Вы находитесь на странице: 1из 6

Behaviour Research and Therapy 48 (2010) 38–43

Contents lists available at ScienceDirect

Behaviour Research and Therapy


journal homepage: www.elsevier.com/locate/brat

Fear generalization in humans: Impact of verbal instructions


Bram Vervliet a, b, *, Merel Kindt a, Debora Vansteenwegen b, Dirk Hermans b
a
University of Amsterdam, Roetersstraat 15, 1018 WB Amsterdam, The Netherlands
b
University of Leuven, Belgium

a r t i c l e i n f o a b s t r a c t

Article history: Fear generalization lies at the heart of many anxiety problems, but little is known about the factors that can
Received 26 June 2009 influence this phenomenon. The present study investigated whether verbal instructions about specific
Received in revised form stimulus features can influence conditioned fear generalization. All participants were fear conditioned to
2 September 2009
a yellow triangle, using an electric shock. Participants had received pre-experimental instructions saying
Accepted 5 September 2009
that the shapes (or colours) of the stimuli were informative for the occurrence of shock (group Shape
and group Colour, respectively). Next, generalization was tested to presentations of a blue triangle (same
Keywords:
shape) as well as a yellow square (same colour). Fear reactions were measured through skin conductance
Fear conditioning
Fear generalization and online ratings of shock-expectancy. The results showed strongest generalization to the same shape
Associative learning stimulus in group Shape, versus the same colour stimulus in group Colour. Hence, the same learning
Psychophysiology experience can have opposite effects in terms of fear generalization, depending on verbally transmitted
Experimental psychopathology information about the relative importance of individual stimulus features.
Ó 2009 Elsevier Ltd. All rights reserved.

Fear generalization is an important characteristic of various Hermans, & Eelen, 2005; Vervliet, Vansteenwegen, & Eelen, 2004).
anxiety disorders. In the case of post-traumatic stress disorder If we want to understand the process of fear generalization, it is
(PTSD), for example, strong fear reactions are typically elicited by important to know what factors influence it. The present study
a wide variety of cues that bare only vague similarity with the actual investigated the impact of verbally transmitted information about
trauma cues (Ehlers & Clark, 2000; Feldner, Monson, & Friedman, individual stimulus features.
2007). This unbridled generalization clearly augments the impact Generalization occurs when a conditioned response is elicited by
of the traumatic event in the daily life of the victim, and may well a stimulus different from the conditioned stimulus (CS). Thus, when
contribute to the development and/or maintenance of the disorder. a yellow triangle is paired with an electric shock (the unconditioned
One could even argue that the fear learning mechanism itself is stimulus, US) and starts eliciting fear reactions on future confron-
highly adaptive and promotes survival, but that fears become irra- tations, generalization occurs when the fear reaction is then also
tional when they are overgeneralized to realistically non-dangerous elicited by a blue triangle (the generalization stimulus, GS). The
cues (Feldner et al., 2007; Lissek et al., 2005). It follows that more process of generalization can be conceptualized in a variety of ways,
insight in the phenomenon and process of fear generalization will but one fruitful approach is to view stimuli as sets of features, which
enhance our understanding of anxiety problems in general and can individually enter the conditioning process (Rescorla, 1976).
may eventually inspire new ways for treatment and/or prevention Thus, when a stimulus AX consists of features A and X, conditioning
of anxiety disorders. Generalization has been studied widely in the experiences with that stimulus will affect both features A and X.
older animal and human conditioning literature (until the 1970s, Imagine that generalization is tested with stimulus BX that has
see Razran, 1949, Kalish, 1969, Honig & Urcuioli, 1981, and Guirlanda feature X in common; the amount of generalization from AX to BX
& Enquist, 2003, for reviews). However, it has only recently regained will then be defined by the amount of conditioning that occurred to
interest in the field of human anxiety (see Lissek et al., 2008, who X. In other words, the amount of generalization between two
developed an elegant procedure to examine interindividual differ- stimuli is defined by the amount of conditioning that has accrued to
ences in fear generalization; see also Dunsmoor, Mitroff, & LaBar, the features they have in common. Fear generalization from
2009; Hajcak et al., 2009; Vervliet, Vansteenwegen, Baeyens, a yellow to a blue triangle will depend on the fear conditioning that
has accrued to the feature ‘triangle’.
Hence, if we want to ask what factors influence the generaliza-
* Corresponding author at: University of Amsterdam, Roetersstraat 15, 1018 WB
tion between two stimuli, we must ask what factors determine the
Amsterdam, The Netherlands. Tel.: þ31 20 525 67 68; fax: þ31 20 639 13 69. amount of conditioning to their common feature(s). This is exactly
E-mail address: b.vervliet@uva.nl (B. Vervliet). the type of question that is central to contemporary associative

