Challenging the Mechanism for the Implicit Association Test

Implicit biases are stereotypes and attitudes that impact our decisions and actions, playing an important role in discrimination and societal inequities. The most widely used tool for measuring implicit bias is the Implicit Association Test (IAT), which assesses response time in sorting stimuli into labeled categories. Most interpretations of the IAT assume that performance (D-scores) is driven by the activation of conflicting associative memories, or decision ease. We challenged this assumption by decomposing D-scores into additional processes that may impact IAT results: Response caution, which is people’s tendency to sacrifice speed for accuracy, and non-decision time. We used Racing Diffusion Models to differentiate the effects of decision ease from response caution and non-decision time on IAT D-scores across 39 topics (N = 115,601). We found that response caution explained significantly more variance in the D-score, above and beyond both decision ease and non-decision time. Response caution also best predicted people’s explicitly reported biases. These findings challenge the traditional interpretation of D-scores as largely reflective of associative memory activation and highlight the importance of considering multiple cognitive processes when assessing implicit biases.