Can Social Interaction Constitute Social Cognition?

Abstract

An important shift is taking place in social cognition research, away from a focus on the individual mind and toward embodied and participatory aspects of social understanding. Empirical results already imply that social cognition is not reducible to the workings of individual cognitive mechanisms. To galvanize this interactive turn, we provide an operational definition of social interaction and distinguish the different explanatory roles - contextual, enabling and constitutive - it can play in social cognition. We show that interactive processes are more than a context for social cognition: they can complement and even replace individual mechanisms. This new explanatory power of social interaction can push the field forward by expanding the possibilities of scientific explanation beyond the individual.

Full text

Can social interaction constitute social
cognition?
Hanne De Jaegher
1
, Ezequiel Di Paolo
2
and Shaun Gallagher
3,4
1
Marie Curie Project DISCOS, Department of Psychiatry, University of Heidelberg, VossStrasse 4, D-69115 Heidelberg, Germany
2
Ikerbasque: Basque Science Foundation, Department of Logic and Philosophy of Science, University of the Basque Country
(UPV/EHU), Av de Tolosa 70, 20080 Donostia –San Sebastia´ n, Spain
3
Department of Philosophy, Institute of Simulation and Training, University of Central Florida, Orlando, FL 32816-1352, USA
4
Humanities Division, University of Hertfordshire, Hatfield, AL10 9AB, UK
An important shift is taking place in social cognition
research, away from a focus on the individual mind
and toward embodied and participatory aspects of social
understanding. Empirical results already imply that
social cognition is not reducible to the workings of
individual cognitive mechanisms. To galvanize this inter-
active turn, we provide an operational definition of social
interaction and distinguish the different explanatory
roles –contextual, enabling and constitutive –it can play
in social cognition. We show that interactive processes
are more than a context for social cognition: they can
complement and even replace individual mechanisms.
This new explanatory power of social interaction can
push the field forward by expanding the possibilities of
scientific explanation beyond the individual.
The interactive turn in social cognition research
Research in social cognition increasingly focuses on how
people act together and understand each other in inter-
active situations [1–3]. The role of interaction has also been
a central theme in developmental studies [4–7], and in
recent discussions in philosophy of mind [6,8–11]. In spite
of this trend, most investigations of social cognition still
concentrate on individual mechanisms and the observa-
tional perspective (in which one subject observes others
and tries to explain or predict their behaviour). They
consider mainly third-person aspects of social-cognitive
processes [12], although many have argued for the import-
ance of second-person, participatory capabilities [2,4,5,7,
10,11,13–15]. Inasmuch as social interaction figures in
current accounts, it is typically viewed as the end goal
that individual cognitive functions should achieve (e.g.
through online or implicit mentalizing [16,17]), not as part
of the cognitive processes themselves.
In this article, we argue that investigating interaction is
central to understanding social cognition. This focus is not
meant to be normative, as if interactive explanations were
always preferable. Our proposal is rather that the role of
interactive and individual elements in social cognition
must be systematically re-evaluated. To this end, we pro-
vide a definition of social interaction and identify the
possible roles that it can play in social cognitive perform-
ance. These conceptual tools will facilitate the assessment
of the factors that shape social cognition and motivate
novel experimental designs.
Does social cognition research not already take social
interaction seriously?
We use social cognition as a general term to describe
cognition involving others, for example understanding
others’ emotions, intentions and actions and acting
Opinion
Glossary
Autonomous system: A network of co-dependent, precarious processes able to
sustain itself and define an identity as a self-determined system. The same
systemic relation can be found on many different levels. Examples include
living cells, immune networks, sensorimotor flows of neural and bodily activity,
habits, social institutions and so on.
Coordination: A non-accidental correlation in the activity of two or more
systems that are coupled at present or were coupled in the past, or are or were
coupled to another system in common, over and above what is expected from
their normal behaviour in the absence of such couplings. A typical example of
coordination between two people is synchronization of speech and bodily
movements during a conversation. A situation where two people not directly
influencing each other turn their attention to the same object at the same time
because a strange sound is coming from it, is an example of coordination
because of an external event.
Coupling: The influence between a system’s variables and another system’s
parameters. It can be mutual, for instance a person walking a dog held by a
leash.
Engagement: The qualitative aspect of a social interaction as it starts to ‘take
over’ and acquires a momentum of its own. This can happen for example in
conversations or contagious laughter. There can be coupling between agents
without engagement, for instance heat exchange between people waiting at a
crowded bus stop.
Individualist (or internalist) explanation: One that relies solely on individual
factors, for example neural mechanisms, and for which social interaction plays
no role, or at most a contextual role. Example: an interaction is judged to be
‘live’ by an infant, and not a playback of a previous interaction, by means of an
internal ‘social contingency detection’ module implemented in her brain.
