# Sleep restriction increases reward sensitivity during sequential updating

## TL;DR

One week of sleep restriction makes people more sensitive to rewards and less consistent in decision-making, which could affect performance in high-stakes environments.

## Contribution

The study reveals that sleep restriction increases reward sensitivity and choice randomness during learning from feedback, using computational models.

## Key findings

- Sleep restriction increased sensitivity to reward feedback and choice stochasticity.
- Trial accuracy remained unchanged between sleep conditions.
- Findings suggest possible dose-dependent effects of sleep restriction on decision-making.

## Abstract

Chronic sleep restriction is a common phenomenon in the real world, yet its impact on how individuals learn and adapt to dynamic patterns of reward in uncertain environments are not well understood. The present study examined how sleep restriction alters the cognitive mechanisms underlying sequential updating, defined as the process by which individuals revise beliefs over time in response to outcome feedback. Thirty-six healthy adults completed two conditions in counterbalanced order: well-rested (seven nights of 9 h time-in-bed) and sleep restriction (seven nights of 5 h time-in-bed). On the final day of each condition, participants performed a probabilistic reversal learning task, requiring flexible behavioral adjustments in response to changing reward contingencies. Computational models grounded in reinforcement learning and Bayesian frameworks were fitted to trial-level data using hierarchical Bayesian estimation. Our model comparison indicated the Asymmetric Rescorla–Wagner model best accounted for behavior, indicating feedback-driven updating with distinct learning rates for reward and nonreward outcomes on the probabilistic reversal learning task. Importantly, sleep restriction increased sensitivity to reward feedback and choice stochasticity, indicating greater reward-chasing behavior and less goal-directed decision making under sleep restriction. Trial accuracy did not differ between conditions. These findings diverge from previous evidence indicating reduced sensitivity toward negative outcomes following short-term sleep restriction, pointing to possible dose-dependent sleep restriction effects and those arising from differences in task requirements. Our findings may have practical implications for decision quality in real-world domains where performance depends on optimal feedback-based updating under uncertain decision environments.

Statement of SignificanceWhile sleep loss is known to affect decision making, little is understood about how it impacts people’s ability to learn from experience over time. In many real-world situations, like financial trading and medical contexts, optimized decisions rely on updating beliefs and strategies based on incoming feedback. In this study, we show after one week of sleep restriction, individuals became more influenced by rewards and demonstrated increased choice randomness. This suggests they were more likely to chase immediate gains but had difficulty sustaining consistent decision strategies. These findings help explain how sleep restriction can undermine learning and flexible decision making when faced with uncertainty, and carry implications for performance and accountability in roles where repeated, high-stakes decisions are required.

While sleep loss is known to affect decision making, little is understood about how it impacts people’s ability to learn from experience over time. In many real-world situations, like financial trading and medical contexts, optimized decisions rely on updating beliefs and strategies based on incoming feedback. In this study, we show after one week of sleep restriction, individuals became more influenced by rewards and demonstrated increased choice randomness. This suggests they were more likely to chase immediate gains but had difficulty sustaining consistent decision strategies. These findings help explain how sleep restriction can undermine learning and flexible decision making when faced with uncertainty, and carry implications for performance and accountability in roles where repeated, high-stakes decisions are required.

Graphical Abstract