How stereotypes shape our perceptions of other minds

 by Evan Westra — Ph.D. Candidate, University of MarylandAmbiguous Pictures Task

McGlothlin & Killen (2006) showed groups of (pre­dom­in­antly white) American ele­ment­ary school chil­dren from ages 6 to 10 a series of vign­ettes depict­ing chil­dren in ambigu­ous situ­ations. For instance, one pic­ture (above) showed two chil­dren by a swing set, with one on the ground frown­ing, and one behind the swing with a neut­ral expres­sion. Two things might be going on in this pic­ture: i) the child on the ground may have fallen off by acci­dent (neut­ral scen­ario), or ii) the child on the ground may have been inten­tion­ally pushed by the one stand­ing behind the swing (harm­ful scen­ario). Crucially, McGlothlin and Killen var­ied the race of the chil­dren depic­ted in the image, such that some chil­dren saw a white child stand­ing behind the swing (left), and some saw a black child (right). Children were asked to explain what had just happened in the scen­ario, to pre­dict what would hap­pen next, and to eval­u­ate the action that had just happened. Overwhelmingly, chil­dren were more likely to give the harm­ful scen­ario inter­pret­a­tion — that the child behind the swing inten­tion­ally pushed the oth­er child — when the child behind the swing was black than when she was white. The race the child depic­ted, it seems, influ­enced wheth­er or not par­ti­cipants made an infer­ence to harm­ful inten­tions.

This is yet anoth­er depress­ing example of how racial bias can warp our per­cep­tions of oth­ers. But this study (and oth­ers like it: Sagar & Schofield 1990; Burnham & Harris 1992; Condry et al. 1985)) also hint at a rela­tion­ship between two forms of social cog­ni­tion that are not often stud­ied togeth­er: mindread­ing and ste­reo­typ­ing. The ste­reo­typ­ing com­pon­ent is clear enough. The mindread­ing com­pon­ent comes from the fact that race did­n’t just affect kids’ atti­tudes towards the tar­get — it affected what they thought was going on in the tar­get’s mind. Although these two ways of think­ing about oth­er people — mindread­ing and ste­reo­typ­ing — both seem to play an import­ant role in how we nav­ig­ate the social world, curi­ously little atten­tion has been paid to under­stand­ing the way they relate to one anoth­er. In this post, I want to explore this rela­tion­ship. I’ll first briefly explain what I mean by “mindread­ing” and “ste­reo­typ­ing.” Next, I’ll dis­cuss one exist­ing pro­pos­al about the rela­tion­ship between mindread­ing and ste­reo­typ­ing, and raise some prob­lems for it. Then I will lay out the begin­nings of a dif­fer­ent way of cash­ing out this rela­tion­ship.

*          *          *

 First, lets get clear on what I mean by “mindread­ing” and “ste­reo­typ­ing.”


In order to achieve our goals in highly social envir­on­ments, we need to be able to accur­ately pre­dict what oth­er people will do, and how they will react to us. To do this, our brains gen­er­ate com­plex mod­els of oth­er people’s beliefs, desires, and inten­tions, which we use to pre­dict and inter­pret their beha­vi­or. This capa­city to rep­res­ent oth­er minds is known vari­ous as the­ory of mind, mindread­ing, men­tal­iz­ing, and folk psy­cho­logy. In human beings, this abil­ity begins to emerge very early in devel­op­ment. As adults, we use it con­stantly, in a fast, flex­ible, and uncon­scious fash­ion. We use it in many import­ant social activ­it­ies, includ­ing com­mu­nic­a­tion, social coordin­a­tion, and mor­al judg­ment.


