What Implicit Measures of Bias Can Do

Abstract

Gawronski, Ledgerwood, and Eastwick (this issue; GLE) bring much needed attention to the limitations of currently available implicit measures as tools for studying unconscious bias. We agree with the authors of the target article that the current state of the literature offers little reason to believe that commonly used implicit measures, such as sequential priming (Fazio, Sanbonmatsu, Powell, & Kardes, 1986), the Implicit Association Test (Greenwald, McGhee, & Schwartz, 1998), and the Affect Misattribution Procedure (Payne, Cheng, Govorun, & Stewart, 2005), capture unconscious influences of social category cues on behavioral responses. If anything, the evidence suggests the opposite: Participants may well be aware of how their responses are influenced by social cues on implicit measures (Hahn, Judd, Hirsh, & Blair, 2014; Hahn & Gawronski, 2019), although to what degree and as a result of what type of process or processes remains to be investigated (Morris & Kurdi, 2022). Nonetheless, even if the extent of awareness differs depending on the specific conditions of the task, the lack of compelling evidence for the ability of currently available implicit measures to index unconscious bias is surprising. As GLE observe, the concepts of unconscious bias and bias on implicit measures have been, and continue to be, conflated, both in the empirical literature and popular discourse. This conundrum will prompt many readers to wonder: If implicit measures of bias are not useful for measuring unconscious bias, are they useful at all? They are. Whether or not they shed light on unconscious bias, implicit measures have been, and we believe will remain, essential to the study of social cognition. We suspect that the lead author of the target article, who has used implicit measures of bias to make numerous contributions to the understanding of social information processing, would agree. But what is it, exactly, that implicit measures of bias are good for, if not probing the human unconscious? This is the question we address in the current commentary. Broadly speaking, implicit measures of bias have been and continue to be critical for addressing two related questions: (i) what is the nature of unintentional bias? and (ii) what is the cognitive architecture of bias? In what follows, we show how implicit measures fuel progress on both fronts while, crucially, also advancing the translational goal of revealing the nature of, and reducing, groupbased inequality. Using Implicit Measures of Bias to Reveal the Nature of Unintentional Bias