0005-7967/$ – see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.brat.2009.09.005
B. Vervliet et al. / Behaviour Research and Therapy 48 (2010) 38–43 39

learning research: When multiple stimuli (features) are present on in group Shape, versus more generalization to the yellow square in
a conditioning trial, what then determines the rate of conditioning group Colour. Finally, we also added a small test phase where we
to the different stimuli (features)? It was already shown by Pavlov presented the original CSs again (yellow triangle and black cross),
(1927) that a CS elicits less conditioned responding when it was each three times without shock. This extra test was included in order
accompanied by another stimulus during the conditioning episode to assess the survival of conditioned fear responding after extinction
(‘‘overshadowing’’). He further showed that the strength of this of both generalization stimuli. Previous research has indicated
effect depended on the perceptual salience of the accompanying that extinction of generalization stimuli has no impact on fear
stimulus: The more salient the accompanying stimulus, the weaker responding to the original CS (Vervliet et al., 2005, 2004). We had no
the conditioning will be to the target CS. Years later, Kamin (1969) differential predictions for this fear recovery test phase.
showed that the overshadowing effect is even enhanced when
the accompanying stimulus is conditioned on its own (‘‘blocking’’). Methods
That is, when the accompanying stimulus already carries sufficient
information about the occurrence of the US (e.g., shock), there is less Participants
need to learn additionally about the CS–US relationship (see, e.g.,
Rescorla & Wagner, 1972). Taken together, the amount of condi- Thirty-two students participated either to earn course credits or
tioning to the CS depends crucially on the salience and/or the a small monetary award. Participants were randomly assigned to the
informational relevance of the accompanying stimulus. two experimental groups (group Colour: N ¼ 16 of which 13 females,
This line of reasoning can also be applied to the generalization mean age ¼ 19.06, SD ¼ 1.34; group Shape: N ¼ 16 of which 9 females,
issue: the extent to which conditioning with a CS will accrue to its mean age ¼ 19.44, SD ¼ 2.63). They all gave informed consent and
unique or common features will depend on the relative salience were informed that they could decline to participate at any time.
and informational relevance of those features. When the unique
features are relatively more salient and/or have more informational Apparatus
relevance, little conditioning will accrue to the common features
and lead to weak generalization. Conversely, when the common An isosceles triangle with black outlines and coloured in yellow
features are relatively more salient and/or have more informational served as the CSþ for all participants. The CS was a black cross. One
relevance, most conditioning will accrue to the common features generalization test stimulus, GSc, was a square with black outlines and
and hence produce strong generalization. It follows that general- coloured in the same yellow. The other generalization test stimulus,
ization can be influenced by factors that change the relative salience GSs, was the same isosceles triangle as the CSþ, but coloured in blue.
and/or informational value of the different features of the CS. All stimuli were presented on a computer screen, located on eye-level
In the present study, we aimed at manipulating the relative in front of the participant at approximately 500 mm. The stimulus
salience/relevance of stimulus features through pre-experimental sequence, the presentation of the stimuli and the ITIs were controlled
instructions. Humans have the capacity to talk and to transmit by Affect3 software designed in our lab, which can be downloaded
information through verbal interaction. It has long been shown that freely (Hermans, Clarysse, Baeyens & Spruyt, 2002). A 2 ms electro-
verbal instructions can have a large impact on the conditioning of cutaneous stimulus delivered to the wrist of the left hand served as
fear reactions. Firstly, giving verbal information about the occur- unconditional stimulus (US). It was administered by a Digitimer DS7A
rence of shock is a powerful way to induce fear reactions (e.g., constant current stimulator (Hertfordshire, UK) via a pair of 11-mm
Phelps, O’Connor, Gatenby, Grillon, & Davis, 2001). Secondly, post- Fukuda Standard Ag/AgCl electrodes. The electrodes were filled
conditioning instructions can determine the amount of fear with K-Y Jelly. The intensity of the shock was individually selected to
responding to individual stimuli when these stimuli were condi- a level where it was ‘‘uncomfortable but not painful’’. Participants
tioned in compound (Lovibond, 2003). Thirdly, instructions that the were seated in an armchair in a sound attenuated experimental room,
US will no longer follow can speed up extinction of conditioned adjacent to the experimenter’s room.
fear (e.g., McNally, 1981; Soares & Ohman, 1993). Verbal instruc- Electrodermal activity was recorded using a skin conductance
tions can serve as an experimental tool to investigate the effects of coupler manufactured by Coulbourn Instruments (model V71-23,
cognitions (beliefs, ideas, assumptions, schemata) on experience- Allentown, PA). During skin conductance measurement, the coupler
based learning phenomena like classical conditioning. Differences applied a constant voltage of .5 V across a pair of sintered-pellet silver
in such cognitions may contribute to the differential impact of chloride electrodes (8 mm), attached to the hypothenar palm of the
traumatic experiences in different persons (leading to a chronic non-preferred hand, which was cleaned with tap water. The inter-
course of PTSD or not). In the present study, pre-experimental electrode distance was 7 mm. The electrodes were filled with K-Y Jelly.
instructions focused attention to a particular feature of the CS, The resulting conductance signal was submitted through a Labmaster
which was shared by one generalization test stimulus but not by DMA 12-bit analog-to-digital converter (Scientific Solutions, Solon,
the other. We expected that the instructions would promote fear Ohio) and digitized at 10 Hz from 2 s prior to CS onset until 6 s after CS
generalization to the stimulus with the shared feature. offset. Participants used their right hand to record their subjective
Thirty-two participants received exactly identical conditioning expectancy of a shock on a continuous, 180-degree, rotary dial placed
experiences, and differed only in the pre-experimental instructions. on the table in front of them. The left extreme of the semicircular dial
Participants from group Shape were told that the shape of the figures was labelled Certain no shock, the right extreme was labelled Certain
that they were about to see would be instructive for the occurrence shock and the intermediate point, at 90 degree, was labelled Uncertain.
of the electric shock (US), and that they had to learn to distinguish The rotary dial generated a voltage signal between 0 and 5 V, which
between ‘‘good’’ shapes and ‘‘bad’’ shapes. Participants from group was transmitted through a Labmaster DMA 12-bit analog-to-digital
Colour received similar instructions, but about the colour of the converter (Scientific Solutions, Solon, Ohio) and digitized at 10 Hz
figures. Next, all participants received 4 presentations of a yellow from 2 s prior to CS onset until 6 s after CS offset.
triangle (the CSþ) that co-terminated with shock. A black cross
functioned as the control CS (4 times presented, without shock). Procedure
Generalization was then tested to a blue triangle (same shape) and
a yellow square (same colour), which were both 6 times presented Following general instructions and completion of the informed
without shock. We expected more generalization to the blue triangle consent, participants were fitted with electrodes and were led
40 B. Vervliet et al. / Behaviour Research and Therapy 48 (2010) 38–43

