Visual Circuits

National Eye Institute

Investigators

Linked publications & trials

Abstract

Introduction Our section investigates several inter-related fundamental brain functions: the selection of relevant visual signals, the active ignoring of irrelevant signals, and the ability to adaptively use visual signals to control appropriate actions. Disruptions of these functions are implicated in a variety of disorders, including attention deficit hyperactivity disorder (ADHD) and autism. Scientists in my section investigate the neuronal circuits involved in these visual functions using a range of techniques, in both non-human primates and mice, in order to understand how these neuronal circuits operate under normal conditions, to identify how breakdowns in these mechanisms cause disorders of sensory-motor coordination, and to uncover how mechanisms of learning and plasticity establish normal function and possible recovery of function. Standard models of these visual functions emphasize the role of the cerebral cortex. In contrast, our results demonstrate that higher-order visual functions are built on top of conserved subcortical circuits in the superior colliculus (SC), thalamus and basal ganglia, that play a central role in action selection. Our aims are to understand the operation of the subcortical structures and how they interact with the cortex, with the long-term goal of identifying the detailed neuronal circuit so that more specific therapeutic interventions can be developed. 1) Neuronal circuits for the control of selective attention in primates Much of our work is done using non-human primates, whose close homology with humans makes them the best animal model of visual function in humans. 1a) Rapid face preference during visual object processing by the primate superior colliculus Face processing is central to primates survival. Neural mechanisms of foveal face processing have been extensively studied, leading to the discovery of face-selective regions in temporal cortex (i.e., face patches). To engage these circuits, primates must first decide to foveate a potential face, a process about which much less is known. Here, we report that the superior colliculus (SC), a midbrain structure involved in target selection and orienting, contains neurons that exhibit preferences for faces at short latencies. In a project completed over the past year, we recorded visually responsive neurons in the superficial and intermediate layers of the SC of two rhesus macaques while they passively viewed images presented in their receptive fields. We used 150 grayscale images of objects belonging to one of five categories previously used to test face processing in the temporal cortex: faces, bodies, hands, fruits/vegetables, and human-made objects. The 30 exemplar images in each category had matched distributions of low-level features (RMS contrast, size, power in three spatial frequency bands). This allowed us to evaluate object selectivity in SC while controlling for low-level visual features. Many SC neurons exhibited a preference for faces within 50ms of stimulus onset, well before neurons in cortical face patches. Based on these short-latency responses, we trained a linear classifier to distinguish faces from other visual objects; the classifier achieved cross-validated accuracies of around 80%. Next, to investigate the visual circuits underlying this short latency face preference, we inactivated the lateral geniculate nucleus, which relays visual signals from the retina to the visual cortex. Surprisingly, this manipulation largely abolished visual responses in SC, including any face-related selectivity. We surmise that the short-latency face preference in SC depends on signals routed through early visual cortex. These results reveal an unexpected circuit in the primate visual system for rapidly detecting and possibly orienting toward, face-like stimuli, complementing the higher-order visual areas needed for recognizing foveated faces. 1b) Correlated variability in primate superior colliculus depends on functional class Correlated variability in neuronal activity (spike count correlations, rSC) can constrain how information is read out from populations of neurons. Traditionally, rSC is reported as a single value summarizing a brain area. However, single values, like summary statistics, stand to obscure underlying features of the constituent elements. We predict that in brain areas containing distinct neuronal subpopulations, different subpopulations will exhibit distinct levels of rSC that are not captured by the population rSC. We tested this idea in macaque superior colliculus (SC), a structure containing several functional classes (i.e., subpopulations) of neurons. We found that during saccade tasks, different functional classes exhibited differing degrees of rSC. Delay class neurons displayed the highest rSC, especially during saccades that relied on working memory. Such dependence of rSC on functional class and cognitive demand underscores the importance of taking functional subpopulations into account when attempting to model or infer population coding principles. 2) Role of subcortical neuronal circuits in visual guided behavior in mice Mice provide opportunities to work out the details of neuronal circuits in ways that are not yet possible in nonhuman primates and will help us identify worthwhile genetic and molecular targets in primates. 2a) Circuits underlying striatal dopamine transmission during visual learning The primary input area of the basal ganglia, the striatum, plays a role in integrating signals from the cortex, midbrain, and thalamus to make associations between stimuli, actions, and rewards. The canonical view has been that midbrain activity drives all dopamine signals in the striatum. However, recent findings have forced the field to reconsider this viewpoint: cortical and thalamic inputs to the striatum can also produce large, local dopamine signals indirectly through striatal cholinergic interneurons. Here we aimed to determine how dopamine, specifically cholinergic-evoked dopamine, may be involved in visual learning. This would reveal a novel mechanism for learning specific associations that is not explicable with current models and would be consistent with recent evidence linking visual cortico-striatal circuits to visual learning. Mice injected with dLight1.2 in the dorsomedial striatum were trained on a unilateral orientation-change detection task requiring them to report a stimulus change by licking a spout. We saw that while all mice had dopamine responses to the stimulus onset (wait cue), only mice that learned that task with a high degree of proficiency developed a dopamine response to the stimulus change (go cue). We then used ex vivo voltammetry to measure electrically and optogenetically-evoked dopamine transients in trained animals. We found striking differences in electrically evoked striatal dopamine release that depended on visual task training, which could be attributed to the cholinergic-evoked component of the overall dopamine signal. Interestingly, visual corticostriatal inputs failed to evoke dopamine transients through cholinergic interneurons in either naive or trained animals. The inability to evoke dopamine signals by visual corticostriatal inputs was explained by a lack of direct synaptic connections onto cholinergic interneurons, as measured with cell-attached electrophysiology. Similarly, somatosensory and auditory cortical inputs could not evoke striatal dopamine for the same reason. However, frontal cortical inputs reliably made direct synaptic connections onto cholinergic interneurons, suggesting that these inputs could serve as potential mediators of dopamine signaling in this striatal sub-region. These findings provide new insights into the role of dopamine and corticostriatal circuits in visual learning and sensory perception.