A model-based analysis of impulsivity using a slot-machine gambling paradigm

Area of Science:

• Neuroscience
• Computational Psychiatry
• Behavioral Economics

Background:

• Impulsivity is a key factor in decision-making, particularly in gambling behaviors like problem gambling (PG).
• Existing questionnaire-based impulsivity assessments have limitations due to the multifaceted nature of impulsivity.
• Understanding which specific facets of impulsivity drive gambling behavior is unclear.

Purpose of the Study:

• To investigate how impulsivity manifests in gambling behavior using computational modeling.
• To identify specific decision-making mechanisms linked to individual impulsivity traits in a gambling context.
• To explore the utility of computational models for assessing and potentially preventing problem gambling.

Main Methods:

• Utilized a virtual slot-machine gambling task with 47 healthy male volunteers.
• Measured impulsivity using the Barratt Impulsiveness Scale (BIS-11).
• Applied and compared hierarchical Bayesian belief-updating models, including the Hierarchical Gaussian Filter (HGF) and Rescorla-Wagner reinforcement learning (RL) models, to behavioral data.

Main Results:

• Impulsivity (BIS-11 scores) significantly correlated with gambling responses like bet increases, machine switches, casino switches, and double-ups.
• A three-level HGF model best explained the behavioral data, with parameters reflecting uncertainty-dependent belief updates.
• Decision noise in this model was directly related to trial-wise uncertainty about winning probability, explaining significant variance in BIS-11 scores.

Conclusions:

• Hierarchical Bayesian models can effectively characterize decision-making mechanisms associated with individual impulsive traits.
• Novel computational indices derived from gambling tasks can provide insights into at-risk gambling behavior.
• These findings support the development of computational approaches for online prevention and assessment of pathological gambling.