through the work-up procedure to select a ‘‘definitely uncomfort- Online shock-expectancy ratings
able, but not painful’’ shock level. Next, participants were instruc-
ted that geometrical figures would be presented to them on the Acquisition
computer screen. They were also told that they had to learn to The left panel of Fig. 1 suggests a smooth development of
accurately predict the shock on the basis of the figures. Importantly, differential shock-expectancy to the CSþ and the CS over the
participants from Group Colour were told that the colour of the course of the acquisition phase, without differences between the
figures would be instructive about the occurrence of the shock, and groups. This was confirmed by an ANOVA with one between-
that they should learn to distinguish ‘‘good’’ from ‘‘bad’’ colours. subjects factor (Group, 2 levels), one within-subjects factor (CS, 2
Inversely, the participants from Group Shape were told that the levels) and one repeated measures factor (Trial, 4 levels). This
shape of the figures would be instructive about the occurrence of ANOVA revealed a main effect of Stimulus, F(1,30) ¼ 258.95, p < .001,
the shock, and that they should learn to distinguish ‘‘good’’ from partial h2 ¼ .90, a main effect of Trial, F(3,90) ¼ 30.1, p < .001, partial
‘‘bad’’ shapes. All participants also received the information that h2 ¼ .50, and most importantly a significant Trial * CS interaction,
a black cross would sometimes be presented, and never followed F(3,90) ¼ 116.68, p < .001, partial h2 ¼ .80. This interaction shows
by shock. This last instruction was added in order to avoid that that the differential shock-expectancy developed gradually over
participants would learn about certain perceptual dimensions trials. In addition, the Group * Trial * CS interaction was far from
that differentiate the CSþ from the CS (which could in principle significant, F(3,90) < 1, confirming that the groups did not differ in
influence the direction of generalization to the GSs as well). the acquisition phase.
Next, the operation of the expectancy pointer was explained. The
participants were asked to indicate their expectancy of shock upon Test
each stimulus presentation, and to return the dial to the starting The middle panel of Fig. 1 shows the expectancy ratings to the two
point (the middle of the scale, Uncertain) during inter trial interval. generalization test stimuli, over the course of 6 nonreinforced trials.
When the experimenter had left the room, written instructions The Figure suggests opposite effects in the two groups, in the expected
appeared on the computer screen recapitulating the verbal instruc- direction. This was confirmed by an ANOVA with one between-
tions: ‘‘Remember well: (1) the black cross will never be followed subjects factor (Group, 2 levels), one within-subjects factor (GS, 2
the electrical stimulus, (2) the COLOUR [SHAPE] of the geometrical levels) and one repeated measures factor (Trial, 6 levels). Most
figures is important to know when the electrical stimulus will importantly, the analysis reveals a significant Group * GS interaction,
follow.’’ After approximately 10 s, these instructions disappeared and, F(1,30) ¼ 22.94, p < .001, partial h2 ¼ .43, showing that the direction of
following an interval of 5 s, the actual experiment started. generalization was influenced by the verbal instructions that distin-
The stimuli were always presented for 8 s; the inter trial interval guished the groups. The main effect of GS itself was not significant,
varied between 13 s and 17 s, with a mean of 15 s. During the F(1,30) < 1, nor the effect of Group, F(1,30) < 1, showing that the
acquisition phase, the CSþ and the CS were presented 4 times each, level of expectancy was not directly defined by stimulus or group
in a randomized order with the restriction of no more than 2 membership. There was a main effect of Trial, F(5,150) ¼ 205.68,
consecutive stimulus presentations being identical. The termination p < .001, partial h2 ¼ .87, indicating that the level of shock-expectancy
of every CSþ presentation was immediately followed by the shock. decreased over the 6 (nonreinforced) trials. This points to extinction
During the subsequent generalization test phase, the GSc and the GSs learning to both GSs. The CS * Trial interaction was not significant,
were each presented 6 times without the shock US. The first GS F(5,160) < 1, but the Group * CS * Trial interaction was significant,
presentation was counterbalanced across participants and groups. F(5,160) ¼ 23.99, p < .001, partial h2 ¼ .44. This is not surprising, as the
The rest of the stimulus presentations were randomized, with the direction of generalization was clearly different in the two groups,
restriction of no more than two consecutive presentations being leading to different extinction curves. In order to confirm that the
identical. The experiment ended with a fear recovery test phase, in group differences were produced by differences in generalization,
which the original CSþ and CS were presented, three times each we conducted a separate ANOVA on the first test trial alone, with one
without shock. The first CS presentation was counterbalanced across between-subjects factor (Group, 2 levels) and one within-subjects
participants and groups, and the rest of the stimulus presentations factor (GS, 2 levels). This resulted again in a significant Group * GS
were randomized, with the restriction of no more than two interaction, F(1,30) ¼ 73.03, p < .001, partial h2 ¼ .71, and the absence
consecutive presentations being identical. of a significant main effect of GS, F(1,30) < 1, or Group, F(1,30) < 1.

