On the Optimal Temporal Resolution for Information Representation in Neural Activity: A Theoretical Analysis

Area of Science:

• Neuroscience
• Computational Neuroscience
• Theoretical Neuroscience

Background:

• Neural activity is organized across temporal and spatial scales, but principles of information representation fidelity are unclear.
• Mesoscopic optimality in neural decoding has been observed empirically, yet lacks theoretical explanation.

Purpose of the Study:

• Develop a theoretical framework to understand optimal temporal scales for neural information representation.
• Investigate the dependence of representational accuracy on signal and noise dynamics.

Main Methods:

• Formulated a multiscale theoretical model of neural population activity.
• Modeled neural responses with temporally correlated signal and noise.
• Quantified representational quality using the sensitivity index (d-prime) at microscopic, mesoscopic, and macroscopic resolutions.

Main Results:

• Derived closed-form expressions for the sensitivity index, highlighting signal and noise correlations as key determinants.
• Identified two regimes: extreme scales (micro/macro) are optimal when correlations are absent/persistent.
• Demonstrated that mesoscopic scales emerge as optimal when signal and noise correlations decay, due to a trade-off in temporal integration.

Conclusions:

• Provided a theoretical explanation for optimal temporal scales in neural representation based on signal and noise correlation dynamics.
• Established temporal integration as a mechanism linking neural dynamics to information representation.
• Offered testable predictions for various neural systems and recording modalities.