Discussion on ‘Interactionism in personality and social psychology: An integrated approach to understanding the mind and behaviour’ by Reynolds, Turner, Branscombe, Mavor, Bizumic, and Subašić


Interactionism Spoken Like a True Situationist


Department of Psychology, University of Texas at Austin, TX, USA


Reynolds et al. offer a version of interactionism based on social identity theory. Although we applaud both interactionism and the social identity approach, we suspect that the marriage the authors propose is unlikely to succeed. The core problem is that interactionism is optimized when the situation and person are on equal footing and the authors' model weds robust situational influences to a feckless, empty self. The result is a win for the social identity approach at the cost of what may have been an important new approach to interactionism. Copyright © 2010 John Wiley & Sons, Ltd.

Just when we thought that the nature-nurture (and trait-situation, etc.) debate was dead and buried, it has risen up yet again. This version offers an interesting twist: At first blush, this incarnation of the situationist perspective has the appearance of being balanced, nauanced and integrative. Indeed, as we made our way through Reynolds et al.'s thoughtful analysis, we found very few individual passages with which we disagreed. For example, the authors repeatedly stress the importance of the situation in general and group in particular; ‘Regardless of how “ready” a perceiver is to judge a situation in a particular way, however, there must be evidence in the situation that sustains such a categorization’ (p. 19), or ‘one's sense of self as an individual (‘who I am’) is forged through social comparison and can vary depending on the social comparative context’ (p. 25). It is hard to imagine anyone arguing against such perfectly sensible remarks.

Which brings us to one of our reservations with this paper: At times, it seems like they have targeted a straw man. For example, the authors provide a detailed account of Onorato and Turner's (2004) evidence that participants' self-ratings varied as a function of the context in which they were rating themselves (e.g. first as individuals and then as women compared to men). The fact that this manipulation produced between-condition differences would not surprise advocates of individual differences, for it is widely acknowledged that average group ratings will vary with the situation. What advocates of individual differences would predict, however, is that the rank ordering of self-ratings would remain relatively stable across diverse situations, producing a significant correlation. If the manipulation were strong enough to obliterate such correlations, then the notion that situations swamp individual differences would be supported. This would indeed be newsworthy but it was not what Onorato and Turner demonstrated.

Another curious feature of the paper is the authors' restrictive interpretation of the nature of interactionism. The authors contend that the relationship between personality and context is reciprocal, but they consider only one direction of this reciprocal process: ‘An aim also is to investigate using more naturalistic longitudinal designs the interplay between social identity processes and personality outcomes in a more reciprocal dynamic way. This work is focused on how personality is affected by (a) changes in the meaning of a particular group membership and its psychological significance to members (e.g. as a function of organizational change) and (b) one's experiences as a group member and processes of social influence amongst group members across time’ (p. 27). In our experience, reciprocal relationships are ones in which the flow of influence moves in both directions, not just one.

A related curiosity is the authors' seeming ambivalence when they discuss stability in self-knowledge. On page 30 they urge that ‘There needs to be a movement away from the view that there are 'pre-formed, already stored self-concepts (whose meaning is defined prior to their activation)’ towards an acceptance that the self is a ‘flexible, constructive process of judgment and meaningful inference in which varying self-categories are created to fit the perceiver's relationship to social reality (Turner et al. 1994: p. 458)’. Yet, on the very next page, ‘pre-existing knowledge’ seems to have become more potent and viable: ‘Turner et al. (2006, p. 256) argue that self-categorization “is not free to vary in any which way, but is always constrained by the motives, goals, values, experience, theories and knowledge the perceiver brings to the situation”’. We are confused here. Granted, ‘pre-formed, already formed self-concepts’ are not exactly the same thing as ‘the motives, goals, values, experience, theories and knowledge the perceiver brings to the situation’, but it seems safe to infer that in both instances the authors are referring to ‘pre-existing mental structures’. If so, then why do the authors wish to banish the very pre-existing mental structures that they acknowledge constrain people's self-categorizations?

But if the authors' approach is at times confusing and unconvincing, we should emphasize that our reservations are not with the social identity approach more generally (Tajfel & Turner, 1979). As one of us has noted elsewhere (Swann & Bosson, in press), this approach has been extraordinarily generative and has played a salutary role in illuminating the role of social identity in interpersonal relations. Nevertheless, the meta-theoretical model that the authors are advocating here is far too narrow and restrictive to offer a viable theory of interactionism. Indeed, the field has considered the ‘empty self’ approach in which chronic self-concepts are eschewed (e.g. Bem, 1972) and ultimately found it wanting.

At the end of the day, card-carrying interactionists readily acknowledge the impact of both context and chronically accessible self-views. Indeed, although it is essential to acknowledge the role of context, in those instances in which one wishes to understand individual differences in how people respond within a given context, measurement of some type of self-structure is the only game in town. Moreover, measurement of self-structures lends insight to organismic interactions as compared to the mechanistic interactions that the authors (quite appropriately) regard as limited. Specifically, there is evidence that people work to stabilize their chronic self-views by choosing and transforming their social settings in ways that confirm and stabilize their self-knowledge, including not only their personal self-views (Swann, in press), but their social identities (Chen, Chen, & Shaw, 2004; Gomez, Seyele, Huici, & Swann, 2009) and the confluence of their personal and social identities (Swann, Gomez, Dovidio, Hart, & Jetten, 2010; Swann, Gomez, Seyle, & Morales, 2009). With true interactionism finally gaining a foothold within psychology, it is time to broaden rather than narrow the bandwidth of the conceptual lenses through which we strive to understand behaviour.