Results Fear recovery


The right panel of Fig. 1 shows the expectancy elicited by the
Data analysis original CSþ and CS, over the course of 3 nonreinforced test trials.
The figure suggests a complete return of shock-expectancy. This
The expectancy ratings on the rotary dial were registered as was confirmed by an ANOVA with one between-subjects factor
analog-to-digital units (AD units), and averaged over the last (Group, 2 levels), one within-subjects factor (CS, 2 levels) and one
second of each stimulus presentation before statistical analysis. repeated measure (Trial, 3 levels). This analysis revealed a main
This generated 1 expectancy rating per trial. effect of CS, F(1,30) ¼ 226.26, p < .001, partial h2 ¼ .88, and the
Skin conductance response amplitudes were defined as the peak absence of a significant Group * CS interaction, F(1,30) < 1. The
value within the 1–4 s interval after CS onset, relative to a baseline main effect of Trial was significant, F(2,60) ¼ 45.91, p < .001, partial
averaged over the 2 s prior to CS onset. This generates the first h2 ¼ .61, as was the Trial * CS interaction, F(2,60) ¼ 58.06, p < .001,
interval response (FIR, see Prokasy & Raskin, 1973). Negative changes partial h2 ¼ .66. This shows that the level of shock-expectancy
were scored as zero and included in the analyses. Amplitudes were extinguished over the course of three nonreinforced test trials, and
range-corrected using the largest response elicited by the US (peak that this occurred differentially for the CSþ and the CS. Again, the
amplitude between 9 and 14 s after CS onset) as the maximum range Group factor did not influence this interaction, F(2,60) ¼ 1.25,
for each individual. The corrected response magnitudes were p ¼ .30. In order to test whether there was any effect of the
subjected to a square-root transformation in order to normalize the extinction phase, we conducted a separate ANOVA to compare the
distribution prior to statistical analysis. The alpha-level was set at .05 size of the discrimination on the last acquisition trial and the first
for all analyses. renewal test trial. This ANOVA with one between-subjects factor
B. Vervliet et al. / Behaviour Research and Therapy 48 (2010) 38–43 41

