Integrating Reward Information for Prospective Behavior.

Area of Science:

• Neuroscience
• Cognitive Science
• Decision Science

Background:

• Value-based decision-making often occurs in static environments.
• Real-world decisions require optimizing selection timing for maximum reward.
• Agents may use latent reward tracking or distance-to-goal tracking for timing decisions.

Purpose of the Study:

• To dissociate between latent reward tracking and distance-to-goal tracking strategies.
• To investigate neural mechanisms underlying optimal action timing.
• To demonstrate the feasibility of using EEG and representational similarity analysis (RSA) to identify internal decision variables.

Main Methods:

• Participants performed a novel decision task with dynamically changing latent rewards.
• Electroencephalography (EEG) recorded brain activity during decision-making.
• Representational similarity analysis (RSA) compared EEG patterns to predicted neural signals for each strategy.

Main Results:

• The study successfully dissociated the two timing strategies using RSA.
• EEG signals reflected a distance-to-goal tracking strategy, based on starting value and growth rate.
• Latent reward information could not be independently decoded from the neural data.

Conclusions:

• Noninvasive neural recordings, like EEG, can identify internally computed decision variables.
• Representational similarity analysis is a viable method for distinguishing between cognitive strategies.
• The brain appears to use a distance-to-goal representation for optimizing action timing in dynamic reward scenarios.