges from classic Kanner-type autism at one end of the
spectrum to high-functioning autism and Asperger syndrome at the other (Parsons
& Mitchell, 2002). Although the concept of a spectrum captures well the
behavioural variability that is a central feature of autism and its related
developmental disorders (Lawson, 2003), a debate remains about the validity and
usefulness of a spectrum view (Ozonoff & McMahon-Griffith, 2000; Volkmar &
Klin, 2000). In the present thesis, we will therefore use the general term pervasive
developmental disorder or PDD.
THEORY OF MIND
The past two decades, several psychological accounts have emerged as potential
explanations of the various aspects of PDD, such as the theory of mind hypothesis,
executive dysfunction, weak central coherence and others (Rapin, 1997). However,
it is a characteristic of each existing account that it only explains a specific subset
of the overall features of PDD (Booth, Charlton, Hughes, & Happ, 2003; Jarrold,
Butler, Cottington, & Jimenez, 2000; Lawson, 2003; Rapin, 1997). While some
Introduction
theories mainly focus on the social features, the focus of others lies more on the
non-social features of PDD.
One of the most influential of all accounts is the theory of mind hypothesis
(Baron-Cohen, Tager-Flusberg, & Cohen, 2000; Klin, 2000; Rapin, 1997). This
hypothesis suggests that the impairments of subjects with PDD can be explained in
terms of a failure to attribute mental states (such as beliefs, desires, or intentions)
and feelings to oneself and others (Baron-Cohen et al., 2000). Without a theory of
mind, people are unable to develop a normal understanding that others have mental
states and feelings, which results in socially and communicatively inadequate
behaviour. Theory of mind is also referred to as mind-reading (Whiten, 1991).
Studies with children with PDD largely support this hypothesis. Researchers
usually assess the theory of mind of children with PDD with first-order or secondorder false belief tasks (Baron-Cohen, ORiordan, Stone, Jones, & Plaisted, 1999).
First-order false belief tasks present participants with a situation in which they
know the true identity of a hidden object but must deduce that someone else,
without that knowledge, will misidentify the object. The tasks are called first-order
tasks because they only involve inferring one persons mental state (Baron-Cohen
et al., 2000). Second-order tasks demand that participants reason about what one
person thinks about another persons thought (Perner & Wimmer, 1985). Typically
developing children pass first-order tasks at about 4 years old, whereas secondorder theory of mind tasks are passed by 6 or 7 year olds (Perner, 1991; Perner &
Wimmer, 1985; Wellman, 1990; Wimmer & Perner, 1983).
In one of the earliest studies, Baron-Cohen, Leslie & Frith (1985) found that a
large number of children with PDD was unable to attribute a first-order false belief
to another individual. This finding has been replicated several times using slightly
different paradigms (e.g. Baron-Cohen, 1989; Baron-Cohen, Leslie, & Frith, 1986;
Leekam & Perner, 1991; Leslie & Frith, 1987; Perner, Frith, Leslie, & Leekam,
1989). Moreover, several studies have shown that, with the exception of somewhat
older, high-functioning children (e.g. Bauminger & Kasari, 1999; Dahlgren &
Trillingsgaard, 1996; Prior, Dahlstrom, & Squires, 1990), they all fail second-order
theory of mind tasks (for an overview, see Happ, 1995).
However, a substantial number of adolescents and adults with PDD were found
to pass the traditional second-order tasks (e.g. Bowler, 1992; Happ, 1994; Tager-
Chapter 1
Flusberg & Sullivan, 1995). These studies cannot be taken as a conclusive evidence
of an intact theory of mind in individuals with PDD, because second-order tasks
can easily produce ceiling effects if used with individuals with a mental age above
6 or 7 years (Wimmer & Perner, 1983). They might only indicate that the
traditional mind-reading tasks, which are designed for children with PDD, are not
subtle enough to detect impairments in adults with PDD (Baron-Cohen, Jolliffe,
Mortimore, & Robertson, 1997; Rutterford, Baron-Cohen, & Wheelwright, 2002).
Happs (1994) study was one of the first to take this issue seriously. She used a
series of complex, naturalistic stories, each followed by a comprehension question
and a justification question. The justifications of the participants were scored as
either involving mental states or physical states. The task of Happ was pitched at
the level of 8-9 years old, and therefore can be considered as more advanced than
the previous mind-reading tasks. She found that adults with PDD had more
difficulties with the mental state stories than controls, and that even the most able
participants with PDD tended to give context-inappropriate mental state
explanations.
Baron-Cohen and colleagues (1997) extended Happs line of research and
developed The Eyes Task, which consisted of a number of photographs of the
region around the eyes, together with two basic or more complex mental state
terms. Participants were asked to make a forced choice between which of the two
words best describes what the person in the photograph might be thinking or
feeling. The task revealed subtle mind-reading impairments in subjects with PDD,
relative to IQ-matched controls.
The last few years, an increasing amount of research is devoted to the
development of more advanced adult tests of theory of mind (Baron-Cohen,
Wheelwright, Hill, Raste, & Plumb, 2001; Kaland, Moller-Nielsen, Callesen,
Mortensen, Gottlieb, & Smith, 2002; Kleinman, Marciano, & Ault, 2001;
Rutterford et al., 2002). These advanced tasks have a more mature and challenging
content, which distinguishes them from mind-reading tasks developed for younger
subject populations (Heavey, Phillips, Baron-Cohen, & Rutter, 2000).
However, we have to take into account that the current tasks developed to
measure the mind-reading of adolescents and adults with PDD have little ecological
validity. Indeed, it can hardly be argued that the use of static stimuli (such as tales,
Introduction
drawings, photographs and others) is a naturalistic way to evaluate the mindreading of adolescents or adults with or without autism (Baron-Cohen et al., 1997;
Baron-Cohen, Wheelwright, Hill, et al., 2001; Kleinman et al., 2001). In order to
measure the mind-reading of adults with PDD in contexts that closely mirror reallife social interactions, many authors have called for more dynamic tasks (BaronCohen, Wheelwright, Hill, et al., 2001; Bauminger & Kasari, 1999; Fein, Lucci,
Braverman, & Waterhouse, 1992 ).
EMPATHIC ACCURACY
Empathic accuracy in typically developing people
A naturalistic research paradigm that offers the opportunity for a systematic
investigation of the mind-reading in a laboratory situation is the empathic accuracy
design of Ickes and colleagues (Ickes, 1993, 2003; Ickes, Bissonnette, Garcia, &
Stinson, 1990).
Over the years, there has been a variety of ways to approach empathic accuracy,
and the term empathy has been defined in different ways by different authors
(Eisenberg, Murphy, & Sheppard, 1997). According to Ickes and colleagues,
empathic accuracy can be defined as the extent to which someone is able to
accurately infer other peoples thoughts and feelings (Ickes, Stinson, Bissonnette,
& Garcia, 1990). The term empathy includes both a cognitive and an emotional
component. While the cognitive component refers to the capacity to understand
others internal states, the emotional component refers to the emotional response
resulting from the recognition of other persons emotional state or condition
(Eisenberg, Shea, Carlo, & Knight, 1997).
As described by Ickes (1997), the most unequivocal way to measure empathic
accuracy is by rating the similarity between the content of a target persons actual
thoughts and feelings with the content of the corresponding inferred thoughts and
feelings reported by the perceiver. There are two major research paradigms to
measure empathic accuracy. The first is the standard stimulus paradigm, in which
individual participants each view the same set of videotaped interactions and
attempt to infer the thoughts and feelings of the same set of target persons (Gesn &
Chapter 1
Ickes, 1999; Marangoni, Garcia, Ickes, & Teng, 1995). The second design is the
dyadic interaction paradigm, in which dyad members attempt to infer each others
thoughts and feelings from a videotaped record of their spontaneous interaction
(Ickes, Bissonnette, et al., 1990; Ickes, Gesn, & Graham, 2000; Stinson & Ickes,
1992). Whereas the participants in the standard stimulus design serve only as
perceivers and not as targets; in the dyadic interaction design, each participant
becomes an active and interacting member, instead of being a passive observer.
Therefore, empathic accuracy can be characterised as the emergent product of
social interaction processes that occur at the level of the dyad (Ickes, 1997; Ickes,
Bissonnette, et al., 1990).
Over the years, the standard stimulus design as well as the dyadic interaction
design have consistently proved to be reliable and valid methods to assess the
empathic accuracy of other persons, such as strangers, friends or dating couples
(Buysse & Ickes, 1999; Ickes, Stinson, et al., 1990; Marangoni et al, 1995;
Simpson, Ickes, & Blackstone, 1995).
Empathic accuracy in adults with PDD
In a recent study, we tested high-functioning adults with PDD and control adults
using two static mind-reading tasks and a more naturalistic Empathic Accuracy
Task (Roeyers, Buysse, Ponnet, & Pichal, 2001). The two static mind-reading tests
used were an adapted and shortened version of Happs (1994) Strange Stories, and
an adapted version of Baron-Cohen et al.s (1997) Eyes Task. The procedure used
in the Empathic Accuracy Task was the standard stimulus paradigm, in which
individual participants each viewed the same set of two videotaped interactions
between two strangers (Roeyers et al, 2001).
The results suggest that the adults with PDD were not impaired on the static
tasks compared to control adults. On the other hand, the Empathic Accuracy Task
was able to distinguish clearly between both groups, although a significant
between-group difference was found only for one of the empathic accuracy
videotapes in favour of the control adults. A possible explanation for these mixed
results is the different nature of both stimulus tapes. While both targets in the first
videotape were playing a board game, the targets in the second videotape were
Introduction
having a conversation of the getting acquainted type. Consequently, it can be

suggested that the conversation of the targets in video 1 is more structured and
more predictable than the conversation between the targets in the second videotape.
The findings provide evidence that the Empathic Accuracy Task is a valid
method to measure more subtle mind-reading impairments of adults with PDD.
However, it should be noted that unlike the Eyes Task or the Strange Stories Task,
which are relatively pure theory of mind task, deprived of contextual information
(Baron-Cohen et al., 1997), the Empathic Accuracy Task uses movies, which
involve executive function and central coherence components as well as mindreading abilities.
RESEARCH OVERVIEW
The goal of the present thesis is to extend our knowledge on the mind-reading
abilities and disabilities of adolescents and adults with PDD. The approach of this
thesis adopts the sensible line that we should aim to measure the mind-reading
ability of persons with PDD to comment on social events using the language of
internal mental states, without (explicitly or implicitly) committing oneself to any
particular perspective. Therefore, over the chapters, we jump back and forth among
methodologies that come from a variety of theoretical perspectives. For instance,
we use methodologies that come from the theory of mind perspective of BaronCohen and colleagues as well as methodologies that come from the social
psychological approach of Ickes and colleagues. Below we present a short
overview of the different chapters (see also Table 1).
Chapter 2. In this study, we assess whether the findings of Roeyers et al. (2001)
keep upright with adults with Asperger syndrome. The rationale behind this study
is that the empathic accuracy paradigm will be especially suitable to the case of
impaired empathic accuracy in adults with PDD if it proves able to capture even the
more subtle impairments in adults with Asperger syndrome, a milder form of
autism. We improve the methodology of the Empathic Accuracy Task by digitising
the analogue version of Roeyers et al. (2001). Therefore, we use VIDANN (De
Chapter 1
Clercq, Buysse, Roeyers, Ickes, Ponnet, & Verhofstadt, 2001), a system

specifically designed to computerise each of the phases in the empathic accuracy
design. We further explore some differences between the targets actual thoughts
and feelings of both videotapes, and calculate more detailed empathic accuracy
scores in terms of content of thoughts and feelings.
Table 1
Overview of the chapters
Chapter 2
Groups
Materials
19
PDD
Strange Stories
19
Control
Eyes Task
Empathic Accuracy Task (standard design)
Chapter 3
Chapter 4
Chapter 5
Empathic Accuracy Task (dyadic interaction design)
11
PDD
11
Control
30
PDD
30
Control
30
PDD
Interpersonal Reactivity Index (self- & other-report)
30
Control
Emotional Quotient Inventory (self-report)
New Empathic Accuracy Task (standard design)
Revised Eyes Task

Chapter 6
30
PDD
30
Control
Levels of Processing Task
Chapter 3. In this chapter, we switch from the standard stimulus empathic

accuracy design to the dyadic interaction design. Since this design is supposed to
match more closely the real demands of actual social situations, we expect
pronounced within-dyad differences. In the first part of the study, eleven normally
intelligent adults with PDD participate and are videotaped with a concealed camera
while having an initial conversation with a typically developing stranger. The
dyadic interaction empathic accuracy paradigm allows us to study both the overt
behaviour of the interacting dyads (e.g. gazing, verbalisation or stimulatory
Introduction
gestures) and the covert thoughts and feelings of the participants. We assess the
difficulty, abstractness and thematic topic of the participants unexpressed thoughts
and feelings, and work out the empathic accuracy scores in terms of content of
thoughts and feelings. In the second part of the study, we attempt to sort out
whether or not the combination between some behavioural characteristics and the
content of the dyad members thoughts and feelings affects the inference ability of
both groups.
Chapter 4. Based on the results of Chapter 3, we attempt to explore the possible
significance of structure to the empathic accuracy of subjects with PDD and control
subjects. The procedure used is the standard stimulus empathic accuracy paradigm
in which the participants each view two videotaped getting acquainted
conversations between two typically developing strangers, and attempt to infer the
thoughts and feelings of the same set of target persons. The structure of both
videotapes is manipulated so that one videotape is more structured than the other.
We calculate the content accuracy as well as the valence accuracy of the
participants. We further refine the content empathic accuracy scores by calculating
the scores along variables such as difficulty, abstractness, and affective tone of the
videotaped targets thoughts and feelings, and the videotaped strangers
verbalisation and their direction of gaze.
Chapter 5. In the previous studies, we made use of performance-based mindreading tasks. In this chapter, we broaden our scope and investigate both the
perceived and performed mind-reading in adults with PDD and typically
developing controls (i.e. how empathic people think they are versus how empathic
they actually are). We use the Emotional Quotient Inventory (EQ-i) of Bar-On
(1997) and the Interpersonal Reactivity Index (IRI) of Davis (1980) as perceived
mind-reading tasks. While the EQ-i is used as a self-report questionnaire, the IRI is
used as joint self-report/other-report measure. The performed mind-reading is
tested using the Revised Eyes Task of Baron-Cohen and colleagues (Baron-Cohen,
Wheelwright, et al., 2001).
Chapter 1
10
Chapter 6. A logical hypothesis derived from the previous studies is that the
schematic knowledge of adults with PDD differs from that of typically developing
adults. Since each of the empathic accuracy tasks made use of the same script (i.e.
the videotaped targets had to make acquaintance with each other), it can be argued
that the presence of a getting acquainted script in participants may improve their
performance on empathic accuracy tasks. The aim of this study is to investigate the
schematic knowledge (including the getting acquainted script) and levels of
processing (LOP) in adults with PDD and typically developing controls. Therefore,
we develop two LOP Tasks in which two types of independent measures can give
us insight into information processing: a content measure and a reaction time
measure.
Chapter 7. In this chapter, we take up the thread of the empathic accuracy story
and explore some alternative explanations to the non-supportive empathic accuracy
findings. Therefore, we combine parts of the data-sets of the previous studies. We
reanalyse the existing data using Social Relations Model analyses, explore the
power of empathic accuracy studies, and question the reliability and validity of the
empathic accuracy design.
Chapter 8. The final chapter contains an overview and general discussion of the
main findings. Implications for clinical practice and directions for future research
are worked out.
It should be noted that this thesis consists of several papers, which have been
submitted for publication, are under editorial review or are currently published.
Since each of the chapters is a self-contained manuscript, the text of some of the
chapters may partially overlap.
Introduction
11
REFERENCES
American Psychiatric Association. (1994). Diagnostic and statistical manual of mental
disorders. Fourth edition. DSM-IV. Washington DC : Author.
disorders. Fourth edition. DSM-IV-TR. Washington DC : Author.
Bar-On, R. (1997). BarOn Emotional Quotient Inventory: Technical manual. Toronto:
Multi Health Systems.
Baron-Cohen, S. (1989). The autistic childs theory of mind: A case of specific
developmental delay. Journal of Child Psychology and Psychiatry, 30, 285-297.
Baron-Cohen, S., Jolliffe, T., Mortimore, C., & Robertson, M. (1997). Another advanced
test of theory of mind: Evidence from high-functioning adults with autism or Asperger
syndrome. Journal of Child Psychology and Psychiatry, 38, 813-822.
Baron-Cohen, S., Leslie, A., & Frith, U. (1985). Does the autistic child have a theory of
mind? Cognition, 21, 37-46.
Baron-Cohen, S., Leslie, A., & Frith, U. (1986). Mechanical, behavioural and intentional
understanding of picture stories in autistic children. British Journal of Developmental
Psychology, 4, 113-125.
Baron-Cohen, S., ORiordan, M., Stone, V., Jones, R., & Plaisted, K. (1999). Recognition
of faux pas by normally developing children and children with Asperger syndrome or
high-functioning autism. Journal of Autism and Developmental Disorders, 29, 407-418.
Baron-Cohen, S., Tager-Flusberg, H., & Cohen, J. (2000). Understanding other minds:
Perspectives from developmental cognitive neuroscience. Oxford: University Press.
Baron-Cohen, S., Wheelwright, S., Hill, J., Raste, Y., & Plumb, I. (2001). The Reading
the Mind in the Eyes Test revised version: A study with normal adults, and adults with
Asperger syndrome or high-functioning autism. Journal of Child Psychology and
Psychiatry, 42, 241-253.
Baron-Cohen, S., Wheelwright, S., Skinner, R., Martin, J., & Clubley, E. (2001). The
autism-spectrum quotient (AQ): Evidence from Asperger syndrome/high-functioning
autism, males and females, scientist and mathematicians. Journal of Autism and
Developmental Disorders, 31, 5-17.
Bauminger, N., & Kasari, C. (1999). Brief Report: Theory of mind in high-functioning
children with autism. Journal of Autism and Developmental Disorders, 29, 81-86.
Chapter 1
12
Bolton, P., MacDonald, H., Pickles, A., Rios, P., Goode, S., Crowson, M., Bailey, A., &
Rutter, M. (1994). A case-control family history study of autism. Journal of Child
Psychology and Psychiatry, 35, 877-900.
Booth, R., Charlton, R., Hughes, C., & Happ, F. (2003). Disentangling weak coherence
and
executive
dysfunction:
Planning
drawing
in
autism
and
attention
deficit/hyperactivity disorder. Philosophical Transactions of the Royal Society of

London Series B-Biological Sciences, 358, 387-392.
Bowler, D. (1992). Theory of mind in Aspergers syndrome. Journal of Child Psychology
and Psychiatry, 33, 877-893.
Buysse, A., & Ickes, W. (1999). Topic-relevant cognition and empathic accuracy in
laboratory discussions of safer sex. Psychology and Health, 14, 351-366.
Dahlgren, S. , & Trillingsgaard, A. (1996). Theory of mind in non-retarded children with
autism and Aspergers syndrome. Journal of Child Psychology and Psychiatry, 37,
759-763.
Davis, M. H. (1980). A multidimensional approach to individual differences in empathy.
JSAS Catalog of Selected Documents in Psychology, 10, 85.
De Clercq, A., Buysse, A., Roeyers, H., Ickes, W., Ponnet, K., & Verhofstadt, L. (2001).
VIDANN: A video annotation system. Behaviour Research Methods, Instruments, &
Computers, 33, 159-166.
Eisenberg, N., Murphy, B., & Shepard, S. (1997). The development of empathic accuracy.
In W. Ickes (Ed.), Empathic accuracy (pp.73-116). New York: Guilford Press.
Eisenberg, N., Shea, C., Carlo, G., & Knight, G. (1991). Empathy-related responding and
cognition: a chicken and the egg dilemma. In W. Kurtines & J. Gewirtz (Eds.),
Handbook of moral behavior and development: Vol. 2. Research (pp. 63-68). Hillsdale
NJ: Lawrence Erlbaum.
Fein, D., Lucci, D., Braverman, M., & Waterhouse, L. (1992). Comprehension of affect in
children with pervasive developmental disorders. Journal of Child Psychology and
Psychiatry, 33, 1157-1167.
Gesn, P.R., & Ickes, W. (1999). The development of meaning contexts for empathic
accuracy: Channel and sequence effects. Journal of Personality and Social Psychology,
77, 746-761.
Introduction
13
Happ, F. (1994). An advanced test of theory of mind: Understanding of story characters

thoughts and feelings by able autistic, mentally handicapped, and normal children and
adults. Journal of Autism and Developmental Disorders, 24, 129-154.
Happ, F. (1995). The role of age and verbal ability in the theory of mind task
performance of subjects with autism. Child Development, 66, 843-855.
Heavey, L., Phillips, W., Baron-Cohen, S., & Rutter, M. (2000). The Awkward Moment
Test: A naturalistic measure of social understanding in autism. Journal of autism and
Developmental Disorders, 30, 3, 225-236.
Ickes, W. (1993). Empathic Accuracy. Journal of Personality, 61, 587-610.
Ickes, W. (1997). Empathic accuracy. New York: Guilford Press.
Ickes, W. (2003). Everyday mind reading: Understanding what other people think and
feel. Amherst, NY: Prometheus Books.
Ickes, W., Bissonnette, V., Garcia, S., & Stinson, L. (1990). Implementing and using the
dyadic interaction paradigm. In C. Hendrick & M.S. Clark (Eds.), Review of personality
and social psychology: Research methods in personality and social psychology, Vol.
11, pp. 16-44. Newbury Park, CA: Sage.
Ickes, W., Gesn, P., & Graham, T. (2000). Gender differences in empathic accuracy:
differential ability or differential motivation? Personal Relationships, 7, 95-109.
Ickes, W., Stinson, L., Bissonnette, V., & Garcia, S. (1990). Naturalistic social cognition:
Empathic accuracy in mixed-sex dyads. Journal of Personality and Social Psychology,
59, 730-742.
Jarrold, C., Butler, D., Cottington, E., & Jimenez, F. (2000). Linking theory of mind and
central coherence bias in autism and in the general population. Developmental
Kaland, N., Moller-Nielsen, A., Callesen, K., Mortensen, E.L., Gottlieb, D., & Smith, L.
(2002). A new advanced test of theory of mind: Evidence from children and
adolescents with Asperger syndrome. Journal of Child Psychology and Psychiatry, 43,
517-528.
Kleinman, J., Marciano, P., & Ault, R. (2001). Advanced theory of mind in highfunctioning adults with autism. Journal of Autism and Developmental Disorders, 31,
29-36.
Chapter 1
14
Klin, A. (2000). Attributing social meaning to ambiguous visual stimuli in higherfunctioning autism and Asperger syndrome: The social attribution task. Journal of
Child Psychology and Psychiatry, 41, 831-846.
Lawson, J. (2003). Depth accessibility difficulties: An alternative conceptualisation of
autism spectrum conditions. Journal for the Theory of Social Behaviour, 33, 189-202
Leekam, S., & Perner, J. (1991). Does the autistic child have a metarepresentational
deficit? Cognition, 40, 203-218.
Leslie, A., & Frith, U. (1987). Metarepresentation and autism: How not to lose ones
marbles. Cognition, 27, 291-294.
Marangoni, C., Garcia, S., Ickes, W., & Teng, G. (1995). Empathic accuracy in a clinically
relevant setting. Journal of Personality and Social Psychology, 68, 854-869.
Ozonoff, S., & McMahon-Griffith, E. (2000). Neuropsychological function and the
external validity of Asperger syndrome. In A. Klin, F. Volkmar & S. Sparrow (Eds.),
Aspergers syndrome (pp. 72-96). New York: Guilford Press.
Parsons, S., & Mitchell, P. (2002). The potential of virtual reality in social skills training
for people with autistic spectrum disorders. Journal of Intellectual Disability Research,
46, 430-443.
Perner, J. (1991). Understanding the representational mind. Cambridge, MA: MIT Press.
Perner, J., Frith, U., Leslie, A., & Leekam, S. (1989). Exploration of the autistic childs
theory of mind: Knowledge, belief, and communication. Child Development, 60, 689700.
Perner, J., & Wimmer, H. (1985). John thinks that Mary thinks that Attribution of
second-order beliefs by 5-10 year old children. Journal of Experimental Child
Prior, M. R., Dahlstrom, B., & Squires, T. L. (1990). Autistic childrens knowledge of
thinking and feeling states in other people. Journal of Child Psychology and
Rapin, I. (1997). Classification and causal issues in autism. In D. Cohen & F. Volkmar
(Eds.), Handbook of autism and developmental disorders. Second Edition (pp. 846867). New York: John Wiley & Sons.
Roeyers, H., Buysse, A., Ponnet, K., & Pichal, B. (2001). Advancing advanced mindreading tests: Empathic accuracy in adults with a pervasive developmental disorder.
Journal of Child Psychology and Psychiatry, 42, 271-278.
Introduction
15
Rutterford, M. D., Baron-Cohen, S., & Wheelwright, S. (2002). Reading the mind in the
voice: A study with normal adults and adults with Asperger syndrome and highfunctioning autism. Journal of Autism and Developmental Disorders, 32, 189-194.
Simpson, J., Ickes, W., & Blackstone, T. (1995). When the head protects the heart:
Empathic accuracy in dating relationship. Journal of Personality and Social
Stinson, L., & Ickes, W. (1992). Empathic accuracy in the interactions of male friends
versus male strangers. Journal of Personality and Social Psychology, 62, 787-797.
Szatmari, P. (2003). The causes of autism spectrum disorder. British Medical Journal,
326, 173-174.
Tager-Flusberg, H., & Sullivan, K. (1995). Attributing mental states to story characters: A
comparison of narratives produced by autistic and mentally retarded individuals.
Applied Psycholinguistics, 16, 241-256.
Volkmar, F., & Klin, A. (2000). Diagnostic issues in Aspergers syndrome. In A. Klin, F.
Volkmar and S. Sparrow (Eds.), Aspergers syndrome (pp. 60-63). New York: Guilford
Press.
Wellman, H. (1990). The childs theory of mind. Cambridge, MA: MIT Press.
Wimmer, H., & Perner, J. (1983). Beliefs about beliefs: Representation and constraining
function of wrong beliefs in young childrens understanding of deception. Cognition,
13, 103-128.
Wing, L. (1988). The continuum of autistic characteristics. In E. Schopler & G. Mesibov
(Eds.), Diagnosis and assessment in autism (pp. 91-110). New York: Plenum Press.
Whiten, A. (1991). Natural theories of mind. Oxford: Basil Blackwell.
CHAPTER
Advanced mind-reading in adults with

1
Asperger syndrome
This chapter is based on Ponnet, K., Roeyers, H., Buysse, A., De Clercq, A., & Van der Heyden, E. (in
press). Advanced mind-reading in adults with Asperger syndrome. Autism.
Chapter 2
18
ABSTRACT
This study investigated the mind-reading abilities of 19 adults with Asperger
syndrome and 19 typically developing adults. For this aim, two static mind-reading
tests and a more naturalistic empathic accuracy task were used. In the empathic
accuracy task, participants attempted to infer the thoughts and feelings of target
persons, while viewing a videotape of the target persons in a naturally occuring
conversation with another person.
The results are consistent with earlier findings (Roeyers, Buysse, Ponnet, &
Pichal, 2001). The empathic accuracy task enabled us to find significant betweengroup differences, where static mind-reading tasks failed. More specifically, a
difference in empathic accuracy between the Asperger group and the control group
was found on the second videotape of the task.
The most innovating finding of the present study is that the inference ability of
adults with PDD and control adults depends on the focus of the targets thoughts
and feelings, and that the empathic accuracy of adults with Asperger syndrome and
control adults might be different in terms of quantity and quality.
Empathic accuracy in adults with Asperger syndrome
19
INTRODUCTION
One of the most striking characteristics of persons with autism or a related
pervasive developmental disorder (PDD) appears to be their qualitative impairment
in social interaction, which affects almost every aspect of their daily-life
functioning (Howlin, 1998; Zager, 1999). Among the theories that have been
proposed to account for the social interaction problems are an inability to process
emotional information and an inability to comprehend the perspectives of others, to
be known as the theory of mind hypothesis.
The socio-affective theory was proposed by Hobson and colleagues, who found
children with autism to be impaired in their comprehension of facial, vocal and
bodily affect, and in their associations of these expressions with each other
(Hobson, 1986, 1991, 1999; Hobson & Lee, 1989; Weeks & Hobson, 1987).
Consequently, Hobson has argued that the primary psychological deficit in autism
is an early failure of direct perception of bodily expressions, including emotions.
However, the socio-affective theory has been criticized for the inconsistent results
that were found when subjects with autism were matched with typically developing
retarded control subjects (Rutter & Bailey, 1993). The theory of mind hypothesis
suggests that the impairments of persons with PDD can be explained in terms of a
failure to attribute mental states to oneself and to others (for an overview, see
Baron-Cohen, Tager-Flusberg, & Cohen, 2000). Without a theory of mind, people
are unable to develop a normal understanding that other peoples have mental states,
which results in socially inadequate behaviour. Over the past decades, a large
amount of studies has demonstrated the robustness of the theory of mind
hypothesis, although a substantial proportion of able children (and older people)
with PDD were found to pass second-order theory of mind tasks (Bowler, 1992;
Ozonoff, Pennington, & Rogers, 1991; Tager-Flusberg & Sullivan, 1995).
However, we have to take into account that the current tests developed to
measure the social difficulties of persons with PDD have little ecological validity.
Indeed, it can hardly be argued that the use of static stimuli (such as tales,
drawings, photographs and others) is a naturalistic way to evaluate the social
understanding abilities of subjects with or without autism (Baron-Cohen, Jolliffe,
Mortimore, & Robertson, 1997; Baron-Cohen, Wheelwright, Hill, Raste, & Plumb,
Chapter 2
20
2001; Kleinman, Marciano, & Ault, 2001). Therefore, the approach of the present
study adopts the sensible line that since individuals with autism experience
difficulties with social interaction, we should attempt to measure their social
functioning in contexts that mirror as closely as possible real-life social
interactions. Moreover, we should aim to measure the ability of persons with PDD
to comment on social events using the language of internal mental states, without
(explicitly or implicitly) committing oneself to any particular perspective. The
approach of the present study is therefore designated as one of eclecticism pending
improved methodologies.
In our quest for improved methodologies, we ran into the empathic accuracy
design of Ickes and colleagues (Ickes, 1997; Ickes, Bissonnette, Garcia, & Stinson,
1990). The development of empathic accuracy has deep roots in developmental
psychology, but the terminology used varied over the years (Eisenberg, Murphy, &
Sheppard, 1997). According to Ickes and colleagues, empathic accuracy can be
defined as the degree to which someone is able to infer accurately the specific
content of another persons thoughts and feelings (Ickes, Stinson, Bissonnette, &
Garcia, 1990). As described by Ickes (1997), the most unequivocal way to measure
empathic accuracy is by rating the similarity between the content of targets actual
thought or feeling and the content of the perceivers inferences. The term empathy
includes a cognitive as well as an emotional component. While the cognitive
component refers to the capacity to understand others internal states, the emotional
component refers to the emotional response resulting from the recognition of other
persons emotional state or condition (Eisenberg, Shea, Carlo, & Knight, 1997).
Recently, Roeyers, Buysse, Ponnet and Pichal (2001) tested high-functioning
adults with a pervasive developmental disorder using two static mind-reading tests
and a more naturalistic Empathic Accuracy Task. The two static mind-reading tests
used were an adapted and shortened version of Happs (1994) Strange Stories, and
an adapted version of Baron-Cohen et al.s (1997) Eyes Task. The more naturalistic
test was derived from the empathic accuracy design (Ickes, 1997; Ickes,
Bissonnette, et al., 1990). Based on this approach, Roeyers et al. (2001) designed
an Empathic Accuracy Task in which individual participants each attempt to infer
aspects of a target persons actual subjective experience while viewing a videotape
of the target person in a naturally occurring conversation with another person.
21
Roeyers et al. (2001) found the adults with PDD not to be impaired with regard
to the static tasks compared to control adults. These findings differ from other
studies (Baron-Cohen et al., 1997; Happ, 1994; Jolliffe & Baron-Cohen, 1999;
Kleinman et al., 2001), although it should be mentioned that participants with PDD
in the Roeyers et al. (2001) study had higher cognitive abilities (mean IQ: 113) than
participants in the other studies. On the other hand, the naturalistic Empathic
Accuracy Task was able to distinguish clearly between both groups,
notwithstanding the fact that the task produced mixed results. Although the
Empathic Accuracy Task consists of two videotapes, a significant difference was
only found between both groups on the second videotape in favour of the control
adults. A possible explanation for these mixed results is the different nature of both
videotapes. While both targets in the first videotape were playing a board game, the
targets in the second videotape were having a conversation of the getting
acquainted type. Consequently, the authors suggest that the conversation of the
targets in video 1 is structured around the game to a large extent. The conversation
of the targets in video 2 may be less predictable and more complex than the
conversation between the targets in the first videotape because there is no support
by any tangible board game material.
In the present study, we tried to extend the study of Roeyers et al. (2001) in two
ways. First of all, our participants with an autism spectrum disorder were all
clinically diagnosed as having Asperger syndrome. ICD-10 (World Health
Organisation, 1992) defined Asperger syndrome as a variant of autism in which
there are abnormalities in reciprocal social interaction and in patterns of behaviour,
interests and activities. The lack of any clinically significant delay in (spoken or
receptive) language or cognitive development distinguishes Asperger syndrome
from autism. The language criteria relate to the onset of single words by age 2 years
and phrase speech by age 3 years. Despite the diagnostic difficulties that clinicians
experience to distinguish both syndromes from each other, or the controversy
whether or not autism and Asperger syndrome lie on the same continuum (Bishop,
2000; Mayes & Calhoun, 2001; Wing, 1998), Asperger syndrome is generally
understood as a milder form of autism (Attwood, 1998; Gilchrist, Green, Cox,
Burton, Rutter, & Le Couteur, 2001; Miller & Ozonoff, 2000; Ozonoff, South, &
Miller, 2000; Rinehart, Bradshaw, Moss, Brereton, & Tonge, 2001; Szatmari,
Chapter 2
22
Archer, Fisman, Streiner, & Wilson, 1995). Therefore, obtaining the same results as
Roeyers et al.s (2001) study would strengthen the case that empathic accuracy is
able to capture subtle mind-reading differences of adults with Asperger syndrome
from those of control adults. In this study, the participants with Asperger syndrome
were compared with typically developing adults.
Second, the mixed results of Roeyers et al. (2001) induced us to explore the
focus or theme(s) of the targets original thoughts and feelings. Therefore, we
examined the thematic topic of the targets actual thoughts and feelings for both
videotapes. More specifically, we assessed whether the focus of the original
thoughts and feelings was on (a) the self, (b) the interaction partner, (c) other
person(s), (d) the research context, (e) a tangible environmental object or event, or
(f) a past memory. Consequently, with respect to the topic, we expect to gather
insight about potential differences between the targets actual thoughts/feelings.
According to the content, we further worked out the empathic accuracy scores of
each participant for all the thoughts and feelings belonging to the 6 different
categorical classifications of the actual thoughts and feelings. By doing so we
examined the possibility of different between-group patterns with regard to the
level of empathic accuracy.
We improved the methodology of the Empathic Accuracy Task by digitising the
analogue version used before. This was realised with VIDANN, a system
specifically designed to computerise each of the phases in the empathic accuracy
design (see De Clercq, Buysse, Roeyers, Ickes, Ponnet and Verhofstadt (2001) for
more technical details). The advantages of a digitised Empathic Accuracy Task are
numerous. Logically, digitisation leads to an increased standardisation and, in turn,
results in more accurate empathic accuracy scores. Furthermore, using VIDANN
minimises the inaccuracies related to the use of a manual remote control system to
stop a videotape at specific time moments (De Clercq et al., 2001). In addition, a
less active role is required of the research assistant, which decreases the possibility
of distraction. People with PDD have been described as more easily distracted by
trivial cues, for instance involuntary sounds such as coughing, sneezing or shuffling
ones feet (Frith, 1989; Howlin, 1998).
Based on the Roeyers et al. (2001) study, we predict no between-group
differences with regard to the static mind-reading tasks, but we expect significant
23
differences with regard to the Empathic Accuracy Task between the Asperger
group on the one hand and the control group on the other hand. Furthermore, we
want to investigate different patterns between the Asperger group and the control
group with regard to their empathic accuracy for the thoughts and feelings with a
specific topic.
METHOD
Participants
Two groups of normally intelligent adolescents and adults participated in the
present study: 19 subjects with Asperger syndrome and a control group of 19
typically developing persons. Male/female ratio (14/5) was similar in each group.
The participants were recruited with the help of parent associations and schools. All
subjects with Asperger syndrome had been diagnosed by a multidisciplinary team
of experienced clinicians and fulfilled the ICD-10 criteria as stringent as possible,
given the nosologically controversial diagnosis of Asperger syndrome (see
Eisenmajer, Prior, Leekam, Wing, Gould, Welham, & Ong, 1996; Leekam, Libby,
Wing, Gould, & Gillberg, 2000; Mayes & Calhoun, 2001). They showed qualitative
deficiencies in reciprocal social interaction and restricted stereotyped patterns of
behaviour, interests and activities. Subjects with a history of delay in development
of spoken language were excluded (i.e. phrase speech was present by age 3 years).
Motor clumsiness was presented by some of the subjects, but not by all. None of
the subjects had comorbid diagnosis of attention deficit disorder and/or Tourettes
disorder. All subjects with Asperger syndrome were studying or employed, and
some of them lived independently or were married.
Only Asperger subjects with total IQ-scores above 80 were included in the
sample. The control subjects were assessed for intelligence using a shortened
WAIS version, comprising Block Design, Vocabulary, Arithmetic, and Picture
Arrangement. The estimated IQ scores for the subjects with Asperger syndrome
were: Total IQ, M = 106.58 (SD = 15.11), Verbal IQ, M = 108.32 (SD = 13.99), and
Performance IQ, M = 104.05 (SD = 18.08). The estimated IQ scores for the control
Chapter 2
24
subjects were: Total IQ, M = 114.05 (SD = 15.84), Verbal IQ, M = 116.00 (SD =
18.39), and Performance IQ, M = 110.58 (SD = 14.03). A series of univariate
analyses of variance (ANOVAs) revealed no significant between-group differences
for Total IQ, F(1, 36) = 2.21, for Verbal IQ, F(1, 36) = 2.10, and for Performance
IQ, F(1, 36) = 1.54. Furthermore, no significant difference (ANOVA) was found
between the mean chronological age of the subjects with Asperger syndrome (M =
21.06, SD = 4.77) and the mean chronological age of the control subjects (M =
21.93, SD = 6.60), F(1, 36) < 1.
Materials
Stories Task
The Stories Task of Roeyers et al. (2001) is an adapted and shortened version of
Happs (1994) Strange Stories. The presented stimuli consisted of six different
stories. Each story is accompanied by an illustrative picture and contained two
questions, i.e. an open-ended question (Was it true, what X said ?) and a
justification question (Why did X say that ?).
Eyes Photographs Task
The Eyes Photographs Task (Roeyers et al., 2001) is an adapted version of the
Eyes Task of Baron-Cohen et al. (1997) and consists of 25 photographs of the eye
region of the faces of male and female adults, together with three basic (e.g. afraid,
sad) or more complex (e.g. guilty, blas) mental states. Participants were asked to
make a forced choice between the three mental state terms. To be sure that every
participant was looking attentively at the photograph, they were also asked to
identify the gender of the person in each photograph.
Empathic Accuracy Stimulus Tapes
The two videotapes used in the Empathic Accuracy Task were exactly the same
as those used in the study of Roeyers and colleagues (see Roeyers et al., 2001, p.
273). The stimulus tapes were prepared as follows. Four typically developing
persons (targets) were invited to come to our laboratory. They came two by two
(male-female dyads) and, initially, were strangers to each other. They were told in
advance that they were about to meet another stranger and that their interaction
25
would be videotaped for use in future research. Once they arrived at the laboratory,
the research assistant explained they were to play a board game together. However,
when both targets were seated, the research assistant pretended to have a problem
with the sound equipment. Promising to return in a few minutes and telling the
participants that he had to look for technical assistance, he left the room. At this
point, a video camera was activated and recorded the initial conversation between
the two targets. After approximately 11 minutes, the research assistant returned and
partly debriefed the participants. The research assistant further informed the targets
that their written consent was required for the tape to be used. The stimulus tapes
consisted of two initial conversations between two strangers. In the first videotape,
the participants immediately started to play a board game. The targets in video 2
did not touch any of the board game materials and had a conversation of the
getting acquainted type.
After signing the written consent form, the targets were asked to view their
videotape and to make a written record of all their unexpressed thoughts and
feelings (see also Ickes, Bissonnette, et al., 1990; Roeyers et al., 2001). More
specifically, the targets were instructed to stop the videotape at each point during
the interaction when they remembered having had a specific thought or feeling.
They were asked to write down (a) the time when the thought or feeling occurred
(as displayed by a digital clock on the videotape), (b) whether the entry was a
thought or a feeling, and (c) the specific content of the thought or feeling. Two
additional instructions were given. First of all, the targets were instructed to report
all thoughts and feelings they remembered having had as accurately and honestly as
possible and, secondly, they were instructed not to report any new thoughts or
feelings that may have occurred to them while they were viewing the videotape.
Furthermore, the targets were assured of complete anonymity of their data and were
informed that their interaction partner would never be shown their thought/feeling
responses. After recording the thoughts and feelings, the targets were debriefed
more thoroughly and were asked to sign a written consent.
The first videotape contained 85 thoughts and feelings and videotape 2 contained
36 thoughts and feelings. Because of the large number of thoughts and feelings in
video 1, the number of thoughts and feelings was randomly reduced to 36 thoughts
and feelings (see Roeyers et al., 2001). By doing so, video 1 contained 18 male and
Chapter 2
26
18 female thoughts and feelings and video 2 contained 13 male and 23 female
thoughts and feelings.
Procedure
All participants were seen individually and were given the three different mindreading tasks. They first viewed each of the videotapes, with the order of videotape
presentation counterbalanced among participants. While viewing the videotape, the
tape was paused at the precise moments when a target had recorded a specific
thought or feeling. The videotapes presented to the participants were paused
automatically with the use of VIDANN (De Clercq et al., 2001). Whenever the
videotape was paused, the participants were asked to make inferences about the
specific content of the unexpressed thought/feeling and to write down (a) whether
the entry was presumed to be a thought or feeling, and (b) the specific content of
the thought or feeling. To ensure that the participants clearly understood the
procedure of the Empathic Accuracy Task, an advanced preparatory session was
given.
After the participants had finished the Empathic Accuracy Task, the two static
mind-reading tasks were given. First, each of the six stories was placed in front of
the participants. After each of the stories had been read out in a fixed order, the
comprehension and the mental state question were asked. Following the Stories
Task, the 25 photographs of the eye region were presented in a fixed order, together
with three mental state terms. Each participant had to make a forced choice
between the three mental state terms and was asked to identify the gender of the
person in the photograph. Because no recent intelligence scores were available
from the participants, they were assessed on verbal and non-verbal mental age,
using the shortened WAIS.
27
Measures
Stories Task
In accordance with the scoring procedure of Happ (1994) and Roeyers et al.
(2001), participants were scored on their answers to the justification question. The
justification could be a physical or a mental explanation, and both explanations
could be correct or incorrect.
Participants were scored on their correct answers (1 point for a correct answer)
on both the emotion recognition and gender recognition task (see Baron-Cohen et
al., 1997). Because 25 photographs were presented, the minimum and maximum
scores of each participant could vary between 0 to 25, both for the emotion
recognition and gender recognition task.
Empathic Accuracy Task : Measuring Actual Thoughts and Feelings
Six independent coders watched the videotapes that were stopped at the times
the targets had reported having had a thought or feeling. The coders were instructed
to assess for each of the targets thoughts and feelings whether the focus of the
thought/feeling was on (a) the self, (b) the interaction partner, (c) other person(s),
(d) the research context, (e) a tangible environmental object or event, or (f) a past
memory. The interrater reliabilities (Cronbachs alpha) for five of the six
categorical classifications were above .80, except for the tangible environmental
object or event category ( = .57).
Empathic Accuracy Task : Measuring Inferred Thoughts and Feelings
Following the logic and procedures described by Ickes and colleagues (Ickes,
1997; Ickes, Bissonnette et al., 1990; Marangoni, Garcia, Ickes, & Teng, 1995), the
global empathic accuracy scores were computed by comparing each participants
inference with the corresponding thoughts and feelings obtained from the targets.
Five nave and independent coders were asked to compare each participants
inferred thought/feeling with the corresponding original thought/feeling and to rate
the level of similarity on a 3-point scale, ranging from 0 (essentially different
content) through 1 (somewhat similar but not the same content) to 2 (essentially the
Chapter 2
28
same content), with I dont know and missing responses rated 0. The internal
consistency (Cronbachs alpha) of the five judges content accuracy ratings was .80
for Video 1 and .82 for Video 2. Then, as explained more detailed in Appendix A,
the global empathic accuracy was calculated for each participant and for each
videotape.
Similarly, twelve empathic accuracy scores for each participant for each video
were computed for all the thoughts and feelings whether or not belonging to the 6
different categorical classifications of the actual thoughts and feelings, more
specifically (a) the self, (b) the interaction partner, (c) other person(s), (d) the
research context, (e) a tangible environmental object or event, or (f) a past memory.
The internal consistency (Cronbachs alpha) was calculated for each topic and per
videotape. All internal consistencies were acceptable, with one exception (see
Appendix B). We refer to Appendix A for the procedural details.
RESULTS
Stories Task
The ANOVA testing revealed no between-group differences for the total number
of correct and incorrect mental justifications (see Table 1). Similarly, no betweengroup differences were found for the total number of correct and incorrect physical
justifications. The mean scaled scores and standard deviations obtained by each
group are presented in Table 1.
The mean scores of both groups were almost identical on the Eyes Photographs
and the Gender Task (see Table 1). Univariate analyses of variance (ANOVAs)
revealed no between-group differences on the Eyes Photographs and the Gender
Task.
29
Table 1
Between-group differences on static mind-reading tasks
Asperger
Control
SD
F(1, 36)
SD
Mental
4.57
1.21
4.61
.92
.01
Physical
1.10
.94
1.19
.67
.11
Mental
.22
.73
.13
.32
.23
Physical
.10
.46
.06
.23
.13
Eyes Photographs
18.63
3.27
19.31
3.20
.42
Gender Task
21.37
1.50
21.53
1.58
.10
Video 1
32.63
20.38
40.76
13.01
2.15
Video 2
19.59
13.23
31.55
12.77
8.03 **
Stories Task
Correct Justifications
Incorrect Justifications
Global EA
*p < .05; **p < .01; *** p < .005
Empathic Accuracy Task