Keeping the Person in Person-Situation Integration


1Michigan State University, MI, USA

2University of California, Davis, CA, USA

Reynolds et al. (2010) highlight the important influence of group identity processes on person-situation integration. We make several recommendations for improving their conceptualization of the person side of the integration. First, a clearer distinction should be made between personality and identity and between personality and responses to self-report questionnaires. Second, conclusions drawn from research showing transient contextual effects on self-reported personality should be circumscribed because it has no direct relevance to real personality change. Third, evidence that personality has a relatively stable biological core should not be dismissed but rather should serve as a building block for future theorizing about person-situation integration. Finally, greater theoretical precision is needed to clarify how the core constructs of the dynamic interactionist perspective should be operationalized in empirical research. Copyright © 2010 John Wiley & Sons, Ltd.


The alliance between personality and social psychology has always been an uneasy one. The two areas study many of the same phenomena—aggression, emotion, relationships, self-esteem—but they do so using different assumptions, methods and procedures (Tracy, Robins, & Sherman, 2009). For decades, interactionism has been repeatedly presented as the theoretical glue that will truly unite the two areas (Swann & Seyle, 2005) and virtually all contemporary social-personality researchers subscribe to some form of interactionism. Nonetheless, disagreements occur over the preferred ways that an interactionist approach should be implemented in empirical research (Lucas & Donnellan, 2009). In their recent article, Reynolds et al. (2010) revive one of the more conceptually complex variants of interactionism—dynamic interactionism—and inject it with novel theoretical ideas related to group identity processes. In this comment we reflect on one of the main virtues of their perspective and then identify a few limitations with how they conceptualize the person side of the person-situation integration.


Reynolds et al. highlight an important role for identity processes in person-situation integration and we found those aspects of their work to be particularly compelling (see also McAdams & Pals, 2006). However, Reynolds et al. seem to imply that personality should be equated with identity, which is antithetical to the way personality is conceptualized by most personality theorists. Likewise, they seem to equate personality with self-report responses on questionnaire measures. Although personality is often assessed by self-reports, the gold standard in personality research is to adopt a multi-method approach by also collecting ratings by knowledgeable informants, behavioural responses in standardized situations, physiological measures and data on life outcomes (Funder, 2010). This approach operationalizes traits and other personality constructs independently of a person's subjective experience. A theoretical framework aiming to integrate dispositional and situational influences should not assume that personality and identity are interchangeable constructs, just as personality should not be equated with any single method of assessment.

A metaphor might help to clarify the distinction between identity processes and ‘true’ personality. The social world is like a carnival house of mirrors with each interaction partner and each social group reflecting a partially veridical and partially distorted view of an individual's true personality. The individual has the same underlying dispositions regardless of which self is reflected in the mirror. Of course, spending most of one's life in the part of the carnival house that makes one feel incompetent or neurotic might lead to gradual changes in actual competence and anxiety levels. However, this kind of personality change occurs through very different processes than simply priming a salient identity. Indeed, research suggests that personality changes gradually across years (rather than moments) and in concert with often sustained changes in important life contexts and adult roles (Roberts, Woods, & Caspi, 2008).

Reynolds et al. discuss research showing that contextual factors produce shifts in self-reports of personality, and conclude that ‘personality is dynamic and shifts as a function of a person's salient social identity’ (p. 27). Such findings underscore the vagaries of the self-report method of personality assessment but they do not unambiguously demonstrate that personality itself changes. For example, a study showing large mean differences in self-reported intelligence when individuals rate their intelligence relative to either the average person or the average Nobel prize winner does not demonstrate that rating context influences intelligence. Unless one adopts a social constructionist view of personality, the same interpretational problem holds for research showing contextual effects on self-reported personality traits. To conclude that group identity processes can change personality, we would like to see evidence that these changes are apparent not only in self-reports but also in informant ratings, behavioural responses and physiological measures. We would also like to see that such changes are associated with real-world outcomes such as health, achievement and relationship functioning. Thus, the findings described by Reynolds et al. may provide important insights into the influence of contextual factors on identity processes, but their relevance to personality processes remains unclear. In the interests of furthering the integration of dispositional and situational approaches, this seems to be an important area for future research using multi-method studies.


Reynolds et al. apparently reject a cornerstone of much of contemporary personality psychology when it comes to acknowledging the biological core of dispositional traits. They suggest that accepting such a proposition presents ‘an obstacle to real engagement with [their] ideas’ (p. 34). In contrast, we believe that a complete account of person-situation interaction must be constrained by findings from behavioural genetics pointing to substantial genetic influences on human individuality and from findings from personality development pointing to an appreciable degree of rank-order stability in personality across years, and even decades of life. These ‘facts’ need not present any obstacles to integrating personality and social psychology but rather they should serve as the building blocks for future theorizing about person-situation integration.

Genetic factors contribute to individual differences in a wide range of personality characteristics, including not only core traits such as the Big Five dimensions but also self-esteem, narcissism, adult attachment styles, religiosity and other more complex aspects of personality such as the person X situation profiles that figure prominently in social-cognitive approaches (Borkenau, Riemann, Spinath, & Angleitner, 2006). Genetic factors contribute to the complex interplay between persons and situations that plays out across the life span as individuals react differently to the same objective events, select environments and continuously evoke responses from their environments, in part, because of their genetic predispositions.

