High-dimensional geometry of population responses in visual cortex

By C. Stringer, Marius Pachitariu, Nicholas A. Steinmetz, Matteo Carandini, Kenneth D. Harris. 2018

647 citations recorded by the source corpus.

A neuronal population encodes information most efficiently when its stimulus responses are high-dimensional and uncorrelated, and most robustly when they are lower-dimensional and correlated. Here we analysed the dimensionality of the encoding of natural images by large populations of neurons in the visual cortex of awake mice. The evoked population activity was high-dimensional, and correlations obeyed an unexpected power law: the nth principal component variance scaled as 1/n. This scaling was not inherited from the power law spectrum of natural images, because it persisted after stimulus whitening. We proved mathematically that if the variance spectrum was to decay more slowly then the population code could not be smooth, allowing small changes in input to dominate population activity. The theory also predicts larger power-law exponents for lower-dimensional stimulus ensembles, which we validated experimentally. These results suggest that coding smoothness may represent a fundamental constraint that determines correlations in neural population codes. Analysis of the encoding of natural images by very large populations of neurons in the visual cortex of awake mice characterizes the high dimensional geometry of the neural responses.

Subfields: Cellular and Molecular Neuroscience, Cognitive Neuroscience

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Cellular and Molecular Neuroscience

In Cognitive Neuroscience

Citations in the corpus

Re-evaluating circuit mechanisms underlying pattern separation (2019, positive): “Moreover, recent theoretical advances may help estimate the dimensionality from noisy neural data (Machens et al., 2010; Stringer et al., 2018), or based on the local connectivity motifs of neural circuits (Recanatesi et al., 2018).”
Neuropixels 2.0: A miniaturized high-density probe for stable, long-term brain recordings (2020, background): “…et al., 2018; Vélez-Fort et al., 2018; Bennett et al., 2019; Kostadinov et al., 2019; Musall et al., 2019; Park et al., 2019; Schröder et al., 2019; Stringer et al., 2019a; Liu et al., 2020; Sauerbrei et al., 2020), rats (Krupic et al., 2018; Gardner et al., 2019; Böhm and Lee, 2020; Luo et al.,…”
Sequential and efficient neural-population coding of complex task information (2019, neutral): “In addition, we identified thirteen variables that spanned execution and psychophysics of the task: (1&2) the running tally #ipsi and #contra of cue-counts on the sides ipsilateral and contralateral to the recorded brain hemisphere; (3&4) the final tally of ipsi/contra cue counts from the previous t”
Interpreting neural computations by examining intrinsic and embedding dimensionality of neural activity (2021, background): “However, embedding information in arbitrarily high dimensional subspaces can have adverse effects for generalization [64].”
Manifold GPLVMs for discovering non-Euclidean latent structure in neural data (2020, background): “Most models in neuroscience justiﬁably assume that neurons are smoothly tuned (Stringer et al., 2019).”
Low-dimensional dynamics for working memory and time encoding (2020, background): “67–69) or as high-dimensional as it could be ( 70 Lowdimensionality allows for better generalization, but this comes at a cost.”
Intrinsic dimension of data representations in deep neural networks (2019, background): “This result is qualitatively consistent with those obtained for stimulus-representations in the primary visual cortex [19].”
The representation of abstract goals in working memory is supported by task-congruent neural geometry (2024, unknown): “For example, when naturalistic objects or pictures were used, the neural responses in the visual cortex can be high-dimensional [49].”