Correlated Variability in the Neurons With the Strongest Tuning Improves Direction Coding

Area of Science:

• Neuroscience
• Computational Neuroscience
• Sensory Systems

Background:

• Sensory perception relies on neuronal populations to represent external stimuli.
• Information capacity is influenced by individual neuron tuning and shared trial-by-trial variability.
• The relationship between neural correlations and information encoding is complex.

Purpose of the Study:

• To investigate how single-neuron properties and population-level correlations affect neural code efficacy.
• To determine the impact of correlated versus uncorrelated neural populations on encoding stimulus motion direction.

Main Methods:

• Recorded responses from motion-sensitive neurons in marmoset middle temporal area to visual stimuli.
• Trained decoders to assess information encoding in both correlated and uncorrelated neural populations.
• Analyzed single-neuron tuning, variability, and pairwise spike-count correlations.

Main Results:

• In uncorrelated populations, highly responsive, direction-selective, and low-variability neurons are crucial.
• In correlated populations, specific neurons significantly shape shared variability, aiding decoding.
• Temporally stable correlations enhance decoding performance.
• Least variable neurons with strong stimulus representations improve population coding by providing a robust signal and shaping correlations.

Conclusions:

• The efficacy of neural population coding is modulated by both intrinsic neuronal properties and the structure of neural correlations.
• Specific neurons play dual roles: contributing a strong signal and influencing population-level variability.
• Optimizing neural variability through stable correlations can enhance sensory information representation.