In his book Social Perception and Social Reality Jussim writes:
Is the glass half full or half empty? As everyone knows the optimist says “half full,” the pessimist says “half empty.” … The parable is NOT “The glass is half empty but the optimist sees it as 90% full. […] the parable is in sharp conflict with many social science and social psychological researchers, who do indeed often claim that our beliefs and expectations powerfully influence and distort our perceptions of objective social reality. (Jussim Reference Jussim2012, p. 3)
While Jussim seems to disagree with social scientists and social psychologists, research in neuroscience on social person perception also supports the claim that indeed, categorization can change how humans see an individual. This commentary argues that (a) stereotype activation is mostly automatic and (b), that stereotype activation leads to person perception being inaccurate as (c) it is insufficient. The commentary will refer to Part Six of Jussim's (Reference Jussim2012) book, “Stereotypes,” and particularly Chapter 18, titled “Stereotypes and Person Perception.”
In Part Six, Jussim describes numerous studies and discusses the notion that “all beliefs about groups – all stereotypes – are not necessarily wrong, irrational, and malevolent” (p. 270). Jussim argues that many researchers, lay people, and others believe that stereotypes are inaccurate whilst he does not think they are. He mentions multiple examples for stereotypes, such as the notion that men are more likely to commit murder than women, and “Jews really are, on average, richer” (p. 272). A recent mainstream UK Channel 4 documentary entitled “Things We Won't Say About Race That Are True” discussed similar issues; for example, that one “could not say” that “Romanians in the UK are far more likely to be pickpockets” as this was seen as being a racist remark by many people. It is surprising, that we are recently (again) discussing this issue, as in 1954 the famous social psychologist Gordon Allport had already suggested that stereotypes might have “a kernel of truth” (Allport Reference Allport1954b, 1954 p. 195). I want to argue that the discussion of whether stereotypes are true (or not true, or partly true) is irrelevant, because even if one were to agree that stereotypes were 100% accurate, the conclusion that this therefore makes person perception accurate, seems problematic. The activation of stereotypes – as recent social neuroscience research supports – is mostly automatic, and subsequently distorts person perception, thus making it inaccurate. In particular, I highlight recent social neuroscience research which contradicts the statement: “If the stereotype is approximately accurate and one only has a small bit of ambiguous information about an individual, using the stereotype as a basis for judging the person will likely enhance accuracy” (Jussim Reference Jussim2012, p. 365, emphasis in original).
Jussim mentions several behavioural experimental studies to support his claim. For instance, Jussim discussed a study by Cohen (Reference Cohen1981), who presented participants with a videotaped conversation and gave the information that the target person was a librarian or waitress either before or after the video. Cohen (Reference Cohen1981) found that those participants who received the information before the video remembered more (stereotype-consistent) attributes. Jussim argues that this supports the idea that using the stereotype increased accuracy. However, one should note that in this experimental scenario, the participants had either one single aspect of information or no information at all. Indeed, it should be considered that in no real life scenario – compared to a behavioural experimental scenario – would one have only one aspect of information (say race) available, and nothing else. The same criticism applies to the study by Macrae et al. (Reference Macrae, Bodenhausen, Milne and Jetten1994), in which either the information about the profession of the protagonist was given, or no information was given at all. Thus, using stereotypical information seems to increase accuracy if this one aspect is the only information one has; however, this would ideally need to be compared to non-stereotypical information that one might have, which is difficult to construct in behavioural experiments.
In the 2012 book, Jussim also mentions his own experimental work (e.g., Jussim et al. Reference Jussim, Eccles and Madon1996), in which he found that teachers were rating their pupils performance accurately, and that they were more likely to use past performance and motivation rather than gender stereotypes. In fact, Jussim et al. (Reference Jussim, Eccles and Madon1996) found that teachers predicted that girls would outperform boys in that school year, which was an accurate prediction of their real performance. Jussim (Reference Jussim2012) argues that this experimental evidence supports the claims that (a) teachers were not relying solely on the stereotype, and (b) that their perceptions lead to accurate prediction. One key problem here is the use of self-report measures, which could have led teachers to give socially desirable answers, in order to indicate that they were not applying stereotypical views. Indeed, self-report data might be weak in assessing whether a stereotype was applied or not. I argue that the activation of stereotypical information is mostly automatic and precedes subsequent information processing. Jussim (Reference Jussim2012) himself supports the theory that “people may sometimes receive stereotype information before individuating information” (p. 381).
