Академический Документы
Профессиональный Документы
Культура Документы
Received 1 April 2003; received in revised form 4 November 2003; accepted 6 November 2003
Abstract
In order to investigate the neurophysiological mechanisms related to the infant’s preference of maternal stimuli
over other stimuli, auditory event-related potentials (ERPs) were recorded in responses to the mother’s voice and to
a voice of an unfamiliar female in 15 infants at the age of 4 months. Stimuli were presented in intermittent and
alternating trains of four identical stimuli (mother’s voice or an unfamiliar voice). A significant amplitude difference
was observed in the responses. This was seen as a negative ‘shift’ in the responses to mother’s voice after
approximately 350 ms. The finding suggests that the infants allocate more attention to process their own mothers’
voices compared to unfamiliar voices and it may work in favor of establishing and strengthening an emotional tie
between the infant and its mother.
䊚 2003 Elsevier B.V. All rights reserved.
Keywords: Auditory event-related potential (ERP); Infant; Human stimuli; Mother’s voice; Mother–infant interaction; Subject
matched stimuli
0167-8760/04/$ - see front matter 䊚 2003 Elsevier B.V. All rights reserved.
doi:10.1016/j.ijpsycho.2003.11.003
258 M. Purhonen et al. / International Journal of Psychophysiology 52 (2004) 257–266
between the infant and the caretakers (Bowlby, number of variables modulate the amount of atten-
1991; Stern, 1985). tion a stimulus allocates. Subjectively, significant
The neurophysiology of infant–mother interac- ‘signal stimuli’ tend to elicit strong orientation
tion is poorly understood, partly because of the reactions and the orientation reaction attenuates
difficulty in studying young human subjects. Brain more slowly with the repetition compared to indif-
physiology related to cognitive processes can be ferent stimulus (Sokolov et al., 2002).
studied in humans with event-related potentials In order to study how infants process their own
(ERPs). ERPs are small electrical potentials that mother’s voice and to compare it with the proc-
can be recorded non-invasively from human scalp. essing of an unfamiliar human voice, we studied
They result from change in the ongoing electrical auditory ERPs to mother’s voice and to an unfa-
activity of neuronal networks in response to stim- miliar female’s voice in 4 month old infants. Both
ulation of a sensory pathway, or as a result of a stimuli were delivered in sequences composed of
cognitive process. In general, the auditory ERPs intermittent trains of the same stimulus type. In
of a new-born and an infant have no resemblance order to simulate real life situations, were the
to the adult ERP waveform (Kushnerenko et al., model of a certain stimulus is build up by encoun-
2001). For example the N100 component, reflect- tering this stimulus in different occasions, the
ing activation of the auditory cortex approximately sequences were repeated one after the other in an
100 ms after stimulus (Naatanen¨¨ ¨ and Picton, alternating manner. We assumed that during the
1987), is not reliably obtained in children before study session a neural model is build up for both
6 years of age (Goodin et al., 1978). During stimulus types. However, the experiences and fam-
infancy the ERP waveforms increase in complex- iliarisation towards mother’s voice in real life
ity, the ERP peak amplitudes increase and the before the study session makes it anyhow more
latencies shorten with increasing age (Kushneren- familiar compared to the other voice, heard for the
ko et al., 2001). In infants, speech stimuli evoke first time during the experiment. On the basis of
positivity at a latency of approximately 85–120 this proportional familiarity of mother’s voice
ms and negativity at approximately 200–400 ms compared to the other voice, the well known
(Shibasaki and Miyazaki, 1992) reflecting detec- infants preference to maternal stimuli over other
tion of a stimulus and processing of stimulus stimuli, and the fact that significant stimuli tend
features in the cortex. to attract attention, we hypothesized that mother’s
When a stimulus is presented to infants repeat- voice might allocate more attention in general
edly in a behavioural setting, they will in general leading to higher ERP amplitudes, and to show
respond to it progressively less as the original less amplitude decrement with stimulus repetition.
stimulus loses its novelty (Stern, 1985). Electro-
physiologically, stimulus repetition results in a 2. Materials and methods
generation of a neural representation of the repeat- 2.1. Subjects
ed stimulus in the cortex, also called a ‘memory
trace’ of the stimulus. This trace becomes stronger Fifteen (seven boys and eight girls) healthy
¨¨ ¨
with further repetitions (Naatanen, 1990). In four-month-old infants (average 114 days, range
infants, repetition of spoken syllables causes 101–134) were studied with the approval of the
amplitude decrement in ERP components at local ethical committee. Informed consent was
approximately 200 ms and at 400 ms (Dehaene- received from the parents.
