The lab’s research addresses a deep, yet poorly understood problem that straddles cognitive neuroscience and mechanistic neurobiology: how an organism discovers and learns the structure of its world. What objects are out there in the world? And how do those objects relate to one another?
Most studies of learning rely on task training or the explicit pairing of sensory cues with reward or punishment. But in the wild animals can learn the structure of their worlds absent such extrinsic reinforcement. While such unsupervised or task free learning has been captured in the laboratory for over a century (e.g. Tolman’s latent learning), we still lack a basic grasp of the neurobiological mechanisms that govern this process. This fundamental form of learning may be orchestrated by learning rules and biological mechanisms distinct from those engaged by canonical learning paradigms, and it has proved more resistant to progress in artificial intelligence than have supervised and reinforcement learning. Yet evolution has evidently discovered robust solutions to this problem.
Our lab uses the rodent olfactory circuit to explore the neurophysiological, circuit, and network processes that endow biological organisms with this rich cognitive faculty. To do this we have developed behavioral approaches for studying unsupervised, task-free learning in both head-fixed (eLife, 2019) and more recently freely moving animals (Neurobiology of Learning and Memory, forthcoming). We have also established methods for long-term observation of neurophysiological activity. This recently led to the discovery that the olfactory cortex of the mouse (the piriform) exhibits representational drift (Nature, 2021), suggesting the hypothesis that the piriform functions as a highly plastic, unsupervised olfactory learning system that continually learns and continually overwrites itself. In close collaboration with theoreticians, have developed advanced methods for inferring synaptic connectivity and plasticity in vivo to test this hypothesis. This opens up unprecedented opportunity to study how the connectivity of a network evolves over time and how that plasticity produces changes in network function and behavior.