Decision-making dynamics are predicted by arousal and uninstructed movements

Area of Science:

• Neuroscience
• Animal Behavior
• Computational Modeling

Background:

• Animal decision-making exhibits dynamic performance states despite stable environmental conditions.
• Factors influencing these performance fluctuations remain poorly understood.
• Externally observable measures like movement and arousal are hypothesized predictors.

Purpose of the Study:

• To investigate the relationship between uninstructed behaviors, arousal, and task performance states in mice.
• To determine if external measures can predict transitions between optimal and sub-optimal performance.

Main Methods:

• Computational modeling of visual and auditory task performance data from mice.
• Application of hidden Markov models to analyze behavioral choices.
• Analysis of pupil diameter and movement as indicators of arousal and engagement.

Main Results:

• Mice transition between distinct, minutes-long performance states: optimal, sub-optimal, and disengaged.
• Optimal performance epochs correlate with intermediate arousal levels and reduced movement variability.
• Uninstructed movements and arousal levels predict optimal decision-making states.

Conclusions:

• Externally observable behaviors, including uninstructed movements and arousal, are reliable predictors of optimal task performance.
• Mice appear to actively regulate their arousal levels during periods of optimal performance.
• These findings offer insights into the neural and behavioral underpinnings of dynamic decision-making.