Interactive explanation: One that relies on social interacti on playing an
enabling or constitutive role. Example: an infant behaves differently in a ‘live’
interaction because the coupling is more dynamically stable and disposes the
infant to keep interacting as opposed to a playback of a previous interaction,
which is dynamically less stable and easier to disengage from.
Regulated coupling: Motivated changes that an agent makes to the constraints
and parametrical conditions that influence the coupling between the agent and
another system. The other system can be an agent that could itself be
regulating the coupling, in which case we speak of a ‘co-regulated’ coupling. A
simple example: moving closer to someone speaking in a low voice to hear him
better.
Social cognition: General term used to describe different forms of cognition
about, or actions in regard to, agents or groups of agents, their intentions,
emotions, actions and so on, particularly in terms of their relation to other
agents and the self.
Social interaction: Two or mo re autonomous agents co-re gulating their
coupling with the effect that their autonomy is not destroyed and their
relational dynamics acquire an autonomy of their own. Examples: conversa-
tions, collaborative work, arguments, collective action, dancing and so on.
Corresponding author: Di Paolo, E. (ezequiel@sussex.ac.uk).
1364-6613/$ –see front matter ß2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.tics.2010.06.009 Trends in Cognitive Sciences, October 2010, Vol. 14, No. 10 441

towards and with them in social settings. Social cognition
is more than figuring out the other. It involves understand-
ing others but also understanding with others [11,15].
‘Understanding’ in this context does not require a capa-
bility for verbalising reasons for actions, but rather a
pragmatic ability to act appropriately in a particular situ-
ation. Following embodied approaches [18,19], we take
social cognition to involve the know-how that allows us
to sustain interactions, form relations, understand each
other, and act together.
The current understanding of the role of social inter-
action in social cognition is limited. Most empirical
research in psychology and neuroscience focuses on indi-
vidual mechanisms in the absence of interaction (e.g.
functional imaging research typically examines passive
differential understanding of social stimuli [20] –but see
Box 1 for exceptions). The importance of interactive pro-
cesses, however, has been highlighted in the study of
different forms of coordination in dynamical systems
approaches [3,21] and developmental studies [4,5,22] indi-
cating that complex coordination patterns result from the
mutual regulation of a social encounter.
However, it remains unclear how to incorporate such
findings. The prevailing view is that interactive patterns
figure in explanations of social cognition merely as inputs
to be processed by individual mechanisms (e.g. contingency
detection modules tuned to pick up social contingency [23]).
Thus, the study of processes of interaction remains on the
margins of the supposed central question that asks how
individual cognitive mechanisms work.
Another reason for not sufficiently examining the role of
social interaction is the lack of a definition. It is often un-
controversially assumed to signify no more than the co-
presence of more than one individual. A more adequate
definition is needed.
Defining social interaction; capturing engagement
Social interactions are complex phenomena involving
different dimensions of verbal and nonverbal behaviour,
varying contexts, numbers of participants and –frequently
–technological mediation. They impose strict timing
demands, involve reciprocal and joint activity, exhibit a
mixture of discrete and continuous events at different
timescales, and are often robust against external disrup-
tions. Essential to interaction is that it involves engage-
ment between agents.
The notion of engagement [7,14] is meant to capture the
qualitative aspect of social interaction once it starts to ‘take
over’ and acquires a momentum of its own. It also reflects
the way this experience is described in everyday language.
A definition of social interaction should capture these
intuitions concerning the engagement of at least two
agents in a complex co-regulated pattern. Engagement
can correspond to fluctuating feelings of connectedness
with the other –whose meaning sometimes seems trans-
parent and sometimes opaque –and of increasing and
decreasing possibilities for participation [11]. This experi-
ence sometimes turns into that of being taken up in the
flow of the interaction (e.g. getting caught up in an argu-
ment or a flirtation). Indeed, interaction with complex, but
not fully autonomous systems (such as virtual characters
or some social robots) can provoke an experience of engage-
ment. It is also true that engagement can occur without a
clear experience of there being another person (see Auvray
et al.’s experiment below). The reasons for these phenom-
ena deserve examination, but the phenomena would
remain obscured if we integrated the subjective element
directly into the definition. Thus, although these subjective
aspects are important, our definition needs to focus on
objective aspects, because only then can the link between
interactive patterns and the experience of interacting be
scientifically examined instead of assumed.
