Distributed Neural Activity Patterns Underlying Practice-Based Learning - PROJECT SUMMARY To survive, animals must learn appropriate associations between sensory cues and motor actions through a process of trial and error. We expect that this learning will strengthen synaptic connections between neurons representing the sensory cue and neurons initiating the motor action. The strengthened synapses may be direct connections between these neuronal populations or via systems intermediate between them, i.e., a “plastic brain circuit” or “pathway.” Synaptic plasticity has been observed in many brain areas, and the mechanisms are moderately well understood. However, we have struggled to identify which plastic brain circuit underlies, specifically, the sensory cue-to-motor action association learned through trial and error. This is due, in part, to the fact that many brain areas undergo plastic changes during learning, as the experience recruits a variety of cognitive processes, including sensory detection, motor control, reinforcement learning, and memory recall—processes that all engage different brain areas and distributed networks. I have developed an approach to assign these cognitive functions to distinct brain circuits for a case of trial and error learning in mice. Specifically, sensory detection requires the visual cortex; motor control requires brainstem circuits to control the forelimb; reinforcement learning requires the posterior dorsomedial striatum tail (pDMSt); and memory recall requires the superior colliculus. We have discovered a clear division of labor—the learning (pDMSt) and memory recall (superior colliculus) brain centers are spatially separated. We will now identify the specific cells within the superior colliculus that store the learned association to mediate memory recall. Hence, we aim to identify the precise site of an associative memory in the brain. Finally, we will study interactions between the learning center (pDMSt) and the memory recall center (superior colliculus) to understand the “neural handshake” between learning and memory systems. To do this, we will optogenetically inactivate pDMSt while recording neural activity in superior colliculus. This will reveal the component of the neural activity pattern in superior colliculus that depends on input from pDMSt, which drives learning. This will allow us to characterize the “learning signal” as it is transferred to the memory storage site, analogous to how information in computer RAM is written to the hard drive. Finding the neural basis and mechanisms of such learned, sensory cue-motor action associations will be essential to treat harmful associations, such as in PTSD, OCD, autism and anxiety, without generally disrupting sensory or motor behavior.