Stereotypes are ways of stor­ing gen­er­ic inform­a­tion about social groups (includ­ing races, genders, sexu­al ori­ent­a­tion, age-groups, nation­al­it­ies, pro­fes­sions, polit­ic­al affil­i­ation, phys­ic­al or men­tal abil­ity, and so on) (Amodio 2014). A par­tic­u­larly import­ant aspect of ste­reo­types is that they often con­tain inform­a­tion about stable per­son­al­ity traits. Unfortunately, it is all too easy for us to think of ste­reo­types about how cer­tain social groups are lazy, or greedy, or aggress­ive, or sub­missive, and so on. According to Susan Fiske and col­leagues’ Stereotype Content Model (SCM), there is an under­ly­ing pat­tern to the way we attrib­ute per­son­al­ity traits to groups (Cuddy et al. 2009; Cuddy et al. 2007; Fiske et al. 2002; Fiske 2015). Personality trait attri­bu­tion, on this view, var­ies along two primary dimen­sions: warmth and com­pet­ence. The warmth dimen­sion includes traits like (dis-)honesty, (un-)trustworthiness, and (un-)friendliness. These are traits that tell you wheth­er or not someone is liable to help you or harm you. The com­pet­ence dimen­sion con­tains traits like (un-)intelligence, skill­ful­ness, per­sist­ence, lazi­ness, clum­si­ness, etc. These traits tell you how effect­ively someone is at achiev­ing their goals.

Together, these two dimen­sions com­bine to yield four dis­tinct clusters of traits, each of which picks out a dif­fer­ent kind of ste­reo­type:

the stereotype content model

*          *          *

So what do ste­reo­typ­ing and mindread­ing have to do with one anoth­er? There are some obvi­ous dif­fer­ences, of course: ste­reo­types are mainly about groups, while mindread­ing is mainly about indi­vidu­als. But intu­it­ively, it seems like know­ing about some­body’s social group mem­ber­ship could tell you a lot about what they think: if I tell you that I am a lib­er­al, for instance, that should tell you a lot about my beliefs, val­ues, and social pref­er­ences — valu­able inform­a­tion, when it comes to pre­dict­ing and inter­pret­ing my beha­vi­or.

Some philo­soph­ers and psy­cho­lo­gists, such as Kristin Andrews, Anika Fiebich and Mark Coltheart, have sug­ges­ted that ste­reo­types and mindread­ing may actu­ally be altern­at­ive strategies for pre­dict­ing and inter­pret­ing beha­vi­or (Andrews 2012; Fiebich & Coltheart 2015). That is, it may be that some­times we use ste­reo­types instead of mindread­ing to fig­ure out what a per­son is going to do. According to one such pro­pos­al (Fiebich & Coltheart 2015), ste­reo­types allow us to pre­dict beha­vi­or because they encode asso­ci­ations between social cat­egor­ies, situ­ations, and beha­vi­ors. Thus, one might form a three-way asso­ci­ation between the social cat­egory police, the situ­ation donut shops, and the beha­vi­or eat­ing donuts, which would lead one to pre­dict that, when one sees a police officer in a donut shop, he or she will likely be eat­ing a donut. A more com­plex ver­sion of this asso­ci­ation­ist approach would be to asso­ci­ate social groups with par­tic­u­lar traits labels (as per the SCM), and thus con­sist in four-way asso­ci­ations between social cateo­gires, trait labels, situ­ations, and beha­vi­ors (Fiebich & Coltheart 2015; Andrews 2012). Thus, one might come to the trait of gen­er­os­ity with leav­ing large tips in res­taur­ants, and asso­ci­ate the social cat­egory of uncles with gen­er­os­ity, and thereby come to expect uncles to leave large tips in res­taur­ants. One might then explain this beha­vi­or by refer­ring to the uncle’s gen­er­os­ity. The key thing to notice about these accounts is that their pre­dic­tions do not rely at all upon mental-state attri­bu­tions. This is by design: these pro­pos­als are meant to show that we often don’t need mindread­ing to pre­dict or inter­pret beha­vi­or.

