Mismatch Receptive Fields in Mouse Visual Cortex

Abstract

In primary visual cortex, a subset of neurons responds when a particular stimulus is encountered in a certain location in visual space. This activity can be modeled using a visual receptive field. In addition to visually driven activity, there are neurons in visual cortex that integrate visual and motor-related input to signal a mismatch between actual and predicted visual flow. Here we show that these mismatch neurons have receptive fields and signal a local mismatch between actual and predicted visual flow in restricted regions of visual space. These mismatch receptive fields are aligned to the retinotopic map of visual cortex and are similar in size to visual receptive fields. Thus, neurons with mismatch receptive fields signal local deviations of actual visual flow from visual flow predicted based on self-motion and could therefore underlie the detection of objects moving relative to the visual flow caused by self-motion. VIDEO ABSTRACT.

Evaluating the neurophysiological evidence for predictive processing as a model of perception

“…Hence, the amplitude of error responses will reflect both differences in top-down predictions and differences in predicted precision. Indeed, as mentioned above, many studies have reported that stimulus-evoked responses are inversely proportional to the degree to which the stimulus was expected, 47,75,76,[83][84][85]98 although this effect is not universally observed. 79,82 In PP, stimulus salience and attention are considered emergent properties of this precisionweighting mechanism.…”

“…Acquiring recordings from distinct cortical layers, whether directly or indirectly, may be essential to overcoming many of these obstacles and adding to what is currently a limited evidence base supporting the existence of functionally distinct subpopulations (e.g., Refs. 47 and 98).…”

“…The V1 neurons targeted by these backward connections have been found to signal the mismatch between predicted and actual sensory input for their corresponding region of the visual field, characteristic of PP error units. 48,98 These predictions do not simply modulate visual responses but appear to drive the activity of V1 neurons even in the absence of bottomup input. 48,136 Moreover, these axonal influences are found to be experience dependent.…”

“…In addition, PP mismatch responses are typically considered to be the product of a simple subtraction (error = actual input – predicted input), but this scheme would appear to necessitate negative firing rates for omitted stimuli or for weaker than expected sensory inputs. In fact, several rodent studies have observed just such a response to an expected but omitted stimulus 47,48,98,123 . Keller and Mrsic‐Flogel 41 accommodated the representation of signed prediction errors by proposing that PP is subserved by two classes of error unit: positive error units (excited by bottom‐up sensory signals and inhibited by top‐down prediction) and negative error units (with the opposite mapping).…”

Opposing influence of top-down and bottom-up input on different types of excitatory layer 2/3 neurons in mouse visual cortex

“…Another idea is that layer 2/3 (L2/3) neurons use a difference between bottom-up visual input and a top-48 down prediction to compute visuomotor prediction errors (Keller and Mrsic-Flogel, 2018). This was based 49 on the finding that a subset of L2/3 neurons strongly respond to a sudden mismatch between visual flow 50 feedback and locomotion speed (Keller et al, 2012;Zmarz and Keller, 2016). This last interpretation is at 51 the core of the framework of predictive processing.…”