Assessing underlying processes of implicit weight bias: a diffusion model analysis

Abstract Weight bias is a widespread issue, yet its cognitive mechanisms remain unclear. Using cognitive modeling across two studies ( $$N_1$$ = 164, $$N_2$$ = 60), we investigated the processes underlying implicit weight bias. We assessed implicit weight bias with an Affective Priming-like Task, in which participants see a prime and then classify adjectives as positive or negative. Despite instructions to ignore the prime, responses are typically faster when prime and target share the same valence. In our studies, participants were primed with four gender-matched body shapes, of which we used the difference between the intermediate weight and higher weight primes to operationalize weight bias. To distinguish between response (starting point) and stimulus bias (drift rate), we applied a Bayesian Hierarchical Diffusion Decision Model. Results from both studies revealed an effect on the starting point: After intermediate weight primes, participants showed a stronger a priori tendency to choose the positive response compared with after higher weight primes. By contrast, drift rate differences between the primes were not credible. These findings suggest that implicit weight bias when measured with an Affective Priming-like Task manifests primarily as response, not as stimulus bias. This response bias is often linked to judgment-related rather than perceptual evidence-related processes and may inform the development of process-based anti-bias trainings.