Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session C: Friday, August 15, 2:00 – 5:00 pm, de Brug & E‑Hall

Revisiting Cost Functions in Sensorimotor Decision-Making

Tobias F. Niehues¹, Dominik Straub², Constantin A. Rothkopf¹; ¹Technische Universität Darmstadt, ²University of Cambridge

Presenter: Tobias F. Niehues

Human decision-making in sensorimotor tasks is characterized by perceptual uncertainty, motor variability, prior beliefs, and the goal of a task, as well as by other factors like the effort required to act. Distributions and costs in these tasks are usually assumed to be normally distributed or of quadratic shape to maintain analytical tractability for mathematical convenience. However, there are no guarantees whether these assumptions correctly represent the functional relations underlying human behavior. Recent work on inverse decision-making makes it possible to overcome these limitations while still allowing inference of behavioral parameters from data with arbitrary cost functions and sensory encoding. Here, we extended this approach to a hierarchical model, thereby allowing model comparison at the task level instead of a per-subject level. In all data sets, asymmetric cost functions describe human behavior better than quadratic costs, and in four out of five cases, the cost function contains explicit effort costs, contrary to previous investigations.

Topic Area: Predictive Processing & Cognitive Control

Extended Abstract: Full Text PDF