Differences between both Videotapes in Targets Actual Thoughts and
Feelings
A series of univariate analyses (ANOVAs) with the different topics of the
selected thoughts and feelings belonging to the targets as dependent variable and
videotape as factor, revealed significant differences between both videotapes. Table
2 shows that, compared with the targets of video 2, the targets of video 1 had
significantly more thoughts and feelings that focussed on the research context and
on a tangible environmental object or event. Conversely, the targets of video 1 had
less thoughts and feelings focussed on self and a past memory. There are no
differences between both videotapes with regard to the number of thoughts and
feelings concerning the interaction partner, and other person(s).
Chapter 2
30
Table 2
The percentages of thoughts and feelings with a specific topic
Total
Video 1
Video 2
Self
61.3
47.6
76.3
6.49 **
Interaction partner
45.0
35.7
55.3
3.73
Other person(s)
21.3
19.0
23.7
.98
Research context
61.3
83.3
36.8
15.97 ***
Object or event
50.0
64.3
34.2
10.96 ***
6.3
11.9
0.0
4.13 *
Past memory
F(1, 68)
*p < .05; **p < .01; ***p < .005
Differences between both Groups in Participants Inferred Thoughts and

Feelings
As shown in Table 1, the ANOVA testing for between-group differences
regarding the Empathic Accuracy task revealed no significant difference with
regard to the inferences of the unexpressed thoughts and feelings of the targets in
video 1. As expected, a significant between-group difference was found on video 2.
The mean empathic accuracy scores of thoughts and feelings belonging to a
specific topic are shown in Table 3, represented separately for both videotapes and
for each group. We conducted a series of 2 (Group: Asperger versus Control) x 2
(Topic: Presence versus Absence of a particular focus) ANOVAs, with Group as
between-subjects factor and Topic as within-subjects factor.
As was shown in prior analyses (see Table 1), video 1 did not distinguish
between both groups, whereas video 2 did. The detailed analyses displayed in Table
4 confirm our previous findings and reveal no significant between-group
differences for the more detailed empathic accuracy scores in video 1, whereas a
significant between-group difference was found for each of the empathic accuracy
scores in video 2, with control subjects reaching higher empathic accuracy than
subjects with Asperger syndrome.
31
Table 3
Empathic accuracy along topic-relevant thoughts and feelings (in percentages)
Asperger
Thoughts/Feelings:
Control
With
Without
With
Without
a focus on
a focus on
a focus on
a focus on
(SD)
(SD)
(SD)
(SD)
Video 1
Self
63.00 (25.02)
55.84 (22.04)
63.79 (20.87)
60.62 (15.37)
Partner
48.49 (20.80)
65.22 (25.16)
56.14 (16.53)
65.46 (18.81)
Other person(s)
78.55 (36.20)
54.52 (21.04)
84.08 (28.54)
57.12 (16.64)
Research context
61.95 (22.74)
45.71 (30.79)
65.17 (17.23)
46.92 (20.25)
Object or event
62.42 (24.00)
53.54 (24.31)
64.99 (17.84)
56.98 (19.20)
Past memory
57.05 (32.68)
59.54 (21.96)
76.42 (27.74)
60.20 (17.08)
Self
35.72 (16.74)
31.23 (22.52)
44.21 (12.97)
45.61 (15.63)
Partner
32.68 (16.87)
36.84 (16.07)
43.06 (13.06)
46.26 (16.55)
Other person(s)
24.21 (15.27)
37.82 (16.50)
35.79 (14.38)
47.16 (13.28)
Research context
50.03 (18.58)
25.97 (14.97)
63.37 (20.33)
32.76 (11.05)
Object or event
30.53 (20.01)
34.41 (12.93)
44.14 (12.06)
41.34 (13.46)
Video 2
As shown in Table 4, the analyses revealed 5 main effects of topic for video 1,
indicating that the participants reached higher accuracy rates for the thoughts and
feelings focussed on the self, other persons, the research context and a tangible
environmental object or event, than those without such a focus. On the contrary, the
thoughts and feelings of video 1 that focussed on the interaction partner are less
accurately inferred than the thoughts and feelings without this focus.
Furthermore, an interaction effect (Topic x Group) was found for video 1
between group and absence versus presence of past memory as focus of the
thought/feeling. The Asperger group displayed less empathic accuracy when the
focus of the thoughts and feelings was on a past memory, whereas the control
group displayed more empathic accuracy when the focus of the thoughts and
feelings was on a past memory.
Chapter 2
32
Table 4
F-values for Topic and Group effect
Topic
Group
Topic x Group
F(1, 36)
F(1, 36)
F(1, 36)
4.54 *
.68
.68
Interaction partner
25.93 ***
.41
2.10
Other person(s)
35.06 ***
.29
.11
Research context
23.70 ***
.11
.08
Object or event
8.56 **
.22
.02
Past memory
2.92
1.90
5.43 *
.23
6.18 *
.83
2.25
4.90 *
.04
30.69 **
5.97 *
.25 a
109.52 ***
5.01 *
1.14 a
.07
5.53 *
2.51
Video 1
Self
Video 2
Self
Interaction partner
Other person(s)
Research context
Object or event
*p < .05; **p < .01; ***p < .005
a
p > .25, which indicates that the main effects of Topic and Group are additive
With regard to video 2, two main effects of topic were found. The mean
empathic accuracy scores of thoughts and feelings that focussed on the research
context reached higher accuracy rates than those of thoughts and feelings without
such a focus and, on the other hand, the thoughts and feelings that focussed on
other persons are less accurately inferred than the thoughts and feelings without
such a focus. In addition, the group and topic effects were found to be additive for 2
out of the 5 empathic accuracy scores in video 2. While the empathic accuracy rates
of both groups were the lowest for thoughts and feelings that focussed on other
persons (see Table 3), the significant main effect of Group revealed that adults with
Asperger syndrome had lower empathic accuracy scores for thoughts and feelings
that focussed on other persons than control adults. So, the additive effect suggests
that the lowest empathic accuracy among all other empathic accuracy rates was
33
obtained by the Asperger group inferring thoughts and feelings that focussed on
other persons.
Furthermore, while the empathic accuracy rates of both groups were the lowest
for thoughts and feelings that did not focus on the research context (see Table 3),
the significant main effect of Group revealed that adults with Asperger syndrome
had generally lower empathic accuracy scores than control adults. So, the additive
effect suggests that adults with Asperger syndrome inferring thoughts and feelings
that did not focus on the research context revealed the lowest empathic accuracy
rates.
Time required to complete the Eyes Task and the Empathic Accuracy Task
We administered the time required to complete the Eyes Photograph Task and
the Empathic Accuracy Task (see Table 5). The ANOVAs testing for betweengroup differences revealed no differences between both groups in the average time
required to complete the Eyes Photograph Task. However, time differences were
found for video 1 and video 2 of the Empathic Accuracy Task.
Table 5
Required time to complete the tasks (in seconds)
Asperger
Control
F(1, 36)
SD
SD
Video 1
2794.11
1156.80
2178.11
476.52
4.61 *
Video 2
3341.26
1273.79
2313.74
533.27
10.52 ***
418.32
175.83
345.68
150.08
Empathic Accuracy Task
Eyes Photograph Task
1.88
*p < .05; **p < .01; ***p < .005
Correlates of the Mind-Reading Tasks

The correlations and intercorrelations between the mind-reading tasks,
chronological age and the IQ-scores were calculated. Significant positive
correlations (Pearson) were found between video 1 and video 2, r(38) = .56, p <
.001, and between the Strange Stories and the Eyes Task, r(38) = .33, p < .05. The
other associations yielded not significant.
Chapter 2
34
DISCUSSION
In the present study, two static mind-reading tasks and a more naturalistic task
were performed on high-functioning adults with Asperger syndrome and control
adults. The results are consistent with those reported in the study of Roeyers et al.
(2001). The adults with Asperger, i.e. a group showing milder communication
problems than the group in Roeyers et al. (2001), were not impaired with regard to
the static tasks compared to the control group. However, on the second videotape of
the Empathic Accuracy Task the control group performed significantly better than
the Asperger group.
As a result of the fact that we found no between-group differences on the first
videotape of the Empathic Accuracy Task and significant between-group
differences on the second videotape, we explored some differences between the
targets actual thoughts and feelings of both videotapes. Analyses revealed that the
percentages of the first videotapes original thoughts and feelings belonging to a
specific topic differ from those of video 2. When the mean empathic accuracy was
calculated in terms of content of the thoughts and feelings, analyses confirmed our
previous findings and revealed no significant between-group differences on the
detailed empathic accuracy scores in video 1, whereas significant between-group
differences were found for all detailed empathic accuracy scores in video 2.
Furthermore, analyses revealed some main effects of topic on empathic accuracy
for video 1 and video 2, and a significant interaction effect for video 1 between
Topic and Group, indicating that the control persons are more accurate in inferring
thoughts and feelings that focussed on a past memory, whereas the person with
Asperger syndrome are more accurate in inferring thoughts and feelings without the
focus on a past memory, which -in a way- is consistent with the results of Bowler,
Gardiner and Grice (2000). Finding an interaction effect with the empathic
accuracy design is quite rare and, although only 11.9 percent of all thoughts and
feelings of video 1 are focussed on a past memory, this suggests that persons with
Asperger syndrome use qualitatively different strategies than control persons to
infer thoughts and feelings that focus on a past memory. In addition, the additive
effects suggest that the adults with Asperger syndrome probably use the same
mechanism as the control adults to infer the thoughts and feelings of video 2 that
35
focus on other persons or on the research context. As such, we speculate that the
empathic accuracy of both groups is also quantitatively different, in addition to the
aforementioned qualitative differences.
In sum, the study represents an initial attempt to investigate the mind-reading
abilities of persons with Asperger syndrome, while using a naturalistic design. The
design of the empathic accuracy task has enabled us to find significant differences
between high-functioning persons with Asperger syndrome and the control group,
where static mind-reading tasks failed. More specifically, a difference in empathic
accuracy between the Asperger group and the control group was found on video 2.
Further analyses of the content of the original thoughts and feelings did not reveal
what could account for this between-group difference but suggested that there are
some qualitative and quantitative differences in empathic accuracy scores of the
group with Asperger and those of the control group.
One might suggest that the presentation of static versus dynamic tasks was not
counterbalanced, and that this lack of counterbalancing might have affected the
results. Based on prior research (Roeyers et al., 2001), the authors were well aware
that the Empathic Accuracy Task required quite some mental strain. Therefore, this
task was presented first. The fact that no between-group differences were found on
the static tasks suggests that the required mental strain of the Empathic Accuracy
Task has not affected more the subjects with Asperger syndrome than the control
subjects, otherwise the Asperger group would have performed less on the static
tasks than the control group. However, if the Empathic accuracy Task required
more mental strain from the control subjects than from the subjects with Asperger
syndrome (although there is no reason to assume this), then this can explain the
lack of between-group differences on the static tasks.
In future research, it might be worthwhile to examine the role of the targets
behaviours on the inference abilities of persons with or without a developmental
disorder. It is reasonable to think that persons with an autism spectrum disorder rely
on different behavioural cues to infer the thoughts and feelings of interacting
persons. For instance, in a study of Koning and Magill-Evans (2001) adolescents
with Asperger syndrome reported that they use facial cues more often than other
cues to infer emotions and, in general, they reported using fewer of all available
cues to infer emotional states. Therefore, it might be interesting to assess how much
Chapter 2
36
the targets of our videotapes verbalise and make eye-contact with each other.
Moreover, it might be interesting to examine the congruence between the
verbalisation of the interacting targets and the specific content of the targets
thoughts and feelings. By doing so, it is possible to calculate for each participant
the empathic accuracy belonging to the congruent and incongruent thoughts and
feelings. The hypothesis is that typically developing persons are better at inferring
incongruent thoughts and feelings than persons with PDD. Furthermore, we can
presume that the supporting gesticulation of two interacting persons helps us to
infer their unexpressed thoughts and feelings. The question is raised whether these
targets arm or body movements make it easier or, in contrast, more difficult for
persons with PDD to infer other peoples thoughts and feelings.
Nevertheless, it must be emphasised that our naturalistic design still differs from
the real social world in several ways. First of all, our design permits all participants
to use as much time as needed to infer the thoughts and feelings of the targets. As
mentioned above, our participants with Asperger syndrome needed more time than
the control persons to administer the empathic accuracy task. However, in real life,
people have to infer the thoughts and feelings of other persons instantly. In a way,
this implies that our design is much easier than normal life. Secondly, in our
design, participants serve only as perceivers, not as targets. This differs from a
daily dyadic conversation where participants not only have to infer the thoughts and
feelings of their interaction partner, but their interaction partner also has to infer
their thoughts and feelings. By doing so, participants are perceiver and target at the
same time. At the moment, we are conducting a study in which adults with a
pervasive developmental disorder are videotaped using a concealed camera, while
having an initial conversation with a typically developing stranger. The procedure
of the empathic accuracy design (Ickes, Bissonnette, et al., 1990) enables us to
investigate how each of them is able to infer the thoughts and feelings of their
interaction partner. We assume that making an online assessment of the
unexpressed thoughts and feelings of another person during an unstructured
interaction will be more difficult for adults with PDD than for control adults.
37
REFERENCES
Attwood, T. (1998). Aspergers syndrome: A guide for parents and professionals. London:
Jessica Kingsley Publishers.
Bishop, D. (2000). Whats so special about Asperger syndrome ? The need for further
exploration of the borderlands of autism. In A. Klin, F. Volkmar & S. Sparrow (Eds.),
Asperger syndrome. (pp. 254-277). New York: Guilford Press.
Bowler, D. (1992). Theory of mind in Aspergers syndrome. Journal of Child Psychology
and Psychiatry, 33, 877-893.
Bowler, D., Gardiner, J., & Grice S. (2000). Episodic memory and remembering in adults
with Asperger syndrom. Journal of Autism and Developmental Disorders, 30, 295-304.
VIDANN: A video annotation system. Behavior Research Methods, Instruments, &
Computers, 33, 159-166.
In W. Ickes (Ed.), Empathic accuracy. (pp.73-116). New York: Guilford Press.
Eisenberg, N., Shea, C. L., Carlo, G., & Knight, G. (1991). Empathy-related responding
and cognition: a chicken and the egg dilemma. In W. Kurtines & J. Gewirtz (Eds.),
Handbook of moral behavior and development: Vol. 2. Research (pp. 63-68). Hillsdale
NJ: Lawrence Erlbaum.
Eisenmajer, R., Prior, M., Leekam, S., Wing, L., Gould, J., Welham, M., & Ong, B.
(1996). Comparison of clinical symptoms in autism and Aspergers Disorder. Journal
of the American Academy of Child and Adolescent Psychiatry, 35, 1523-1531.
Frith, U. (1989). Autism: Explaining the enigma. Oxford: Basil Blackwell.
Chapter 2
38
Gilchrist, A., Green, J., Cox, A., Burton, D., Rutter, M., & Le Couteur, A. (2001).
Development and current functioning in adolescents with Asperger syndrome: A
comparative study. Journal of Child Psychology and Psychiatry, 42, 227-240.
Hobson, R. (1986). The autistic childs appraisal of expressions of emotion. Journal of
Hobson, R. (1991). Methodological issues for experiments on autistic individuals
perception and understanding of emotion. Journal of Child Psychology and Psychiatry,
32, 1135-1158.
Hobson, R. (1999). Beyond cognition: A theory of autism. In P. Lloyd & C. Fernyhough
(Eds.), Lev Vygotsky: Critical assessments: Future directions, Vol. IV. Florence: Taylor
& Francis/Routledge.
Hobson, R., & Lee, A. (1989). Emotion-related and abstract concepts in autistic people:
Evidence from the British Picture Vocabulary Scale. Journal of Autism and
Howlin, P. (1998). Practitioner review: Psychological and educational treatments for
autism. Journal of Child Psychology and Psychiatry, 39, 307-322.
dyadic interaction paradigm. In C. Hendrick, & M.S. Clark (Eds.), Review of
personality and social
psychology: Research methods in personality and social
psychology, Vol. 11, pp. 16-44. Newbury Park, CA: Sage.

59, 730-742.
Jolliffe, T., & Baron-Cohen, S. (1999). The strange stories test: A replication with highfunctioning adults with autism or Asperger syndrome. Journal of Autism and
29-36.
39
Koning, C., & Magill-Evans, J. (2001). Social and language skills in adolescent boys with
Asperger syndrome. Autism, 5, 23-36.
Leekam, S., Libby, S., Wing, L., Gould, J., & Gillberg, C. (2000). Comparison of ICD-10
and Gillbergs criteria for Asperger syndrome. Autism, 4, 11-28.
Mayes, S., & Calhoun, S. (2001). Non-significance of early speech delay in children with
autism and normal intelligence and implications for DSM-IV Aspergers disorder.
Autism, 5, 81-94.
Miller, J., & Ozonoff, S. (2000). The external validity of Asperger disorder: Lack of
evidence from the domain of neuropsychology. Journal of Abnormal Psychology, 109,
227-238.
Ozonoff, S., Pennington, B. F., & Rogers, S. J. (1991). Executive function deficits in highfunctioning autistic individuals: Relationship to theory of mind. Journal of Child
Ozonoff, S., South, M., & Miller, J. (2000). DSM-IV-defined Asperger syndrome:
Cognitive, behavioral and early history differentiation from high-functioning autism.
Autism, 4, 29-46.
Rinehart, N., Bradshaw, J., Moss, S., Brereton, A., & Tonge, B. (2001). A deficit in
shifting attention present in high-functioning autism but not Aspergers disorder.
Autism, 5, 67-80.
Rutter, M., & Bailey, A. (1993). Thinking and relationships: Mind and brain: some
reflections on theory of mind and autism. In S. Baron-Cohen, H. Tager-Flusberg, & D.
Cohen (Eds.), Understanding other minds: Perspectives from autism (pp. 481-504).
Oxford: University Press.
Szatmari, P., Archer, L., Fisman, S., Streiner, D., & Wilson, F. (1995). Aspergers
syndrome and autism: Differences in behavior, cognition, and adaptive functioning.
Journal of the American Academy of Child and Adolescent Psychiatry, 34, 1662-1671.
Chapter 2
40
Weeks, S. J., & Hobson, R.P. (1987). The salience of facial expression for autistic
children, Journal of Child Psychology and Psychiatry, 28, 137-152.
Wing, L. (1988). The continuum of autistic characteristics. In E. Schopler & G. Mesibov
(Eds.), Diagnosis and assessment in autism (pp. 91-110). New York: Plenum Press.
Wing, L. (1998). The history of Asperger syndrome. In E. Schopler, G. Mesibov & L.
Kunce. (Eds.), Asperger syndrome or high-functioning autism ? (pp. 11-28). New
York: Plenum Press.
World Health Organization (1992). The ICD-10 classification of mental and behavioural
disorders: Clinical descriptions and diagnostic guidelines. Geneva: World Health
Organization.
Zager, D. (1999). Autism: Identification, education, and treatment. Second Edition. New
York: Lawrence Erlbaum Associates.
41
Appendix A
Empathic Accuracy Task: Measuring Inferred Thoughts and Feelings
The mean of the accuracy scores rated by the five judges was calculated for each
individual inference. We then calculated aggregated accuracy scores from the
averaged scores for each inference. The mean ratings were therefore summed
across the thoughts and feelings for each of the two videotapes. These summed
scores were divided by the total number of thoughts and feelings for each target and
multiplied by 100. By doing so, the minimum and maximum empathic accuracy
score of each participant could vary between 0 and 100.
As in previous studies (Ickes, Bissonnette et al., 1990; Ickes, Stinson et al.,
1990), the baseline empathic accuracy was derived by randomly pairing the actual
and inferred thoughts and feelings and rating the content of these randomly paired
actual/inferred thoughts and feelings on similarity. The five judges internal
consistency (Cronbachs alpha) for the baseline empathic accuracy scores was .78
for Video 1 and .71 for Video 2. Similarly to the procedure described above, the
baseline accuracy scores rated by the five judges were averaged for each individual
thought/feeling, divided by the total number of thoughts and feelings and multiplied
by 100. By subtracting the baseline empathic accuracy score for each participant
from the original score for that participant, we derived a revised measure of global
empathic accuracy.
Following this logic, the twelve empathic accuracy scores for each participant
for each video were computed for all the thoughts and feelings whether or not
belonging to the 6 different categorical classifications of the actual thoughts and
feelings. In the further of the study, the empathic accuracy scores are presented in
percentages. For each participant and each topic, the accuracy score is divided by
the number of corresponding thoughts and feelings and then multiplied by 100.
Chapter 2
42
Appendix B
Internal consistency of the topics per video
Video 1
Video 2
Self
Presence
.79
.82
Absence
.72
.85
Presence
.71
.80
Absence
.79
.83
Presence
.82
.76
Absence
.74
.82
Presence
.77
.82
Absence
.84
.79
Presence
.78
.87
Absence
.26
.77
Presence
.79
Absence
.78
.82
Interaction partner
Other person(s)
Research context
Tangible environmental object

or event
Past memory
CHAPTER
Empathic accuracy in adults with a

pervasive developmental disorder during
an unstructured conversation with a
1
typically developing stranger
This chapter is based on Ponnet, K., Buysse, A., Roeyers, H., & De Corte, K. (2004). Empathic accuracy in
adults with a pervasive developmental disorder during an unstructured conversation with a typically
developing stranger. Manuscript under editorial review.
Chapter 3
44
ABSTRACT
The present paper consists of two parts. In the first part, eleven high-functioning adults
with PDD participated and were videotaped with a concealed camera while having an
initial conversation with a typically developing stranger. Analyses of the overt behaviour
revealed that the dyad members behaviours significantly differed. Contrary to our main
hypothesis, analyses of the covert behaviour revealed that the subjects with PDD did not
differ from the typically developing subjects in the ability to infer the thoughts and
feelings of their interaction partner.
The second part indicated that the inference ability of both groups was independent of
the dyad members behavioural characteristics and the content of the dyad members
thoughts and feelings. Issues addressed in this paper include the relation of scriptal
knowledge to social functioning, and the advantage subjects with PDD take from more
structured interactions compared with less structured interactions.
Empathic accuracy in adults with PDD during a dyadic conversation
45
INTRODUCTION
The past few years there has been great interest in developing tasks which are
able to measure the social functioning of normally intelligent subjects with autism
in contexts that approximate closely the real social world (Baron-Cohen, Jolliffe,
Mortimore, & Robertson, 1997; Kaland, Moller-Nielsen, Callesen, Mortensen,
Gottlieb, & Smith, 2002; Kleinman, Marciano, & Ault, 2001). A validated design
that encounters these demands is the empathic accuracy design of Ickes and
colleagues (Ickes, 1997; Ickes, Bissonnette, Garcia, & Stinson, 1990; Marangoni,
Garcia, Ickes, & Teng, 1995). Empathic accuracy is the degree to which someone is
able to accurately infer the specific content of another persons thoughts and
feelings and, in addition, is the product of a specific conversation between two or
more interacting persons (Ickes, Bissonnette, et al., 1990). As described by Ickes
(1997), the most unequivocal way to measure empathic accuracy is by rating the
similarity between the content of the targets actual thought or feeling and the
content of the perceivers inference. Over the years, the empathic accuracy design
has proved to be a reliable and valid method to measure peoples inference abilities
(Ickes, Stinson, Bissonnette, & Garcia, 1990; Simpson, Ickes, & Blackstone, 1995).
In two recent studies (Ponnet, Roeyers, Buysse, De Clercq, & Van der Heyden,
2004; Roeyers, Buysse, Ponnet, & Pichal, 2001), we tested adults with a pervasive
developmental disorder (PDD) and typically developing adults using a naturalistic
Empathic Accuracy Task. In this task, each participant attempted to infer aspects of
a target persons actual subjective experience, while viewing a videotape of the
target person in a naturally occurring conversation with a stranger. The procedure
used in the studies was the standard stimulus paradigm, in which individual
participants each viewed the same set of two videotaped interactions between
strangers. In both studies, the Empathic Accuracy Task was able to distinguish
clearly between the group with PDD and the control group, although a significant
difference between both groups was only found on one of the two empathic
accuracy videotapes.
The two studies indicate that the above-mentioned standard stimulus empathic
accuracy design is a promising and valuable method to study the mind-reading
abilities of adults with PDD. We consider the empathic accuracy design to be more
Chapter 3
46
naturalistic than static mind-reading tasks since it allows participants to make

inferences about other peoples thoughts and feelings on the basis of verbal and
non-verbal cues, whereas in the static mind-reading tasks participants have to rely
on either verbal or non-verbal cues. In contrast with common paper and pencil
mind-reading tasks, the empathic accuracy task forces people to infer on-line the
thoughts and feelings of interacting persons, which reduces the possibility that
participants successfully pass the task without a genuine social-cognitive
understanding of the situation.
However, it should be noticed that the real social world is still more complex
than the standard stimulus design of the Empathic Accuracy Task. In the standard
stimulus design, participants serve only as perceivers, not as targets, which is
different from any social dyadic interaction where participants have to infer the
thoughts and feelings of their interaction partner and vice versa. This lack of
bidirectional influences induces the authors to use an alternative empathic accuracy
design, the dyadic interaction design, in which each participant becomes an active
and interacting member, instead of being a passive observer. The empathic
accuracy, also defined as the degree to which the participants inference matches
the targets subjective experience, is therefore, in a way, best characterised as the
emergent product of social interaction processes that occur at the level of the dyad
(Ickes, 1997; Ickes, Bissonnette, et al., 1990).
Since the dyadic interaction design requires various social skills, which are at the
core of the PDD diagnosis, the social situation of the dyadic interaction design can
be considered as more complex than that of the standard stimulus design. In order
to assure that the PDD participants possess a minimum of the acquired social skills,
we therefore worked with a sample of high-functioning subjects with PDD of the
Roeyers et al.s study (2001). Although there were significant between-group
differences found in the Empathic Accuracy Task of Roeyers et al. (2001), some
subjects with PDD performed as good as subjects without PDD. This upper-class
of PDD subjects (with empathic accuracy scores above the mean) was invited to
participate in the present study.
The present investigation consists of two parts. In the first part, eleven adults
with PDD participated in a laboratory study in which they were videotaped with a
concealed camera while having an initial conversation with a typically developing
47
stranger. The hypotheses and research questions of the first part are based on
previous studies with the empathic accuracy design as well as other studies with
adults with PDD. A prior study of Stinson and Ickes (1992) with typically
developing adults revealed that the empathic accuracy of male strangers was
strongly dependent on the level of interactional involvement (in casu verbalisations,
gazes, positive affect, gestures and interpersonal distance) in their immediate
interaction. Stinson and Ickes (1992, p. 795) reason that this is not surprising
because the strangers immediate interaction provided them with their only source
of information about each other. In the present study, we assume differences
between the adults with PDD and the control adults with regard to their behavioural
characteristics, and we expect reciprocity between the behaviours of the dyad
members. We further assume that the association between the behavioural
characteristics and the empathic accuracy scores will be different for both groups.
We strengthen the latter expectation with a recent study of Koning and MaggilEvans (2001) who found that subjects with PDD made use of facial cues as
frequently as typically developing controls, but subjects with PDD used facial cues
more often than other cues (such as tone of voice) for inferring emotions. Koning
and Maggil-Evans (2001, p. 32) further noticed that the difficulties of the subjects
with PDD to infer the affective state of others may become only apparent when
dealing with the simultaneous presentation of facial, vocal, body and situational
cues. Based on our previous studies with the empathic accuracy design (Ponnet et
al., 2004; Roeyers et al., 2001), we expect within-dyad differences in empathic
accuracy. In addition, we expect different patterns between the dyad members with
regard to the focus/foci or theme(s) of the original thought/feeling entries and
with regard to the empathic accuracy for thought/feeling entries circling around a
specific focus or theme.
The procedure used was the dyadic interaction design (Ickes, Bissonnette, et al.,
1990) through which we were able to study both the overt behaviour of the
interacting dyad members, and the covered thoughts and feelings of the
participants. First, the dyadic interaction paradigm permits the study of the overt
behavioural characteristics of the dyad members during their conversation. It is
common knowledge that people with PDD often exhibit socially and emotionally
inappropriate behaviours, but the way in which they specifically differ from
Chapter 3
48
typically developing control subjects is not always clear and probably varies from
person to person (see Zager, 1999). The uniqueness of the present observational
study lies in the combination of our naturalistic design and the procedure that is
used to code the behaviours: (a) both the person with PDD as well as the person
without PDD were unaware that they were videotaped, so that the naturalistic
character of the recorded interaction was guaranteed, and (b) instead of using a
rating procedure, the behaviours of each adult were thoroughly (second-to-second
or even slower) observed and coded, using a validated computer program.
Second, in the dyadic interaction paradigm the participants have to make a
written record of all their unexpressed thoughts and feelings on a standardised
coding form, by which we were able to explore the nature of the dyad members
original thought/feeling entries. This permitted us to attribute the different patterns
in the dyad members empathic accuracy. We assessed how difficult it was to infer
the specific content of each thought/feeling entry for each dyad member and how
concrete or abstract the content of each thought/feeling entry was. Furthermore, we
examined the thematic topic of each dyad members actual thought/feeling entries.
Thereafter, we calculated the global empathic accuracy for each dyad member.
According to the content of each thought/feeling entry, the empathic accuracy
scores of each participant could be computed for all the thought/feeling entries
classified as highly easy/concrete and for all the thought/feeling entries classified as
highly difficult/abstract. Furthermore, we worked out the empathic accuracy scores
of each participant for all the thought/feeling entries belonging to the different
topics of the original thoughts and feelings.
PART 1
METHOD
Participants
Two groups of normally intelligent adults participated in this study, i.e. 11
subjects with PDD and a control group of 11 typically developing subjects.
49
Male/female ratio (9/2) was similar in both groups. The subjects of the clinical
group were drawn from a sample of individuals who participated in a research
project two years ago at our laboratory (Roeyers et al., 2001). More specifically,
fifteen subjects with PDD with the highest performance on the Empathic Accuracy
Task of Roeyers et al. (2001) were selected and invited for a follow-up task. Four
of them refused to participate. The eleven remaining participants had all been
diagnosed by a multidisciplinary team of experienced clinicians and fulfilled
established DSM-IV criteria for autism, Asperger syndrome or PDD-NOS (APA,
1994). The comparison group was recruited by personal contact.
The adults with PDD were matched on a one-to-one basis with the control adults
on sex, chronological age, level of education and main interests (e.g. computer and
internet, music, cars). This results in 11 dyads. Given the bidirectional nature of the
communication and the empathic accuracy design (see above), the dyad is used as
unit of analyses and within-dyad analyses will be used.
All subjects with PDD had IQ scores in the normal range as indicated by their
records of previous IQ testing using the full Wechsler Scale of Intelligence. The
control subjects were given the shortened WAIS version, comprising Block Design,
Vocabulary, Arithmetic, and Picture Arrangement. Two subjects of the control
group refused the intelligence testing, but on the basis of their study or professional
level it can be taken for granted that they are of normal intelligence.
The IQ scores for the subjects with PDD were: Total IQ, M = 121.44, SD =
12.49, Verbal IQ, M = 117.67, SD = 10.57, and Performance IQ, M = 122.56, SD =
12.97. The estimated IQ scores for the control subjects were: Total IQ, M = 124.78,
SD = 12.26, Verbal IQ, M = 121.67, SD = 11.25, and Performance IQ, M = 121.78,
SD = 13.06. Paired t-tests revealed no significant within-dyad differences for Total
IQ, t(8) = .46, for Verbal IQ, t(8) = .71., and for Performance IQ, t(8) = -.11.
Furthermore, a paired t-test revealed no significant differences between the mean
chronological ages of the subjects with PDD (M = 24.60, SD = 8.11) and the mean
chronological ages of the control subjects (M = 24.53, SD = 7.89), t(10) = -.10.
Chapter 3
50
Setting and equipment

The observation room where the dyad members interactions were recorded was
furnished with three armchairs, a coffee table and two small tables placed before a
one-way screen. The two tables were decorated with some plants, in support of
covering the one-way screen. Behind the one-way screen was a control room used
to house a camera and audio equipment. The video camera focused on the area of
the coffee table and the two armchairs of the interaction partners and was
connected with two identical VCRs, by which the interaction was directly recorded
on two analogue videotapes. Nearby the observation room, two offices were
equipped with an identical 30 inch colour TV monitor and a VCR.
Procedure
The procedures used were based on those developed by Ickes and colleagues
(Ickes, Bissonnette, et al., 1990; Simpson et al., 1995; Stinson & Ickes, 1992). The
dyad members, previously unacquainted to each other, were scheduled to come to
the research center at the same time. They were asked to participate in a study on
"people's perceptions about others during different situations. They were told in
advance that they would meet another person, but they were kept unaware that they
were to be videotaped. Ethical permission was sought and granted for this study.
Furthermore, the typically developing dyad member was not informed that his/her
interaction partner had a PDD diagnosis. Even so, the dyad member with PDD was
kept unaware whether or not his/her interaction partner had any kind of diagnosis.
Phase 1: Collection of the videotape data
Once they arrived at the research center, both participants were brought to
different waiting rooms so that they would not meet and interact before the
experiment began. When they were brought together, the research assistant told the
participants they had to fill in some inquiry forms. However, when the research
assistant reached for the inquiry forms, he discovered that some of the copies
were not well printed. In order to get some proper copies, the experimenter left the
room. Promising to return in a few minutes, he suggested both participants to seek
acquaintance with each other during his absence. At this point, a second research
51
assistant activated the concealed camera. After approximately 8 minutes, the first
research assistant returned and partly debriefed the participants. The subjects were
told they had been unobtrusively videotaped in order to study the spontaneous
occurring interaction that takes place between two strangers. The subjects were
then asked whether their tape could be used as data. It was made clear that if either
of them did not want the videotape used for any reason, they could erase it
themselves immediately. None of the subjects refused the tape to be released and
each of them signed a consent form. Thereafter, both subjects were asked to
participate in a second phase of the study.
In approximately two third of the cases, the conversation was started by the
control adult. Examples of utterances given by the adults with PDD were: Where
do you come from ?, What is your name ?, What are you studying ?, Its the
second time that Im participating in a research study. Examples of utterances
given by the control adults were: How do you come to be here ?, Where do you
come from ?, Do you have a job?, Phew, its very hot outside.
Phase 2: Collection of the thought/feeling data
In the second phase, the subjects were brought to the above-mentioned separate
offices. Both subjects were asked to view their videotape and to make a written
record of all the unexpressed thoughts and feelings they remembered having had
during the interaction period that they were left alone. Following the procedures of
previous studies (e.g. Ickes, Bissonnette, et al., 1990, Buysse & Ickes, 1999), the
research assistant instructed them to stop the videotape at each point during the
interaction when they remembered having had a specific thought or feeling. At
each of those stops, the subjects were asked to write down on a standardised
thought/feeling coding form (a) the time when the thought or feeling occurred (as
displayed by a digital clock on the under-right corner of the videotape), (b) whether
the entry was a thought or a feeling, and (c) the specific content of the
thought/feeling entry.
Furthermore, the research assistant explicitly encouraged the subjects to report
all the thoughts and feelings they remembered having had as accurately and
honestly as possible, and assured the subjects that their unexpressed thoughts and
feelings would never be shown to their interaction partner. In addition, the research
Chapter 3
52
assistant asked the subjects to report only those thoughts and feelings they
distinctly remembered having had during the interaction period and not to report
any new thoughts and feelings that occurred to them while they were viewing the
videotape. No restriction of time was given. Considerable evidence for the
construct validity of this method of thought/feeling assessment has been established
(see Ickes, Robertson, Tooke, & Teng, 1986).
Sample thought/feeling entries reported by the adults with PDD were: What do
I have to say ?, I feel a bit uncomfortable, It is quiet here, This situation
looks suspicious to me, Where does he come from ?. Sample entries reported by
the control adults were: Its taken too long here for me, What can I say ?,
Perhaps she knows something about the coming experiment, His accent is
funny, Where does he come from ?, Ouch, we both fell silent, I hope the
experimenter will arrive soon, Strange shoes hes wearing.
Phase 3: Collection of the empathic accuracy data
After the collection of all the unexpressed thought/feeling entries, the research
assistant requested the subjects to view their videotape a second time. This time,
the research assistant paused the videotape at precisely those moments during
which their interaction partner had reported having had a specific thought or
feeling. The subjects task was to make inferences about the content of their
partners thought/feeling entries. More specifically, the subjects were asked to
report (a) whether the interaction partners entry was a thought or a feeling, and (b)
the specific content of the thought/feeling entry. No restriction of time was given.
When this task was fulfilled, each subject was debriefed more completely and was
given a small monetary reward for his or her participation.
Measures
Behavioural Measures
The verbalisations, gazing and stimulatory gestures of the subjects with PDD
and those of the control subjects were recorded from videotape by using The
Observer Video-Pro Analysis System (Noldus, 1991; Noldus, Trienes,
53
Hendriksen, Jansen, & Jansen, 2000). This system reads time codes directly from
videotape, which allows accurate event timing at slower playback speed.
Verbalisations are defined as all meaningful words, utterances, questions,
remarks and requests. Vocalisations such as sounds which do not form words,
sneezing or coughing were excluded. Gazing is defined as looking at the face of
the other person. Stimulatory gestures were all hand movements that had to do
with touching some object, body part of the self or body part of the other person.
The codings were performed by one rater. To prevent bias, each of the two
videotaped interactants was rated independently for each behaviour (verbalisation,
gazing and stimulatory gestures), so that each videotape was viewed six times (i.e.
2 interactants x 3 behaviours) by the rater. The behavioural coding was very time
consuming, because the rater scored each behaviour at a slower playback speed
(1/2 normal) and made multiple passes through the videotape if necessary. A
second rater coded a randomly chosen videotape, in order to validate the judgement
of the first coder. Consistent with the first rater and using the same procedure, this
videotape was viewed six times by the second rater. The degree of inter-rater
concordance (based on frequency and sequence) was calculated for each behaviour,
by tallying the frequency of agreements and disagreements between the
observations of both coders. The tolerance window, which defines how accurate the
timing of a record must be to be considered a match or not, was set on 2 seconds.
This means that the program links the events in the two observations and searches 2
seconds around a time code. These links may result in an error (disagreement) or a
match. All kappa values were satisfactory (between .76 and .84).
For each behavioural characteristic of each participant, we calculated (a) the
total duration, (b) the frequency, and (c) the mean duration of each specific
behaviour. The mean duration of each specific behaviour was derived by dividing
for each perceiver the duration of the specific behaviour by the frequency of that
behaviour. In the present paper, the mean duration of each behavioural
characteristic will be presented in seconds, while the total duration of each
behaviour will be presented in percentages, by dividing for each participant the
total duration of the specific behaviour by the total duration of each videotape.
Chapter 3
54
Actual thought/feeling measures