One prob­lem for this sort of view comes from its invoc­a­tion of “situ­ations.” What inform­a­tion, one might won­der, is con­tained with­in the scope of a par­tic­u­lar “situ­ation”? Surely, a situ­ation does not include everything about the state of the world at a giv­en moment. Situations are prob­ably meant to pick out loc­al states of affairs. But not all the facts about a loc­al state of affairs will be rel­ev­ant to beha­vi­or pre­dic­tion. The pres­ence of mice in the kit­chen of a res­taur­ant, for instance will not affect your pre­dic­tions about the size of your uncle’s tip. It might, how­ever, affect our pre­dic­tions about the beha­vi­or of the health inspect­or, should one sud­denly arrive. Which loc­al facts are pre­dict­ively use­ful will ulti­mately depend upon their rel­ev­ance to the agent whose beha­vi­or we are pre­dict­ing. But wheth­er or not a fact is rel­ev­ant to an agent will depend upon that agent’s beliefs about the loc­al state of affairs, as well as her goals and desires. If this is how rep­res­ent­a­tions of pre­dict­ively use­ful situ­ations are com­puted, then the pur­portedly non-mentalistic pro­pos­al giv­en above really includes a tacit appeal to mindread­ing. If this is not how situ­ations are com­puted, then we are owed an explan­a­tion for how the non-mentalistic behavior-predictor arrives at pre­dict­ively use­ful rep­res­ent­a­tions of situ­ations that do not depend upon con­sid­er­a­tions of rel­ev­ance.

*          *          *

Instead of treat­ing mindread­ing and ste­reo­types as sep­ar­ate forms of behavior-prediction and inter­pret­a­tion, we might instead explore the ways in which ste­reo­types might inform mindread­ing. The key to this approach, I sug­gest, lies in the fact that ste­reo­types encode inform­a­tion about per­son­al­ity traits. In many ways, per­son­al­ity traits are like men­tal states: they are unob­serv­able men­tal prop­er­ties of indi­vidu­al, and they are caus­ally related to beha­vi­or. But they also dif­fer in one key respect: their tem­por­al sta­bil­ity. Beliefs and desires are inher­ently unstable: a belief that P can be changed by the obser­va­tion of not‑P; a desire for Q can be extin­guished by the attain­ment of Q. Personality traits, in con­trast, can­not be extin­guished or aban­doned based on every­day events. Rather, they tend to last through­out a per­son’s life­time, and mani­fest them­selves in many dif­fer­ent ways across many dif­fer­ent situ­ations. A unique fea­ture of per­son­al­ity traits, in oth­er words, is that they are highly stable men­tal entit­ies (Doris 2002). So when ste­reo­types ascribe traits to groups, they are ascrib­ing a prop­erty that one could reas­on­ably expect to remain con­sist­ent across many dif­fer­ent situ­ations.

The tem­por­al prop­er­ties of men­tal states are extremely rel­ev­ant for mindread­ing, espe­cially in mod­els that employ Bayesian Predictive Coding (Kilner & Frith 2007; Koster-Hale & Saxe 2013; Hohwy & Palmer 2014; Hohwy 2013; Clark 2015). To see why, let’s start with an example:

Suppose that we believe that Laura is thirsty, and have attrib­uted to her the goal of get­ting a drink (G). As goals go, this one is rel­at­ively short-term (unlike, say, the goal of get­ting a PhD). But we know that in order to achieve (G), we pre­dict that Laura must form a num­ber of even shorter-term sub-goals: (G1) get the juice from the fridge, and (G2) pour her­self a glass of juice. But each of these requires the form­a­tion of still shorter-term sub-sub-goals: (G1a) walk over to kit­chen, (G1b) open fridge door, (G1c) remove juice con­tain­er, (G2a) remove cup from cup­board, (G2b) pour juice into cup. Predicting Laura’s beha­vi­or in this con­text thus begins with the ascrip­tion of a longer-duration men­tal state (G), fol­lowed by the ascrip­tion of suc­cess­ively shorter-term mental-state attri­bu­tions (G1, G2, G1a, G1b, G1c, G2a, G2b).

As mindread­ers, we can use attri­bu­tions of more abstract, tem­por­ally exten­ded men­tal states to make top-down infer­ences about more tran­si­ent men­tal states. At each level in this action-prediction hier­archy, we use higher-level goal-attributions to con­strain the space of pos­sible sub-goals that the agent might form. We then use our pri­or exper­i­ence to select the most likely sub-goal from the hypo­thes­is space, and the pro­cess repeats itself. Ultimately, this yields fairly fine-grained expect­a­tions about motor-intentions, which mani­fest them­selves as mirror-neuron activ­ity (Kilner & Frith 2007; Csibra 2008). Action-prediction thus plays out as a des­cent from more stable mental-state attri­bu­tions to more tran­si­ent ones, which ulti­mately bot­tom out in highly con­crete expect­a­tions about beha­vi­or.

