A separable two-dimensional random field model of binary response data from multi-day behavioral experiments.

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Biology

Background:

• Studying population learning offers insights into brain changes related to aging, drug effects, and psychiatric conditions.
• Understanding learning dynamics is crucial for interpreting cognitive processes across the lifespan.

Purpose of the Study:

• To introduce a novel separable two-dimensional random field (2D RF) model for analyzing binary response data from multi-day learning experiments.
• To quantify performance variability within and across days and detect abrupt learning changes.

Main Methods:

• A separable 2D random field (RF) model was developed to analyze binary response data from object-reward association learning across multiple days.
• The model was applied to data from young and aged macaque monkeys performing a reversal-learning task.
• Model estimates were used as features for clustering monkeys based on performance.

Main Results:

• Older monkeys required more trials to learn object discriminations and exhibited lower cognitive flexibility than younger monkeys.
• The 2D RF model successfully quantified within-day and across-day performance variability and learning rates.
• Clustering based on model performance estimates largely aligned with age-based groups, indicating captured inter-individual differences.

Conclusions:

• The developed 2D RF model effectively discriminates between-group differences and identifies subgroups in multi-day behavioral experiments.
• This method offers advantages over generalized linear models in capturing the two-dimensional nature of learning data and identifying group distinctions.
• The findings highlight age-related differences in cognitive flexibility and learning efficiency in macaque monkeys.