Genetically based temperamental factors shape how individuals perceive and react to situations, yet their influence can be teased apart from the situation, as has been demonstrated in numerous studies. Accordingly, we see no need to conceptualize the person-situation unit as an ‘irreducible whole’. Genetic factors are of course relatively distal influences on behaviour that transmit their effects through more proximal affective, motivational, physiological and social-cognitive processes. Nonetheless, contemporary perspectives on person-situation integration must consider genes, gene-environment correlations, gene-environment interactions and intriguing new evidence for the ways that environments shape the expression of genes through epigenetic processes (e.g. Zhang & Meaney, 2010).


Reynolds et al. make a fairly strong claim about the necessity of conceptualizing persons and situations as an ‘irreducible “whole” that must be studied as one continuously interdependent unit’. (p. xx). In contrast, we believe it is often useful to distinguish between ‘person’ based and ‘situation’ based factors when attempting to understand behaviour. Such an approach has lead to a considerable accumulation of knowledge about the influence of dispositional and situational factors on a wide range of social behaviours. Although Reynolds et al. point to limitations of what they term ‘mechanical’ interactionism, this strategy has played an important role in guiding empirical research, as reflecting in recent studies and theories that examine how personality traits interact with situational manipulations (e.g. Marshall & Brown, 2006). It also seems to us that dynamic interactions will necessarily lead to mechanical interactions in most cases.

One of the chief virtues of the mechanical interactionist approach is that researchers know how to test these kinds of hypotheses using regression-based strategies. In contrast, researchers may not have a firm grasp as to how to test the core propositions implied by the dynamic interactionist perspective. In part, this reflects the lack of a formal theoretical model characterizing the relevant processes but it also reflects the difficulty of modelling dynamic processes using traditional statistical techniques. Finally, as Reis (2008) recently noted, all forms of interactionism—both dynamic and mechanical—are hampered by the absence of any consensual and systematic way of classifying situations, which would serve the same purpose as the Big Five taxonomy does for personality dispositions.


All theoretical models offer a simplified account of reality (Rodgers, 2010) and should be evaluated in terms of their usefulness for advancing psychological science. Important considerations include how easy it is to test a given model, how well the model clarifies previous findings, the accuracy of model-based predictions for future behaviour and the kinds of new questions generated by the model. We believe that the approach offered by Reynolds et al. will generate interesting new questions and we strongly support the idea that group identity processes play an important role in person-situation integration. Given that identity is shaped by individual dispositions, salient group memberships and intergroup relations, it would seem to be a natural point for integrating personality and social psychology. Our hope is simply that the person does not get lost in the integration.

Personality, Social Comparison and Self-categorization


1Laboratoire de Psychologie Sociale et Cognitive, Clermont Université, Université Blaise Bascal, France

2Department of Psychology, University of Geneva, Geneva, Swizerland


Reynolds and colleagues (this issue) make several valuable points in their discussion of dynamic interactionism as a basis for an integration of personality and social psychology. We agree that self-categorization theory provides some extremely important insights into the nature of individuality and personality. However, we also believe that some of the real problems that exist must be spelled out more clearly. To this end, we wish to point out a number of conceptual, methodological and theoretical gaps in existing theory and research. Copyright © 2010 John Wiley & Sons, Ltd.

Reynolds, Turner, Branscombe, Mavor, et al. (2010) stress the importance of dynamic interactionism as oppose to mechanical interactionism in order to develop a better understanding of the mind and behaviour. They point out how social identity and self-categorization theories can help define a more dynamic approach to this problem. There is much to agree in this analysis. We are afraid however that without confronting more directly the methodological and theoretical gaps that exist between personality psychology and social psychology, any effort to build bridges between the two is likely to fall short.


We all agree that we should study the ‘person-in-context’. However, very few researchers do so, probably because there is no agreed-upon method to study the person-in-context. As Zakriski, Wright, and Underwood (2005) observed, even when, in theory, it is argued that personality traits need to be considered as a function of the social context, the method used to study personality traits often reveals the tacit assumption of immutability. If dynamic interactionism is to serve as a conceptual model, there is a need to be more specific about how this conceptualization can be translated into testable hypotheses and about the specific nature of the methodology needed to test these hypotheses.


One promising methodological approach to study the person-in-context was introduced, perhaps inadvertently, by Heine, Lehman, Peng, and Greenholtz (2002). Using social comparison theory (see Festinger, 1954), they argued that people evaluate themselves by using some comparison standards. Their main point was that this raises some methodological problems when comparing mean scores on Likert-scale across cultures because people in a given culture are using different reference group standards than those in another culture. On this basis, they have challenged the validity of many studies comparing mean level of personality traits across cultures (see Heine & Buchtel, 1999; McCrae, Terracciano, Realo, & Allik, 2007). However, this emphasis of Heine and colleagues on methodological artefacts and confound misses an important point that emerges from a different conceptualization than the traditional one based on mechanical interactionism.

Consider first Heine et al.'s (2002) own findings. Research has lead to believe that there are some fundamental psychological differences between people in individualistic cultures compared to those in collectivistic cultures. Heine et al. (2002) asked Canadians (who had lived in Japan) and Japanese (who had lived in Canada) to describe themselves using 16-item from Singelis's (1994) Independent (individualist) versus interdependent (collectivist) self-scale. The basic hypothesis would be that Canadians are more independent and Japanese are more interdependent. However, Heine et al. (2002) found that this difference varied in significant ways when the reference-group standard was manipulated. The hypothesis was clearly supported when participants were asked to compare themselves with members of the other culture but not when participants were asked to compare themselves with members of their own culture. They concluded that ‘cultural comparisons using standard measures are confounded by the reference effect’ (Heine et al., 2002: p. 911).

