You are on page 1of 9

Infant Behavior & Development 27 (2004) 195203

Infants sensitivity to social contingency: a double video


study of face-to-face communication between 2- and
4-month-olds and their mothers
Kjell Morten Stormark , Hanne C. Braarud
Centre for Child and Adolescent Mental Health, University of Bergen,
P.O. Box 7800, 5020 Bergen, Norway

Received 6 May 2003; received in revised form 12 August 2003; accepted 2 September 2003

Abstract

Infants and mothers gaze during three sequences of live face-to-face interaction were compared to sequences
where their interaction was set out of phase, by presenting either the infant or the mother with a replay of their
partners behavior during earlier live interaction in a closed circuit TV setup. There was a significant interac-
tion between sequence and foci of the infants gaze, reflecting that infants looked significantly more at their
mother during the live than the replay sequences. The infants also looked significantly more at their mother
than at other foci during the live but not the replay sequences, which suggests that neither infant fatigue nor
memory traces of earlier maternal behavior can account for the findings. These findings support the idea that
young infants are capable of distinguishing between contingent and non-contingent social interaction with their
mothers.
2004 Elsevier Inc. All rights reserved.

Keywords: Social contingency; Motherinfant communication; Infant inter-subjectivity; Social development

1. Introduction

Trevarthen (1979) put forward the theory of innate inter-subjectivity in humans to account for ob-
servations that infants and mothers were mutually regulating one anothers interests and feelings in a
rhythmic pattern during early interaction (e.g., Bateson, 1975; Brazelton, Tronick, Adamson, Als, &
Wise, 1975). A procedure that examines infants sensitivity to social contingency is the double video

Corresponding author.
E-mail address: kjell.stormark@psych.uib.no (K.M. Stormark).

0163-6383/$ see front matter 2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.infbeh.2003.09.004
196 K.M. Stormark, H.C. Braarud / Infant Behavior & Development 27 (2004) 195203

procedure, developed by Murray and Trevarthen (1985), where infants and their mothers can communi-
cate by seeing one anothers facial expressions and hear one anothers vocalizations through a closed TV
circuit.
Murray and Trevarthen (1985) found that four 2-month-olds looked significantly less at their mother
when the interaction was set out of phase by presenting the infant with a televised replay of the mothers
behavior compared to when they engaged in televised live interaction. Several authors have, however,
pointed out that since the replay always followed the live sequence, changes in the infants level of
contingent interaction could have been confounded with a number of factors, such as changes in in-
fant fussiness (Rochat, Neisser, & Marian, 1998), familiarity of the adult (Bigelow & Birch, 1999),
memory of the previous maternal behavior (Hains & Muir, 1996), expectations of certain amounts
of non-contingent behavior from their mothers (Bigelow, MacLean, & MacDonald, 1996; Hains &
Muir, 1996), or differences in the degree the mothers mirror the infants affect (Legerstee & Varghese,
2001).
The failures to replicate the original findings of Murray and Trevarthen (1985) could, however, also be
attributable to aspects in the experimental procedures. Nadel, Carchob, Kervella, Marcelli, and Rserbat-
Plantey (1999) have shown that when it was ensured that the mother and the infant really were engaged
in active interaction at the start of the experiment and that the infants attention at the mother was not
diverted in the transitions between the Live and Replay sequences the initial findings of Murray and
Trevarthen (1985) were supported. The most important finding was that infants evidenced increased
gaze at the mother during a second Live sequence compared to the preceding Replay sequence in-
dicating that fatigue cannot account for the results, although this comparison only reached statistical
significance when three (of the original 10) infants who became upset during the replay sequence were
excluded.
The main purpose of this study was to examine whether the decline in infant gaze at the mother
during the replay sequence reported by Murray and Trevarthen (1985) and Nadel et al. (1999) could
be accounted for by infant memory of the maternal behavior depicted on the video replay. Hains
and Muir (1996) acknowledged the same possible explanation and controlled for this in their sec-
ond study by letting half of the infants encounter a replay of a female strangers behavior during
interaction with another infant, while the other half were presented with the regular replay proce-
dure. Hains and Muir (1996) found that the infants gaze at the stranger declined during the replay
sequence, in both groups of infants. However, none of the groups showed any increase in visual at-
tention at the adult stranger in subsequent sequences of contingent interaction, which is also later
found by Bigelow and DeCoste (2003) in relation to both strangers and mothers. This could suggest
that responding during the non-contingent interaction carried over to subsequent sequences of con-
tingent interaction, but also reflect fatigue or a decline in the infants interest in looking at the same
person.
An alternative way to control for memory is to add a second replay sequence, where it is the mothers
who are presented with a replay of their infants; while the infants see their mothers live. This arrangement
represents equally non-contingent interaction as in the traditional replay sequence but the important
difference is that the infants responses during this second replay sequence can not be accounted for by
memory traces of previous maternal responses. The experimental setup in this study thus included five
sequences, three live and two replay sequences in a Live1Replay1Live2Replay2Live3 order. If 2- to
4-month-olds are sensitive to social contingency and detects disruptions in social interaction, one would
expect a decline in infant gaze at mother during both replay sequences.
K.M. Stormark, H.C. Braarud / Infant Behavior & Development 27 (2004) 195203 197