Research in developmental psychology and primatology has demonstrated that even very young babies, and primates also, show a tendency to categorize individuals into groups (e.g., Kinzler & Spelke Reference Kinzler and Spelke2011). fMRI research supports this theory, by demonstrating that in particular the fusiform face area correlates not only to person perception, but also distinguishes faces on the bases of age, gender, and race (e.g., Contreras et al. Reference Contreras, Banaji and Mitchell2013). Numerous neuroscience studies found differential activation patterns, for example, in the amygdala, fusiform gyrus, dorso-lateral prefrontal cortex, and insula, elicited by passive viewing of faces of different, races, gender, attractiveness, and body shape (e.g., Cunningham et al. Reference Cunningham, Johnson, Raye, Gatenby, Gore and Banaji2004). We recently conducted an fMRI study, in which we used region of interest as well as time-course analyses, confirming that fusiform gyrus activity differs for black versus white faces, as well as showing a different time pattern (e.g., fusiform gyrus activity to white faces over time decreases, but increases for black faces) (Terbeck et al. Reference Terbeck, Kahane, McTavish, McCutcheon, Hewstone, Savulescu, Chesterman, Cowen and Norbury2015). The fusiform gyrus – and in particular the fusiform face area – is associated with automatic processing of faces (as compared to objects, for example), suggesting that the categorization of attributes such as age and gender, is closely associated with immediate face perception.
However, further neuroimaging studies also indicated that social cognitive goals reduced amygdala activity differences to faces of different races (Wheeler & Fiske Reference Wheeler and Fiske2005). The authors found that when engaging in a classification task (such as deciding the race of the face), as compared to a control task (such as searching for a dot in the picture) could change the effect, suggesting that automatic stereotype activation was not always inevitable. We also determined how basic emotional arousal, mediated by neurotransmitter activity, such as activity of noradrenaline, could reduce social perception differences. In one study we found, using the Implicit Association Test (IAT), that racial biases were reduced in healthy volunteers after a single dose of beta-adrenoreceptor blocker propranolol (Terbeck et al. Reference Terbeck, Kahane, McTavish, Savulescu, Cowen and Hewstone2012). More specifically, we found that the IAT score (a computer response-time–based classification task measuring implicit racial biases) was significantly lower after propranolol intervention, suggesting that noradrenergic-based emotional arousal might contribute to generating implicit racial biases.
Our own neuroscience research (Terbeck et al. Reference Terbeck, Kahane, McTavish, McCutcheon, Hewstone, Savulescu, Chesterman, Cowen and Norbury2015) also showed that basic emotions, such as aggression and fear, might impact on application of stereotypes/categories and the processing of faces in the brain. In this study, participants either received propranolol or placebo before viewing unfamiliar black and white faces during the fMRI scan. We found that differences in fusiform gurus activity were reduced after administration of noradrenergic blockade, which reduced fusiform gyrus as well as thalamic (also involved in attention processing) activity to black but not to white faces (Terbeck et al. Reference Terbeck, Kahane, McTavish, McCutcheon, Hewstone, Savulescu, Chesterman, Cowen and Norbury2015). Furthermore, time course analysis revealed that sensitization to black faces was reduced, while the activity pattern to white faces was not affected. This suggests that the arousal level can implement on the extent to which attention is focused on the category level (e.g., racial aspect).
As the above discussion demonstrates, it might be quite difficult to simply instruct in a behavioural experiment “Ignore the stereotype” or to investigate whether participants applied stereotypes in their decisions or not. Our research supports the theory that the processing of categories in faces, and the processing of stereotypes, is mostly automatic and immediate. Furthermore, besides the automatic application of the stereotype (e.g., classification of faces according to age, race, or gender), the focus of attention (i.e., differences in thalamic activity was observed with noradrenergic intervention) might be shifted to stereotypical information (which, as we observed, could be modulated by emotional arousal, such as noradrenaline-mediated fight-or-flight responses). Thus, automatic stereotype activation makes person perception insufficient. Additionally, if it is just focused on one aspect (say race, and not also age, gender, facial characteristics etc.) it is not reliable for decision making, either. More specifically, person perception would be more accurate if more information about the person would be processed, which implies that application of a stereotype makes person perception inaccurate as it is insufficient. However, since automatic stereotype activation within the brain leads to differential activation patterns for in- and out-groups, it could be suggested that indeed some additional information was not processed or not seen.
To conclude, regardless of whether the stereotype content is accurate or not, the automatic activation biases in face processing makes it insufficient (focused on one aspect) and thus inaccurate. Indeed, 70 years of research after Gordon Allport has demonstrated both the relevance and the limitations of stereotypes in social perception.
