Home / Fellows / Gabriela Airenti


Gabriela Airenti

Université de Turin, Italie
Cognitive development of Intersubjectivity
25 January 2010 -
21 February 2010

Son projet de recherche s'inscrit dans le cadre du programme collectif : "Anthropologie d'hier à aujourd'hui" coordonné par Claude Imbert.

In recent years two trends of research, previously separated have converged on a common topic. On one side there is the work of developmental psychologists interested in explaining how young humans develop an attitude toward intersubjectivity. On the other side there is the work of the researchers who try to understand what makes the difference between humans and nonhuman primates.

This turn of research is due to the work of Tomasello and his collaborators who, after years of investigation with young children and chimpanzees have come to the conclusion that what makes the difference between human and nonhuman primates is precisely the attitude towards shared action which, according to them is typically human (Tomasello and Racoczy, 2003; Tomasello et al., 2005). It must be noted that this result due to years of experimental studies corroborates what in a theoretical way had already been stated by Premack and Premack (1994) who considered sharedness the feature that makes the distinction between humans and other primates as well as the basis for cultural transmission.

To clarify what lies behind the concept of sharedness it is useful to reconstruct how Tomasello and his collaborators have “discovered” shared intentionality. In their quest of what makes the difference between human and nonhuman primates their first candidate has been intentionality. The main hypothesis at that time was that the crucial point is the link between imitation and intentional action. Children since nine months have a comprehension of others as intentional actors. This is at the basis of human cultural transmission from one generation to the other by the so-called ratchet effect. The fact that children since a precocious age are able to understand others as intentional agents allows them to perform imitative learning both regarding object-directed actions and the use of communicative symbols. To see the other as an intentional agent is the first step in the development of the concept of person whose successive developments will be to see others as mental agents and then as reflexive agents (Tomasello, Kruger and Ratner, 1993; Tomasello, 1999). Primates do not acquire the concept of person as an intentional agent and do not perform imitation. This does not allow the transmission of knowledge that is constitutive of societies.

The results of research in the last years have led Tomasello and his collaborators to revise their position and to support the conclusion that the comprehension of intentionality is not sufficient to characterize cultural cognition. In fact different studies have shown that nonhuman primates have a rather interesting comprehension of intentional action (Tomasello, Call and Hare, 2003). In particular, it has been shown that in experimental situations chimpanzees understand attempted and accidental actions, i.e. actions that do not attain the expected result (Call et al., 2004). Moreover, chimpanzees understand that perception has an influence on action. It has been shown that in a situation of competition for food a subordinate individual tried to get food that was visible only to it and ignored food that was visible both to it and to a dominant individual (Hare et al., 2000; Hare, Call and Tomasello, 2001). If primates have these abilities and still they do not produce a social engagement, this means that the difference has to be sought elsewhere. Nonhuman primates may have an intense social activity but are not motivated to share with their conspecific as it is the case for human beings since a very precocious phase in their life. Human beings have not only intentionality but also shared intentionality (Tomasello and Racoczy, 2003; Tomasello et al., 2005).

In this way the work of Tomasello rejoins the classical philosophical position opposing individual intention and collective intention (see for instance, Gilbert, 1989; Searle, 1990; Bratman, 1992; Tuomela, 1995). According to this point of view humans have two different mental states, intentions and - taking Searle’s terminology - we-intentions. These are causal in performing individual actions and collective actions, respectively. Tomasello’s work finds in ontogenetic development the appearance of we-intentions. In this view what is shared is the intentionality of action. This is what makes the distinction between humans and other primates.

An objection to this point of view is that it still underestimates the possibilities of primates regarding collective action. In fact, as we have seen, experimental research has already extended significantly the possibilities we can attribute to primates in terms of individual action. As regards collective action, the common view has been that primates are able to act in situations of competition, but not to collaborate. But even this truth is now challenged. In nonhuman primates observed in the wild it has been shown that competition and cooperation are interrelated (Muller and Mitani, 2005). Chimpanzees have been observed coordinating and synchronizing their behaviors in hunting situations and also shifting roles and anticipate each other behaviors (Boesch, 1994). Recently the skill to collaborate has been at least partly proven also in an experimental setting with semi-free-ranging chimpanzees. The chimpanzees understood when to attain a goal a collaborator was necessary and chose the partner who had previously proven more effective (Melis, Hare and Tomasello, 2006).

