Corson Areshenkoff (Queen’s University)

Presenter: Corson Areshenkoff

Title: Subjects will learn your task however they damn well please — Strategy use, non-stationarity, and task-impurity during Human (motor) learning

Abstract:
Goal directed behaviour — even in simple tasks — involves the coordinated activity of multiple cognitive and neural systems, all leading to the motor output measured by the experimenter. We often hope to design tasks which isolate a particular component of these systems (a “pure” measure), but Humans are quite clever, and often succeed in finding ways to make a task easier by leveraging their capacity for sophisticated cognitive control. In this talk, we’ll consider performance in an extremely simple task, designed to study Human sensorimotor learning, and which is classically believed to rely on adaptation within motor structures in the brain. However, there is substantial variability in the way that individual subjects appear to perform this task, and pinning down how subjects actually learn turns out to be quite difficult. In attempting to accomplish this feat, I’ll touch on matters of executive functions, reinforcement learning, and the cognitive systems supporting the representation of the overarching structure of a task, all of which allow subjects to display highly strategic behaviour. Somewhat more polemically, I’ll argue that many tasks which are often considered to measure specific processes actually show evidence of evolving, strategic processes. In this sense, all tasks are measures of executive functions.