Mathematical Model of Emotions Evoked by Novelty

Abstract

I introduce our mathematical model of emotions evoked by novel events. The core idea is to use information contents processed in the brain through experiencing an event to represent novelty and its emotions. The model explains human surprise of novelty as Kullback–Leibler divergence of Bayesian posterior and prior, i.e. information gain. We formalized habituation to novelty as decrement of the information gain through repeated exposure of the same event and demonstrate how initial condition of uncertainty affects speed of the habituation. Furthermore, I consider information theoretic free-energy as arousal potential decomposed into two terms: novelty and perceived complexity. As the model predicted, an empirical study shows that emotional valence (positive-negative) shapes as an inverse-U function of summation of novelty and perceived complexity.