2. Method

2.1. Subjects

Fifteen mothers and infants were recruited via maternal and child health centers and from advertise-
ments at the University campus. Data from two motherinfant dyads were excluded because of technical
problems or that the mother and infant did not establish contact during the initial live sequence. Thus,
the study included 13 mothers with their infants. Of these, 6 were girls and 7 were boys with a mean age
of 13 weeks and 1 day (range 818 weeks).

2.2. Apparatus

Fig. 1 shows the experimental setup. The mothers and the infants interacted through a closed-circuit
TV-system in two adjacent sound- and light-proof rooms, which retained full-face life size images and
the voice of the companion and allowed them to have direct eye-contact with each other. During the
experiment, the mother and the infant sat inside two separate three-sided booths, where the inside of the
walls were isolated with foam rubber plates to ensure optimal audio quality. Both the mother and the
infant sat approximately 50 cm from the one-way mirrors that were placed diagonally inside the booths.
The mirrors reflected the screen of 14 in. TV monitors that were placed on top of the booths facing down
towards the mirrors. Behind the one-way mirrors, Sony CCX-ZIIE digital color video cameras were
placed and connected to a Pinnacle Miro DC 30+ videocard and AKG D230 microphones which were
connected to a PC via an amplifier for the microphone. The mother received the audio input from the
infant through earplugs to minimize the risk that TV-speakers in the mothers booth would pick up her
voice from the microphone in the infants booth.
The audio- and video-signals from each partner were transmitted to the others TV-monitor, through a
local PC network, which consisted of three PCs equipped with the Microsoft Transaction Server software
to synchronize the transmission and recording of the signals in the different sequences of the experiment.
During all five experimental sequences, the audio- and video-signals from the infant and the mother were
recorded on PC1 and PC2, and stored on separate files. Both PCs were equipped with Pinnacle System
Miro version DC 30+ video-card. The files depicting the behavior of the mother and the infant during the
first contingent live interaction sequence (Live1) were copied from PC1 and PC2, respectively, to PC3
equipped with Pinnacle Systems Miro DC10 videocard which controlled the playback of the mothers
and the infants behavior from the Live1 sequences during Replay1 and Replay2.
The transmission of the signals was controlled manually via two control boxes where the experimenter
could switch between live signals between the partners via PC1 and PC2 and replay signals from PC3.

2.3. Procedure

In the laboratory the mothers were explained that the purpose of the study was to investigate young
infants sensitivity to social interaction using a closed-circuit TV-system. The mothers were told that the
experiment would consist of several sequences, but not informed about the live-replay distinction, as in
Murray and Trevarthen (1986). The mothers and the infants were then seated in front of the TV monitors
inside the booths where they could see and hear each other through the closed circuit TV system. The
mothers were asked to try to establish contact with their infants via the TV-monitors. This typically took
198
Video
Video and
and audio
audio output Mothers monitor
Babys monitor

K.M. Stormark, H.C. Braarud / Infant Behavior & Development 27 (2004) 195203
output to (inverted picture)
(inverted picture)
to baby mother sees and hears infant.
sees and hears mother.

Babys PC3 Mothers


camera camera

Perspex mirror at PC1 PC2


50 angle. Camera Perspex mirror at i
films through the 45 angle. Camera
mirror. films through the
mirror.

Live 1 Baby sees mother live (in realtime). Mother sees baby live (in realtime).
Camera records babys current behavior. Camera records mothers current behavior.

Replay1 Baby sees a replay of mothers previous behavior. Mother sees baby live (in realtime).
Camera records babys current behavior Camera records mothers current behavior.

Live 2 Baby sees mother live (in realtime). Mother sees baby live (in realtime).
Camera records babys current behavior Camera records mothers current behavior.

Replay 2 Baby sees mother live (in realtime). Mother sees a replay of babys previous behavior.
Camera records babys current behavior Camera records mothers current behavior

Live 3 Baby sees mother live (in realtime). Mother sees baby live (in realtime).
Camera records babys current behavior Camera records mothers current behavior.