As suggested by Ickes and colleagues (Ickes, Bissonnette, et al., 1990), the
subjects thought/feeling entries were coded by five independent judges (i.e.
undergraduate college students). The judges watched all videotapes that were
stopped each time one of the interacting persons had reported having had a thought
or a feeling. The judges rated for each thought/feeling entry how difficult they
thought it was to infer the specific content of each thought/feeling entry on a 7
point scale ranging from 1 (very easy) to 7 (very difficult). Because of the
satisfactory interrater reliability (Cronbachs alpha = .75) of the five judges, we
averaged the judges scores for each though/feeling entry.
Similarly, the same five judges rated for each thought/feeling entry how concrete
or abstract the specific content of each thought/feeling entry was on a 7 point scale
ranging from 1 (very concrete) to 7 (very abstract). Because the interrater reliability
(Cronbachs alpha) of the five judges was .72, the judges scores were averaged for
each thought/feeling entry.
Finally, the same five judges assessed for each thought/feeling entry whether the
focus of the thought/feeling entry was (a) on the self, (b) the interaction partner, (c)
other person(s), (d) the research context, or (e) a tangible environmental object or
event. The interrater reliabilities (Cronbachs alpha) for these mutually nonexclusive categories varied between .86 and .95.
Empathic accuracy measures
Following the logic and procedure described by Ickes and colleagues (Ickes,
Bissonnette, et al., 1990; Marangoni et al., 1995), the empathic accuracy was
computed by comparing the written content of each actual thought/feeling entry
with that of the corresponding inferred thought/feeling entry.
The same five judges were instructed to compare each inferred thought/feeling
entry with the corresponding thought/feeling entry and to rate the level of similarity
on a 3-point scale ranging from 0 (essentially different content) through 1
(somewhat similar but not the same content) to 2 (essentially the same content).
The internal consistency (Cronbachs alpha) of the five judges content accuracy
ratings was .90 for the PDD group and .92 for the control group. Because the high
reliability of the judges ratings, the mean of the empathic accuracy scores rated by
55
the five judges was calculated for each particular inference. In order to derive an
overall empathic accuracy score for each perceiver, the mean empathic accuracy
scores were summed across all thought/feeling inferences and then divided by the
total number of accuracy points that could be obtained for a given number of
inferences, and multiplied by 100. As in previous studies (Buysse & Ickes, 1999;
Ickes, Bissonnette, et al., 1990; Ickes, Stinson, et al., 1990), the baseline empathic
accuracy for each of the subjects was estimated by randomly pairing each set of the
actual thought/feeling entries with the corresponding set of the partners inferences
and rating the content of these randomly paired entries on similarity. The
Cronbachs alpha of the baseline accuracy provided by the five judges was .94 for
the PDD group and .90 for the control group. The mean of the baseline accuracy
scores rated by the five judges was further calculated for each inference. Then,
these baseline accuracy scores were summed across all thought/feeling inferences
and then divided by the total number of accuracy points that could be obtained for a
given number of inferences and multiplied by 100. Finally, we obtained an adjusted
measure of empathic accuracy for each perceiver, by subtracting the baseline
accuracy scores for each subject from the empathic accuracy scores for that subject
(see also Ickes, Bissonnette, et al., 1990).
Following this logic, the empathic accuracy scores for each perceiver were
calculated for all thought/feeling entries belonging to (a) the most easy-to-infer
thought/feeling entries, (b) the most-difficult-to-infer thought/feeling entries, and
(c) the remaining moderate difficult-to-infer thought/feeling entries. Similarly, the
empathic accuracy scores for each perceiver were calculated for all thought/feeling
entries belonging to (a) the most abstract thought/feeling entries, (b) the most
concrete thought/feeling entries, and (c) the moderate abstract thoughts and
feelings. Finally, we computed for each perceiver the empathic accuracy scores for
all the thought/feeling entries whether or not belonging to the 5 categorical
classifications of the actual thoughts and feelings, more specifically (a) the self, (b)
the interaction partner, (c) other person(s), (d) the research context, and (e) a
tangible environmental object or event.
Chapter 3
56
RESULTS
Within-dyad differences with regard to the behavioural characteristics
Since the empathic accuracy design allows us to study the overt behaviour of the
participants, we tested our research question whether or not the verbalisation,
gazing and stimulatory gestures differed between the dyad members. As shown in
Table 1, analysis revealed that the within-dyad difference in total duration of
verbalisation approached significance, t(10) = -2.05, p = .07. There were no
significant within-dyad differences in the total duration of gazing and the total
duration of stimulatory gestures.
Table 1
The total duration (in percentages), the frequency, and the mean duration (in seconds) of the
verbalisation, gazing and stimulatory gestures
PDD
Control
SD
SD
t(10)
Total duration (%)
18.17
20.64
8.33
7.62
-2.05
Frequency
30.18
26.78
28.73
24.67
-.35
1.92
1.50
1.14
.69
Total duration (%)
17.00
12.49
25.75
23.99
1.84
Frequency
31.73
24.04
25.18
18.63
-1.30
2.73
1.68
5.02
3.57
Total duration (%)
36.99
29.28
24.90
21.24
-1.53
Frequency
14.09
8.90
15.45
9.85
.32
Mean duration
32.11
72.82
9.76
9.71
-1.08
Verbalisation
Mean duration
-2.38 *
Gazing
Mean duration
2.36 *
Stimulatory gestures
*p < .05
Furthermore, a series of paired t-tests revealed that there were no significant

within-dyad differences in the frequency of verbalisation, in the frequency of
gazing, and in the frequency of stimulatory gestures. Significant within-dyad
57
differences were found with regard to the mean duration of verbalisation, indicating
that subjects with PDD verbalise longer than control subjects at the moment they
get up to speak, and with regard to the mean duration of gazing, indicating that the
period of each look at the interaction partner is shorter for subjects with PDD than
for control subjects. No significant within-dyad difference was found with regard to
the mean duration of stimulatory gestures.
In order to know more about the level of interactional involvement, two sets of
correlations are of interest. First, we are interested in the within-dyad correlations
of the behavioural characteristics. As shown in Table 2, the verbalisation of the
PDD subjects significantly relates to the verbalisation of the control subjects, which
implies that more verbalisation by one interaction partner is reciprocated by more
verbalisation by the other interaction partner. The same pattern is true for gazing,
but not for stimulatory gestures. The fact that several other significant within-dyad
associations were found (see Table 2) suggests that reciprocity is a main
characteristic of the behavioural interactions.
Table 2
Intercorrelations between the total durations of the behavioural characteristics
Verbalisation
PDD
Gazing
Control
PDD
Control
PDD
Control
.73 *
.77 **
.83 **
-.47
-.25
.74 **
.62 *
-.41
-.39
.80 **
-.72*
-.19
-.62*
-.18
Verbalisation
PDD
Control
Gazing
PDD
Control
PDD
.50
Control
*p < .05, **p < .01
Second, we are interested in the within-group correlations of the different

behaviours. Two significant positive associations were found between verbalisation
and gazing, indicating that the more a dyad member verbalises, the more this
Chapter 3
58
person is looking at the interaction partner. Furthermore, the significant negative

association between the stimulatory gestures and the gazing of the subjects with
PDD indicates that that the more a subjects with PDD is touching some object or
body part, the lesser he is looking at the interaction partner, and vice versa. By
using the formula of Hays2 (1994), transformation of the correlations into z-scores
revealed that the associations between stimulatory gestures and gazing were equally
strong in both groups.
Within-dyad differences with regard to the actual thought/feeling entries

A paired t-test revealed no significant within-dyad difference between the
number of thoughts and feelings reported by the PDD subjects (M = 16.55, SD =
11.99) and the control subjects (M = 20.45, SD = 10.43), t(10) = -.86.
Within-dyad analysis revealed that the mean difficulty of the actual
thought/feeling entries (aggregated over the 5 judges and all entries) did not differ
significantly, t(10) = .77, with M = 4.92 (SD = 1.21) for the PDD subjects and M =
4.93 (SD = 1.14) for the control subjects. Then, based on the quartile distribution of
both groups thought/feeling entries and independent of the empathic accuracy
variable, the thought/feeling entries were divided into (a) the 25.2% most easy-toinfer
thought/feeling
entries,
(b)
the
51.5%
moderate
difficult-to-infer
thought/feeling entries, and (c) the 23.3% most-difficult-to-infer thought/feeling

entries. Analysis revealed that there were no significant within-dyad differences
(2(2) = 1.16) with regard to the percentages of thought/feeling entries classified as
easy-to-infer thought/feeling entries (24.3% for the PDD subjects and 25.8% for the
control subjects), moderately difficult-to-infer thought/feeling entries (50.0% for
the PDD subjects and 52.5% for the control subjects), and difficult-to-infer
thought/feeling entries (25.7% for the PDD subjects and 21.7% for the control
subjects).
Within-dyad analysis revealed that the mean abstractness of the actual
thought/feeling entries (aggregated over the 5 judges and all entries) was not
significantly different, t(10) = 1.36, with M = 3.98 (SD = 1.13) for the PDD
z1 z 2
2
1
1
+
n1 3 n 2 3
59
subjects and M = 4.34 (SD = 1.08) for the control subjects. Following the abovementioned procedure, the thought/feeling entries were divided into (a) the 27.9%
most abstract thought/feeling entries, (b) the 45.2% moderate abstract
thought/feeling entries, and (c) the 26.7% most concrete thought/feeling entries.
Significant within-dyad differences were found (2(2) = 12.23, p < .01) with regard
to the percentages of thought/feeling entries classified as abstract thought/feeling
entries (35.4% for the PDD subjects and 22.7% for the control subjects),
moderately abstract thought/feeling entries (43.4% for the PDD subjects and 46.7%
for the control subjects), and concrete thoughts and feelings (21.2% for the PDD
subjects and 30.6% for the control subjects), indicating that the subjects with PDD
had more abstract thought/feeling entries than the control subjects and had less
concrete thoughts and feelings than the control subjects.
Table 3
The mean percentages of thought/feeling entries belonging to the PDD group and the control
group
PDD
Control
t(10)
SD
SD
Self
44.61
22.10
45.90
19.34
.12
Interaction partner
18.09
17.32
35.31
22.39
2.58 *
Other person(s)
12.12
14.62
11.96
15.82
-.02
Research context
16.05
17.24
16.30
9.96
.04
8.73
9.81
4.26
5.72
-1.79
Topic
Tangible environmental
object or event
*p < .05
Furthermore, within-dyad comparison of the topic of the thought/feeling entries

revealed no significant differences with regard to the thought/feeling entries that
focussed on the self, other person(s), the research context, and a tangible
environmental object or event (see Table 3). However, a significant within-dyad
difference was found with regard to the thoughts and feelings that focussed on the
interaction partner, t(10) = 2.58, p < .05, indicating that the control subjects had
Chapter 3
60
more thoughts and feelings that focussed on the interaction partner than the subjects
with PDD.
Finally, there was no significant within-dyad difference in the total amount of
time required to write down their original thoughts or feelings, t(10) = .06. The
mean time was 36.31 minutes (SD = 17.57) for the subjects with PDD and 35.96
minutes (SD = 19.84) for the control subjects. By dividing for each participant the
time required to write down all original thought/feeling entries by the number of
thought/feeling entries, we derived for each dyad member the average time required
to write down a thought or feeling. Analysis revealed that the within-dyad
difference in average time (in minutes) required to write down a thought or feeling
approached significance, t(10) = -2.11, p = .06, with M = 2.47 (SD = 0.78) for the
PDD subjects and M = 1.84 (SD = 0.63) for the control subjects.
Within-dyad differences with regard to empathic accuracy

We compared the empathic accuracy scores of the subjects with PDD with those
of the control subjects, to test for our primary hypothesis, that the control subjects
would display significantly more empathic accuracy than the subjects with PDD.
Contradictory to our hypothesis, there was no significant within-dyad difference
(see Table 4). The mean adjusted accuracy score was 25.70% for the PDD subjects
and 25.14% for the control subjects. Table 4 contains data that provide comparisons
of the original content accuracy scores, the baseline accuracy scores and the
adjusted measures of empathic accuracy.
Table 4
Components of Content Accuracy Measure (in percentages)
PDD
Control
SD
SD
t(10)
Original content accuracy
37.67
13.26
41.41
19.49
-.45
Baseline content accuracy
11.97
10.56
16.27
11.84
-1.38
Adjusted content accuracy
25.70
15.73
25.14
19.99
.06
61
The mean empathic accuracy scores of the thought/feeling entries belonging to

the most easy-to-infer thought/feeling entries, the most difficult-to-infer
thought/feeling entries and the moderately difficult-to-infer thought/feeling entries
are shown in Table 5, represented separately for each group. We conducted a 2
(Group: PDD versus Control) x 3 (Difficulty: Easy, Moderate and Difficult)
ANOVA on the empathic accuracy scores, with Group and Difficulty as withinsubject factors. The analysis revealed no significant main effect for Group, F(1, 10)
< 1. However, a significant main effect for Difficulty was found, indicating that the
most difficult-to-infer thought/feeling entries are less accurately inferred than the
most easy-to-infer thought/feeling entries, F(2, 9) = 6.27, p < .01. The interaction
between Group and Difficulty yielded no significance, F(2, 9) = 2.07, indicating
that both groups have the same tendency to be less accurate in inferring more
difficult thought/feeling entries.
Table 5
Empathic accuracy scores along the nature of the thought/feeling entries (in percentages)
PDD
Control
SD
SD
Most easy to infer
48.94
28.16
41.39
22.55
Moderate difficult to infer
27.75
16.11
34.99
25.60
Most difficult to infer
15.68
11.41
27.70
22.80
Most concrete to infer
31.82
23.69
36.01
23.65
Moderate abstract to infer
36.27
20.67
29.66
22.12
Most abstract to infer
33.83
27.75
36.86
24.37
Difficulty
Abstractness
Furthermore, Table 5 presents for each group the mean empathic accuracy scores
of the thought/feeling entries belonging to one of the three categories of
abstractness. We conducted a 2 (Group: PDD versus Control) x 3 (Abstractness:
Concrete, Moderate and Abstract) ANOVA on the empathic accuracy scores, with
Group and Abstractness as within-subject factors. The analysis revealed no
significant main effect for Group, F(1, 10) < 1. We found no significant main effect
Chapter 3
62
for Abstractness, F(2, 9) < 1, indicating that there were no differences in empathic
accuracy between the most concrete-to-infer, the moderate concrete-to-infer and the
most abstract-to-infer thought/feeling entries. The interaction between Group and
Abstractness was not significant, F(2, 9) < 1.
For each group, the mean empathic accuracy scores of thought/feeling entries
belonging to a specific topic are shown in Table 6. A series of 2 (Group: PDD
versus Control) x 2 (Topic: Presence versus Absence of a particular focus)
ANOVAs on the empathic accuracy scores were conducted, with Group and Topic
as within-subject factors. As shown in Table 6, the analyses revealed no significant
effect of group for each of the more detailed empathic accuracy scores.
Furthermore, the analyses revealed only one significant main effect of topic,
indicating that the thought/feeling entries that focussed on a tangible environmental
object or event are less accurately inferred than the thought/feeling entries without
this focus. No significant interaction effects (Topic x Group) were found.
We administered the total amount of time (in minutes) required to infer and
write down the other persons thoughts and feelings. No significant within-dyad
difference was found, t(10) = 1.53, with M = 42.43 (SD = 27.95) for the subjects
with PDD and M = 29.71 (SD = 19.29) for the control subjects. By dividing for
each participant the time required to write down all inferences by the number of
inferences, we derived for each dyad member the average time required to infer and
write down a thought or feeling. No significant within-dyad difference was found
in the average time required to infer the other persons thought or feeling, t(10) =
1.11. The mean time was 2.17 minutes (SD = .80) for the subjects with PDD and
1.88 minutes (SD = .55) for the control subjects.
2 (Topic: Presence versus Absence of a particular focus) x 2 (Group: PDD versus Control) ANOVAs on the empathic accuracy scores along the
specific topic of the thought/feeling entries (in percentages)
Effects
df
Group
Topic
Presence
Absence
Topic x
PDD
Control
PDD
Control
Group
Mean (SD)
Mean (SD)
Mean (SD)
Mean (SD)
Self
1,10
0.34
0.10
0.21
34.67 (18.24)
45.14 (47.68)
35.67 (11.20)
38.80 (29.46)
Interaction partner
1,10
0.00
2.46
0.01
49.81 (29.25)
51.23 (57.66)
34.72 (18.32)
35.10 (23.13)
Other person(s)
1,5
0.57
1.41
0.47
48.94 (34.56)
29.22 (34.92)
33.66 (10.38)
42.38 (26.57)
Research context
1,10
0.09
0.02
0.72
38.45 (33.43)
36.05 (31.34)
34.33 (12.08)
42.80 (29.94)
Tangible environmental
1,4
0.94
10.44 *
0.26
10.00 (20.00)
28.33 (39.76)
35.17 (12.75)
41.44 (25.01)
object or event
* p < .05
Table 6
63
Chapter 3
64
Correlates of the Empathic Accuracy Measures

We found a significant association between the mean empathic accuracy of the
control group and the total duration of the stimulatory gestures of the group with
PDD, indicating that the more one dyad member is touching some object or body
part, the better the interaction partner can infer the thoughts and feelings. No other
significant associations between empathic accuracy and the total duration of a
specific behaviour were found (see Table 7). However, a significant negative
association was found between the empathic accuracy of the PDD group and the
mean duration of the PDD groups gazing, indicating that the empathic accuracy of
the subjects with PDD decreases when the mean time of their gazes increases and
vice versa. We further used the formula of Hays to transform the Pearson
correlations into z-scores. As shown in Table 7, the transformation revealed that the
strength of the associations were equally strong in both groups.
Finally, we calculated for both groups the correlations between empathic
accuracy scores, chronological ages, IQ-scores and inference times. As shown in
Table 8, Pearson correlations revealed no significant associations between the
empathic accuracy scores and the time needed to infer another persons
thought/feeling entry. Furthermore, no significant associations were found between
empathic accuracy scores and IQ scores, or between empathic accuracy scores and
ages. Since we had the empathic accuracy scores of the 11 participants with PDD
on both videotapes of the Empathic Accuracy Task of Roeyers et al. (2001), we
were able to correlate these scores with the present empathic accuracy scores of the
PDD subjects. Pearson correlations revealed that the empathic accuracy scores of
the 11 participants with PDD correlated significantly with their performance on
videotape 1 of the Empathic Accuracy Task (Roeyers et al., 2001).
65
Table 7
Correlations between empathic accuracy and behavioural characteristics
Empathic Accuracy
PDD
Control
z1 z2
1
1
+
n1 3 n2 3
Verbalisation of PDD group

Total duration
.32
-.24
1.16
Frequency
.29
-.28
1.17
Mean duration
.21
-.28
1.02
Total duration
.21
-.10
.63
Frequency
.09
-.21
.62
Mean duration
.21
-.22
.88
Total duration
.34
-.50
1.80
Frequency
.45
-.40
1.80
-.61*
-.24
-.93
Total duration
.26
-.28
1.12
Frequency
.08
-.10
.37
Mean duration
.07
-.36
.90
Verbalisation of control group
Gazing of PDD group
Mean duration
Gazing of control group
Stimulatory gestures of PDD group

Total duration
-.25
.60 *
-1.91
Frequency
-.21
.39
-1.23
.03
.38
-.73
Total duration
-.28
.04
-.65
Frequency
-.03
-.52
1.11
Mean duration
-.26
.33
-1.23
Mean duration
Stimulatory gestures of control group
* p < .05
Chapter 3
66
Table 8
Correlations between standard empathic accuracy scores1, dyadic empathic accuracy scores, age,
IQ scores and (mean) time needed to infer another persons thought/feeling entry
Empathic
Age
IQ score
Inference time
Accuracy
PDD
Control
PDD
Control
PDD
Control
PDD
Control
Standard EA Task1
Video 1
.63 *
Video 2
.54
Empathic Accuracy
PDD
Control
-.44
.30
.27
-.05
-.40
.49
-.13
-.49
-.52
.05
.43
-.39
.40
.96 **
.30
-.39
.56
.27
.25
-.44
.49
.28
-.53
.02
.38
-.25
-.41
Age
PDD
Control
IQ-score
PDD
Control
Inference time
PDD
.19
Control
1
The scores on the Empathic Accuracy Task of Roeyers et al. (2001)
*p < .05, **p < .001
DISCUSSION
The dyadic empathic accuracy design allowed us to study both the overt and the
covert behaviour of the dyad members. As expected, within-dyad analyses of the
overt behaviour of the dyad members revealed that there were significant and
nearly significant differences between the total and mean duration of verbalisation
and gazing. While the within-dyad analyses demonstrated that the social interaction
between subjects with and without PDD is characterised by a strong level of
behavioural reciprocity, further within-group analyses revealed that the different
67
behaviours correlated with each other and that the associations between these
behaviours were equally strong in both groups. Contrary to expectations, the main
finding of the within-dyad analyses of the covert behaviour was that the subjects
with PDD did not differ from the control adults in the ability to infer the thoughts
and feelings of their interaction partner.
However, when we are interested in the inference abilities of adults with or
without PDD while interacting with each other, the peculiar behavioural
characteristics of adults with PDD should be taken into consideration (e.g. Howlin,
1997, 1998). Several people, who meet for the first time a normally intelligent adult
with PDD, often notice that this person is -in a way- somewhat strange, but because
of their unfamiliarity with the PDD diagnosis, they can not figure out what exactly
makes the other person so oddly (Wing, 1992; Attwood, 1998). Moreover, it has
proved surprisingly difficult to determine accurately what is abnormal about
autistic subjects social behaviour through systematic studies (Hobson & Lee,
1998). Sometimes, the oddity is caused by a single word or gesture that occurs at an
inappropriate moment. Otherwise, it is possible that the behaviour of the adult with
PDD does not occur at adequate strength or is not exhibited at all and which
absence is inappropriate (Howlin, 1997; Lord & Magill-Evans, 1995; Schreibman,
1994; Tsai, 1992). When a typically developing adult interacts with an adult with
PDD, we can assume that the peculiarities of the person with PDD have an
influence either on the behaviour of the person without PDD or on the perception
that the person forms on the PDD adult or both. Even so, the behaviour of the
typically developing adult will influence the behaviour and thoughts or feelings of
the person with PDD (Lord & Magill-Evans, 1995). In the present study it is
possible to investigate whether or not the combination between some behavioural
characteristics and the content of the dyad members thought/feeling entries
affected the inference ability of both groups. Therefore, part 2 was conducted.
PART 2
In the second part, the procedure used was the standard stimulus paradigm
(Marangoni et al., 1995). A panel of typically developing persons was asked to
Chapter 3
68
view all eleven videotapes of the dyadic interaction and to make inferences about
the specific content of the thought/feeling entries of each dyad member. The
perceivers were kept unaware that in each videotape one of the interacting persons
was having a PDD diagnosis. The standard stimulus paradigm enabled us to assess
whether or not perceivers reached different accuracy rates for thought/feeling
entries belonging to persons with PDD and for those belonging to typically
developing persons. On the basis of the characteristics of persons with PDD, we
assume that the accuracy scores of the perceivers will be higher for thought/feeling
entries belonging to the typically developing persons than for those belonging to
the persons with PDD.
METHOD
Participants
The participants were thirteen typically developing subjects who were recruited
by a temping agency. All participants (8 male and 5 female) were students. On the
basis of their study we can assume that they are of normal intelligence. The mean
chronological age of the group was 21.12 year (SD = 2.07).
Procedure
The procedure used was based on Ickes and colleagues (Ickes, Bissonette, et al.,
1990; Marangoni et al., 1995). The thirteen participants were invited to come
together to our laboratory. They were randomly divided into two groups. The first
group comprised 7 participants who viewed six of the eleven above-mentioned
videotapes during a day. The second group comprised 6 participants who viewed
eight of the eleven videotapes. By doing so, three videotapes were seen by both
groups. In order to avoid bias, all participants were kept unaware of the purpose of
the study.
The procedure was based on the standard stimulus paradigm and was essentially
the same as in the first part. The participants of each group were seated before a
video-screen and were instructed to view each videotape one time without an
69
interruption. After viewing the videotape in its entirely, the experimenter asked
them to view the videotape a second time. However, this time the experimenter
manually interrupted the videotape at each of those points at which one of the two
interaction partners had reported a specific unexpressed thought or feeling.
Whenever the videotape was paused, the members of the panel were asked to make
inferences about the specific content of the unexpressed thought/feeling entries and
to write down (a) whether the entry was presumed to be a thought or a feeling, and
(b) the specific content of the thought/feeling entry. To ensure that the perceivers
clearly understood the procedure, a preparatory session was given with other
material. When the participants had viewed all tapes and had completed the task,
they were debriefed more fully and thanked for their participation in the study.
Empathic Accuracy Measure
The empathic accuracy scores of the members of the panel were computed by
using the same logic and procedure as described in the first part. The same abovementioned five independent judges had to compare each perceivers inferred
thought/feeling entry with the corresponding original thought/feeling entry and to
rate the level of similarity on a 3-point scale, ranging from 0 (essentially different
same content). The internal consistency (Cronbachs alpha) of the five judges
content accuracy ratings was .86 for the thought/feeling entries belonging to the
PDD subjects and .94 for the thought/feeling entries belonging to the control
subjects. Similar to part 1, the original empathic accuracy scores were calculated
for each perceiver.
The baseline level of empathic accuracy for each of the panel members was
estimated by randomly pairing each set of the perceivers inferred thought/feeling
entries with the corresponding set of the original thought/feeling entries and rating
the content of these randomly paired original/inferred entries on similarity. The
internal consistency of the baseline accuracy provided by the five judges was .87
for the thought/feeling entries belonging to the PDD group and .67 for the
thought/feeling entries belonging to the control group. Following the logic of the
first part, the baseline empathic accuracy scores were calculated for each member
of the panel. By subtracting for each panel member the baseline empathic accuracy
Chapter 3
70
scores from the original content scores, we derived a measure of global empathic
accuracy for each panel member.
RESULTS
The mean original content accuracy score (M = 32.60%, SD = 9.28%) of the
panel for thought/feeling entries belonging to the PDD subjects did not differ
significantly from the mean original content accuracy score (M = 28.72%, SD =
11.56%) of the panel for thought/feeling entries belonging to the control subjects,
t(10) = -.85. Furthermore, the mean adjusted empathic accuracy score of the panel
was 19.83% (SD = 9.82%) for thought/feeling entries belonging to the PDD
subjects and 19.25% (SD = 9.29%) for thoughts and feelings belonging to the
control subjects. Analysis revealed no significant differences in inferring the
thoughts and feelings of adults with PDD and those of control adults, t(10) = -.15.
GENERAL DISCUSSION
This paper attempted to measure the social functioning of eleven normally
intelligent adults with PDD during a naturalistic conversation with a typically
developing stranger. The paper differs from previous research in that the procedure
used enabled us to study the overt and the covert behaviour of the interacting
participants. The study consisted of two parts.
In the first part, analyses of the overt behaviour revealed that during a
naturalistic dyadic conversation subjects with PDD verbalise longer than control
subjects at the moment they get up to speak and that the period of each look at the
interaction partner is shorter for subjects with PDD than for control subjects.
Furthermore, subjects with PDD tended to speak more than control subjects. No
significant within-dyad differences were found with regard to the stimulatory
gestures. These data are consistent with other behavioural studies in which was
found that the differences between subjects with PDD and control subjects in the
frequency of non-verbal expression during social interaction are much less than
expected (Tantam, Holmes, &
71
Cordess, 1993; Willemsen-Swinkels, Buitelaar,
Weijnen, & Van Engeland, 1998; Van Engeland, Bodnar, & Bolhuis, 1985). As is
expected in studies with typically developing dyads (e.g. Stinson & Ickes, 1992),
we further found several reciprocities between the verbalisations, gazes and
stimulatory gestures of the dyad members, indicating that most behaviours of the
dyad member are reciprocated by one or more behaviours of the interaction partner.
The dyadic interaction design of the first part enabled us to explore the theme(s)
of the dyad members original thoughts and feelings. Analyses indicated that the
percentages of thought/feeling entries belonging to a specific topic were merely the
same for the PDD subjects and the control subjects, with the exception of thoughts
and feelings that focussed on the interaction partner. Control subjects had twice as
much thought/feeling entries that focussed on the interaction partner than subjects
with PDD. Further analyses demonstrated that the thought/feeling entries of the
adults with PDD were as difficult as those of the typically developing adults.
Although the mean level of abstractness was similar for thoughts and feelings
belonging to subjects with PDD and for those belonging to control subjects,
subjects with PDD had more abstract thoughts and feelings than control subjects
and had less concrete thoughts and feelings than control subjects.
Contrary to our main hypothesis, the subjects with PDD did not differ from the
typically developing subjects in the ability to infer the thoughts and feelings of their
interaction partner. Further analyses indicated that the empathic accuracy scores
decreased with an increasing level of difficulty, although this effect was similar for
both groups. Only one main effect of topic on the empathic accuracy scores was
found, indicating that thought/feeling entries that focussed on a tangible
environmental object or event are less accurately inferred than the thought/feeling
entries without this focus. Furthermore, the findings do not endorse the assumption
that the association between the participants behavioural characteristics and the
empathic accuracy scores is different for subjects with PDD and control subjects,
which -in a way- is inconsistent with the findings of Koning and Magill-Evans
(2001).
In the second part, we explored whether or not the inference ability of both
groups was affected by the combination between some behavioural characteristics
and the content of the interaction partners thoughts and feelings. The analyses
Chapter 3
72
revealed that the thought/feeling entries of the PDD subjects were as difficult to
infer as the thought/feeling entries of the control subjects, which indicated that the
inference ability of both groups was independent of the dyad members behavioural
characteristics and the content of the dyad members thoughts and feelings.
The results of both parts are surprising given the daily life perspective taking
difficulties of adults with PDD and implicate that, under some circumstances, some
high-functioning adults with PDD are able to read the thoughts and feelings of
others during a naturalistic conversation. However, it should be noted that the highfunctioning adults with PDD had an intelligence level far above the normal range
and were invited to participate in the present study on the basis of their good
performance on previous mind-reading tasks (Roeyers et al., 2001). Many
researchers have stressed the role of both verbal and chronological age in the
performance on theory of mind tasks and have suggested that intelligence might
compensate for conceptual perspective taking strategies (Bowler, 1992; Happ,
1995; Prior, Dahlstrom, & Squires, 1990; Yirmiya & Shulman, 1996). The fact that
the IQ-scores of the participants did not correlate with the empathic accuracy
scores, suggests that it can not be taken for granted that the good empathic accuracy
scores of the present PDD group are solely due to their higher level of intelligence.
The higher intelligence of the PDD persons can play a necessary but not a sufficient
role in their task success.
There are, however, two possible alternative explanations. A first explanation is
based on social psychological research. According to Eisenberg, Murphy and
Shepard (1997), individuals use a variety of types of information to decipher what
other people are thinking or feeling. Karniol (1995) found that older children use a
greater variety of strategies than younger children while they are inferring the
thoughts of other persons, and that the strategies of older children rely more on
personal information, whereas the younger childrens strategies rely more on
situational information. Even so, Gnepp, Klayman, and Trabasso (1982) found that
the use of type of information to rely on when evaluating the emotional states of
others, varies with age. Increasing with age, people prefer to use personal
information (i.e. specific information about the individual) over normative
information (i.e. information about the group to which the individual belongs), and
normative information more than situational information (i.e. information about
73
anothers physical or social environment) (Gnepp et al., 1982). Although the

present investigation does not provide information about the specific type of
information the participants with or without PDD are relying on, it can be
considered that all participants are familiar with the script of the present study (i.e.
an initial conversation of the getting acquainted type). This familiarity might be
derived from experience or (with regard to adults with PDD) by having learned the
script previously. Therefore, cues in the situation may lead to the retrieval of
information from memory about similar situations to that of the target person, as
well as social scripts or other socially relevant knowledge (see Karniol, 1995).
However, there has been as yet no systematic examination of the relation of scriptal
knowledge of adults with PDD to their mind-reading performance. Moreover, very
little is known about the capabilities of subjects with autism in scripts. While a
recent study of Trillingsgaard (1999) with a small sample of children with autism
suggested that children with autism have significantly fewer well-organized scripts
for familiar social routines (such as make a cake or celebrate a birthday) than
normal control children, the results of a study of Volden and Johnston (1999)
suggested that the basic scriptal knowledge of children with autism appears to be
intact. The question remains whether these results can be generalised to adults with
PDD. This could be an area of further investigation.
The second explanation is allied to the first explanation. In a previous study with
the empathic accuracy design, Hancock and Ickes (1996) already suggested that an
8-minute interaction period does not provide enough time for the interactants to
import different meaning contexts into the interaction, and that it might be that two
strangers, when they interact for the first time, employ rather a generic meaning
context for the first minutes of the interaction, until they feel sufficiently
acquainted with each other to go beyond it. Although we believe that the design of
the present study is an advantage for the naturalistic study of the mind-reading
abilities of persons with PDD, every medal has two sides. On the one hand the
empathic accuracy design implies that the mind-reading abilities can be studied
ecologically, on the other hand the ecological design implies that the participants
can structure the social situation according to their own possibilities or desires. For
instance, during the initial conversation the participants can talk when they are
willing to do, but if one of the participants does not want to talk, he/she does not
Chapter 3
74
have to talk. This implies that the participants can influence the conversation and,
in a way, have control over the situation. Even so, if one of the interaction partners
has lower social skills than the other, then the level of the conversation will drop as
far as the level of social abilities of the first interaction partner. However, the
advantage of being able to structure the present 8-minute social interaction can be
considered as non-recurrent, because in daily life it is impossible to structure the
miscellany of protracted social interactions. Furthermore, it should be noted that the
ability to structure the situation has an impact on both the content of the
thought/feeling entries of each dyad member and the inference ability of each
interaction partner. Suppose that the dyad members decided to have only small
talks with each other and did not go beyond the generic meaning context of the
getting acquainted type, then the content of their thoughts and feelings would
also be rather generic. Inferring generic thoughts and feelings requires different
(and probably less) social-cognitive acquirements from the interaction partner, than
inferring thoughts and feelings that do not (or less) represent familiar experiences.
The fact that the present mind-reading performance of the eleven participants
with PDD correlates with their performance on the first empathic accuracy
videotape in the study of Roeyers et al. (2001) and does not correlate with their
performance on the second videotape, does not rule out either explanation. As
noted by Roeyers et al. (2001) the first videotape was more structured and more
predictable than the second videotape. In order to obtain more information about
the relationship between scriptal knowledge, structuralisation and mind-reading
performance, we are at the moment conducting a study in which adults with a
pervasive developmental disorder each have to view different videotapes with
varying level of structure and with different scripts, and have to infer the thoughts
and feelings of the videotaped targets.
The present study may contribute towards a better understanding of the mindreading abilities of adults with PDD during a naturalistic conversation with a
stranger. Nevertheless, it should be noted that the design of the study still differs
from any real life social situation in several ways. First, the participants had to infer
the thoughts or feelings of their interaction partner in a retrograde phase (i.e. while
they were viewing the videotape of their own conversation for a second time). The
demands of daily life do not permit us to review our interactions and expect to
75
make quick inferences about the thoughts and feelings of the other at the specific
moment these thoughts and feelings occur. Second, the design permits the
participants to use as much time as needed to infer the thoughts and feelings of
their interaction partner. Although the subjects with PDD did not need more time
than the control subjects to infer the others thoughts and feelings, in daily life they
may not have the time they need and it might be that therefore their impaired mindreading abilities become more apparent. Third, in daily life we do not have access
to the original thoughts and feelings of the interactants. It can be suggested that
persons with PDD might have had more difficulties to write down their original
thought/feeling entries, although there is no conclusive evidence to support this
hypothesis. On the one hand this hypothesis can be supported by the fact that
subjects with PDD tended to need more time to write down a single thought/feeling
entry. On the other hand, the analysis revealed no significant within-dyad
difference in the total amount to write down the thought/feeling entries. Besides the
mental strain to write down the original thoughts and feelings, it should be pointed
out that there is no reason to believe that persons with PDD are less accurate in
writing down their own thoughts and feelings. We have previously shown that the
mean difficulty and the mean abstractness of thought/feeling entries of the subjects
with PDD were similar to those of the control persons, and no significant
differences were found in the inference ability of the thirteen typically developing
subjects with regard to the thought/feeling entries belonging to subjects with PDD
and those belonging to control subjects.
Finally, it seems to be that being in the interaction yields higher empathic
accuracy scores than perceiving the interaction. While the mean empathic accuracy
scores of the thirteen perceivers in part 2 were around 19%, the mean empathic
accuracy scores of the dyad members in part 1 were around 25%. This suggests that
the empathic accuracy scores are not only mediated by components of the perceiver
and components of the target, but also by the intersubjective meaning context of
both dyad members (who are target and perceiver at the same time). Although
highly hypothetical, this suggests that the intersubjective meaning context, which is
created through the conversation and non-verbal behaviours of the dyad members,
enhances the empathic accuracy of the dyad members or, conversely, affects
negatively the empathic accuracy of participants who serve only as perceivers.
Chapter 3
76
In sum, the present study found no support for the main hypothesis that the
inference ability of persons with PDD would be more hampered than the inference
ability of control persons while having a naturalistic conversation with each other.
The study differs from previous research in that both the overt behaviour and the
covert thoughts and feelings of subjects with and without PDD were meticulously
analysed in a standardised manner. The results of the study underline the
importance of exploring the impact of previous scriptal knowledge on the mindreading performance and the possible advantage the adults with PDD turn from
more structured situations compared with less structured situations.
REFERENCES
American Psychiatric Association (1994). Diagnostic and statistical manual of mental
disorders (fourth edition). Washington, DC : Author.
Attwood, T. (1998). Aspergers syndrome: A guide for parents and professionals. Jessica
Kingsley Publishers: London.
Bowler, D. M. (1992). Theory of mind in Aspergers syndrome. Journal of Child
In W. Ickes (Ed.), Empathic accuracy (pp. 73-116). New York: Guilford Press.
Gnepp, J., Klayman, J., & Trabasso, T. (1982). A hierarchy of information sources for
inferring emotional reactions. Journal of Experimental Child Psychology, 33, 111-123.
Hancock, M., & Ickes, W. (1996). Empathic accuracy: When does the perceiver-target
relationship make a difference? Journal of Social and Personal Relationships, 13, 179199.
Happ, F. G. A. (1995). The role of age and verbal ability in the theory of mind task
77
Hays, W.L. (1994). Statistics. Fifth edition. Harcourt Brace College Publishers: Orlando.
Hobson, P., & Lee, A. (1998). Hello and goodbye: A study of social engagement in
autism. Journal of Autism and Developmental Disorders, 28, 117-127.
Howlin, P. (1997). Autism: Preparing for adulthood. Routledge: New York.
Howlin, P. (1998). Practitioner review: Psychological and educational treatments for
autism. Journal of Child Psychology and Psychiatry, 39, 307-322.
dyadic interaction paradigm. In C. Hendrick & M.S. Clark (Eds.), Review of personality
and social psychology: Research methods in personality and social psychology (Vol.
11, pp. 16-44). Newbury Park, CA: Sage.
Ickes, W., Robertson, E., Tooke, W., & Teng, G. (1986). Naturalistic social cognition:
Methodology, assessment, and validation. Journal of Personality and Social
59, 730-742.
Kaland, N., Moller-Nielsen, A., Callesen, K., Mortensen, E.L., Gottlieb, D., & Smith, L.
(2002). A new advanced test of theory of mind: Evidence from children and
adolescents with Asperger syndrome. Journal of Child Psychology and Psychiatry, 43,
517-528.
Karniol, R. (1995). Developmental and individual differences in predicting others
thoughts and feelings: Applying the transformation rule model. In N. Eisenberg (Ed.),
Review of personality and social psychology: Vol. 15. Social development (pp. 27-48).
Thousand Oaks, CA: Sage.
Kleinman, J., Marciano, P., &
Ault, R. (2001). Advanced theory of mind in high-
functioning adults with autism. Journal of Autism and Developmental Disorders, 31,
29-36.
Lord, C., & Magill-Evans, J. (1995). Peer interactions of autistic children and adolescents.
Development and Psychopathology, 7, 611-626.
Chapter 3
78
Noldus, L.P. (1991). The observer: A software system for collection and analysis of
observational data. Behavior Research Methods, Instruments, & Computers, 23, 415429.
Noldus L.P., Trienes, R., Hendriksen, A., Jansen, H., & Jansen, R.G. (2000). The Observer
Video-Pro: New software for the collection, management, and presentation of timestructured data from videotapes and digital media files. Behavior Research Methods,
Instruments, & Computers, 32, 197-206.
Ponnet, K., Roeyers, H., Buysse, A., De Clercq, A., & Van der Heyden, E. (2004).
Advanced mind-reading in adults with Asperger syndrome. Autism, in press.
Schreibman, L. (1994). General principles of behavior management. In E. Schopler & G.
Mesibov (Eds.), Behavioral issues in autism (pp. 13- 38). Plenum Press: New York.
Simpson, J. A., Ickes, W., & Blackstone, T. (1995). When the head protects the heart:
Tantam, D., Holmes, D., & Cordess, C. (1993). Nonverbal expression in autism of
Asperger type. Journal of Autism and Developmental Disorders, 23, 111-133.
Trillingsgaard, P. (1999). The script model in relation to autism. European Child &
Adolescent Psychiatry, 8, 45-49.
Tsai, L. (1992). Diagnostic issues in high-functioning autism. In E. Schopler and G.
Mesibov (Eds.), High-functioning individuals with autism (pp. 11-40). Plenum Press:
New York.
79
Van Engeland, H., Bodnar, F. A., & Bolhuis, G. (1985). Some qualitative aspects of the
social behavior of autistic children: An ethological approach. Journal of Child
Psychology, Psychiatry and Allied Disciplines, 26, 879-893.
Volden, J., & Johnston, J. (1999). Cognitive scripts in autistic children and adolescents.
Journal of Autism and Developmental Disorders, 29, 203-211.
Willemsen-Swinkels, S., Buitelaar, J., Weijnen, F., & Van Engeland, H. (1998). Timing of
social gaze behaviour in children with a pervasive developmental disorder. Journal of
Autism and Developmental Disorders, 28, 199-210.
Wing, L. (1992). Manifestations of social problems in high-functioning autistic people. In
E. Schopler and G. Mesibov (Eds.), High-functioning individuals with autism (pp. 1140). Plenum Press: New York.
Yirmiya, N., & Shulman, C. (1996). Seriation, conservation, and theory of mind abilities
in individuals with autism, individuals with mental retardation, and normally
developing children. Child Development, 67, 2045-2059.
Zager, D. B. (1999). Autism: Identification, Education, and Treatment. Lawrence Erlbaum
Associates, Publishers: New Jersey.
CHAPTER
Empathic accuracy in young adults with

