5 The person model theory
Before expounding the new account, let me highlight two main criteria of adequacy for any plausible candidate theory and some open questions. (i) The theory should account for two levels of understanding others from a phenomenological perspective, namely intuitive understanding and inference-based understanding. This was first clearly discussed by Gallagher (2001), while Goldman (2006) described it in his distinction between low-level and high-level mindreading. What, we may then ask, would be an adequate way of establishing this distinction? (ii) We learned from Gallagher (2005) that we should distinguish understanding others by observation from understanding by interaction.
There are also a number of open research questions that can potentially be answered in developing the alternative account: (a) What is the relation between understanding oneself and understanding others? Here the ST claims that understanding oneself is the basis for all understanding of others, while TT is neutral; Carruthers, for example, has famously argued that understanding others is the source of our self-understanding (2009). (b) What is the relation between understanding persons and understanding objects or situations? (c) How can we best account for the difference between understanding a well-known person, on the one hand, and a complete stranger, on the other?
The new alternative theory, which promises to deal with these open questions, is the person model theory. The central claim of this theory is that we organize our prior knowledge that is used to understand others into something we can call person models, and that accounting for our way of using person models is the most informative factor when analyzing our everyday understanding of others. A person model[7] is a unity of properties or features that we represent in memory as belonging to one person or a group (resp. type) of persons. To account for the difference between two types of understanding others (intuitive versus inference-based understanding), I suggest that there are two types of person models in use: implicit person models,[8] which we shall call person schemata; and explicit person models, which we shall call person images. Very early in life we develop person schemata: a person schema is an implicit person model and can typically be described as a unity of sensory-motor abilities and basic mental phenomena[9] realized by basic representations and associated with one human being (or a group of humans), where the schema typically functions without any explicit considerations and is activated when directly seeing or interacting with another person. A person schema is thus the unity of implicitly-available information about a person that is thus not easily accessible in terms of being reportable but is nevertheless used in a specific situation. In other words, a person schema is the basic unit that enables a practical knowledge (a knowledge how) for dealing with another human being while this ability relies mainly upon social perception and interaction. Person schemata can be developed step by step into person images. A person image is a unity of explicitly represented and typically consciously available mental and physical phenomena related to a human being (or a group of people). Thus, a person image is the unity of rather easily and explicitly available information about a person, including the person’s mental setting. Both person schemata and person images can be developed for an individual, e.g., one’s mother, brother, best friend, etc., as well as for groups of people, e.g., medical doctors, homeless people, managers, etc. Furthermore, person models are created for other people but also for oneself.[10] In the case of modelling oneself we can speak of a self-model that we develop implicitly as a self-schema and explicitly as a self-image. Thus, we have the following varieties of person models (see Table 1).
Person models are characterized here as memorized units of person features, ignoring the difference between long-term or short-term memorization.[11] Person models are distinguished from the result of understanding in a situation, which may be either a person impression that mainly relies on person schemata, or a person judgment that mainly relies on person images. Let me illustrate one clear virtue of adopting the distinction between person schema and person image by reference to the fact that it can account for the difference between intuitive understanding and inference-based understanding of others.