Fig. 1. Schematic outline of the experimental setup. Both audio and video outputs from the mother and the infant were recorded through microphones and
video-cameras positioned behind the one-way mirrors and stored on separate PCs connected in a network. During the live sequence, the output from the mother
and the infant was sent to the companion, using a loop-through function. Prior to the replay sequences, the audio and the video output from the images recorded
during Live1, from the infant and the mother, respectively, was transferred from PC1 and PC2, respectively, to PC3, where the output from the mother was
replayed to the infant during Replay1 (mother saw infant live) and to the mother during Replay2 (the infant saw mother live).
K.M. Stormark, H.C. Braarud / Infant Behavior & Development 27 (2004) 195203 199

around 23 min. The experiment started when the experimenter, who saw real time images of the mothers
and the infants on separate monitors, judged the mother and the infant to have established eye-contact
with each other.
The experiment consisted of five face-to-face interaction sequences, Live1Replay1Live2Replay2
Live3, each with duration of 30 s, with seamless shifts between the sequences. The three live sequences
consisted of contingent face-to-face interaction, where 2- to 4-month-olds and their mothers saw and
heard each other in real time. In the two replay sequences the interaction between the mother and the
infant was made non-contingent upon the others behavior, by presenting one of the companions with a
replay of the others behavior from the Live1 sequence, while the other saw the companion in real time.
Thus, in Replay1, the infant was presented with a replay of the mothers behavior during Live1, while
the mother saw the infants responses in real time. In Replay2, the mother was presented with a replay
of the infants behavior during Live1 while the infant saw the mothers responses in real time.
The transition between each of the sequences was identical in that the experimenter switched between
the output channels, so that both the mother and the infant saw a blank screen for a brief (less than 1 s)
period, which marked the difference between the sequences.

2.4. Coding

The behaviors of the mother and the infant were scored, using The Observer Video Pro version 3.0, on
a frame-by-frame basis (each second consisted of 24 frames), yielding a temporal resolution of 41.67 ms.
The coders could move between frames, back and forth, and chose one among the mutually exclusive
and exhaustive items for the behavior category of gaze, consisting of the following elements: gaze at
companion, gaze at surroundings, gaze at own body and eyes closed. The lenses of the video-cameras
which recorded the infants and the mothers behavior were placed directly behind the mirror images of
the face of the partner, which meant that when the coder saw the subjects looking directly at her, it meant
that they looked at each other during the experiment.
The coders were blind to which of the sequences they scored, which was secured by renaming the
video-files in a randomized order and scoring the files depicting the behavioral responses in the mothers
and the infants separately. This left no external cues as to which of the sequences contained recordings
of live and replay sequences. Inter-rater reliability was assessed between the main coder and a secondary
coder (blind to the hypotheses of the study as well), comprising an inter-rater agreement for infants gaze
of 0.90 and for the mothers gaze of 0.96. The behavior element closed eyes was excluded from the
statistical analyses, since it does not imply any direction in the infants or the mothers gaze.

2.5. Design and statistical analyses

The experimental design for both infants and mothers gaze was a two-way within-group factorial
design, comprising Sequence (Live1, Replay1, Live2, Replay2, Live3) Foci (companion, own body,
surroundings).
To investigate whether changes in infants gaze at the mothers between the live and the replay sequences
were specific to the experimental manipulation of setting the interaction out of phase, live sequences had to
be presented both before and after the replay sequences. Thus, the number of live sequences exceeded the
number of replay sequences in this experiment, which meant that the question addressed in this study could
not be evaluated in terms of an omnibus ANOVA, but was examined in a priori comparisons (planned
200 K.M. Stormark, H.C. Braarud / Infant Behavior & Development 27 (2004) 195203

80
Mother
70
Surroundings
Duration of gaze in %

60
Own body
50

40

30

20

10

0
Live1 Replay1 Live2 Replay2 Live3

Fig. 2. Mean proportional duration of infants gaze at the mother, surroundings and own body during the Live1, Replay1, Live2,
Replay2 and Live3 sequences.

comparisons of contrast variables; Hays, 1988), where the live and replay sequences was weighted
equally.

