Group Patterns, Joint Action and Social Cognition: the

Intellectica, 2007/2-3, 46-47, pp. 207-219
Group Patterns, Joint Action and Social Cognition: the
Simmelian Hypothesis
Bernard Conein
RÉSUMÉ : Schémas de groupe, action conjointe et cognition sociale : l'hypothèse
simmelienne. Il y a peu de consensus sur la signification du terme ‘groupe social’.
Une vision dominante en psychologie du développement et en philosophie met en
avant l’attention sociale et l’action conjointe sans considérer la nature des buts
sociaux (sélection des partenaires, acquisition de rang). Pour Simmel et les théoriciens
de l’action conjointe, un groupe social est le résultat des actions réciproques de deux
agents. Dans cette perspective (que j’appellerai l’hypothèse simmelienne) l’action
conjointe mutuelle est un mécanisme constitutif générant à la fois des dyades et des
groupes plus larges. L’autre, dominante en écologie comportementale et en éthologie,
se focalise sur le comportement d’affiliation (coalition) et sur la gestion des relations
sociales multi-niveaux. Selon cette vision (l’hypothèse de « liaison »), un groupe
social est fondé sur une sélection de partenaires, des buts sociaux et des évaluations de
relations. L’opposition entre les deux visions peut-elle être surmontée ou bien cellesci impliquent-elles des divergences intrinsèques portant sur des schémas de groupes et
des habiletés sociales sous-tendant la liaison sociale entre humains ?
Je montrerai que les deux approches s’appuient sur des figures distinctes lorsqu’elles
caractérisent les schémas de groupe et les savoir-faire sociaux impliqués. Je
suggérerai cependant qu’il n’y a pas nécessairement conflit entre les deux visions car
les coalitions portent sur des actions coopératives conjointes. J’avance qu’une partie
de la difficulté avec l’hypothèse simmelienne a à voir avec la proéminence donnée à
la coopération sur la coalition et sur la négligence du contrôle social.
MOTS-CLÉ : Comportement d’affiliation, Action conjointe, Cognition de groupe,
ABSTRACT: There is little consensus on what the term ‘social group’ means. One
prominent view in Developmental Psychology and Philosophy emphasizes social
attention and joint action without considering the nature of social goals (partner’s
selection, rank acquisition). A social group for Simmel and joint action theorists is the
result of reciprocal actions of two agents. In that perspective (which I will call the
Simmelian hypothesis) joint mutual action is a constitutive mechanism for generating
both dyads (pair bonds) and larger groups. The other, dominant in Behavioral Ecology
and Ethology, focuses on affiliative behavior (coalition) and the monitoring of multilevel social relationships. A social group in that view (the bonding hypothesis) is
based on partner selection, social goals and relationships assessment.
Can the opposition between the two views be relaxed or do they imply intrinsic divergences on group patterns and social skills that underlie human social bonding?
I will show that the two approaches focus on distinct features when they characterize
the group patterns and social skills involved. However I shall suggest that there are no
necessary conflicts between the two views as coalitions bear upon joint cooperative
actions. I propose that part of the difficulty with the Simmelian hypothesis has to do
Nice-Sophia Antipolis University, LAMIC.
© 2008 Association pour la Recherche Cognitive.
with the prominence given to cooperation over coalition and the neglect of social
KEY WORDS: Affiliative behavior, Joint action, Group cognition, Coalition
Social cognition has begun to be seen as a combination of several distinct
simple skills dedicated to the processing of information about social relationships. This view has important implications on the way social groupings should
be studied, on the conception of social skills, and how social objects are
constructed in social sciences. Recent efforts within cognitive ethology and
developmental psychology challenge traditional assumptions regarding group
structure. One point of departure is the idea that social life is made-up by joint
action based on perceptual and attentional mechanisms. However, the relationship between social skills and group patterns is difficult to characterize because
there are various sources that sustain that argument. I would like to examine
one aspect of the difficulty: what relation should the notion of ‘social group’
have with regard to joint action?
One aspect of living in groups concerns doing things together but other
aspects of social groups are related to membership, dominance and affiliative
behaviors. The relation between the two aspects are often attributed to distinct
traditions. I would like to underscore interesting convergences between relational sociology in the tradition of Simmel and Goffman and current works in
ethology and social cognition that show how the two aspects could be conceived in a new perspective.1
I would like to consider at least two views on the relation between group
and cognition: the first I called the Simmelian hypothesis which is dominant in
developmental psychology and philosophy of mind, the other I called the
bonding hypothesis which is prominent in ethology and behavioral ecology.