Our definition must capture the notion of an encounter
‘taking on a life of its own’ –for this we draw on the
systemic concept of autonomy. Accordingly, we define
social interaction as a co-regulated coupling between at
least two autonomous agents, where: (i) the co-regulation
and the coupling mutually affect each other, constituting
an autonomous self-sustaining organization in the domain
of relational dynamics and (ii) the autonomy of the agents
Box 1. Methods for studying social interaction
Social interactions show some form of autonomy, or ‘take on a life
of their own’. This makes them difficult to study under controlled
conditions, but not impossible. It is crucial that experimental
designs do not prevent engagement from developing [28]. Although
several methods cannot guarantee this, there are some ways in
which a situation of engagement can be approached.
Imaging studies tend to be restricted to noninteractive situations
because of their low time resolution [20]. However, they can
approach interaction through games in which participation is
simulated [17,42,43] or by studying self-involvement through
differential response to communicative vs. noncommunicative
stimuli [44–46]. The use of virtual characters is a promising route
for exploring contingent stimulation, although it does not yet
amount to social interaction [47,48]. Dual EEG studies are more
suitable for studying fine temporal aspects of interactive coordina-
tion and their corresponding neural support [49,50].
Naturalistic studies can adapt techniques from conversation and
gesture analysis [51–54] and measure degrees of coordination
between interactors to test hypotheses regarding cognitive and
affective aspects of engagement. For instance, Motion Energy
Analysis [55] has been applied to psychotherapy sessions to
measure bodily coordination between patient and therapist, finding
it to be a good predictor of subjective assessments of session
quality [56]. Similar links between coordination and affect have been
studied in developmental psychology [57].
Dynamical systems tools can be useful for analysing the structure
of interaction patterns [2,3,21]. Degrees of coordination can be
measured in different ways, and associated with the dimensionality
of systems involving several interactors, which can be estimated, for
instance, by Principal Component Analysis [3].Measuresof
influence between bi-variate time series [58–60] could be used to
study intra- and inter-individual coordination between neural, bodily
and environmental variables.
Synthetic modelling techniques [61] can distil the essence of
complex experimental situations into simpler models that are better
suited to dynamical analysis, making it possible to generate novel
hypotheses. Di Paolo et al.’s [30] model of Auvray et al.’s study has
yielded predictions that were later verified [32].
The study of engagement can demand the development of second
person methodologies, in which the experimenter intervenes
directly as a participant in the interaction [7,14]. Such methodolo-
gies are challenging but potentially powerful, as long as they are
kept rigorous by complementary third person measures.
This variety of methods indicates that it is already feasible to
approach the study of social interaction. What remains to be seen is
which of these methods can be used to systematically assess the
roles of individual and interactive factors.
Opinion Trends in Cognitive Sciences Vol.14 No.10
442

involved is not destroyed (although its scope can be aug-
mented or reduced) (see [11], p. 493).
The notion of autonomy here means a self-sustaining
network of processes under precarious conditions
[11,19,24,25]: a self-sustaining identity. It applies here
both to the agents and the relational dynamics of their
coupling. Autonomy can happen on different levels (meta-
bolic, neural, cognitive and social) and different timescales,
and autonomous agents can interact at various levels.
Interactions are social as long as the autonomy of the
agents is not dissolved. If one agent becomes the sole
regulator of the coupling, as in the use of a tool, this is
no longer social interaction. The definition excludes such
cases, for example situations of strong coercion. Moreover,
mere co-presence or even some forms of coupling between
agents do not automatically guarantee a co-regulated, self-
sustaining pattern of joint activity. The definition also
excludes cases in which there is no mutuality, such as
remotely observing a social scene, the mere presence of
another, or the belief that another is present. Such situ-
ations are social in an obvious sense and have measurable
cognitive effects [17,26], but do not involve interactions. We
do not restrict social interaction to the human species. As
long as the terms of the definition can be verified, they can
apply to cross-species interactions or interactions with
robots that are autonomous in the sense intended.
The cognitive processes involved in social interaction
can be sophisticated. Even though social interaction seems
to come naturally most of the time, interactive processes
are not automatic and higher cognitive processes such as
reflection, imagination and self-monitoring can influence
them.
Factors, conditions and constitutive elements
Towards a clarification of the possible explanatory roles of
social interaction, consider the collection of past and pre-
sent events, processes and relations that are observed with
a phenomenon X. We call this the set of circumstances, or
situation, in which X occurs. X could occur in different
situations, for instance different historical paths might
lead to an instance of X. Given a specific situation, some
elements in this set will play no role in the explanation of X,
whereas others will; we call the latter contextual factors.
Moreover, other elements will be necessary for the
phenomenon to occur; we call them enabling conditions.