3. Results

There was a significant Foci Sequence interaction (F(1, 12) = 6.89, P < 0.05), which was explained
by infants looking significantly more to their mothers during the Live than during the Replay sequences
(F(1, 12) = 6.31, P < 0.05). There was no significant effect for the other foci of the infants gaze.
The drop in the infants gaze during Replay1 was not statistically different from either the previous
Live1 or the subsequent Live2, but the infants level of gaze at the mother during Replay2 tended to be
significantly lower than the previous Live2 (F(1, 12) = 3.77, P = 0.07), and was significantly lower
than the subsequent Live3 (F(1, 12) = 5.04, P < 0.05). In addition, the infants looked significantly
more at their mothers than at other foci during all the three live sequences, Live1 (F(1, 12) = 10.46,
P < 0.01), Live2 (F(1, 12) = 11.18, P < 0.01) and Live3 (F(1, 12) = 17.19, P < 0.01), but not during
either Replay1 or Replay2 (see Fig. 2).
As expected, the mothers looked significantly more at their infants than at other foci (F(1, 12) =
9272.77, P < 0.0001). There was a tendency towards a Sequence Foci interaction, although not
significantly so (F(1, 12) = 3.27, P = 0.09). Comparing individual sequences, there was a difference
between the Replay2 and Live3 sequences, in that the mothers looked significantly more at the infants
(F(1, 12) = 6.73, P < 0.05) during Live3 than they did during the preceding Replay2 sequence (see
Fig. 3).

4. Discussion

To sum up the results; the infants looked more at their mothers during the sequences of live interaction
than during the replay sequences where the communication between the infant and the mother was set
K.M. Stormark, H.C. Braarud / Infant Behavior & Development 27 (2004) 195203 201

10 0
Duration of gaze in%

98

96

94

92

90
Live1 Replay1 Live2 Replay2 Live3
Sequence

Fig. 3. Mean proportional duration of mothers gaze at infant during the five sequences.

out of phase. This effect was substantiated by an overall significant decline in gaze at mother during the
replay versus the live sequences, and by the significant preference the infants exhibited towards looking
at their mothers compared to other foci during all three live sequences, but not during the two replay
sequences. However, comparing individual live and replay sequences yielded significant differences only
between the Replay2 and the Live3 sequences. The mothers had a marked focus at their infants. Still there
was a decline in the mothers gaze at the infants during the Replay2 sequence, where the mothers were
presented with a video replay of their infants behavior.
The overall difference between the infants proportional gaze at the mothers during the live compared to
the replay sequences accords with Murray and Trevarthens (1985) and Nadel et al.s (1999) findings that
young infants are sensitive to social contingency. When comparing individual live and replay sequences,
it was only the difference between Replay2 and Live3 which yielded a statistically significant effect. It
could thus be argued that the infants did not discriminate between the other sequences of contingent and
non-contingent interaction. However, such an interpretation does not accommodate the fact that the infants
looked proportionally longer at their mother than at other foci only during the three live sequences and
not during any of the replay sequences. This suggests that the infants at some level distinguished between
sequences of contingent and non-contingent interaction, but that this difference was most pronounced
between the last replay and live sequences.
As expected, the mothers maintained visual attention at their infants, but nevertheless there was a
significant difference from Replay2 to Live3 in their proportional amount of gaze at the infants. This
probably suggests that the mothers experienced that the interaction during Replay2 was out of phase,
which accords with Murray and Trevarthen (1986) who found that mothers utterances were more centered
on the infant during two sequences of live interaction than during two sequences when the mother saw a
replay of her infant. Whether this reflects that the mothers understood that they witnessed a replay or it
was due to other factors is not clear as we did not question the mothers specifically about their experiences
during the experiment.
The most interesting aspect of maternal responses is, however, the fact that the mothers did not evidence
a similar decline in gaze during the Replay1 sequence. It could be argued that it presumably were easier
for the mothers to detect that the interaction during the Replay2 sequence was set out of phase because
they were presented with a replay of their infants during good interaction than during Replay1 where
202 K.M. Stormark, H.C. Braarud / Infant Behavior & Development 27 (2004) 195203