The two hypotheses can be distinguished through their use of different criteria:
conception of relations between group and cognition, group patterns and types
of social skills. According to the first hypothesis, group patterns are grounded
in dyadic units or pair bonds constituted by co-orientation, cooperation and
joint action. According to the second, group patterns are multi-layered and
based on coalition and affiliative behavior. It seems that we have two different
ways of characterizing social groups. For the Simmelian hypothesis, groups are
made up with sequences of certain actions and mental coordination. The
bonding hypothesis characterizes groups in terms of durable relational
structures existing both at the micro and the meso level and shaping the social
environment of the agent.
So the two hypotheses can be seen as defending two different responses to
the same question: ‘what is a social group?’ At first glance, the two views
seem difficult to reconcile.
The Simmelian hypothesis conceives joint action of two agents as forming a
group under certain normative conditions (joint commitment cf. Gilbert, 1996;
collective commitment cf. Tuomela, 2000) and certain mental conditions (co1
This convergence has been explicitly traced by a recent paper of Michael Tomasello , Carpenter, M.,
Call, J., Behne, T. & Moll, H. (2005) when they mention the importance of the contributions of Gilbert
and Tuomela for understanding the nature of joint action.
Group Patterns, Joint Action and Social Cognition: the Simmelian hypothesis
intention). A social group is the result of reciprocal actions of peoples doing
things together. Joint mutual action can be seen as a unique constitutive
mechanism for generating both pair bonds and larger groups. What is important here is not the size of the group but the nature of the coordination that
produces cooperation. Two persons can act as a group if they are related in a
certain mode like the we-perspective for Tuomela (2000). An other Simmelian
aspect of the hypothesis is the connexion between the social group and sociation (Vergesellschaftung) based on the idea that sociality and group life are the
same thing.
The relation between joint action and social grouping has recently become a
focus of interest in cognitive science and social theory (Gilbert, 1996, 2003;
Tuomela, 1995, 2000). Although in principle joint action theorists
acknowledge the importance of group patterns, the vast majority of works in
the field has been restricted to the analysis of dyadic interactions and to
mechanisms of co-action and co-orientation. Some ideas about joint action and
face-to-face interaction as a basic locus of sociality are similar to those in
social science, especially in the writings of Simmel and Goffman. One reason
to define that view as Simmelian can be found in Gilbert (1996, 2003) as
Simmel (1950, 1992) appears to be the precursor of the idea that the co-action
of two persons can be treated as a social group.
The bonding hypothesis conceives social groups in quite different ways.
One of the central ideas of the bounding hypothesis is that social grouping is
multi-level layered: “primate societies should be viewed as multi-layered sets
of coalitions based on relationships that differ in intensity, character and function” (Dunbar, 1988, p. 106).2 The idea that multi-level structures characterize
primate societies bears some implications for the conception of primate cognition as there is, for most ethologists, a relation between sophisticated cognition
and social complexity. Concerning social cognition, group patterns have direct
impact on cognitive architecture. Dunbar (1993), among others, considers that
the human brain evolved to make sense of and react to a complex social environment. Multi-level bonding shapes the primate social environment in a
certain way by setting new constraints on the species cognition that explain
why apes and human possess a complex social cognition. Not only the increase
of group size but also of complexity of group patterns exerted a strong pressure
for the emergence of sophisticated social skills.
For the bounding hypothesis, social group cannot be reduced to local joint
action systems. Even if for most ethologists like Byrne & Whiten (1988)
Dunbar (1993), Wrangham (1987), primate social life bears upon closed social
networks in order to sustain group cohesion (“African apes and people share a
strong tendency towards closure of their social networks” (Wrangham, 1987,
p. 58), the main feature of primate sociality is that the social group can be
extended to the community level.
Relation between cognition and group
The real divergence between the two hypotheses does not only concern
group scale but the relation between group patterns and cognitive architecture.
