Starting with Steele and Aronson (1995), research documents the performance decrements resulting from the activation of a negative task-relevant stereotype. I suggest that negative stereotypes can generate better performance, as they produce a prevention focus (Higgins, 2000; Seibt & Förster, 2004), because a prevention focus leads to greater cognitive flexibility in a task where points are lost (Maddox, Markman, & Baldwin, 2006). My prior work, Experiments 1 and 2, done in collaboration with Arthur B. Markman, W. Todd Maddox, and Grant C. Baldwin, used a category learning task that requires the participant test different explicit rules to correctly categorize stimuli. Half of the participants gained points for correct responses while half of the participants lost points for correct responses. We primed a positive or a negative gender stereotype. The negative prime matches the losses environment while the positive prime matches the gains environment. The match states are assumed to increase dopamine release into frontal brain areas leading to increased cognitive flexibility and better task performance whereas the mismatch states should not. Thus, we predict and obtain a 3-way interaction between Stereotype (Positive, Negative), Gender (Male, Female), and Reward structure (Gains, Losses) for accuracy and strategy. Experiments 3 and 4 used a category learning task, which requires the implicit learning system to govern participant responses. This task had an information-integration category structure and involves the striatum (e.g., Maddox & Ashby, 2004). Importantly, cognitive flexibility will hurt performance using this category structure. I therefore predicted that regulatory match states, created by manipulating Stereotype and Reward structure, will produce worse performance than mismatch states. I did not completely reverse the effects described in Experiments 1 and 2 as predicted. I found evidence supporting my predictions using computational models to test for task strategy in Experiment 3 and found results consistent with the flexibility hypothesis in Experiment 4. Importantly, I believe that stereotype threat effects should not be conceptualized as a main effect with negative stereotypes producing worse performance than positive stereotypes, but instead as an interaction between the motivational state of the individual, task environment, and type of task performed.