An information theoretic approach to the functional classification of\n neurons

2002/12/31 by Elad Schneidman, William Bialek, Schneidman, Elad +3
Biochemistry, Genetics and Molecular Biology · Neuroscience · #Biological Physics (physics.bio-ph) #Data Analysis #FOS: Biological sciences #FOS: Physical sciences #Neural dynamics and brain function #Neurons and Cognition (q-bio.NC) #Photoreceptor and optogenetics research #Retinal Development and Disorders #Statistics and Probability (physics.data-an)

paper · pdf · doi:10.48550/arxiv.physics/0212114

openalex publication_date 2002/12/31 · openalex created_date 2025/10/24 · openalex updated_date 2026/07/28

Abstract

A population of neurons typically exhibits a broad diversity of responses to\nsensory inputs. The intuitive notion of functional classification is that cells\ncan be clustered so that most of the diversity is captured in the identity of\nthe clusters rather than by individuals within clusters. We show how this\nintuition can be made precise using information theory, without any need to\nintroduce a metric on the space of stimuli or responses. Applied to the retinal\nganglion cells of the salamander, this approach recovers classical results, but\nalso provides clear evidence for subclasses beyond those identified previously.\nFurther, we find that each of the ganglion cells is functionally unique, and\nthat even within the same subclass only a few spikes are needed to reliably\ndistinguish between cells.\n

Citations

• Divergence measures based on the Shannon entropy
• Universality and individuality in a neural code

Related

• Neural and perceptual signatures of efficient sensory coding
• Learning to represent signals spike by spike
• Noise-Robust Modes of the Retinal Population Code have the Geometry of "Ridges" and Correspond with Neuronal Communities
• Revealing structure components of the retina by deep learning networks
• Neuronal identity is not static: An input-driven perspective