Grant: Discovery Grants Program - Individual

The ability to learn about the environment and adapt behavior using predictions based on learned experiences is fundamental to brain function. Yet it's not clear how the brain is able to accomplish this ability. Comprehensive theories of brain function, such as predictive coding theory, have suggested that the brain is constantly building a virtual model of one's environment to produce predictions that can better guide an individual towards optimal outcomes. An essential feature of these theories is that mental representations of the environment are tuned, through learning, to match external information passing through the sensory systems; and if they do not, such mental representations will eventually drift and no longer be tethered to the physical world. To maintain a valid virtual model, the brain must determine mismatches between model predictions and sensory inputs and adjust the model accordingly. Support for these theories primarily arises from fMRI data in humans and precise mapping of cortical circuit structures from anatomical and in vitro experimental techniques. Although these approaches are and have been essential for understanding this issue, causal evidence showing that mismatches between predictions and sensory stimuli can support learning is lacking. The research in this proposal addresses two related questions. First, whether sensory cortical regions are involved in generating mental representations; and second, whether disruption of these predictive representations can artificially drive new learning. We propose to address these questions in rodents due to the availability of precise neurobiological tools within the rodent research field. These tools can allow for monitoring of many neurons at a high temporal resolution and can support precise neural manipulations using new technologies such as optogenetics. We use the gustatory (taste) system as our model sensory system since tastes are both inherently motivating and expected stimuli. This allows for behavioral procedures to be used that can ensure the rodent subjects develop predictions of the stimuli. This proposal describes a general program of research to tackle this fundamental hypothesis and lays out long-term goals in addition to a specific 5-year plan of feasible and carefully designed experiments. These experiments address questions of how the gustatory network 1) generates predictions of taste stimuli, 2) compares predictions of tastes with the actual taste stimulus delivered, and 3) distributes signals about mismatches between predictions and actual stimuli - what are referred to as prediction errors - in order to adjust and refine taste predictions. Understanding how the brain creates models of the environment and corrects these models when they are inaccurate is fundamental to the understanding of healthy brain function. This program of research is designed to understand how that world model is learned about and kept up-to-date within the brain.