Mentalization,embodiment,and narrative: Critical comments on the social ontology of mentalization theory

The assumption that the mind is essentially closed off: Forgetting embodiment

One line of critique that has been consistently directed against ToM approaches in recent years has a phenomenological heritage (e.g., Heidegger, 1927/2001; Merleau-Ponty, 1945/2012; Scheler, 1931/1973). According to this perspective, the view of the mind as an impenetrable interiority, only accessible through mediation, is phenomenologically inadequate. Rather, if one resigns on this assumption and attempts a phenomenological description of social cognition, the emerging view is that the basic understanding of the other person is not a developmental achievement, but given directly. In the contemporary debate, this view has been put forth primarily through an appeal to what is referred to as the direct perception hypothesis (e.g., Gallagher, 2008; Zahavi, 2011), which is often elaborated through an appeal to the Merleau-Pontian notion of “Intercorporeity” (e.g., Gallagher, 2005; Tanaka, 2015). The claim from phenomenology is that once we attend to the structure of experience, there is only rarely recourse to such things as theory or simulation in our daily interactions with and understanding of other people. Instead, the other presents himself directly to me through his embodied expressions. According to Merleau-Ponty: “I do not perceive the anger or the threat as a psychological fact hidden behind the gesture, I read the anger in the gesture. The gesture does not make me think of anger, it is the anger itself” (1945/2012, p. 190). In these experiences, the other person is not given as a split entity, divided between a hidden interiority and a mere physical appearance, but rather as what Scheler calls an “expressive unity” (Ausdruckseinheit). However, as Zahavi (2011) and Gallagher (2008) have recently pointed out, it is important to specify what precisely is meant by “direct” in this context. The point is not to understand direct as if the other person’s mind would conflate with mine in full transparency, but rather that my experience of others is not mediated through conscious inferences or simulations, as ToM would have it. My experience of others is indeed often contextualised by my personal history or social background, knowledge, etc., but not in the sense of an epistemic operation. ² As Merleau-Ponty specifies:

The sense of the gestures is not given but rather understood, which is to say taken up by an act of the spectator. The entire difficulty is to conceive of this act properly and not to confuse it with an epistemic operation [emphasis added]. Communication or the understanding of gestures is achieved through the reciprocity between my intentions and the other person’s gestures, and between my gestures and the intentions, which can be read in the other person’s behaviour. Everything happens as if the other person’s intention inhabited my body, or as if my intentions inhabited his body. (1945/2012, pp. 190–191)

This reciprocity between my intentions and the other person’s gesture, which must not be mistaken for an epistemic operation, is what Merleau-Ponty (1964) terms intercorporeity (intercorporéité) in his later writings. The important point is that the problem of understanding other minds only becomes an epistemic question in so far as one thinks of the mind as a Cartesian interiority, invisible from the outside. But this assumption only allows for two modes of being: subjective interiority and objective exteriority, which in turn entirely misses the phenomenon of expressivity, resulting from us being embodied beings. From this fact of embodiment, it follows that the very distinction between interior/exterior distinction is problematic; not only is the Other not given as mere exteriority with a hidden interiority but I am also given to myself ambiguously as both object and subject. Merleau-Ponty explains:

If my left hand can touch my right hand while it palpates the tangibles, can touch it touching, can turn its palpation back upon it, why, when touching the hand of another, would I not touch in it the same power to espouse the things that I have touched in my own? It is true that “the things” in question are my own, that the whole operation takes place (as we say) “in me,” within my landscape, whereas the problem is to institute another landscape. When one of my hands touches the other, the world of each opens upon that of the other because the operation is reversible at will, because they both belong (as we say) to one sole space of consciousness, because one sole man touches one sole thing through both hands. (1964, p. 148)

The dual aspect of embodiment, simultaneously constituted as both subjectivity (Leib) and objectivity (Körper), is co-given with the very structure of experience itself. The expressions of others are therefore not given in the mode of “a visible shell we need to move beyond” (Bateman & Fonagy, 2012b, p. 3) through the use of our imagination or inferences, as MT would have it. Qua embodied, the boundary between self and Other is not hermetically closed, but deeply intertwined in intercorporeity:

There is, between my consciousness and my body such as I live it, and between this phenomenal body and the other person’s phenomenal body such as I see it from the outside, an internal relation that makes the other person appear as the completion of the system. Others can be evident because I am not transparent for myself, and because my subjectivity draws its body along behind itself. (Merleau-Ponty, 1945/2012, p. 368)

It is important to realise that was this not the case, if the basic expressions of Others did not show themselves directly as the presence of another subjectivity, social cognition would never be able to get off the ground, as both Merleau-Ponty and Scheler have repeatedly argued. Rather, any resort to ideas of simulation or theory in social cognition already presupposes that which they are supposed to achieve. This is easily seen when one considers what would motivate interpreting the Other as Other in the first place and not only as yet another physical occurrence. Hence, TT and ST are not constitutive of the social, but presuppose it. This is, of course, not to say that the phenomenological account denies the possibility of hiding intentions and feelings, or that we cannot utilise simulations or theories in order to understand others. Hiding intentions and feelings is, of course, possible but the claim is that they are secondary acts and that our basic interaction is in no need of such deciphering or inferences. Resorting to means of mediation is secondary to our primary interaction, and only becomes necessary when our basic understanding breaks down.

