By Chad E. Forbes

Past research suggests that while boys and girls perform fairly comparably on tests of mathematic aptitude, girls tend to systematically report more negative math attitudes and emotions. New research out of my lab suggests that this paradox might arise in part because performance situations that prime negative math stereotypes about women’s math ability, i.e., stereotype threatening situations, alter how the brain processes information during the performance situation at a fundamental level. Our work suggests that when negative feedback is received by women in stereotype threatening situations their brains interpret it in a manner consistent with any physiological threat. This results in an attentional and encoding bias towards negative feedback very early in the information processing stream that undermines performance among women on future math tests and prompts feelings of post-test anxiety accordingly. 

To better understand how these effects manifest let’s take a more comprehensive look at the problem. A recent meta-analysis conducted by Else-Quest, Hyde, and Linn (2010) examined gender differences in mathematic aptitude. Among over 493,000 boys and girls across the globe they essentially found that there was a negligible difference between boys and girls in mathematic achievement (the magnitude of the effect, cohen’d d, was less than 0.15-a large effect would be considered north of .80). When there was a difference in math achievement between boys and girls in a given country, it was largely because these countries were more misogynistic in general; the extent to which there was gender equity in school enrollment, research oriented jobs and parliamentary representation were key predictors of gender differences. 

Where boys and girls did systematically differ, however, was with respect to how they felt about math. Girls in general tended to report more negative math attitudes and emotions compared to boys. Thus while boys and girls tended to perform comparably to one another on tests of math achievement, girls walked away from these situations feeling more negative about math. An important ingredient, one would imagine, for keeping girls motivated to pursue careers in math-intensive fields. This begs the question: How in the world do such paradoxical patterns emerge between boys and girls?  

Recent work out of my lab provides a potential mechanism for how exactly these patterns emerge. Some of it may have do with how societal beliefs and expectations, or stereotypes, about women’s math ability negatively effects women specifically in situations where their math merit is on the line. According to a seminal theory termed stereotype threat (Steele & Aronson, 1995), when targets of negative stereotypes (like women in math domains) find themselves in situations where they’re reminded of their negative group status (like when women mark their gender before taking the SAT or GRE) and they have to perform in the stigmatized domain, they experience a fear that their behaviors may inadvertently confirm the negative group stereotype. Ironically this additional burden results in confirmation of the very stereotype women are trying to disprove. Interestingly, these effects appear to be most pronounced among the women who care most about math and may manifest even in settings where math merit is not on the line, e.g., women just sitting in a math classroom where they are outnumbered by men (termed social identity threat; Murphy, Steele, & Gross, 2007).  

Some of our past work suggests that stereotype threatening situations may create a “perfect storm” for altering how women attend to and ultimately encode or remember information in situations where their math merit is on the line. A lot of this has to do with the likelihood that stereotype threatening contexts are stressful, i.e., physiologically arousing and ripe with negative emotions, in and of themselves. We know from extant literature in the realm of cognitive and clinical psychology that when individuals are stressed, not only does it alter what information is attended to, it alters how that information is ultimately remembered. Specifically, according to research on mood congruent or emotional memory encoding, when individuals are placed in contexts that elicit negative mood states they tend to pay more attention to negative information and encode or remember that information better than information that is at odds with the mood state (i.e., neutral or positive information). This suggests that when women are in stereotype threatening situations, experiencing physiological arousal and negative affect or stress accordingly, they may be more likely to attend to and encode negative, stereotype consistent information better than positive, stereotype inconsistent information.

Some of our initial work demonstrated just how attention grabbing stereotype confirming information is when administered to individuals in stereotype threatening situations (Forbes, Schmader, & Allen, 2008; Forbes & Leitner, 2014). For instance, in one study we had women complete a math task that provided them with feedback indicating they had solved a problem correctly or not. In the stereotype threat condition this task was framed as a diagnostic math test and in the control condition it was framed as a problem solving task. Women in the stereotype threat condition also marked their gender before the test, theoretically priming stereotype threat. Findings revealed that stereotype threatened women exhibited greater activity in neural regions associated with fundamental attentional processes in response to negative feedback compared to positive feedback as early as 50ms after receiving the feedback. More importantly, stereotype threatened women tended to underperform on the same math task to the extent they exhibited a neural attentional bias towards the negative feedback, suggesting this bias taxed cognitive resources otherwise necessary for optimal performance. Those in stereotype neutral contexts actually tended to exhibit a bias towards positive feedback (though this effect has proven less reliable).

But how does this attentional bias ultimately manifest in to more negative feelings among women even if their performance is comparable to their peers? Well, we think a lot of it has to do with how negative information is not only attended to but encoded.  In a separate set of studies (Forbes, Duran, Magerman & Leitner, 2015) we had men and women complete an initial math task (framed as either a diagnostic math test in the stereotype threat condition or a problem solving task in the non-threat condition) that provided them with veridical feedback about their performance and was designed to elicit comparable amounts of correct and incorrect answers.  And it did-women and men performed comparably on the math task. Importantly, the feedback was provided in a unique font on every trial (e.g., CORRECT vs. CORRECT or WRONG vs. WRONG -- and we counterbalanced all of these things like good scientists), which allowed us to test the idea that any information (such as a given font) associated with stereotype confirming feedback would be better encoded by women under threat. Participants then completed a standard difficult math test comprised of GRE problems (which yielded the traditional stereotype threat effects) and then we gave them a surprise memory test to determine whether individuals accurately remembered font-feedback pairings they were exposed to during the feedback task.