Table 1: Varieties of person models
5.1 Person schemata
In detail, then, what are person schemata? A person schema is an intuitively formed, implicit model of a person; it is a memorized unity of characteristic features of a person including facial features and expression, voice, moving pattern, body posture, gestures, and other perceivable features of a person. The function of clustering these features is to allow us to evaluate a person very quickly in a situation according to evolutionarily-important aspects: is a person familiar, dangerous, aggressive, helpful, or attractive? The evaluation is either expressed in a type of interaction, or it can simply be memorized in an implicit unitary structure for future retrieval, including recognizing the person and activating the former evaluation (Reddy 2008). Our main access to others in everyday life is through perceiving a person and forming an impression (see the review published as a book chapter by Macrae & Quadflieg 2010). To form a person impression, (i) we typically pick up these basic features by means of a quick visual evaluation, even when seeing a person for the first time, where (ii) most features are directly associated with socially-relevant information, and (iii) they are clustered at the level of perceiving the whole person. Let me offer some support for all three characteristics of the process of forming a person impression in a situation that is memorized as a person schema:
(i) Quick evaluation even with parsimonious information: Evaluations of threat (which is of strong evolutionary relevance) can be made on the basis of exposure to an unfamiliar face lasting as little as 39 milliseconds (Bar et al. 2006). If the exposure to the unfamiliar face lasts about 100 milliseconds, we are able to evaluate likeability, trustworthiness, competence, and aggressiveness with subjective reliability levels that are similar to those generated under longer viewing times (Willis & Todorov 2006).[12]
(ii) Most features are associated with socially relevant information: looking into the face is a very rich source of information about a person. Between 3 and 7 months of age, infants learn to recognize the face of the mother and to distinguish it from the faces of strangers, and they start to categorize people according to emotional expression and sex (Nelson 2001). One important source of information that children use from 4 months onwards is the gaze-direction of a person, it having been shown that they can distinguish a direct from an averted gaze (Vecera & Johnson 1995). From 9 months onwards, infants learn to register the joint attention of the infant and an adult as directed towards an object (Cleveland & Striano 2007). Thus, on the basis of gaze-interaction they evaluate whether joint attention towards an object has been established or not, and learn to direct the attention of the other if necessary (Tomasello 1999). Between the ages of 9 and 18 months, children start to use gaze-information to register the goal of the action of the other human: they attend immediately to the eyes when the intentions of an actor are ambiguous (Phillips et al. 1992).
Let me now pick out some results based on studies of adults that illustrate the informational value of single cues. To start with facial expression: in emotion recognition, highly informative features include knitted eyebrows for sadness, a smile for happiness, and a frown for anger (Ekman 1972, 1999). To prevent this remark giving the wrong impression, I here highlight some individual features and will argue in the next step that they are part of an integrated view at the level of persons. Salient biological visual markers allow us to easily identify the “big three” categories in person perception (Brewer 1988; Fiske & Neuberg 1990), i.e., sex, race, and age. In the same way, we can illustrate highly informative single features such as body posture: if the other is bending her head in a communicative context, this is unconsciously registered as signalling sympathy (Frey 1999).[13] One important data source here is biological motion-detection as investigated by point light studies. If a person has lights on her hands, feet, and ankles, and some other significant parts of her body, we can videotape her bodily movement in the dark. Such artificial pure biological movement information allows us to register social features, e.g., we can recognize emotions (Ambady & Rosenthal 1992) and attribute personality features (Heberlein et al. 2004) on the basis of seeing dynamic movements alone. Furthermore, there is evidence that social information can be taken from the combination of gesture and body posture alone. In an intercultural study (Bente et al. 2010), an interaction between an employer and an employee (played by two students of one type of culture) was filmed for a short period. Then the film was edited to show only gesture and body posture. This was realized by showing idealized wooden puppets, representing the real interaction while abstracting from facial information, speech, clothing etc. The question to be addressed was, what we can read from seeing the body postures and gestures. The interactions were filmed with students from UAE (United Arabic Emirates), Germany, and the United States; and the test subjects were also drawn from all three countries. With this film, people could determine whether the people in the scene were nervous or not, as well as the dominance relation, i.e., they could see who was the boss. This is an interculturally-shared social understanding of otherwise culturally variable cues of body posture and gesture (the US students moved a lot while the UAE students moved rarely). They furthermore could perceive the level of friendliness in the interaction, although the study showed that we are good at this only in assessing our own culture.[14] Furthermore, there are many more complex culturally-dependent visual features that (according to other studies) we use for evaluating the other—e.g., physical attractiveness, where attractive people are evaluated as possessing more desirable characteristics than their less attractive counterparts, a phenomenon that has been labelled the beauty-is-good stereotype (Dion et al. 1972; Eagly et al. 1991). These kinds of stereotypes are especially connected with racial classifications: African–Americans are stereotypically assumed to be lazy, criminal, and uneducated, but also musical and athletic (Devine & Elliot 1995), whereas Asian–Americans are considered to be intelligent, industrious, conservative, and shy (Lin et al. 2005). Most observers in our culture assume that people with stylish hair and extravagant clothing are highly extrovert (Borkenau & Liebler 1992). We live with a lot of these deeply culturally-anchored stereotypes, and they are often applied without the perceivers’ intention or conscious awareness (Macrae & Bodenhausen 2000). This last point relates to the third aspect of person schemata. Person schemata are unities of characteristic features integrated at the level of persons. All these singular features are integrated into person models that enable us to develop detailed and extensive expectations of behaviour.