You have
Access
In his book Social Perception and Social Reality Jussim writes:
Is the glass half full or half empty? As everyone knows the optimist says “half full,” the pessimist says “half empty.” … The parable is NOT “The glass is half empty but the optimist sees it as 90% full. […] the parable is in sharp conflict with many social science and social psychological researchers, who do indeed often claim that our beliefs and expectations powerfully influence and distort our perceptions of objective social reality. (Jussim Reference Jussim2012, p. 3)
While Jussim seems to disagree with social scientists and social psychologists, research in neuroscience on social person perception also supports the claim that indeed, categorization can change how humans see an individual. This commentary argues that (a) stereotype activation is mostly automatic and (b), that stereotype activation leads to person perception being inaccurate as (c) it is insufficient. The commentary will refer to Part Six of Jussim's (Reference Jussim2012) book, “Stereotypes,” and particularly Chapter 18, titled “Stereotypes and Person Perception.”
In Part Six, Jussim describes numerous studies and discusses the notion that “all beliefs about groups – all stereotypes – are not necessarily wrong, irrational, and malevolent” (p. 270). Jussim argues that many researchers, lay people, and others believe that stereotypes are inaccurate whilst he does not think they are. He mentions multiple examples for stereotypes, such as the notion that men are more likely to commit murder than women, and “Jews really are, on average, richer” (p. 272). A recent mainstream UK Channel 4 documentary entitled “Things We Won't Say About Race That Are True” discussed similar issues; for example, that one “could not say” that “Romanians in the UK are far more likely to be pickpockets” as this was seen as being a racist remark by many people. It is surprising, that we are recently (again) discussing this issue, as in 1954 the famous social psychologist Gordon Allport had already suggested that stereotypes might have “a kernel of truth” (Allport Reference Allport1954b, 1954 p. 195). I want to argue that the discussion of whether stereotypes are true (or not true, or partly true) is irrelevant, because even if one were to agree that stereotypes were 100% accurate, the conclusion that this therefore makes person perception accurate, seems problematic. The activation of stereotypes – as recent social neuroscience research supports – is mostly automatic, and subsequently distorts person perception, thus making it inaccurate. In particular, I highlight recent social neuroscience research which contradicts the statement: “If the stereotype is approximately accurate and one only has a small bit of ambiguous information about an individual, using the stereotype as a basis for judging the person will likely enhance accuracy” (Jussim Reference Jussim2012, p. 365, emphasis in original).
Jussim mentions several behavioural experimental studies to support his claim. For instance, Jussim discussed a study by Cohen (Reference Cohen1981), who presented participants with a videotaped conversation and gave the information that the target person was a librarian or waitress either before or after the video. Cohen (Reference Cohen1981) found that those participants who received the information before the video remembered more (stereotype-consistent) attributes. Jussim argues that this supports the idea that using the stereotype increased accuracy. However, one should note that in this experimental scenario, the participants had either one single aspect of information or no information at all. Indeed, it should be considered that in no real life scenario – compared to a behavioural experimental scenario – would one have only one aspect of information (say race) available, and nothing else. The same criticism applies to the study by Macrae et al. (Reference Macrae, Bodenhausen, Milne and Jetten1994), in which either the information about the profession of the protagonist was given, or no information was given at all. Thus, using stereotypical information seems to increase accuracy if this one aspect is the only information one has; however, this would ideally need to be compared to non-stereotypical information that one might have, which is difficult to construct in behavioural experiments.
In the 2012 book, Jussim also mentions his own experimental work (e.g., Jussim et al. Reference Jussim, Eccles and Madon1996), in which he found that teachers were rating their pupils performance accurately, and that they were more likely to use past performance and motivation rather than gender stereotypes. In fact, Jussim et al. (Reference Jussim, Eccles and Madon1996) found that teachers predicted that girls would outperform boys in that school year, which was an accurate prediction of their real performance. Jussim (Reference Jussim2012) argues that this experimental evidence supports the claims that (a) teachers were not relying solely on the stereotype, and (b) that their perceptions lead to accurate prediction. One key problem here is the use of self-report measures, which could have led teachers to give socially desirable answers, in order to indicate that they were not applying stereotypical views. Indeed, self-report data might be weak in assessing whether a stereotype was applied or not. I argue that the activation of stereotypical information is mostly automatic and precedes subsequent information processing. Jussim (Reference Jussim2012) himself supports the theory that “people may sometimes receive stereotype information before individuating information” (p. 381).