The more sophisticated are the experiments the subtler becomes the distinction between human and nonhuman primates as regards the comprehension of action, and of collective action particularly. Recent experimental research and fieldwork seem to show that collaboration and then collective action is possible in chimpanzees not only in competition situations but also to achieve a goal together. So if we follow Tomasello, sharing intentions are at the origins of cultural cognition. However, sharing intentions cannot coincide with the intentions underlying simple cooperation on pain of loosing the distinction between humans and other primates. Therefore, I will argue that there is another form of sharedness that does not coincide with collective action and is characteristic of human interactions.


Cooperation and reciprocity

Let us come back to sharedness studied both from an experimental point of view and through the analysis of observations made in the fieldwork. Werneken, Chen and Tomasello (2006) have shown the difference between young children and chimpanzees that participated to an experiment where the tasks required collaboration with an adult competent partner. Both children 18-24 months of age and chimpanzees achieved the coordination requested to collaborate. The important difference was that the children were motivated not only by the goal but also by cooperation itself. This was shown by the fact that when the children attained the goal, which was to retrieve a toy, they put again the toy in the previous position to start the game again.

Matsuzawa (2007) has studied the development of chimpanzees raised by their biological mother with the Participation Observation method. With respect to the problem we are discussing he maintains that the children have the intrinsic motivation to copy the mother’s behavior. But a simple gesture that is so common in humans, the child who responds to her mother’s behavior of feeding her trying in turn to put food in the mother’s mouth, has never been registered in infant chimpanzees. He concludes that reciprocity may be the fundamental difference between humans and chimpanzees.

We find in the mentioned works two key concepts that can be used to define sharedness, i.e. cooperation and reciprocity. To analyze them in more detail may be useful to better understand what characterizes human interactions.

In fact cooperation has been designed within philosophy of language as the concept describing the bases of human interactions. Humans are rational beings and to pursue their goals they have to cooperate. So cooperation is a principle of general rationality (Kasher, 1976, 1982). It can be performed by collective action or by that particular form of collective action that is communication. To explain the cooperation principle that in his view is at the basis of human communication, Grice (1975, 1978) explicitly mentions a situation where two persons are acting together. To communicate, in his example, is the same case of two persons mending a car. There is a common goal to which the participants cooperate, both doing the part they have been assigned in the most adequate way. This is true even if their ultimate goals can be completely divergent. For instance, one of them can look forward the car being mended only to flee with it.

On a similar position is Searle. He defines social behavior, of which communication is an instance, as dependent on a mental state that he calls collective intentionality. When two persons do something together, like to push a car or speak, collective intentions take the place of individual intentions that normally cause action. In this case each actor performs her action as part of a shared action (Searle, 1990). Collective intentionality causes cooperation. According to this point of view people cooperate even in situations of conflict or competition. Two persons who insult each other at a party realize a form of higher-level cooperation, in the same way that two players in competition cooperate in participating to a match. Therefore, for Searle the use of language is part of human action, which has two dimensions, an individual and a collective one. For these two dimensions the mind provides for two different primitive mental states, intentions and collective intentions.

Thus the definition of cooperation within the philosophy of language is an engagement to make one’s part in an activity shared with others and communication is one of these possible activities. The stress put on the actors’ engagement clearly indicates that we are within the field of rational action.

Let us focus now on the concept of engagement. What is engagement in this acceptation? Engagement is to have the explicit plan to do one’s part of a cooperative action. In the famous example of a music ensemble, the violin plays her score because she knows that someone else will play the alto and the cello scores of the same trio. The quality of the result or the way it may be acquired are not relevant. The result can be the product of long sessions of coordination. Moreover, in principle it is possible to imagine that there is no difference in the performance of the violin playing alone or playing with her partners. What distinguishes a collective action is that every participant intends to do her part to attain a common goal and holds the belief that the others too do their part and share the same intentions and beliefs (Clark, 1996). Collective action is then the product of an agreement that has been stipulated before the action or is included in a conventional situation known to the participants. For instance, everybody knows the sequence of actions to be performed in a supermarket to check out.

This is not the case for chimpanzees. As both experiments and fieldwork show they make collective actions because coordination with others simply proves useful to attain specific goals. Their behavior shows that they have expectations on others’ behavior. There is no reason to interpret their action as guided by reciprocal engagement.

What about children? Very young children should be in the same situation of chimpanzees, since they have not stipulated explicit agreements leading to shared actions. Actually, this is not the case. Even infants behave, in a different way with respect to chimpanzees. To shed light on this point it is interesting to examine what happens in their first months. As it appears in various studies on infants, they establish interactions with adults very precociously. More precisely they establish interactions with adults before nine months, i.e. before the appearance of language and before the comprehension of intentional action. Trevarthen (1977) has well described these interactions calling them interactions without object. They are the implementation of intersubjectivity per se. A six-month-old infant is not able to communicate about a specific meaning. However, the infant is able to perform her part in interactions with adults using gestures, sounds, and face expressions.