Personality traits, which are dis­tin­guished by their high degree of tem­por­al sta­bil­ity, fit nat­ur­ally into the upper levels of this action-prediction hier­archy. Warmth traits, for instance, can tell us about the gen­er­al pref­er­ences of an agent: a gen­er­ous per­son prob­ably has a gen­er­al pref­er­ence for help­ing oth­ers, while a greedy per­son prob­ably has a gen­er­al desire to enrich her­self. These broad preference-attributions can in turn inform more imme­di­ate goal-attributions, which can then be used to pre­dict beha­vi­or.

This role for rep­res­ent­a­tions of per­son­al­ity traits in mental-state infer­ence fits well with what we know about how we reas­on about traits more gen­er­ally. For instance, we often make extremely rap­id judg­ments about the warmth and com­pet­ence traits of indi­vidu­als based on fairly super­fi­cial evid­ence, such as facial fea­tures (Todorov et al. 2008); we also tend to over attrib­ute the causes of beha­vi­or to per­son­al­ity traits, rather than situ­ation­al factors — a phe­nomenom com­monly known as the “fun­da­ment­al attri­bu­tion error” or the “cor­res­pond­ence bias (Gawronski 2004; Ross 1977; Gilbert et al. 1995). Prioritizing per­son­al­ity traits makes a lot of sense if they form the infer­en­tial basis for more com­plex forms of beha­vi­or pre­dic­tion. It also makes sense that this aspect of mindread­ing would need to rely on fast, rough-and-ready heur­ist­ics, since per­son­al­ity trait inform­a­tion would need to be inferred very quickly in order to be use­ful in action-planning.

From a com­pu­ta­tion­al per­spect­ive, thus, using per­son­al­ity traits to make infer­ences about beha­vi­or makes a lot of sense, and might make mindread­ing more effi­cient. But in exchange for this effi­ciency, we make a very dis­turb­ing trade. Stereotypes, which can be activ­ated rap­idly based on eas­ily avail­able per­cep­tu­al cues provide the mindread­ing sys­tem with a rap­id means for stor­ing trait inform­a­tion (Mason et al. 2006; Macrae et al. 1994). With this speed comes one of the most mor­ally per­ni­cious forms of human social cog­ni­tion, one that helps to per­petu­ate dis­crim­in­a­tion and social inequal­ity.

*          *          *

 The pic­ture I’ve painted in this post is, admit­tedly, rather pess­im­ist­ic. But just because the roots of dis­crim­in­a­tion are cog­nit­ively deep, we should not con­clude that it is inev­it­able. More recent work from McGlothlin and Killen (2010) should give us some hope: while chil­dren from racially homo­gen­eous schools (who had little dir­ect con­tact with mem­bers of oth­er races) ten­ded to show signs of biased intention-attribution, McGlothlin and Killen also found that chil­dren from racially het­ero­gen­eous schools (who had reg­u­lar con­tact with mem­bers of oth­er races) did not dis­play such signs of bias. Evidently, inter­group con­tact is effect­ive in curb­ing the devel­op­ment of ste­reo­types — and, by exten­sion, biased mindread­ing.



Amodio, D.M., 2014. The neur­os­cience of pre­ju­dice and ste­reo­typ­ing. Nature Reviews: Neuroscience, 15(10), pp.670–682.

Andrews, K., 2012. Do apes read minds?: Toward a new folk psy­cho­logy, Cambridge, MA: MIT Press.

Burnham, D.K. & Harris, M.B., 1992. Effects of Real Gender and Labeled Gender on Adults’ Perceptions of Infants. Journal of Genetic Psychology, 15(2), pp.165–183.

Clark, A., 2015. Surfing uncer­tainty: Prediction, action, and the embod­ied mind, Oxford: Oxford University Press.

Condry, J.C. et al., 1985. Sex and Aggression : The Influence of Gender Label on the Perception of Aggression in Children Development Sex and Aggression : The Influence of Gender Label on the Perception of Aggression in Children. Child Development, 56(1), pp.225–233.

