Extracting the dynamics of behavior in sensory decision-making experiments

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Biology

Background:

• Decision-making strategies evolve during training and can vary even in well-trained animals.
• Traditional studies often use fixed psychometric functions, fitted only after training completion.
• This approach overlooks the dynamic nature of strategy evolution during learning.

Purpose of the Study:

• To introduce PsyTrack, a novel, flexible method for inferring decision-making strategy trajectories from choice data.
• To analyze trial-to-trial fluctuations in stimulus weighting, bias, and task-irrelevant covariates.
• To quantify time-varying behavior in diverse animal models and human subjects.

Main Methods:

• PsyTrack was developed to analyze behavioral choice data.
• The method was applied to training data from mice, rats, and human subjects performing auditory and visual decision-making tasks.
• PsyTrack infers the trajectory of decision-making strategies, capturing dynamic changes.

Main Results:

• PsyTrack successfully captured trial-to-trial fluctuations in sensory stimulus weighting and bias.
• The method identified significant differences in learning rates across mouse subjects.
• Rapid adaptation to changing task statistics was observed, highlighting behavioral flexibility.

Conclusions:

• PsyTrack offers a powerful tool for quantifying time-varying decision-making strategies.
• The method reveals dynamic behavioral changes and individual learning differences.
• PsyTrack is scalable and applicable to a wide range of animals and tasks.