PDD: Does structure matter ?1
This chapter is based on Ponnet, K., Buysse, A., Roeyers, H., & De Clercq, A. (2004). Empathic accuracy
in young adults with PDD: Does structure matter ? Manuscript submitted for publication.
Chapter 4
82
ABSTRACT
This study further elaborates on the mind-reading impairments of young adults
with PDD. The hypothesis is that differences in mind-reading abilities between
subjects with PDD and control subjects become more apparent when they have to
infer thoughts and feelings of other persons in a less structured or more chaotic
conversation, than when they have to do so in a more structured conversation.
Conform to the empathic accuracy design, subjects viewed two videotaped
interactions depicting two strangers and attempted to infer the thoughts and feelings
of these persons. The videotaped interactions were manipulated in such a way that
the conversation in one of the videotapes was less structured than in the other.
The results suggest that subjects with PDD have poorer mind-reading skills than
control subjects when confronted with the less structured conversation. No
between-group differences were found when subjects had to infer the thoughts and
feelings of other persons in the more structured conversation. The findings
underscore the significance of structure to the mind-reading abilities of young
adults with PDD.
Empathic accuracy in young adults with PDD
83
INTRODUCTION
It is well established that the majority of children with autism or a related
pervasive developmental disorder (PDD) have impairments in the development of a
theory of mind (Bailey, Phillips, & Rutter, 1996; Baron-Cohen, 1995; BaronCohen, Leslie, & Frith, 1985). The theory of mind (ToM) account of autism
suggests that the impairments of subjects with autism or PDD can be explained in
terms of a failure to attribute mental states (such as beliefs, desires, or intentions)
and feelings to oneself and others, which results in socially and communicatively
inadequate behaviour (Baron-Cohen, Tager-Flusberg, & Cohen, 2000). Without a
theory of mind, people are unable to develop a normal understanding that other
people have mental states and feelings. Theory of mind is also referred to as mindreading. Over the past decades, an increasing amount of research was devoted to
the development of mind-reading tasks for use with children with PDD (Happ,
1994a; Sodian & Frith, 1992). With a few exceptions (e.g. Bauminger & Kasari,
1999; Dahlgren & Trillingsgaard, 1996; Prior, Dahlstrom, & Squires, 1990), these
studies largely support the theory of mind hypothesis.
However, a substantial number of adolescents and adults with PDD were found
to pass the traditional mind-reading tasks (e.g. Bowler, 1992; Happ, 1994a; TagerFlusberg & Sullivan, 1995). These studies cannot be taken as a conclusive evidence
of an intact social-cognitive understanding in individuals with PDD because
typically developing children pass such tasks at about 6 years (Baron-Cohen,
Jolliffe, Mortimore, & Robertson, 1997; Perner & Wimmer, 1985). They might
only indicate that the traditional mind-reading tasks, which are designed for
children with PDD, are not subtle enough to detect impairments in adults with PDD
(Rutterford, Baron-Cohen, & Wheelwright, 2002).
The last few years, an increasing amount of research is devoted to the
development of more advanced adult tests of theory of mind (Baron-Cohen et al.,
1997; Baron-Cohen, Wheelwright, Hill, Raste, & Plumb, 2001; Rutterford et al.,
2002). These advanced tasks have a more mature and challenging content, which
distinguishes them from mind-reading tasks developed for younger subject
populations (Heavey, Phillips, Baron-Cohen, & Rutter, 2000). However, the
ecological validity of most of the advanced ToM tasks is still debatable, because
Chapter 4
84
the greater part of these tasks makes use of static stimuli (i.e. verbal or non-verbal
stimuli). In order to measure the mind-reading of adults with PDD in contexts that
approximate closely the real social world, several authors have called for more
dynamic tasks, in which verbal and non-verbal stimuli (e.g. videomaterial) are used
(Baron-Cohen et al., 2001; Kleinman, Marciano, & Ault, 2001; Roeyers, Buysse,
Ponnet, & Pichal, 2001).
A validated research paradigm that offers the opportunity for a systematic
investigation of the mind-reading in a laboratory situation is the empathic accuracy
design of Ickes and colleagues (Ickes, 1993; Ickes, Bissonnette, Garcia, & Stinson,
1990). Empathic accuracy is the extent to which someone is able to infer accurately
the specific content of another persons thoughts and feelings (Ickes, Stinson,
Bissonnette, & Garcia, 1990; Simpson, Ickes, & Blackstone, 1995). According to
Ickes (1993), the most unequivocal way to measure empathic accuracy is to
compare the content of a target persons actual thoughts and feelings with the
content of the corresponding inferred thoughts and feelings reported by the
perceiver. Over the years, the empathic accuracy paradigm has consistently proved
to be a reliable and valid way to measure peoples inference abilities (Buysse &
Ickes, 1999; Ickes, Stinson, et al., 1990; Marangoni, Garcia, Ickes, & Teng, 1995;
Stinson & Ickes, 1992).
In two recent studies (Ponnet, Roeyers, Buysse, De Clercq, & Van der Heyden,
2004; Roeyers, et al. 2001), we tested adults with PDD and control adults using a
naturalistic Empathic Accuracy Task. In this task, individual participants each
attempted to infer the thoughts and feelings of a target person, while viewing a
videotape of the target person in a naturally occurring conversation with a stranger.
The procedure used in the studies was the standard stimulus design (Gesn & Ickes,
1999; Marangoni et al., 1995), in which the participants each view the same set of
two videotaped interactions between two strangers. In both studies, the Empathic
Accuracy Task was able to distinguish between the adults with PDD and the
control adults, although a significant between-group difference was found only for
one of the empathic accuracy videotapes in favour of the control adults. The mixed
results can be explained by the different nature of both videotape tasks. Although
we had not the intention to manipulate the videotapes degree of structure, the
85
social interaction in the first videotape was more structured than that in the second
one.
Since the empathic accuracy paradigm has proven to be a reliable way to
measure adults inference abilities as well as to distinguish adults with PDD from
typically developing adults, we want to test a more fine-grained hypothesis. More
specifically, the present study attempts to explore the possible significance of
structure to the empathic accuracy of subjects with PDD. As is well known,
subjects with PDD prefer activities and situations that are more structured and
provide clear, explicitly stated rules (Bauminger & Shulman, 2003; Howlin, 1997;
Zager, 1999). Although the consensus is that subjects with PDD benefit from
structured situations (Howlin, 1997; Mesibov, 1992), structure has seldom
explicitly been manipulated in empathic situations. This study is related to this
issue.
In the present study, we developed a new Empathic Accuracy Task in which
young adults with PDD have to infer the thoughts and feelings of other persons.
The procedure used was the standard stimulus empathic accuracy paradigm
(Marangoni et al., 1995), in which individual participants (i.e. subjects with PDD
and control subjects) each view two videotaped getting acquainted conversations
between two strangers and attempt to infer the thoughts and feelings of the same set
of target persons. We manipulated the structure of both videotapes in such way that
one videotape was more structured than the other, and we suggest that the inference
impairments of young adults with PDD may only become apparent when
confronted with the less structured conversation.
In addition to the structure manipulation, the present Empathic Accuracy Task
extends our previous studies in two ways. First, two empathic accuracy measures
were calculated, whereas in our previous studies we only calculated one. Although
empathy is believed to be based upon the cognitive ability to understand others
emotions and cognitions, empathy is also characterised by a substantial emotional
component (Eisenberg, Murphy, & Sheppard, 1997). Consequently, we studied the
content accuracy as well as the valence accuracy of subjects with PDD and control
subjects. Content accuracy refers to the degree to which the perceivers description
of the targets thoughts or feelings matches the content of the specific thoughts or
feeling reported by the target. Valence accuracy is part of the emotional component
Chapter 4
86
and refers to the degree to which the perceivers inferences about the emotional
tone of the targets actual thoughts or feelings matches the actual valence of the
specific thoughts or feeling reported by the target (Ickes, Stinson, et al., 1990).
Second, we can expect that variables other than structure might influence the mindreading abilities of subjects with PDD as well. In a conversation, the subjects
verbal and non-verbal behaviours provide information about their inner thoughts
and feelings. So, subjects who are good interpreters of behavioural cues will be
better at inferring other peoples thoughts and feelings. Although many subjects
with PDD are aware that verbal and non-verbal cues play a crucial role in social
interactions, it might be difficult to them to interpret these cues correctly (Howlin,
1997). Therefore, we studied both the overt behaviour of the videotaped strangers
as well as the covert thoughts and feelings of the interacting targets. With regard to
the overt behaviour, we calculated the congruence between the videotaped
strangers verbal communication and their written thoughts and feelings, and the
congruence between the strangers direction of gaze and their thoughts and
feelings. Subsequently, we calculated for each participant more refined accuracy
scores along these topics. With regard to the latter, we assessed the level of
difficulty and concreteness, and the affective tone of the written original thoughts
and feelings, and for each participant we further worked out more refined empathic
accuracy scores along the nature of the original thoughts and feelings.
In the present study, we expect significant differences between adults with PDD
and control adults when they have to infer the content of other persons thoughts
and feelings in a more chaotic and less structured conversation. However, no
significant between-group differences are expected when subjects have to infer the
content of other persons thoughts and feelings in a more structured conversation.
In addition, we expect to get a clearer picture of the empathic accuracy scores of
subjects with PDD when more refined content accuracy scores and valence
accuracy scores are calculated.
87
METHOD
Participants
present study: 30 subjects with PDD and a control group of 30 typically developing
subjects. Male/female ratio (22/8) was similar in each group. The subjects were
recruited with the help of parent associations and schools. All subjects with PDD
had been diagnosed by a multidisciplinary team of experienced clinicians and
fulfilled established DSM-III-r or DSM-IV criteria for autism, Asperger syndrome
or PDD-NOS (APA, 1987, 1994), depending upon which was most recent at the
time of diagnosis. The mean general level of intelligence was above average. Only
subjects with a Verbal IQ of at least 75 and a minimum chronological age of 15
years were included in the sample. The typically developing control subjects were
matched meticulously on a one-to-one basis on sex, chronological age and IQ. All
subjects were administered 11 of the 14 subtests of the WAIS-III (Wechsler, 2000),
so that we could calculate the four indices and that the three IQ-scores could be
estimated. Further background information for both groups is presented in Table 1.
Table 1
Descriptive characteristics of the sample
PDD
Control
SD
SD
F(1, 59)
Chronological age
19.73
3.69
19.64
3.13
.01
Verbal IQ
98.13
13.69
95.57
10.29
.67
Performance IQ
90.70
15.96
94.27
12.96
.90
Total IQ
94.47
13.89
94.40
11.81
.00
Verbal Comprehension
99.90
13.19
96.03
10.53
1.57
Perceptual Organisation
93.63
13.91
95.53
12.29
.31
Working Memory
93.93
16.55
97.13
14.02
.65
Processing Speed
83.33
17.04
94.00
22.31
4.33 *
*p < .05
Chapter 4
88
Materials
Empathic Accuracy Stimulus Tapes
The stimulus materials consisted of 2 videotapes of 2 volunteers, with varying
level of structure. For each of the videotapes, two opposite-sex individuals
(targets), previously unacquainted with each other, were scheduled to come to our
research centre at the same time. They were asked to participate in a study on
peoples perceptions of others during different situations. They were told in
advance that they were about to meet another stranger and that their interaction
would be videotaped for use in future research. Once they arrived at the research
centre, the two targets were brought to different waiting areas so that they would
not meet and interact before the session began. They were then brought together.
In the first videotape, the research assistant told the participants they had to fill
in some inquiry forms. However, when the research assistant reached for the
inquiry forms, he discovered that some of the copies were not well printed. In
order to get some proper copies, the experimenter left the room. Promising to return
in a few minutes, he suggested both participants to seek acquaintance with each
other during his absence. At this point, a concealed video camera was activated and
recorded the initial conversation between the two targets.
In the second videotape, the research assistant explained the targets they were to
play a board game together, but that the rules of the game require that both
participants get to know each other personally before starting the game. In order to
become acquainted with each other in a decent and less stressful manner, the
experimenter proposed to leave the room for approximately 8 minutes. Before
leaving, he gave the targets an 8-point list with questions they surely have to know
from each other (see Appendix A), and encouraged the targets to get acquainted
with each other as closely as possible. At the moment the experimenter left the
room, a second research assistant activated a concealed video camera.
In both videotapes, the research assistant returned after approximately 8 minutes
and partly debriefed the targets. The research assistant further informed the targets
that their written consent was required for the tape to be used. The stimulus tapes
consisted of two initial conversations between two strangers. The targets were all
white, 18 or 19 years old (within the age-range of the participants), and college
89
students. We presented the preparation of the stimulus tapes in the abovementioned contexts for several reasons. First, both contexts served as a plausible
reason to invite the targets to come to the research centre and enabled us to
videotape them unobtrusively. Second, the explicit instruction in video 2 to get to
know each other more personally increased the possibility of a more structured
conversation between the targets, whereas the absence of such an explicit
instruction in video 1 did not. Third, the questions of the eight-point list (see
Appendix A) can be considered as highly typical questions for an initial
conversation of the getting acquainted type. Therefore, instructing the targets in
video 2 to base their conversation on the list was a guarantee that the conversation
of the targets would be one of the getting acquainted type and also structured the
conversation. Finally, as will be described in the procedure, the list was presented
to the participants as a guideline, which also enhances the structuralisation.
Immediately following the videotape session, the targets were asked to view
their videotape and to make a written record of all the unexpressed thoughts and
feelings they remembered having had during the acquaintance. The targets were
each seated at a separate small table in one of two respective areas of the room,
where each of them had a clear view of a single, large TV monitor. Following the
procedure of previous studies (Buysse & Ickes, 1999; Ickes, Bissonnette, et al.,
1990; Roeyers et al., 2001), the videotape was activated and the targets were
instructed to say stop at each point during the interaction when they remembered
having had a specific thought or feeling. At each of those stops, the targets were
asked to write down on a standardised thought/feeling coding form (a) the time
when the thought or feeling occurred (as displayed by a digital clock on the underright corner of the videotape), (b) whether the entry was a thought or a feeling, (c)
the specific content of the thought/feeling entry, and (d) whether the entry was
positive, negative, or neutral in its overall affective tone. The targets were
instructed to provide an accurate, honest and complete account of all the thoughts
and feelings they remembered having had. In addition, the targets were guaranteed
that their actual thought/feeling entries would never be shown to their interaction
partner. On the other hand, the instruction cautioned them to report only those
thoughts and feelings they distinctly remembered having had during the interaction
Chapter 4
90
period and not to report any new thoughts and feelings that occurred to them while
viewing the videotape.
Video 1 contained 43 thought/feeling entries (22 female/21 male) and video 2
contained 74 thought/feeling entries (36 female/38 male). After recording the
thoughts and feelings, the targets were debriefed more thoroughly. All targets
agreed to sign a release form allowing their videotape and thought/feeling data to
be used as stimuli in the present study.
Because of the large number of thoughts and feelings in both videotapes, we
reduced the number of thoughts and feelings in a random stratified way to 28
thought/feeling entries per videotape, so that each videotape contained 14 male and
14 female thought/feeling entries. Video 1 lasted for 7 min 31 s and video 2 lasted
for 6 min 53 s.
An ANOVA revealed that the percentages of thought/feeling entries indicated as
positive (35.7% in video 1 and 39.3% in video 2), neutral (32.1% in video 1 and
28.6% in video 2) or negative (32.1% in video 1 and 32.1% in video 2) did not
differ significantly between both videotapes, F(1, 54) < 1.
Sample thought/feeling entries reported by the male target in the first videotape
were: I have to say something, What a lovely girl, Shall I say something
funny ?. Sample entries reported by the female target in the first videotape were:
What is he saying ?, I have to ask him something ?. Sample thought/feeling
entries reported by the male target in the second tape were: Ouch, we have to get
to know each other personally, Which are our mutual hobbies ?, Wow, I
promote myself a lot.. Sample entries reported by the female target in the second
videotape were: Shall I laugh to decrease the tension ?, I feel more at ease than
only a minute ago.
As mentioned above, video 2 was more structured than video 1. To check the
videotapes level of structure, 60 typically developing persons (30 male/30 female)
between 16 and 32 years old (M = 22.60, SD = 3.39) were invited to come to our
laboratory. All of them were students who were nave to the aims of the study.
They were instructed to view the videotapes with the order of videotape
presentation counterbalanced between subjects. After each videotape, the subjects
had to rate on a 10-point scale, ranging from 1 (not at all typical) to 10 (very
typical), to what degree 40 items were typical for the videotaped interaction.
91
Among these 40 items, 13 items2 assessed how structured each videotape was. The
internal consistency (Cronbachs alpha) of the 13 items provided by the 60 subjects
was .80 for video 1 and .90 for video 2. Then, mean scores were calculated for each
participant and for each videotape, by adding up for each videotape the
participants scores on the 13 items and by dividing this aggregate score by the total
number of items. The mean structure of video 1 was 4.94 (SD = 1.14) on a total
score of 10, and the mean structure of video 2 was 7.04 (SD = 1.32). A paired t-test
revealed a significant difference between both videotapes, t(59) = -12.06, p < .001,
indicating that video 2 was more structured than video 1.
Procedure
All participants were tested individually in a quiet laboratory room at the
University. Each participant was tested in two phases. In the first phase, the WAISIII was administered. Approximately two weeks later, subjects were invited to
come for a second time, during which they were instructed to view the empathic
accuracy stimulus tapes with the order of videotape presentation counterbalanced
between the participants. While viewing the videotape, the tape was paused at
precisely those moments during which a target had recorded a specific thought or
feeling. The stimulus tapes presented to the subjects were paused automatically
with the use of VIDANN (De Clercq, Buysse, Roeyers, Ickes, Ponnet, &
Verhofstadt, 2001). Whenever the videotape was paused, the subjects were asked to
make inferences about the specific content of the unexpressed thought/feeling
entries and to write down on a standardised coding form (a) whether the entry was
presumed to be a thought or feeling, (b) the specific content of the thought/feeling
entry, and (c) whether the entry was presumed to be positive, negative, or neutral in
its overall affective tone. Before each videotape, the participants were presented a
powerpoint-presentation with minimal background information about the
construction of the videotapes and, with regard to video 2, the eight-point list (see
Appendix A). Furthermore, to ensure that the subjects clearly understood the
The 13 items were synonyms or were connected semantically with each other (e.g. structured,
organized, surveyable, coherent, predictable).
Chapter 4
92
procedure of the Empathic Accuracy Task, an advanced preparatory session was

given.
Measures
Empathic Accuracy Task: Actual Thought/Feeling Measures
As suggested by Ickes, Bissonnette, et al. (1990), the subjects thought/feeling
entries were coded by five independent coders. The coders watched the videotapes
that were stopped each time one of the targets had reported having had a specific
thought or feeling. The coders were instructed to rate for each thought/feeling entry
how difficult they thought it was to infer the specific content of each
thought/feeling entry on a 7 point scale ranging from 1 (very easy) to 7 (very
difficult). Because the internal consistency (Cronbachs alpha) of the five judges
was .83 for all thought/feeling entries, the coders scores were averaged for each
thought/feeling entry. A univariate analysis (ANOVA) revealed no significant
difference between the mean difficulty of the thought/feeling entries of video 1 (M
= 4.17, SD = 1.15) and the mean difficulty of the thought/feeling entries of video 2
(M = 4.05, SD = 1.20), F(1, 54) < 1. Then, based on the quartile distribution of both
videotapes thought/feeling entries, the thought/feeling entries were divided into (a)
the 25.0% most easy-to-infer thought/feeling entries, (b) the 48.2% moderate
difficult-to-infer thought/feeling entries, and (c) the 26.8% most-difficult-to-infer
thought/feeling entries.
Similarly, the same five coders rated for each thought/feeling entry how concrete
or abstract the thought or feeling was on a 7 point scale ranging from 1 (very
concrete) to 7 (very abstract). Because the interrater reliability (Cronbachs alpha)
was .77 for all thoughts and feelings, the coders scores were averaged for each
thought/feeling entry. The mean abstractness of the thought/feeling entries of video
1 was 5.16 (SD = 1.07) and the mean abstractness of the thought/feeling entries of
video 2 was 5.09 (SD = .65). A univariate analysis (ANOVA) revealed no
significant difference between both videotapes, F(1, 54) < 1. Following the abovementioned procedure, the thought/feeling entries were divided into (a) the 25.0%
most abstract thought/feeling entries, (b) the 51.8% moderate abstract
thought/feeling entries, and (c) the 23.2% most concrete thought/feeling entries.
93
Finally, the same five coders assessed for each targets thought/feeling entry
whether or not the content of each thought/feeling entry was congruent with (a) the
verbalisation of the target and (b) with the targets direction of gaze. In the first
videotape 28.6% thought/feeling entries of the targets were congruent with their
verbalisation. In the second videotape, 53.6% targets thought/feeling entries were
congruent with their verbalisation. A univariate analysis (ANOVA) revealed that
the difference between both videotapes approached significance, F(1, 54) = 3.73, p
= .06. Furthermore, the number of thought/feeling entries that was congruent with
the targets direction of gaze was similar in both videotapes (i.e. 35.7%).
Empathic Accuracy Task: Empathic Accuracy Measures
Two measures of empathic accuracy were computed from comparisons of the
actual and inferred thought/feeling entries. The first measure, valence accuracy, is
an index of the proportion of instances in which each inference about the overall
emotional tone (negative, neutral or positive) matches the actual valence label
assigned to each entry. Following the logic and procedure described by Ickes and
colleagues (Ickes, Stinson, et al., 1990), we computed the number of matches and
mismatches between the rated valence of the actual entries and the corresponding
inferences for each perceiver. We then computed the percentage of correct matches
for each perceiver.
The second measure, content accuracy, is an index of the degree to which a
perceivers written description of the inferred content matches the targets actual
content of the sentence. As suggested by Ickes, Bissonnette, et al. (1990), five nave
and independent judges were instructed to compare each perceivers inferred
thought/feeling entry with the corresponding original thought/feeling entry and to
rate the level of similarity on a 3-point scale, ranging from 0 (essentially different
same content), with I dont know and missing responses rated 0. The internal
consistency (Cronbachs alpha) of the five judges content accuracy ratings was .94
for Video 1 and .90 for Video 2. The mean of the accuracy scores assigned by the
five judges was calculated for each individual inference. We then calculated
aggregated accuracy scores from the averaged scores for each inference. The mean
ratings were therefore summed across the thought/feeling entries for each of the
Chapter 4
94
two videotapes. These summed scores were divided by the total number of
thought/feeling entries for each target and multiplied by 100. By doing so, the
minimum and maximum empathic accuracy scores of each perceiver could vary
between 0 (total inaccuracy) and 100 (total accuracy).
Following the logic and procedures described by Ickes and colleagues (Ickes,
Bissonnette, et al., 1990; Ickes, Stinson, et al., 1990), a measure of baseline
empathic accuracy was similarly derived for each of the perceivers. This logic
suggested that perceivers could simply make a fortuitous guess about the content of
the thought/feeling entry and be sometimes correct because of chance alone (Ickes,
Stinson, et al., 1990). A straightforward method to estimate the baseline accuracy
level that reflects this false accuracy component is to randomly pair each set of the
actual thought/feeling entries with the corresponding set of the perceivers
inferences and to ask independent judges to make similarity ratings of these
randomly paired actual/inferred entries. The Cronbachs alpha of the baseline
accuracy provided by the same five judges was .60 for Video 1 and .77 for Video 2.
The mean of the baseline empathic accuracy scores rated by the five judges was
further calculated for each individual inference and aggregated accuracy scores
were calculated from the averaged scores for each inference. We then summed the
mean ratings and these values were divided by the total amount of thought/feeling
entries and multiplied by 100. By subtracting for each perceiver the baseline
empathic accuracy scores from the original content scores, we derived a revised
measure of global empathic accuracy score for each perceiver.
Following this logic, we calculated some more refined empathic accuracy scores
for each videotape. First, the empathic accuracy scores were calculated for all
thoughts and feelings belonging to (a) the most easy-to-infer thought/feeling
entries, (b) the most difficult-to-infer thought/feeling entries, and (c) the remaining
moderate difficult-to-infer thought/feeling entries. Second, the empathic accuracy
scores for each videotape were calculated for all thought/feeling entries belonging
to (a) the most abstract thought/feeling entries, (b) the most concrete
thought/feeling entries, and (c) the moderate abstract thoughts and feelings. Third,
we calculated the empathic accuracy scores for all thought/feeling entries judged by
the targets as (a) negative, (b) neutral, and (c) positive in affective tone. Finally, we
computed for each videotape the empathic accuracy scores for all the
95
thought/feeling entries with the content of the thought/feeling entries whether or

not to be congruent with (a) the targets verbal communication, and (b) with the
targets direction of gaze.
RESULTS
Empathic Accuracy: Content Accuracy
The MANOVA testing for between-group differences on both videotapes
revealed that the content accuracy was significantly different for both groups, F(2,
57) = 6.92, p < .005. Consistent with our expectations, the subjects with PDD were
found to make less accurate inferences of the thoughts and feelings of the targets in
video 1, F(1, 58) = 13.69, p < .001. The mean empathic accuracy of the subjects
with PDD was 18.36% (SD = 12.41) and the mean empathic accuracy of the control
subjects was 31.98% (SD = 15.89). No significant between-group difference was
found on video 2, F(1, 56) = 2.63. In video 2, the mean empathic accuracy of the
PDD group was 23.02% (SD = 15.86) and the mean empathic accuracy of the
control group was 29.12% (SD = 13.11).
We further analysed the more refined empathic accuracy scores. The mean
empathic accuracy scores of the thought/feeling entries belonging to the most easyto-infer thought/feeling entries, the most difficult-to-infer thought/feeling entries
and the moderately difficult-to-infer thought/feeling entries are shown in Table 2,
represented separately for both videotapes and for each group. For each videotape,
a 2 (Group: PDD versus Control) x 3 (Difficulty: Easy, Moderate and Difficult)
ANOVA was conducted on the empathic accuracy scores, with Group as betweensubject factor and Difficulty as within-subject factor. Confirming our prior analysis,
a significant main effect for Group was found for video 1, F(1, 58) = 12.89, p =
.001, but not for video 2, F(1, 58) = 2.71. Furthermore, a significant main effect for
Difficulty was found for video 1, F(2, 57) = 49.88, p < .001, and for video 2, F(2,
57) = 35.54, p < .001.
Chapter 4
96
Table 2
Empathic accuracy along the nature of the thought/feeling entries (in percentages)
PDD
Level of
Difficulty
Video 1
Unstructured
Video 2
Structured
Level of
Abstractness
Video 1
Unstructured
Video 2
Structured
Level of
Affective Tone
Video 1
Unstructured
Video 2
Structured
Control
Easy
Moderate
Difficult
Easy
Moderate
Difficult
(SD)
(SD)
(SD)
(SD)
(SD)
(SD)
34.67
15.52
7.71
54.86
29.33
14.38
(23.05)
(15.00)
(17.15)
(28.70)
(19.21)
(14.71)
31.43
25.69
11.33
42.38
30.10
15.92
(27.79)
(17.04)
(19.68)
(25.16)
(16.20)
(16.91)
Concrete
Moderate
Abstract
Concrete
Moderate
Abstract
(SD)
(SD)
(SD)
(SD)
(SD)
(SD)
26.22
17.59
14.22
33.44
34.46
27.41
(21.49)
(16.56)
(16.88)
(27.19)
(19.56)
(24.16)
35.81
21.96
6.40
38.76
30.04
11.87
(25.60)
(16.13)
(14.58)
(24.80)
(13.77)
(22.15)
Negative
Neutral
Positive
Negative
Neutral
Positive
(SD)
(SD)
(SD)
(SD)
(SD)
(SD)
19.63
21.70
14.20
44.67
26.52
25.47
(15.59)
(16.08)
(20.08)
(22.72)
(19.23)
(16.80)
21.56
23.67
23.76
30.81
30.67
26.61
(18.67)
(19.19)
(22.11)
(24.58)
(22.21)
(13.68)
Follow-up analyses revealed that the most easy-to-infer thought/feeling entries

were more accurately inferred than the moderate difficult-to-infer thought/feeling
entries and that the moderate difficult-to-infer thoughts and feelings were more
accurately inferred than the most difficult-to-infer thoughts and feelings (see Table
3). The interaction between Group and Difficulty yielded not significant for video 1
and video 2, with F(2, 57) = 2.22 and F(2, 57) < 1 respectively, indicating that
both groups have the same tendency to be less accurate in inferring more difficult
97
Table 3
Follow-up analyses (F-values) along the nature of the thought/feeling entries
Video 1: Unstructured
Source
df
Video 2: Structured
df
Thoughts and feelings

Difficulty
Easy versus Moderate
1,58
43.42 ***
1,58
6.57 **
Moderate versus Difficult
1,58
19.40 ***
1,58
24.51 ***
Easy versus Difficult
1,58
101.00 ***
1,58
63.72 ***
Abstractness
Concrete versus Moderate
1,58
1.26
1,58
14.12 ***
Moderate versus Abstract
1,58
2.78
1,58
37.53 ***
Concrete versus Abstract
1,58
5.48 *
1,58
73.06 ***
Negative versus Neutral
1,58
9.61 ***
1,58
.09
Neutral versus Positive
1,58
2.31
1,58
.35
Negative versus Positive
1,58
1,58
.12
Affective Tone
22.72 ***
*p < .05, **p < .01, ***p < .005
Similarly, Table 2 presents for each group the mean empathic accuracy scores of
the thoughts and feelings belonging to one of the three categories of abstractness.
For each videotape, a 2 (Group: PDD versus Control) x 3 (Abstractness: Concrete,
Moderate and Abstract) ANOVA was conducted on the empathic accuracy scores,
with Group as between-subject factor and Difficulty as within-subject factor. As
expected, the analysis revealed a significant main effect for Group for video 1, F(1,
58) = 13.47, p < .001, but not for video 2, F(1, 58) = 2.46. No significant main
effect for Abstractness was found for video 1, F(2, 57) < 1. However, a significant
main effect for Abstractness was found for video 2, F(2, 57) = 33.30, p < .001.
Follow-up analyses revealed that the most concrete thought/feeling entries were
more accurately inferred than the moderate abstract thought/feeling entries and that
the moderate abstract thoughts and feelings were more accurately inferred than the
most abstract thoughts and feelings (see Table 3). The interaction between Group
Chapter 4
98
and Abstractness was not significant for video 1 (F(2, 57) < 1), and for video 2
(F(2, 57) < 1).
Furthermore, the mean empathic accuracy scores of the thought/feeling entries
judged by the targets as negative, neutral and positive in affective tone are shown in
Table 2, presented separately for both videotapes and for each group. For each
videotape, a 2 (Group: PDD versus Control) x 3 (Affective Tone: Negative, Neutral
and Positive) repeated measure (ANOVA) was conducted, with Group as betweensubject factor and Affective Tone as within-subject factor. The analysis revealed
that video 1 distinguished between both groups, F(1, 58) = 13.85, p = .001, whereas
video 2 did not, F(1, 58) = 2.85. Furthermore, a significant main effect of Affective
Tone was found for video 1, F(2, 57) = 11.59, p = .001. Follow-up analyses
indicated that negative thought/feeling entries were better inferred than neutral
thought/feeling entries, and that negative thoughts and feelings were better inferred
than positive thoughts and feelings (see Table 3). No significant main effect of
Affective Tone was found for video 2, F(2, 57) < 1. Moreover, a significant Group
x Affective Tone interaction effect was found for video 1, F(2, 57) = 7.51, p = .001.
Paired t-tests suggest that subjects with PDD were slightly better at inferring
thoughts and feelings with a neutral tone than thoughts and feelings with a positive
tone, t(29) = 1.80, p = .08, whereas thoughts and feelings with a negative tone were
not significantly better inferred than those with a neutral tone, t(29) = -.57, or
positive tone, t(29) = 1.42. However, the control group inferred thoughts and
feelings with a negative tone significantly better than thoughts and feelings with a
neutral tone, t(29) = 4.95, p < .001, or a positive tone, t(29) = 5.51, whereas the
difference between the empathic accuracy belonging to thought/feeling entries with
a neutral versus positive tone was not significant, t(29) = .28. No significant Group
x Affective Tone interaction effect was found for video 2, F(2, 57) < 1.
Finally, the mean empathic accuracy scores of thought/feeling entries along the
congruence with the targets verbal communication and the targets gazes are
shown in Table 4, represented separately for both videotapes and each group. We
conducted two 2 (Group: PDD versus Control) x 2 (Congruence: Congruent versus
Non-Congruent) repeated measures (ANOVA) for each videotape, with Group as
between-subject factor and Congruence as within-subjects factor. The detailed
analyses displayed in Table 5 confirm our previous findings and revealed no
99
significant between-group differences for empathic accuracy in video 2, whereas

significant between-group differences were found for both empathic accuracy
scores in video 1, with control subjects reaching higher empathic accuracy than
subjects with PDD. As shown in Table 5, the analyses revealed 2 significant main
effects of Congruence for both videotapes, indicating that the subjects reached
higher accuracy rates for thoughts and feelings that are congruent with the targets
verbal communication and the targets direction of gaze than those that are not
congruent with the targets communication and direction of gaze. No significant
interaction effects (Congruence X Group) were found for both videotapes (see
Table 5), but the main effects of Group and Congruence in video 1 were found to
be additive (p-value > .25). The additive effect suggests that the lowest empathic
accuracy among all other empathic accuracy rates was obtained by the PDD group
inferring thoughts and feelings that are not congruent with the targets
communication (see Table 4).
Table 4
Empathic accuracy for congruent and non-congruent thought/feeling entries (in percentages)
PDD
Congruent
M
(SD)
Control
Non-congruent
M
(SD)
Congruent
M
(SD)
Non-congruent
M
(SD)
Verbalization
28.75 (22.23)
20.67 (11.10)
42.08 (22.52)
35.27 (14.55)
Gazing
25.00 (16.41)
21.85 (13.68)
43.47 (20.12)
33.74 (16.06)
Verbalization
31.11 (17.65)
20.82 (17.46)
36.22 (13.70)
27.69 (13.93)
Gazing
32.47 (18.02)
22.93 (16.69)
39.00 (17.93)
28.52 (11.48)
Video 2: Structured
Chapter 4
100
Table 5
F-values for Group and Congruence effect
Source
df
Video 2: Structured
df
3.05
Verbalization
Between Subjects
Group
1,58
12.30 ***
1,58
Congruent Verbalization
1,58
8.65 **
1,58
Group x Congruence
1,58
.06 a
1,58
.16
1,58
17.86 ***
1,58
2.94
Congruent Gazing
1,58
7.19 **
1,58
Group x Congruence
1,58
1.88
1,58
Within Subjects
18.23 ***
Gazing
Between Subjects
Group
Within Subjects
19.65 ***
.04
*p < .05, **p < .01, ***p < .001

a
p > .25
Empathic Accuracy: Valence Accuracy

The mean valence accuracy of the subjects with PDD was 35.24% (SD = 8.48)
for video 1 and 36.90% (SD = 7.35) for video 2. The mean valence accuracy of the
control subjects was 34.40% (SD = 10.12) for video 1 and 39.52% (SD = 8.54) for
video 2. In order to exclude the possibility that the subjects made a lucky guess
about the affective tone of the thoughts and feelings and were correct on a few rare
occasions because of chance alone (Ickes, Stinson, et al., 1990), we conducted for
each videotape and for each group a one-sample t-test with 33.33% as test value. In
the first videotape, the valence accuracy of both groups did not significantly differ
from 33.33%, with t(29) = 1.23 for the PDD group and t(29) = .58 for the control
group, indicating that the valence accuracy scores were not significantly different
from coincidence. In the second videotape, significant differences were found for
the group with PDD, t(29) = 2.66, p = .01 and for the control group, t(29) = 3.97, p
< .0001. An ANOVA with the valence accuracy scores of video 2 as dependent
101
variable and group as factor, revealed no between-group differences, F(1, 59) =

1.62.
Time required to complete the Empathic Accuracy Task

We administered the time required to infer and write down the targets
thought/feeling entries. The ANOVA testing revealed that the between-group
difference in time required to infer the thought/feeling entries of the targets of
video 1 approached significance, F(1, 58) = 3.14, p = .08. The average time to infer
the thought/feeling entries of the targets of video 1 was 28 min 31 sec for the
subjects with PDD and 25 min for the control group.
Similarly, an ANOVA revealed that the between-group difference in time
required to infer the thought/feeling entries of the targets of video 2 was nearly
significant, F(1, 58) = 3.69, p = .06. The average time to infer the thought/feeling
entries of the targets of video 2 was 30 min 32 sec for the subjects with PDD and
25 min 54 sec for the control subjects.
Since the Processing Speed (WAIS-III) was significantly different for both
groups, we conducted two ANCOVAs with the time needed to make inferences
for both videotapes as dependent variables, Group as between-subject factor and
Processing Speed as covariate. Analyses revealed significant between-group
differences for video 1, F(2, 57) = 9.40, p < .05, and for video 2, F(2, 57) = 9.31, p
< .05.
Correlates of the Empathic Accuracy Task

The correlations and intercorrelations between the content empathic accuracy
scores, the valence accuracy scores and the total IQ scores were calculated for both
groups (see Table 6). A significant positive correlation (Pearson) was found
between the content accuracy score on video 1 and that on video 2 for the control
group, but not for the group with PDD. Furthermore, we found a significant
positive correlation between the total IQ scores of the control adults and their
content accuracy scores on video 1, and between the total IQ scores of the controls
and their valence accuracy scores on video 2. No other significant correlations were
found.
Chapter 4
102
Table 6
Content Accuracy
Valence Accuracy
Video 1:
Video 2:
Video 1:
Video 2:
Unstructured
Structured
Unstructured
Structured
PDD
.30
.24
-.15
.00
Control
.53 **
.09
.43 *
.07
PDD
.17
.16
.25
Control
.38 *
.17
-.02
PDD
.12
.14
Control
.19
-.07
Total IQ
Content Accuracy: Video 1
Content Accuracy: Video 2
Valence Accuracy: Video 1

PDD
-.01
Control
-.15
*p < .05, **p < .01
We further used the formula of Hays3 (1994) to transform the Pearson

correlations into z-scores. The transformations revealed that the strength of the
associations were equally strong in both groups, with the exception of the
significant association between Total IQ and valence accuracy on Video 1 (z =
2.24, p < .05).
DISCUSSION
This study investigated the empathic accuracy of subjects with PDD when
confronted with two conversations between two strangers: a highly structured
conversation and a less structured or more chaotic one. The results suggest that
structure does matter for the mind-reading abilities of subjects with PDD, in that
z1 z2
3
1
1
+
n1 3 n2 3
103
way that mind-reading differences between young adults with PDD and typically
developing controls were more pronounced when subjects had to infer thoughts and
feelings of other persons in the less structured conversation.
Since the mind-reading impairments of the subjects with PDD were only
apparent in the less structured conversation, we tried to understand more fully these
findings by studying mind-reading along variables such as difficulty, abstractness,
and affective tone of the videotaped strangers thoughts and feelings, the
videotaped persons verbalisation and their direction of gaze. These more refined
analyses revealed that young adults with PDD use, at least to some extent, the same
strategies as typically developing controls to infer other peoples thoughts and
feelings. Support for this comes from the absence of any interaction effect between
group and the variables discussed, with the exception of the Group and Affective
Tone interaction effect for video 1. The latter interaction effect suggests that adults
with PDD were best in inferring when the affective content of the conversation was
neutral, whereas control persons were better able to infer when the target
individuals expressed negative affect. One possibility for this is that the more
disturbing affective cues might help clarify others inner states for typically
developing subjects, whereas subjects with PDD may find expressions of negative
or positive affect more distracting than helpful. Yet, several main effects were
indicating that in both videotapes the empathic accuracy scores of all subjects
decreased with an increasing level of difficulty. In the more structured
conversation, the empathic accuracy scores of the subjects decreased with an
increasing level of abstractness, whereas in the more chaotic conversation, the most
concrete thoughts and feelings were better inferred than the most abstract thoughts
and feelings. These results agree with the notion that subjects with PDD have
greater difficulty with symbolic or abstract language concepts than with concrete
facts and descriptions (Mesibov & Shea, 2001). However, it should be noted that
this is also the case for the control group. With regard to congruence, in both
videotapes it was found that thoughts and feelings that are congruent with the
targets communication or the targets eye gazes are better inferred than
incongruent thoughts and feelings. Words and eye gazes are just two ways to show
our intentions. Other non-verbal actions (e.g. body movements, smiles) can speak
as loud as words because they too can act as clues to our intended meaning (Happ,
104
Chapter 4
1994b). The fact that the inference ability of both groups was better when the
verbalisations and eye gazes supported the targets covert thoughts and feelings,
suggests that young adults with PDD are able to recognise, at least to some part, the
supportive character of these variables. None of the above-mentioned variables are
thus able to explain the differences in mind-reading performance between adults
with PDD and control adults in the more chaotic conversation.
With regard to the valence accuracy, we first calculated whether or not the
scores of the participants differed from chance. The analyses revealed that the
valence accuracy scores of both groups were only significantly different from
chance in the more structured videotape. One possibility is that subjects have more
difficulties in the more chaotic conversation with inferring the affective tone of
other persons thoughts and feelings, which might have given them cause to guess.
The associations between valence accuracy on Video 1 and Total IQ were not
equally strong in both groups, which is probably caused by coincidence.
In sum, this study demonstrated the significance of structure to the empathic
accuracy of young adults with PDD and showed that the empathic accuracy design
offers opportunities to test more refined hypotheses with regard to the on line
mind-reading abilities of subjects with PDD. Furthermore, the findings might
explain why some high-functioning subjects with PDD are able to pass through
mind-reading tasks, although they have severe social-functioning problems in daily
life. While laboratory tasks provide clear structure, daily life situations are
characterized by circumstances beyond our control. One implication is that, if
mind-reading difficulties of subjects with PDD become mainly apparent in less
structured situations, then it seems more appropriate to develop mind-reading tasks
in which the structure is reduced, so that even more subtle mind-reading
impairments can be detected.
It should be noted that the mind-reading performance of normally intelligent
individuals with PDD is subject to disparity in age (Happ, 1995; Yirmiya,
Solomonica-Levi, Shulman, & Pilowsky, 1996). Several studies have demonstrated
that the majority of children with PDD fail to pass traditional as well as more
advanced mind-reading tasks (Baron-Cohen et al., 2000). However, the greater part
of high-functioning adults with PDD seems to have no difficulties with traditional
mind-reading tasks. Their mind-reading impairments become mainly apparent with
105
the use of more advanced or naturalistic mind-reading tasks (Kleinman et al., 2001;
Ponnet et al., 2004; Roeyers et al., 2001). Therefore, it can be argued that adults
with PDD have more subtle mind-reading impairments than young children with
PDD. It is unclear whether adults with PDD have developed compensatory
strategies to camouflage their mind-reading impairments or whether their mindreading skills improved spontaneously with growing age. One way to find out is to
administer traditional and more naturalistic mind-reading tasks from teenagers with
PDD. It has been suggested that teenagers with PDD have no problems with
traditional mind-reading tasks (Dahlgren & Trillingsgaard, 1996). However, since
little research has been done with this population, it has to be seen how they
perform on more naturalistic mind-reading tasks. One might suggest that the use of
naturalistic tasks would clarify even more the mind-reading impairments in
teenagers with PDD than those in adults with PDD.
A limitation of the present study is that the Empathic Accuracy Task can not be
considered as a highly naturalistic real-life task, at best the task is a highly dynamic
one. Both videotapes were manipulated so that one of the tapes was more structured
than the other. This manipulation was needed to test the hypothesis, but also
restricts the naturalistic character of the task. Furthermore, inherent to the
methodology of the empathic accuracy design, subjects are permitted to use as
much time as needed to infer the thoughts and feelings of other persons. Consistent
with Ponnet et al. (2004), the results suggest that adults with PDD need more time
than typically developing adults to infer other persons thoughts and feelings.
However, in real life, thoughts and feelings have to be inferred hic et nunc. In a
way, this implies that the design does not mirror normal life.
The question remains which variables can account for the mind-reading
differences between subjects with PDD and typically developing subjects. The
correlations suggest that the various indices of the content empathic accuracy are,
with a single exception, independent of each other. The observed correlation
suggests that only 15% of variability in performance is shared across both
videotape tasks. What this pattern of results might suggest is that performance on
both videotapes is not related to the ability of individuals: even within the control
condition, subjects who achieve higher scores on one videotape are not that likely
to achieve higher scores on the other videotape. Therefore, one might argue that
106
Chapter 4
achievement is largely task-dependent, and that characteristics other than difficulty,

abstractness, congruence and so on limit the performance on one videotape and
affect performance differently on the other videotape. However, it should be noted
that even studies with typically developing individuals still failed to identify
reliable and replicable individual-difference correlates of good versus poor
perceivers (Ickes, Buysse, et al., 2000). One of the problems in the present study is
that we do not know to what specific condition we are able to attribute the fact that
there is minimal evidence of any one individual differences dimension being
measured by both videotapes. In a way, this means that we are not able to attribute
the between-group differences in performance to group differences in some ability
that affects performance on both tapes.
In future empathic accuracy studies, it might be interesting to sample more
interactions, so that raters have to judge a larger number of targets, and to analyse
the relative contribution of perceiver, target and relationship effects to the empathic
accuracy scores, for instance using the Social Relations Model (see Kenny, 1994).
Furthermore, it might be interesting to explore the significance of subjects with
PDD personal variables to their mind-reading performance. It seems plausible that
the mind-reading performance of subjects with PDD is related to variables such as
the severity of subjects social-communication disabilities, their motivation, or the
subjects perception of their own mind-reading abilities. Therefore, the use of
multiple instruments at the same time may contribute towards a better
understanding of the mind-reading impairments in subjects with PDD.
REFERENCES
disorders (3rd ed., rev.). Washington, DC: Author.
disorders (4th ed., rev.). Washington, DC: Author.
Bailey, A., Phillips, W., & Rutter, M. (1996). Autism: Towards an integration of clinical,
genetic, neuropsychological, and neurobiological perspectives. Journal of Child
107
Baron-Cohen, S. (1995). Mindblindness: An essay on autism and theory of mind.