GROUP COLOUR
CSp
CSm
450 GS1
GS2
400

rated shock-expectancy
350

300

250

200

150

100

50

0
acq1 acq2 acq3 acq4 ext1 ext2 ext3 ext4 ext5 ext6 test1 test2 test3

GROUP SHAPE
CSp
CSm
450
GS1
400 GS2
rated shock-expectancy

350

300

250

200

150

100

50

0
acq1 acq2 acq3 acq4 ext1 ext2 ext3 ext4 ext5 ext6 test1 test2 test3

Fig. 1. Mean expectancy ratings per trial over the entire experiment. Ratings are represented as digitized voltage outputs (AD units) from the rotary expectancy dial. High levels
correspond to more certainty of imminent shock (where 450 ¼ ‘‘Certain shock’’), lower levels correspond to more certainty of no shock (where 0 ¼ ‘‘Certain no shock’’). The upper
graph represents the data from group Colour, the lower graph the data from group Shape. The left panel shows the data from the acquisition phase for both the CSþ and the
CS (acq1-acq4); the middle panel shows the data from the generalization test phase for both the GSc and the GSs (ext1-ext6); the right panel shows the data for the fear recovery
test phase for both the CSþ and the CS (test1-test3).

(Group, 2 levels), one within-subjects factor (CS, 2 levels) and one one repeated measure (Trial, three levels). Most importantly, this
repeated measures factor (Trial, 2 levels) revealed a significant analysis revealed the expected Group * CS interaction, F(1,30) ¼ 4.95,
main effect of CS, F(1,30) ¼ 500.92, p < .001, partial h2 ¼ .94, and p < .05, partial h2 ¼ .14, showing that skin conductance responding
the absence of a main effect of Trial, F(1,30) < 1, and of a Trial * CS to the two GSs differed in function of Group. There was no main
interaction, F(1,30) ¼ 1.63, p ¼ .21. Moreover, the Group factor had effect of CS, F(1,30) ¼ 1.99, p ¼ .17, but a main effect of Trial,
no influence on this interaction, F(1,30) < 1. This analysis thus F(2,60) ¼ 4.25, p < .05, partial h2 ¼ .12, showing that extinction was
confirms that there was no detectable effect of the extinction phase already initiated. Interestingly, there was also a marginally signifi-
on the differential shock-expectancy to the original CSs. cant Group * CS * Trial interaction, F(2,60) ¼ 3.15, p ¼ .05, partial
h2 ¼ .01, showing that the rate of extinction differed between the two
Skin conductance responding GSs in function of Group. This difference in rate of extinction is most
likely a consequence of the difference in the level of generalization.
Acquisition
The left panel of Fig. 2 suggests a smooth development of differ- Fear recovery
ential skin conductance responding in the two groups. This was The right panel of Fig. 2 shows the skin conductance responses
confirmed by an ANOVA with one between-subjects factor (Group, 2 to the CSþ and the CS over the course of 3 nonreinforced test
levels), one within-subjects factor (CS, 2 levels) and one repeated trials. The figure suggests clear differential responding to the two
measures factor (Trial, 4 levels). This analysis revealed a main effect of CSs, unaffected by the extinction phase. This was tested first with
CS, F(1,30) ¼ 23.27, p < .001, partial h2 ¼ .44, and a significant CS * Trial an ANOVA with one between-subjects factor (Group, 2 levels), one
interaction, F(3,90) ¼ 2.8, p < .05, partial h2 ¼ .09. This shows that the within-subjects factor (CS, 2 levels) and one repeated measures
differential response was indeed an effect of the conditioning proce- factor (Trial, 3 levels). Most importantly, this analysis revealed
dure. As expected, the Group * CS * Trial was not significant, F(3,90) < 1, a significant main effect of CS, F(1,30) ¼ 21.95, p < .001, partial
suggesting that the groups did not differ over the course of acquisition. h2 ¼ .42, without a significant interaction with Group, F(1,30) < 1.
The main effect of Trial, F(2,60) ¼ 3.52, p < .05, partial h2 ¼ .11,
Test indicates that extinction was initiated by the three nonreinforced
The middle panel of Fig. 2 shows a capricious curve, but suggests test trials. In order to test whether there was any effect of the
a difference between the groups. This was analyzed by conducting extinction phase with the generalization stimuli, we conducted
an ANOVA over the first three test trials, with one between-subjects a separate ANOVA to compare the size of the discrimination on the
factor (Group, 2 levels), one within-subjects factor (GS, 2 levels) and last acquisition trial with the first renewal test trial. This ANOVA
42 B. Vervliet et al. / Behaviour Research and Therapy 48 (2010) 38–43