This confound argument makes sense when people are assumed to function with a unique sense of self that is relatively stable and enduring, the personal self. From this perspective, Canadians are either more individualists than Japanese, or they are not. It is out of the question to consider that they may be more individualists in some context but not in others. However, as Reynolds et al. (2010) point out, from the perspective of Self-categorization Theory (SCT), a perspective not considered by Heine et al. (2002), people's sense of self varies in levels of abstraction. People can define themselves in terms of their personal identity (individual self or personality) but also in terms of their social identity (collective self). The collective self is no less valid than the individual self. It is simply a different way of defining the self that emerges under some specified conditions. The question that needs to be asked then is: Are cultural differences in individualism/collectivism occurring at the level of the individual self or at the level of the collective self? To be Canadian or Japanese is obviously to be a member of a particular cultural group and thus, it is when this group is used as a basis to define the self that significant differences between Canadians and Japanese would be expected. On the other hand, when Canadians and Japanese define themselves in terms of their personal self, SCT would predict little differences in individualism/collectivism. Thus, the findings of Heine et al. (2002) can be directly derived from SCT.


The fact that the method used by Heine et al. (2002) is a valid one to study the person-in-context can be demonstrated by considering our own research on gender differences in the self (see Guimond, Chatard, Martinot, Crisp, & Redersdorff, 2006). We showed when, and why, gender differences in the self on independent/agentic versus interdependent/communal dimensions emerge. As predicted by SCT, gender differences emerged in between-gender social comparison context, not in within-gender social comparison context, because the process of self-stereotyping occurred only in the former context. Indeed, in between-gender social comparison context, the ingroup stereotype mediated gender differences on both the independent/agentic/individualist dimension and the interdependent/communal/collectivist dimension. The evidence was clearly consistent with the fact that the personal self was involved in one condition and the collective self in the other.

A critical question is what happens after people describe themselves. If the reference-group effect is merely a methodological artefact, then little effect of a reference-group manipulation should be observed on the behaviour of the participants subsequent to the self-ratings. If Canadians are more individualists than Japanese even though this is not observed in their self-ratings because of a methodological confound (i.e. the reference-group effect), then Canadians should behave in more individualistic ways than Japanese after the self-ratings. Heine et al. (2002) do not provide any evidence that this is the case. In contrast, this question was examined in two experiments (see Guimond et al., 2006, Study 3 and Study 4). The reference-group standard was manipulated directly on the self-rating task, as in Heine et al. (2002). Subsequently, the behaviour of all participants was assessed, in the same way, regardless of experimental condition. If the reference-group effect is merely a confound, such that it produces shift in self-ratings that are purely artefactual, one would predict no significant differences across conditions in the subsequent behaviour of the participants. This is the prediction that can be derived from Heine et al. (2002). On the other hand, SCT suggests that changes in self-ratings will have an impact on subsequent behaviour (see Turner, Oakes, Haslam, & McGarty, 1994). This prediction was confirmed in both Study 3 and Study 4 (see Guimond et al., 2006). These results imply that the manipulation of reference-group standard is a valid method to study the person-in-context.


Personality is a core concept in psychology. It can be defined in many different ways. A traditional definition has been that personality refers to some enduring psychological dispositions that can explain behaviour. In their discussion of dynamic interactionism, Reynolds et al. suggest that personality is more relative and context dependent than previously considered. From our point of view, the existence of both stability and variability in the self and personality is beyond reasonable doubt (see e.g. Akrami, Ekehammar, Bergh, Dahlstrand, & Malmsten, 2009). Evidence for stability stems from research showing strong product-moment or rank-order correlations in personality variables (Terracciano, Costa, & McCrae, 2006; Perugini & Richetin, 2007). Evidence for variability stems from research showing that even a minimal change in instructions (varying the comparative frame of reference) can affect scores on standard personality measures (see Reynolds et al., 2010). Clearly, these two positions are complementary and not mutually exclusive. To illustrate, research indicates that there are very strong rank-order correlations in the Big Five dimensions of personality. Yet, there is also strong variability as a function of situations. Ramírez-Esparza, Gosling, Benet-Martínez, Potter, and Pennebaker (2006) demonstrated, for instance, that people exhibit different personalities when using different languages to fulfil the Big Five scale. According to us, such findings enlighten the importance of self-categorization processes in how the self is perceived at a given moment.

We agree with SCT that to better understand the self and personality, we need to take into account the different standards of comparisons that people use. However, we also believe that further research is needed to identify these different standards (see Buunk & Gibbons, 2006; Lorenzi-Cioldi & Chatard, 2006; Zagefka & Brown, 2006). Of course, sometimes these standards are quite explicit, because they are imposed by the situation. Thus, a woman may perceive herself as being very sociable in her professional context, because this particular context involves only comparison with men who are less sociable. However, we now know that people can also avoid some comparison standards, something that often compromises our ability to predict which comparison standards people spontaneously consider (see Guimond, 2006). For example, an issue that has attracted considerable attention on the part of social psychologists and for which the contribution of personality psychologists would appear essential, is the extent to which members of stigmatized social groups suffer from damage to their self-esteem and psychological integrity as a result of being discriminated against (e.g. Branscombe, Schmitt, & Harvey, 1999). Jones (2005) concludes his review of the evidence on this issue by noting: ‘An important general research direction is to explore the conditions under which intergroup versus interpersonal comparisons are made by stigmatized social groups. That is, when does one compare with mainstream standards or ingroup standards as a measure of self-worth’? (p. 168). This is a problem for which the integration of personality and social psychology may be well suited.