And yet others will be part of the phenomenon itself; we
call them constitutive elements. Accordingly, given X, and
a particular situation in which X occurs:
Fisacontextual factor if variations in F produce
variations in X,
Cisanenabling condition if the absence of C prevents X
from occurring and
Pisaconstitutive element if P is part of the processes
that produce X.
A contextual factor is simply something that has an
effect on X, and can be determined by observing how X is
changed when the factor is changed. An enabling condition
not only influences the phenomenon (therefore also being
contextual), but is also necessary (either contempora-
neously or historically) for X to occur. A constitutive
element is part of the phenomenon (it must be present
in the same time frame as the phenomenon). The set of all
the constitutive elements is the phenomenon itself. The
presence of these elements is necessary, and therefore also
enabling.
What exact role (if any) an element plays in X depends
on how one chooses to describe and observe X. This imposes
a requirement of added precision on the case-by-case
description of the phenomenon.
The three roles follow a scale of increased specificity. For
instance, a change in air pressure can affect the process of
boiling water in an electric kettle, thus making it a con-
textual factor. An enabling condition, in turn, is required
for the phenomenon to exist, but it might not be constitu-
tive. Boiling water in an electric kettle requires the inven-
tion of such an appliance and a supply of electricity,
although these are not themselves part of the phenomenon
–they are enabling conditions. In normal conditions, boil-
ing is constituted by an appropriate heat exchange be-
tween a metal plate and the water provoking a phase
transition from liquid to steam.
Using these concepts we can now examine the explana-
tory role assigned to social interaction in some examples.
Social interaction as contextual factor
Consider Murray and Trevarthen’s double TV monitor
experiment [27] in which 2-month-old infants interact with
their mothers via a live television link. When presented
with a recorded replay of their mother’s previous actions,
infants disengage and become distracted and upset. Repli-
cations of these results have eliminated explanations such
as infants’ fatigue [28,29].
On one explanation, the phenomenon (the difference in
the infant’s behaviour) results from the infant’s use of a
contingency detection module [23]. This module takes as
input the relevant aspects of the interaction (e.g. infor-
mation about the timing of one’s own and the other’s
actions) and outputs the integrated information as knowl-
edge about the presence or absence of social contingency.
On this explanation, the cognitive mechanism is not consti-
tuted by social interaction. If contingency detection
modules are considered innate, past social interactions
do not play an enabling role either. As input to this module,
interaction is only a contextual factor: variations in the
interaction pattern change the output of the module.
It is important to clarify that even though this is an
individualist explanation (because all the enabling and
constitutive elements are internal; see Glossary), it is
possible to imagine noninternalist explanations in which
social interaction nevertheless plays only a contextual role.
This would be the case if some external factor other than
social interaction plays an enabling or constitutive role.
For instance, a person who has recently lost his hearing
and speech is able to communicate with others using a
notepad and pen. This person can in this way ask someone
for directions in an unknown city. The reply he obtains has
an influence on his success in finding the way, but it can be
construed as only providing information: the interaction
here is no more than contextual. However, the notepad and
pen are necessary for the successful performance of the
task and therefore play at least an enabling role. In this
Opinion Trends in Cognitive Sciences Vol.14 No.10
443

case, the explanation is noninternalist and noninteractive.
By contrast, and by definition, if interaction were to play an
enabling or constitutive role (see examples below) the
explanation could not be internalist.
Social interaction as enabler of social cognition
An alternative explanation of the double video results
based on an evolutionary robotics model reveals an
enabling role for social interaction [30]. Artificial simu-
lated agents controlled by small dynamical neural net-
works moving along a one-dimensional space interact
with each other by repeatedly crossing their positions.
When the real-time movement of one of the agents is
replaced by a recording of its previous movement, the other
agent quickly retreats from the interaction (behaving ana-
logously to the infants when shown a recording of their
mothers).
During normal interaction, both agents keep each other
in dynamic transient activity, which prevents them from
losing contact. They oscillate back and forth and repeatedly
stimulate each other’s sensors. This self-sustaining coordi-
nation is robust and resistant to significant amounts of
noise. However, when trying to interact with a recording,
this robustness is lost and a single agent is unable to
sustain the coordination pattern. Eventually it moves to
one side and disengages.
There is no internal module for detecting contingencies
in the agents’ neural controllers. Rather, it is a property of
the interaction dynamics (the difference in the stability of
the coordination pattern in 2-way and 1-way coupling) that
allows the agents to either sustain interaction or disen-
gage. Without interaction the necessary processes would
not function and the phenomenon would not occur, making
interaction an enabling condition.
