Objective
A prominent question has emerged about how individual differences shape the way decision results drive subsequent risk-taking behavior. Sensation seeking (SS) and reward sensitivity (RS) are important pathological personalities for behavioral disorders such as gamble and material addictive disorders. However, previous studies have shown behavioral heterogeneity is a pervasive feature of risk-taking and decision-making, yet a neural trait approach can at least partially explain the heterogeneity in behavior by stable brain-based characteristics of individuals. Hence, to study their impacts on reward-driven risk-taking behaviors, we combined the Reinforcement Learning (RL) model and the neural measure of a dynamic risky decision task to explore the relationship between SS, RS, and risk adjustment (RA) to rewards.
Methods
A task characterized by the unknown but ordered risk was designed to quantify the RA with the RL model and adapted from the Balloon Analog Risk Task. In Study 1, 43 young participants completed the task; in Study 2, 37 young participants finished the task while wearing an electroencephalography device. The recorded behavioral data and EEG signal were analyzed using the computational model, event-related potentials and spectral perturbations analysis, and bayesian multi-model linear regression.
Results
Results of Study 1 showed the choice deviations were larger in the higher SS participants with a lower level of RS. Meanwhile, results from Event-related potential and Time-frequency analysis of Study 2 showed higher SS participants were less sensitive to the reward feedback. Based on Study 1 and Study 2, Bayesian multi-model linear regression showed the saliently direct effect of RS on RA and the moderating effect of SS.
Conclusions
SS might indirectly relate to RA through RS. In conclusion, RS impacts the entire process of reward prediction and learning and is vital for intervening in risky behaviors, especially in individuals with high SS.