ABSTRACT
Though implicit measures often are portrayed as process-pure measures of automatic attitudes, instead, they reflect the joint contributions of automatic and controlled processes. As such, automatic and controlled components of attitudes are better measured, not with two separate measures, but with process dissociation (PD) techniques that extract independent estimates of automatic and controlled influences from performance on a single task. I will describe such approaches for analyzing responses on implicit tasks, concentrating on our own Quadruple Process (Quad) model (Sherman, Gawronski, Hugenburg, & Groom, 2005; Sherman, Gawronski, Conrey, Hugenburg, & Groom, 2008). The application of the Quad model provides important insights into central questions surrounding the conceptualization, measurement, and interpretation of implicit attitudes. Among these, Quad model analyses shed new light on Bargh’s (1999) provocative claim that the “cognitive monsters” of implicit attitudes and stereotypes cannot be controlled. I will reexamine the status of this claim, concluding that, in general, the controllability of such biases has been underestimated. However, some forms of control may be more attainable than others. In particular, I
will argue that a failure to consider the multifaceted nature of implicit task performance has led to both an overestimation of the ease with which people can control the initial automatic activation of relevant associations and an underestimation of the extent to which the expression of those associations subsequently can be controlled.