Csibra, G., 2008. Action mir­ror­ing and action under­stand­ing: an altern­at­ive account. In P. Haggard, Y. Rossetti, & M. Kawato, eds. Sensorymotor Foundations of Higher Cognition. Attention and Performance XXII. Oxford: Oxford University Press, pp. 435–459.

Cuddy, A.J.C. et al., 2009. Stereotype con­tent mod­el across cul­tures: Towards uni­ver­sal sim­il­ar­it­ies and some dif­fer­ences. British Journal of Social Psychology, 48(1), pp.1–33.

Cuddy, A.J.C., Fiske, S.T. & Glick, P., 2007. The BIAS map: beha­vi­ors from inter­group affect and ste­reo­types. Journal of per­son­al­ity and social psy­cho­logy, 92(4), pp.631–48.

Doris, J.M., 2002. Lack of char­ac­ter: Personality and mor­al beha­vi­or, Cambridge, UK: Cambridge University Press.

Fiebich, A. & Coltheart, M., 2015. Various Ways to Understand Other Minds: Towards a Pluralistic Approach to the Explanation of Social Understanding. Mind and Language, 30(3), pp.235–258.

Fiske, S.T., 2015. Intergroup biases: A focus on ste­reo­type con­tent. Current Opinion in Behavioral Sciences, 3(April), pp.45–50.

Fiske, S.T., Cuddy, A.J.C. & Glick, P., 2002. A Model of (Often Mixed Stereotype Content: Competence and Warmth Respectively Follow From Perceived Status and Competition. Journal of per­son­al­ity and social psy­cho­logy­er­son­al­ity and social psy­cho­logy, 82(6), pp.878–902.

Gawronski, B., 2004. Theory-based bias cor­rec­tion in dis­pos­i­tion­al infer­ence: The fun­da­ment­al attri­bu­tion error is dead, long live the cor­res­pond­ence bias. European Review of Social Psychology, 15(1), pp.183–217.

Gilbert, D.T. et al., 1995. The Correspondence Bias. Psychological Bulletin, 117(1), pp.21–38.

Hohwy, J., 2013. The pre­dict­ive mind, Oxford University Press.

Hohwy, J. & Palmer, C., 2014. Social Cognition as Causal Inference: Implications for Common Knowledge and Autism. In M. Gallotti & J. Michael, eds. Perspectives on Social Ontology and Social Cognition. Dordrecht: Springer Netherlands, pp. 167–189.

Kilner, J.M. & Frith, C.D., 2007. Predictive cod­ing: an account of the mir­ror neur­on sys­tem. Cognitive Processes, 8(3), pp.159–166.

Koster-Hale, J. & Saxe, R., 2013. Theory of Mind: A Neural Prediction Problem. Neuron, 79(5), pp.836–848.

Macrae, C.N., Stangor, C. & Milne, A.B., 1994. Activating Social Stereotypes: A Functional Analysis. Journal of Experimental Social Psychology, 30(4), pp.370–389.

Mason, M.F., Cloutier, J. & Macrae, C.N., 2006. On con­stru­ing oth­ers: Category and ste­reo­type activ­a­tion from facial cues. Social Cognition, 24(5), p.540.

McGlothlin, H. & Killen, M., 2010. How social exper­i­ence is related to children’s inter­group atti­tudes. European Journal of Social Psychology, 40(4), pp.625–634.

Mcglothlin, H. & Killen, M., 2006. Intergroup Attitudes of European American Children Attending Ethnically Homogeneous Schools. Child Development, 77(5), pp.1375–1386.

Ross, L., 1977. The Intuitive Psychologist And His Shortcomings: Distortions in the Attribution Process. Advances in Experimental Social Psychology, 10©, pp.173–220.

Sagar, H.A. & Schofield, J.W., 1990. Racial and beha­vi­or­al cues in Black and White chil­dren’ s per­cep­tion of ambigu­ously aggress­ive acts. Journal of per­son­al­ity and social psy­cho­logy, 39(October), pp.590–598.

Todorov, A. et al., 2008. Understanding eval­u­ation of faces on social dimen­sions. Trends in Cognitive Sciences, 12(12), pp.455–460.


Thanks to Melanie Killen and Joan Tycko for per­mis­sion to use images of exper­i­ment­al stim­uli from McGlothlin & Killen (2006, 2010).