Cambridge, MA: MIT Press/Bradford Books.
Baron-Cohen, S., Leslie, A. M., & Frith, U. (1985). Does the autistic child have a theory
of mind ? Cognition, 21, 37-46.
Bauminger, N., & Kasari, C. (1999). Brief report: Theory of mind in high-functioning
Bauminger, N., & Shulman, C. (2003). The development and maintenance of friendship in
high-functioning children with autism. Maternal perceptions. Autism, 7, 81-97.
Dahlgren, S., & Trillingsgaard, A. (1996). Theory of mind in non-retarded children with
autism and Aspergers syndrome. Journal of Child Psychology and Psychiatry, 37,
759-763.
Computers, 33, 159-166.
In Ickes, W. (Ed.), Empathic accuracy (pp.73-116). New York: Guilford Press.
Gesn, P.R., & Ickes, W. (1999). The development of meaning contexts for empathic
accuracy: Channel and sequence effects. Journal of Personality and Social Psychology,
77, 746-761.
Chapter 4
108
Happ, F. (1994a). An advanced test of theory of mind: Understanding of story characters

Happ, F. (1994b). Autism: An introduction to psychological theory. London: University
College London Press.
Happ, F. (1995). The role of age and verbal ability in the theory of mind task
Heavey, L., Phillips, W., Baron-Cohen, S., & Rutter, M. (2000). The awkward moment
test: A naturalistic measure of social understanding in autism. Journal of Autism and
Howlin, P. (1997). Autism: Preparing for Adulthood. Routledge: New York.
Ickes, W., Bissonnette, V., Garcia, S., & Stinson, L. L. (1990). Implementing and using
the dyadic interaction paradigm. In C. Hendrick & M.S. Clark (Eds.), Review of
psychology (Vol. 11, pp. 16-44). Newbury Park, CA: Sage.

Ickes, W., Buysse, A., Pham, H., Rivers, K., Erickson, J., Hancock, M., Kelleher, J., &
Gesn, P. (2000). On the difficulty of distinguishing good and poor perceivers: A
social relations analysis of empathic accuracy data. Personal Relationships, 7, 219-234.
59, 730-742.
Kenny, D. (1994). Interpersonal perception: A social relations analysis. New York:
Guilford Press.
29-36.
Mesibov, G. (1992). Treatment issues with high-functioning adolescents and adults with
autism. In E. Schopler and G. Mesibov (Eds.), High-functioning individuals with
autism. (pp.143-156). New York: Plenum Press.
109
Mesibov, G., & Shea, V. (2001). Understanding Asperger syndrome and high-functioning
autism. Netherlands: Kluwer Academic Publishers.
Perner, J., & Wimmer, H. (1985). John thinks that Mary thinks that Attribution of
second order beliefs by 5-10 year old children. Journal of Experimental Child
Rutterford, M. D., Baron-Cohen, S., & Wheelwright, S. (2002). Reading the mind in the
voice: A study with normal adults and adults with Asperger syndrome and highfunctioning autism. Journal of Autism and Developmental Disorders, 32, 189-194.
Simpson, J., Ickes, W., & Blackstone, T. (1995). When the head protects the heart:
Sodian, B., & Frith, U. (1992). Deception and sabotage in autistic, retarded and normal
children. Journal of Child Psychology and Psychiatry, 33, 591-605.
Wechsler, D. (2000). WAIS-III: Wechsler Adult Intelligence Test. Third Edition. Lisse:
Swets & Zeitlinger.
Yirmiya, N., Solomonica-Levi, D., Shulman, C., & Pilowsky, T. (1996). Theory of mind
abilities in individuals with autism, Down syndrome, and mental retardation of
unknown etiology: The role of age and intelligence. Journal of Child Psychology and
Psychiatry and Allied Disciplines, 37, 1003-1014.
110
Chapter 4
Zager, D. (1999). Autism: Identification, education, and treatment. Second Edition. New
Appendix A
The eight-point list to structure video 2
1. Whats your name ?
2. Where do you live ?
3. How do you come to be here ?
4. Where do you work or study ?
5. What are your leisure activities ?
6. How old are you ?
7. Do you have specific hobbies ?
8. What are our common interests ?
111
CHAPTER
Perceived and performed mind-reading

in adults with a pervasive developmental
disorder 1
This chapter is based on Ponnet, K., Roeyers, H., Buysse, A., & De Corte, K. (2004). Perceived and
performed mind-reading in adults with a pervasive developmental disorder. Manuscript submitted for
publication.
Chapter 5
114
ABSTRACT
This study investigated the perceived and performed mind-reading abilities of
adults with a pervasive developmental disorder and typically developing controls.
For this aim, the Interpersonal Reactivity Index (Davis, 1983) and the Emotional
Quotient Inventory (Bar-On, 1997a) were used as perceived mind-reading
assessments and the Revised Eyes Task (Baron-Cohen, Wheelwright, Hill, Raste, &
Plumb, 2001) as performance-based assessment.
The results suggested that subjects with PDD perceive their mind-reading as
more impaired than typically developing controls do. However, no between-group
differences were found on the performance-based Eyes Task. Furthermore, the
analyses revealed that the associations between the different mind-reading
assessments were equally strong in both groups.
On the one hand, the results count in favour of more advanced performancebased mind-reading assessments. On the other hand, the results demonstrate the
need for further studies in which perceived as well as performance-based mindreading assessments are used.
Perceived and performed mind-reading in adults with PDD
115
INTRODUCTION
People with autism or a related pervasive developmental disorder (PDD)
experience profound social deficits, which tend to persist over time. Even the most
able individuals with autism have difficulties in understanding and responding
appropriate to peoples feelings or emotions (Howlin, 1997; Lord, 1993; Zager,
1999). A psychological theory of autism that has been proposed to account for
these social problems is the theory of mind hypothesis. This hypothesis suggests
that the impairments of persons with PDD can be explained in terms of a failure to
attribute mental states to oneself and others (for an overview, see Baron-Cohen,
Tager-Flusberg, & Cohen, 2000). Without a theory of mind, people are unable to
develop a normal understanding that other people have mental states, which results
in socially inadequate behaviours. Theory of mind is also referred to as mindreading. Over the past decades, an increasing amount of research is devoted to the
development of mind-reading tasks for use with children with PDD (Bauminger &
Kasari, 1999; Happ, 1994; Sodian & Frith, 1992). These studies largely support
the theory of mind hypothesis. More recently, research begins to focus on the mindreading ability of adults with PDD. However, there are few tasks that can measure
if an adult with PDD has a mild deficit in mind-reading (Baron-Cohen,
Wheelwright, Hill, Raste, & Plumb, 2001; Mesibov & Handlan, 1997).
The aim of this study is to examine the mind-reading ability of young adults with
PDD compared to typically developing control adults. There appeared to be three
general approaches that could be used to assess this ability (Ickes, 2003; Mayer,
Caruso, & Salovay, 2001). In the first approach, people are simply asked to tell
how empathic they are. Data are obtained with self-administered questionnaires,
through personal interviews or diaries. Advantages of self-report measures are that
they are quick to administer and that they offer the most direct estimate of how
people experience their mind-reading abilities. Limitations of self-reports are that
they do not assess the actual (on-line) mind-reading ability, and therefore,
perceptual biases and distortions are possible. Furthermore, self-reports can give
cause to social desirability response biases, as some individuals tend to want to
appear socially appropriate in their behaviour, rather than abnormal and deviant
116
Chapter 5
(Canary, Cupach &, Messman, 1995). In the second approach, information is

obtained through other-report measures. The use of knowledgeable informants (e.g.
family, friends, teachers) yields information about how a person is perceived by
others (Mayer et al., 2001). Although it can be assumed that the second approach is
less subject to social desirability, the advantages and disadvantages of the second
approach are to a large extent the same as those in the first approach (Ickes, 2003).
On the one hand other-report measures are as quick and easy to administer as selfreport measures, on the other hand the informant approach can be subject to
distortion as it measures a persons reputation (i.e. the informants beliefs about the
others mind-reading abilities) (Mayer et al., 2001). In the third approach,
performance-based tasks are used. The main benefit of performance measures is
that they can be evaluated against objective, predetermined scoring criteria.
Whereas the third approach assesses the mind-reading performance, the other two
approaches assess the perceived mind-reading (Ciarrochi, Chan, Caputi, & Roberts,
2001). As people may not always have an accurate understanding of their own or
others mind-reading abilities, questionnaire measures consistently failed to predict
participants scores on performance-based tasks (Ickes, 2003). However, both
perceived and performed mind-reading abilities may be important predictors of
peoples social functioning in daily life.
In the present study, we relied on each of the above-mentioned approaches to
examine the mind-reading in adults with PDD. First, we used the Emotional
Quotient Inventory (EQ-i), a self-report measure of emotional intelligence
developed by Bar-On (1997a). The relatively new term emotional intelligence
was first introduced in 1990 by Salovey and Mayer and was later popularised by
Goleman (1995). The concepts origins trace back to Thorndikes (1920) definition
of social intelligence. According to Bar-On (1997a, p.2), emotional intelligence
can be defined as an array of emotional, personal, and social abilities and skills
that influence ones ability to succeed in coping with environmental demands and
pressures. The key factors involved in this definition include intrapersonal
capacity, interpersonal skills, adaptability, stress management strategies, and
motivational and general mood factors (Bar-On, Brown, Kirkcaldy, & Thom,
2000). The EQ-i renders a total EQ score and 5 EQ composite scale scores (i.e. the
key factors stated above) comprising fifteen subscale scores (Bar-On, 2000).
117
Whereas emotional intelligence may be discussed as a general factor, the construct

is clearly considered by its advocates as a multidimensional one (McCallum &
Piper, 2000).
Second, the Interpersonal Reactivitiy Index (IRI) of Davis (1980) was used as a
joint self-report/other-report measure. The IRI is an individual difference measure
of empathy, based on a multidimensional approach (Davis, 1983). Rather than
treating empathy as a single unidimensional concept, the rationale underlying the
IRI is that empathy can best be considered as a set of constructs, related in that they
all concern responsivity to others, but are also clearly discernable from each other.
Two constructs, Perspective Taking and Fantasy, involve cognitive responses,
whereas two other constructs, Empathic Concern and Personal Distress, both reflect
emotional responses (Davis, 1983, 1994).
Third, we used the Revised Eyes Task of Baron-Cohen and colleagues (BaronCohen et al., 2001), a performance-based task in which subjects have to look at a
series of photographs of the eye-region of adult faces and have to make a forced
choice between four mental state terms that describe what the person might be
thinking. According to Baron-Cohen et al. (2001), the Revised Eyes Task can best
be described as an advanced adult test of theory of mind, although it should be
noted that the task only involves the attribution of the relevant mental state and not
the inference of the content of that mental state (see Flavell, 1992).
The hypothesis of the present study is that adults with PDD will be more
impaired than typically developing controls on each of the three perceived mindreading assessments and on the performed mind-reading assessment.
METHOD
Subjects
A total of 60 adults were included in the study: 30 subjects with PDD and 30
typically developing controls. Male/female ratio (22/8) was similar in each group.
The subjects with PDD were recruited with the help of parent associations,
diagnostic centres and schools. All subjects had been previously diagnosed by a
Chapter 5
118
multidisciplinary team of experienced clinicians according to established criteria

for autism, Asperger syndrome or PDD-NOS, as specified in DSM-III-r or DSMIV (APA, 1987, 1994). The typically developing subjects were matched with the
subjects with PDD on a one-to-one basis on gender, chronological age and IQ.
Exclusion criteria for both groups were an estimated Verbal IQ below 75. The
intelligence level of the subjects was estimated using 11 of the 14 subtests of the
WAIS-III (Wechsler, 2000). The three not-administered subtests were Picture
Arrangement, Object Assembly and Comprehension.
Further background
information for both groups is presented in Table 1.

Table 1
PDD
Control
F(1, 59)
SD
SD
Chronological age
19.73
3.69
19.64
3.13
.01
Verbal IQ
98.13
13.69
95.57
10.29
.67
Performance IQ
90.70
15.96
94.27
12.96
.90
Total IQ
94.47
13.89
94.40
11.81
.00
Materials
Emotional Quotient Inventory
The Emotional Quotient Inventory (EQ-i) comprises 133 items, which have to be
scored on a 5-point scale ranging from very seldom or not true of me to very
often true of me or true of me (Bar-On, 1997a). The inventory renders a Total EQ
score, five composite scale scores, fifteen EQ subscale scores and four validity
scores. The definitions of the composite scales and EQ subscales, as defined by
Bar-On and colleagues (Bar-On, 1997a; Bar-On, Brown, Kirkcaldy, & Thom,
2000), are presented in Appendix A. Subjects completed the Dutch translation of
the EQ-i (Derksen, 1998).
The scores of the EQ-i are computer generated. Raw scores are automatically
tabulated and converted to standard scores based on a mean of 100 and standard
deviations of 15. Subjects are excluded if one or more of the four validity scores
119
(i.e. Omission Rate, Inconsistency Index, Positive and Negative Impression Scales)
suggest that the results are invalid (see Bar-On, 1997a). More specifically, subjects
are excluded if (a) the Omission Rate is more than 5 %, (b) the Inconsistency Index
score is more than 12, or (c) when the Positive or (d) Negative Impression scales
scores are more than two standard deviations above the mean (i.e. greater than
130). In the present study, 12 subjects with PDD and 9 controls were rejected on
the basis of these criteria.
Several studies provided support for the reliability and validity of the EQ-i (BarOn, 1997b; Dawda & Hart, 2000; Derksen, Kramer, & Katzko, 2002; Parker,
Taylor, & Bagby, 2001). For the present sample, the internal consistency
(Cronbachs alpha) was .63 for the PDD group and .76 for the control group.
Interpersonal Reactivity Index
The Interpersonal Reactivity Index (IRI; Davis, 1983, 1994) includes four sevenitem subscales, each of which aims to tap a separate aspect of the global concept of
empathy. The Perspective-taking scale (PT) measures the tendency to adopt the
points of views of others. The Fantasy scale (FS) measures the tendency to
transpose oneself into the feelings and actions of fictitious characters. The
Empathic Concern scale (EC) measures the tendency to experience other-oriented
feelings of warmth compassion and concern. The Personal Distress scale (PD) taps
ones own feelings of personal unease and discomfort in reaction to the emotions of
others. All of the items are scored on a 5-point Likert scale ranging from 0 (does
not describe me well) to 4 (describes me very well). The scores on the seven items
are summed up, so that the minimum and maximum score of each subscale can
vary between 0 and 28.
Subjects completed the Dutch translation of the IRI, which was achieved by
translation of the items from English to Dutch by four persons independently.
These translations were integrated into a provisional Dutch version of the IRI. This
version was back-translated by a bilingual native English speaker. Finally, in
consultation with the test author, a number of additional revisions were made.
The items of the IRIs other-report version are adapted to a third persons point
of view (e.g. I would describe him/her as a pretty soft-hearted person or In
emergency situations, he/she feels apprehensive and ill-at-ease).
120
Chapter 5
Research has demonstrated that the four factors are internally consistent and are
relatively independent of one another (Cliffordson, 2001; Davis, 1983, 1994;
Yarnold, Bryant, Nightingale, & Martin, 1996). For the present sample, the internal
consistency (Cronbachs alpha) of the self-report version was .86 for the subjects
with PDD and .87 for the control subjects. The Cronbachs alpha of the other-report
version was .80 for the acquaintances of the subjects with PDD and .76 for the
acquaintances of the control subjects.
Revised Eyes Test
The Revised Eyes Test is an authorised Dutch translated version2 of the
Reading the Mind in The Eyes Test-Revised Version (Baron-Cohen et al., 2001)
and consists of 36 photographs of the eye region of the faces of male and female
adults, together with a target mental state word and three foils. In order to be sure
that the participants understand the mental states, a glossary of the words is
provided (Baron-Cohen et al., 2001). In the present study the Revised Eyes Task
was computerized, so that we were able to record the response time in the subjects
judgements.
The computer version of the Revised Eyes Task was implemented by using
Inquisit (Inquisit 1.31, 2002). Each trial began with a glossary of four mental states.
Participants were encouraged to read attentively the particular meanings of the
words. By pushing the backspace, a blank interval of 850 ms followed. After that
interval, the photograph was presented in the middle of the computer screen and
was accompanied with the four mental state terms that were presented in the
corners of the screen. By pushing the 1, 3, 7, or 9 button of a numerical
keyboard, the correct response was registered. Each trial was separated by a blank
interval of 1450 ms. A practice session with other material ensured that the
participants understood the procedure before performing the computerised Revised
Eyes Test.
2
The Dutch translation of the Revised Eyes Task was achieved by translating the items and glossary from
English to Dutch by four persons independently. These translations were integrated into a provisional Dutch
version of the Revised Eyes Task. Then, this provisional version was back-translated by a bilingual native
English speaker. Back-translated words that differed from the original version were, at recommendation of
the test author (Baron-Cohen), piloted on a group of eight judges (4 female/4 male), conform the criterion of
the study of Baron-Cohen et al. (2001).
121
Subjects obtained 1 point for each correct answer, so that the minimum and
maximum scores of each subject could vary between 0 and 36. The program
Inquisit (Inquisit 1.31, 2002) allowed us to record the subjects reaction times. For
each subject, we calculated the time required to match the mental states with the
photographs and the time required to read the glossaries. Consequently, for each
subject four different reaction times could be calculated: (a) we calculated the mean
time to match correctly a mental state term with a photograph by dividing for each
subject the total reaction time of all correct matches through the number of correct
matches, (b) we calculated the mean time to match incorrectly a mental state term
with a photograph by dividing for each subject the total reaction time of all
incorrect matches through the number of incorrect matches, (c) the mean time
required to read the glossary of a correct match was obtained by dividing the total
time to read the glossary if a correct match was registered by the number of correct
matches, and (d) the mean time required to read the glossary of an in correct match
was calculated similarly.
Procedure
Subjects were tested individually in a laboratory room at Ghent University. The
IRI and the computerised Revised Eyes Task were administered first. Then,
subjects completed the EQ-i. Finally, the 11 subscales of the WAIS-III were
administered.
The other-report version of the IRI was completed by the mothers of the
participants. In case that the mothers could not be found, the other-report form was
administered from the participants father (1 subject with PDD/2 control subjects)
or the participants elder brother or sister (2 subjects with PDD/1 control subject).
RESULTS
Emotional Quotient Inventory
A series of ANOVAs were conducted for the Total EQ-i score, for each of the 5
composite scales and for each of the 15 subscales, with the scale as dependent
Chapter 5
122
variable and Group (PDD versus control) as factor. The mean scores and the Fvalues are shown in Table 2.
Table 2
Performance on the EQ-i
PDD (n = 18)
Control (n = 21)
SD
F(1,38)
95.86
10.93
5.67 *
9.82
95.95
14.43
.44
94.00
12.75
96.76
18.91
.28
Assertiveness
96.78
16.79
101.38
16.56
.74
Self-Regard
97.94
16.14
97.33
16.28
.01
101.33
10.47
94.38
12.14
3.60
91.28
13.08
94.43
14.80
.49
80.61
20.82
95.95
14.43
6.72 **
Empathy
79.72
15.75
96.24
16.10
10.40 **
Interpersonal Relationship
89.44
19.68
104.71
15.82
7.21 **
Social Responsibility
75.83
20.99
87.90
21.88
3.06
Adaptibility Scale
86.44
12.57
91.95
13.63
1.70
Problem Solving
85.11
13.42
84.05
16.86
.05
Reality Testing
93.06
14.57
92.95
13.20
.00
Flexibility
90.28
15.76
103.40
13.03
91.89
15.39
94.62
11.19
.41
Stress Tolerance
92.00
12.13
97.29
13.83
1.58
Impulse Control
94.89
19.32
94.00
14.98
.03
96.56
13.01
103.33
14.47
2.33
Happiness
100.44
14.92
106.48
12.38
1.90
Optimism
92.83
10.59
98.52
15.40
1.75
SD
Total EQ-i
87.83
9.95
Intrapersonal Scale
93.28
Emotional Self-Awareness
Self-Actualisation
Independence
Interpersonal Scale
Stress Management Scale
General Mood Scale
8.08 **
*p < .05, **p < .01
Analyses revealed a significant between-group difference on the Total EQ-i,

suggesting that the control subjects perform generally better on the EQ-i than the
subjects with PDD. With regard to the composite scales, analyses revealed a
123
significant between-group difference on the Interpersonal Scale, indicating that

subjects with PDD reported to have lower interpersonal skills than control subjects.
None of the other composite scales were significant. With regard to the subscales,
significant between-group differences were found for Empathy and Interpersonal
Relationship, subscales that belong to the Interpersonal composite scale, and on
Flexibility, a subscale that belongs to the Adaptability composite scale. The
significant differences suggest that subjects with PDD perceived themselves less
able to understand and appreciate the feelings of other persons than control subjects
(Empathy), and that subjects with PDD perceived themselves less able than
controls to establish and maintain mutually satisfying relationships that are
characterised by emotional closeness, intimacy, and by giving and receiving
affection (Interpersonal Relationship). The significant between-group difference on
the subscale Flexibility indicates that adults with PDD perceived themselves less
able than control adults to adjust their thoughts, feelings, and behaviour to changing
situations and conditions. No other between-group differences were found.
Interpersonal Reactivity Index
For each subscale of the IRI a 2 (Informant: Self-Report versus Other-Report) x
2 (Group: PDD versus Control) ANOVA was conducted, with Informant as withinsubject factor and Group as between-subject factor. The performance scores and Fvalues of the ANOVAs are presented in Table 3 and Table 4 respectively.
Table 3
Performance on the IRI
PDD (n = 30)
Self-report
M
(SD)
Control (n = 30)
Other-report
M
(SD)
Self-report
Other-report
(SD)
(SD)
Perspective-taking
12.20 (4.48)
4.75 (4.30)
14.47 (3.98)
14.00
(4.82)
Fantasy
14.03 (6.40)
11.64 (6.55)
14.73 (5.77)
11.66
(5.20)
Empathic Concern
15.70 (6.42)
12.25 (5.64)
16.63 (4.29)
18.34
(4.64)
Personal Distress
14.07 (6.05)
17.75 (4.97)
11.63 (5.04)
11.66
(5.14)
Chapter 5
124
Table 4
F-values for Group and Informant effect
Perspective-taking
Fantasy
Group
Informant
Group x Informant
F(1, 59)
F(1, 59)
F(1, 59)
34.26 ***
.13
37.06 ***
29.69 ***
7.27 *
.21
10.12 **
Empathic Concern
9.25 **
.96
Personal Distress
18.11 ***
4.16
3.22
*p < .05, **p < .01, ***p < .001
Group Effects. The analysis revealed a significant main effect of Group on the
Perspective-taking scale, indicating that subjects with PDD rated themselves and
are rated by their parents to embrace less other persons psychological point of
view than control adults. Furthermore, a significant main effect of Group was found
on the Empathic Concern scale and the Personal Distress scale, suggesting that
PDD subjects perceived themselves and are perceived by their parents to
experience less regard and sympathy for anothers feeling than control subjects, and
that, independent of Informant, subjects with PDD are perceived to experience
more self-oriented feelings of distress and discomfort in response to the emotions
of others than typically developing controls. No significant main effect of Group
was found on the Fantasy scale.
Informant Effects. A significant main effect of Informant was found on the
Perspective-taking scale and the Fantasy scale, suggesting that subjects gave
themselves higher ratings on the Perspective-taking scale than their parent, and
indicating that, independent of Group, parents judged their child to experience less
feelings and actions in creative work than their child did. No significant main effect
of Informant was found on the Empathic Concern scale and the Personal Distress
scale.
Group x Informant Effects. A significant interaction effect between Group and
Informant on the Perspective-taking scale revealed that the PDD groups ratings
were more than twice as high than those of their parents (see Table 3), whereas the
mean scores of the control subjects and their parent were nearly the same.
Furthermore, a significant interaction between Group and Informant on the
125
Empathic Concern scale indicated that subjects with PDD gave themselves higher
ratings on the Empathic Concern scale than their parent, whereas control subjects
gave themselves lower ratings on the Empathic Concern scale than their parent. No
significant interaction effect between Group and Informant was found on the
Fantasy scale and the Personal Distress scale.
Revised Eyes Task
An univariate analysis of variance (ANOVA) revealed no significant betweengroup difference on the Revised Eyes Task, F(1, 59) = 2.13. The mean score of the
subjects with PDD was 22.43 (SD = 3.16) and the mean score of the control
subjects was 23.70 (SD = 3.58). As shown in Table 5, a series of univariate
analyses revealed no between-group differences with regard to the time required to
match the mental states with the photographs and the time required to read the
glossaries.
Table 5
Time required to match the photograph with the mental state and to read the glossaries (in
seconds)
PDD (n = 30)
Control (n = 30)
SD
SD
F(1,59)
Correct answers
7.87
2.99
7.10
2.29
1.25
Incorrect Answers
9.45
3.88
9.01
3.62
.21
Correct answers
9.31
5.79
7.84
5.38
1.04
Incorrect Answers
9.16
5.86
7.18
5.14
1.93
Time required to match the

photo with the mental state
Time required to read the

glossaries
Furthermore, for each group we correlated the task performance with the mean
time to correctly or incorrectly match the mental state terms with the photographs.
For the control group, no correlations were found between the Eyes Task
performance and the time required to correctly, r(30) = -.13, p =.49, or incorrectly,
Chapter 5
126
r(30) = -.05, p =.80, match the mental states with the photographs. However, we
found significant negative associations between the task performance of the group
with PDD and their time required to correctly, r(30) = -.49, p < .01, or incorrectly,
r(30) = -.38, p < .05, match the mental states with the photographs.
Other correlates of the perceived and performed mind-reading assessments
The correlations and intercorrelations between the mind-reading tasks and Total
IQ were calculated for both groups (see Table 6). For the group with PDD,
significant positive correlations (Pearson) were found between the Revised Eyes
Task and Total EQ-i, the Revised Eyes Task and Personal Distress (self-report
version), Empathic Concern (self-report version) and Perspective Taking (selfreport version), and Empathic Concern (self-report version) and Fantasy (selfreport version). The Empathic Concern (other-report version) associates
significantly with Perspective Taking (other-report version) and with Fantasy
(other-report version). Furthermore, several associations were found between the
subscales of the IRIs other-report version and those of the IRIs self-report version
(see Table 6).
For the control group, significant positive correlations were found between the
Revised Eyes Task and Total IQ, and the EQ-I and Empathic Concern (other-report
version). As shown in Table 6, several significant associations were found within
the different subscales of the IRIs self-report version, and between the subscales of
the IRIs self-report version and those of the IRIs other-report version.
correlations into z-scores. The transformations revealed that the strength of the
associations were equally strong in both groups.
z1 z2
3
1
1
+
n1 3 n2 3
Correlations between IQ, Total EQ-i, subscales of the IRI and the Revised Eyes Task
EQ-ia
Eyes
PT
IRI Self-report (SR)

FS
EC
PD
PT
IRI Other-report (OR)

FS
EC
.10
.04
.31
.09
-.06
-.20
.38*
.20
.36
.48**
-.06
.14
.46**
.21
.39*
-.29
.21
.16
-.16
-.07
.06
.08
.14
-.10
.61**
.55**
.45*
.64 **
.11
-.25
.12
.34
.29
-.06
127
TIQ
PDD
-.23
-.08
.34
.10
-.17
.20
.01
-.19
Control
.19
.42 *
-.02
.21
-.33
-.34
.27
-.19
Eyes
PDD
.48*
.11
.03
.27
.40*
.40*
.35
Control
.05
-.09
.06
-.02
.03
.08
.03
IRI Self-report
PT - PDD
.16
-.08
.36*
-.11
.32
.11
PT - Control
.20
.33
.30
-.32
.47**
.51**
FS - PDD
.04
.37*
.29
-.17
.28
FS - Control
.06
.09
-.03
.22
.24
EC - PDD
.21
.33
.28
.38 *
EC - Control
.32
.43 *
-.23
.34
PD - PDD
-.23
.48**
.39 *
PD - Control
-.26
-.27
-.16
IRI Other-report
PT - PDD
-.17
.36
PT - Control
.24
.28
FS - PDD
.03
FS - Control
.33
EC - PDD
-.37
EC - Control
.53**
PD - PDD
.05
PD - Control
-.13
*p < .05, **p < .01, a Correlations with the EQ-i were calculated on the basis of 18 subjects with PDD and 21 control subjects
PD
Table 6
128
Chapter 5
DISCUSSION
In the present study, we tested the mind-reading in high-functioning adults with
PDD and control adults using three perceived mind-reading assessments and one
performance-based assessment.
On the Emotional Quotient Inventory, the total EQ-i score of the adults with
PDD was significantly lower than that of the control adults. However, the total EQi score supplies only general information about the respondent, because it is
composed of a broad range of skills. Therefore, we can best evaluate the
respondents performance at deeper levels, based on examination of the composite
scores and subscales (Bar-On, 1997a). Analyses revealed that adults with PDD
rated themselves significantly lower than control adults on the Interpersonal
composite scale and on the subscales Empathy, Interpersonal Relationship and
Flexibility. This confirms that subjects with PDD perceived their mind-reading
ability significantly lower than control adults. Based on clinical accounts of the
social impairments of subjects with PDD (Grandin, 1995; Howlin, 1997), one
might expect that subjects would also rate themselves less on the other subscales.
However, the analyses revealed that this was not the case.
On the Interpersonal Reactivity Index, subjects with PDD rated themselves and
are rated by their parents significantly lower than control adults on the subscales
Perspective Taking and Empathic Concern, suggesting that they are seen to adopt
less others points of views and to experience less other-oriented feelings of
warmth, compassion and concern. Furthermore, we found that the ratings on the
Personal Distress scale were higher for the subjects with PDD than for the control
subjects, indicating that subjects with PDD perceived themselves and are perceived
by their parent to experience lower regard and sympathy for anothers feelings than
control subjects. The latter is consistent with the study of Davis (1983) with a
normal population, who reported that higher levels of Personal Distress are
associated with higher levels of social dysfunction and lower level of social
competence. Thus, except for the Fantasy scale, the results on both self-report and
other-report version of the IRI suggest that the perceived mind-reading is more
hampered in adults with PDD than in typically developing controls.
129
The results on the EQ-i and the self-report version of the IRI are in line with a
study of Baron-Cohen and colleagues (Baron-Cohen, Richler, Bisarya, Gurunathan,
& Wheelwright, 2003) in which was found that adults with PDD rated themselves
significantly lower than controls on the Empathizing Quotient (a self-report
empathy scale), and with Koning and Magill-Evans (2001) who suggested that
adolescents with PDD are aware of some of their social skills deficits as they rated
themselves significantly lower than controls on the Social Skills Rating System
(SSRS; Gresham & Elliot, 1990).
When we compared the IRI self-report version with the IRI other-report version,
the analyses revealed a considerable contrast between the ratings of the adults and
those of the parents. The results indicated that the PDD groups ratings on the
Perspective-taking scale were more than twice as high than those of their parent.
Furthermore, the subjects with PDD gave themselves higher ratings on the
Empathic Concern scale than their parent did, whereas the control subjects gave
themselves lower rating on the Empathic Concern scale than their parent did. Both
adults with PDD as well as typically developing adults rated themselves higher on
the Fantasy scale than their parent did.
It should be noted that the results on the perceived mind-reading assessments are
to some extent paradoxical. On the one hand, subjects with PDD perceived
themselves to be less able in mind-reading than typically developing adults,
although the higher ratings of their parents suggested that they still might
underestimate their impairments. On the other hand, the subjects with PDD do have
some mind-reading abilities given that they rated themselves lower on the
perceived mind-reading assessments.
On the Revised Eyes Task, we failed to find a difference between both groups.
On the one hand, this differs from previous studies in which subjects with PDD had
to match mental states with photographs of the eyes (Baron-Cohen, Jolliffe,
Mortimore, & Robertson, 1997; Baron-Cohen et al., 2001; Kleinman, Marciano, &
Ault, 2001). On the other hand, the present results are in line with other Eyes Task
studies in which no differences were found between high-functioning subjects with
PDD and control subjects (Ponnet, Roeyers, Buysse, De Clercq, & Van der
Heyden, 2004; Roeyers, Buysse, Ponnet, Pichal, 2001). A possible explanation for
130
Chapter 5
these inconclusive findings is that the Eyes Task has little ecological validity.
Indeed, it can hardly be argued that the use of static photographs is a naturalistic
way to evaluate the social understanding abilities of subjects with or without autism
(Kleinman et al., 2001; Ponnet et al., 2004). Consequently, it might be that the Eyes
Task is not subtle enough to measure more mild deficits of high-functioning adults
with PDD. One way to cope with this is through the development of more
naturalistic mind-reading tasks, in which dynamic stimuli (e.g. videomaterial) are
used. To our knowledge, the few studies that made use of videomaterial
consistently found mind-reading differences between subjects with PDD and
typically developing controls (Heavey, Phillips, Baron-Cohen, & Rutter, 2000;
Ponnet et al., 2004; Roeyers et al., 2001; Yirmiya, Sigman, Kasari, & Mundy,
1992).
The computerized Eyes Task also allowed us to record the time to match the
mental states with the photographs. No between-group differences were found in
the time required to match (correctly or incorrectly) the mental states with the
photographs. However, whereas no significant associations were found between the
control groups performance on the Eyes Task and their time to correctly or
incorrectly match the mental states with the photographs, this was not the case for
the subjects with PDD. Negative associations indicated that the more time subjects
with PDD needed to match the mental states with the photographs (either correct or
incorrect), the worse their performance on the Eyes Task was. It should be noted
that these findings do not fit in with the notion that subjects with PDD might use
other processes to deal with problems that are normally dealt with in the affective
and emotional domains, and that this use of other processes might result in a
slowing of responses (Bowler, 1992; Happ, 1995). In addition, the analysis
revealed an association between the Eyes Task and Total IQ in the control group
but not in the group with PDD, which suggests that intellectual abilities might more
advantage the task performance of control subjects than that of subjects with PDD.
In sum, the results on the Interpersonal Reactivity Index and the Emotional
Quotient Inventory revealed that subjects with PDD perceived their mind-reading
as more impaired than typically developing controls did. However, no betweengroup differences were found on the performance-based Eyes Task. On the one
hand, this might suggest that the Eyes Task is not subtle enough to measure mild
131
mind-reading deficits in adults with PDD. This counts in favour of more advanced
performance-based mind-reading assessments, such as the naturalistic Empathic
Accuracy Task (Ponnet et al., 2004; Roeyers et al., 2001). On the other hand, both
the fact that between-group differences were found using the perceived mindreading assessments and that the associations between the different mind-reading
assessments were equally strong in both groups, demonstrated the need for further
studies in which perceived as well as performance-based mind-reading assessments
are used.
REFERENCES
Baron-Cohen, S., Richler, J., Bisarya, D., Gurunathan, N., & Wheelwright, S. (2003). The
systemizing quotient: An investigation of adults with Asperger syndrome or highfunctioning autism, and normal sex differences. Philosophical Transactions of the
Royal Society of London Series B-Biological Sciences, 358, 361-374.
Bar-On, R. (1997a). BarOn Emotional Quotient Inventory: A measure of emotional
intelligence. Facilitators resource manual. Toronto: Multi Health Systems.
Bar-On, R. (1997b). BarOn Emotional Quotient Inventory: Technical manual. Toronto:
Multi Health Systems.
132
Chapter 5
Bar-On, R. (2000). Emotional and social intelligence: Insights from the emotional quotient
inventory. In R. Bar-On, J. D. Parker, & D. Goleman (Eds.), The handbook of
Emotional Intelligence (pp. 363-388). San Francisco: Josey-Bass Inc. Publishers.
Bar-On, R., Brown, J., Kirkcaldy, B., & Thom, E. (2000). Emotional expression and
implicational stress; an implication of the Emotional Quotient Inventory. Personality
and Individual Differences, 28, 1107-1118.
Bauminger, N., & Kasari, C. (1999). Brief report: Theory of mind in high-functioning
Canary, D., Cupach, W., & Messman, S. (1995). Relationship Conflict. Conflict in parentchild, frienship, and romantic relationships. London: Thousand Oaks.
Ciarrochi, J., Chan, A., Caputi, P., & Roberts, R. (2001). Measuring Emotional
Intelligence. In J. Ciarrochi, J. Forgas, & J. Mayer (Eds.), Emotional intelligence in
everyday life. (pp. 25-45). New York: Psychology Press.
Cliffordson, C. (2001). The hierarchical structure of empathy: Dimensional organization
and relations to social functioning. Scandinavian Journal of Psychology, 43, 49-59.
Davis, M. H. (1980). A multidimensional approach to individual differences in empathy.
JSAS Catalog of Selected Documents in Psychology, 10, 85.
Davis, M. H. (1983). Measuring individual differences in empathy: Evidence for a
multidimensional approach. Journal of Personality and Social Psychology, 44, 113126.
Davis, M. H. (1994). Empathy: A social Psychological Approach. Madison, WI: Brown
and Benchmark.
Dawda, D., & Hart, S. (2000). Assessing emotional intelligence: Reliability and validity of
the Bar-On emotional Quotient Inventory (EQ-i) in university students. Personality and
Individual Differences, 28, 797-812.
Derksen, J. (1998). EQ en IQ in Nederland. Nijmegen: Pen Test Publisher.
Derksen, J., Kramer, I., & Katzko, M. (2002). Does a self-report measure for emotional
intelligence assess something different than general intelligence? Personality and
Individual Differences, 32, 37-48.
Flavell, J. (1992). Perspectives on perspective taking. In H. Beilin & P. Pufall (Eds.),
Piagets theory: Prospects and possibilities (pp. 107-139). Hillsdale, NJ: Erlbaum.
133
Goleman, D. (1995). Emotional Intelligence. New York: Bantam Books.

