Decoding reward-curiosity conflict in decision-making from irrational behaviors.

Area of Science:

• Decision-making models
• Neuroscience
• Computational psychiatry

Background:

• Humans and animals exhibit curiosity-driven exploration beyond rational reward exploitation.
• The underlying mechanisms of this 'irrational' behavior remain largely unexplained.
• Curiosity plays a crucial role in understanding novel environments.

Purpose of the Study:

• To develop a decision-making model based on the free energy principle to explain curiosity-driven behavior.
• To propose a machine learning method for decoding temporal curiosity from behavioral data.
• To investigate the role of curiosity in animal decision-making and its potential clinical applications.

Main Methods:

• Developed a decision-making model for a two-choice task grounded in the free energy principle.
• Integrated recognition and action selection within the model to describe behavior.
• Proposed and applied a machine learning algorithm to decode temporal curiosity from rat behavioral data.

Main Results:

• The model successfully described irrational behaviors influenced by varying curiosity levels.
• Rats exhibited 'negative curiosity,' indicating a preference for more certain options.
• Curiosity levels in rats increased with expected future information from uncertain environments.

Conclusions:

• The developed decoding approach serves as a foundational tool for identifying the neural underpinnings of reward-curiosity conflicts.
• This method holds potential for diagnosing various mental disorders.
• Understanding curiosity mechanisms can advance cognitive science and clinical neuroscience.