The model suggests a new hypothesis for the double
video experiment: the infant’s involvement is partly sus-
tained by the stability of the live interaction and is lost
without it. If it were empirically supported (for instance, by
testing the infant’s susceptibility to external distraction
when engaged vs. disengaged from the interaction), this
would indicate that social interaction plays an enabling
role in the infant’s behaviour. Implicit support for this
hypothesis is found in a study that failed to replicate
the original results because it did not allow mothers and
infants to develop sufficient engagement before the intro-
duction of the replay [28,31]. This illustrates how our
definition of social interaction (that insists on the auto-
nomy of the mutually regulated coupling) can be used to
examine existing explanations and generate novel hypoth-
eses, making the presence of engagement (and not only
contingency) the appropriate independent variable.
Social performance constituted by social interaction
An experiment by Auvray, Lenay and Stewart [32] shows a
constitutive role for interaction. Two blindfolded partici-
pants sitting in different rooms interact by moving a sensor
along a shared virtual line using a computer mouse. Every
time the sensor encounters an object on this line, the
participant receives a tap on the finger (all objects are
the same size and produce the same stimulation). Each
participant can in this way sense three different objects
(Figure 1a): a static object, the sensor of the other partici-
pant, and a ‘shadow’ object that copies the movements of
the other’s sensor at a fixed distance. Participants are
unaware of this connection between sensor and shadow.
They are told that there is a moving object controlled by the
other participant. Their task is to click on the mouse
whenever they judge that they are in contact with the
other participant.
Even in such an impoverished sensory situation, partici-
pants find each other and concentrate their mouse clicks on
each other’s sensors (65.9% of clicks) and not on the iden-
tically moving, but non-contingent shadow objects (23%)
(Figure 1b). We could postulate an individual capability for
recognizing stimuli as social based on their contingency.
However, this explanation is ruled out. An analysis of the
Figure 1.(a) Set-up for perceptual crossing experiment [32]. Both participants are
isolated, each controlling the position of a sensor along a shared virtual 1-D line
using a computer mouse (the ends of the line meet making it a circle). The squares
on each side of the line represent the objects that can be sensed by each
participant respectively. Objects are identical in size. When the sensor touches an
object the participant gets a tactile feedback on the finger (green circle). Each
participant can sense only three objects, a static one (black square), the sensor of
the other participant (red square) and a ‘shadow’ object that copies exactly the
movement of the other’s sensor at a fixed distance (blue square). (Copyright 2010
H. De Jaegher, E. Di Paolo and S. Gallagher. Licenced under Creative Commons
Attribution 3.0 Unported [http://creativecommons.org/licences/by/3.0]). (b)
Participants are asked to signal with mouse clicks when they judge they are in
contact with an object controlled by the other person. The distribution of clicks as a
function of the distance between the two sensors (thin line) shows a significant
peak at zero (sensor-sensor contact) as does the distribution of distances during
the interaction (regardless of clicking events) between the two sensors (thick line).
A small peak (arrow) at a distance of 50 pixels indicates response to the position of
the shadow object. Reproduced, with permission, from Ref. [32].
Opinion Trends in Cognitive Sciences Vol.14 No.10
444

clicks that follow each type of stimulation indicates that
participants individually cannot tell the difference be-
tween touching the other’s sensor and the other’s shadow.
Both events are followed by clicks with comparable prob-
abilities. The larger number of clicks on each other’s sen-
sors is explained by the higher frequency of sensor-sensor
encounters (52.2% versus 15.2% for sensor-shadow). The
higher frequency of sensor-sensor encounters, in turn, is
explained by the collective dynamics. Participants scan
objects in a back and forth pattern. This strategy helps
differentiate static from moving objects but by itself it
cannot differentiate between sensors and shadows: it can-
not detect contingency in the stimulus. Only when both
participants scan each other’s sensors –crossing their
positions repeatedly –does the situation stabilize. By
contrast, a situation in which one participant scans the
other’s shadow is an unstable disengaged 1-way coupling.
The other participant will move away because she is still
searching and there is no interaction.
Thus, participants consistently find each other’s sensors
in spite of their individual inability to tell the difference
between sensors and shadows (they sometimes even report
doubts about the presence of the other). Of course, indi-
vidual factors enable this social performance (e.g. imple-
menting a scanning search strategy), but they function the
same whether scanning sensors or shadows. The variation
in the number of clicks is attributable only to the differ-
ences in the stability of the coupling and not to individual
strategies. This experiment shows that the interaction
process is not only enabling but plays a constitutive role.
The phenomenon is a manifestation of the properties of the
interaction pattern.
