Search Papers | Poster Sessions | All Posters
Poster C147 in Poster Session C - Friday, August 9, 2024, 11:15 am – 1:15 pm, Johnson Ice Rink
A computational account optimism and pessimism as means of self-control
Boaz Rosenberg1 (), Eran Eldar1; 1The Hebrew University of Jerusalem
From a decision-making perspective, self-deception doesn’t seem to be a beneficial behavioral characteristic. Nevertheless, in this article, we demonstrate how, under certain viable constraints, agents learn that biasing their perceived state can be advantageous. This demonstration is based on recognizing the value of self-control in cases where cached and prospection-based evaluations result in conflicting action selection. When this conflict arises from evaluations of resource states, self-biasing could serve as an effective means of self-control, diminishing the influence of potentially incorrect cached values, and thus improving decision-making processes.
Keywords: Model-Based Cached value Biased perception Reinforcement Learning