GROUP COLOUR
CSp
0,6 CSm
GS1
0,5 GS2

FIR-range corr-sqrt
0,4

0,3

0,2

0,1

0
acq1 acq2 acq3 acq4 ext1 ext2 ext3 ext4 ext5 ext6 test1 test2 test3

GROUP SHAPE
CSp
0,6
CSm
GS1
GS2
0,5
FIR-range corr sqrt

0,4

0,3

0,2

0,1

0
acq1 acq2 acq3 acq4 ext1 ext2 ext3 ext4 ext5 ext6 test1 test2 test3

Fig. 2. Mean first interval skin conductance responses (F.I.R.) per trial over the entire experiment. Responses are range-corrected and square-root transformed. The upper graph
represents the data from group Colour, the lower graph the data from group Shape. The left panel shows the data from the acquisition phase for both the CSþ and the CS
(acq1-acq4); the middle panel shows the data from the generalization test phase for both the GSc and the GSs (ext1-ext6); the right panel shows the data for the fear recovery test
phase for both the CSþ and the CS (test1-test3).

with one between-subjects factor (Group, 2 levels), one within- in the development of anxiety disorders, these cognitions may
subjects factor (CS, 2 levels) and one repeated measures factor therefore influence that development.
(Trial, 2 levels) revealed a main effect of CS, F(1,30) ¼ 13.16, p < .01, In the procedure of the present experiment, all features of the CS
partial h2 ¼ .31, and surprisingly, a significant CS * Trial interaction, (a yellow triangle) were present in the two generalization test stimuli
F(1,30) ¼ 4.33, p < .05, partial h2 ¼ .13. This interaction indicates (a yellow square and a blue triangle). Further nonreinforced presen-
that the size of the discrimination actually increased from acqui- tations of the two generalization stimuli provided an opportunity to
sition to test (see Fig. 2). Hence, extinguishing the GSs had certainly test whether fear of a stimulus can be extinguished by extinguishing
not a decreasing effect on the original discrimination. fear of its features. The results were very clear in this regard: There was
no indication whatsoever of fear extinction to the CS during final
Discussion testing. One interpretation of this finding is that fear extinction tends
to bind to the unique features of the stimulus under extinction (blue,
The present study was designed to test whether verbal instruc- square), rather than to the features it has in common with the CS
tions can influence the direction of fear generalization in humans. (yellow, triangle). An alternative viewpoint is that fear extinction
The instructions focused attention to a specific stimulus feature of the promotes configural processing of the stimulus under extinction; the
conditioned stimulus, and generalization was tested with stimuli that fear extinction effects then become confined to that specific percep-
either shared that feature or not. Two groups differed in the partic- tual constellation (see Vervliet, Vansteenwegen, Hermans, & Eelen,
ular feature of the CS that was instructed to focus on. The results 2007; Lovibond, Davis, & O’Flaherty, 2000). Irrespective of the exact
clearly showed stronger generalization towards the stimulus that mechanism, the present finding shows once again that generalization-
shared the ‘instructed feature’, both in the online shock-expectancy of-extinction is difficult to achieve, and that it is clearly weaker than
ratings and in the skin conductance responses. This suggests that fear the generalization-of-acquisition (see also Vervliet et al., 2004, 2005;
generalization is influenced by cognitions (expectations, beliefs, etc.) Vervliet, Vansteenwegen, & Eelen, 2006; Vervliet, Vansteenwegen,
about the relative importance of the different features of a feared Hermans, & Eelen, 2007, and Rowe & Craske, 1998). This parallels the
stimulus. Under the assumption that fear generalization plays a role established asymmetry in contextual generalization between fear
B. Vervliet et al. / Behaviour Research and Therapy 48 (2010) 38–43 43

