Gastvortrag von Prof. C. Shawn Green (University of Wisconsin-Madison, USA) am 02.12.2022 um 13:45 Uhr in Raum 417 (Schiffbauergasse).
Abstract: Humans’ thoughts and behaviors are embedded in an ever-changing world, and effectively engaging with that world involves constant adaptation and learning of new information and skills. Yet, although learning occurs continuously as we interact with the world, an enormous number of theories in psychology have their roots in experimental approaches and/or data analytic techniques that (at least implicitly) assume humans are unchanging over the course of psychological experiments. This is true for instance, of approaches where performance is aggregated over some number of trials (i.e., where an “average accuracy” or “average reaction time” is taken across trials, whether in “blocks,” “sessions,” “days”, or the full task) as well as in cases where performance is assessed via adaptive procedures meant to find a single value describing participant performance (e.g., staircase procedures to find staircases). In recent work in the lab we have taken to modelling all behavior as continuously changing in response to experience – even in cases where traditional approaches would take a simple average across trials. In this talk, I’ll first describe our general approach for this type of modelling and then highlight some of the key benefits both in terms of describing the behaviors themselves (i.e., that continuous approaches better capture how participants are actually behaving) and in terms of the inferences that can be drawn in three unique contexts: (1) in assessing how perceptual learning generalizes after training; (2) in assessing patterns of individual difference-level correlations across tasks (in particular re-examining known links between working memory task performance and fluid intelligence task performance); and finally, to show the wide-ranging utility of the approach, (3) in assessing performance in a common implicit bias task from social psychology.