they witnessing their infants being less responsive. However, the fact that maternal gaze at the infant
remained high during all sequences could also reflect that maternal gaze in this study did not simply
reflect sensitivity to social contingency, but also reciprocal functions of the mother and the infant during
interaction (see Tronick, Als, & Adamson, 1979).
The fact that the infants evidenced the same preferential looking at the mother during all three sequences
of live interaction, and significantly so between the last replay and live sequences argues against Rochat
et al.s (1998) claim that infants discrimination between live and replay sequences is due to fatigue or
natural decline in social interaction as a function of lapsed time. In fact, infants amount of gaze at mother
was greater during the last compared to the first sequence of live interaction, which supports the findings
of Nadel et al. (1999).
The most important finding in this study was that the infants evidenced equally pronounced decline in
their proportional gaze at the mother during Replay2 as during Replay1. While the decline in the infants
gaze at the mother during Replay1 in principle could be attributed to memory traces of prior maternal
behavior, it cannot account for the decline during Replay2 where the infants saw the mother in real-time.
This suggests that it is the lack of contingency between the behavior of the mother and the infant in itself
which is the essential factor in explaining the decline in the infants gaze at the mother during the replay
sequences.
The fact that 30 s intervals are sufficient for 2- to 4-month-olds to recognize this major experiential,
yet subtle, sensory difference between live and replay sequences, further attests to the infants sensitivity
to social contingency, as also shown by Nadel et al. (1999). The importance of contingency is well
documented both in the area of social (e.g., Watson, 1985), emotional (Stern, 1985) and cognitive (e.g.,
Rovee-Collier, 1997) development. Whether the same principles apply to explain infants discrimination
between contingent and non-contingent interaction is not clear (see Trevarthen & Aitken, 2001). The
importance of infant sensitivity to social contingencies is, however, that it implies that 2- to 4-month-olds
perceive the effects of their own actions during interaction and thus are capable of understanding their
partners internal states and adapt to the partners subjective control (Trevarthen, 1979).

References

Bateson, M. C. (1975). Motherinfant exchanges: The epigenesis of conversational interaction. In D. Aronson & R. W. Rieber
(Eds.), Developmental psycholinguistics and communication disorders: Annals of the New York Academy of Sciences (Vol.
263, pp. 6377). London: Cambridge University Press.
Bigelow, A. E., & Birch, S. A. J. (1999). The effects of contingency in previous interactions on infants preference for social
partners. Infant Behavior and Development, 22, 367382.
Bigelow, A. E., & DeCoste, C. (2003). Sensitivity to social contingency from mothers and strangers in 2-, 4-, and 6-month-old
infants. Infancy, 4, 111140.
Bigelow, A. E., MacLean, B. K., & MacDonald, D. (1996). Infants responses to live and replay interactions with self and mother.
Merrill-Palmer Quarterly, 42, 596611.
Brazelton, T. B., Tronick, E., Adamson, L., Als, H., & Wise, S. (1975). Early motherinfant reciprocity. In M. Hofer (Ed.),
Parentinfant interaction (pp. 137154). Amsterdam: Elsevier.
Hains, S. M. J., & Muir, D. W. (1996). Effects of stimulus contingency in infantadult interactions. Infant Behavior and
Development, 19, 4961.
Hays, W. L. (1988). Statistics (4th ed.). New York: Holt, Rinehart & Winston.
Legerstee, M., & Varghese, J. (2001). The role of maternal affect mirroring. Child Development, 72, 13011313.
Murray, L., & Trevarthen, C. (1985). Emotional regulation of interaction between two-months-old and their mothers. In M. Field
& N. A. Fox (Eds.), Social perception in infancy (pp. 177198). New Jersey: Ablex Publishing Corporation.
K.M. Stormark, H.C. Braarud / Infant Behavior & Development 27 (2004) 195203 203

Murray, L., & Trevarthen, C. (1986). The infants role in motherinfant communications. Journal of Child Language, 13, 1529.
Nadel, J., Carchob, I., Kervella, C., Marcelli, D., & Rserbat-Plantey, D. (1999). Expectancies for social contingency in
2-month-olds. Developmental Science, 2, 164173.
Rochat, P., Neisser, U., & Marian, V. (1998). Are young infants sensitive to interpersonal contingency? Infant Behavior and
Development, 21, 355366.
Rovee-Collier, C. (1997). Dissociations in infant memory: Rethinking the development of implicit and explicit memory. Psy-
chological Review, 104, 467498.
Stern, D. (1985). The interpersonal world of the infant: A view from psychoanalyses and developmental psychology. New York:
Basic Books
Trevarthen, C. (1979). Communication and cooperation in early infancy. A description of primary intersubjectivity. In M. Bullowa
(Ed.), Before speech: The beginning of human communication (pp. 321347). London: Cambridge University Press.
Trevarthen, C., & Aitken, K. J. (2001). Infant intersubjectivity: Research, theory and clinical applications. Journal of Child
Psychology and Psychiatry and Allied Disciplines, 42, 348.
Tronick, E., Als, H., & Adamson, L. (1979). Structure of early face-to-face communicative interaction. In M. Bullowa (Ed.),
Before speech: The beginning of human communication (pp. 349372). London: Cambridge University Press.
Watson, J. S. (1985). Contingency perception in early social development. In T. M. Field & N. A. Fox (Eds.), Social perception
in infants (pp. 156176). Norwood, NJ: Albex.

You might also like