Reconsidering individual mechanisms
If social interaction can play more than a contextual role,
then this implies, minimally, a reassessment of individu-
alist explanations (Figure 2). At the very least, they need to
be complemented with an interactional component. Mirror
neurons, for example, could function differently in inter-
active contexts. Evidence already indicates differential
activation according to whether an interactive situation
presents conflict or not [33]. Combined with evidence for
their plasticity [34], this indicates that mirror neurons
could develop as a result of the agent’s skilful involvement
in social interaction rather than being the wellspring of
capacities for social understanding. Most findings on mir-
ror neurons do not arise out of interactive situations. Their
proposed functionality, such as for recognizing goal inten-
tions in others [35], needs to be re-evaluated in interaction
contexts. Their activity might even turn out to be epiphe-
nomenal.
In another case, an interactive perspective on the de-
velopment of self–other awareness can resolve apparent
mysteries that are hard to elucidate from an individualist
viewpoint. In infants, the gap between self-experience and
awareness of others as intentional beings can be bridged
by interactive elements. Reddy argues that at 2 months,
before evidence of shared attention, infants already
respond to being the object of another’s attention in
emotion-rich interactions [6]. Thus, social interaction fur-
nishes infants with know-how about others as bearers of
intentions, a step towards understanding others’
perspectives.
Maximally, if we take seriously the idea that inter-
action can enable and constitute social cognition, we can
conceive of interaction dynamics as, in some cases, deli-
vering the necessary cognitive performance. There is no
need to duplicate their effects by an individual mechan-
ism. For instance, stability of the interactive dynamics
could explain infants’ attention in the double video test,
with no need for contingency detection modules. More
strongly, in Auvray et al.’s [32] experiment, sensitivity
to social contingency as an individual accomplishment is
ruled out. Only the interaction process fully explains the
result.
Conclusion: social interaction as explanatory tool
Resistance to interactive explanations may be partly due to
a false impression that interaction patterns are abstract.
They seem hard to pin down because they are not easily
associated with material structures such as brain regions;
but this is doubly misleading. Interaction dynamics do
have a measurable material basis and recruit not only
interacting bodies (including brains) but also elements of
social technologies and cultural norms (e.g. tools and toys
[36]). It is also misleading to conceptualize neural function
as less abstract given that it is increasingly understood as a
matter of complex relational dynamics ([37–40]). There is
no reason to privilege skull-bound theories over what are
essentially explanations at the same level of concreteness.
We have illustrated the explanatory roles of social
interaction considered in general, without focusing on
features relevant to particular cases. Sometimes fluid
engagement enables a specific social cognitive task; some-
times, instead, a history of coordination breakdowns and
recoveries is the essential feature. Interaction could be
present, but a crucial feature such as fluency or asymmetry
might be absent, thus disabling or reducing social cognitive
performance.
Social cognition has noninteractive aspects (e.g. remote
observations of social scenes, reflection on others’ actions).
But even here, social interaction is likely to play an
enabling role in the processes involved, at least in a
developmental sense [7].
It is imperative to investigate the ways in which inter-
action dynamics influence, enable, or constitute social
cognition, using and expanding current methods (Box 1).
Candidate phenomena can be found in studies linking
Figure 2. Possible roles played by social interaction in social cognition. The dotted
line represents the point from which individual mechanisms need to be re-thought
in terms of the explanatory role of the interaction process. (Copyright 2010 H. De
Jaegher, E. Di Paolo and S. Gallagher. Licenced under Creative Commons
Attribution 3.0 Unported [http://creativecommons.org/licences/by/3.0]).
Opinion Trends in Cognitive Sciences Vol.14 No.10
445

interpersonal coordination with social cognition
[2,3,7,32,41].
Previously, if thought to have an impact at all, the role of
social interaction in social cognition remained unclear.
With its definition and the clarification of its different
roles, unanswered problems in social cognition research
(Box 2) can benefit from a new explanatory tool, putting
social cognition back where it belongs: between individuals
and not only in their heads.
Acknowledgements
Many thanks to three anonymous reviewers for their suggestions, and to
Alan Costall, Vasu Reddy and the Arbeitsgruppe Pha
¨nomenologie und
Psychiatrie, University of Heidelberg for helpful conversations on the
ideas presented here. HDJ was supported by the EU Marie Curie
Research Training Network 035975 DISCOS: Disorders and coherence of
the embodied self.
References
1 Sebanz, N. et al. (2006) Joint action: bodies and minds moving together.
Trends Cogn. Sci. 10, 70–76
2 Marsh, K.L. et al. (2006) Contrasting approaches to perceiving and
acting with others. Ecol. Psychol. 18, 1–38
3 Shockley, K. et al. (2009) Conversation and coordinative structures.
Topics in Cognitive Science 1, 305–319
4 Fogel, A. and Garvey, A. (2007) Alive communication. Infant Behavior
and Development 30, 251–257
5 Trevarthen, C. and Aitken, K.J. (2001) Infant intersubjectivity:
research, theory, and clinical applications. J. Child Psychol.
