Learning decouples accuracy and reaction time for rapid decisions in a transitive inference task.

Area of Science:

• Cognitive Neuroscience
• Computational Neuroscience
• Animal Behavior

Background:

• The drift diffusion model (DDM) is widely used for decision-making.
• Its application to rapid serial learning with atypical response patterns requires adaptation.
• Latent variables in decision-making during learning are not fully understood.

Purpose of the Study:

• To adapt the drift diffusion model (DDM) for serial learning tasks.
• To investigate the role of latent variables in rapid decision-making.
• To model macaque monkey behavior in a transitive inference transfer task.

Main Methods:

• Behavioral data from three macaque monkeys were fitted using the PyDDM software.
• Monkeys were trained on a transitive inference task involving learning the order of novel picture lists.
• Saccadic eye movements were used to indicate choices.

Main Results:

• Monkeys learned list orders within 200-300 trials, achieving 80-90% accuracy.
• A symbolic distance effect was observed, with accuracy varying based on item proximity in the learned list.
• Reaction times remained invariant despite learning and accuracy improvements.
• The generalized drift-diffusion model successfully fit both accuracy and reaction time data by adjusting evidence accumulation and a collapsing bound.

Conclusions:

• Decision-making in transitive inference (TI) tasks during learning and transfer can be described by a "variable collapsing bound" DDM.
• This suggests a unique dynamic regime within the DDM framework for serial learning.
• The findings offer insights into the neural mechanisms underlying learning and decision-making.