Grandin, T. (1995). How people with autism think. In E. Schopler & G. Mesibov (Eds.),
Learning and cognition in autism (pp. 137-156). New York: Plenum Press.
Gresham, F., & Elliott, S. (1990). Social Skills Rating System Manual. Circle Pines, MN:
American Guidance Service.
Happ, F. (1995). Annotation: Current psychological theories of autism: The theory of
mind account and rival theories. Journal of Child Psychology and Psychiatry, 35, 215229.
Inquisit 1.31 [Computer software]. (2002). Seattle, WA: Millisecond Software.
Kleinman, J., Marciano, P., &
Ault, R. (2001). Advanced theory of mind in high-
functioning adults with autism. Journal of Autism and Developmental Disorders, 31,
29-36.
Lord, C. (1993) The complexity of social behaviour in autism. In S. Baron-Cohen, H.
Tager-Flusberg and D. Cohen (Eds.), Understanding other minds. (pp. 292-316). New
York: Oxford University Press.
Mayer, J., Caruso, D., & Salovay, P. (2001). Selecting a measure of emotional
intelligence. In R. Bar-On, J. D. Parker, & D. Goleman (Eds.), The handbook of
Emotional Intelligence (pp. 92-117). San Francisco: Josey-Bass Inc. Publishers.
134
Chapter 5
McCallum, M., & Piper, W. (2000). Psychological mindedness and emotional intelligence
In R. Bar-On, J. D. Parker, & D. Goleman (Eds.), The handbook of Emotional
Intelligence (pp. 118-135.). San Francisco: Josey-Bass Inc. Publishers.
Mesibov, G., & Handlan, S. (1997). Adolescents and adults with autism. In D. Cohen and
F. Volkmar (Eds.). Handbook of Autism and Pervasive Developmental Disorders.
Second Edition. (p.309-324). New York: John Wiley & Sons Inc.
Parker, J., Taylor, G., & Bagby, R. (2001). The relationship between emotional
intelligence and alexithymia. Personality and Individual Differences, 30, 107-115.
Salovey, P., & Mayer, J. (1990). Emotional intelligence. Imagination, cognition, and
personality, 9, 185-211.
Sodian, B., & Frith; U. (1992). Deception and sabotage in autistic, retarded and normal
children. Journal of Child Psychology and Psychiatry, 33, 591-605.
Thorndike, E. (1920). Intelligence and its uses. Harpers Magazine, 140, 227-235.
Swets & Zeitlinger.
Yarnold, P. R., Bryant, F. B., Nightingale, S. D., & Martin, G. J. (1996). Assessing
physician empathy using the Interpersonal Reactivity Index: A measurement model and
cross-sectional analysis. Psychology, Health, and Medicine, 1, 207-221.
Yirmiya, N., Sigman, M., Kasari, C., & Mundy, P. (1992). Empathy and cognition in highfunctioning children with autism. Child Development, 63, 150-160.
Zager, D. (1999) Autism: Identification, education, and treatment. Second Edition. New
135
Appendix A
Definitions of the 5 EQ-i composite scales and 15 subscales
1. Intrapersonal Scale. The ability to be aware and understand oneself, ones emotions and
to express ones feelings and ideas
Emotional self-awareness (ES).The ability to recognize and understand ones emotions
Assertiveness (AS). The ability to express feelings, beliefs, and thoughts, and to defend
ones rights in a non-destructive manner
Self-regard (SR). The ability to be aware of, understand, accept, and respect oneself
Self-actualization (SA). The ability to realise ones potential and to do what one wants
to do, enjoys doing, and can do
Independence (IN). The ability to be self-directed and self-controlled in ones thinking
and actions and to be free of emotional dependency
2. Interpersonal Scale. The ability to be aware of, understand and to appreciate others
feelings as well as to establish and maintain mutually satisfying and responsible
relationships with others
Empathy (EM). The ability to be aware of, understand, and appreciate the feelings of
others
Interpersonal relationship (IR). The ability to establish and maintain mutually
satisfying relationships that are characterized by emotional closeness, intimacy, and by
giving and receiving affection
Social responsibility (RE). The ability to demonstrate oneself as a cooperative,
contributing, and constructive member of ones social group
3. Adaptability Scale. The ability to verify ones feelings with object external cues and
accurately size up the immediate situation, flexibly to alter ones feelings and thoughts
with changing situations, and to solve personal and interpersonal problems
Problem solving (PS). The ability to identify and define personal and social problems
as well as to generate and implement potentially effective solutions
Reality testing (RT). The ability to assess the correspondence between what is internally
and subjectively experienced and what externally and objectively exists
136
Chapter 5
Flexibility (FL). The ability to adjust ones feelings, thoughts, and behavior to changing
situations and conditions
4. Stress Management Scale. The ability to cope with stress and to control strong
emotions
Stress tolerance (ST). The ability to withstand adverse events, stressful situations, and
strong emotions without falling apart by actively and positively coping with stress
Impulse control (IC). The ability to resist or delay an impulse, drive, or temptation to
act, and to control ones emotions
5. General Mood. The ability to be optimistic, to enjoy oneself and others, and to feel and
express positive feelings
Happiness (HA). The ability to feel satisfied with ones life, to enjoy oneself and
others, and to have fun and express positive emotions
Optimism (OP). The ability to look at the brighter side of life and to maintain a
positive attitude, even in face of adversity
CHAPTER
Schematic knowledge and levels of

processing in young adults with PDD1
This paper is based on Ponnet, K., Buysse, A., Roeyers, H., Peene, O., & De Clercq, A. (2004). Schematic
knowledge and levels of processing in young adults with a pervasive developmental disorder. Manuscript
submitted for publication.
Chapter 6
138
ABSTRACT
This study investigated the schematic knowledge and levels of processing (LOP)
in adults with a pervasive developmental disorder (PDD) and typically developing
adults. For this aim, two computerized LOP Tasks were developed. The
presupposition of a LOP Task is that peoples schemata consist of associative
networks in which several constructs are related to each other, and that these
constructs, and the way they are linked together, lead people in their encoding,
storage and retrieval of personally relevant information.
Within each LOP Task, two independent measures were used: a content measure
that focuses on recall and a reaction time measure. If subjects with PDD are
schema-driven, than (a) their schemata will guide recall so that relevant information
will be preferentially recalled over irrelevant information, and (b) priming one
construct from an associative network will facilitate the accessibility of the other
constructs from the same network.
The results on both measures suggest that adults with PDD are as schematic as
typically developing adults and show intact levels of processing effects.
Furthermore, a more qualitative approach of the data revealed no between-group
differences.
Levels of processing in young adults with PDD
139
INTRODUCTION
Autism is a developmental disorder typified by profound impairments in
reciprocal social relationships, communication, imagination, and by a restricted
range of interests and behaviours (APA, 1994; Wing, 1988). Besides these
impairments, several individuals with autism or a related pervasive developmental
disorder (PDD) also show characteristic strengths in certain areas (Frith, 2003). In
his initial report of autism, Kanner (1943) already spoke of islets of abilities. The
most frequently reported skills are those that have to do with low-level perceptual
and visuospatial processes (Frith & Happ, 1994; Shah & Frith, 1993). The
presence of such idiosyncratic abilities provides evidence that individuals with
PDD process information differently than individuals without PDD (Mottron,
Burack, Stauder, & Robaey, 1999; Mottron, Peretz, & Mnard, 2000).
Recent studies suggest that subjects with PDD are less top-down driven
compared to typically developing controls (Mottron & Burack, 2001; Ropar &
Mitchell, 2002). In the normal course of events, any significant mental task
involves both bottom-up and top-down processing (Ashcraft, 1994). When the
information comes up from the sensory receptors and is then assembled into a
whole, this is bottom-up or data-driven processing (Gillam, Hoffman, Marler, &
Wynn-Dancy, 2002). In bottom-up processing, little of our existing knowledge is of
any use, and we rely almost exclusively on the data itself. However, during their
life, people develop several cognitive structures or schemata that consist of
associative networks in which several constructs are related to each other
(Anderson, 1990). These constructs, and the way they are linked together, lead
people to encode, process, store and recall some types of information rather than
others (Baldwin, 1992; McClelland & Rumelhart, 1981). When a construct from an
associative network is primed, the accessibility of the other constructs from the
same network is facilitated (Higgins & Bargh, 1987). Furthermore, schemas allow
people to go beyond the information available in a situation and to fill in the gaps
with expected information (Baldwin, 1992; Farrar & Boyer-Pennington, 1999). On
the other hand, schemas can also distort memory (Bower, Black, & Turner, 1979;
Fiske, Haslam, & Fiske, 1991; Wyer & Radvansky, 1999).
140
Chapter 6
It seems that people with PDD are less influenced by prior knowledge and less
able to integrate information in context, due to difficulties in forming meaningful
connection between different stimuli (Happ, 2000; Ropar & Mitchell, 2001). For
instance, it has been demonstrated that subjects with PDD show impairments in the
use of context information when reading homographs in a sentence-processing task
(Frith & Snowling, 1983; Jolliffe & Baron-Cohen, 1999), show enhanced visual
search compared to typically developing controls (Mottron & Burack, 2001;
ORiordan, Plaisted, Driver, & Baron-Cohen, 2001), and do no not perceive visual
illusions to the same extent as controls (Happ, 1996). Furthermore, subjects with
PDD detect hidden figures more easily than control subjects within larger
meaningful drawings (Jolliffe & Baron-Cohen, 1997; Shah & Frith, 1993), and
their perception is less influenced by prior knowledge (Ropar & Mitchell, 2002).
However, the tendency of subjects with PDD to be less top-down driven has
mainly been demonstrated in tasks in which non-social cues are used. The area of
social knowledge has received less attention (Volden & Johnston, 1999). Since
people with PDD may be less top-down driven than typically developing subjects,
it seems plausible that they also have trouble with relational schemas. To date, very
little is known about the capabilities of subjects with PDD in their relational
schemas. While a recent study of Trillingsgaard (1999) with a small sample of
children with autism suggests that children with autism have fewer well-organized
scripts2 for familiar social routines (such as make a cake or celebrate a birthday)
than normal control children, the results of a study of Volden and Johnston (1999)
suggest that the basic scriptal knowledge of children with autism appears to be
intact. Furthermore, studies in which subjects with PDD have to remember
semantically related verbal material delivered inconclusive findings: on the one
hand subjects with PDD seem to be less susceptible to the influence of
associatively and semantically related words (e.g. Beversdorf et al., 2000; Bowler,
Matthews, & Gardner, 1997; Klinger & Dawson, 2001; Minshew & Goldstein,
1993; Tager-Flusberg, 1991); on the other hand several studies gave evidence that
Schemas are basic units of knowledge, generalizations about the world that form through experience.
Schemas about familiar sequences of activities are called scripts (Baldwin, 1999; Shank & Abelson, 1977).
Scripts describe particular behavioural sequences likely to occur in the situation (Baldwin, 1999).
141
the ability to connect items on the basis of meaning is intact in autism in both visual
and verbal domains (e.g. Brian & Brysan, 1996; Lopez & Leekam, 2003).
The aim of the present study is to compare the social information processing and
the schematic knowledge of adolescents and adults with PDD with those of
typically developing control subjects. Examples of scripts that have been studied in
typically developing people are going to a restaurant, a movie, or a grocery store,
getting acquainted with someone else, a birthday party, being in a hospital and
so on (Farrar & Goodman, 1990; Hudson, Fivush, & Kuebli, 1992; Trafimow &
Wyer, 1993;Welch-Ross & Smidt, 1996). In this study, we picked out two scripts
(i.e. a conversation of a getting acquainted type and a scientific experiment)3.
For both scripts, we developed a Levels of Processing (LOP) Task, an incidental
recall task that is based on the Depth of Processing paradigm (Craik & Tulving,
1975; Rogers, Kuiper, & Kirker, 1977; Rudolph, Hammen, & Burge, 1995). The
presupposition of the LOP task is that schemas drive attentional processes and
memory search, thereby facilitating the encoding, storage, and retrieval of
personally relevant information (Rudolph et al., 1995). Over the years, the Depth of
Processing approach has proved to be very robust (Richardson-Klavehn, Gardiner,
& Ramponi, 2002), and levels of processing effects are found in typically
developing children as well as in adults (Geis & Hall, 1976, 1978; Owings &
Baumeister, 1979). However, it is still inconclusive whether individuals with PDD
show the levels of processing effect (Toichi & Kamio, 2002; Toichi, Kamio, et al.,
2002).
In a LOP Task subjects are directed to encode words under different levels of
processing, e.g. under a semantic instruction (e.g. Is this word typical for a
scientific experiment?) or under a structural instruction (e.g. Is the word in
capital letters?). Half of the presented words are schema-relevant, and the other
half is schema-irrelevant. After completing the ratings, subjects are asked
unexpectedly to recall as many words as possible. The LOP task enables us to
assess subjects level of schematicity using two independent measures: a content
3
It can be suggested that one script is more familiar than the other. However, the difference between
prototypic and exemplar-based representations of knowledge is often difficult to determine (Barsalou, 1990),
and is beyond the scope of the present investigation. In this study, we are primarily concerned whether
subjects with PDD show intact levels of processing effects.
142
Chapter 6
measure and a speed measure. With regard to the first, the depth of processing
paradigm postulates that words encoded under the structural instruction are only
processed to a shallow level in memory, while the semantic encoded words are
subjected to more intentional and meaningful processing (Ashcraft, 1994).
Therefore, semantically encoded words will be better recalled than structurally
encoded words (Anderson, 1990). Furthermore, study after study has shown that
schema-relevant information is better recalled than schema-irrelevant information
(Baldwin, 1992). With regard to the latter measure, Craik and Tulving (1975)
suggested that deeper levels of processing require longer processing times.
Furthermore, according to the accessibility feature of the information processing
paradigm, priming one construct from an associative network facilitates the
accessibility of the other constructs from the same network (Swann & Read, 1981).
Therefore, we extended the methodology of previous LOP studies by
computerizing both LOP tasks. The advantage of digitised tasks is that accurate
reaction times can be registered.
In this study, we examined the presence of two scripts in adults with PDD. If
subjects with PDD have organised schemas that guide processing, then this will be
demonstrated by an interaction effect between Relevance and Instruction on the
content as well as the reaction time measure. With regard to the content measure,
the presence of schemas will be supported by an enhanced recall of semantically
encoded schema-relevant words over structurally encoded schema-relevant words,
and by an enhanced recall of semantically encoded schema-relevant words over
semantically encoded schema-irrelevant words. With regard to the recall measure,
larger reaction times will be registered for ratings made under the semantic
instruction versus ratings made under the structural instruction. Furthermore, we
explored the schematic knowledge of both groups on a more qualitative way.
143
METHOD
Subjects
present study: 30 subjects with PDD and 30 typically developing subjects.
Male/female ratio (22/8) was similar in each group. The subjects with PDD were
recruited with the help of parent associations, diagnostic centres and schools. All
subjects with PDD had been previously diagnosed by a multidisciplinary team of
experienced clinicians according to established criteria for autistic disorder,
Asperger syndrome or PDD-NOS, as specified in DSM-III-r or DSM-IV (APA,
1987, 1994). Only subjects with an IQ of at least 75 and a minimum chronological
age of 15 years were included in the sample. The typically developing control
subjects were matched meticulously on a one-to-one basis on sex, chronological
age and IQ. All subjects were administered 11 of the 14 subtests of the WAIS-III
(Wechsler, 2000). The three non-administered subtests were Picture Arrangement,
Object Assembly and Comprehension. Further background information for both
groups is presented in Table 1.
Table 1
PDD
Control
SD
SD
F(1, 59)
Chronological age
19.73
3.69
19.64
3.13
.01
Verbal IQ
98.13
13.69
95.57
10.29
.67
Performance IQ
90.70
15.96
94.27
12.96
.90
Total IQ
94.47
13.89
94.40
11.81
.00
Levels of Processing Task

Two Levels of Processing Tasks (LOP) were made to tap subjects organisation
and processing of their scriptal knowledge. The tasks were based on previous
studies (Rudolph et al., 1995) and were implemented using Inquisit (Inquisit 1.31,
2002), in order to register reaction times.
The first LOP Task allows the
144
Chapter 6
assessment of the presence and organization of subjects scriptal knowledge of a

dyadic conversation of the getting acquainted type. The second LOP Task was
made to assess subjects scriptal knowledge of a scientific experiment.
In each of the LOP Tasks, subjects were directed via instructions on a computer
screen to encode 40 constructs under two different instructions, reflecting different
levels of depth of processing: (a) structural instruction: Is the following word in
capital letters? and (b) semantic instruction: Is the following word typical for an
initial dyadic conversation of the getting acquainted type? (LOP 1) or Is the
following word typical for a scientific experiment? (LOP 2). Each question
remained on the screen for 3000 ms, followed by a blank interval of 500 ms. After
that interval, the construct was presented. Each trial was separated by a blank
interval of 2000 ms. The subjects had to answer with yes or no. Therefore, a
keyboard was placed in front of the subject. By pushing the 1 button of the
numerical keyboard (coloured green), a yes-response was registered; by pushing
the 3 button of the numerical keyboard (coloured red), a no-response was
registered. Although no time limit was imposed for the task, the subjects were
encouraged to work fast and steadily. The computer task was followed by an
incidental recall task in which subjects were asked to write down as many
constructs as they remembered.
The presented constructs in the first LOP Task were obtained as follows. Thirtyfive typically developing college students (15 male/20 female) between 20 and 25
years old were asked to write down as many words as possible that they considered
as highly typical for an initial conversation of the getting acquainted type between
two strangers. The time to write down the constructs was limited to 3 minutes.
Thereafter, each subject was asked to rank his written words in order of
significance. By dividing for each subject the rank number by the total number of
words, a rank proportion was obtained for each word. On the basis of frequency
and order of ranking, the 20 most relevant words were retained as items for the
LOP Task. For each of these consistent words, we sought an irrelevant word (i.e. a
word that does not refer to an acquaintanceship with two strangers) on the basis of
two criteria. First, it is known that memory for longer words is poorer than that for
shorter words (Baddeley, Thomson, & Buchanan, 1975), so the length of each
145
irrelevant word had to match with that of the relevant word. For each match, we
allowed a variance of one character4. Second, the irrelevant words were selected in
such a manner that they were comparable as stringent as possible for lexical
familiarity measured as the frequency with which a word occurs in printed text. For
Dutch words, such objective frequency norms have been provided by the Centre for
Lexical Information (Baayen, Piepenbrock, & Gulikers, 1995). The list of words is
presented in Appendix A.
To obtain the presented constructs in the second LOP Task, thirty-five college
students (18 male/12 female) between 20 and 25 years old were asked to write
down as many words as possible that they considered as highly typical for a
scientific experiment. The same procedure of the first LOP Task was used to
arrive at the list of constructs (see Appendix B).
For both LOP Tasks, two counterbalanced versions were developed. For each
version, 10 relevant and 10 irrelevant constructs were randomly presented under the
semantic instruction, and the same number of relevant and irrelevant constructs was
randomly presented under the structural instruction. Two constructs were added at
the beginning and at the end of each task, but were not included in the analyses to
minimise the primacy and recency memory effects.
For each subject, we added up all yes and no ratings, so that 4 recall scores were
obtained: (a) the recalled schema-relevant words encoded under the semantic
instruction, (b) the recalled schema-irrelevant words encoded under the semantic
instruction, (c) the recalled schema-relevant words encoded under the structural
instruction, and (d) the recalled schema-irrelevant words encoded under the
structural instruction. Each of these scores has a range of 0 to 10. The sum of the
four recall scores represents the total recall score.
Furthermore, the design allowed us to calculate the reaction times for each
subject. Four reaction time scores were calculated for each subject as functions of
instruction (semantic versus structural) and word relevance (relevant versus
irrelevant).
This criterium was successfully accomplished with the exception of the construct asking questions (see
Appendix A).
146
Chapter 6
Procedure
All subjects were individually tested in a quiet laboratory room and were
randomly assigned to either the first or second LOP Task, so that in each LOP Task
15 subjects with PDD and 15 controls participated.
RESULTS
Recall Scores on the LOP Tasks
An ANOVA was conducted with the total recall score as dependent variable and
Group (PDD versus control) and Task (LOP 1 versus LOP 2) as factor. The
ANOVA testing revealed that the interaction between Group and Task was not
significant, F(1, 59) < 1. Furthermore, the main effect of Group, F(1, 59) < 1, and
Task, F(1, 59) = 1.16, was not significant. In the first LOP Task, the mean recall of
the PDD group was 5.60 (SD = 4.15) and the mean recall of the control group was
5.20 (SD = 2.48). In the second LOP Task, the mean recall of the PDD group was
4.80 (SD = 2.34) and the mean recall of the control group was 4.47 (SD = 1.25).
A 2 (Relevance: yes versus no) x 2 (Instruction: semantic versus structural) x 2
(Group: PDD versus control) x 2 (Task: LOP 1 versus LOP 2) repeated measures
analysis of variance (ANOVA) was computed on the four recall scores, with
Relevance and Instruction as within-subjects factors and Group and Task as
between-subjects variables. The mean recall scores of both groups are presented in
Table 2.
The analysis revealed that the four-way interaction effect (Relevance x
Instruction x Group x Task) was not significant, F(1, 56) = 1.20. Furthermore, none
of the three-way interaction effects were significant. We found that the Group x
Relevance interaction (F(1,56) < 1), the Group x Instruction interaction (F(1,56) <
1), the Instruction x Task interaction (F(1, 56) = 3.62) and the Relevance x Task
interaction (F(1, 56) = 2.68) were not significant. However, the Relevance x
Instruction interaction was significant, F(1,56) = 9.36, p < .005, indicating that the
relevant words under the semantic instruction are better recalled than the relevant
words under the structural instruction, while the differences between recall of the
147
irrelevant words under the semantic instruction and recall of the irrelevant words
under the structural instruction are less pronounced. This confirms that subjects
have organised schemas of both a conversation of the getting acquainted type and
experimental situations. Furthermore, the analysis revealed a significant main
effect of Relevance, F(1, 56) = 20.91, p < .0001, indicating that subjects recalled
significantly more relevant words than irrelevant words. We further found a
significant main effect of Instruction, F(1, 56) = 123.29, p < .0001, indicating that
more words are recalled under the semantic instruction than under the structural
instruction. No significant main effect of Group, F(1, 56) < 1, and Task, F(1, 56) =
1.16, was found.
Reaction time scores on the LOP Tasks
Similarly to above, a 2 (Instruction: semantic versus structural) x 2 (Relevance:
yes versus no) x 2 (Group: PDD versus control) x 2 (Task: LOP 1 versus LOP 2)
repeated measures analysis of variance (ANOVA) was computed on the four
reaction time scores, with Instruction and Relevance as within-subjects factors and
Group and Task as between-subjects variables. The mean reaction time scores are
presented in Table 3.
The analysis revealed that the four-way interaction effect and the three-way
interaction effects were not significant. We further found that the Group x
Relevance interaction (F(1, 56) < 1), the Group x Instruction interaction (F(1, 56) <
1) and the Relevance x Task interaction (F(1, 56) < 1) were not significant.
However, the interaction between Instruction and Relevance was significant, F(1,
56) = 6.27, p < .05, which suggests that the irrelevant words under the semantic
instruction were rated faster than the relevant words under the semantic instruction,
while the differences between the ratings of the irrelevant words under the
structural instruction and the ratings of the relevant words under the structural
instruction were less pronounced. The Instruction x Task interaction was
significant, F(1, 56) = 5.47, p < .05, indicating that the second LOP Tasks words
under the semantic instruction were rated faster than the first LOP Tasks words
under the semantic instruction, while the differences between the ratings of the
second LOP Tasks words under the structural instruction and the ratings of the
first LOP Tasks words under the structural instruction were less pronounced.
148
Table 2
Recall scores on the LOP Tasks
Script 1:
Script 2:
Acquaintanceship
Scientific Experiment
PDD
Control
PDD
Total Scores
Control
PDD
Control
SD
SD
SD
SD
SD
SD
2.60
2.29
2.53
1.46
2.87
1.41
2.27
1.28
2.73
1.87
2.40
1.35
1.93
1.39
1.87
1.64
.87
.64
1.13
.91
1.40
1.19
1.50
1.36
.67
1.34
.47
.64
.60
.83
.73
1.16
.63
1.09
.60
.93
.40
.51
.33
.49
.47
.64
.33
.49
.43
.57
.33
.48
Recall under semantic instruction

Relevant words a
Irrelevant words
Recall under structural instruction

Relevant words a
Irrelevant words
a
min-max score: 0-10
Chapter 6
Reaction times (RT) of the constructs (in seconds)

Script 1:
Script 2:
Acquaintanceship
Scientific Experiment
PDD
Control
PDD
Total Reaction Times
Control
PDD
Control
SD
SD
SD
SD
SD
SD
Total RT - Relevant words
33.27
21.07
29.30
8.91
23.97
8.59
22.62
8.03
28.62
16.50
25.96
9.00
Total RT - Irrelevant words
27.83
13.46
24.73
5.40
19.38
6.03
21.09
9.96
23.60
11.11
22.91
8.09
Total RT - Relevant words
20.17
17.95
17.77
4.79
17.03
8.45
13.48
3.36
18.60
13.88
15.62
4.61
Total RT - Irrelevant words
18.74
12.08
15.67
3.98
15.26
4.47
13.54
3.23
17.00
9.12
14.61
3.72
RT under semantic instruction
RT under structural instruction
Table 3
149
150
Chapter 6
The analysis revealed a significant main effect of Instruction, F(1, 56) = 120.76,
p < .0001, indicating that ratings made under the structural instruction occurred
significantly faster than those made under the semantic instruction. A significant
main effect of Relevance was found, F(1, 56) = 10.86, p < .005, indicating that
subjects needed more time to rate relevant words than irrelevant words. A
significant main effect of Task was found, F(1, 56) = 5.09, p < .05, indicating that
subjects needed more time to rate the words of the first LOP Task than those of the
second LOP Task. No significant main effect of Group was found, F(1, 56) < 1.
Further analyses of schema-relevant and schema-irrelevant words
We analysed the degree to which the schema-relevant and schema-irrelevant
words were actually rated as schema-relevant and schema-irrelevant by both
groups. Table 4 presents for each group and for each script the mean scores of the
yes-rated schema-relevant and the yes-rated schema-irrelevant words. Furthermore,
we conducted two ANOVAs with the yes-rated schema-relevant and yes-rated
schema-irrelevant words as dependent variables, and Group and Script as factor. As
shown in Table 4, no significant effects were found.
Then, we pooled the data over the subjects for each group (PDD versus control)
and each condition (two versions for each LOP Task). Table 5 presents an
overview. For each word in each condition, we pitted the number of times the word
was recalled in the PDD group and the control group against the total number of
words recalled. As such, we conducted forty 2 (Group: PDD/Control) x 2 (Word:
recalled/not-recalled) 2 analyses. Only one out of the 40 2 analyses was
significant. More specifically, a between-group difference was found with regard to
the word friendly, suggesting that this word was significantly more recalled by
control subjects than by subjects with PDD (2(1) = 4.24, p < .05).
Yes-rated schema-relevant and yes-rated schemairrelevant words

Effects
Script 1
Script 2
PDD
Control
PDD
Control
Group
Script
Script x Group
M (SD)
M (SD)
M (SD)
M (SD)
Schema-relevant words
1, 56
.37
1.46
2.77
6.80 (2.45)
7.80 (1.42)
8.07 (1.58)
7.60 (1.05)
Schema-irrelevant words
1, 56
.33
1.07
.85
9.00 (1.13)
9.07 (1.58)
9.53 (0.74)
9.13 (0.83)
Table 4
151
152
Table 5
Recall (frequencies) of the relevant words under the semantic instruction
LOP 1 V1
PDDa
Controla
LOP 1 V2
PDDb
Controlb
LOP 2 V1
PDDb
Controlb
Name
Talk about
weather
Distant
Work
Silences
Personal
Superficial
Polite
Reticent
Ask
questions
1
3
4
4
2
0
2
3
0
2
1
0
0
1
1
1
1
0
2
4
Formal
Leisure
Friendly
Tense
Unsure
Studies
Ill at ease
Age
Place of
residence
Interest
1
0
2
0
1
5
1
1
4
3
3
3
4
2
3
2
5
3
1
1
Truth
Scientist
Test
subject
Tests
Control
Statistic
Evaluation
Reliable
Theory
Dull
5
4
0
1
6
5
3
0
1
1
4
2
1
1
0
2
0
1
2
2
Data
Repeatable
Hypothesis
Money
Artificial
Accurate
Laboratory
Objective
Guinea
pig
Structure
3
3
14
18
Total
25
20
Total
22
19
Total
21
15
Chapter 6
LOP 2 V2
PDDa
1
1
2
3
0
2
2
3
4
a
Control
0
1
0
0
0
1
4
4
2
LOP 1V1: LOP Task 1 (getting acquainted) Version 1; LOP 1V2: LOP Task 1 (getting acquainted) Version 2;
LOP 1V1: LOP Task 2 (experimental situation) Version 1; LOP 1V2: LOP Task 2 (experimental situation) Version 2
a
N = 7; b N = 8
Total
153
DISCUSSION
This study was conducted to examine the schematic knowledge and the presence
of two scripts in adults with PDD using two levels of processing tasks. Within each
task, two independent information-processing parameters were used: a content
measure and a reaction time measure.
Based on the content measure, no between-group differences were found. The
results revealed that both groups showed the level of processing effect: semantic
processing of meaningful verbal materials facilitated episodic memory better than
shallow processing. Furthermore, the results suggested that both groups had
organized schemas of social situations, such as a conversation of the getting
acquainted type and experimental situations. We further found that schemarelevant information was better recalled than schema-irrelevant information.
Analyses of the reaction time scores largely confirmed the results on the recall
measure: no between-group differences were found and both groups showed the
levels of processing effect. Furthermore, words under the structural instruction (i.e.
the shallow level) required less time to rate than those under the semantic
instruction. However, contrary to the findings on the recall measure, the analyses of
the reaction times revealed that relevant words required more time to rate than
irrelevant words.
Analyses of the reaction times further revealed that the words in the second LOP
task were rated faster than the words in the first LOP task. More specifically, we
found that the second LOP tasks words under the semantic instruction were rated
faster than the first LOP tasks words under the semantic instruction, while the
differences between the ratings of the second LOP tasks words under the structural
instruction and the ratings of the first lop tasks words under the structural
instruction were less pronounced. Although this result was not anticipated and is
not confirmed with the findings on the recall measure, it seems quite logical that
the subjects script of a scientific experiment was more activated than that of a
getting acquainted conversation, since the subjects were actively participating in a
laboratory experiment.
The findings of this study provide strong evidence for unimpaired levels of
processing effects in subjects with PDD, which is consistent with a LOP study of
154
Chapter 6
Toichi, Kamio, et al. (2002). However, it should be noted that the present LOP task
differed from the LOP task of Tochi, Kamio, et al. (2002) in several ways. First, the
task of Toichi and colleagues was an incidental recognition task, whereas the
present task was an incidental recall task. Second, Toichi, Kamio, et al (2002)
aimed to investigate self-consciousness in autism and therefore used three levels
(i.e. self-referent, semantic and phonological), whereas this study aimed to assess
the schematic knowledge of adults with PDD, using two levels (i.e. semantic versus
structural). While Toichi and colleagues found that subjects with PDD showed a
lack of self-reference effects despite intact levels of processing effects, we found
that the schematic knowledge of subjects with PDD was unimpaired and that they
had intact levels of processing effects. Third, Toichi, Kamio, et al. (2002) used only
one measure (a recognition measure), while we used two measures (a recall as well
as a reaction time measure).
The findings of this study do not support the hypothesis that subjects with PDD
benefit less than typically developing subjects from semantic meaning (Frith, 2003;
Minshew & Goldstein, 2001). Furthermore, the analyses revealed no betweengroup differences with regard to the degree to which schema-relevant and schemairrelevant words were actually considered to be schema-relevant and schemairrelevant. We further found that -with the exception of the word friendlysubjects with PDD recalled the same schema-relevant words as typically
developing controls. In a way, this is contrary to the hypothesis that the semantic
network of subjects with PDD may be associated in an aberrant manner
(Beversdorf et al., 2000; Bowler et al., 1997).
One might suggest that the scripts in the present LOP Tasks are basic or simple
ones, and that adults with PDD would be more impaired when using more complex
scripts. However, adults differ in the amount of experience they have with events
and social situations. A simple script for one individual, might be a complex one
for another. So, the difference between basic and complex scripts is difficult to
determine (see also Barsalou, 1990).
In sum, none of the findings provided evidence that subjects with PDD are less
schematic than typically developing subjects, nor do they support the claim that
subjects with PDD are more impaired than controls when verbal materials to be
remembered are semantically or contextually related. Furthermore, the schematic
155
knowledge of subjects with PDD and control subjects did not differ in terms of
quality.
REFERENCES
Anderson, J. R. (1990). Cognitive psychology and its implications (Third edition). New
York: W. H. Freeman.
Ashcraft, M. H. (1994). Human memory and cognition. (Second edition). Harper Collins:
College Publisher.
Baayen, R.H., Piepenbrock, R., & Gulikers, L. (1995). The CELEX lexical database (CDRom). Linguistic Data Consortium, University of Pennsylvania, Philadelphia, PA.
Baddeley, A., Thomson, N., & Buchanan, M. (1975). Word length and the structure of
short-term memory. Journal of Verbal Learning and Verbal Behavior, 14, 575-589.
Baldwin, M.W. (1992). Relational schemas and the processing of social information.
Psychological Bulletin, 112, 461-484.
Baldwin, M. (1999). Relational schemas: Research into social-cognitive aspects of
interpersonal experience. In D. Cervone & Y. Shoda (Eds.), The coherence of
personality: Social-cognitive bases of consistency, variability, and organization (pp.
126-154). New York: Guilford.
Barsalou, L. (1990). On the indistinguishability of exemplar memory and abstraction in
category representations. In T. Srull & R. Wyer (Eds.), Content and process specificity
in the effects of prior experiences: Advances in social cognition (pp. 61-88). Hillsdale,
NJ: Erlbaum.
Beversdorf, D., Smith, B., Crucian, G., Anderson, J., Keillor, J., Barrett, A., Hughes, J.,
Felopulos, G., Bauman, M., Nadeau, S., & Heilman, K. (2000). Increased
discrimination of "false memories" in autism spectrum disorder. Proceedings of the
National Academy of Sciences of the United States of America, 97, 8734-8737.
156
Chapter 6
Bower, G.H., Black, J.B., & Turner, T.J. (1979). Scripts in memory for text. Cognitive
Bowler, D., Matthews, N., & Gardiner, J. (1997). Aspergers syndrome and memory:
Similarity to autism but not to amnesia. Neuropsychologia, 35, 65-70.
Brian, J., & Bryson, S. (1996). Disembedding performance and recognition memory in
autism/PDD. Journal of Child Psychology and Psychiatry, 37, 865-872.
Craik, F. I., & Tulving, E. (1975). Depth of processing and the retention of words in
episodic memory. Journal of Experimental Psychology: General, 104, 268-294.
Farrar, M., & Boyer-Pennington, M. (1999). Remembering specific episodes of a scripted
event. Journal of Experimental Child Psychology, 73, 266-288.
Farrar, M., & Goodman, G. (1990). Developmental differences in the relation between
scripts and episodic memory: do they exist? In R. Fivush & J. Hudson (Eds.), Knowing
and remembering in young children (pp. 30-64). New York: Cambridge University
Press.
Fiske, A. P., Haslam, N., & Fiske, S. T. (1991). Confusing one person with another: What
errors reveal about the elementary forms of social relations. Journal of Personality and
Social Psychology, 60, 656-674.
Frith, U. (2003). Autism: Explaining the enigma. Second edition. Oxford: Basil Blackwell.
Frith, U., & Happ, F. (1994). Autism: Beyond theory of mind. Cognition, 50, 115-132.
Frith, U., & Snowling, M. (1983). Reading for meaning and reading for sound in autistic
and dyslexic children. Journal of Developmental Psychology, 1, 329-342.
Geis, M., & Hall, D. (1976). Encoding and incidental memory in children. Journal of
Experimental Child Psychology, 22, 58-66.
Geis, M., & Hall, D. (1978). Encoding and congruity in childrens incidental memory.
Child Development, 49, 857-861.
Gillam, R., Hoffman, L., Marler, J., & Wynn-Dancy, M. (2002). Sensitivity to increased
task demands: Contributions from data-driven and conceptually driven information
processing deficits. Topics in Language Disorders, 22, 30-48.
Happ, F. (1996). Studying weak central coherence at low levels: Children with autism do
not succumb to visual illusions. A research note. Journal of Child Psychology and
Happ, F. (2000). Parts and wholes, meaning and minds: central coherence and its relation
to theory of mind. In S. Baron-Cohen, H. Tager-Flusberg & D. Cohen (Eds.),
157
Understanding other minds: Perspectives from developmental cognitive neuroscience.

