Decision criterion dynamics in animals performing an auditory detection task

Area of Science:

• Neuroscience
• Cognitive Science
• Animal Behavior

Background:

• Classical signal detection theory posits bias arises from a fixed threshold criterion.
• Optimal criterion setting is challenged by time-varying signal levels and subject naivety.
• Subjects must adapt thresholds based on feedback to optimize performance.

Purpose of the Study:

• To investigate how decision bias in ferrets tracks changing auditory stimulus levels.
• To explore the time scales of criterion dynamics in a signal detection task.
• To model the underlying policy governing bias adjustments.

Main Methods:

• Conducted a yes-no auditory signal-in-noise detection task with ferrets.
• Grouped trials into blocks with alternating easy and hard-to-detect signals.
• Applied a computational model to analyze criterion dynamics and bias changes.

Main Results:

• Ferrets exhibited both short-term (trial-to-trial) and long-term (block-level) criterion dynamics.
• Bias shifted towards 'yes' for harder trials and away from 'yes' for easier trials.
• A single criterion update policy, incorporating trial outcomes and decay to a resting value, explained both dynamics.

Conclusions:

• Decision bias dynamically adapts to changing stimulus statistics via a unified update policy.
• Block-level bias stabilization emerges from the interplay of trial outcomes and shifts.
• The framework offers insights into decision dynamics across species and psychological domains.