Psychiatry 42, 3–48
6 Reddy, V. (2003) On being the object of attention: implications for self-
other consciousness. Trends Cogn. Sci. 7, 397–402
7 Reddy, V. (2008) How Infants Know Minds, Harvard University Press
8 Gallagher, S. (2008) Inference or interaction: Social cognition without
precursors. Philosophical Explorations 11, 163–174
9 Gallagher, S. and Hutto, D.D. (2008) Understanding others
through primary interaction and narrative practice. In The Shared
Mind: Perspectives on Intersubjectivity (Zlatev, J. et al., eds), John
Benjamins
10 Ratcliffe, M. (2007) Rethinking Commonsense Psychology: A Critique
of Folk Psychology, Theory of Mind and Simulation, Palgrave
Macmillan
11 De Jaegher, H. and Di Paolo, E. (2007) Participatory Sense-Making: An
enactive approach to social cognition. Phenomenology and the
Cognitive Sciences 6, 485–507
12 Frith, C.D. (2008) Social cognition. Philos. Trans. R Soc. London B Biol.
Sci. 363, 2033–2039
13 Hutto, D.D. (2004) The limits of spectatorial folk psychology. Mind and
Language 19, 548–573
14 Reddy, V. and Morris, P. (2004) Participants don’t need theories:
knowing minds in engagement. Theory Psychology 14, 647–665
15 Gallagher, S. (2009) Two problems of intersubjectivity. Journal of
Consciousness Studies 16, 298–308
16 Frith, U. and Frith, C.D. (2003) Development and Neurophysiology of
Mentalizing. Philos. Trans. R Soc. London B Biol. Sci. 358, 459–473
17 Gallagher, H.L. et al. (2002) Imaging the intentional stance in a
competitive game. NeuroImage 16, 814–821
18 Clark, A. (1997) Being There: Putting Brain, Body and World Together
Again, MIT Press
19 Varela, F.J. et al. (1991) The Embodied Mind: Cognitive Science and
Human Experience, MIT Press
20 Van Overwalle, F. (2009) Social cognition and the brain: A meta-
analysis. Human Brain Mapping 30, 829–858
21 Richardson, M.J. et al. (2007) Rocking together: Dynamics of
intentional and unintentional interpersonal coordination. Human
Movement Science 26, 867–891
22 Tronick, E.Z. (2007) The Neurobehavioral and Social Emotional
Development of Infants and Children, Norton
23 Gergely, G. and Watson, J.D. (1999) Infant’s sensitivity to imperfect
contingency in social interaction. In Early Social Cognition (Rochat, P.,
ed.), Erlbaum
24 Di Paolo, E. (2009) Extended life. Topoi 28, 9–21
25 Thompson, E. (2007) Mind in Life: Biology, Phenomenology, and the
Sciences of Mind, Harvard University Press
26 Sebanz, N. et al. (2003) Representing others’ actions: just like one’s
own? Cognition 88, B11–B21
27 Murray, L. and Trevarthen, C. (1985) Emotional regulation of
interactions between 2-month-olds and their mothers. In Social
Perception in Infants (Field, T.M. and Fox, N.A., eds), pp. 177–197,
Ablex
28 Nadel, J. et al. (1999) Expectancies for social contingency in 2-month-
olds. Developmental Science 2, 164–173
29 Stormark, K.M. and Braarud, H.C. (2004) Infants’ sensitivity to social
contingency: A ‘‘double video’’ study of face-to-face communication
between 2- and 4-month-olds and their mothers. Infant Behavior
and Development 27, 195–203
30 Di Paolo, E. et al. (2008) Sensitivity to social contingency or stability of
interaction? Modelling the dynamics of perceptual crossing. New Ideas
in Psychology 26, 278–294
31 Rochat, P. et al. (1998) Are young infants sensitive to interpersonal
contingency? Infant Behavior and Development 21, 355–366
32 Auvray, M. et al. (2009) Perceptual interactions in a minimalist virtual
environment. New Ideas in Psychology 27, 32–47
33 Fujii, N. et al. (2007) Dynamic social adaptation of motion-related
neurons in primate parietal cortex. PLoS ONE 2, e397
34 Catmur, C. et al. (2007) Sensorimotor learning configures the human
mirror system. Curr. Biol. 17, 1527–1531
35 Fogassi, L. et al. (2005) Parietal lobe: from action organization to
intention understanding. Science 29, 662–667
36 Sinha, C. (2009) Objects in a storied world: Materiality, normativity,
narrativity. Journal of Consciousness Studies 16, 167–190
37 Varela, F.J. et al. (2001) The brainweb: Phase synchronization and
large-scale integration. Nature Reviews Neuroscience 2, 229–239
38 Engel, A.K. et al. (2001) Dynamic predictions: oscillations and
synchrony in top-down processing. Nature Reviews Neuroscience 2,
704–716
39 Lungarella, M. and Sporns, O. (2006) Mapping information flow in
sensorimotor networks. PLoS Comput. Biol. 2, 1301–1312
40 Sporns, O. et al. (2004) Organization, development and function of
complex brain networks. Trends Cogn. Sci. 8, 418–425
41 Richardson, M.J. et al. (2005) Effects of visual and verbal interaction on
unintentional interpersonal coordination. J. Exp. Psychol: Hum.