A concluding point that can perhaps summarize this entire discussion is the critical importance of distinguishing between different levels of analysis (Doise, 1986). In order to fruitfully integrate personality and social psychology, it is vital to keep in mind the different psychologies that exist at different levels of analysis so that we can proceed to study the central problem of articulating these different levels of analysis. From this point of view, to integrate personality and social psychology is essentially to articulate how different levels of analysis can be combined to achieve a better explanation of human behaviour (see Guimond, 2006). On the other hand, puzzlement can result from a failure to consider this matter. This can be illustrated by considering the study of Terracciano et al. (2005).

Using samples from 49 cultures, Terracciano et al. (2005) provided the most compelling test to date of the extent to which national stereotypes reflect the actual personality profiles of the members of a given nation. In each culture, measures of national stereotypes were obtained, revealing as anticipated that Australians for example are stereotyped as extraverted, German-Swiss as conscientious and Canadians as having an agreeable nature. However, the NEO-PI-R inventory assessing the Big five personality dimensions was translated into 27 different languages and also used in each of the 49 cultures. The findings revealed that for 45 of the 49 cultures, there was no reliable relation between the perception of the national character and the personality traits of the people constituting each nation. The authors concluded: ‘Although social scientists have long been sceptical about the accuracy of national stereotypes, the present study offers the best evidence to date that in-group perceptions of national character may be informative about the culture, but they are not descriptive of the people themselves’ (p. 99).

Another lesson should perhaps be derived from this unusual study. Social groups and culture, as Gestalt psychologists would say, are ‘wholes’ that cannot be reduced to their constituent parts. The social-psychological study of stereotypes involves a distinct level of analysis that cannot be reduced to the psychology of personality. The integration of different levels of analysis in psychology may not be easy but there is little doubt that it represents the way forward.


This research was supported by grant ANR-06-CONF-007.

The Requirement for a Non-individualistic Psychology of Individual Differences: Evidence from Studies of Tyranny and Oppression


1University of Exeter, UK

2University of St Andrews, Scotland


To flesh out the principles of dynamic interactionism presented by Reynolds et al. (2010), we discuss traditional approaches to the psychology of tyranny and oppression. We argue that models which place personality and situational factors in opposition, together with those which combine these elements mechanically, signally fail to capture the dynamism that is characteristic of tyrannical systems. We identify four aspects of this dynamism. These suggest that individual differences (a) draw people towards particular contexts, (b) are given meaning by salient social identities, (c) are transformed by intense group experiences and (d) become most potent when representative of group identity. Copyright © 2010 John Wiley & Sons, Ltd.

Reynolds et al.'s review and analysis of interactionism in personality and social psychology provides a powerful and timely re-examination of a range of prejudices that have defined contemporary understandings of the nature of individual differences and of their relationship to important forms of social behaviour. Possibly the most serious of these is the mistaken belief that personality and social psychology exist in opposition to each other and that the more is explained by the one, the less is left for the other.

The key contribution of Reynolds et al.'s paper, and of the various intellectual traditions within personality and social psychology upon which it draws, is to suggest not only that it is possible to move beyond this unsatisfactory state of affairs, but that the theoretical tools for advance are already available. In contrast to mechanical models which suggest that behaviour is simply a product of person and situation, their paper outlines a dynamic interactionist position which acknowledges both that people are formed in social relations and that the way people are formed shapes their social relations. There may thus be differences in the ways in which different individuals respond to any given situation, but those differences are in part a product of their different social experiences. To understand this process, we therefore need to advance a non-individualistic psychology of individual differences.

In this brief response, we shall seek to extend the discussion initiated by Reynolds et al. by considering how an interactionist approach can be used to shed light on the psychology of tyranny and oppression. Our choice of topic is deliberate, as previous responses to this question—framed initially by the motivation to understand the Nazi Holocaust—have come to exemplify the dualism between personality and social psychology.

In psychology, the immediate response to the Holocaust was to explain perpetrator behaviour as a manifestation of individual pathology—more specifically, as an expression of authoritarian personality (Adorno, Frenkel-Brunswik, Levinson, & Sanford, 1950). However, from the late 1950s onwards, a series of influential field studies by Sherif, Milgram and Zimbardo were used to emphasise instead the power of the situation—group contexts—in producing oppressive behaviours. These studies showed that ordinary schoolboys could be turned violent (Sherif & Sherif, 1956), that ordinary Americans could be induced to deliver lethal electric shocks (Milgram, 1963) and that ordinary students could be transformed into callous abusers (Haney, Banks, & Zimbardo, 1973). In accounts of this work, situation was invoked at the expense of personality or indeed of any attention to individual difference. The broader implication was to suggest that even the most ordinary people are capable of committing abhorrent acts—the prototypical case being drawn from a (mis)reading of Hannah Arendt's (1963) account of Adolf Eichmann as an embodiment of ‘the banality of evil’. Indeed, according to Zimbardo (2007; Haney et al., 1973), ‘the Lucifer effect’ is such that it is ‘natural’ for decent people to succumb to role requirements and thereby become agents of oppression.

Yet while the classic field studies do indeed demonstrate the power of context, if one looks closely they also reveal the importance of individual difference. In Zimbardo's Stanford Prison Experiment (SPE), for example, some of the Guards were brutal, some were ‘firm but fair’, and some wanted no part in any brutality. So how can we provide an integrated analysis of the contribution of both individual and situational factors rather than emphasizing one only to neglect the other? Based on the research of others and ourselves (notably, the BBC Prison Study; see Reicher & Haslam, 2006; www.bbcprisonstudy.org) we propose a number of pointers (see also Haslam & Reicher, 2007).

