Keynotes | K&Ts | GACs | Talks | Posters | Search

Poster C in Poster Session C: Wednesday, August 5, 9:30 – 11:15 am, Kimmel Center, Shorin & Rosenthal Rooms

Belief Updating as Emotional Value-Based Decision-Making

Kanji Shimomura^1,2, Kenji Morita¹, Yuichi Yamashita³; ¹The University of Tokyo, ²Tokyo Institute of Technology, ³National Center of Neurology and Psychiatry

Presenter: Kanji Shimomura

Humans tend to incorporate desirable information more than undesirable information when updating their self-related beliefs, a phenomenon known as positive bias. However, recent studies have reported the opposite, negative, bias in tasks involving repeated feedback. Existing accounts typically describe these biases as asymmetries in static learning rates, leaving their underlying mechanisms unclear. Here we focus on the possibility that the number of feedback opportunities influences the direction of bias and propose a novel computational framework that provides a unified account of the two biases from the perspective of emotional utility maximization. The core idea is that agents update beliefs to maximize the combined emotional utility of current beliefs and future prediction errors. Simulations show that the model reproduces both positive bias in one-shot settings and negative bias in repeated settings. These findings suggest that belief updating biases reflect rational trade-offs between immediate and future emotional outcomes.

Topic Area: Decision-Making, Cognitive Control & Event Cognition