A Computational Analysis of the Effect of Hard Choices on the Individuation of Values

Experimental studies show that when an individual makes choices, they affect future decisions. Future choices tend to be consistent with past ones. This tendency matters in the context of ambivalent situations because they may not lead to clear choices, often leading people to make "arbitrary" decisions. Thus, because of choice consistency with the past, people's decision-making values diverge. Thus, hard choices may contribute to the individuation of values.

Here, we develop a Bayesian framework for the effects of cognitive choice consistency on decision-making. This framework thus extends earlier cognitive-science Bayesian theories, which focus on other tasks, such as inference. The minimization of total surprisals considering the history of stimuli and chosen actions implements choice consistency in our framework. We then use a computational model based on this framework to study the effect of hard choices on decision-making values.

The results for action selection based on sensory stimuli show that hard choices can cause the spontaneous symmetry breaking of the decision-making space. This spontaneous symmetry breaking is different across individuals, leading to individuation. If in addition, rewards are given to certain choices, then the direction of the symmetry breaking can be guided by these incentives. Finally, we explore the effects of the parametric complexity of the model, the number of choices, and the length of choice memory.

Considering the brain's mechanism of choice consistency and the number of hard choices made in life, we hypothesize that they contribute to individuality. We assess this hypothesis by placing our study in the context of the cognition-of-individuality literature and proposing experimental tests of our computational results.