(iii) Integration of characteristic features at the level of perceiving the whole person: Although I have presented evidence that some single features are very salient for transferring social information, there is also much evidence that these features are normally combined with a variety of others to form an integrated impression of a person that I call a person schema. We have seen evidence for the key role of gaze detection in registering another person’s direction of attention (see ii). But there is further evidence that gaze alone is not the critical source of information; we actually seem to rely on an integrated evaluation on the basis of perceiving gaze, head, and body position (Frischen et al. 2007). The same holds for evaluation of the basic features sex, race, and age. Although isolated facial features are often sufficient to determine a person’s sex, research has indicated that sex categorization is based on the integration of several features (Baudoin & Humphreys 2006; Bruce et al. 1993; Brown & Perrett 1993; Roberts & Bruce 1988; Schyns et al. 2002). Concerning face, the best available theory of face recognition seems to be Haxby’s account (Haxby et al. 2000), according to which there are two distinguishable processes, one leading to face identification by focussing more on invariant core features, and the other leading to registering facial expression by relying on varying features. Furthermore, there is evidence that there are two different neural circuits for face perception and body perception (see the review by Macrae & Quadflieg 2010), both playing a core role in registering face or body identity, and playing an extended role in registering face or body expression in a given situation. And the integration processes are not limited to this level (Martin & Macrae 2007). Since we know that information about facial and bodily features is integrated, e.g., in the evaluation of emotional expression, we can therefore characterize a sequence of integration processes as leading finally to a person impression in a situation, which may be stored as a person schema in memory.
5.2 A model of forming a person schema
How can we best describe this process of forming a person schema? In general terms, the same complex process takes place in the case of perceiving a person and forming a person impression in a given situation as takes place when we perceive an object. I describe the process according to the model of object perception developed by Ernst & Bülthoff (2004), and I have already shown in detail that it can do justice to our recognition of emotions (Newen et al. forthcoming). The overall process comprises bottom-up processes starting with basic visual features that are modulated either by feature combination (if two features provide complementary information), or by feature integration. The latter can be modelled as a Bayesian weighting process that leads to the most probable intermediate estimate given the input. Further integration processes then lead from the most probable estimate to a stable percept of an object in the case of object perception, and to a stable person impression in the case of person perception. This model explicitly accounts not only for bottom-up but also for top-down processes, in the form of so-called cognitive penetration. I have sketched a plausible but in no way complete model of the formation of a person impression (see figure below). According to the evidence I have presented so far, it is plausible to suggest that at the level of intermediate estimates in the process of forming an impression of a person, we find (a) an estimation of a core person identity, (b) an estimate of situational emotions, intentions, and actions, as well as (c) an estimation of social status, person abilities, and individual personality traits. An important step in the model is the association of visual features with socially-anchored stereotypes (see above) which allows us to develop rich intermediate estimates, e.g., of the other’s emotional situation, social status, etc.