First, experimental evidence (Carnahan & McFarland, 2007) shows that a particular type of individual will be drawn to a particular type of situation: You have to have to be attracted in some way to tension and hardship in order to volunteer for a prison study. You have to be brutal to choose a brutal system. Thus, as Cesarani (2004) shows, Eichmann was no innocent but rather actively sought out and embraced Nazi Party membership.

Second, though, even if one chooses to enter a particular situation, not everyone will find the situation conducive to social identification. As we found in the BBC study, several of our Guards were highly uncomfortable with their group membership, especially where role requirements were at odds with other identities that they valued. More generally, behaviour in any given social situation will be different for individuals as a function of the identities they hold in other situations, especially where what they do in the one context is visible to those in others. In this sense, individual difference is not intra-psychic but a consequence of the fact that everyone's configuration of social identities is unique.

Third, once people are in a group and they identify with it, their personality can be transformed by the experience of group membership—especially where this experience is intense. Eichmann, like other Nazis, was radicalized through his membership of the Nazi Party. One of the most striking findings of the BBC study was that, in response to the dynamic exigencies of the system they were in, liberal participants became more authoritarian over time to the point that, by the end, they hardly resisted the imposition of an authoritarian regime.

Fourth, however, people do not simply respond to group experiences, they also initiate them. Thus, in the BBC study the idea of an authoritarian regime did not emerge out of the blue. It was proposed by particular individuals, and, from the start, these were individuals who were more authoritarian than the rest. Yet at the start, they were not influential. It was only later, as others moved towards them and they became more prototypical of the group as a whole that they were in a position to exert influence (Turner & Haslam, 2001). To understand the impact of individuals with particular orientations (in this case authoritarianism) upon a social system it is therefore essential to understand the group-level dynamics that transform individual differences into collective leadership (Haslam, Reicher, & Platow, 2010).

We do not suggest that these dynamics are exhaustive, either in explaining tyranny and oppression or in understanding the interplay between individual and group-level dynamics. What we do suggest however, is that unless one adopts a dynamic interactionist approach, the task of advancing understanding is doomed from the outset. The same, we believe, is true when it comes to explaining more mundane forms of human behaviour. And for that reason, we welcome Reynolds et al.'s incisive contribution to the debate.

Interactionism in Personality and Social Psychology: A Whole That is Less Than the Sum of its Parts


Department of Psychology, University of Illinois, Urbana-Champaign, IL, USA


Reynolds et al. encourage integrating personality and social psychology perspectives on interactionism, to develop a Gestalt across the fields. While we concur with their stated aims, the authors seem to select only the ‘parts’ that fit their ‘whole’, and in turn neglect well-established findings in the personality literature. Indeed, there is a wealth of evidence for personality consistency, and that personality change is gradual rather than frequent. We welcome future collaborative efforts to solve the interactionism puzzle, as long as they use all the pieces in the box. Copyright © 2010 John Wiley & Sons, Ltd.


Reynolds et al. attempt to persuade researchers to move beyond the usual superficial agreement that both person factors and situational factors matter, to a more integrated approach where both the person and situation are considered together. We agree (see Roberts, 2009; Roberts & Jackson, 2008). However, it appears that there are many aspects of personality psychology that the authors are reluctant to incorporate to make this a truly successful integration. For example, statements such as ‘What is being proposed is a non-reductionist model of mind and behaviour where the causes of behaviour are not located within an inner essence of the person (temperament, biology, limitations of the cognitive system: Turner & Oakes, 1997: p. 35)’; and ‘An obstacle to real engagement with these ideas, though, is the view that there is a stable “essence” or “core” within the person that is biologically determined (e.g. genes, temperament)’ (p. 34); and, ‘It is difficult to reconcile these levels of self-categorization and the concept of personality defined as the stable characteristics of the (individual) person’. (p. 17) run counter to a number of well-validated findings in personality psychology and suggests a lack of desire for a true Gestalt between different perspectives.

A truly integrative model of psychology that merges personality and social psychology to better understand the mind and behaviour would reconcile self-categorization with stable individual differences that are biologically based and influenced by temperament. Why? Because the empirical facts are irrefutable. Personality traits are quite consistent across the lifespan with long-term test-retest correlations typically exceeding .5 (Roberts & DelVecchio, 2000). Even the personality traits of children show consistency when assessed 40 years later (Hampson & Goldberg, 2006). Furthermore, personality traits are also heritable, and associated with specific regions of the brain and genetic polymorphisms (Canli, 2008). Finally, one of the most robust findings in longitudinal behavioural genetics research is the fact that the majority of variance in personality trait continuity is genetic in origin (Krueger, Johnson, & Kling, 2006). These findings do not suggest that personality traits are entirely genetically determined or impervious to environmental influence. Rather these findings make it undeniable that there exists an ‘inner essence or core of a person’. Moreover, it is undeniable that this ‘inner essence’ or ‘core’ is, at least partially, due to biology and genetics.