The works that more clearly explain the infant’s contribution to these interactions are the studies on interruption. It has been proven in experimental settings with 6/12-week-olds that if the adult during an interaction averts her attention, the infant will try to recuperate it (Tronick, Als and Adamson, 1979; Murray and Trevarthen, 1985). In particular, in Murray and Trevarthen (1985) it was shown that not only infants acknowledged the interruption but also that they distinguished between a “natural” interruption when the researcher distracted the mother and the unnatural situation in which the interruption consisted in the mother suddenly presenting a blank face. While in the first case there was only a reduction of positive excitement, in the second situation the infants appeared disturbed passing from protest to signs of distress and finally to withdrawal. But there was a third experimental condition that is particularly interesting. In this case the mother and the infant were in two separate rooms and they interacted viewing each other in a life-sized video image immediately before them. After some minutes of normal interaction, the communication was perturbed showing mother’s behaviors that occurred in a previous time and were not correlated with the present infant’s behavior. While during live communication the infant behaved as in normal face-to-face interactions, in the replay phase the reaction of the infant was one of distress similar to the reaction shown in the blank phase situation. This third condition has been utilized also to study the effect of perturbed interactions on mothers. In this case it was the mother who unknowingly was presented with her infant’s reaction to her previous behavior and then unrelated with her current one. Several mothers remarked that the interaction was odd and all of them changed their communication focusing more on their own experience than on the infants’ one (Murray and Trevarthen, 1986).

These results show that infants, when participating to their first interactions already have the ability to acknowledge:

  •  what an interaction is


  • what counts as an interruption.

The question is then how do they get this knowledge from. In fact there is no necessity to think that this is an explicit knowledge. What the infants react to is the format of the interaction, i.e. alternation: any gesture, sound of one of the participants is responded to by other gestures, sounds, etc. (Airenti, 2001). If this is not the case there has to be a good reason. The fact that the partner has started an interaction with someone else is a reason. On the contrary, for the infant a sudden blank face is distressing. The infant responds like an adult would do. What is interesting is that to perform this completely normal behavior there is no necessity to understand a specific meaning. The compliance or not to the format of interaction is sufficient. But the previous experiments tell us something more. In fact, alternation is not sufficient to denote the situation. Would the infant be able to acknowledge only reactions in a frame of alternation, any reaction should be acceptable. Actually, the infant is not satisfied if the response is taken in the repertoire of her mother’s responses but unrelated with her current behavior. As we have seen, submitted to the same conditions the mother too finds the situation incomprehensible and disturbing. Therefore there is a quality of the response that has to be attuned to the actor’s behavior (Murray, 1998). From this fact, it follows that to the feature of alternation another feature has to be added. We can denominate it reciprocity. The example mentioned before gives us an idea about what reciprocity means: when the mother puts food in the child’s mouth, the child in turn will try to put food in the mother’s mouth. Or if the mother tickles the child, the child will try to tickle the mother. This form of reciprocity starts very early and is perceived as a condition of successful interaction. In adults’ communication we expect that others respond to the point and in this case meaning is involved. In precocious interactions we can see that before real communication is possible, the format of reciprocity is already established. It is not easy to define what characterizes reciprocity in this sense. Trevarthen refers to expressive reciprocity (Trevarthen, 1998). Other authors consider that the main point is the mutual attunement of emotions (Stern, 1985; Hobson, 2002).

I would suggest that regarding at the problem in a more formal way, we can formulate the hypothesis that the reciprocal response has both to contain something of the behavior to which it responds and something new. Let us consider what happens in imitation. Reciprocal imitation is an important part of precocious interactions. Imitation is the simplest way to establish intersubjectivity because nothing is required except for the ability to express emotions by face expressions. Imitation in the context of reciprocity, not imitation alone characterizes the behavior of mother/infant pairs. Actually, imitation is performed not simply mimicking the other’s expression, but introducing little variations. For instance, often mothers repeat the infant’s gestures exaggerating them. Likewise, from one expression another similar or opposite can come out. Again, we can find here the pure format of what later will become real communication. In adult communication the partner has to acknowledge comprehension and to answer. The mother who finds her infant odd in the experiment does not find these two features in her infant’s response.

Collective action and communication

The previous considerations lead us to an alternative point of view with respect to the one expressed by Tomasello. Tomasello thinks that shared intentionality establishes continuity in child’s development from dyadic exchanges to collaboration. In fact, we can argue that an intrinsic motivation to reciprocity exists, which is specifically human and not linked to purposeful action and it appears in the first months. If instead of taking as the turning point the ninth month when the interpretation of intentional action and triadic exchange develop, we consider the preceding phase, we can see that the fundamental difference between humans and other primates emerges in the first months when children show their motivation to relate with others before action and cooperation are possible. Seen from this stance collaboration in action and sharedness due to the motivation to be related to another human follow two different paths.