Numerous lines of research (Albright, Kenny, & Malloy, 1988; Ambady & Rosenthal, 1992; Behling & Williams, 1991; Borkenau & Liebler, 1992; Kenny, Horner, Kashy, & Chu, 1992; Norman & Goldberg, 1966; Secord, Dukes, & Bevan, 1954) have provided compelling evidence that trait evaluations are readily drawn from a person’s physiognomy (i.e., facial features), outer appearance (i.e., clothing), or demeanor (i.e., posture, walking, style). (Macrae & Quadflieg 2010, p. 433)
Finally, I highlight that the top-down processes are able to interfere in this process of combination and integration very early in the visual information processes: for example, it has been shown that the activation of a race concept on the basis of the form of a face (African versus European face format) changes the perception of colour in the face, while colour is known to be represented in V4 as part of early visual brain processes. The same hue of colour is seen as more dark in the African face than in the European face (Levin & Banaji 2006). Thus we have to admit that the process of feature-combination and integration is highly dynamic, involving simultaneous activation of features rooted in bottom-up and top-down processes, finally reaching the most probable and usually stable person impression. The dynamic is described in detail for the case of object perception in Vetter & Newen (2014); it is postulated for person categorization in Macrae & Martin (2007), and analysed according to the levels of processing that lead to person construal in Freeman & Ambady (2011). Figure 1 is a sketch of the formation of a person impression according to my account.
Figure 1: A model of the dynamics of bottom-up and top-down processes leading to a stable person impression by relying on person images and/or person schemata
A person schema emerges as the result of direct perception of a person, where this may be either basic or relatively smart perception; yet it usually remains implicit, and is not amenable to linguistic description. A typical example of person schema based on basic perception is the everyday experience of seeing a person only briefly in a single situation, whereupon it is difficult for us to describe the person—particularly her face. While we can often easily recognize the person, it may take hours with a professional to end up with an adequate “identikit” picture such as those produced at police stations. A person schema based on smart perception might be, for instance, a person schema that includes a lot of top-down activation—for example, while on campus, perhaps I see a person of typical student age dressed like a law student, and thus activate the “rich person” schema that is the basis for my everyday smooth interaction with law students, and which differs (despite overlaps) from my person schema for students in natural sciences. If we not only develop implicit practical knowledge regarding our use of the person impression (independent from its richness), but also develop explicit knowledge pertaining to the relevant person information, or at least develop easy explicit access to it, then we go beyond a person schema. We can characterize this new unified information as a person image.
5.3 Person images
In detail, then, what is a person image? A person image is a unity of relatively easily and explicitly available information about a person, including her mind-set. On the basis of typically implicit person schemata, young children learn to develop explicit person images. These are models of individual subjects or groups. In the case of individual subjects, they may include names, descriptions, stories, whole biographies, and visual images highlighting both mental and physical dispositions as well as episodes. Person images are essentially developed not only by observation but also by telling, exchanging, and creating stories (or “narratives”).[15] Person images presuppose the capacity to explicitly distinguish the representation of my own mental and physical phenomena from the representation of someone else’s mental and physical phenomena. This ability develops gradually, reaching a major and important stage when children acquire the so-called explicit theory-of-mind ability (operationalized by the false-belief task, see Wimmer & Perner 1983).[16] Then they are able to construct explicit person images by characterizing a person such that they attribute a biography to an individual. There is strong folk-psychological evidence that we have explicit person models of the people we deal with extensively, e.g., family members, and people about whom we tend to have a lot of explicit knowledge. The same is true for relevant groups of persons we deal with often. Even in professional contexts this leads to judgments that can be inadequate: the apparent association between wearing revealing clothes and immodesty and promiscuity has been shown to cause not only laypeople but also police officers and judges to hold victims of rape to be responsible for their having been assaulted (Lennon et al. 1999). An essential part of becoming an adult is learning to interact socially with other humans, by developing sophisticated and explicit person images of the groups of professions we have to come to any sort of arrangement with. We often have explicit beliefs about medical doctors, managers, secretaries, craftspeople, etc., and we try to deploy these beliefs to deal with these people in a smooth and efficient way. When we have stored a person image in memory, and are placed in a new situation in which we see and recognize the person, there is evidence that we immediately activate the biographical knowledge we have available. For example, when test persons were asked to judge the traits of target individuals from photographs, the test persons’ responses continue to be influenced by what they have explicitly learned about the people in question (Uleman et al. 2005). A recent neuroimaging study (Hassabis et al. 2013) indicated that when test persons were asked to predict the behaviour of persons, they essentially relied on prior knowledge of personality