acquisition and fear extinction (the latter being much more context- Dunsmoor, J. E., Mitroff, S. R., & LaBar, K. S. (2009). Generalization of conditioned
fear along a dimension of increasing fear intensity. Learning & Memory, 16,
specific, see Bouton, 2002; Vansteenwegen et al., 2005). This context-
460–469.
specificity of fear extinction has often been viewed as an experimental Ehlers, A., & Clark, D. (2000). A cognitive model of posttraumatic stress disorder.
model to study relapse of anxiety following successful therapy Behaviour Research and Therapy, 38, 319–345.
completion (e.g., Bouton, 1988, 2002; Craske & Mystkowski, 2006). Feldner, M., Monson, C., & Friedman, M. (2007). A critical analysis of approaches to
targeted PTSD prevention: current status and theoretically derived future
The present results suggest that the stimulus-specificity of fear directions. Behavior Modification, 31, 80–116.
extinction may represent yet another road to relapse. Guirlanda, S., & Enquist, M. (2003). A century of generalization. Animal Behaviour,
The emotional index (skin conductance) and the cognitive index 66, 15–36.
Hajcak, G., Castille, C., Olvet, D. M., Dunnning, J. P., Roohi, J., & Hatchwell, E. (2009).
(shock-expectancy) showed remarkably similar results in this Genetic variation in brain-derived neurotrophic factor and human fear condi-
experiment. They only differed on one point. The skin conductance tioning. Genes, Brain and Behavior, 8, 80–85.
data revealed hardly any difference between the two GSs in group Hall, G. (2003). Learned changes in the sensitivity of stimulus representations:
associative and nonassociative mechanisms. The Quarterly Journal of Experi-
Colour (although the crucial Group * GS interaction was statistically mental Psychology: Comparative and Physiological Psychology, 56B(1), 43–55.
significant), whereas the effect was equally strong in both groups Hermans, D., Clarysse, J., Baeyens, F., & Spruyt, A. (2002). Affect (Version 3.0)
for the expectancy ratings. It is unclear why the effect was absent in [computer software] from University of Leuven, Belgium, http://www.psy.
kuleuven.be/leerpsy/affect3. Retrieved20.04.06.
the skin conductance, but one possibility is that the shapes were Honig, W., & Urcuioli, P. (1981). The legacy of Guttman and Kalish (1956): 25 years
generally more salient than the colours. Under this assumption, the of research on stimulus generalization. Journal of the Experimental Analysis of
colour instructions succeeded in weakening the ‘‘normal’’ gener- Behavior, 36, 405–445.
Kalish, H. (1969). Stimulus generalization. In M. Marx (Ed.), Learning: Processes (pp.
alization on the basis of the shape, but failed to redirect it on the
205–297). Oxford, England: Macmillan.
basis of the colour. This seems a plausible explanation for the skin Kamin, L. J. (1969). Predictability, surprise, attention, and conditioning. In
conductance data, but it remains unclear why the expectancy R. M. Church (Ed.), Punishment and aversive behaviour (pp. 279–296). New York:
ratings are different (more successful) in this regard. Appleton-Centuries-Crofts.
LePelley, M. E., Oakeshott, S. M., & McLaren, I. P. (2005). Blocking and unblocking in
The experimental protocol of the present study provides a basic human causal learning. Journal of Experimental Psychology: Animal Behavior
tool for further investigations into fear generalization and the factors Processes, 31, 56–70.
that influence it. The present study focused on the effect of verbal Lissek, S., Biggs, A. L., Rabin, S. J., Cornwell, B. R., Alvarez, R. P., Pine, D. S., et al.
(2008). Generalization of conditioned fear-potentiated startle in humans:
instructions, but other strategies to influence the salience/relevance experimental validation and clinical relevance. Behaviour Research and Therapy,
of stimulus features can be investigated as well. For instance, the 46, 678–687.
verbal instructions may be replaced by prior conditioning experi- Lissek, S., Powers, A. S., McClure, E. B., Phelps, E. A., Woldehawariat, G., Grillon, C., et
al. (2005). Classical fear conditioning in the anxiety disorders: a meta-analysis.
ences in which certain stimulus features acquire more relevance Behaviour Research and Therapy, 43, 1391–1424.
than others (cf. the literature on selective attention in animal Lovibond, P. F. (2003). Causal beliefs and conditioned responses: retrospective
discrimination learning, see Mackintosh, 1965; see also LePelley, revaluation induced by experience and by instruction. Journal of Experimental
Psychology: Learning, Memory, and Cognition, 29(1), 97–106.
Oakeshott, & McLaren, 2005, for a demonstration of associability Lovibond, P. F., Davis, N. R., & O’Flaherty, A. S. (2000). Protection from extinction in