The authors provide no compellingly new findings to convince personality psychologists to abandon conceptions of an ‘inner essence’ or ‘core’ and accept a fully situated, contextualized worldview. They suggest that it is difficult to reconcile self-concepts/categorizations/behaviours that can change across situations and the stable characteristics of a person. We disagree. The idea that personality traits result in monolithically unchanging behaviour across situations misrepresents the true definitions of personality traits (Roberts, 2009). As has been pointed out numerous times, the fact that situations or social factors, such as group memberships, roles and experimental manipulations, shape behaviour is orthogonal to the existence of stable individual differences across situations and time (Funder & Colvin, 1991; Roberts & Pomerantz, 2004; Roberts & Jackson, 2008). Therefore, pointing out that behaviour is variable across situations does nothing to provide evidence for the superiority of a purely contextualized position. Temporal consistency, not cross-situational consistency, is the key to the existence of personality traits, and there is little argument over its existence. Additionally, the levels of cross-situational consistency in behaviour are entirely consistent with the definitional basis of personality traits and states (Borkenau, Mauer, Riemann, Spinath, & Angleitner, 2004; Fleeson, 2001).

The authors highlight the proposed ‘dynamic’ nature of personality by shifting self-rating of personality (p. 27) and in doing so conclude that personality is highly malleable. We agree that self-perceptions can be shifted from situation to situation but take a more sober approach for ascribing true, lasting changes in personality. The evidentiary base demonstrates quite strongly that change in personality traits is slow and gradual, not dramatic (Roberts, Caspi, & Moffitt, 2001; Robins et al., 2002). Changes in personality traits are therefore likely due to enduring environmental factors rather than quixotic shifts in molecular situations (Roberts & Jackson, 2008). For example, going to college, starting a new career and getting married do not dramatically shape personality traits.

The reason for the modest effects of life experience on personality traits can be seen in our sociogenomic model of personality that the authors briefly review (Roberts & Jackson, 2008). Personality traits act as a brake on variability, providing a homeostatic function to the personality system. Environments and life experiences will impinge themselves on states directly and traits indirectly. Minor shifts in states will leave no lasting effect on the homeostatic function of traits. Rather, only long-term changes in states will lead to changes in personality traits. This discrepancy in what is seen as ‘dynamic’ boils down to how much change or continuity exists in a person across situation and over time. We feel that across situations there can be large amounts of change in behaviour that hold little or no consequence for the existence of personality traits. However, over time the effects of environments are so modest that it questions whether the term ‘dynamic’ should be applied.


We do genuinely agree with the authors that personality psychologists could do a better job of integrating group-level phenomena. Including ideas about self-perceptions linked to large social groups (e.g. female, New Zealander, etc.) could provide keen insights into the interface between the individual and society. That being said, we are not sure whether it has not already been accomplished. There are theoretical systems, such as the neo-socio-analytical model of personality development (Roberts & Wood, 2006), that explicitly incorporate culture, society and social roles into the conceptual framework. Moreover, a long line of research in personality psychology has investigated the multiplicity of the self and identity structure around role identities and has integrated these perspectives with classic trait models (Block, 1961; Roberts & Donahue, 1994; Stryker, 2007; Wood & Roberts, 2006). At first blush it appears that many, if not all of the ideas proposed from a social identity structure have been anticipated by this work with the caveat that we could always do a better job. Nonetheless, improving existing theoretical structures is less of a substantive contribution than a purported new integration.

From our perspective, the next step in integrating the social world in models of personality is to identify crucial environments responsible for long-term changes in personality. Currently we do not know what environments are causally responsible for changes in personality traits. It should be noted that many measures of the environment are also heritable, most likely due to gene-environment correlations. This further ties the environment closer to the person, as people are likely to select and evoke specific environments that are consistent with their personality, and highlights the theme of the paper that the person and the environment are best considered together.


To conclude, we laud the authors' intentions. Creating a working model of human nature that can move seamlessly between personality traits and more contextualized aspects of human functioning would indeed be a constructive achievement. Unfortunately, we do not feel the approach outlined in Reynolds et al. (2010) achieves the desired Gestalt between person and situation necessary to accomplish this goal. A successful integration has to integrate all components of each perspective rather than picking and choosing among the offerings deemed to be more desirable.

On the Diversity of Dynamic Person × Situation Interactions


Department of Psychology, University of Koblenz-Landau, Germany


Five propositions are offered in reaction to the target paper on interactionism in personality and social psychology. (1) The possible meanings of ‘dynamic interaction’ need to be considered carefully. (2) Dynamic person × situation interactions that generate behaviour are different from interactions that cause personality change. (3) Dynamic person × situation interactions cannot be generalized across different domains of the human personality such as attitudes and beliefs, the self-concept, temperament and ability. (4) The concept of identity and the concept of personality need to be distinguished. (5) More attention needs to be paid to the causal mechanisms that generate interactions, and these mechanisms need to be included in the models that are being tested empirically. Copyright © 2010 John Wiley & Sons, Ltd.

Reynolds, Turner, Branscombe, Mavor, Bizumic, and Subasic (2010) have made an interesting proposal to link the classic state-trait model of personality with social identity theory (SIT) and self-categorization theory (SCT). More specifically, they propose that the frameworks be combined into a dynamic person × context interaction model of behaviour and personality development. The target paper goes far beyond the many instances of lip service that have been paid to interactionism since Cronbach's (1957, 1975) plea to bridge the gap between the two disciplines of psychology: General psychology and differential psychology. If Lee Cronbach was still alive, he certainly would have agreed with the authors' proposal that dynamic person × situation interaction models of behaviour and personality development are far more convincing than are models that explain behaviour solely with static personality traits or solely with situation factors. In fact, it seems difficult to disagree with the basic premise of interactionism that behaviour can only be understood fully if interactions between the person and the situation are taken into account. It also seems easy to agree with the authors' proposition that the joint impact of personality factors and situation factors is dynamic in nature.

