Variations in neuronal selectivity create efficient representational geometries for perception

Area of Science:

• Neuroscience
• Computational Neuroscience
• Visual Perception

Background:

• Neural response properties in visual areas underpin visual capabilities.
• The diversity of neuronal tuning properties and its functional significance remain largely unexplained.
• Understanding neuronal diversity is key to understanding visual perception.

Purpose of the Study:

• To investigate how neuronal response diversity influences the information capacity of neural populations.
• To determine the impact of amplitude and bandwidth diversity on representational geometry and perceptual tasks.
• To elucidate the role of neuronal diversity as a fundamental neural coding strategy.

Main Methods:

• Theoretical modeling of neuronal population responses.
• Computer simulations to test theoretical predictions.
• Analysis of electrophysiological recordings from macaque primary visual cortex (V1).

Main Results:

• Amplitude and bandwidth diversity induce complementary changes in representational geometry.
• Amplitude diversity separates representations, enhancing encoding efficiency.
• Bandwidth diversity decorrelates representations, further improving encoding.
• Both diversity types enhance performance on discrimination and identification tasks, with complementary benefits.

Conclusions:

• Neuronal tuning diversity is crucial for efficient neural coding and perception.
• Amplitude and bandwidth diversity offer distinct advantages for different perceptual tasks.
• These findings suggest a fundamental neural coding strategy applicable across species and brain areas.