In most studies on social cognition, the relation between group and cognition tends to be conceived in several forms and we will here consider three of
The idea that a social group consists of patterning and quality of relationships or social ties can be seen
as a network approach to group patterns (see Dunbar, 1988).
them. One can examine group cognition (common beliefs, joint intention, weattitude, collective goals) as mostly do Bratman (1999), Gilbert (1996), Pettit
(2003) and Tuomela (1995). One can also describe group action, how people
are bounded together in order to act as a group and the way groups are made up
by actions of their members. One can also examine cognition on group relationship, the type of cognitive processes that assess or monitor social
relationships and the way social objects are detected as relational objects
(Cheney and Seyfarth, 1990; Premack, 1991).
The three types of inquiry could be accomplished independently. One can
accept that a social group is constituted through reciprocal and joint actions
without asserting that groups have a mind on their own. One can consider that
social cognition is cognition about social relationships without endorsing the
view that groups are constituted by joint actions of their members through pair
bonds. But more important, it leads us to distinguish between two distinct and
challenging questions: (i) in what sense is a form of group cognition required
to construct a social group? (ii) why cognition on group, eg. recognition of
group relationships, shall be necessary to live in group?
Most recent works on joint action are concerned with the first question: the relation between group cognition and group action by stating that
common actions presuppose a form of mental coordination, we-mode or cointention, and for some writers mental coordination can be conceived as group
mind distinct from personal mind (Pettit, 2003).3
Works in primatology and behavioral ecology examine the relation between
cognition on group and group patterns. In addressing the second question, one
can introduce a new problem: does the maintenance of complex bounding need
a cognition on group built of abilities to qualify social relationships and assess
their quality?
Cognition on group relations
If cognition on group is made up of cognitive capabilities dedicated to specific stimuli, it needs to process particular informations that are distinct from
purely physical stimuli (Spelke & Kinzler, 2007). How social information is
detected or derived from specialized input coming from facial expressions,
tone of voice or body movements? Visual and auditory systems seem to be
highly sensitive to facial expression and tone of voice (Jackendoff, 1992, p.
73). For Premack & Premack (1995) and Leslie (1995) character of movements
is crucial for the understanding of group.
If there is a cognition on group or a perception of group, it does not imply
that a group should be taken as something that is there in the world. The notion
of group is a relational notion. Regularities in behavior towards other
conspecifics must be linked to a system dedicated to keeping track of those
regularities. It means that a social environment is made up with a particular
class of objects, relational and intentional objects, that are distinct from other
physical objects.
3 The notion of mental coordination does not imply group mind if we conceive coordination in the Imode as coordination between two individuals with their own minds. It is a matter of debate as to
whether we could conceive the we-mode like two persons sharing a common goal as a case of group
Group Patterns, Joint Action and Social Cognition: the Simmelian hypothesis
Premack and Premack (1995) propose that we should conceive that system
by treating the notion of group as an output of a module having relational
objects as an input:
“The concept of group, like that of possession, concerns the
relation among objects, but differs in that not only are all
objects in a group intentionnal, but they are also equal in
power, i.e no object controls the movement of any other”
(Premack & Premack, 1995, p. 194).
Under that view the existence of social groups presupposes specific expectations toward and sensibilities to certain properties of the world. In cognition
on group, things being processed are relations, ties or types of relational/intentional objects:
Group /Power
Reciprocation or mutuality between two objects A and B can be seen as a
group because it displays equal power between intentional entities. In that
view, the idea of cognition on group must be seen as an aspect of a core architecture that processes relational/intentional beings as distinct class of objects
(Leslie, 1995).
Social skills and group patterns
The bonding hypothesis explicitly defends a conception of social groups
where the kind of social information processed comprise exocentric relations
even if there are debates among ethologists about the nature of skills and the
type of cognitive architecture that parse social relationships.4
On the contrary, for the Simmelian hypothesis, social information is not in
itself information about social relations but rather mechanisms which express
co-orientation or mutuality between two individuals through motion, gaze
detection, reciprocal alignment or cooperation.
It seems that the two hypotheses express distinct views about the relation
between group and cognition; the focus is not only on different group patterns
but on two contrasting models of social skills.