Therefore, we can distinguish between two different ways to see the development of intersubjectivity. One considers that there are different phases. One early phase, before nine months, characterized by sharing behavior and emotions followed by a phase at around nine months where goals are shared and preluding to the possibility of collaborative engagement. According to this view the child that in the first phase just responds to behavior and emotions afterwards becomes able to understand action and to share goals with others. Thus, the first phase is overcome from the subsequent ones. The point of view I propose is different. Humans are endowed with a specific attitude to mutually interact. This attitude manifests itself in the first months and overstays all life long. This means that, contrary to the classical position that one finds in the philosophy of language, we can define communication separately from collective action. Communication is not just a form of collective action. We can change the formal features that we attribute to communication. Within philosophy of language the line of reasoning is that of logical derivation. In this perspective, communication derives its features from the features of rational action. This implies two major consequences. One is that to communicate the ability to acknowledge others’ mental states is necessary; the other is that it is impossible to comprehend what does the infant in the first interactions (Airenti, 2003). I think that we should look instead to what characterizes precocious interactions, i.e. interactions realized not only before the development of language but also before the development of the comprehension of the intentionality of action. At that stage we find three main features defining communication. It must be noted that these features are independent of action and language. Moreover, they do not disappear with development. What is in place since first interactions is the format of communication. We can define this format as follows:

  • participants share attention on each other’s acts
  • the modality of interaction is alternation
  • the acts must follow the frame of reciprocity

These features must be simultaneously present to characterize an interaction as communicative.

This model postulates continuity between infant and adult communication (Airenti, 2004). Actually, what is characteristic of precocious interactions is that the format is performed without the necessity of meaning. This is what happens in the experiment mentioned before where children of 18/24 months, after having retrieved a toy, put the toy where it was before to start the game again. In adults the possibility exists to have interactions where no meanings are involved but the only goal is to confirm or establish relatedness. This can be realized either by gestures, smiles, sounds, precisely as infants - and adults in relation with infants – do, or using language. Examples are the perfectly interchangeable utterances performed when for any reasons one is compelled to stay with unknown people (lift, doctor waiting room, etc.), or desires to comfort someone in distress, and so on. In adults exchanges as in children’s ones not any possible response is accepted within the format of reciprocity. Things are made more difficult by the fact that while infants have interactions with a very limited number of people, adults interact with many possible partners. Thus what counts as an acceptable response depends on the background constructed by all the preceding interactions (Airenti, 2005).

Engagement and reciprocity

We can now come back to the concept of engagement. As we have seen, this concept is generally used to designate two different phenomena, i.e. to explain both the involvement of infants in interactive situations and of adults in collective actions. The hypothesis I present here is that the two situations are fundamentally different. In fact infants involved in interactions with adults are experimenting the first basic forms of communications. In this sense communication is characterized as the intention to be in interaction with the other whether or not a specific meaning is present. Communication is a form of intersubjectivity (Airenti, in press). On the contrary, as far as action is involved we have two possibilities. One is that one actor just happens to have the occasion to profit of the action of someone else. If my neighbor is entering the front door before me, she will probably keep the door open for me to come in. We do not have a shared intention. We just adjust our actions. In this case the motivation is typically human, politeness. In the majority of situations, on the contrary, when two people do something together they share the intention to attain a specific goal. This can be very simple as lifting up an object too heavy for one person or more complex as in the case of musicians trying to perform a good interpretation of a trio. In this case there is the mutual engagement that each one will perform the part of the task she has been assigned.

Therefore, we can contrast two models. In the classical model we have two situations: individual actions, which are caused by intentions vs. collective actions (including communication), which are caused by we-intentions and imply mutual engagement. Collective actions are the manifestation of cooperation.

I propose an alternative model. Three kinds of action are possible, individual action, caused by individual intention, collective actions, which are caused by individual intention plus expectations about others’ actions, and shared actions. A collective action can fail even if one of the individual actions has been completely performed. The failure can be due to two different reasons. Either one of the actors does not perform her part, or the cooperation is not sufficient to achieve the goal, for instance the weight is too heavy also for two persons. Shared actions require mutual engagement. Often the other’s engagement is attributed on the basis of a presumed common interest to achieve the goal. Sometimes, to be sure that an actor will participate to a shared action an explicit engagement is requested, a promise, for instance or, even more formally, a contract.

