The neural representation of time: an information-theoretic perspective

Area of Science:

• Neuroscience
• Computational Neuroscience
• Psychophysics

Background:

• Weber's law, a key finding in time perception, describes linear scaling of timing errors with duration.
• Reproducing this scaling is a criterion for validating neurocomputational models of time perception.
• The statistical origin of Weber's law and its deviations is currently unknown.

Purpose of the Study:

• Investigate the statistical origin of Weber's law in time perception using an information-theoretical framework.
• Explore the reasons behind frequently observed deviations from Weber's law.
• Classify existing neurocomputational models based on their temporal information processing mechanisms.

Main Methods:

• Utilized an information-theoretical framework treating neuronal mechanisms of time perception as stochastic processes.
• Assumed the brain computes optimal time estimates.
• Analyzed how estimates based on temporal changes in mean, variance, and correlations affect timing error scaling.
• Examined models involving multiple stochastic processes, including covariance-based and synfire chain models.

Main Results:

• Weber's law holds exactly when time estimates are based on temporal changes in variance.
• Sublinear scaling of timing errors occurs when estimates use systematic changes in the mean (common in many models).
• Superlinear scaling results from estimates based on temporal correlations.
• This hierarchy of temporal information processing is maintained when multiple information sources are available.

Conclusions:

• The study provides a statistical explanation for Weber's law and its deviations in time perception.
• Neurocomputational models can be categorized as mean-, variance-, or correlation-based, predicting distinct timing error scaling.
• This framework offers predictions for the scaling of timing errors in various time perception models.