In addition, other distinctions can be mentioned concerning group patterns
and social skills. If we look at group patterns, it turns out that group size must
be considered seriously. Group patterns concerns what is the appropriate basic
unit which constitutes primate societies. If the main feature of grouping is
mutuality, the dyad or pair bonds tend to be viewed as a basic unit for group
patterning. But if the main feature of grouping is coalition, alliance or partner
4 The modular approach defended by Baron-Cohen, Leslie and Premack is not shared by all ethologists
even if most of them conceive social cognition as a specific form of cognition.
selection, the basic unit of group patterning would rather be triadic (Kummer
For social skills, we can identify other possibilities. The social skills related
to group cognition are made up of cognitive capabilities of maintaining or sustaining co-orientation and mental coordination (joint intention, we-perspective)
through time by group actions (co-movement, gaze-following, attention-getting
behavior, joint attention, collective commitment). But if social skills are seen
as capabilities for detecting and assessing social relationships, other group
actions must be considered (mutual grooming, visual monitoring, empathy,
valuation, assessment of social ties).
The German sociologist Georg Simmel can be seen as one of the first to
introduce the idea that social groups find their locus in the mutual action of two
agents. When joint action theorists argue that groups are the result of their
members’ action, they are acting as Simmelian lines even though they never
throught they were.5
The simmelian inspiration
Tuomela suggests that “social groups are identified by their members’
action” or “action of groups are ‘made up’ of or constituted by joint actions of
persons” (Tuomela, 1995, p. 52). Joint actions are what create groups. Consequently, to identify a group by a form of social co-orientation between two
individuals implies that it is defined by a joint action.
The ‘Simmelian hypothesis’ consists primarily of an argument bearing both
on group patterns and on bonding mechanisms. Gilbert considers that Simmel
can be thought as the precursor of two main claims about what constitutes a
social group: (i) dyadic joint action is the paradigm of social bonding, (ii) both
dyads and large groups bear upon a common mechanism (“analysis of our
concept of shared action discovers a structure that is constitutive of social
group as such” (Gilbert, 1996, p. 178).
Joint engagement and joint commitment
What theorists of joint action like Gilbert and Tuomela add to Simmel’s
theory is two important ideas:
• a degree of social coordination between two individuals has to be correlated to a degree of mental coordination like joint intention or common
• for a shared action to emerge, the two participants in a joint action
must communicate their willingness to each other to be engaged together
or “jointly committed”:
“Each party to a desired joint commitment must express to the
other his or her readiness to be jointly committed. When this
readiness has been expressed in conditions of common
knowledge, the joint commitment is in place” (Gilbert, 2000).
Those two other ideas fit quite well with works in developmental
psychology on gaze exchange, gaze following and joint attention behavior.
During social interaction, the partners manifest their willingness to be engaged
Margaret Gilbert (1996, 2003) is an exception as she clearly makes a link between joint action and
Group Patterns, Joint Action and Social Cognition: the Simmelian hypothesis
in a relationship with each other by visual cues such as gaze orientations, head
and trunk motions. In a recent paper, Gilbert (2000) presents an interesting
example of the sort of bonding mechanisms that are required in order to form a
joint commitment. The example can easily be described as depicting a joint
attentional scene where group actions are accomplished by careful alignment to
the attentional states of others (Tomasello, 1999; Whiten, 1997).
Example: pulling a cart up on a hill
In this example, presented by Gilbert, one person has difficulty pulling a
cart up on a hill. Another person observes her difficulty and starts to help her.
The person who receives help catches the gaze of the helper and smiles. The
helper reciprocates the smile by saying “you are welcome”.
The example can be re-described as a joint attentional scene where two persons are jointly engaged to act on an object (the cart) by tuning in to the
attentional attitude and behavior of the other persons. What Tomasello (1999)
describes as ‘joint engagement’ seems to be a mechanism very similar to ‘joint
Both joint action theorists and developmental psychologists seem to accept
a common view on bonding mechanisms: to be socially linked, two persons
must be mutually engaged in an action sharing a common goal. But beside the
similarity between the two views, we can draw two distinct objections.
One would be that joint engagement in mutual attention might be a bonding
mechanism that does not require any group formation (see Goffman, 1961).
The degree of mental coordination in joint attention is weak compared to the
notion of joint commitment based upon common goal acceptance.
The other objection is that joint commitment for Gilbert requires a moral
obligation between the two participants that is absent from joint attentional
engagement. Will this be sufficient as an argument to sustain that indeed joint
commitment is related to group formation?6
Two claims of the Simmelian hypothesis fall under critics that both concern
the relationship between joint action and group formation.
