Feature selectivity can explain mismatch signals in mouse visual cortex.

Tomaso Muzzu,Aman B. Saleem

doi:10.1016/j.celrep.2021.109772

Abstract

SummarySensory experience often depends on one’s own actions, including self-motion. Theories of predictive coding postulate that actions are regulated by calculating prediction error, which is the difference between sensory experience and expectation based on self-generated actions. Signals consistent with prediction error have been reported in the mouse visual cortex (V1) when visual flow coupled to running was unexpectedly stopped. Here, we show that such signals can be elicited by visual stimuli uncoupled to an animal running. We record V1 neurons while presenting drifting gratings that unexpectedly stop. We find strong responses to visual perturbations, which are enhanced during running. Perturbation responses are strongest in the preferred orientation of individual neurons, and perturbation-responsive neurons are more likely to prefer slow visual speeds. Our results indicate that prediction error signals can be explained by the convergence of known motor and sensory signals, providing a purely sensory and motor explanation for purported mismatch signals.

Highlights

Sensation and action are two intertwined processes that the brain continuously executes and adjusts (Claxton, 1975; Wolpert et al, 1995; Rao and Ballard, 1999; Koster-Hale and Saxe, 2013; Friston, 2018)
It has been suggested that errors in the prediction of visual flow are encoded as early as in the primary visual cortex (Keller et al, 2012; Keller and Mrsic-Flogel, 2018), based on large responses to sudden stops of visual flow that were normally coupled to an animal running
In agreement with theories of predictive coding, such activity would provide the mouse visual cortex (V1) with the ability to encode the error between the actual sensory feedback and the expected one, i.e., a visuomotor mismatch signal (Keller et al, 2012)

Summary

Introduction

Sensation and action are two intertwined processes that the brain continuously executes and adjusts (Claxton, 1975; Wolpert et al, 1995; Rao and Ballard, 1999; Koster-Hale and Saxe, 2013; Friston, 2018). Theories of predictive coding postulate that sensation is an active process that uses information about one’s own actions to distinguish between self-generated and external sensory stimuli. One feature of such predictive coding is the computation of prediction error—the difference between observed features and those expected based on one’s own actions. In agreement with theories of predictive coding, such activity would provide the mouse visual cortex (V1) with the ability to encode the error between the actual sensory feedback and the expected one, i.e., a visuomotor mismatch signal (Keller et al, 2012). An alternate and untested hypothesis is that responses to sudden stops of visual flow are due to the convergence of motor and visual inputs and do not arise from the precise coupling between an animal’s actions and the visual stimulus

Methods

Results

Discussion

Conclusion