traits, which in this particular study were implemented in two ways, namely as agreeableness (the tendency toward altruism, cooperation, and the valuing of harmony in interpersonal relationships as opposed to antisocial and exploitative behaviours) and as extroversion (in contrast to introversion). The test person became acquainted with four types of personalities that had been constructed from combinations of high and low versions of agreeableness, on the one hand, and high and low versions of extroversion, on the other. In the test situation they had to predict the behaviour of four specific persons who were exemplars of the four personality types. The authors report that the predictions of behaviour were mainly based on personality traits and that the latter also had rather clear neural correlates: by using functional magnetic resonance imaging (fMRI) the authors showed that there is a neural correlate for recognizing (and imagining) high agreeableness (in contrast to low), namely in the left LTC (lateral temporal cortex) and dorsal mPFC (medial prefrontal cortex), as well as for recognizing (and imagining) high extroversion (in contrast to low), namely in the pCC (posterior cingulate cortex); in addition the recognition (and imagination) of one of the four personality types was correlated with four distinctive patterns in the anterior medial prefrontal cortex (mPFC). In line with my proposal, the authors of the fMRI study write: “Different patterns of activation in the anterior mPFC could reliably distinguish between the different people whose behavior was being imagined. It is hypothesized that this region is responsible for assembling and updating personality models” (Hassabis et al. 2013). Since the study was based on explicit evaluation of personality features or types, I take this to support the existence of person images. Yet even if the reader accepts the idea of person models, she may be sceptical about whether we need to distinguish person schemata and person images.
5.4 Why should we distinguish person schemata and person images?
A very convincing case that forces us to make a distinction between person schemata and person images comes from taking a closer look at a typical patient suffering from Capgras syndrome, a misidentification syndrome. Sufferers have the delusional belief that one of their closest relatives, e.g., their wife, has been replaced by an impostor. Such a patient typically says things like “this person looks exactly like my wife, she even speaks and behaves like my wife and she expresses her typical desires but she is not my wife” (Davies et al. 2001); thus, one aspect of this mental disorder is the observation that all the features explicitly believed to be possessed by the wife are correctly attributed. We can account for this by asserting that the patient has an intact person image of his wife. Nevertheless, the usual person identification has gone wrong. According to a standard analysis, what is lacking in the case of the Capgras patient is a feeling of familiarity that normally comes with perceiving a well-known person. How can we account for this in the new framework? When perceiving his wife, the subject intuitively develops and activates a person schema. One aspect of the person schema is the person’s identity.[17] As the Capgras case nicely illustrates, the registration of a person’s identity is a result of an integration process that relies not only on visual features but also on an implicit emotional evaluation, and that these together trigger an explicit judgment. While the visual recognition fits, here the emotional evaluation is inadequate and the feeling of familiarity is lacking; and in the case of this disorder, the Bayesian integration process for these features leads to an implausible result, since the emotional mistake overrides the visual adequateness. Thus, the Capgras patient has an adequate person image of his wife but an incorrect person schema, and the tension between the two is solved by developing the (implausible) hypothesis that she is an imposter. This analysis is in line with two-factor theories of the Capgras disorder, according to which two distinct factors cause the phenomenon[18]: first, the lack of familiarity, and, second, a local breakdown of rationality that enables the irrational belief-formation on the basis of a severely disturbed person schema (Davies et al. 2001).[19] Several other cases seem to be accounted for if we accept the evidence for a two-factor theory of person modelling—namely a first level of intuitive and implicit person impression and a second level of explicit person evaluation, which are described respectively as intuitive person schemata and explicit person images.
A contrast case to Capgras syndrome is the Fregoli syndrome, wherein a patient has the delusional belief that one and the same person, usually a persecutor, is following her, who is able to radically change his outer appearance. The sufferer then connects people with rather different outer appearances and treats them as the same persecutor. One explanation, still in need of testing, is that this time the feeling of familiarity is developed too often, probably by top-down initiation due to the delusional belief that the subject is being persecuted. The delusional belief, together with an inadequate feeling of familiarity, may explain the syndrome.[20] But again we need to distinguish the two factors: a level of implicit feeling or impression, and a level of explicit judgment. This time the delusion produces a breakdown of rational judgment formation, i.e., the person model of the other is strongly influenced by the delusion: the person schema formation may be largely intact but has a local defect due to being dominated by the delusional belief. In general, monothematic delusions (delusions about a single belief content) seem to rely on two factors (Coltheart et al. 2007): “[o]ne factor has to explain the strange experiences patients claim to have, while the other factor has to explain the misattribution of actions and thoughts” (Vosgerau & Newen 2007, p. 40).
