Final Report Summary - PERCEPT (Cortical circuits of visual perception)
Any sensory experience, such as the perception of a visual scene, relies on the coordinated activity of neurons in sensory areas of cerebral cortex. Sensory cortical areas, including the primary visual cortex (V1), have traditionally been understood as 'feature detectors', meaning that neurons in these areas largely represent the physical properties of a visual stimulus, such as its shape or color. While these neurons undoubtedly represent such stimulus features, it has become clear that their activity also depends on internal states, such as the focus of attention, memories, or expectations. Responses of sensory neurons to an identical stimulus are shaped by its behavioral relevance.
Investigating the neural mechanisms by which behavioral relevance can shape visual sensory processing is challenging for at least two reasons. First, in cerebral cortex a variety of different types of neurons are connected in a highly structured way, forming local cortical circuits. Different elements of these circuits might have different functional roles, which requires making measurements from identified circuit elements. Second, the experimenter needs to control precisely the sensory input and manipulate its behavioral relevance. Addressing these challenges requires an animal model, in which one can target specific circuit elements for measurements while the animal is performing a well-controlled visual perception task.
In this project, we investigated the neural mechanisms underlying visual perception in the mouse model, where we combined rigorous analyses of behavior with electrophysiological measurements and optogenetic manipulations of neural activity. To investigate how V1 sensory processing is shaped by a stimulus’ behavioral relevance, we established three well-controlled perceptual tasks.
In one set of experiments, we focused on learning and asked how the V1 representation of a stimulus changes, as the the behavioral relevance of that stimulus became clear. We trained mice to discriminate two visual stimuli, and precisely quantified, in individual mice, when learning happened. We first confirmed, using optogenetic suppression of V1, that orientation discrimination in this task relied on V1 activity. We then measured, during learning, the neural representation of these stimuli in area V1. We observed learning-related improvements in V1 processing across the depth of cortex, which were fully expressed before discrimination was evident in the animals behavior. Based on these results we proposed that V1 plays a key role early in discrimination learning to enhance behaviorally relevant sensory information.
In a second set of experiments, we adopted for the mouse model a classic backward masking paradigm, where a briefly presented visual stimulus goes unnoticed if it is followed, closely in time, by the presentation of a second stimulus. Combining behavior, electrophysiology, and precisely timed optogenetic suppression of V1 activity, this paradigm allowed us to identify those components of V1 responses that are critical for visual perception.
In a third set of experiments, we asked how reward expectancy can modulate sensory processing in mouse V1. To isolate effects of reward we developed a novel foraging task, in which two stimuli provided identical sensory drive to V1 neurons, but differed in reward contingencies. During task performance, we recorded extracellular activity from many V1 neurons simultaneously across the depth of cortex, and found a substantial number of neurons, for which the processing of the stimulus depended on reward expectancy. To tackle the neural circuitry potentially implementing these effects we used optogenetic tools to identify, in the recorded population, specific types of interneurons. Interneurons have been hypothesized to play a key role in the modulation of sensory processing by cognitive operations, and our data will allow us to provide a direct test of this hypothesis.
Together, these experiments help establish the mouse as a powerful model to study the neural mechanisms underlying visual perception. Precise control over the availability, or the behavioral relevance, of incoming visual information has effects on mouse behavior that are similar to those observed in primates. The behavioral relevance of a visual stimulus can also affect how that stimulus is processed in V1, indicating that, in the mouse, early sensory processing reflects a combination of bottom-up sensory signals with top-down information.
Investigating the neural mechanisms by which behavioral relevance can shape visual sensory processing is challenging for at least two reasons. First, in cerebral cortex a variety of different types of neurons are connected in a highly structured way, forming local cortical circuits. Different elements of these circuits might have different functional roles, which requires making measurements from identified circuit elements. Second, the experimenter needs to control precisely the sensory input and manipulate its behavioral relevance. Addressing these challenges requires an animal model, in which one can target specific circuit elements for measurements while the animal is performing a well-controlled visual perception task.
In this project, we investigated the neural mechanisms underlying visual perception in the mouse model, where we combined rigorous analyses of behavior with electrophysiological measurements and optogenetic manipulations of neural activity. To investigate how V1 sensory processing is shaped by a stimulus’ behavioral relevance, we established three well-controlled perceptual tasks.
In one set of experiments, we focused on learning and asked how the V1 representation of a stimulus changes, as the the behavioral relevance of that stimulus became clear. We trained mice to discriminate two visual stimuli, and precisely quantified, in individual mice, when learning happened. We first confirmed, using optogenetic suppression of V1, that orientation discrimination in this task relied on V1 activity. We then measured, during learning, the neural representation of these stimuli in area V1. We observed learning-related improvements in V1 processing across the depth of cortex, which were fully expressed before discrimination was evident in the animals behavior. Based on these results we proposed that V1 plays a key role early in discrimination learning to enhance behaviorally relevant sensory information.
In a second set of experiments, we adopted for the mouse model a classic backward masking paradigm, where a briefly presented visual stimulus goes unnoticed if it is followed, closely in time, by the presentation of a second stimulus. Combining behavior, electrophysiology, and precisely timed optogenetic suppression of V1 activity, this paradigm allowed us to identify those components of V1 responses that are critical for visual perception.
In a third set of experiments, we asked how reward expectancy can modulate sensory processing in mouse V1. To isolate effects of reward we developed a novel foraging task, in which two stimuli provided identical sensory drive to V1 neurons, but differed in reward contingencies. During task performance, we recorded extracellular activity from many V1 neurons simultaneously across the depth of cortex, and found a substantial number of neurons, for which the processing of the stimulus depended on reward expectancy. To tackle the neural circuitry potentially implementing these effects we used optogenetic tools to identify, in the recorded population, specific types of interneurons. Interneurons have been hypothesized to play a key role in the modulation of sensory processing by cognitive operations, and our data will allow us to provide a direct test of this hypothesis.
Together, these experiments help establish the mouse as a powerful model to study the neural mechanisms underlying visual perception. Precise control over the availability, or the behavioral relevance, of incoming visual information has effects on mouse behavior that are similar to those observed in primates. The behavioral relevance of a visual stimulus can also affect how that stimulus is processed in V1, indicating that, in the mouse, early sensory processing reflects a combination of bottom-up sensory signals with top-down information.