Percept. Perform 31, 62–79
42 McCabe, K. et al. (2001) A functional imaging study of cooperation in
two-person reciprocal exchange. Proc. Natl. Acad. Sci. U. S. A. 98,
11832–11835
43 Eisenberger, N. et al. (2003) Does rejection hurt? An fMRI study of
social exclusion. Science 302, 290–292
44 Lotze, M. et al. (2006) Differential cerebral activation during
observation of expressive gestures and motor acts. Neuropsychologia
44, 1787–1795
45 Dietrich, S. et al. (2007) Semiotic aspects of human nonverbal
vocalizations: a functional imaging study. Neuroreport 18, 1891–1894
46 Tyle
´n, K. and Allen, M. (2009) Interactive sense-making in the brain.
In Enacting Intersubjectivity, International Workshop (Carassa, A.
et al., eds), pp. 224–241
Box 2. Questions for future research
What are the characteristics of asymmetric interactions (e.g.
mother–infant, teacher–student, doctor–patient) and how does
significant asymmetry influence each participant’s possibilities for
understanding and acting?
What are the diagnostic and therapeutic implications of adopting
an interactive perspective on pathologies such as autism,
schizophrenia and Moebius syndrome?
How do more observational forms of social understanding (e.g.
understanding a film, passing false belief tests) develop? What is
the role of interaction in these capacities?
How do the autonomy of the interaction partners and of the
interaction process itself relate? How do social institutions and
cultural norms affect this relationship?
Opinion Trends in Cognitive Sciences Vol.14 No.10
446

47 Schilbach, L. et al. (2009) Minds made for sharing: initiating joint
attention recruits reward-related neurocircuitry. J. Cogn. Neurosci.,
doi:10.1162/jocn.2009.21401
48 Wilms, M. et al. (2010) It’s in your eyes. Using gaze-contingent stimuli
to create truly interactive paradigms for social cognitive and affective
neuroscience. Social Cognitive and Affective Neuroscience 5, 98–107
49 Tognoli, M. et al. (2007) The phi complex as a neuromarker of human
social coordination. PNAS 104, 8190–8195
50 Lindenberger, U. et al. (2009) Brains swinging in concert: cortical
phase synchronization while playing guitar. BMC Neuroscience 10, 22
51 Kendon, A. (1990) Conducting Interaction: Patterns of Behavior in
Focused Encounters, Cambridge University Press
52 Sacks, H. et al. (1974) A simplest systematics for the organization of
turn-taking for conversation. Language 50, 696–735
53 Bavelas, J.B. et al. (2002) Listener responses as a collaborative process:
the role of gaze. J. Commun. 52, 566–580
54 Goodwin, C. (2007) Participation, stance, and affect in the organizat ion
of activities. Discourse & Society 18, 53–73
55 Grammer, K. et al. (1999) Fuzziness of nonverbal courtship
communication unblurred by motion energy detection. J. Pers. Soc.
Psychol. 77, 487–508
56 Ramseyer, F. and Tschacher, W. (2006) Synchrony: A core concept for a
constructivist approach to psychotherapy. Constructivism in the
Human Sciences 11, 150–171
57 Jaffe, J. et al. (2001) Rhythms of Dialogue in Infancy: Coordinated
Timing in Development, Blackwell
58 Chen, Y. et al. (2004) Analyzing multiple nonlinear time series with
extended Granger causality. Phys. Lett. A 324, 26–35
59 Schreiber, T. (2000) Measuring information transfer. Phys. Rev. Lett.
85, 461–464
60 Arnhold, J. et al. (1999) A robust method for detecting
interdependences: application to intracranially recorded EEG.
Physica D 134, 419–430
61 Harvey, I. et al. (2005) Evolutionary robotics: a new tool for studying
cognition. Artif. Life 11, 79–98
Opinion Trends in Cognitive Sciences Vol.14 No.10
447