Indeed, we found that women under threat remembered fonts paired with feedback telling them they got a question wrong (i.e., stereotype confirming feedback) better than fonts paired with feedback telling them they got a question correct (i.e, stereotype-disconfirming feedback) and better than anyone else in every other condition; non-threatened women did not display this bias in recall, nor did men in either condition (in fact they were pretty much at chance). In a follow up study this effect even extended to neutral information presented in conjunction with the negative feedback. That is, when the feedback was embedded within pictures of cities or the woods, stereotype threatened women exhibited more accurate memory for these city and nature pictures when they were paired with negative compared to positive feedback.  

More importantly, we found evidence that biases in encoding had downstream consequences on both how stereotype threatened women performed on future math tests and felt about themselves after the performance. Consistent with our attentional bias findings we found across studies that more accurate encoding of negative feedback predicted underperformance on the future difficult math test (the difficult GRE test), which in turn predicted greater post-test self-reported anxiety. This pattern was only evident among stereotype threatened women. Thus despite performing comparably on the initial math task, the attentional and encoding bias exhibited towards negative feedback appeared to be cognitively taxing in and of itself, facilitating underperformance on future difficult problems and feelings of anxiety after the test.

These findings could provide one explanation for the findings reported in Else-Quest et al.’s (2010) meta-analysis. There are important distinctions of course, namely that boys and girls in standardized testing situations typically do not receive feedback. It’s possible that in these situations feedback is simply replaced with more negative thoughts, worry, concern and self-doubt in general-all of which have been demonstrated to occur in stereotype threatening situations (for a review see Schmader, Johns, & Forbes, 2008). Current work in my lab is finding some evidence for this possibility.

These findings highlight the subtle yet deleterious effects that stereotypes can have on the stigmatized in our society and point to the possibility that even when the field appears to be level, stigmatized individuals may leave performance situations with something more than just a grade-a memory or perception that suggests they don’t have what it takes to be successful in a field that is important to them. Current work in our lab is examining the long-term consequences of these memory biases but also ways to combat them. For instance, in past work we’ve found that stereotype threatened minorities’ post-task performance perceptions vary as a function of how well brain regions involved in self-regulatory and coping processes communicate with one another before performing in a stereotype threatening situation: Better copers have more accurate performance perceptions and experience less self-doubt (Forbes, Leitner, Jordan, Magerman, Schmader, & Allen, 2014). Of course having individuals perform in stereotype neutral contexts or implement any number of successful strategies used to combat stereotype threat effects, e.g., self-affirmation, could also help reverse these effects. And it probably wouldn’t hurt to bask a little in the glory of past accomplishments and experiences in the field as well.


Else-Quest, N. M., Hyde, J. S., & Linn, M. C. (2010). Cross-national patterns of gender differences in mathematics: a meta-analysis. Psychological bulletin, 136, 1,103.

Forbes, C. E., Duran, K. A., Leitner, J. B., & Magerman, A. (2015). Stereotype Threatening Contexts Enhance Encoding of Negative Feedback to Engender Underperformance and Anxiety. Social Cognition, 33(6), 605-625.

Forbes, C. E., & Leitner, J. B. (2014). Stereotype threat engenders neural attentional bias towards negative feedback to undermine performance. Biological Psychology. DOI: 10.1016/j.biopsycho.2014.07.007

Forbes, C. E., Leitner, J. B., Jordan, K., Magerman, A., Schmader, T., & Allen, J. J. B. (2015). Spontaneous default mode network phase locking moderates performance perceptions under stereotype threat. Social Cognitive and Affective Neuroscience.

Forbes, C. E., Schmader, T., & Allen, J. J. (2008). The role of devaluing and discounting in performance monitoring: a neurophysiological study of minorities under threat. Social Cognitive and Affective Neuroscience, 3, 253-261.

Murphy, M. C., Steele, C. M., & Gross, J. J. (2007).  Signaling threat: How situational cues affect women in math, science, and engineering settings.  Psychological Science, 18, 879-885.

Schmader, T., Johns, M., & Forbes, C. (2008).  An integrated process model of stereotype threat effects on performance. Psychological Review, 115, 336-356.

Steele, C. M., & Aronson, J. (1995).  Stereotype threat and the intellectual test performance of African Americans.  Journal of Personality and Social Psychology, 69, 797-811.

Chad E. Forbes (Ph.D., University of Arizona) is an Assistant Professor of Psychological and Brain Sciences at the University of Delaware. With a background spanning from molecular biology to complex social processes, Dr. Forbes utilizes behavioral methodologies as well as electroencephalography (EEG), functional magnetic resonance imaging (fMRI), lesion studies and genetic approaches to investigate social phenomena. Specifically, he examines how priming negative stereotypes affects our perceptions as well as stigmatized individuals in our society, e.g. minorities and women, to ironically engender situations that inadvertently reinforce the stereotype. Dr. Forbes is currently funded by the National Science Foundation to develop a model that outlines how and why minorities and women are more likely to leave academics and STEM (Science Technology Engineering and Mathematics) fields respectively, how negative feelings can be transmitted to others in STEM groups, as well as how these phenomena can be reversed. He has numerous publications on these topics in a wide array of journals, including Annual Reviews of Neuroscience, Cerebral Cortex, and the Journal of Personality and Social Psychology, and was recently recognized as a “Rising Star” by the American Psychological Association.