In recognition of a similar line of critique, Davidsen and Fosgerau (2015) have recently argued that the concept of “implicit mentalization” could be interpreted as an expression of this phenomenological level of social understanding. According to Davidsen and Fosgerau, the phenomenon of implicit mentalization suffers the somewhat unfortunate fate within MT of being attributed a gross significance in the literature while never receiving a systematic theoretical articulation. In order to remedy this, they suggest that the phenomenological account of social cognition could serve this purpose.

I strongly endorse the suggestion of integrating phenomenological insights into MT. However, I do not believe that the phenomenological account of basic social cognition can readily be added to the existing framework of MT without imposing theoretical inconsistencies. In order to achieve a theoretically consistent integration of phenomenological accounts of social cognition, MT would need to revise the fundamental Cartesian assumption that the mind is essentially closed off. Furthermore, whereas MT sees social cognition as a developmental achievement, one that “is not constitutionally guaranteed” (Fonagy & Allison, 2014, p. 372), but is crucially dependent upon the attachment context, the most basic social cognition is exactly constitutionally guaranteed and primordial on the phenomenological account. This dispute is not merely terminological but an expression of the fundamental social-ontological assumption made in MT.

The assumption that social cognition is rooted in inner mental representations: Forgetting the enacted dimensions of social understanding

A similar criticism of ToM approaches has been raised from the field of enactivism, arguing that the idea of social cognition in ToM essentially remains a solitary enterprise; one in which one individual tries to decipher or simulate the other person’s mind through inner mental representations. This radically misses the phenomenon of the social and how social understanding is situated in interaction (De Jaegher & Di Paolo, 2007; Fuchs & De Jaegher, 2009). According to enactivism, basic social understanding should rather be understood as embodied and non-representational, which is elaborated as a dynamic coupling and coordination between two inter-actors—not unlike the idea of intercorporeity (Fuchs & De Jaegher, 2009). According to this account, social cognition is something that is constituted in-between, and which is “arising in the moment-to-moment interaction of two subjects” (p. 466). The intentions involved and expressed in interaction are often generated and transformed throughout the process of interaction. In this way, the interaction process gains “a life of its own” and a “kind of autonomy” (p. 471). These patterns of interaction can again sediment into identifiable patterns; a kind of disposition of the participants based on the history of interaction.

How does this critique apply to MT? At face value, MT should not be alien to the notion of interaction due to its significant emphasis on the importance of early interaction between caretaker and infant in an attachment context. But the vital difference is, once again, that MT starts from a notion of the mental as closed off and rooted in inner mental representations. Mentalizing is essentially an ability attributed to the individual that can be more or less developed, rendering social understanding an individual matter and not something arising out of the particular interaction. Mentalizing is an individualised ability to develop adequate inner representations of others, rooted in the mentalizing models established in early childhood. In the case of BPD, this capacity is supposedly impaired and further weakened in heightened states of arousal (Bateman et al., 2012). As Fuchs and De Jaegher remark, this differs fundamentally from their account of interaction:

This account of inter-subjectivity is quite different from concepts of mentalization and mind-reading. To illustrate this, let us take the example of an infant of a borderline mother, which has learnt to withhold its impulses to approach the mother because of her repeatedly aggressive behaviour. According to Fonagy’s mentalization concept, the infant would inhibit its ability to reflect on the mother’s affect because of the unbearable content of the representation of her aggressive affects. The alternative view is that the mother’s hate is expressed through the quality and dynamics of the embodied interaction, e.g., repeatedly rejecting the infant’s approach or overriding its initiative. As a result, these interaction sequences are taken up in the infant’s implicit memory in their process form, not their content form, e.g., as a bodily tension and resistance which the child builds up against his own impulses to approach the mother. He does not need to create a representation of the mother’s inner state but enacts his implicit knowledge in the situational context, as an avoidant coping style. (Fuchs & De Jaegher, 2009, p. 481)

From the enactivist approach to social cognition, we are, furthermore, perfectly able to account for the consequences of problematic attachment patterns as seen in BPD cases without reference to inner mental representations. Rather, attachment issues could be seen as resultant from the particular history of interactions in the individual, which sediment into a specific, embodied response and interaction repertoire; one that affords itself to be re-enacted in attachment relations with adequate similarity: explaining part of the aberrant attachment style found in BPD.