Research in developmental psychology and primatology has demonstrated that even very young babies, and primates also, show a tendency to categorize individuals into groups (e.g., Kinzler & Spelke Reference Kinzler and Spelke2011). fMRI research supports this theory, by demonstrating that in particular the fusiform face area correlates not only to person perception, but also distinguishes faces on the bases of age, gender, and race (e.g., Contreras et al. Reference Contreras, Banaji and Mitchell2013). Numerous neuroscience studies found differential activation patterns, for example, in the amygdala, fusiform gyrus, dorso-lateral prefrontal cortex, and insula, elicited by passive viewing of faces of different, races, gender, attractiveness, and body shape (e.g., Cunningham et al. Reference Cunningham, Johnson, Raye, Gatenby, Gore and Banaji2004). We recently conducted an fMRI study, in which we used region of interest as well as time-course analyses, confirming that fusiform gyrus activity differs for black versus white faces, as well as showing a different time pattern (e.g., fusiform gyrus activity to white faces over time decreases, but increases for black faces) (Terbeck et al. Reference Terbeck, Kahane, McTavish, McCutcheon, Hewstone, Savulescu, Chesterman, Cowen and Norbury2015). The fusiform gyrus – and in particular the fusiform face area – is associated with automatic processing of faces (as compared to objects, for example), suggesting that the categorization of attributes such as age and gender, is closely associated with immediate face perception.
However, further neuroimaging studies also indicated that social cognitive goals reduced amygdala activity differences to faces of different races (Wheeler & Fiske Reference Wheeler and Fiske2005). The authors found that when engaging in a classification task (such as deciding the race of the face), as compared to a control task (such as searching for a dot in the picture) could change the effect, suggesting that automatic stereotype activation was not always inevitable. We also determined how basic emotional arousal, mediated by neurotransmitter activity, such as activity of noradrenaline, could reduce social perception differences. In one study we found, using the Implicit Association Test (IAT), that racial biases were reduced in healthy volunteers after a single dose of beta-adrenoreceptor blocker propranolol (Terbeck et al. Reference Terbeck, Kahane, McTavish, Savulescu, Cowen and Hewstone2012). More specifically, we found that the IAT score (a computer response-time–based classification task measuring implicit racial biases) was significantly lower after propranolol intervention, suggesting that noradrenergic-based emotional arousal might contribute to generating implicit racial biases.
Our own neuroscience research (Terbeck et al. Reference Terbeck, Kahane, McTavish, McCutcheon, Hewstone, Savulescu, Chesterman, Cowen and Norbury2015) also showed that basic emotions, such as aggression and fear, might impact on application of stereotypes/categories and the processing of faces in the brain. In this study, participants either received propranolol or placebo before viewing unfamiliar black and white faces during the fMRI scan. We found that differences in fusiform gurus activity were reduced after administration of noradrenergic blockade, which reduced fusiform gyrus as well as thalamic (also involved in attention processing) activity to black but not to white faces (Terbeck et al. Reference Terbeck, Kahane, McTavish, McCutcheon, Hewstone, Savulescu, Chesterman, Cowen and Norbury2015). Furthermore, time course analysis revealed that sensitization to black faces was reduced, while the activity pattern to white faces was not affected. This suggests that the arousal level can implement on the extent to which attention is focused on the category level (e.g., racial aspect).
As the above discussion demonstrates, it might be quite difficult to simply instruct in a behavioural experiment “Ignore the stereotype” or to investigate whether participants applied stereotypes in their decisions or not. Our research supports the theory that the processing of categories in faces, and the processing of stereotypes, is mostly automatic and immediate. Furthermore, besides the automatic application of the stereotype (e.g., classification of faces according to age, race, or gender), the focus of attention (i.e., differences in thalamic activity was observed with noradrenergic intervention) might be shifted to stereotypical information (which, as we observed, could be modulated by emotional arousal, such as noradrenaline-mediated fight-or-flight responses). Thus, automatic stereotype activation makes person perception insufficient. Additionally, if it is just focused on one aspect (say race, and not also age, gender, facial characteristics etc.) it is not reliable for decision making, either. More specifically, person perception would be more accurate if more information about the person would be processed, which implies that application of a stereotype makes person perception inaccurate as it is insufficient. However, since automatic stereotype activation within the brain leads to differential activation patterns for in- and out-groups, it could be suggested that indeed some additional information was not processed or not seen.
To conclude, regardless of whether the stereotype content is accurate or not, the automatic activation biases in face processing makes it insufficient (focused on one aspect) and thus inaccurate. Indeed, 70 years of research after Gordon Allport has demonstrated both the relevance and the limitations of stereotypes in social perception.
ACKNOWLEDGMENT
Thanks to Adam Chekroud (at Yale University) and Gerald Lang (at Leeds University) for comments.