Improving the Reliability of the Pavlovian Go/No-Go Task for Computational Psychiatry Research

Area of Science:

• Cognitive Neuroscience
• Computational Psychiatry
• Psychometric Research

Background:

• The Pavlovian go/no-no task measures individual differences in learning and biases.
• It's used in computational psychiatry but suffers from low reliability.
• Prior research indicated unacceptable reliability for computational model-based measures.

Purpose of the Study:

• To enhance the reliability of the Pavlovian go/no-go task for individual-differences research.
• To improve computational model-based performance measures.
• To provide a reliable tool for computational psychiatry.

Main Methods:

• Two experiments with adult participants (N=103, N=110) using a gamified Pavlovian go/no-go task.
• Hierarchical Bayesian modeling for reinforcement learning model-based indices.
• Modification of the task in Experiment 2 to reduce practice effects.

Main Results:

• Experiment 1 showed significant practice effects and low reliability (0.379).
• Experiment 2, with a modified task, demonstrated reduced practice effects.
• Improved test-retest reliability estimates ranged from 0.696 to 0.989.

Conclusions:

• Model-based measures from the modified Pavlovian go/no-go task are reliable for individual-differences research.
• The task code is provided for the computational psychiatry community.
• Further validation in diverse populations and settings is recommended.