Neurophysics of Space, Time and Learning
Coordinators: Mayank Mehta, Peter Latham, Kechen Zhang
Scientific Advisors: Chuck Stevens
All events occur in space and time. How are space and time perceived by the brain? What biophysical and cellular mechanisms govern the formation of these neural representations? How does behavior shape the neural representation of space-time? These questions have puzzled the humanity for centuries. While much progress was made in the last century, there has been an explosion of experimental and theoretical techniques and discoveries in the last decade that have generated a substantial progress and provided many unsuspected answers.
More than half a century of research suggests that perception of space occurs by an interaction between two large and distinct brain regions over two, very different behavioral states. During spatial exploration the neocortex is thought to send the information about environmental stimuli to the hippocampus which is thought to rapidly learn the associations between these stimuli via the mechanisms of synaptic plasticity. Then, during the subsequent period of sleep, the hippocampus is thought to send this recently learned information back to the neocortex for long-term storage by a process called consolidation. Research also showed that the hippocampal neurons show spatially selective activity and an ensemble of a few hundred hippocampal neurons contain accurate information about the position of the animal.
How are these spatial maps of the world formed? Synaptic plasticity is crucial for the formation of spatial maps, but how does plasticity at a synapse change the emergent dynamics of ensembles of neurons and generate a map of space? A range of experimental and theoretical advances have revealed the Neurophysics of this fascinating process. Recent advances in genetic techniques has allowed us to perturb distinct components of this vast and complex circuit and probe the contribution of individual receptors to the map formation and learning. This, combined with theoretical studies, has revealed the surprising contribution of inhibitory synapses to the network dynamics and learning.
Advances in mapping neural circuit connectivity has thrown up several surprising findings. For example, the brain region between the neocortex and the hippocampus, called the entorhinal cortex, not only shows spatial selectivity but also shows spatially periodic firing, arranged as a hexagonal lattice, called the grid cells. A flurry of experimental and theoretical papers has investigated how the grid cells could arise, how they are modulated and what would be their behavioral outcome. These ideas include interference between multiple oscillators, resonance, dual lattice and spinglass models. These models are being challenged by fascinating sets of experiments including measurements of place cells and grid cells in flying bats using telemetry, and the use of transgenic rodents.
Advances in hardware have enabled scientists to probe the place cells and grid cells using novel techniques such as in vivo whole cell patch clamp during navigation, and in vivo two photon imaging during natural and virtual navigation, thereby providing an unprecedented view of the circuit dynamics. These findings have challenged the theories of map formation, thereby further refining them, and showing how the precise spike timing required for map formation can be generated in a network of noisy neurons. This combination of techniques also suggests that the neurons representing space in the hippocampus and the entorhinal cortex may also represent time, either directly or indirectly through the representation of speed. In fact, recent papers show that along with a neural code for position or time, there may be a concurrent and independent code for speed based on the Fourier components.
Recent in vivo measurement of identified neurons during sleep has also revealed that the brain behaves like a two state system during sleep and that the transitions between these two states can provide valuable information about the functional connectivity of this complex circuit. These have also revealed a novel role for the persistent activity that can be generated by a hysteresis type of mechanism in a stochastic network, which could facilitate the modification of spatial map memory during sleep and improve the long-term learning.
Thus, over the past decade, advances in diverse fields --such as hardware, behavioral measurement, anatomical tracings, in vivo measurement of neural ensemble activity, data analysis techniques, computational models, numerical simulations, and analytical theories-- have resulted in rapid progress in the field of Neurophysics of learning and neural representation of space and time. The meeting will bring together experts from these diverse fields and generate a much needed dialogue between them which would enable a coordinated evolution of the fields.
This program is funded by the Gordon and Betty Moore Foundation grant no. 2919 and National Institutes of Health grant no. R25 GM067110