An objection from Goffman (1961)
A first objection has been formulated by Goffman (1961) in the preface to
Encounters. Goffman argued that joint action exhibits different properties than
those of a social group. Joint actions are expressed by mutual gaze and social
attention. But mutuality is not enough to build up a social group, as social
attention mostly prepares to an encounter or maintains a gathering of its participants. Goffman defends the autonomy of ‘the interactional order’ as a
situated activity system based on participation rather than on membership. We
participate to a joint action and we are members of a social group. For that
purpose, in the preface, he gives a list of a series of properties that distinguish
small groups from focused gatherings. What Goffman underlines is that face
engagement and mutual attention are bonding mechanisms not for constituting
the social group but for maintaining social contact or a ‘focused gathering’:
“A crucial attribute of focused gatherings- the participant’s
maintenance of continuous engrossment in the official focus of
Tuomela (2000), objects that joint commitment require automatically moral obligation.
activity- is not a property of social groups in general”
(Goffman, 1961, p. 11).
Thus for Goffman, mechanisms of group membership and group affiliative
behaviors are distinct from mechanisms for maintaining engagement in a social
activity, even if there are “frequent congruence between the structure of a
group and the structure of a gathering of its members”.
An objection from Cosmides & Tooby (1992)
Two evolutionary psychologists Cosmides & Tooby (1992) present another
objection based on their model of how a cognitive architecture of the social
domain should be shaped. The human social cognitive architecture is a
compound of two interrelated but distinct evolved social modules. In their
view, there are two main distinct kinds of social modules: one module
processes social information without pursuing any social goals and the other
processes social information in the pursuit of a social goal such as rank
acquisition, coalition and social exchange. As in the Machiavellian hypothesis,
defended by Byrne & Whiten (1988), the pursuing of a social goal is
instrumental as social goals are adaptative and species-specific.
Only the second kind of module can be thought of as being intrinsically
social in the sense of being dedicated to the solving of problems of a social
kind (“coalition formation is specifically social”).
This second objection leads us to consider the claims propounded by the
bonding hypothesis as it is mainly focused on coalitional groupings and group
The bonding hypothesis is mostly defended by cognitive ethologists and
behavioral ecologists. Its distinctive feature is that it proposes different conceptions of the relation between cognition and the group, of what constitutes
the unit of a social group, and how social skills should be defined.
The relation between group and cognition
The bonding hypothesis tends to view the relation between cognition and
group as a cognition on group, or cognition made up of observed regularities in
the social behavior of conspecifics. Those behavioral regularities that can be
observed in the social environment of primate societies are always seen as the
product of ecological pressures on primate cognition:
• If primate social cognition is a cognition ‘on group’ or about social
relations, this is mainly due to the fact that for most cognitive ethologists
and behavioral ecologists7, the social environment of multi-level
bonding creates cognitive demands and selective pressures on primate
• The main aspect of the selective pressures is that most primates face
the challenge of increasingly complex and multiplex links and networks
patterns that might explain why they possess both complex social cognition and high order cognitive capacities such as social monitoring and
relationships assessment.
7 For Dunbar (1993b) sociality is “the core of primate existence” because primates use sociality as a tool
for survival: sociality is the “mark” that makes primates distinct from other species.
Group Patterns, Joint Action and Social Cognition: the Simmelian hypothesis
Thus, one of the main ideas of the bonding hypothesis is that social
environment contrasts with physical environment. The “special nature of the
social environment” (Byrne & Whiten, 1988), is that it is more challenging and
demanding for natural cognition.
Group patterns and social skills
To characterize the species-specificity of primate social environments, we
must carefully look at the nature of relational patterns that produce effects on
the cognitive architecture of primates. The presumption is that living in close
knit groups produces tensions but also requires attention and monitoring. The
detectable differences in bonding patterns of primate societies are formulated
most clearly by Dunbar (1993a) in the following terms:
• Close-knit groups (dyads and triads) maintain and reinforce a
preferential alliance to exclude a third party, as they are mainly the locus
of coalitional networks;
• Most primate bonding structures are multi-layered systems that include
both small coalitional groups at the micro level and larger groups at the
meso level (such as bands and communities);
• Coalitional groups are mutually dependant with larger groups because
tighly bonded group (cliques) are the main instrument of social survival
within social systems made up with large group size. Group closure
occurs in order to avoid harassment and overcrowding;
• The time devoted to maintaining mutual social contacts at the micro
level (such as grooming for apes and conversation for humans) is related
to the size of the grouping structure at the meso level. Maintaining
bonding stability inside a local social group and increasing group size at
the meso level requires that partners devote a large amount of time to
increasing both the degree of binding inside the group and the social
monitoring of others outside the group. If boundary activities link
selected partners inside the group or within the coalition, at the same
time they decouple potential links with other partners.