From the emphasis placed on interaction in the MT literature, I believe that enactivist analysis, such as this, would be welcomed in MT. This would, however, once more, require a fundamental revision of the social ontological assumptions made in MT. MT would need to revise the strong emphasis on inner representations and allow for more enacted and embodied views on social cognition; a view that sees social understanding as emerging from situated patterns of interaction in context.

The assumption that social understanding can be tracked back to the infant/caretaker dyad: Forgetting the sociocultural basis of social understanding

The critical comments made so far have primarily addressed the levels of social cognition pertaining to our everyday face-to-face social interaction. However, an account of social cognition that does not include how we understand each other through more mediated settings and situations would hardly be convincing. In accounting for this dimension of social cognition, we can once again find ground for questioning the social ontology of MT. As we remember, according to MT, models of social understanding owe their ontogenesis to the early infant/caretaker interaction, placing a tremendous emphasis on the quality of this early relation. Some proponents of MT take this to express no less than a Hegelian position, in so far as they see something like the dialectics of recognition in this early interaction (e.g., Jurist, 2010). Apart from rousing speculations as to how Hegel might have felt about being associated with the broader framework on social cognition found in MT, I believe it is important to raise the fundamental question of whether our capacity for more complex social understanding should exclusively be traced back to the early dyadic attachment context.

From within philosophy, two lines of criticism have been raised to this point, proposing a different social ontology which emphasises that complex social understanding has a sociocultural basis. The first line of criticism points out that in most everyday social situations, we do not employ any such thing as mentalizing models in understanding others; rather, the intentions of other people suggest themselves from the affordances and the norms regulating the pragmatic contexts we engage in (Gallagher & Hutto, 2008; Ratcliffe, 2007). In order to understand why a person stops her car before a red light, I need not consult either theory or simulation to read the intention of others; the intention is directly available through familiarity with the norm-governed practical context. Similarly, the implicit social rules governing how to wait in line at an English supermarket structure that everyday activity and relieve us from any speculation about what the other person might be thinking. Only in rare cases where these norm-governed practices break down might we need to re-coordinate or, in some case, even theorise about the other people’s intentions. What is important to notice is that the social understanding emerging in these situations should in no way be understood as inner mentalizing models, but rather as broad sociocultural frameworks that the child gradually appropriates through engaging with others in a variety of contexts. One way of specifying the alternative social-ontological framework pointed to here, is to follow Bernhard Waldenfels’ (2015) distinction between “frontal” vs. “lateral” variants of sociality. Whereas frontal sociality refers to the type of “face-to-face interaction” which completely dominates the framework of MT, lateral sociality designates the equally prominent dimension that we find ourselves “side-by-side” in a shared relation to a cultural world of practices, norms, objects, etc., which are saturated with social reference and meaning. Whereas the early attachment context might supply the child with a basic understanding of the rhythm and structure of this lateral dimension, it hardly seems persuasive to claim that these forms of social understanding exclusively or even primarily owe their ontogenesis to the early infant/caretaker interaction. Rather, these norms are learned from actual engagement in these social practices in their proper contexts.

The second line of criticism starts from the obvious fact that in spite of our more immediate understanding of others—present in expressions, participatory sense-making, and norm-governed interaction—social understanding does sometimes break down, forcing me to speculate about the intentions and feelings of other persons. But, as Daniel Hutto (2008) has recently argued, in such cases—much less frequently than we tend to assume—the attribution of beliefs and desires, or attempting simulations, would be a radically individualised and decontextualised means of social understanding. The problem is, as Hutto points out, that “having an understanding of beliefs is logically distinct from having an understanding of what it is to act for a reason” (p. 26). Whereas belief attribution is a somewhat rudimentary skill, usually considered to be developed around the age of 4–5, the ability to understand actual reasons for acting is not only a much more complex phenomenon with a considerably longer developmental trajectory, it is also highly dependent on a second-person perspective. This is not to say that belief/desire attribution is not part of the process of understanding reasons for acting but that a lot more is always implied for this to amount to an actual understanding. What we need, according to Hutto, is an understanding of how these propositions contextualise in a particular person’s life. The medium for this is narrative. ³

