Related Experiment Videos
Uncovering the 'state': Tracing the hidden state representations that structure learning and decision-making
Angela J Langdon1, Mingyu Song1, Yael Niv1
1Princeton Neuroscience Institute and Department of Psychology, Princeton University, Princeton, NJ, 08544, United States.
Behavioural Processes
|August 6, 2019
Summary
Reinforcement learning agents use internal states to summarize relevant environmental information for decision-making. These states often include past timing and context, influencing behavior in complex ways.
Area of Science:
- Neuroscience
- Cognitive Science
- Artificial Intelligence
Background:
- Reinforcement learning theories propose internal 'state' representations for agents.
- These states summarize environmental features relevant to future behavior and decision-making.
Purpose of the Study:
- To investigate the nature of 'state' representations in learning and decision-making.
- To explore neurobiological and behavioral evidence defining internal states.
Main Methods:
- Review of neurobiological literature.
- Analysis of behavioral findings.
Main Results:
- State representations extend beyond immediate environmental cues.
- They incorporate temporal and contextual information from past experiences.
- This richer information influences decision-making and goal-directed behaviors.
Conclusions:
- Internal states are complex representations crucial for adaptive behavior.
- Understanding these states offers insights into learning and decision-making across biological and artificial agents.