Lambertz and Dehaene, 1994). This amplitude Of the 15 infants studied, three fell asleep during
decrement likely results from the refractory prop- the recording and were excluded from the statisti-
erties of activated neuronal populations to further cal analysis.
stimulation (Picton and Hillyard, 1988), and loss 2.2. Stimuli and procedure
of arousal when the novelty value of the stimulus
¨¨ ¨
decreases with repetition (Naatanen and Picton, ERPs were recorded in response to digitized
1987). However, besides the effect of repetition a word stimuli spoken either by the infant’s own
M. Purhonen et al. / International Journal of Psychophysiology 52 (2004) 257–266 259
mother or by an unfamiliar female voice. The The continuous data were transformed off-line
stimulus consisted of the Finnish word ‘hei’, mean- to epochs of y100 to 900 ms relative to the onset
ing ‘hi’ and pronounced ‘hay’. A standard un- of each stimulus. Epochs containing eye movement
familiar female voice (stimulus duration 250 ms) artefacts were rejected using both automatic ("75
was used for all infants. Several ‘hei’ stimuli mV) and manual checking of the data. The epochs
spoken by the infant’s mother were recorded prior were digitally filtered with a low pass cut-off
to the ERP study, and the recording physically best frequency at 20 Hz, and the baseline was corrected
matching the standard unfamiliar female voice between 0 and 400 ms.
stimulus, and lasting approximately 250 ms (aver- The responses were averaged in two different
age 238 ms, range 210–256 ms) was used in the ways. (See also Fig. 1). First, all responses to the
study. same type of stimulus (mother’s voice or the
Fig. 1 shows the pattern of stimulus presenta- unfamiliar voice) were averaged resulting in two
tion. Stimuli were presented in sequences that different ‘grand averaged’ responses, and second,
consisted of four trains of four identical stimuli responses were averaged according to the number
(mother’s voice or unfamiliar female’s voice). of the stimulus in the train (the 1st, 2nd, 3rd and
This sequence was followed by a similar sequence 4th) resulting in four ‘subaverage’ responses to
using the other stimulus (mother’s or unfamiliar both mother’s voice and to the unfamiliar voice.
female’s voice). Both sequences were repeated 10 The midline electrodes Fz, Cz and lateral tem-
times in an alternating manner. The inter-stimulus poral electrodes C3 and C4 were chosen for
interval (ISI) was 1 s and the inter-train interval statistical analysis. These channels were selected
(ITI) was 12 s. as visual inspection of the responses showed the
The recordings were made in a quiet laboratory. amplitude differences to be greatest at these sites.
The stimuli were presented to the right ear through The amplitudes were measured as the maximum
an inserted earphone. In order to keep the infants peak deflection from baseline and the peaks were
feeling safe, they were held by their mother during labeled according to their latency and polarity. We
the recording. used the same latency ranges as used in previous
infant studies (Kushnerenko et al., 2001). The
N250 peak was measured between 180 and 280
2.3. Data collection and analysis
ms, P350 peak between 250 and 400 ms and N450
peak between 350 and 600 ms from stimulus onset.
The EEG was recorded with disposable-elec-
trodes from parietal, central and frontal midline 2.4. Statistical analyses of data
sites (Pz, Cz and Fz) and left and right central
and temporal sites (C3, T3 and C4, T4) of the All statistical analyses were computed with the
international 10–20 system. All electrodes were SPSS for Windows 9.0 statistical program. Repeat-
referred to the right mastoid. ed measures anovas, separately for each peak
Potentials registering vertical eye movements (N250, P350, N450) were applied to both the
were recorded between electrodes placed above amplitude and latency data with stimulus type
and below the right eye, and those registering (mother’s voice vs. unfamiliar voice) and channels
horizontal eye movements between electrodes (Fz, Cz, C3 and C4) as within-subject factors for
placed at the outer canthus of each eye. EEG and the grand averaged waveforms. The corresponding
eye movement signals were amplified and filtered models for subaverages had stimulus type (moth-
by a Synamp amplifier at 0.5–100 Hz. All signals er’s voice vs. unfamiliar voice), running number
were digitized continuously at 250 Hz by a Neu- of stimulus on the train (1st, 2nd, 3rd and 4th
roscan Acquisition System PyN 1098. Trigger presentations of the stimuli), and channels (Fz,
pulses from a Neuroscan Stim Audio System PyN Cz, C3 and C4) as within-subject factors. When
1105 controlling stimuli were stored with the spherical assumptions were not fulfilled the Green-
electrophysiological data. house-Geisser correction was used. Significant dif-
260 M. Purhonen et al. / International Journal of Psychophysiology 52 (2004) 257–266
Fig. 1. The stimulus presentation and averaging protocol used and the respective event-related potentials for Grand Averaged
waveforms and for Subaverages. Two stimulus types (mother’s voice and unfamiliar voice) were used. Both stimulus types were
presented in trains of four identical stimuli. The train was repeated four times thus forming a sequence. Corresponding sequences
for mother’s voice and for unfamiliar voice were repeated one after the other in an alternating manner. The ERPs for Grand Averaged
waveforms are shown at FZ, CZ, C3 and C4 and for Subaverages at CZ only. The numbers I–IV under the ERPS for Subaverages
show the ordinal stimulus number within a stimulus train. Black lines for mother’s voice and grey lines for unfamiliar voice.