Are there nonpathological everyday cases that support the distinction between person schema and person image? One illustration can be drawn from Mark Twain’s “Huckleberry Finn.” At first Huck helps the slave Jim to escape from slavery; but then he rethinks his support in the light of the law, and forms the judgment that he should turn him in to the slave-hunters. But when he has the opportunity to do so, Huck actually ends up protecting Jim. Why does he do this? Huck has a person schema of Jim that is constituted by a person impression according to personal interactions that are dominated by empathy; thus he has a positive impression of Jim and there exists between them a growing friendship. On the other hand, he has a person image of Jim that is dominated by the fact that he is a slave, such that he has to accept his role in society, to do the hard work, to live without freedom, and thus that it is forbidden to aid his escape. Cases of tension between an intuitive person impression (being helpful, being peaceful) and a person image dominated by the knowledge that the same person is a pathological murderer are often reported by judges and policemen. A less dramatic tension seems to be part of our everyday experience of “false” friends (we may still think of someone as a friend while implicitly already noticing signs of unfair treatment), though of course the tension can also exist the other way around. As illustrated above, the visual features of a person are often loaded with social information, and often involve the activation of negative prejudices which, after a more careful investigation of the person, can be opposed by a positive person image. The general functional role of person models is to simplify the structuring and evaluation of social situations, to enable a quick evaluation of the person in a given situation, and to initiate adequate behaviour. An additional special functional role of person models consists in stabilizing my self-estimation, since there is a strong tendency to have positive stereotypes of one’s own in-group members and negative stereotypes of the out-groups’ (see Volz 2008, p. 19). These examples illustrate not only that we need to distinguish the person schema and person image, but also that we have a tendency towards harmonizing both. Thus, if one of them is disturbed we tend to adjust the other, which may result not only in wrong judgments about persons, but in extreme cases may become an aspect of a mental disorder, as described above. Finally, to distinguish them is compatible with the claim that a person image may often gradually evolve on the basis of a person schema such that partially the same information about a person changes the status of accessibility from implicit to explicit. But we also have to distinguish both kinds of person models because often an implicit representation of a person as unfriendly exists simultaneously with an explicit evaluation of the same person as friendly.
5.5 Person model theory (PMT) and its relation to other main theories
The central claim of PMT is that we organize information about others by forming person models. We account for a multiplicity of epistemic access strategies, while direct perception and interaction are the main source for person schema formation. Person image formation is based on all the epistemic strategies we have examined, including theory-based inferences and (high-level) simulation strategies. Why, then, is PMT not a version of TT? Person models are more general and allow for a unification of rather parsimonious information about a person, which does not warrant being called a theory since it does not form even a minimal package of systematically-interconnected beliefs. As we learn more and more about the same person, our person model may develop into a theory. Thus, this is not to deny that we often have rich person models that are theories; and thus I can account for the empirical evidence that supporters of TT tend to rely on. A further question concerns how PMT is related to ST. Simulation is one epistemic strategy in which person models are used to understand others: if I have evidence that another person is similar to me in relevant respects, then I may use my self-model, either the self-schema or the self-image, to produce an explanation or a prediction of the other’s behaviour. But I also often have clear knowledge that the other is different from me in relevant respects, especially when there are great differences in the three main categories—sex, age, and race—or in cultural background. In such cases simulation is not used. Although simulation is a worthy epistemic strategy, it is only of limited and constrained use in everyday understanding. How is PMT related to interaction theory and direct perception theories? It explicitly accepts the important role of both as epistemic strategies, but insists that in addition to understanding others in situations of direct interaction there is also often an understanding of others just by observation. The use of these two strategies seems to depend heavily on the personality traits of the person who aims to understand another: while extroverts mainly rely on interaction, introverts (who avoid social contact) mainly rely on observation. Furthermore, these theories do not offer an answer to the main question addressed in this article, namely how we organize the information about other people that we already have. The narrative account offers one answer here, and again we can account for the role of narratives that in the case of rather rich person models may be sources for creating or enriching the models further, or they may also concern the way a person model is memorized. But the narrative account alone ignores the strong relevance of our intuitive understanding of others as it is anchored in person schemata. This short overview, then, indicates that all of the evidence that representatives of other theories put forward can be integrated into this view, while there is further evidence for my theory, e.g., rich evidence that there is an integration of information into person models by person perception. Notably, PMT allows us to account for certain mental disorders, and I have cited evidence from a very recent fMRI study that is further supportive of the organization of information according to person models.
