Interpretation of individual differences in computational neuroscience using a latent input approach

Area of Science:

• Computational Neuroscience
• Cognitive Neuroscience
• Neuroscience of Behavior

Background:

• Computational neuroscience models neural mechanisms of behavior.
• Interpreting individual differences, like developmental variations, in these mechanisms is challenging.
• Reinforcement learning studies often use computational models to link behavior to neural activity.

Purpose of the Study:

• To address the complexities in interpreting individual differences in computational neuroscience, particularly in reinforcement learning.
• To clarify the origins of observed individual differences in neural coding and brain resource utilization.
• To provide recommendations for advancing the understanding and interpretation of individual differences in computational neuroscience.

Main Methods:

• Utilized computational modeling to generate individual-specific prediction error regressors.
• Analyzed individual differences in regression weights, typically interpreted as neural coding differences.
• Conducted simulations to explore alternative explanations for observed individual differences.

Main Results:

• Demonstrated that the absence of individual differences in neural coding is not problematic, as these are captured in the individual-specific regressor.
• Illustrated through simulations that individual differences can arise from factors beyond neural coding, including prediction error standardization, network effects, response duration, outcome valuation, and model parameters.
• Showed that observed individual differences are not solely indicative of differences in brain resource use.

Conclusions:

• The interpretation of individual differences in computational neuroscience requires careful consideration of multiple factors.
• Recommendations are provided to enhance the accurate interpretation of individual differences in computational neuroscience research.
• Advancing the field necessitates a nuanced approach to understanding individual variations in neural mechanisms.