Abstract
The context in which decisions are learned can influence what choices we prefer in new situations. Such dependencies are well replicated and often lead to decision biases, i.e. choices that deviate from rational choice theory. We propose a simple computational model of such biases. To test the model, we analyzed behavioral data from 351 male and female participants in a series of nine value-based decision tasks and re-analyze six previously published datasets (n = 350 participants). Our results show that the combination of two basic principles, learning by reward, and repetition of decisions, is sufficient to explain biased preferences across all 15 datasets. Using standard analysis and hierarchical Bayesian model comparison we found that the proposed model provides a better explanation than previous accounts. In addition, our results show that higher choice frequency is linked to higher subjective valuation and lower value uncertainty. Results indicate that repetition is an important mechanism in shaping preferences.
Significance Statement Human decision-making often deviates from rational choice theory. Understanding how decision biases emerge is essential for interpreting real-world human behavior. Our study shows how repeating decisions can shape preference in novel situations. We adapt a simple computational model that can explain biased preferences based on reward learning and decision repetition. By testing the model across 15 datasets, we demonstrate that it outperforms previous models, also offering new insights into how preferences affect subjective valuation and uncertainty. In sum, these findings provide a deeper understanding of how context-dependent preferences emerge and persist.
Competing Interest Statement
The authors have declared no competing interest.