Communication is a different form of interaction. Caused by communicative intention (Airenti, Bara and Colombetti, 1993), it implies sharedness and has reciprocity as a condition. In case of failure, the failure of attaining reciprocity implies the failure of communication. This is because the acts of the participants take their pragmatic meaning only as part of the shared situation of reciprocity. If the partner does not acknowledge a communicative act, the act is not accomplished. This is why the non-response situation is far more distressing than any negative answer.

If we accept these theoretical distinctions, the differences that we observe among infants, adults and chimpanzees become clear. Adults may perform according to all the possibilities we have just described. Chimpanzees participate to individual actions and collective actions without engagement, relying on the expectation of others’ actions. They do not have communicative exchanges. They can attract the attention of others’ to obtain their help but they do not perform their actions within the reciprocity frame. Infants have only the communicative intention and they afterward develop the comprehension of the intentionality of action and later the comprehension of collective intentionality. They do not use explicit engagement in cooperation till rather late.

In conclusion collective intentionality does not make the difference between young humans and chimpanzees. What makes the difference is reciprocity, which infants develop very early in life and that is the fundamental aspect of intersubjectivity all life long.

Humans are social beings and they develop their relationships with others since very early in life. If this is an obvious fact for everybody, there are many questions that are still open to discussion. In particular, three problems are controversial: Which are the bases of human intersubjectivity? Is it possible to distinguish human intersubjectivity from the relations that nonhuman primates establish with their conspecifics? Finally, is there continuity between the precocious forms of intersubjectivity and the more developed ones of the adult? I have suggested that the infants not only share emotions with their caregivers, but also already apply a cognitive format that is the basic structure of all communicative interactions. It is this format, based on reciprocity that constitutes the continuity between precocious and adult interactions. How this format can be compelling for humans is shown by the fact that it is imposed any time a possible interaction is conceived even with animals, plants or objects. In fact, anthropomorphism, this odd phenomenon where we see humans for instance proposing deals or asking for excuses, to their pets or even to objects of everyday use, like a car or a coffee-maker, becomes more explicable if we postulate that even in these cases humans use the only format they have at their disposal (Airenti, 2007). I have then argued that the bases of human intersubjectivity are also the cognitive bases of communication. To cooperate with others in view of attaining a common goal is a very different way to interact. This kind of interaction is based on collective action and it is common to human and nonhuman primates.

Starting from this general framework, during my stay at IEA I propose the organization of three seminars.

1 - Imitation, reciprocity and action in the development of intersubjectivity.

I will discuss

  • the game of imitation, considered as the first non linguistic form of acknowledgment of reciprocity
  • the relation between conflict and cooperation in infants and nonhuman primates
  • the origin and the development of anthropomorphism

11 - Sincerity in communication: from Grice to theory of mind.

In pragmatics sincerity is considered a constitutive aspect of communication. For Grice (1975, 1978) sincerity is one of the maxims that realize conversational cooperation. In Searle’s work on speech acts sincerity is one of the conditions of an illocutory act (Searle, 1969). In folk psychology it is an obvious fact that there are several forms of communicative acts where sincerity is not prescribed or is even banished. Thus, besides proper deceit, we have story telling, pretending, the use of metaphors, ironic statements, jokes, utterances where politeness demands not to be sincere. In all these cases insincere communicative acts are performed but the mental states underlying these utterances are different. There is no structural feature allowing to understand if a false statement was performed by error and then unintentionally – for instance due to insufficient information - or intentionally. And in case of an intentional insincerity, if it was aimed to realize a deceit or was intended as an ironic statement, etc. Therefore, in all these cases mindreading is necessary to comprehension.

A way to better understand these phenomena is to study how they develop from childhood. When do children learn to deal with these different situations? What is easy and what is difficult for them? How the developing capacity of mindreading contributes to this comprehension?

I shall present the results of my theoretical and experimental research on the development in young children of comprehension and production of insincere communicative acts: story telling, pretense, deceit, politeness, irony (Airenti et Angeleri, 2009a, 2009b).

In the seminar I shall focus in particular on:

  • what is shared in the comprehension and production of insincere communicative acts
  • the use of the different forms of insincere communicative acts by young children
  • the relationship between these acts and the development of the theory of mind
  • the relationship between truth, deceit and secret in children’s play.

I shall suggest that the use of these communicative modalities allows children to experience their first complex social relations.

Seminar organized by Gabriela Airenti (Paris IAS fellow)
05 Feb 2010 09:00 -
19 Feb 2010 13:00,
Paris :
Le développement cognitif et l'intersubjectivité

Other or several periods