5.6 Widening PMT: Person models, situation models and culture
Does PMT give us the complete story about understanding others? What about my understanding of a person whom I only see from behind, when queuing at a self-service restaurant? Here it seems sufficient to predict her behaviour just by expecting her to act according to the social conventions of a self-service restaurant. Understanding the situation alone seems to be sufficient for an understanding of and interaction with the other.[21] This is an important observation that suggests a widening of my theory: we do not only create person models, but also situation models, and our understanding of others uses both types of model as input and selects the most helpful model for evaluating the other person. If I have no person model of this individual, if seeing someone from the back gives me only very parsimonious information, and if I am only interested in getting my lunch, then the situation model may be dominant in dealing with persons in this context. As soon as minimal enrichment of person information is available we naturally tend to rely on person models. The fact that situation models are used at all is supported by successful artificial intelligence (AI) studies working with scripts and frames that can account for human behaviour (Schank & Abelson 1977). Furthermore, in Asian cultures the understanding of other people seems to rely much more on social conventions, since people are strongly expected to behave according to these conventions. In general, situation models are more important for understanding others in “collectivistic” cultures than in individualistic cultures where explanations and predictions of behaviour are usually more reliant on individual belief–desire explanations. Such observations as these require us to give an account of situation models. This can be easily done by widening the theory of understanding others such that it includes situation models, as well as the interdependence of personal models and situation models. It can also include a dynamic, involving bottom-up and top-down processes that lead to an activation or construction of the most plausible person model for interacting with, explaining, or predicting the behaviour of the other person.[22] Here is a rough outline of the process leading to understanding others in the rich sense of interacting, such as in observing, explaining, or predicting (see Figure 2).
Figure 2: Interaction of person models with situation models in understanding others
In general, we should note the important role of culture in shaping our way of modelling persons (Vogeley & Roepstorff 2009). As we have seen, culture modulates the relevance of person models in relation to situation models. But it also influences our formation of person models, for example by shaping our person perception. To illustrate: Japanese individuals are encouraged to be sociable and cooperative (Moskowitz et al. 1994), to be affiliative rather than competitive (Yamaguchi et al. 1995), and to show obligation to others (Oyserman et al. 1998). Concerning dominance and subordination, Japanese people learn to be rewarded for subordinate behaviour, while Americans learn to be rewarded for dominant behaviour. This also shapes the perception of dominance and subordination in others. Typical neurological activations of the mesolimbic reward system can be shown to be shaped by the respective culture: Americans show a higher activation of this system when doing and seeing dominant behaviour (in contrast to subordinate behaviour) while with Japanese people we can observe the opposite: they show a higher activation of exactly the same system when doing and seeing subordinate behaviour (Freeman et al. 2009). Thus, the perception of dominant and subordinate behaviour is connected with opposite evaluations (Americans highly esteem dominance while Japanese people highly esteem subordinate behaviour) and a different set of personality traits. Cultural influences on the psychological and neural level are also reported for self-models: on the psychological level, the difference between an Asian interdependent self and a Western independent self was reported by Markus & Kitayama (1991), while a respective difference in neural correlates was also recently discovered (Sui & Han 2007).