(2nd edn). (pp.203-221). Oxford: Oxford University Press.
Higgins, E., & Bargh, J. (1987). Social cognition and social perception. Annual Review of
Hudson, J., Fivush, R., & Kuebli, J. (1992). Scripts and episodes: The development of
event memory. Applied Cognitive Psychology, 6, 483-505.
Inquisit 1.31 [Computer software]. (2002). Seattle, WA: Millisecond Software.
Jolliffe, T., & Baron-Cohen, S. (1999). A test of central coherence theory; linguistic
processing in high-functioning adults with autism or Aspergers syndrome: Is local
coherence impaired? Cognition, 71 149-185.
Kanner, L. (1943). Autistic disturbances of affective contact. Nervous Child, 2, 217 250.
Klinger, L., & Dawson, G. (2001). Prototype formation in autism. Development and
Psychopathology, 13, 111-124.
Lopez, B., & Leekam, S. (2003). Do children with autism fail to process information in
context ? Journal of Child Psychology and Psychiatry, 44, 285-300.
McClelland, J., & Rumelhart, D. (1981). An interactive activation model of context effects
in letter perception: An account of basic findings. Psychological Review, 88, 375-407.
Minshew, N., & Goldstein, G. (1993). Autism A distributed neural network deficit.
Journal of Clinical and Experimental Neuropsychology, 15, 56-56.
Minshew, N., & Goldstein, G. (2001). The pattern of intact and Impaired memory
functions in autism. Journal of Child Psychology and Psychiatry, 42, 1095-1101.
Mottron, L., & Burack, J. (2001). Enhanced perceptual functioning in the development of
autism. In J. Burack, T. Charman, N. Yirmiya, & P. Zelazo (Eds.), The development of
autism: Perspectives from theory and research (pp. 131-148). Mahwah, NJ: LEA.
Mottron, L., Burack, J., Stauder, J., & Robaey, P. (1999). Perceptual processing among
high-functioning persons with autism. Journal of Child Psychology and Psychiatry, 40,
203-211.
Mottron, L., Peretz, I., & Mnard, E. (2000). Local and global processing of music in
high-functioning persons with autism: Beyond central coherence? Journal of Child
Psychology and Psychiatry, 41,1057-1065.
ORiordan, M., Plaisted, K., Driver, J., & Baron-Cohen, S. (2001). Superior visual search
in autism. Journal of Experimental Psychology: Human Perception and Performance
Issue, 27, 719-730.
158
Chapter 6
Owings, R., & Baumeister, A. (1979). Levels of processing, encoding strategies, and
memory development. Journal of Experimental Child Psychology, 32, 1081-1105.
Richardson-Klavehn, A., Gardiner, J., & Ramponi, C. (2002). Level of processing and the
process-dissociation procedure: Elusiveness of null effects on estimates of automatic
retrieval. Memory, 10, 349-364.
Ropar, D., & Mitchell, P. (2001). Do individuals with autism and Aspergers syndrome
utilise prior knowledge when pairing stimuli? Developmental Science, 4, 433-441.
Ropar, D., & Mitchell, P. (2002). Shape constancy in autism: the role of prior knowledge
and perspective cues. Journal of Child Psychology and Psychiatry, 43, 647-653.
Rogers, T. B., Kuiper, N. A., & Kirker, W. S. (1977). Self-reference and the encoding of
personal information. Journal of Personality and Social Psychology, 35, 677-688.
Rudolph, K. D., Hammen, C., & Burge, D. (1995). Cognitive representations of self,
family, and peers in school-age children: Links with social competence and sociometric
status. Child Development, 66, 1385-1402.
Schank, R., & Abelson, R. (1977). Scripts, plans, goals and understanding. Hillsdale, NJ:
Lawrence Erlbaum Associates.
Shah, A., & Frith, U. (1993). Why do autistic individuals show superior performance on
the block design task ? Journal of Child Psychology and Psychiatry, 34, 1351-1364.
Swann, W., & Read, S. ( 1981). Acquiring self-knowledge: the search for feedback that
fits. Journal of Personality and Social Psychology, 41, 1119-1128.
Tager-Flusberg, H. (1991). Semantic processing in the free recall of autistic children:
Further evidence for a cognitive deficit. British Journal of a Developmental
Trafimow, D., & Wyer, R. (1993). Cognitive representation of mundane social events.
Journal of Personality and Social Psychology, 64, 365-376.
Toichi, M., & Kamio, Y. (2002). Long-term memory and levels-of-processing in autism.
Neuropsychologia, 40, 964-969.
Toichi, M., Kamio, Y., Okada, M., Skihama, M, Youngstrom, E., Findling, R., &
Yamamoto, K. (2002). A Lack of self-consciousness in autism. The American Journal
of Psychiatry, 159, 1422-1424.
Trillingsgaard, P. (1999). The script model in relation to autism. European Child &
Adolescent Psychiatry, 8, 45-49.
159
Volden, J., & Johnston, J. (1999). Cognitive scripts in autistic children and adolescents.
Journal of Autism and Developmental Disorders, 29, 203-211.
Swets & Zeitlinger.
Welch-Ross, M., & Schmidt, C. (1996). Gender-schema development and childrens
constructive memory: Evidence for a developmental model. Child Development, 67,
820-835.
Wing, L. (1988). The continuum of autistic characteristics. In: Schopler, E., & Mesibov,
G. B. (Eds.). Diagnosis and assessment in autism..Current issues in autism. (pp. 91110). New York: Plenum.
Wyer, R., & Radvansky, G. (1999). The comprehension and validation of social
information: The role of situation models. Psychological Review, 106, 89-118.
160
Appendix A
Schema-relevant and schema-irrelevant words of the first LOP Task.
Dutch
English
Naam
Vrije tijd
Werk
Studies
Weerpraatje
Name
Leisure
Work
Studies
Talk about the
weather
Polite
Superficial
Unsure
Distant
Silences
Place of residence
Friendly
Reticent
Age
Ill at ease
Interest
Formal
Ask questions
Tense
Personal
Beleefd
Oppervlakkig
Onzeker
Afstandelijk
Stiltes
Woonplaats
Vriendelijk
Afwachtend
Leeftijd
Onwennig
Interesse
Formeel
Vragen stellen
Gespannen
Persoonlijk
A
FrequencyA
FrequencyA
Dutch
English
294
3
496
19
0
Logarithmic
valueB
2,4683
0,4771
2,6955
1,2788
0
Hart
Fabrikage
Boom
Trommel
Stroomkabel
Heart
Manufacture
Tree
Drum
Electricity cable
183
3
53
14
0
Logarithmic
valueB
2,2625
0,4771
1,7243
1,1461
0
8
11
22
3
99
10
64
4
100
4
17
17
80
24
64
0,9031
1,0414
1,3424
0,4771
1,9956
1
1,8062
0,6021
2
0,6021
1,2304
1,2304
1,9031
1,3802
1,8062
Stoffige
Gearresteerd
Dodelijk
Oorverdovend
Opstand
Uitgeverij
Onbeweeglijk
Hardhandig
Overheid
Sluitend
Vloeistof
Bevoegd
Betalen
Strafbaar
Onzichtbaar
Dusty
Arrested
Lethal
Deafening
Revolt
Publishing company
Immobile
Hard-handed
Authorities
Closing
Liquid
Qualified
Pay
Punishable
Invisible
7
13
13
4
25
9
13
4
117
3
11
14
35
7
15
0,8451
1,1139
1,1139
0,6021
1,3979
0,9542
1,1139
0,6021
2,0682
0,4471
1,0414
1,1461
1,5441
0,8451
1,1761
The frequency of the words, scaled down to a range of 1 to 1,000,000.
Logarithmic value. The effect of the logarithmic scale is to emphasize the importance of lower frequency words in a way that the usual linear scale does not. For example,
(see Baayen et al., 1995).
Chapter 6
the difference between two words, one of frequency 2 and the other of frequency 1 becomes much greater than the difference between two words of frequency 2002 and 2001
Schema-relevant and schema-irrelevant words of the second LOP Task

Dutch
English
Frequency
Objectief
Proefpersoon
Laboratorium
Statistiek
Geld
Structuur
Nauwkeurig
Controle
Proefdier
Herhaalbaar
Hypothese
Saai
Betrouwbaar
Kunstmatig
Evaluatie
Waarheid
Tests
Gegevens
Theorie
Wetenschapper
Objective
Test subject
Laboratory
Statistic
Money
Structure
Accurate
Control
Guinea pig
Repeatable
Hypothesis
Dull
Reliable
Artificial
Evaluation
Truth
Tests
Data
Theory
Scientist
671
201
689
162
11691
2612
1380
2181
17
17
493
352
484
314
597
4365
216
4130
3949
112
Logarithmic
value
1,2041
0,699
1,2041
0,6021
2,4409
1,7924
1,5185
1,7076
0
0
1,0792
0,9031
1,0414
0,8451
1,1461
2,0128
0,699
1,9868
1,9685
0,4771
Dutch
English
Frequency
Verliefd
Prikkeldraad
Boodschappen
Vloerkleed
Eeuw
Vliegtuig
Onschuldig
Schouder
Stamkroeg
Paradijselijk
Schilderij
Lauw
Verleidelijk
Boosaardig
Wandeling
Behoefte
Metro
Telefoon
Familie
Tandenborstel
In love
Barbed wire
Purchase
Carpet
Century
Aeroplane
Innocent
Shoulder
Favourite pub
Heavenly
Painting
Lukewarm
Tempting
Malicious
Walk
Need
Subway
Telephone
Family
Toothbrush
852
321
975
161
7883
1577
767
3098
20
21
762
313
246
171
875
5004
120
2824
5174
139
Logarithmic
value
1,301
0,9031
1,3617
0,6021
2,2695
1,5682
1,2553
1,8633
0
0
1,2553
0,8451
0,7782
0,6021
1,3222
2,0719
0,4771
1,8261
2,0864
0,4771
Appendix B
161
CHAPTER
Alternative explanations to the empathic

accuracy findings
Chapter 7
164
ABSTRACT
In the present chapter, we reconsider our empathic accuracy studies and explore
some alternative explanations to the different findings. First, we look for reasons
that can account for the empathic accuracy differences between some of the control
groups. Then, we reanalyse the existing data using Social Relations Model (SRM)
analyses, so that the relative contribution of perceiver, target and relationship
effects to the overall mind-reading ability of subjects with PDD and control
subjects can be determined. Furthermore, we explore the power of the empathic
accuracy studies, and question the reliability and validity of the empathic accuracy
design.
A potential explanation for the differences between the control groups in
empathic accuracy seems to be the difference in the methodology of the studies.
Moreover, SRM analyses revealed substantial perceiver variance in each of the
studies. We further found that, with the exception of the study of Chapter 4, none of
the empathic accuracy studies had enough power to reveal our expected effects.
Although the interrater reliability ranged from acceptable to good for the content
accuracy ratings, the interrater reliability dropped in some of the studies beneath
the .80 border for the baseline accuracy ratings. Consistent with studies with
typically developing adults (Davis & Kraus, 1997), we failed to find reliable
predictors of empathic accuracy.
Alternative explanations to the empathic accuracy findings
165
INTRODUCTION
In the early chapters of this thesis (Chapter 2 to 4), we started from the
hypothesis that the traditional (static) mind-reading tasks are not subtle enough to
detect impairments in high-functioning adults with PDD. Therefore we made use of
the naturalistic empathic accuracy design, which indeed proved to be an alternative
for the static tasks that have been used so far. Then, in Chapter 5, we changed the
subject from performance-based to perceived mind-reading assessment methods
and, in Chapter 6, we investigated the schematic knowledge of young adults with
PDD. In the present chapter, we take up the thread of our empathic accuracy story
and elaborate on some undiscussed issues. Without the intention of being
exhaustive, we start with the enumeration of some expected, unexpected, and
inconclusive empathic accuracy results, and then explore some alternative
explanations to the findings.
Two findings support the case of impaired empathic accuracy in adults with
PDD. First, a comparison of a group with Asperger syndrome a group expected to
display more subtle impairments- with a typically developing control group
revealed the expected pattern of results. More specifically, a difference in empathic
accuracy between both groups was found on the second videotape of the standard
Empathic Accuracy Task (Chapter 2). Based on the study of Roeyers, Buysse,
Ponnet, and Pichal (2001), this finding was expected.
Second, the impairments in empathic accuracy in adults with PDD were found to
be, at least partly, due to the structure of the conversation. In the study reported in
Chapter 4, we manipulated the structure of both videotapes in such way that one of
the tapes was more structured than the other. The manipulation of structure
revealed the expected pattern of results: mind-reading difficulties between adults
with PDD and typically developing controls become more apparent when they have
to infer thoughts and feelings of other persons in a less structured or more chaotic
conversation, than when they have to do so in a more structured conversation.
On the other hand, there are some findings that do not support the case of
impaired empathic accuracy in adults with PDD. First, we found that during an
166
Chapter 7
initial conversation with a typically developing stranger, subjects with PDD did not
differ from typically developing subjects in the ability to infer the thoughts and
feelings of their interaction partner (Chapter 3). We have argued that one potential
explanation to this finding is that the participants with and without PDD were
familiar with the getting acquainted script of the empathic accuracy design.
Indeed, the study reported in Chapter 6 confirmed the presence of a getting
acquainted schema in control subjects as well as subjects with PDD.
Second, typically developing subjects did not have more difficulties to
accurately infer the unexpressed thoughts and feelings of people with PDD
compared to those of typically developing subjects (Part 2 of Chapter 3).
Third, it seems that accurate mind-reading can not be solely seen as a skill or
disposition belonging to an individual. Evidence for this is found in the studies with
the standard stimulus paradigm (see Appendix A), since the percentages of
variability in performance that is shared across both videotape tasks vary from 15%
(Chapter 4) to 56% (Chapter 2). Moreover, from our attempts to find potential
variables that may account for the subjects mind-reading performance (e.g.
difficulty, abstractness, topic of thought/feeling entry, and so on), we conclude that
characteristics other than these variables limit the performance of test-takers on one
videotape task and affect performance differently on the other videotape task. This
means that we are not (yet) able to attribute any between group differences in
performance to group differences in some ability that affects performance on the
tasks.
Fourth, when we consider each of the studies with the standard empathic
accuracy design as a self-containing study, the analyses clearly revealed differences
between adults with PDD and control adults. However, since the same Empathic
Accuracy Task was used in the Roeyers et al.s study (2001) and Chapter 2, we
were interested to know if the case of impaired empathic accuracy in adults with
PDD sustain when we combine and reanalyse the data sets of both studies, so that
the enlarged data set comprises 43 typically developing controls, 24 adults with
autism/PDD-NOS, 19 Asperger subjects. Combining both data-sets, the Cronbachs
alpha of the content accuracy provided by the judges was .81 for Video 1 and .82
for Video 2. The internal consistency of the baseline empathic accuracy provided
by the judges was .75 for Video 1 and .72 for Video 2.
167
We re-analysed all data with the Empathic Accuracy scores of both videotapes
as dependent variables and Group (i.e. controls, autism/PDD-NOS, Asperger) as
factor. The MANOVA testing for between-group differences on the Empathic
Accuracy Task yielded significance, F(4, 82) = 4.65, p = .001. More specifically,
between-group differences were found with regard to the inferences of the
unexpressed thoughts and feelings of the targets in Video 2, F(2, 83) = 8.9, p <
.001. Follow-up univariate analyses further revealed that the between-group
difference applied only for the comparison of the Roeyers at al.s group with
autism/PDD-NOS and the enlarged control group, p < .001, and not for the
comparisons of the other groups (i.e. Asperger group versus control group, and
autism/PDD group versus Asperger group). No between-group differences were
found on Video 1, F(2, 83) < 1. In sum, the findings suggest that the Empathic
Accuracy Task is able to distinguish adults with autism/PDD-NOS from control
adults, but might be not subtle enough to distinguish adults with Asperger
syndrome (i.e. a milder form of autism) from typically developing controls. The
latter finding is not supportive to what we reported in Chapter 2.
Furthermore, it seems that the empathic accuracy scores of the typically
developing control groups also differ from each other (see Table 1). Therefore, we
reanalysed the data with the empathic accuracy scores of both videotapes as
dependent variables and group (i.e. both control groups) as factor. The MANOVA
testing for between-group differences regarding the Empathic Accuracy Task was
significant, F(2, 40) = 3.63, p < .05. More specifically, a significant between-group
difference was found on Video 2 (see Table 1), which is a rather unexpected
finding.
Table 1
Empathic accuracy scores of the typically developing controls
Roeyers et al.
Chapter 2
N = 24
N = 19
F(1, 41)
SD
SD
Video 1
33.31
14.4
40.76
13.0
2.40
Video 2
23.26
10.2
31.55
12.8
6.73 *
EA Task
*p < .05
Chapter 7
168
Taken these findings together, it seems that there remain some inconclusive and
unexpected findings that merit further investigation. In the further of this chapter,
we look for alternative explanations to these non-supportive findings. First, we
provide two explanations that could account for the difference in empathic
accuracy between the control groups of both studies (see above). Second, we
explore some statistical reasons that could account for the non-supportive findings
in general. Third, we look for some methodological reasons, including reliability
and validity of our designs used.
1. DIFFERENCES
IN EMPATHIC ACCURACY BETWEEN THE TYPICALLY
DEVELOPING CONTROL GROUPS
As mentioned earlier, analysis revealed a difference in empathic accuracy

between the control group of the Roeyers et al.s study (2001) and that of the study
in Chapter 2. Potential statistical reasons and reasons with regard to the reliability
and the validity of the design will be discussed in the further of this chapter.
However, there appeared to be two other explanations that could account for this
between-group difference in empathic accuracy specifically.
One explanation for this difference might be that both groups differ concerning
sex ratio, chronological age, IQ and so on. Since the intelligence scores of the
control subjects of Roeyers et al. (2001) were not available, we were only able to
compare sex ratio and age of both groups. Analysis revealed no between-group
differences with regard to sex ratio (2(1) = 2.52). Furthermore, the ANOVA
testing revealed no between-group differences with regard to age, F(1, 41) < 1.
Therefore, it can be suggested that variables other than sex ratio and chronological
age may underlie the between-group difference on Video 2.
Another potential explanation for the between-group difference is the varying
methodology of both studies. In Chapter 2, we improved the methodology of the
Empathic Accuracy Task by digitising the analogue version used in the Roeyers et
al.s study (2001). As we described in the introduction of Chapter 2, this
digitisation may lead to an increased standardisation and may result in more
accurate empathic accuracy scores. Furthermore, using the computer program
169
VIDANN minimises the inaccuracies related to the use of a manual remote control
system to pause a videotape at the specific time moments that one of the targets had
reported having had a specific thought or feeling (De Clercq, Buysse, Roeyers,
Ickes, Ponnet, & Verhofstadt, 2001). Thus, although not measurable, the difference
between analogue and digital methodology may account -at least partly- for the
between-group difference in empathic accuracy.
Another possible explanation that could account for the between-group
difference in empathic accuracy might be that the coding teams of both studies are
different. We will discuss this in the further of this chapter.
2. STATISTICAL REASONS
In the present section, we explore some statistical reasons that may account for
the different non-supportive empathic accuracy findings.
First, we reanalyse the existing empathic accuracy data using the Social
Relations Model (SRM) of Kenny (1994) and test whether SRM analyses reveal
substantial perceiver variance and substantial target variance in the empathic
accuracy scores. If the level of perceiver variance is found to be consistently low,
this finding may explain why reliable individual-difference correlates of the mindreading ability are difficult to find. On the other hand, if the level of perceiver
variance is found to be substantial in any of the studies, this indicates that the
empathic accuracy design is a valid and valuable way to measure mind-reading,
and this should encourage us to search further for reliable and replicable individualcorrelates of empathic accuracy (see also Ickes et al., 2000).
Second, it can be suggested that the sample size of the studies is rather small,
which decreased the statistical power (i.e. the probability of detecting a
phenomenon under test). We will therefore calculate the power of the tests.
2.1. Social Relations Model analyses
The Social Relations Model (Kenny & La Voie, 1984; Warner, Lenny & Stoto,
1979) is a methodological and statistical model that allows us to partition the
variance found in interpersonal judgments into perceiver, target and relationship
170
Chapter 7
effects. Take, for example, Carol, whose first impression of Susan is that she
appears to be intelligent1. The perceiver effect describes how Carol, as a perceiver,
sees other people. If Carol typically sees other people as intelligent, then there will
be a large perceiver effect. So, the perceiver effect measures an individuals
tendency to see other people similarly on a dimension (Graziano & Tobin, 2002).
The target effect describes how Susan is typically perceived by others. If most
people agree that Susan is more intelligent than anyone else, there will be a large
target effect. In other words, the target effect indicates how much raters agree in
rating targets on a dimension (Graziano & Tobin, 2002). The relationship effect is a
dyadic effect and describes how Carol uniquely perceives Susan. The relationship
effect indicates the unique adjustment one person makes to another person (Cook,
2001). Because the relationship effect emerges after the individual-level effects of
perceiver and target are removed (Kenny, 1994, p. 17), measurement error is
contained in the relationship component. Without multiple measures, it is not
possible to separate relationship from error.
Methods and Results
The study of Roeyers, Buysse, Ponnet, & Pichal (2001)
The design of the Roeyers et al. (2001) study is a half-block design with 48
perceivers (i.e. 24 adults with PDD and 24 typically developing adults) and 4
targets. Consistent with the requirements of the half-block design (Kenny, 1994,
p.232-234), the perceivers did not serve as targets, and the targets did not serve as
perceivers.
Using the empathic accuracy data from the 48 perceivers and the four targets, we
(1) estimated the perceiver effect for each of the 48 perceivers, the target effect for
each of the four targets, and the relationship-plus-residual effect for each of the 192
perceiver-target pairs; (2) we calculated the mean square for the 48 perceivers, the
mean square for the four targets, and the mean square for the 192 perceiver x target
interactions; (3) calculated the average perceiver, target, and relationship-plusresidual variance; (4) tested the statistical significance of the perceiver and targets
1
This example is based on Kenny (1994, p. 16-17) and Marcus & Leatherwood (1998, p. 299).
171
variance using F-tests; and (5) estimated the percentage of variance explained by
the perceiver, target, and relationship-plus-residual effects. This was done for the
total population, for the group with PDD and for the typically developing control
group (see Table 2).
Reanalysis of the empathic accuracy scores of adults with PDD and control
adults reveals substantial perceiver and target variance in each group separately.
However, the target variance in the total population was not significant.
Furthermore, the target variance in the PDD group (38%) was found to be larger
than that in the control group (15%), suggesting the readability of the targets may
be of major influence on the ability of adults with PDD to accurately infer others
The study of Chapter 2
Data from the second chapters study were also amenable to a social relations
analysis using Kennys (1994, p. 232-234) half-block design. Consistent with the
requirements of this design, the 38 perceivers (i.e. 19 adults with Asperger
syndrome and 19 typically developing adults) did not serve as targets, and the four
targets did not serve as perceivers.
(1) estimated the perceiver effect for each of the 38 perceivers, the target effect for
each of the four targets, and the relationship-plus-residual effect for each of the 152
perceiver-target pairs; (2) we calculated the mean square for the 38 perceivers, the
mean square for the four targets, and the mean square for the 152 perceiver x target
interactions; (3) calculated the average perceiver, target, and relationship-plusresidual variance; (4) tested the statistical significance of the perceiver and targets
variance using F-tests; and (5) estimated the percentage of variance explained by
the perceiver, target, and relationship-plus-residual effects. This was done for the
total population, for the group with Asperger syndrome and for the typically
developing control group (see Table 2).
Chapter 7
172
Table 2
Social Relations Model
Sources of Variance
n
Perceiver
Target
Relationship
and Residual
Roeyers et al. (2001)

PDD
24
22.34 **
38.31*
39.35
Control
24
21.60 **
15.51*
62.89
Total Group
48
22.66 **
26.10
51.25
Asperger
19
43.71 **
16.99 *
39.30
Control
19
21.04 *
19.12 *
59.83
Total Group
38
37.03 **
15.57 *
47.10
Subjects with PDD as targets
0.81
40.46 *
58.72
Control subjects as targets
20.27
81.25
Both groups as targets
12
32.64 **
65.92
Subjects with PDD as targets
23.11
80.56
Control subjects as targets
34.37 *
69.90
Both groups as targets
16
27.82 **
71.33
PDD
30
16.67 *
1.89
81.94
Control
30
26.74 **
4.30
68.97
Total Group
60
26.81 **
1.56
71.63
Chapter 2
Chapter 3
Group 1
Group 2
Chapter 4
*p < .05, **p < .01
Reanalysis of the empathic accuracy scores of adults with Asperger syndrome

and control adults reveals substantial perceiver and target variance in both groups.
The perceiver variance in the Asperger group (43.71%) was found to be larger than
that in the control group (21%), suggesting that characteristics of the perceivers
may be of major influence on the ability of adults with Asperger to accurately infer
others thought/feeling entries.
173

The design in the first part of this study is a round-robin design: all eleven
perceivers serve as target and all eleven targets serve as perceiver (Kenny, 1994, p.
236-238). Since each target is only rated by one single perceiver, the data are not
amenable to use the round-robin design. However, the design in the second part of
this study is a half-block design with multiple perceivers and multiple targets. In
this part, a group of 7 typically developing subjects inferred the thoughts and
feelings of 6 pairs of PDD/non PDD subjects, and a group of 6 typically developing
perceivers inferred the thought/feeling entries of 8 pairs of PDD/non PDD subjects.
Using the empathic accuracy data, we (1) estimated the perceiver effect for each
of the perceivers, the target effect for each of the targets, and the relationship-plusresidual effect for each of the perceiver-target pairs; (2) we calculated the mean
square for the perceivers, the mean square for the targets, and the mean square for
the perceiver x target interactions; (3) calculated the average perceiver, target, and
relationship-plus-residual variance; (4) tested the statistical significance of the
perceiver and targets variance using F-tests; and (5) estimated the percentage of
variance explained by the perceiver, target, and relationship-plus-residual effects.
This was done for each group. Whereas in the above-mentioned studies the
status of the perceiver was further estimated (i.e. PDD subjects as perceiver versus
control subjects as perceiver), in this study we further estimated the status of the
target (i.e. PDD subjects as target versus controls as target). The results are shown
in Table 2.
The reanalysis of the empathic accuracy data revealed no significant perceiver
variances, which probably can be attributed to the small number of perceivers. The
target variances were consistently found to be significant.
Data from the fourth chapters study were amenable to a social relations analysis
using Kennys (1994, p. 232-234) half-block design. Consistent with the
requirements of the half-block design (Kenny, 1994), the perceivers did not serve
as targets, and the targets did not serve as perceivers.
(1) estimated the estimated the perceiver effect for each of the 60 perceivers, the
174
Chapter 7
target effect for each of the four targets, and the relationship-plus-residual effect for
each of the 240 perceiver-target pairs; (2) we calculated the mean square for the 60
perceivers, the mean square for the four targets, and the mean square for the 240
perceiver x target interactions; (3) calculated the average perceiver, target, and
relationship-plus-residual variance; (4) tested the statistical significance of the
perceiver and targets variance using F-tests; and (5) estimated the percentage of
variance explained by the perceiver, target, and relationship-plus-residual effects.
This was done for the total population, for the group with PDD and for the typically
developing control group (see Table 2).
Reanalysis of the empathic accuracy scores of adults with PDD and control
adults reveals substantial perceiver variance in both groups. However, the target
variance in both groups was not significant. Furthermore, the perceiver variance in
the PDD group (17%) was found to be smaller than that in the control group (27%),
suggesting that characteristics of the perceivers may be of major influence on the
ability of the control adults to accurately infer others thought/feeling entries.
Conclusions
We found that the perceiver variance was substantial in the studies in which a
relative large set of perceivers inferred the thoughts and feelings of the same set of
multiple target individuals (see also Ickes et al., 2000). Since the social relations
analyses revealed substantial perceiver effects, this finding provides evidence for
the validity of the empathic accuracy design. On the other hand, it should be noted
that, with regard to the amount of perceiver and target variance, the results obtained
were quite variable across the data sets that were analysed.
2.2. Power analyses

The power of a statistical test depends upon three parameters: the reliability of
the sample results, the effect size, and the significance criterion (Cohen, 1988).
These three parameters and the power are so related with each other, that any one
of them is a function of the others. This relationship makes four different types of
power analysis possible. In each analysis, one of the parameters is determined as a
function of the other parameters (Cohen, 1965, p. 97-101). In the present section,
175
we calculate the power as a function of alpha, effect size and sample size (i.e. posthoc power analyses).
Table 3
Post-hoc power analyses
Groups
Total sample size
Power
Small effect size
48
.10
Medium effect size
48
.40
Large effect size
48
.77
Small effect size
38
.09
Medium effect size
38
.32
Large effect size
38
.67
Small effect size
86
.12
Medium effect size
86
.52
Large effect size
86
.91
Small effect size
22
.07
Medium effect size
22
.20
Large effect size
22
.43
Small effect size
60
.12
Medium effect size
60
.48
Large effect size
60
.86
Roeyers et al. (2001)
Chapter 2
Roeyers et al. (2001) and

Chapter 2a
Chapter 3
Chapter 4
The analyses were based on the average group size
Methods and Results

As described by Cohen (1965, p. 98-99), it is proposed as a convention that a
desired power value is .80 (or more). We calculated the power (with the premise
that alpha = .05; F-tests are conducted in the study of Roeyers et al. (2001),
Chapter 2 and Chapter 4; and paired t-tests are conducted in Chapter 3), using the
176
Chapter 7
statistical program of Buchner and colleagues (Buchner, Faul, & Erdfelder, 1992;
Erdfelder, Faul, & Buchner, 1996). As shown in Table 3, post-hoc power analyses
revealed that the sample sizes are large enough to measure large effects in Chapter
4 and to measure large effects in the pooled data of Roeyers et al. (2001) and
Chapter 2, but not in the other studies.
Discussion
As described in the empathic accuracy studies, we believe the empathic accuracy
design to be a valuable alternative for measuring mind-reading in adolescents and
adults with PDD. Because of the severe social impairments of subjects with PDD,
we expected large effects in the study of Roeyers et al. (2001), Chapter 3 and
Chapter 4. Since Asperger syndrome is considered to be a milder form of autism,
minor effects (i.e. medium or small) were expected in Chapter 2. However, posthoc power analyses suggest that Chapter 2 may not have enough power to reveal
small or medium effects. Moreover, even the enlarged data-set is not large enough
to reveal small or medium effects. So, the differences between the studies in sample
size (and the related power) may account for some of the non-supportive findings.
Although it is common practice that, in the experimental planning, the researcher
anticipates a certain effect size, sets a significance criterion alpha, and then
specifies the amount of power he desires, so that the sample size that is necessary to
meet these specifications can be determined (Erdfelder et al., 1996; Cohen, 1965),
this was not always realistic in our studies. For instance, if we set alpha at .05, the
power at .80, and with the premise that the performance of two groups (e.g. adults
with and without PDD) will be compared using F-tests, then 52 participants (or
more) are needed to find large effects, 128 participants are needed to find medium
effects, and 788 participants are needed to find small effects.
It should however be noted that the prevalence rates of autism vary between 1020 per 10,000 cases (Fombonne, 1999, 2003), and that autism is usually associated
with mental retardation in about 70% to 80% of the cases (Rutter, 1979; Fombonne,
2003). The estimated prevalence rate of Asperger syndrome is 2.5 per 10,000 cases
(Fombonne, 2003). Since we worked only with a particular part of the PDD
population (i.e. high-functioning adolescents and adults), the recruitment of
participants was not that easy. Furthermore, considering other studies in which
177
high-functioning adults with PDD participated (e.g. Baron-Cohen, Jolliffe,

Mortimore, & Robertson, 1997; Baron-Cohen, Wheelwright, Hill, Raste, & Plumb,
2001; Happ, 1994; Heavey, Phillips, Baron-Cohen, & Rutter, 2000; Jolliffe &
Baron-Cohen, 1999; Kleinman, Marciano, & Ault, 2001), with the number of
participants with PDD ranging from 15 to 24, it seems not unusual to work with
small samples. On the other hand, there are other ways to increase to possibility of
finding effects (i.e. the power). For instance, through the use of cognitive loads
during the empathic accuracy tasks, or by enhancing the motivation of the
participants.
3. METHODOLOGY OF THE EMPATHIC ACCURACY DESIGN

3.1. Reliability of the empathic accuracy design
In the empathic accuracy design, independent raters are asked to judge the
similarity between each actual thought or feeling and the corresponding inferred
thought or feeling made by the perceivers, and between the randomly paired actual
thought/feeling entry and the inferred thought/feeling entry.
In each of the empathic accuracy chapters, we calculated the interrater reliability
for the content accuracy ratings as well as the baseline empathic accuracy scores.
Although it is common practice in studies with the empathic accuracy design to use
Cronbachs alphas (e.g. Buysse & Ickes, 1999; Ickes, Bissonnette, Garcia, &
Stinson, 1990; Ickes, Stinson, Bissonnette, & Garcia, 1990; Marangoni, Garcia,
Ickes, & Teng, 1995; Simpson, Ickes, & Blackstone, 1995; Stinson & Ickes, 1992),
one could argue that intraclass correlations (where the unit of analysis is the
individual rating) are more appropriate than Cronbachs alphas (where the unit of
analysis is the mean of all the ratings). Since -in the present thesis- we are
interested in between-group differences and not in intra-group differences, we do
not particular adhere to this thesis. Nevertheless, Appendix B presents an overview
of the Cronbachs alphas as well as the intraclass correlations for each of the
studies.
For the content accuracy ratings, the Cronbachs alpha varied from .80 (in
Chapter 2) to .94 (in Chapter 5), and the intraclass correlation varied from .46 to
178
Chapter 7
.75. For the baseline accuracy ratings, Cronbachs alpha ranged from a low of .71
in Chapter 2 to a high of .94 in Chapter 4, and intraclass correlation ranged from
.24 to .77.
So, while the interrater reliability ranged from acceptable (.80) to good for the
content accuracy ratings, the interrater reliability dropped -in some of the studiesbeneath the .80 border for the baseline accuracy ratings. Therefore, one might
suggest that, since we are not able in some studies- to measure the baseline
empathic accuracy scores on a reliable way, we should not try to do. On the other
hand, as described in Ickes, Stinson et al. (1990, p. 734), the logic underlying the
baseline score is that perceivers can make a lucky, uneducated guess about the
content of the thought/feeling entry and, therefore be correct on a few occasions
because of change alone. In order to distinguish accuracy that is based on some
form of inferential ability from a false chance accuracy, the baseline score is
therefore subtracted from the content score. Notwithstanding, intuitively, it seems
logical that chance is more difficult to measure reliably. So, not calculating the
baseline accuracy is no alternative. In either case, we lose.
As mentioned before, it should be noted that the raters in each of the studies are
different ones. Interrater reliability is dependent upon the ability of two or more
raters to be consistent. However, training can enhance interrater reliability, and
therefore, in each of the studies, the actual rating of thought/feeling entries was
preceded by a practical session. The purposes of a practical session are evident: we
want (a) to make the raters comfortable with their task (i.e. rating the level of
similarity on a 3-point scale), (b) to explain them the nuances of score 0, 1 and 2,
and (c) to bring them into line with regard to the criteria to be used. However, it is
not the purpose of a practical session to bring the different raters across all studies
into the same line. Thus, if the raters of Chapter 2 agree in the practical session that
I am sad is essentially the same content as Im depressed (i.e. score 2), and the
raters of Roeyers et al. (2001) decide that in that case a score of 1 (somewhat
similar but not the same content) is more appropriate, then the raters of Chapter 2
and those of Roeyers et al. (2001) use different standards. Although intuitively and
not measurably, this may explain the differences in empathic accuracy between the
control group of Roeyers et al. (2001) and that of Chapter 2.
179
3.2. Validity of the empathic accuracy design

Studies with typically developing subjects have shown that self-report measures
of empathy-related dimensions are rather poor and unreliable predictors of
participants empathic accuracy scores (Davis & Kraus, 1997; Eisenberg, Murphy,
& Sheppard, 1997; Ickes, 2003). In the present section, we attempt to identify some
individual-difference measures that may predict the subjects empathic accuracy
ability. First, we explore whether the degree to which individuals show autistic
traits correlates with their empathic accuracy. Second, we combine and re-analyse
the data-sets of Chapter 4, 5 and 6, and assess whether the subjects empathic
accuracy scores correlate with their perceived mind-reading performance (i.e.
performance on Interpersonal Reactivity Index and Emotional Quotient Inventory)
and their performance on the static Eyes Task.
Are autistic traits associated with empathic accuracy ?
It seems plausible that the empathic accuracy ability of subjects with PDD is
related to the severity of their social-communication disabilities. Therefore, we
asked the participants of Chapter 4 to complete the Autism-Spectrum Quotient
(AQ), a self-administered questionnaire for identifying the degree to which a
normally intelligent adult shows autistic traits (Baron-Cohen, Wheelwright,
Skinner, Martin, & Clubley, 2001).
The AQ includes five 10-items subscales: (1) social skill, (2) attention switching,
(3) attention to detail, (4) communication, and (5) imagination (Baron-Cohen,
Wheelwright, Skinner, et al., 2001). All of the items are scored on a 4-point scale,
ranging from definitely agree to definitely disagree. Each item scores 1 point if
the subject records the autistic-like behaviour either mildly or strongly, so that the
minimum and maximum score of each of the subscales can vary between 0 and 10.
By aggregating the scores on the five subscales, a total score is derived. Subjects
completed an authorized Dutch translated version of the AQ2.
According to Baron-Cohen and colleagues, the AQ has good test-retest
reliability and reasonable construct validity. The internal consistency is moderate to
high (see Baron-Cohen, Wheelwright, Skinner, et al., 2001). For the present
2
Translated by Ponnet, Roeyers, & Buysse (2001).
Chapter 7
180
sample, the internal consistencies (Cronbachs alpha) ranged from .48 to .76 for the
PDD group and from a low of .40 to .79 for the control group.
A series of ANOVAs were conducted for the Total AQ score and for each of the
5 subscales, with the scale as dependent variable and Group (PDD versus control)
as factor. As shown in Table 4, analyses revealed significant between-group
differences on the Total AQ and on each of the subscales, suggesting that the group
with PDD scored significant higher than the control group on each of the scales.
Then, the correlations between the content empathic accuracy scores and the AQ
scores were calculated for both groups separately (see Table 5). Analyses revealed
just one significant negative correlation between the control subjects content
accuracy scores on Video 1 and their scores on the subscale Imagination.
Table 4
Performance on the AQ
PDD (n = 30)
Control (n = 30)
SD
Range
SD
Range
F(1, 59)
25.87
9.53
6-44
14.10
6.25
4-31
32.00***
Social skill
5.70
3.10
0-10
1.70
2.05
0-8
34.75***
Attention switching
5.40
2.75
1-10
3.35
1.83
1-7
9.57**
Attention to details
5.03
2.17
0-9
3.67
2.75
0-10
4.57*
Communication
5.13
2.57
1-10
2.07
1.55
0-7
31.31***
Imagination
4.60
2.03
2-8
3.13
2.22
0-8
7.13*
Total score
*p < .05, **p < .01, ***p < .001
181
Table 5
Correlations between AQ and empathic accuracy
Empathic accuracy
Video 1
Video 2
Unstructured
Structured
PDD
-.19
-.12
Control
-.06
-.09
PDD
-.25
-.11
Control
-.05
-.02
PDD
-.14
-.06
Control
-.09
-.17
PDD
.04
-.31
Control
.24
-.01
PDD
-.21
-.09
Control
.12
.15
PDD
-.10
-.21
Control
-.43 *
-.18
AQ Total
AQ Social Skills
AQ Attention Switching
AQ Attention to details
AQ Communication
AQ Imagination
*p < .05
Which other correlates are associated with empathic accuracy ?

In order to identify some other correlates of empathic accuracy, we combined
the data-sets of Chapter 4, 5 and 6. As shown in Table 6, there are some significant
correlations (Pearson) between the Empathic Accuracy Task and the other tasks,
but the pattern differs in both groups and is rather miscellaneous.
correlations into z-scores. The strength of associations was equal in both groups,
z1 z2
3
1
1
+
n1 3 n2 3
182
Chapter 7
with the exception of significant associations between content accuracy on Video 1

and Total EQ-i (z = 2.21, p < .05), content accuracy on Video 1 and self-reported
Fantasy (z = -2.01, p < .05) and content accuracy on Video 2 and self-reported
Perspective Taking (z = -2.68, p < .01).
Discussion
We failed to find an unequivocal pattern in the associations between empathic
accuracy scores and empathy-related dimensions. Since the Autism Spectrum
Quotient (AQ) aims to measure the degree to which someone shows autistic traits,
the absence of any association between the AQs total score and empathic accuracy
scores was rather unexpected. However, our findings are in a way- consistent with
studies with typically developing subjects (Davis & Kraus, 1997; Ickes, Stinson, et
al., 1990), which failed to find self-report questionnaire measures that can predict
participants empathic ability.
183
Tabel 6
Empathic Accuracy
Video 1
Video 2
Unstructured
Structured
Eyes Task
PDD
.40 *
.01
Control
.31
-.17
.50 *
-.20
Total EQ-I a
PDD
Control
-.21
.10
Perspective-taking PDD
.23
-.32
Perspective-taking Control
.13
IRI Self-report
Fantasy PDD
-.03
.38 *
-.20
Fantasy Control
.48 **
Empathic Concern PDD
.23
-.20
-.23
.07
.18
.12
-.05
.03
Perspective-taking PDD
.11
-.09
Perspective-taking Control
.22
.08
Fantasy PDD
.06
-.45*
-.27
-.02
.05
-.16
Empathic Concern Control
-.06
.02
Personal Distress PDD
-.17
-.33
Personal Distress Control
-.10
-.05
-.10
.21
Empathic Concern Control

Personal Distress PDD
Personal Distress Control
.13
IRI Other-report
Fantasy Control
Empathic Concern PDD
LOP task 1 Semantically encoded

schema-relevant words
PDD
Control
a
.59 *
.36
* p < .05, Correlations with the EQ-I were done on the basis of 18 subjects with PDD and 21
control subjects, b Correlations with the LOP task 1 (i.e. getting acquainted conversation) were
done on the basis of 15 subjects with PDD and 15 controls.
184
Chapter 7
GENERAL CONCLUSION
In the present chapter, we explored some alternative explanations to the
empathic accuracy findings. First, we provided two explanations that could account
for the difference in empathic accuracy between the control group of the Roeyers et
al.s study (2001) and that of Chapter 2. On the one hand, we hypothesized that
both groups differed concerning age, sex, and so on. The analyses revealed no
between-group differences, although the findings should be interpreted with
caution, since some of the subjects descriptive characteristics were not available
(e.g. intelligence scores in the Roeyers et al.s study). On the other hand, the
methodology of both studies differed (analogue versus digital), which may account
-at least partly- for the between-group difference in empathic accuracy.
Then, we explored some statistical issues. We reanalysed the existing empathic
accuracy data using the Social Relations Model of Kenny (1994). Consistent with
Ickes et al. (2000), we found that the perceiver variance was substantial in the
studies in which a relatively large set of perceivers inferred the thoughts and
feelings of the same set of multiple targets. This provides evidence that the
empathic accuracy design is a valid way to measure mind-reading. Post-hoc power
analyses further revealed that the sample size of the studies was relatively small,
which decreased the power of some of the studies, and -in turn- may have
hampered us to find some effects.
Furthermore, we explored the reliability of the empathic accuracy design. We
found that the interrater reliability ranged from acceptable to good for the content
accuracy ratings. However, in some of the studies, the interrater reliability dropped
beneath the .80 border for the baseline accuracy ratings. With regard to the validity
of the design, we failed to find reliable predictors of empathic accuracy.
REFERENCES
185
autism, males and females, scientists and mathematicians. Journal of Autism and
Buchner, A., Faul, F., & Erdfelder, E. (1992). GPOWER: A priori-, post hoc-, and
compromise power analyses for the Macintosh [computer program]. Bonn, Germany:
Bonn University.
Cohen, J. (1965). Some statistical issues in psychological research. In B. B. Wolman
(Ed.), Handbook of clinical psychology (pp. 95-121). New York: McGraw-Hill.
Cohen, J. (1988). Statistical power analysis for the behavioural sciences. Second edition.
Hillsdale, NJ: Erlbaum.
Cook, W. (2001). Interpersonal influence in family systems: a social relations model
analysis. Child Development, 72, 1179-1197.
Davis, M., & Kraus, L. (1997). In W. Ickes (Ed.), Empathic accuracy. (pp. 144-193). New
York: Guilford Press.
Computers, 33, 159-166.
In W. Ickes (Ed.), Empathic accuracy. (pp.73-116). New York: Guilford Press.
Erdfelder, E., Faul, F., & Buchner, A. (1996). Gpower: A general power analysis program.
Behavior Reasearch Methods, Instruments, & Computers, 28, 1-11.
Fombonne, E. (1999). The epidemiology of autism: a review. Psychological Medicine, 29,
769-786.
Fombonne, E. (2003). Epidemiological surveys of autism and other pervasive
developmental disorders: An update. Journal of Autism and Developmental Disorders,
33, 365-382.
Chapter 7
186
Graziano, W., & Tobin, R. (2002). Agreeableness: dimension of personality or social

desirability artifact ? Journal of Personality, 70, 695-727.
Ickes, W., Bissonnette, V., Garcia, S., & Stinson, L. L. (1990). Implementing and using
the dyadic interaction paradigm. In C. Hendrick, & M.S. Clark (Eds.), Review of
psychology, Vol. 11, pp. 16-44. Newbury Park, CA: Sage.

