My work examines how stereotype threat (worries people have about confirming negative stereotypes about their group's ability through their own performance) can negatively impact women's ability to learn and perform in the domain of math. Research in the lab has shown that the stereotype that "women are bad at math" can inhibit women's ability to learn by reducing their ability to encode math-related information; and it can even reduce or eliminate incidental learning (i.e., leaning that is unintended, but that is acquired naturally through repeating the same task several hundred times). The negative gender-based math stereotype also hurts women's ability to perform mathematical tasks that they have learned well. In our work, stereotype threat hurts performance by making women's gender identity more accessible in memory, thereby undermining the cognitive resources needed to solve difficult math problems.

In addition, my lab has conducted a large amount of work trying to understand when evaluations that are measured indirectly (by using implicit attitude measures that are based on response latency and difficult to control) are inconsistent with evaluations measured directly (by asking people their attitudes on a scale). My collaborators and I have found much support for the idea that directly and indirectly measured attitudes can be formed and changed by different processes. Specifically, directly measured attitudes seem most responsive to logic (although sometimes flawed) and verbalizable rules. Indirectly measured attitudes seem most responsive to pairing the attitude object with positive or negative information. When directly and indirectly measured attitudes are inconsistent, our work indicates that this state is uncomfortable and people are motivated to eliminate these negative feelings by, for instance, learning more about the attitude object.