Cognitive Processes Underlying the Weapon Identification Task: A Comparison of Models Accounting for Both Response Frequencies and Response Times

The weapon identification task (WIT) is a sequential priming paradigm designed to assess effects of racial priming on visual discrimination between weapons (guns) and innocuous objects (tools). We compare four process models that differ in their assumptions on the nature and interplay of cognitive processes underlying prime-related weapon-bias effects in the WIT. All four models are variants of the process dissociation procedure, a widely used measurement model to disentangle effects of controlled and automatic processes. We formalized these models as response time-extended multinomial processing tree models and applied them to eight data sets. Overall, the default interventionist model (DIM) and the preemptive conflict-resolution model (PCRM) provided good model fit. Both assume fast automatic and slow controlled process routes. Additional comparisons favored the former model. In line with the DIM, we thus conclude that automatically evoked stereotype associations interfere with correct object identification from the outset of each WIT trial.