These studies investigated how performance on a verbal task can be moderated by varying the salience of social identities. In Experiment 1, female Asian-American undergraduates completed a survey designed to highlight their ethnic identity, their gender identity, or their identity as a student. After completing one of the surveys, the students were told that they would complete a task assessing their verbal ability, inducing possible stereotype threat based on Asian identity and possibly stereotype lift based on female identity. Results showed that Asian women performed best in the condition in which gender had been highlighted, marginally less well in the control condition, and marginally worse still in the Asian identity condition. Performance differed significantly between the gender and ethnic identity conditions. In addition, women in the female identity condition expressed significantly more confidence in their verbal abilities compared with the other two conditions. Experiment 2 focused on identifying the default strength of race versus gender identities in Asian-American female undergraduates who were to take a test focusing on verbal ability. Procedures were identical to Experiment 1, except the students did not take a verbal test (though they had been led to believe that they would). Instead, after completing one of the surveys, they were prompted to generate descriptions of six memories from their past, three based on their gender and three based on their ethnicity. Results showed that in the control condition (when neither gender nor ethnicity had been highlighted), students were equally likely to generate a memory based on gender or ethnicity first. However, the rated positivity of the first memory was higher when it focused on gender than when it focused on ethnicity. This suggests that women recalled memories that might be adaptive given that the upcoming test focused on a domain in which stereotypes favored the performance of women. These studies are consistent with other research suggesting that highlighting one of multiple social identities can affect task performance, depending on the stereotypical expectations associated with the specific salient identity.