processes in human causal learning). Alternatively, the salience of human fear conditioning. Behaviour Research and Therapy, 38, 967–983.
stimulus features can be changed by targeted preexposures of the CS Mackintosh, N. J. (1965). Selective attention in animal discrimination learning.
Psychological Bulletin, 64(2), 124–150.
and another stimulus in alternation. It is known that such pre-
McNally, R. J. (1981). Phobias and preparedness: instructional reversal of elec-
exposure schedules induce perceptual learning that augments the trodermal conditioning to fear-relevant stimuli. Psychological Reports, 48,
salience of the features that distinguish the two stimuli (Hall, 2003). 175–180.
These strategies can be implemented in the present procedure and Pavlov, I. (1927). Conditioned reflexes. London: Oxford University Press.
Phelps, E. A., O’Connor, K. J., Gatenby, J. C., Grillon, C., & Davis, M. (2001). Activation
tested for their effects on conditioned fear generalization in humans. of the left amygdala to a cogitative representation of fear. Nature Neuroscience,
Fear generalization is central to many anxiety problems, but it has 4(4), 437–441.
received relatively little attention in contemporary preclinical and Prokasy, W. F., & Raskin, D. C. (1973). Electrodermal activity in psychological research.
Oxford, England: Academic Press.
clinical research. The present study proposes an experimental Razran, G. (1949). Stimulus generalization of conditioned responses. Psychological
procedure to investigate this phenomenon and the factors that Bulletin, 46, 337–365.
influence it. It was shown here that verbally transmitted information Rescorla, A. R. (1976). Stimulus generalization: some predictions from a model of
Pavlovian conditioning. Journal of Experimental Psychology: Animal Behavior
can drastically influence the direction of generalization following Processes, 2(1), 88–96.
a fearful experience. Hence, cognitions may contribute to the inter- Rescorla, R., & Wagner, A. (1972). A theory of Pavlovian conditioning: variations in
individual differences regarding the impact of a traumatic event, by the effectiveness of rein forcement and nonreinforcement. In A. Black, &
W. Prokasy (Eds.), Classical conditioning II (pp. 64–99). New York: Appleton
determining the generalization of acquired fear reactions. Century Crofts.
Rowe, M. K., & Craske, M. G. (1998). Effects of varied-stimulus exposure training
Acknowledgements on fear reduction and return of fear. Behaviour Research and Therapy, 36,
719–734.
Soares, J. J., & Ohman, A. (1993). Preattentive processing, preparedness and
Preparation of this paper was supported by a VENI-grant phobias: effects of instructions on conditioned electrodermal responses to
(451-07-033) of the Netherlands Organisation for Scientific masked and non-masked fear-relevant stimuli. Behaviour Research and
Therapy, 31(1), 87–95.
Research (NWO), and by a Grant GOA/2007/03 of the University Vansteenwegen, D., Hermans, D., Vervliet, B., Francken, G., Beckers, T., Baeyens, F., et
of Leuven. We would like to thank Ellen Vervoort for assistance al. (2005). Return of fear in a human differential conditioning paradigm caused
in conducting the research. by a return to the original acquisition context. Behaviour Research and Therapy,
43(3), 323–336.
Vervliet, B., Vansteenwegen, D., Baeyens, F., Hermans, D., & Eelen, P. (2005). Return
References of fear in a human differential conditioning paradigm caused by a stimulus
change after extinction. Behaviour Research and Therapy, 43(3), 357–371.
Bouton, M. E. (1988). Context and ambiguity in the extinction of emotional Vervliet, B., Vansteenwegen, D., & Eelen, P. (2004). Generalization of extinguished
learning: implications for exposure therapy. Behaviour Research and Therapy, 26, skin conductance responding in human fear conditioning. Learning & Memory,
137–149. 11, 555–558.
Bouton, M. E. (2002). Context, ambiguity, and unlearning: sources of relapse after Vervliet, B., Vansteenwegen, D., & Eelen, P. (2006). Generalization gradients for
behavioral extinction. Biological Psychiatry, 52(10), 976–986. acquisition and extinction in human contingency learning. Experimental
Craske, M. G., & Mystkowski, J. L. (2006). Exposure therapy and extinction: clinical Psychology, 53(2), 132–142.
studies. In M. Craske, D. Hermans, & D. Vansteenwegen (Eds.), Fear and learning: Vervliet, B., Vansteenwegen, D., Hermans, D., & Eelen, P. (2007). Concurrent excitors
From basic processes to clinical application. Washington, DC: American Psycho- limit the extinction of conditioned fear in humans. Behaviour Research and
logical Association. Therapy, 45, 375–383.

Вам также может понравиться