So the main feature of primate group patterns is the interdependance
between close-knit groups and large groups. Dense cliques need communities
as much as community structures need to be fragmented into smaller substructures in order to maintain various levels of ties between members
(Freeman 1993). The mechanism to increase the size of social groups bears on
local coalitions between a restricted number of selected partners.
Triad and not dyad
One implication for group patterns is that primate society “doesn’t consist
exclusively of one type of group” (Dunbar, 1993a) but mainly is seen as a
balance between small close knit coalitions and large groups at the community
level. Concerning the coalitional group, the paradigm is not dyad but triad, as
the social group really emerges when a pair bond closes its border by setting
social distance from a third party. Kummer (1967) was the first to describe the
nature of coalitional group as intrinsically tripartite: a strong pair bond in a
coalition is always conceived in relation with a third party. The members’
willingness to support each other for help in an alliance is a buffer against
intrusion of a stranger. What is important is not that A wants to be friend with
B but that A observing C wants to be friends with B as a potential ally against
a third party.
Monitoring social relationships
To conceive the triad as a basic unit of grouping has a consequence for the
conception of social skill. Partner selection is the result of capabilities that
primates display to assess and monitor social relationships with others and
between others8. The capabilities of monitoring social links with others are
limited by “the number of relationships an animal can keep track of in a complex continuously changing social world” (Dunbar, 1993a).
If coalitional cohesive groups are preserved by partners’ selections, among
primates the selection is achieved by assessing the qualities of allies:
“Primates use information about potential quality of group
members as partners to establish friendly relationships. Only
primates manipulate the probability of receiving help in the
future from particularly valuable partners. For non primates a
coalition is an end in itself, for primates it is a mean to an end”
(Harcourt, 1993, p. 706).
One main implication of the bonding hypothesis is that by focussing on the
relation between grouping and social goals such as rank acquisition, it presents
different properties that have never been thought of by the Simmelian arguments on joint action. If those properties are so distinct, how can we relax the
opposition between the two hypotheses? Can those various properties be
related in a more integrative framework? If we follow the arguments of
Cosmides and Tooby (1992), the question will be: how heterogeneous social
phenomena such as group maintenance and social monitoring could be related?
I would like to present two claims that establish how the opposition
between the two mentioned hypotheses can be relaxed even if there are few
attempts to fill the gap in that direction.
One claim concerns the fact that the social domain in primate cognition is
made of various interrelated social capacities. Among primate social capacities,
a distinction is often made by cognitive ethologists between two kinds of
cognitive skills. Certain social skills are effectively mainly devoted to the
assessment and monitoring of relationships for partners’ selection. But there is
another class of social skills with different functions, as they are mainly
devoted to the establishment and maintenance of strong ties among partners.
Grooming activities are a good example of group maintenance based on
reciprocity and co-orientation. If we accept that distinction, it has a consequence as group maintenance such as grooming is a type of joint cooperative
action. Grooming is a form joint engagement mainly dedicated to the
establishment and maintenance of social contact between selected partners.
Social monitoring serves to support a function of social vigilance: detection
and assessment of ties between partners.
But even if the two abilities do indeed differ, ethologists consider that they
both contribute to the formation of coalition. For cognitive ethologists, the
relation between the two kinds of skills is a matter of discussion (see Harcourt
1992) but the distinction seems relevant as it has an interesting consequence
for relaxing the opposition between the two hypotheses.
8 For Cheney & Seyfarth (1990), the social monitoring in primate groups is based on specific capacities
of analyzing exocentric relationships with others without reference to self (see also Harcourt, 1988).