The idea is quite intuitive: imagine a man suspects his wife might be cheating on him due to what he considers suspicious activities on her part. In order to dismantle this suspicion, the man would hardly be satisfied by a mere confirmation of her commitment to him (desire to be with him, belief that he is the one she loves, etc.). Instead, what he needs is an account of her concrete actions and for them to be placed into the broader development of events; one that is at least consistent with her character, as well as the context she is currently in. A narrative will achieve exactly this. Hutto frames this notion in what he calls the “Narrative Practice Hypothesis” (NPH). According to NPH, “the normal route [to understanding reasons] is through encounters with stories about people who act for reasons” (Hutto, 2008, p. 28). It should be noted that although Hutto (2008) is very liberal as to what qualifies as a narrative, what he has in mind is the myriad folk narratives that any child’s life will be discursively emerged in. From these narratives, the child gets a basic, yet richly contextualised, understanding of how reasons to act emerge in context from the interacting elements of character, personal history, and broader social and material circumstances (p. 34). This is not to suggest that these stories are idealised types:

They are snapshots of the adventures of situated persons, presented in the kinds of settings in which all the important factors needed for understanding reasons are described – that is, those that are relevant for making sense of what is done and why. (p. 34)

Narratives serve as exemplars to train and rehearse us in contextualised understanding and this kind of narrative competency is a sophisticated, discursive skill in need of a cultivation that undoubtedly transgresses the confines of the infant/caretaker dyad. As Jerome Bruner has famously argued, training a person’s narrative competency should be seen as a broader educational task (Bruner, 2009; see also Gallagher, 2014).

Furthermore, employing narratives in attempting to compensate for the failure of more immediate understanding also implies a strong second-person and interaction-based perspective. As Hutto (2008) remarks, in these situations of failed social understanding, by far the best solution is to engage in an actual conversation with the other person and ask them to explicate and explain why an action was performed. Recently, Tim Henning (2009) has explicitly taken up this line of thought and argued that narratives play a constitutive role in the social epistemology regulating our interpersonal understanding on the level of normative justifications for actions. The intuition Henning is referring to is that whereas we usually rely on a preconceived understanding of the character and actions of the people we are familiar with, based on our interaction history with them, we are sometimes confronted with the fact that a person acts out of character. In these cases, the epistemology governing interpersonal understanding requires an account for this shift that goes beyond mere belief/desire attribution. To illustrate, Henning asks us to imagine a conversation between two people: person A and person B. Throughout his life, person A has consistently prioritised career and personal hobbies, swearing never to settle down in a suburban life with a wife and kids. However, in conversation, it emerges that these priorities have suddenly shifted, and that person A now wants to marry a woman he recently met, have kids, move to the suburbs and quit the job he loves, due to these circumstances. Confronted with this, person B would find himself requiring an explanation in order to understand that this is actually still his friend he is talking to. Person B will attempt such an understanding through prompting what Henning calls critical “since-when-questions” (p. 19). In this case, person A is normatively expected to provide a narrative that gives an adequate, temporal account for the development of this change in his character. Meeting these normative requirements would call for extensive references to the elements taught to us through narrative as justifying reasons for acting in context. Again, in this case, social understanding is not something regulated by inner mentalizing models of how people act, but rather through a contextualising and interactive epistemology that goes beyond basic belief/desire attributions and is extensively regulated by social norms: what exactly counts as a good explanation is regulated by normative expectation.

What does this imply for MT? Whereas NPH and MT share the claim that social understanding is a developmental achievement, they differ exactly in their appreciation of the role of context in this development. First of all, whereas MT reduces social understanding to an affective and rational level in the developed mentalizing model (Fonagy, 2006), NPH places tremendous emphasis on the context of action, focusing on how the particular action was influenced by the particular characters, personal history, larger projects, past choices, existing commitments and circumstances, etc. Though MT sometimes mentions the importance of narrative accounts, it has very little to say about this story-telling context and instead stays at the abstract level of inner mental representations. Second, whereas MT seems to confine the achievement of social understanding to the mentalizing models developed through the infant/caretaker dyad, NPH proposes that our capacity for complex social understanding is a gradually developed narrative competency appropriated throughout the various contexts of operations we engage in and one that is nurtured by the broader sociocultural context. This is not to suggest that primary caretakers do not play an important role in scaffolding this process, but rather that the ontogenesis of complex social understanding is seen as fundamentally sociocultural in nature, and not primarily based on the attachment dyad.

Finally, in restricting the development of mentalizing models to the infant/caretaker dyad, it is worth considering whether or not MT might be susceptible to a variant of what Gallagher (2012) has called “the diversity problem” in simulation theories. Gallagher proposes that the simulationist idea—that we use our own experiences to project ourselves imaginatively into the Other’s perspective—is problematic because it is nowhere given from this account of social cognition that my mental life would be similar to that of the other. In an equivalent manner, one might raise doubt as to whether the caretaker/infant dyad is an adequate model for the diversity of motives, beliefs, and reasons people have for acting across social contexts and cultural diversity, and, therefore, as a model of social understanding. Rather, does the diversity of culturally patterned actions not imply the need for a richer platform for understanding reasons?