Attention Decorrelates Sensory and Motor Signals in the Mouse Visual Cortex

Abstract

Perception is an active process involving continuous interactions with the environment. During such 1 interactions neural signals called corollary discharges (CDs) propagate across multiple brain regions 2 informing the animal whether itself or the world is moving. How the interactions between concurrent 3CDs affect the large-scale network dynamics, and in turn help shape sensory perception is currently 4 unknown. We focused on the effect of saccadic and body-movement CDs on a network of visual 5 cortical areas in adult mice. CDs alone had large amplitudes, 3-4 times larger than visual responses, 6 and could be dynamically described as standing waves. They spread broadly, with peak activations in 7 the medial and anterior parts of the dorsal visual stream. Inhibition mirrored the wave-like dynamics 8 of excitation, suggesting these networks remained E/I balanced. CD waves superimposed sub-linearly 9and asymmetrically: the suppression was larger if a saccade followed a body movement than in the 10 reverse order. These rules depended on the animal's cognitive state: when the animal was most 11 engaged in a visual discrimination task, cortical states had large variability accompanied by increased 12 reliability in sensory processing and a smaller non-linearity. Modeling results suggest these states 13 permit independent encoding of CDs and sensory signals and efficient read-out by downstream 14 networks for improved visual perception. In summary, our results highlight a novel cognitive-15 dependent arithmetic for the interaction of non-visual signals that dominate the activity of occipital 16 cortical networks during goal-oriented behaviors. These findings provide an experimental and 17 theoretical foundation for the study of active visual perception in ethological conditions. 18

Scaling of information in large neural populations reveals signatures of information-limiting correlations

“…These lower discrimination thresholds predict increasingly small choice correlations, in line with recent reports from area V1 of monkeys, where fewer than 7% of V1 neurons were found to feature significant choice correlations (Jasper, Tanabe, & Kohn, 2019). In general, the estimated asymptotic information predicted direction discrimination thresholds compatible with previous behavioral reports in mice (Abdolrahmani et al, 2019;Glickfeld et al, 2013), but the use of different stimuli in these experiments precludes a direct quantitative comparison. A more detailed analysis of the relation between neural activity and choice would require training animals to report their percepts, and then relating these reports to population activity fluctuations.…”

“…To compare our required population size estimates to the total number of neurons in mouse V1, we conservatively estimated the need for about 48,000 neurons (see Methods) to achieve drift direction discrimination performance that most likely exceeds that of the animals (Abdolrahmani, Lyamzin, Aoki, & Benucci, 2019;Glickfeld, Histed, & Maunsell, 2013). Our use of time-deconvolved calcium activity as a noisy proxy for spike counts (T.-W. Chen et al, 2013;Ledochowitsch et al, 2019) makes these estimates upper bounds on required population sizes (see SI).…”

“…To compare the estimated population sizes to the number of neurons in V1, we asked for the number of neurons required to encode 95% of the asymptotic information associated with a direction discrimination threshold of 1°. This threshold most likely exceeds the behavioral performance that mice can reach even for high contrast stimuli (Abdolrahmani et al, 2019;Glickfeld et al, 2013) and thus provides an upper bound on the required population size. Achieving such a low threshold requires an asymptotic information of 4651 rad -2 (Fig.…”

Distributed context-dependent choice information in mouse dorsal-parietal cortex

“…The state axis defined by attentional modulations remained stable throughout the duration of the trial (Extended Data Fig. 2), consistent with periods of high and low attention that persisted across trials [35]. The anterior-medial visual areas and the retrosplenial cortex contributed most significantly to large d discriminability (Fig.…”

“…3), as expected from the lack of orientation domains in mouse visual cortex [34] and the mesoscale spatial resolution of our imaging system. Besides bottom-up visual inputs, imaged dorsal-parietal regions reflected activations associated with general movements of the body and eyes [35]. Therefore, we defined state axes associated with wheel and eye movements.…”