Harnessing behavioral diversity to understand neural computations for cognition

Area of Science:

• Neuroscience
• Computational Neuroscience
• Artificial Intelligence

Background:

• Large-scale neural recordings are increasingly common, posing challenges in inferring computations and their link to behavior.
• Existing methods often lack the richness to fully understand the neural basis of complex behaviors.

Purpose of the Study:

• To review advances in conceptual and technological approaches for understanding neural computations and behavior.
• To highlight the synergistic potential of biological and artificial neural networks in this domain.

Main Methods:

• Integrating neural data acquisition with rich behavioral tasks.
• Quantifying animal movements to understand their influence on neural dynamics and brain states.
• Utilizing artificial neural networks (ANNs) as model systems to bridge neural activity and behavior.

Main Results:

• Recording neural data during complex behaviors allows for modeling cognitive processes and estimating latent variables.
• Detailed movement quantification provides insights into neural dynamics and brain state changes (e.g., arousal, stress).
• ANNs demonstrate potential in explaining diverse behaviors and generating hypotheses about neural mechanisms.

Conclusions:

• A multi-faceted approach combining detailed behavioral analysis and ANNs is crucial for deciphering neural computations.
• This integrated strategy can advance our understanding of how neural activity generates behavior and explain behavioral variability.