Group Patterns, Joint Action and Social Cognition: the Simmelian hypothesis
The second argument concerns the relation between coalition and cooperation. There is strong tendency for some joint action theorists to identify
cooperation and group behavior (Tuomela, 1995; Gilbert, 1996)9. A group
cannot be reduced to cooperation, even if joint engagement should be seen as
an important precondition for goal-oriented activity (Tomasello, 1999), the
difference between cooperative behavior and coalition should be maintained.
But at the same time, the two claims are combined by the ethologists as they
show that group and cooperative behavior must be strongly connected. For the
bonding hypothesis, individuals do not cooperate without partners’ selection.
The other claim is a direct consequence of the first argument: cooperation is an
important path for maintaining a coalitional group.
It seems that group actions like mutual gaze, gaze-following, joint attention, touching, grooming can be viewed as an attempt to service a social
relationship and reinforce bonding. We can relate group actions with all the
mechanisms which are used not only to open and establish but also to reinforce
durable social contacts. Group maintenance happens not only in a sequence of
cooperative actions but in a succession of dyadic encounters for stabilizing
durable linkages between partners. It is the reason why group action as much
as group maintenance should be seen in the context of coalition and partner’s
selection. Joint actions and cooperation matter for coalition and partners’
selection. For the human, as Dunbar (1993b) suggests, verbal conversations
serve for micropolitics as a social tool both to maintain contacts between the
participants and to assess a social link with a third party by gossiping.
Simmel himself (1992/1999) has made insightful remarks to sustain
that if dyadic encounters are the primary form of grouping, social groupings
tend to exist at several levels. The idea that the locus of sociation is grounded
in dyads is compatible, he suggests, with another idea that small groups can
transmit some of their properties (mutuality, reciprocity) at the level of larger
bonding. For Simmel, small groups concentrate some emerging relational
properties that are responsible for properties at the meso level. It is of course a
matter of discussion as to whether the properties should be seen as a result of
the multi-layered nature of primate group patterns as in the bonding hypothesis
or mainly the consequences of local mechanisms that appeared primarly at the
It seems plausible that the social skills dedicated to group maintenance
should be distinguished from related capabilities of monitoring social ties. If
we admit that they both constitute two main means for building
groups (maintenance of cohesion and monitoring of relations), there are no
strong incompatibilities between the Simmelian hypothesis and the bonding
hypothesis. There might constitute a bridge between the two opposed hypotheses. Joint attention and joint commitment are not the only mechanisms of
social cognition but certainly they are part of it.
As several authors in the field of social cognition (Baron-Cohen, 1995;
Cosmides & Tooby, 1992; Premack & Premack, 1995) accept that social primate cognition is made with various interrelated social skills, an important part
In fact Tuomela (2000) makes here an interesting distinction between weak cooperation (joint action
in the I-mode) and strong cooperation (joint action as a group in the we-mode).
of the discussion should be devoted to design the several components of that
architecture by refusing to reduce social cognition to one unique mechanism.
Baron-Cohen, S. (1995). Mindblindness, an Essay on Autism and Theory of Mind.
Cambridge: MIT Press.
Bratman, M. (1999). Faces of Intention. Cambridge: C.U.P.
Byrne, R., Whiten, A., (Eds) (1988). Machiavellian Intelligence: Social Expertise and
the Evolution of Intellect in Monkeys, Apes and Humans. Oxford: Clarendon Press.
Cheney, D., Seyfarth, R. (1990). How Monkeys See the World: Inside the Mind of
Another Species. Chicago & London: University of Chicago Press.
Cheney, D., Seyfarth, R., Smuts, B. (1986). Social Relationships and Social Cognition
in Non Human Primates. Science, 234.
Conein, B. (2005), Les sens sociaux: trois essais de sociologie cognitive. Paris:
Cosmides, L., Tooby, J. (1992). Cognitive Adaptations for Social Exchange. In The
Adaptated Mind. Oxford: Oxford University Press, p. 163-184.
Dunbar, R. (1988). Primate Social Systems. London and Sideny: Croom Helm.
Dunbar, R. (1989). Social Systems as Optimal Strategy Sets. In V. Standen & R.Fooley
(eds.), Comparative Sociobiology. Blackwell, p.131-150.
Dunbar, R. (1993a). Co-evolution of Neocortical Size, Group Size and Language in
Humans. Behavioral & Brain Sciences, Vol. 16, n° 4, p. 631-735.
