Journal of Contextual Behavioral Science (JCBS)
Volume 39, January 2026
Authors
Denise Aparecida Passarelli, Júlio C. de Rose, Táhcita Medrado Mizael, & Bryan Roche
Key Findings
- Racial bias was assessed using the PODT and explicit self-report scales.
- FAST-based relational training simulated diverse, socially relevant contexts.
- Flexibility and stereotype-inconsistent training reduced racial bias on the PODT.
- A pro-Black bias emerged in explicit ratings in the stereotype-inconsistent group.
Abstract
Racial biases can be captured using self-report methods and simulated first-person shooter games, such as Correll's Police Officer's Dilemma Task (PODT). The current study adopts a behavior-analytic approach to reducing racial bias on such measures. The prophylactic intervention involved exposing 116 White adults (mean age: 23.13 years; 67 females) to different configurations of the Function Acquisition Speed Test (FAST) across four conditions, before completing two self-report measures and the PODT. These conditions aimed to create functional response classes that were either consistent with racial bias, inconsistent with racial bias, or included both types (i.e., relational flexibility). A fourth control condition involved no intervention. Participants in the stereotype-consistent condition (C3) were more accurate in shooting armed Black targets compared to White targets. In addition, participants in both the stereotype-consistent (C3) and control (C4) groups showed significantly more false alarms for unarmed Black targets than for unarmed White targets, as well as a significantly more liberal decision criterion when responding to Black targets, meaning participants required less certainty before deciding to shoot Black targets. No such racial bias was observed in relational flexibility (C1) or stereotype-inconsistent (C2) conditions. These findings offer insight into how verbal relational training can influence and reduce racial bias in shoot decisions.