Firing rates in visual cortex show representational drift, while temporal spike sequences remain stable.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Systems Neuroscience

Background:

• Neural responses to stimuli change over time, posing challenges for stable brain coding.
• Existing models often focus on firing rates, potentially overlooking other neural coding mechanisms.

Purpose of the Study:

• To investigate whether spike-timing codes offer a more stable neural representation than firing-rate codes.
• To compare the information content and stability of spike-rate versus spike-timing codes in visual cortical areas.

Main Methods:

• Utilized SpikeShip, a novel method based on optimal transport theory, to analyze information in multi-neuron spike sequences.
• Recorded neural ensemble activity from six visual areas in response to natural video stimuli.
• Quantified information conveyed by population firing-rate vectors and relative spike-timing patterns.

Main Results:

• Temporal spike sequences conveyed more information than population firing-rate vectors in large neural ensembles.
• Firing-rate codes showed significant drift across stimulus repetitions and experimental blocks.
• Spike-timing codes demonstrated remarkable stability over time, unlike firing-rate representations.

Conclusions:

• Relative spike-timing relations among neurons form a stable neural code for sensory information.
• This spike-timing code provides a robust mechanism for neural information processing, overcoming the instability of firing-rate codes.
• Findings highlight the importance of temporal dynamics in high-dimensional neural ensembles for stable brain function.