Ickes, W., Buysse, A., Pham, H., Rivers, K., Erickson, J., Hancock, M., Kelleher, J., &
Gesn, P. (2000). On the difficulty of distinguishing good and poor perceivers: A
social relations analysis of empathic accuracy data. Personal Relationships, 7, 219-234.
Ickes, W., Robertson, E., Tooke, W., & Teng, G. (1986). Naturalistic social cognition:
Methodology, assessment, and validation. Journal of Personality and Social
59, 730-742.
Jolliffe, T., & Baron-Cohen, S. (1999). The strange stories test: A replication with highfunctioning adults with autism or Asperger syndrome. Journal of Autism and
Kenny, D. (1994). Interpersonal perception: A social relations analysis. New York:
Guilford Press.
Kenny, D., & La Voie, L. (1984). The social relations model. In L. Berkowitz (Ed.),
Advances in experimental social psychology (Vol. 18, pp. 142-182). San Diego, CA:
Academic Press.
187
29-36.
Marcus, D., & Leatherwood, J. (1998). The interpersonal circle at zero acquaintance: a
social relations analysis. Journal of Research in Personality, 32, 297-313.
Rutter, M. (1979). Language, cognition and autism. In R. Katzman (Ed.), Congenital and
acquired cognitive disorders (pp. 247-264). New York: Raven Press.
Simpson, J. A., Ickes, W., & Blackstone, T. (1995). When the head protects the heart:
Warner, R., Kenny, D., & Stoto, M. (1979). A new round robin analysis of variance for
social interaction data. Journal of Personality and Social Psychology, 37, 1742-1757.
Chapter 7
188
Appendix A
Pearson correlates between the videotapes of the Empathic Accuracy Tasks
EA Task
New EA Task
Video 2
Video 2
EA Task Video 1
Roeyers, Buysse, Ponnet & Pichal (2001)
PDD
.48 **
Control
.26
Chapter 2
Asperger
.75 ***
Control
.33
Chapter 4
PDD
.17
Control
.38 *
*p < .05, **p < .01
189
Appendix B
Interrater reliability
Cronbachs
Intraclass
alpha
correlation
Content empathic accuracy
.80
.46
Baseline accuracy
.78
.41
.82
.47
Baseline accuracy
.71
.33
.90
.63
Baseline accuracy
.94
.77
.92
.70
Baseline accuracy
.90
.64
.94
.75
Baseline accuracy
.60
.24
.90
.65
Baseline accuracy
.77
.40
Chapter 2
Video 1
Video 2
Chapter 3
PDD group
Control group
Chapter 4
Video 1
Video 2
CHAPTER
General discussion
Chapter 8
192
RESEARCH OVERVIEW
Tuning the empathic accuracy design into the impairments of adults with PDD
In this thesis we aimed to measure the mind-reading ability of adults with a
pervasive developmental disorder (PDD) in an ecologically valid way. However,
linking the impairments in social performance of adults with PDD with their
impairments in the ability to represent minds has so far been an intractable issue.
Most likely because it is difficult to create operational measures that are valid as
well as practical to administer, and that are tuned into adults with PDD. Throughout
this thesis, it became clear that the standard as well as the dynamic empathic
accuracy design is likely to fulfil these requirements.
From the beginning of this thesis, the conclusions of each empathic accuracy
study were taken as point of departure for the next study. In Chapter 2, we tried to
replicate the results of the study of Roeyers et al. (2001) with adults with Asperger
syndrome, a group expected to display more subtle impairments. Consistent with
Roeyers et al. (2001), we found a difference between the Asperger group and the
control group on the second videotape of the standard Empathic Accuracy Task
(Chapter 2). These findings induced us to assess the mind-reading performance of
adults with and without PDD using the dyadic empathic accuracy design (Chapter
3). Since the dyadic empathic accuracy design requires various social skills, this
design can be considered as more complex than that of the standard stimulus
design. As such, we expected pronounced within-dyad differences. Contrary to our
expectations, we found that the subjects with PDD did not differ from the control
adults in the ability to infer their interaction partners thoughts and feelings. From
this, we suggested two alternative explanations. First, we assumed that the adults
with and without PDD were familiar with the getting acquainted script of the
empathic accuracy design, and that this familiarity may have led to an enhanced
mind-reading performance. Since the study reported in Chapter 6 confirmed the
presence of a getting acquainted schema in control subjects as well as in subjects
with PDD, we consider this as a plausible explanation. Second, we hypothesized
that the dyadic interaction design enabled the participants to structure the social
situation according to their own possibilities. In order to assess whether or not
General discussion
193
subjects with PDD benefit from structured situations, we developed a new standard
Empathic Accuracy Task (Chapter 4). The structure of this tasks videotapes was
manipulated in such way that one videotape was more structured than the other.
Consistent with our expectations, we found that mind-reading impairments between
adults with PDD and control adults were more pronounced when subjects had to
infer the thoughts and feelings of other persons in the less structured conversation.
In each of the empathic accuracy studies, we also explored several
characteristics that might affect the performance of subjects with and without PDD
differently. We explored the thematic topic (Chapter 2 and 3), the difficulty
(Chapter 3 and 4), the abstractness (Chapter 3 and 4) and the affective tone
(Chapter 4) of the targets actual thoughts and feelings, and assessed the
congruence between the targets verbal communication and direction of gaze, and
their written actual thoughts and feelings (Chapter 4). Subsequently, we worked out
empathic accuracy scores along these different variables for each participant.
However, the pattern of results gave no hint of individual differences dimensions in
the PDD group and therefore we concluded that characteristics other than these
variables affect performance differently on our empathic accuracy tasks (see
Chapter 7).
Theoretical considerations
We made use of two terms that stem from radically different perspectives. On
the one hand, the term mind-reading comes from the innatist/modularist theory
of mind perspective of Baron-Cohen and his colleagues (Baron-Cohen, Leslie, &
Frith, 1985). Although the theory of mind approach has roots in the psychology of
typical development of the Piagetian era (Piaget, 1929), the term theory of mind
derives from a paper of Premack and Woodruff (1978) entitled Does the
chimpanzee have a theory of mind?. They called the ability to impute mental
states to oneself and others a theory, because mental states are not observable. A
few years later, Wimmer and Perner (1983) were the first who tested the theory of
mind ability in typically developing children. As mentioned earlier (Chapter 1),
Baron-Cohen and colleagues were the first who tested the issue with children with
autism (Baron-Cohen et al., 1985) and found that a large number of children with
194
Chapter 8
autism was unable to attribute a false belief to another individual. This finding has
been replicated several times (e.g. Baron-Cohen, Leslie, & Frith, 1986; Charman &
Baron-Cohen, 1992, 1995; Leekam & Perner, 1991; Leslie & Frith, 1987; Perner,
Frith, Leslie, & Leekam, 1989). However, one of the difficulties in assessing the
theory of mind hypothesis of autism is that there is no clear single statement of the
hypothesis (Boucher, 1989). In its narrow use, the term refers only to the ability to
impute cognitive or volitional states to others (Premack & Woodruff, 1978). In its
broader use, the terms also refers to the work of Flavell on social cognition, and
especially perspective taking (Flavell, 1974; Flavell, Botkin, Fry, Wright, & Jarvis,
1986). Flavell identified four successive knowledge components that children must
acquire for perspective taking (see Flavell, 1992). The existence component refers
to the knowledge that mental states exist. The need component pertains to
childrens recognition that certain situations require them to obtain knowledge
about anothers mental states. The inference component refers to the actual ability
to obtain such knowledge by means of inference. Finally, the application
component relates to the ability to apply this knowledge in social situations. While
theory of mind in its narrow usage mainly refers to the first component of Flavell
(1992), the broader usage of the term refers also to the second component, and -to
some degree- even to the third component. Consequently, the term mind-reading
can be considered synonymous with the term theory of mind in its broader usage.
On the other hand, the term empathy comes from mainstream social
psychology and has as such been defined in different ways by different authors (see
Eisenberg, Murphy, & Sheppard, 1997; Preston & de Waal, 2002). Opinions differ
greatly on the matter whether empathy should be viewed as unidimensional
versus multidimensional, as cognitive versus emotional, or as a global skill versus a
set of component subskills (Ickes, Marangoni, & Garcia, 1997, p. 283). Although it
is beyond the scope of this thesis to review these various opinions in detail,
empathy is probably best referred to as a process of cognitive and affective
perspective taking (Davis, 1994), that can be assessed as a collection of individual,
sequentially related subskills (Ickes et al., 1997). Embedded in the term empathy
are both the cognitive ability to understand others emotions and cognitions, and
the emotional response towards others thoughts and feelings (Eisenberg et al.,
1997).
General discussion
195
It is difficult to determine to what extent the terms mind-reading and empathy

are related to each other (Corcoran, 2000). The most accepted argument is that, to
show empathy, one must be able to appreciate others mental states. From this point
of view, a cognitive mind-reading deficit precludes empathic display in autism. A
second -less supported- explanation suggests that empathy and mind-reading are in
fact dissociable. Probably the best determination of the relationship between both
terms is that empathy involves cognitions as well as emotions, and therefore
overlaps with the term mind-reading that mainly focuses on the cognitive
component.
It is however important not to dilute the distinction between empathy (the
process) and empathic accuracy (the outcome) (Thomas & Fletcher, 1997), which
is the focus of the present thesis. The empathy process consists of three
components: empathic understanding (the ability to accurately infer other peoples
thoughts and feelings), empathic expression (the ability to express these thoughts
and feelings) and empathic communication (the dialectic aspect of the empathy
process) (Barrett-Lennard, 1981; Elliott et al., 1982; Keefe, 1979). When empathic
accuracy is strictly theoretically defined, it is most synonymous with the
component empathic understanding. When empathic accuracy is operationally
defined for the purpose of empirical study, the definition should also include the
empathic expression component (see Ickes, 1993). The rationale behind this is
that it is only when a subjects empathic understanding is expressed in some form
that we are able to assess it (Ickes et al., 1997). From this, Ickes (1993) reasoned
that the most unequivocal way to measure empathic accuracy is by comparing the
content of a targets actual thought or feeling with the content of the perceivers
inference. Empathic accuracy is therefore best defined as the degree to which
someone is able to infer accurately the specific content of another persons thoughts
and feelings (Ickes, Stinson, Bissonnette, & Garcia, 1990).
From Flavells perspective (1992), the term empathy covers the four
knowledge components. Furthermore, subjects have to acquire at least the first
three components of Flavell if they want to come up to empathic understanding.
Figure 1 shows an overview. As mentioned before, there is however a difference
between the theoretical concepts and the operationalisation of these concepts. This
brings us to the section below.
Chapter 8
196
Figure 1
Flavells (1992) knowledge components
COGNITIVE
AFFECTIVE
Theory of mind (narrow use)

Theory of mind (broad use)
Mind-reading
EXISTENCE
Empathic Understanding
Strange Stories
Eyes Task
Standard/Dyadic EA Tasks
Eyes Task
a
Standard/Dyadic EA Tasksa

Mind-reading
NEED
Strange Stories

Mind-reading
INFERENCE
Strange Stories
Eyes Task
Standard/Dyadic EA Tasks
APPLICATION
[Dyadic EA Task] a
Eyes Task
a
Standard/Dyadic EA Tasksa
[Dyadic EA Task] a
EA Task: Empathic Accuracy Task
The grey zones refer to theoretical concepts, the white zones refer to operationalisations
General discussion
197
Considering the empathic accuracy design

In this thesis, we considered the empathic accuracy design to be a valuable way
to assess mind-reading impairments. The question however is whether an impaired
performance on an Empathic Accuracy Task indeed reflects mind-reading
impairments.
Independent of the methodological considerations, the findings on the traditional
mind-reading tasks (Strange Stories and Eyes Tasks), the standard Empathic
Accuracy Tasks and the dyadic Empathic Accuracy Task seem to indicate that
adults with PDD have no problems with the existence component (Flavell, 1992):
adults with PDD seem to know that mental states exist. Since no between-group
differences were found on the traditional mind-reading tasks and mixed results
were found on the Empathic Accuracy Tasks, the findings might suggest that some
adults with PDD have acquired -at least to some extent- the inference component
of Flavell (1992). Furthermore, the Eyes Tasks and the Empathic Accuracy Tasks
(the standard ones as well as the dyadic one) provide no information about Flavells
(1992) need component. However, the findings on the Strange Stories seem to
suggest that adults with PDD are as likely as control adults to use mental
justifications (see Chapter 2), which can be considered as an indication that adults
with PDD have no problems with this component. What both the traditional mindreading tasks and the Empathic Accuracy Tasks are unable to say is whether adults
with PDD have problems with the application component (Flavell, 1992). For
instance, inherent to the methodology of the empathic accuracy design, participants
are permitted to use as much time as needed to infer the thoughts and feelings of
other persons. Indeed, the empathic accuracy findings suggested that adults with
PDD needed more time than control adults to infer other persons thoughts and
feelings (Chapter 2 and 4). However, in real life, thoughts and feelings have to be
inferred instantly. This would imply that the empathic accuracy design is easier
than normal life. On the one hand, one might therefore suggest that the Empathic
Accuracy Tasks tap no more knowledge components than the traditional mindreading tasks do. On the other hand, it should be noted that the dyadic Empathic
Accuracy Task provided at least to some extent information about Flavells
application component, since this design enabled us to explore whether adults with
198
Chapter 8
PDD behave themselves socially adequate during an initial conversation (Chapter

3). In Figure 1, we therefore putted the dyadic Empathic Accuracy Task between
brackets.
Based on the theoretical considerations, one should further expect that the
Empathic Accuracy Tasks pay attention to the cognitive as well as the emotional
component of empathy. However, similar to the traditional mind-reading tasks, the
focus of the Empathic Accuracy Tasks is on the cognitive component. Together
with the paragraph above, this provides evidence that the Empathic Accuracy Tasks
measure actually mind-reading in a more naturalistic way, than that they measure
empathy. Nevertheless, it should be noted that we made an attempt to measure the
valence accuracy, which we consider to be part of the emotional component
(Chapter 4). We have to admit that this attempt was rather unsuccessful, since the
valence accuracy scores were only different from chance in one of the two
videotapes. A challenge for future research might be to make the emotional
component of the empathic accuracy design more operational.
Looking at the issue from a different point of view, it can also be argued that the
Empathic Accuracy Tasks tap more components than traditional mind-reading tasks
do. Whereas the Eyes Task is a relatively pure theory of mind task deprived of
contextual information (Baron-Cohen, Jolliffe, Mortimore, & Robertson, 1997), the
Empathic Accuracy Task uses movies, for which executive function and central
coherence components as well as mind-reading abilities are required. Therefore,
one might suggest that the decreased performance of the subjects with PDD on the
Empathic Accuracy Task can be accounted for by problems with executive
functioning or central coherence, rather than mind-reading. If this is the case, then
one possible solution might be to compare the performance of subjects with PDD
with other clinical groups, such as subjects with ADHD and Tourette syndrome.
Both clinical groups have some considerable similarities with subjects with PDD.
They too have impaired peer relations, both in childhood and adulthood (Gillberg,
1996), and the disorders are all postulated to involve executive functioning
abnormalities (Baron-Cohen, Cross, Crowson, & Robertson, 1994; Ozonoff &
Jensen, 1999; Pennington & Ozonoff, 1996; Sheppard, Bradshaw, Purcell, &
Pantelis, 1999). However, an important difference is that there is no evidence to
assume that children with ADHD and Tourette suffer from mind-reading
General discussion
199
impairments (e.g. Baron-Cohen et al., 1997; Dyck, Ferguson, & Shochet, 2001;
Perner, Kain, & Barchfeld, 2002).
At the moment, we are conducting a study with adults with PDD, ADHD,
Tourette syndrome and typically developing adults, in which we administer the
mind-reading tasks used in the study of Roeyers et al. (2001) and Chapter 2.
Although adults with PDD are not (yet) involved in our dataset, our preliminary
results suggest no significant differences between adults with ADHD, TS and
typically developing adults on the static mind-reading tasks as well as the Empathic
Accuracy Task. The fact that the latter task does not distinguish between the control
group and the clinical groups might suggest that the differences between adults
with PDD and control adults (see Roeyers et al. (2001) and Chapter 2) can not be
put down solely to account of components that are different from mind-reading. In
a way, the preliminary results are consistent with studies in which children with
ADHD and Tourette showed unimpaired mind-reading abilities (Dyck et al., 2001;
Perner et al., 2002). Still, the extent to which the Empathic Accuracy Task taps
other components, such as executive function or central coherence, remains
indistinct. In further studies it might therefore be interesting to assess the relation
between executive function, empathic accuracy and central coherence. Several
studies have found that individuals with relatively strong executive function
abilities perform better on theory of mind tasks than those with weaker executive
abilities (Carlson & Moses, 2001; Carlson, Moses, & Breton, 2002; Frye, Zelazo, &
Palfai, 1995; Hughes, 1998; Hughes, Dunn, & White, 1998; Perner et al., 2002),
although it is not always clear which particular aspects of executive function are
involved. Furthermore, there is evidence for a link between performance on theory
of mind tasks and performance on accepted tasks of central coherence bias (BaronCohen & Hammer, 1997a; Jarrold, Butler, Cottington, & Jeminez, 2000). The
question however remains how strong the three components are linked to one
another.
Correlates of empathic accuracy

One of the most challenging issues of this thesis was the search for reliable
indicators for individual differences in empathic accuracy. There are two ways to
200
Chapter 8
deal with this. One way is to correlate performance-based empathic ability

measures with empathy-related questionnaire measures (Davis & Kraus, 1997;
Eisenberg et al., 1997; Ickes, 2003). So, when we examined the mind-reading
ability of adults with PDD and control adults with the use of self- and other-report
questionnaire measures (Chapter 5), the results on both measures suggested that the
perceived mind-reading is more hampered in adults with PDD than in typically
developing adults. However, when we correlated the subjects empathic accuracy
scores with their perceived mind-reading performance (Chapter 7), we failed to find
an unequivocal pattern in the associations between empathic accuracy scores and
empathy-related dimensions. In a way, these findings are consistent with studies
with typically developing people, which failed to find self-report questionnaire
measures that can predict participants empathic accuracy ability (Davis & Kraus,
1997; Ickes, 2003).
A second way to deal with this issue, is to look at those people whose mindreading ability is most impaired (i.e. subjects with PDD). It can be suggested that a
self-report questionnaire that is based on the specific characteristics of those people
would strongly correlate with empathic accuracy. Recently, Baron-Cohen,
Wheelwright, Skinner, Martin and Clubley (2001) developed the Autism Spectrum
Quotient (AQ), a self-report questionnaire that specifically aims at measuring the
degree to which someone shows autistic traits. As described in Chapter 7, the AQ
clearly distinguished between subjects with and without PDD. However, we failed
to find significant associations between the empathic accuracy scores and the AQ,
with the exception of the significant association for the control group between the
subscale Imagination and the empathic accuracy score on Video 1 (see Chapter 7,
p. 181). In future studies, it would be interesting to investigate which characteristics
of adults with PDD are to transpose into questionnaire measures, and to examine
whether these characteristics can predict subjects empathic accuracy scores.
PRACTICAL IMPLICATIONS
It is compelling to question what implications the findings of this thesis have for
the assessment and treatment of subjects with PDD.
General discussion
201
Assessment. As described in the different chapters, instruments are needed that

can measure the mind-reading impairments of adults with PDD, and -as a resultcan distinguish adults with PDD from typically developing adults and other clinical
subjects. From the start of this thesis, we looked forward to find a reliable and valid
mind-reading task for use with adults with PDD. Since the traditional mind-reading
tasks (e.g. Strange Stories, Eyes Task) were not able to capture the more subtle
impairments of adults with PDD, our hope was placed on the empathic accuracy
design, which -indeed- proved to be a valid alternative. Therefore, the use of the
empathic accuracy design has potential for diagnostic purposes. However, as time
and research progressed, it became also clear that an Empathic Accuracy Task is
time consuming to administer and to calculate. Since the diagnostic process already
requires a considerable amount of time (mostly up to 4 days for one diagnosis of
PDD), the question is rather whether clinicians will make time and effort to fit this
instrument into the process. In the future, the challenge will be to make an
Empathic Accuracy Task that is normed and oriented towards diagnostic settings,
and that can be administered and calculated within reasonable time.
Nevertheless, one might suggest that quick and efficient measures are needed
and/or more appropriate. If it comes to that, then the use of questionnaires, such as
the Interpersonal Reactivity Index and the Autism Spectrum Quotient (Chapter 5
and 7), can be considered. On the one hand, we value self-administered
questionnaires for reasons mentioned above. On the other hand, if questionnaires
are used, then we believe that they should be at least normed for typically
developing as well as autistic populations. For this purpose, we are administering
the AQ and IRI from a larger sample with and without PDD. It should however be
borne in mind that questionnaires measure subjects perception about their
performance and not their actual performance. Furthermore, questionnaire
measures will never reach the level of that of performance-based diagnostic
instruments. Presumably, they will reach -at the most- the level of screening
instruments.
Treatment. The bright side of this thesis is that some high-functioning adults
with PDD performed relatively well on mind-reading measures. This was best
demonstrated in the dyadic interaction study (Chapter 3), which suggested that
some adults with PDD are able to read the thoughts and feelings of their interaction
202
Chapter 8
partner, and that the social interaction between subjects with and without PDD was
characterized by a level of behavioural reciprocity. So, at least some individuals
with PDD can attain high-levels of mind-reading skills, which strengthens the
argument that these skills can be taught. We believe that the empathic accuracy
design can play a part in this process.
However, two lessons can be learned from previous studies that tried to teach
children with PDD about mental states in the hope of improving social
understanding (Parsons & Mitchell, 2002). On the one hand, most studies reported
an improvement on the specific skill or task being taught. On the other hand, there
were frequent failures to generalize knowledge to different tasks or to real life
situations (e.g. Hadwin, Baron-Cohen, Howlin, & Hill, 1997; Ozonoff & Miller,
1995; Swettenham, 1996). With regard to the latter, a training based on the
empathic accuracy approach might have an advantage, since empathic accuracy is
dependent on characteristics of the target, the perceiver, and the unique interplay
between both of them. So, whereas prototypical situations are generally used to
train people, a training based on the empathic accuracy approach will learn people
with PDD that it is not only the specificity of a situation, but also the specificity of
a relationship that influences someones thoughts and feelings, and behaviour. As a
result, it is opportune to involve the individual with PDD as well as all those who
have a significant relationship with him or her (e.g. mother, father, brothers or
sisters, caretakers, colleagues and so on). As such, we can -for instance- learn Jack,
a computer expert of a company, that if he forgets to install a virus firewall, most of
his colleagues will blame him for that, but -considering other situations in which
something went wrong- it was always colleague Dave who gave him a roasting, so
this time -most probably- it will also be Dave who will give him a dressing-down.
Consequently, the focus is not only on the situation itself, but also on the unique
relationship between Jack and each of his colleagues. In addition, since the findings
of the study reported in Chapter 4 suggest that subjects with PDD benefit from
structure, the use of clear, simple and concise guidelines is recommended in social
intervention programs (see also Howlin, 1997, 2003; Howlin & Yates, 1999; Kunce
& Mesibov, 1998; Mawhood & Howlin, 1999; Mesibov, 1992).
General discussion
203
DIRECTIONS FOR FUTURE RESEARCH

In addition to the potential improvements to the empathic accuracy design, there
are a few lines for future research specific to the population of persons with PDD.
Development in mind-reading performance. In Chapter 4 (p. 104-105), we
argued that the mind-reading performance of high-functioning individuals with
PDD is subject to disparity in age. While the majority of children with PDD fail to
pass traditional and more advanced mind-reading tasks, the greater part of normally
intelligent adults with PDD have no difficulties with such tasks. Therefore, one
might suggest that the mind-reading impairments of adults with PDD are more
subtle than those of children with PDD. However, it is still controversial whether
adults with PDD camouflage their mind-reading impairments or whether their
mind-reading skills improved spontaneously with growing age. Therefore, it seems
appropriate to administer traditional and more naturalistic mind-reading tasks from
teenagers with PDD. It can be suggested that the use of naturalistic tasks would
clarify even more the mind-reading impairments in teenagers with PDD than those
in adults with PDD.
Gender differences. In this thesis, we paid no attention to gender differences.
Interestingly, some researchers have suggested that autism is an extreme form of
the male brain (Baron-Cohen, 2002; Baron-Cohen & Hammer, 1997b; BaronCohen, Wheelwright, Griffin, Lawson, & Hill, 2002; Baron-Cohen, Wheelwright,
Stone, & Rutterford, 1999). The extreme male brain (EMB) theory of autism
predicts that on any test of empathising (i.e. the attribution of mental states to
other people, and an appropriate affective response to others affective state),
typically developing males will score lower than typically developing females, and
performance by subjects with PDD will be even lower than typically developing
males (Baron-Cohen & Wheelwright, 2003, p. 510). Similarly, the EMB theory
predicts that on any test of systemizing (i.e. the drive to analyse the variables in a
system, to derive the underlying rules that govern the behaviour of a system),
typically developing males will score higher than typically developing females, and
performance by subjects with PDD will be intact or even higher than typically
developing males (Baron-Cohen, 2002; Baron-Cohen & Wheelwright, 2003).
Evidence for the EMB theory derives from several studies that -for the most part-
Chapter 8
204
made use of questionnaire measures (e.g. Baron-Cohen et al., 2001; Baron-Cohen,

Richler,
Bisarya,
Gurunathan,
&
Wheelwright,
2003;
Baron-Cohen
&
Wheelwright, 2003). It is interesting that none of these studies found gender

differences in adults with PDD. Baron-Cohen and Wheelwright (2003) attributed
this to the relatively small number of female subjects with PDD that was tested1.
In a way, the EMB theory is consistent with the widespread stereotype that
women are more empathic than men. However, a considerable amount of studies
with typically developing people provided evidence against this stereotypic
viewpoint. In a review of gender differences in empathy, Eisenberg and Lennon
(1983) suggested that women are not more empathic than men, but they think that
they are more motivated than men. They found that women report greater selfperceived empathy when questionnaire measures are used, but that gender
differences are generally not found when performance-based tasks are used
(Eisenberg & Lennon, 1983; Lennon & Eisenberg, 1987). In addition, it has been
suggested that women view empathic skills as more important to their self-concept
(Gilligan, 1982; Zahn-Waxler, Cole, & Barret, 1991), and -as such- are more
motivated than men to present themselves as being empathic (Graham & Ickes,
1997; Klein & Hodges, 2001). So, when a gender difference does occur, it is
primarily one of differential motivation rather than differential ability (Graham et
al., 1997; Ickes, Graham, & Gesn, 2000).
This perspective, together with the EMB theory, leads to several questions for
future research. Based on the findings with empathy-related questionnaires that
females score higher than males, and that males score higher than subjects with
PDD (Baron-Cohen et al., 2001; Baron-Cohen, Richler, et al., 2003; Baron-Cohen
& Wheelwright, 2003), it would be interesting to find out whether the same pattern
will occur if performance-based tasks (e.g. the empathic accuracy design) are used.
Furthermore, if autism is indeed an extreme form of the male brain (Baron-Cohen,
2002), is it possible to reduce empathic accuracy differences between subjects with
PDD and typically developing subjects by extremely motivating subjects with
PDD, or -on the contrary- will between-group differences increase if we can
Note that the average male/female ratio in autism is 5.75:1, varying according to the absence or presence of
mental retardation (Fombonne, 2003).
General discussion
205
motivate control subjects more than subjects with PDD? If we are able to assess the
perceived and performed mind-reading ability of a large, female sample with PDD,
will there be a difference in performance compared to a male sample with PDD?
And if so, what will be the relationship between men and women with PDD, and
men and women without PDD?
Intra-group differences. As we described above, the findings suggest that at
least some adults with PDD are (relatively) able to infer the thoughts and feelings
of other people. It would therefore be interesting to investigate in a retrospective
way- to what extent these top adults with PDD differ from the worst performing
adults with PDD. For instance: Is it possible that adults who received a late
diagnosis performed better than those who received an early diagnosis? What
about the impact of previous social skill training? Are there any differences with
regard to the construction of the family (e.g. the more brothers or sisters one adult
with PDD has, the better he performs)? If we are able to map one or more variables
that improve or influence the mind-reading performance of subjects with PDD, then
these variables can be used as clues for intervention purposes.
Furthermore, a constraint that applies to the empathic accuracy studies (Chapter
2 to 4) as well as the other studies (Chapter 5 and 6) is that the criteria that we used
to recruit our samples are arbitrary, and therefore can be subject to criticism. In
each of our studies (Chapter 2 to 6), we mapped the age, sex and intelligence scores
of the subjects with and without PDD, and -except for the studies in Chapter 2 and
3- we matched the subjects with PDD on a one-to-one basis with control subjects
on each of these variables. In the study reported in Chapter 3, the PDD subjects
were matched with the control subjects on chronological age, sex, level of
education and main interests (e.g. music, computer and internet, and so on). In
future studies, it might be interesting to match our subjects (also) on variables other
than these ones (e.g. speed of processing2). Another possibility would be to get a
large and diverse control population, so that norms can be obtained.
The model of the minimal cognitive architecture views intelligence as an innate capacity that is determined
by the speed of a basic processing mechanism (Anderson, 1992). The model addresses general intelligence
from the traditional individual differences point of view and, in addition, from a uniquely developmental
perspective (Anderson, 2001; Scheuffgen, Happ, Anderson, & Frith, 2000).
206
Chapter 8
REFERENCES
Anderson, M. (1992). Intelligence and development: A cognitive theory. Oxford:
Blackwell.
Anderson, M. (2001). Annotations: Conceptions of intelligence. Journal of Child
Baron-Cohen, S. (2002). The extreme male brain theory of autism. Trends in Cognitive
Sciences, 6, 248-254.
Baron-Cohen, S., Cross, P., Crowson, M. & Robertson, M. (1994). Can children with
Gilles de la Tourette syndrome edit their intentions? Psychological Medicine, 24, 2940.
Baron-Cohen, S., & Hammer, J. (1997a). Parents of children with Asperger syndrome:
What is the cognitive phenotype. Journal of Cognitive Neuroscience, 9, 548-554.
Baron-Cohen, S., & Hammer, J. (1997b). Is autism an extreme form of the male brain?
Advances in Infancy Research, 11, 193-217.
Baron-Cohen, S., Leslie, A., & Frith, U. (1985). Does the autistic child have a theory of
mind? Cognition, 21, 37-46.
Baron-Cohen, S., Leslie, A., & Frith, U. (1986). Mechanical, behavioural and intentional
understanding of picture stories in autistic children. British Journal of Developmental
Baron-Cohen, S., Richler, J., Bisarya, D., Gurunathan, N., & Wheelwright, S. (2003). The
systemizing quotient: An investigation of adults with Asperger syndrome or highfunctioning autism, and normal sex differences. Philosophical Transactions of the
Royal Society of London Series B-Biological Sciences, 358, 361-374.
Baron-Cohen, S., & Wheelwright, S. (2003). The Friendship Questionnaire: An
investigation of adults with Asperger syndrome or high-functioning autism, and normal
sex differences. Journal of Autism and Developmental Disorders, 33, 509-517.
Baron-Cohen, S., Wheelwright, S., Griffin, R., Lawson, J., & Hill, J. (2002). The exact
mind: Empathising and systemising in autism spectrum conditions. In U. Goswami
(Ed.), Handbook of Cognitive Development. Oxford: Blackwells.
General discussion
207
autism, males and females, scientists and mathematicians. Journal of Autism and
Baron-Cohen, S., Wheelwright, S., Stone, V., & Rutterford, M. (1999). A mathematician,
a physicist, and a computer scientist with Asperger syndrome: Performance on folk
psychology and folk physics test. Neurocase, 5, 475-483.
Barret-Lennard, G. T. (1981). The empathy cycle: Refinement of a nuclear concept.
Journal of Counseling Psychology, 39, 333-339.
Boucher, J. (1989). The theory of mind hypothesis of autism: Explanation, evidence and
assessment. British Journal of Disorders of Communication, 24, 181-198.
Carlson, S., & Moses, L. (2001). Individual differences in inhibitory control and childrens
theory of mind. Child Development, 72, 1032-1053.
Carlson, S., Moses, L., & Breton, C. (2002). How specific is the relation between
executive function and theory of mind? Contributions of inhibitory control and working
memory. Infant Child Development, 11, 73-92.
Charman, T., & Baron-Cohen, S. (1992). Understanding beliefs and drawings: A further
test of the metarepresentation theory of autism. Journal of Child Psychology and
Psychiatry, 33, 1105-1112.
Charman, T., & Baron-Cohen, S. (1995). Understanding models, photos, and beliefs: A
test of the modularity thesis of metarepresentation. Cognitive Development, 10, 287298.
Corcoran, R. (2000). Theory of mind in other clinical conditions: Is a selective theory of
mind deficit exclusive to autism? In S. Baron-Cohen, H. Tager-Flusberg and D. Cohen
(Eds.), Understanding other minds. Perspectives from developmental cognitive
neuroscience. Second Edition. (pp. 391-421). Oxford: University Press.
Davis, M. H. (1994). Empathy: A Social Psychological Approach. Madison, WI: Brown
and Benchmark.
Davis, M., & Kraus, L. (1997). Personality and empathic accuracy. In W. Ickes (Ed.),
Empathic accuracy (pp.114-168). New York: Guilford Press.
Dyck, M., Ferguson, K., & Shochet, I. (2001). Do autism spectrum disorders differ from
each other and from non-spectrum disorders on emotion recognition tests? European
Child and Adolescent Psychiatry, 10, 105-116.
208
Chapter 8
Eisenberg, N., & Lennon, R. (1983). Sex differences in empathy and related capacities.
Psychological Bulletin, 94, 100-131.
In W. Ickes (Ed.), Empathic accuracy (pp. 73-116). New York: Guilford Press.
Elliott, R., Filipovich, H., Harrigan, L., Gaynor, J., Reimschuessel, C., & Zapadka, J.K.
(1982). Measuring response empathy: The development of a multi-component rating
scale. Journal of Counseling Psychology, 29, 379-387.
Flavell, J. (1974). The development of inferences about others. In T. Mischel (Ed.),
Understanding other persons. Totowa, NJ: Rowan & Littlefield.
Flavell, J. (1992). Perspectives on perspective taking. In H. Beilin and P. Pufall (Eds.),
Piagets theory: Prospects and possibilities (pp. 107-139). Hillsdale, NJ: Erlbaum.
Flavell, J., Botkin, P., Fry, C., Wright, J., & Jarvis, P. (1968). The development of roletaking and communication skills in young children. New York: Wiley.
Fombonne, E. (2003). Epidemiological surveys of autism and other pervasive
developmental disorders: An update. Journal of Autism and Developmental Disorders,
33, 365-382.
Frye, D., Zelazo, P., & Palfai, T. (1995). Theory of mind and rule-based reasoning.
Cognitive Development, 10, 483-527.
Gillberg, C. (1996). The long-term outcome of childhood empathy disorders. European
Child and Adolescent Psychiatry, 5, 52-56.
Gilligan, C. (1982). In a different voice: Psychological theory and womens development.
Cambridge, MA: Harvard University Press.
Graham, T., & Ickes, W. (1997). When womens intuition isnt greater than mens. In W.
Ickes (Ed.), Empathic Accuracy (pp. 117-143). New York: Guilford Press.
Hadwin, J., Baron-Cohen, S., Howlin, P., & Hill, K. (1997). Does teaching theory of mind
have an effect on the ability to develop conversation in children with autism? Journal
of Autism and Developmental Disorders, 27, 519-537.
Howlin, P. (2003). Outcome in high-functioning adults with autism with and without early
language delays: Implications for the differentiation between autism and Asperger
syndrome. Journal of Autism and Developmental Disorders, 33, 3-13.
Howlin, P., & Yates, P. (1999). The potential effectiveness of social skills groups for
adults with autism: Information update. Autism, 3, 299-307.
General discussion
209
Hughes, C. (1998). Executive function in preschoolers: Links with theory of mind and
verbal ability. British Journal of Developmental Psychology, 34, 1326-1339.
Hughes, C., Dunn, J., & White, A. (1998). Trick or treat? Uneven understanding of mind
and emotion and executive dysfunction in "hard-to-manage" preschoolers. Journal of
Ickes, W., Gesn, P., & Graham, T. (2000). Gender differences in empathic accuracy:
Differential ability or differential motivation? Personal Relationships, 7, 95-109.
Ickes, W., Marangoni, C., & Garcia, S. (1997). The development of empathic accuracy. In
W. Ickes (Ed.), Empathic accuracy (pp. 251-281). New York: Guilford Press.
59, 730-742.
Jarrold, C., Butler, D., Cottington, E., & Jeminez, F. (2000). Linking theory of mind and
central coherence bias in autism and in the general population. Developmental
Psychology, 36,126-138.
Keefe, T. (1979). The development of empathic skill: A study. Journal of Education for
Social Work, 15, 30-37.
Klein, K., & Hodges, S. (2001). Gender differences, motivation, and empathic accuracy:
When it pays to understand. Personality and Social Psychology Bulletin, 27, 720-730.
Kunce, L., & Mesibov, G. (1998). Educational approaches to high-functioning autism and
Asperger syndrome. In E. Schopler, G. Mesibov, and L. Kunce (Eds.), Asperger
syndrome or high functioning autism ? (pp. 227-262). New York: Plenum.
Leekam, S., & Perner, J. (1991). Does the autistic child have a metarepresentational
deficit? Cognition, 40, 203-218.
Lennon, R., & Eisenberg, N. (1987). Gender and age differences in empathy and
sympathy. In N. Eisenberg & J. Strayer (Eds.), Empathy and its development:
Cambridge studies in social and emotional development (pp. 195-217). New York:
University Press.
Leslie, A., & Frith, U. (1987). Metarepresentation and autism: How not to lose ones
marbles. Cognition, 27, 291-294.
210
Chapter 8
Mawhood, L., & Howlin, P. (1999). The outcome if a supported employment scheme for
high-functioning adults with autism or Asperger syndrome. Autism, 3, 229-254.
Mesibov, G. (1992). Treatment issues with high-functioning adolescents and adults with
autism. In E. Schopler and G. Mesibov (Eds.), High-functioning individuals with autism
(pp. 143-156). New York: Plenum.
Ozonoff, S., & Jensen, J. (1999). Brief report: Specific executive function profiles in three
neurodevelopmental disorders. Journal of Autism and Developmental Disorders, 29,
171-177.
Ozonoff, S., & Miller, J. (1995). Teaching theory of mind: A new approach to social skills
training for individuals with autism. Journal of Autism and Developmental Disorders,
25, 415-433.
Parsons, S., & Mitchell, P. (2002). The potential of virtual reality in social skills training
for people with autistic spectrum disorders. Journal of Intellectual Disability Research,
46, 430-443.
Pennington, B., & Ozonoff, S. (1996). Executive functions and developmental
psychopathology. Journal of Child Psychology and Psychiatry, 37, 51-87.
Perner, J., Frith, U., Leslie, A., & Leekam, S. (1989). Exploration of the autistic childs
theory of mind: Knowledge, belief, and communication. Child Development, 60, 689700.
Perner J., Kain W., & Barchfeld, P. (2002). Executive control and higher-order theory of
mind in children at risk of ADHD. Infant and Child Development, 11, 141-158.
Piaget, J. (1929). The child's conception of the world. New York: Harcourt, Brace.
Premack, D., & Woodruff, G. (1978). Does the chimpanzee have a theory of mind?
Behavioural and Brain Sciences, 4, 515-526.
Preston, S., & de Waal, F. (2002). Empathy: Its ultimate and proximate bases. Behavioural
and Brain Sciences, 25, 1-72.
Roeyers, H., Buysse, A., Ponnet, K., & Pichal, B. (2001). Advancing Advanced Mindreading Tests: Empathic Accuracy in Adults with a Pervasive Developmental Disorder.
Scheuffgen, K., Happ, F., Anderson, M., & Frith, U. (2000). High intelligence, low
IQ? Speed of processing and measured IQ in children with autism. Development and
Psychopathology, 12, 83-90.
General discussion
211
Sheppard, D., Bradshaw, J., Purcell, R., & Pantelis, C. (1999). Tourettes and comorbid
syndromes: Obsessive compulsive and attention deficit hyperactivity disorder. A
common aetiology ? Clinical Psychology Review, 19, 531-552.
Swettenham, J. (1996). Can children with autism be taught to understand false belief using
computers? Journal of Child Psychology and Psychiatry, 37, 157-166.
Thomas, G., & Fletcher, G. (1997). The development of empathic accuracy. In W. Ickes
(Ed.), Empathic accuracy (pp.194-217). New York: Guilford Press.
Wimmer, H., & Perner, J. (1983). Beliefs about beliefs: Representation and constraining
function of wrong beliefs in young childrens understanding of deception. Cognition,
13, 103-128.
Zahn-Waxler, C., Cole, P., & Barrett, K. (1991). Guilt and empathy: Sex differences and
implications for the development of depression. In J. Garber and K. Dodge (Eds.), The
development of emotion regulation and dysregulation (pp. 234-272).Cambridge, UK:
Cambridge University Press.

Rug01-000813602 2010 0001 Ac

Загружено:

Сведения о документе

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Rug01-000813602 2010 0001 Ac

Загружено:

Авторское право:

Доступные форматы

Mind-reading in adolescents and adults

with a pervasive developmental disorder

Promotor: Prof. Dr. Herbert Roeyers

Proefschrift ingediend tot het behalen van de academische graad

Gutta cavat lapidem non vi, sed saepe cadendo.

Advanced mind-reading in adults with Asperger

Empathic accuracy in adults with a pervasive

developmental disorder during an unstructured

Empathic accuracy in young adults with PDD:

Does structure matter ?

Perceived and performed mind-reading in adults

with a pervasive developmental disorder

Schematic knowledge and levels of processing

in young adults with PDD

Alternative explanations to the empathic accuracy

PERVASIVE DEVELOPMENTAL DISORDERS

having a conversation of the getting acquainted type. Consequently, it can be

Clercq, Buysse, Roeyers, Ickes, Ponnet, & Verhofstadt, 2001), a system

Empathic Accuracy Task (dyadic interaction design)

Interpersonal Reactivity Index (self- & other-report)

Emotional Quotient Inventory (self-report)

New Empathic Accuracy Task (standard design)

Revised Eyes Task

Levels of Processing Task

Chapter 3. In this chapter, we switch from the standard stimulus empathic

deficit/hyperactivity disorder. Philosophical Transactions of the Royal Society of

Happ, F. (1994). An advanced test of theory of mind: Understanding of story characters

Advanced mind-reading in adults with

press). Advanced mind-reading in adults with Asperger syndrome. Autism.

Empathic accuracy in adults with Asperger syndrome

Empathic accuracy in adults with Asperger syndrome

Empathic accuracy in adults with Asperger syndrome

Empathic accuracy in adults with Asperger syndrome

Empathic accuracy in adults with Asperger syndrome

Empathic accuracy in adults with Asperger syndrome

Eyes Photographs Task

*p < .05; **p < .01; *** p < .005

Empathic Accuracy Task

*p < .05; **p < .01; ***p < .005

Differences between both Groups in Participants Inferred Thoughts and

Empathic accuracy in adults with Asperger syndrome

Empathic accuracy in adults with Asperger syndrome

Empathic Accuracy Task

Eyes Photograph Task

*p < .05; **p < .01; ***p < .005

Correlates of the Mind-Reading Tasks

Empathic accuracy in adults with Asperger syndrome

Empathic accuracy in adults with Asperger syndrome

psychology: Research methods in personality and social

psychology, Vol. 11, pp. 16-44. Newbury Park, CA: Sage.

Empathic accuracy in adults with Asperger syndrome

Empathic accuracy in adults with Asperger syndrome

Tangible environmental object

Empathic accuracy in adults with a

Empathic accuracy in adults with PDD during a dyadic conversation

naturalistic than static mind-reading tasks since it allows participants to make

Empathic accuracy in adults with PDD during a dyadic conversation

Empathic accuracy in adults with PDD during a dyadic conversation

Setting and equipment

Empathic accuracy in adults with PDD during a dyadic conversation

Empathic accuracy in adults with PDD during a dyadic conversation

Actual thought/feeling measures

Empathic accuracy in adults with PDD during a dyadic conversation

p < .05; p < .01; p < .005

p < .05; p < .01; p < .005

p < .05; p < .01; p < .005

p < .05, p < .01, p < .005

p < .05, p < .01, p < .001