Surprising Sounds Influence Risky Decision Making

Adaptive behavior necessitates appropriate responses to environmental uncertainty. Unexpected sensory events, or sensory prediction errors, are salient and can capture attention. While such events might simply be distracting, leading to errors, they could alternatively trigger stereotyped responses regardless of their relevance. This study explores these possibilities by examining whether task-irrelevant auditory prediction errors systematically influence risky decision-making in humans. Decision-making under risk is well-understood, with established models allowing parametric variation of option values independently of trial-to-trial changes. This independence permits separation of the influence of incidental sensory prediction errors from the effect of changing option values, avoiding confounds present in learning tasks. Auditory stimuli are easily manipulated, statistically well-defined, and readily elicit surprise with dissociable neurophysiological correlates. The auditory stimuli used in the experiments carry no information about the probability of risky options leading to wins or losses, ensuring decisions based on these stimuli do not enhance future decision-making. Three possibilities were considered: incidental sounds have no systematic influence, incidental sounds increase decision noise, or sensory prediction errors elicit stereotyped responses (e.g., increased risk-taking and option switching). The latter aligns with evidence linking surprise and adaptive behavior, particularly the role of dopamine in value-neutral sensory information and risk-taking.

Prior research has shown that unpredictable sensory events can impair performance in decision-making tasks. Studies have documented the negative impact of incidental visual distractors, changes in visual salience, acoustic frequency, and stimulus loudness on decision-making. However, the possibility of systematic stereotyped responses to such irrelevant sensory prediction errors, independent of their adaptive function, remained largely unexplored. While some studies reported increased exploration in response to task-relevant uncertainty in human learning tasks, the presence of similar responses to irrelevant uncertainty was unknown. The study builds upon existing research on decision-making under risk, particularly the use of Prospect Theory, and the established link between surprise, dopamine, and risk-taking behavior. Studies using pharmacologically-induced dopamine changes have demonstrated impacts on value-independent risk-taking and choice perseverance, providing a theoretical framework for interpreting the potential effects of auditory surprises.

Seven experiments were conducted using an established risky decision-making paradigm combined with auditory stimuli designed to elicit surprise. In each trial, participants chose between a risky and a safe option with equal probabilities of winning or losing for the risky option. A six-tone auditory sequence preceded option presentation; 75% of trials used a common sequence, while 25% used a rare sequence with a different ending. Experiments varied the spatial arrangement of options and the nature of the rare sequence (deviant or standard tones). The experiments involved a total of 1600 participants recruited online. Participants were paid a flat rate for completion. A pre-screening task was used to ensure participants wore headphones. After the main task, participants completed the PHQ-9 and GAD-7 questionnaires. Data analysis involved computational modeling using Prospect Theory, incorporating parameters for risk aversion, loss aversion, and choice perseverance. Model comparison was used to assess the best explanation for the observed behavioral effects. Ranked-based non-parametric tests were employed for statistical analysis.

Experiments 1 and 2 demonstrated that rare auditory sequences increased risk-taking and decreased choice perseverance compared to common sequences. This effect was consistent across different trial types (Gain, Mixed, Loss) and was not dependent on the valence of the outcomes. Computational modeling revealed that these effects were best explained by a model incorporating value-independent parameters for risk bias and perseverance. Experiments 3 and 4, using rare sequences ending in standard tones, showed a continued decrease in perseverance but eliminated the risk-taking effect, indicating dissociable effects. Experiments 5 and 6, using balanced and fully predictable sequences, eliminated both the risk-taking and perseverance effects. Experiment 7 replicated the main findings, demonstrating that the effects were present even when participants did not believe the auditory cues carried information about reward probabilities. The findings were robust across various experimental manipulations, suggesting that the effects of auditory surprise are not explained by changes in choice stochasticity or lapse rate.

The study's findings demonstrate that task-irrelevant sensory prediction errors, even in the form of incidental sounds, systematically influence risky decision-making. The effects are independent of the value of available options and appear to stem from distinct mechanisms underlying risk-taking and choice perseverance, possibly linked to dissociable neurobiological processes involving dopamine. The robustness of the effects across diverse experimental manipulations highlights their significance. The dissociation between risk-taking and perseverance effects suggests different underlying neural mechanisms. The lack of impact in experiments with balanced sequences indicates that the effects depend on expectation violation. The study provides strong support for the hypothesis that sensory surprise impacts decision-making processes independently of value-based considerations.

This research reveals a significant influence of incidental auditory surprise on risk-taking and choice perseverance in decision-making under uncertainty. The findings highlight the importance of considering task-irrelevant sensory information in understanding human decision-making processes. Future research could explore the underlying neurobiological mechanisms, investigate the generalizability to other sensory modalities and real-world settings, and examine the potential implications for clinical populations.

The study used online participants, potentially limiting the generalizability to other populations. While the experimental design controlled for several confounds, the possibility of other unmeasured factors influencing the results cannot be completely ruled out. The use of auditory stimuli might limit the generalizability to other sensory modalities, although the findings suggest a wider applicability given the presence of similar mechanisms in other sensory domains.