Although we agree with the basic idea of the paper, we would like to address a few issues that we believe need close attention, not only from the authors of the target article, but from the community of personality and social psychologists. First, we think that we need to carefully consider the possible meanings of ‘dynamic interaction’. Second, we need to investigate theoretically and empirically whether the interactions that generate behaviour are identical to or different from the interactions that contribute to personality change. Third, we need to rigorously test whether interactions and the mechanisms that generate them can be generalized across different domains of the human personality such as attitudes, values and beliefs versus temperament and ability traits. Fourth, we believe we need to differentiate between the concepts of identity and personality. Fifth, we need to pay more attention to the mechanisms that generate interactions. These mechanisms have to be included in the theoretical models that we test empirically.

(1) An interaction is dynamic in nature only if it reflects a process over time. The authors argue that including the psychological situation versus the objective situation in a person × situation model makes this model a dynamic model. We do not agree completely with this position. From our point of view, an interaction is dynamic only if it reflects the result of a process. Thus, we suggest reserving the term ‘dynamic’ for models that describe and explain how personality affects situations (via selection, interpretation, modification) and how psychological states that result from the (self-)exposure to situations and contexts feed back into personality and lead to personality change on the trait level.

(2) Dynamic interactions that generate behaviour are different from dynamic interactions that contribute to personality change. Dynamic processes can be short termed or long termed, they can result from diverse psychological mechanisms, and they can have diverse results. Therefore, every dynamic interaction has to be specified according to the variables involved, the time dimension and the mechanisms that produce a specific effect. Let us give two examples, one for a short-termed dynamic interaction and another one for a long-termed interaction. (a) The psychological situation is the result of a short-termed interaction between personality and the objective situation. Driven by their personality, individuals attend to certain features of a situation and interpret these features in a specific manner. The exact mechanisms of this meaning-giving process are not yet fully understood. However, a few promising candidates have been identified such as (i) the chronic accessibility of concepts or schemata and (ii) the complexity of the person's conceptual network in the domain that is relevant for attending to, interpreting and encoding information contained in a situation (e.g., Baumert & Schmitt, 2009; Bruner, 1957; Higgins & King, 1981). Giving meaning to a situation creates a psychological state that results in emotional and behavioural reactions in a person. Such a short-termed interaction will usually not result in personality change. (b) Personality change results from other mechanisms. Reynolds et al. (2010) mentioned several ideas and research findings from life-span developmental psychology. Of those, we find the sociogenomic model of Roberts and Jackson (2008) most promising. In a nutshell, this model assumes that long-term personality change on the trait level reflects state density distributions across time (Fleeson, 2001). If people are exposed to or expose themselves repeatedly to situations that produce a specific psychological state, this may eventually result in a personality shift on the trait level (Mathews & MacLeod, 2002; Schmidt, Richey, Buckner, & Timpano, 2009).

(3) The mechanisms that drive dynamic interactions differ between personality domains. Given their background and interest, Reynolds et al. (2010) used examples from the domains of values, attitudes, beliefs, identity and the self-concept. Many studies have shown that the person's salient social and personal identity affects his or her behaviour. Moreover, group membership and chronic social identity have been shown to affect values, attitudes, and beliefs, and, consequently, behaviour corresponding to these dispositions. However, values, attitudes, beliefs and identity are acquired dispositions (Campbell, 1963), and thus more plastic than temperament traits or ability traits (Caspi, Roberts, & Shiner, 2005). We doubt that the mechanisms that have been identified in SIT and SCT research can be generalized to the long-term development of core personality traits, such as extraversion, and ability traits, such as intelligence (see next comment).

(4) Identity and the self-concept must be distinguished from personality traits and ability. Reynolds et al. (2010) presented important research findings showing that a person's identity and self-concept as measured with conventional personality inventories can change depending on the context. For instance, the person's self-concept of intelligence will change depending on the intelligence of people who are available for social comparison (peers, friends, colleagues). It seems highly unlikely, however, that social comparisons will change the person's intelligence as much as they change his or her self-concept of intelligence. Research on the Big Fish Little Pond Effect (BFLP) has shown that social comparisons indeed can result in true achievement differences (Marsh, Hau, & Craven, 2004). However, the BFLP has a much stronger effect on the self-concept of ability than on ability itself. Moreover, the mechanisms that cause self-concept change are not the same ones that cause changes in achievement.

(5) More theoretical attention needs to be paid to the mechanisms that generate dynamic interactions, and these mechanisms need to be included in the models that we test empirically. According to our knowledge of the literature, the mechanisms that are held responsible for dynamic interactions are rarely included in empirical model tests. Let us take anxiety research as an example. We know from many studies that trait anxiety and situational threat jointly cause state anxiety and behavioural reactions such that differences in situational threat have a larger impact on persons high in trait anxiety than on persons low in trait anxiety (e.g., Endler, 1997). We also know that trait-anxious individuals selectively attend to threatening information and tend to selectively interpret ambiguous information as evidence of danger (e.g., MacLeod, Mathews, & Tata, 1986; Richards & French, 1992). Moreover, it is assumed that cognitive bias explains the synergistic interaction between trait anxiety and situational threat on state anxiety. However, direct tests of this moderated mediation model are rare at best.

The shortage of studies that (a) specify the mechanisms of dynamic interactions as moderated mediation processes, (b) measure these mechanisms and (c) test the entire model in one study is not a problem only in anxiety research. Rather, it is a general limitation of personality and social psychological research. It is this type of research that we need in order to understand how and why personality and the physical and social environment generate psychological states and behaviour, and shape each other reciprocally over time.