Dunbar, R.(1993b), Grooming, Gossip and the Evolution of Language. London: Faber
& Faber.
Freeman, L. (1993). Group Structure and Group Size Among Humans and Other
Primates. Behavioral & Brain Sciences, Vol. 16, n° 4, p.703-704.
Gallese, V. (2006). Intentional Attunement: A Neurophysiologcial Perspective on
Social Cognition and its Diruption in Autism. Brain Research, 1079, pp. 15-24.
Gigerenzer, G. (1997). The Modularity of Social Intelligence. In A., Whiten & R.,
Byrne (eds), Machiavellian Intelligence II. Cambridge: Cambridge University
Press, pp. 264-272.
Gilbert, M.(1996). Living Together. Lanham: Rowman & Littlefield Publisher.
Gilbert, M. (2000). A Social Ontological Approach to Political Obligation.
Communication in Dresden, 25 june.
Gilbert, M. (2003). Marcher ensemble, un phénomène social paradigmatique. In
Marcher ensemble. Paris: PUF, pp. 45-71.
Goffman, E. (1961). Encounters: Two Studies in the Sociology of Interaction.
Indianapolis: Bobbs-Merrill.
Harcourt, A. (1988). Alliances in Contests and Social Intelligence. In R., Byrne & A.
Whiten (eds), Machiavellian Intelligence: Social Expertise and the Evolution of
Intellect in Monkeys, Apes, and Humans. Oxford: Clarendon Press, pp. 132-152.
Harcourt, A. (1992). How Monkeys See the World: Why Monkeys? Behavioral &
Brain Sciences, Vol. 15, n° 1, pp.160-161.
Jackendoff, R. (1992). Is There a Faculty of Social Cognition? Languages of the Mind:
Essays on Mental Representation. Cambridge, MA: Bradford/MIT Press, pp. 69-81.
Kummer, H. (1967). Tripartite Relations in Hamadryas Baboons. In S., Altman (ed.),
Social Communications among Primates. Chicago: University of Chicago Press,
pp. 113-121.
Pettit, P. (2003). Groups with Minds on Their Own. In F., Schmitt, Socializing
Metaphysics. Rowan & Lilltefield, pp. 167-193.
Leslie, A. (1995). A Theory of Agency. In D., Sperber, D., Premack, & A., Premack,
(eds.), Causal Cognition. Oxford: Clarendon Press, pp. 121-149.
Premack, D. (1991). Moral ‘Knowledge’ in the Infant. Manuscript, tr. fr. in J-P.
Changeux (ed.), Connaissance morale chez le nourrisson. Fondements naturels de
l'éthique. Editions Odile Jacob, 1993, pp. 139-153.
Group Patterns, Joint Action and Social Cognition: the Simmelian hypothesis
Premack, D., Premack, A.(1995). Intention as Psychological Cause. In D., Sperber, D.,
Premack, & A., Premack (eds.), Causal Cognition. Oxford: Clarendon Press, pp.
Sperber, D. (1996). La contagion des idées. Paris: Odile Jacob.
Simmel, G. (1950). The Sociology of Georg Simmel. New York: Free Press.
Simmel, G. (1992). Soziologie. Frankfurt: Surkampf. Tr. fr Sociologie. Paris: Presses
Universitaires de France, 1999.
Spelke, E., & Kinzler, K. (2007). Core Knowledge. Developmental Science, 10 (1), pp.
Tomasello, M. (1999). The Cultural Origins of Human Cognition. Cambridge: MA,
Harvard University Press.
Tomasello, M., Carpenter, M., Call, J., Behne, T. & Moll, H. (2005). Understanding
and Sharing Intentions: The Origins of Cultural Cognition. Behavioral and Brain
Sciences, 28, pp. 675-735.
Tuomela, R. (1995). The Importance of Us. Stanford: Stanford University Press.
Tuomela, R. (2000). Cooperation: A Philosophical Study. Dordrecht and Boston:
Kluwer Academic Publishers.
Wrangham, R. (1987). The Significance of African Apes for Reconstructing Human
Social Evolution. In G., Kinsey (ed.), The Evolution of Human Behavior: Primate
Models. Albany, N.Y: State University of New York Press, pp. 51-71.