M. Purhonen et al. / International Journal of Psychophysiology 52 (2004) 257–266 261
Mother’s voice and the unfamiliar voice elicited Mean (uV) S.D. Mean (uV) S.D.
waveforms with similar configuration, in which N250 Fz y3.74 2.84 y3.97 2.55
the ERP peaks N250, P350 and N450 were clearly Cz y5.06 1.75 y4.62 2.65
identifiable. Fig. 1 shows the grand averaged and C3 y4.14 3.71 y4.98 4.05
C4 y4.56 3.71 y4.21 3.36
subaverage ERPs in response to mother’s and
unfamiliar female’s voice. P350 Fz 3.33 1.54 4.08 1.75
Cz 3.40 2.15 4.51 1.38
C3 4.91 2.20 6.59 3.43
3.1. Grand averaged waveforms C4 4.41 3.10 5.46 2.53
A summary of statistical analysis for grand N450 Fz y1.60 1.78 y1.30 2.71
Cz y2.47 2.68 y1.53 2.36
averaged waveforms is shown in Table 1 and mean C3 y4.99 2.23 y3.52 2.18
peak amplitudes are shown in Table 2. C4 y2.56 2.88 y1.38 2.92
There was a main effect for both stimulus type
and for channel on the P350 and N450 amplitudes,
the P350 amplitude being smaller and the N450 P350 and N450 amplitudes were highest for both
amplitude being higher in response to the mothers’ stimulus types at C3 and lowest at Fz. For latencies
than to the unfamiliar voice at all channels. The there was a main effect for channel for P350
post hoc paired samples T-test showed that the latencies. Mean P350 latencies are shown in Table
difference between the stimulus types was signifi- 3. For both stimulus types, the latencies were
cant for P350 amplitude at C3 (Ps0.044) and for shortest at C3. There were no other statistically
N450 at C3 (Ps0.034) and C4 (Ps0.041). The significant differences concerning the analysis of
Table 1
A summary of statistical analysis for grand averaged waveforms
Table 4
A summary of statistical analysis for subaverages
suggest that the difference was caused by alloca- order to clarify the role of attention in these
tion of more attentional resources in the processing findings in more detail.
of the mother’s voice compared to the unfamiliar Visual ERP studies also offer support for this
voice. However, more studies will be needed in view of maternal stimuli to allocate attention over
264 M. Purhonen et al. / International Journal of Psychophysiology 52 (2004) 257–266
Table 5
Mean peak amplitudes (uV) of subaverages within a train of four stimuli
Table 6
Mean P350 peak latencies (ms) of subaverages within a train of four stimuli
Table 7
Mean N450 peak latencies (ms) of subaverages within a train of four stimuli
unfamiliar stimuli in infants. Visual recognition (Carver et al., 2003). This is interpreted to reflect
memory studies have shown that in ERPs a nega- different developmental tasks during different ages.
tive deflection called negative component (Nc) In the beginning of life forming a relationship to
reflecting attention allocation (Nelson, 1994) is mother and building up a mental representation of
larger for the mother’s face compared to that for her are the main topics. Later, after establishing
unfamiliar face in children younger than 24 months these aims more recourse can be devoted to unfa-
(de Haan et al., 2003). Interestingly, this attention miliar items.
allocation effect changes with age as in children With the exception of the N250 peak in response
older than 45 months ERPs to unfamiliar face to mother’s voice, all other peaks for both stimulus
show a larger Nc compared to mother’s face types were of maximal amplitude in the contralat-
M. Purhonen et al. / International Journal of Psychophysiology 52 (2004) 257–266 265
eral side to the ear of stimulation at the left central that in infants, the mother’s voice is processed
electrode (C3). This finding is similar to previous differently from, and it may attract more attention
findings in adults that have shown that the N100 than an unfamiliar female’s voice. It is thus pos-
amplitudes are higher in the contralateral side to sible that the auditory processing may be altered
¨¨ ¨
the ear of stimulation (Naatanen and Picton, 1987; by the reciprocal sensitization between infant and
Pekkonen et al., 1999). mother.
The analysis of the subaverages did not reveal In conclusion, the present study shows that in
any statistically significant amplitude change with 4-month-old infants, human word stimuli, identical
stimulus repetition. This finding is different from in content but spoken either by the infant’s own
the well documented phenomenon of N100 ampli- mother or by an unfamiliar female give rise to a
tude decline with stimulus repetition in adults brain response that have the same configuration
(Picton and Hillyard, 1988) and in older children that differ in amplitude after approximately 350
(Karhu et al., 1997). Dehaene-Lambertz et al. ms. This difference may be due to a negative shift,
have also demonstrated a similar amplitude reduc- processing negativity, in the response elicited by
tion in infants at the age of 2 to 3 months the mother’s voice, suggesting that the infants may
(Dehaene-Lambertz and Dehaene, 1994; Dehaene- allocate more attention to process their own moth-
Lambertz and Baillet, 1998). In their studies, the ers’ voices compared to unfamiliar voices. It may
stimuli were delivered in trains of five stimuli, work in favor of developing and strengthening an
with a stimulus onset asynchrony of 600 ms and emotional tie between the infant and its mother.
an intertrain interval of 3100 ms. Respectively, in
our study, the trains had four stimuli, the intersti- Acknowledgments
mulus (ISI) interval was 1 s and the intertrain
interval (ITI) was 12 s. Thus, compared to their This work was financially supported by grants
studies we had fewer repetitions within a train, from the Signe and Ane Gyllenberg foundation
longer silent periods between the stimuli within a and the Finnish Cultural Foundation. The authors
stimulus train and longer intertain intervals. These wish to thank Professor Seppo Saarikoski for his
differences in ISI and ITI possibly explain the help in enrolling the infants in the study and
different findings in these two studies. It is known statistician Pirjo Halonen and professor Pasi Kar-
that in adults, N100 amplitude is sensitive to the jalainen for help with the analyses.
¨¨ ¨
rate of stimulus repetition (Naatanen and Picton,
1987), but generally higher rates of stimulus pres-
References
entation strengthen the N100 amplitude decline
with repetition (Fruhstorfer et al., 1970).
Bowlby, J., repr.1991 Attachment and loss. Vol. 1: Attachment,
The development of a child relies on both 2nd edn., Penguin Books, London, 425 pp.
genetic factors and experiences with the environ- Carver, L.J., Dawson, G., Panagiotides, H., Meltzoff, A.N.,
ment. As the development of the human brain is McPartland, J., Gray, J., et al., 2003. Age-related differences
relatively delayed compared with other mammals, in neural correlates of face recognition during toddler and
it is suggested to be more prone to the influences preschool years. Dev. Psychobiol. 42, 148–159.
DeCasper, A.J., Fifer, W.P., 1980. Of human bonding: new-
of postnatal experiences (Johnson, 2001). In the borns prefer their mothers’ voices. Science 208, 1174–1176.
beginning, family members and especially the de Haan, M., Johnson, M.H., Halit, H., 2003. Int. J. Psycho-
mother are the most important caretakers and the physiol. Article in press.
interplay between mother and infant is bi-direc- Dehaene-Lambertz, G., Baillet, S., 1998. A phonological rep-
tional (Stern, 1985). We have shown earlier that resentation in the infant brain. NeuroReport 9, 1885–1888.
few days after childbirth the mothers of the infants Dehaene-Lambertz, G., Dehaene, S., 1994. Speed and cerebral
correlates of syllable discrimination in infants. Nature 370,
who were studied in the present study showed an
292–295.
increased level of arousal and alertness to own ¨
Fruhstorfer, H., Soveri, P., Jarvilehto, T., 1970. Short-term
infant’s cry and to other auditory human stimuli habituation of the auditory evoked response in man. Elec-
(Purhonen et al., 2001). The present study suggests troen. Clin. Neuro. 28, 153–161.
266 M. Purhonen et al. / International Journal of Psychophysiology 52 (2004) 257–266