Past events

The dynamics of learning in natural and artificial environments is a problem of great interest to both neuroscientists and artificial intelligence experts. However, standard analyses of animal training data either treat behavior as fixed, or track only coarse performance statistics (e.g., accuracy and bias), providing limited insight into the dynamic evolution of behavioral strategies over the course of learning. To overcome these limitations, we propose a dynamic psychophysical model that efficiently tracks trial-to-trial changes in behavior over the course of training. In this talk, I will describe recent work based on a dynamic logistic regression model that captures the time-varying dependencies of behavior on stimuli and other task covariates. We applied our method to psychophysical data from both human subjects and rats learning a sensory discrimination task. We successfully tracked the dynamics of psychophysical weights during training, capturing day-to-day and trial-to-trial fluctuations in behavioral strategy. We leverage the model's flexibility model to investigate why rats frequently make mistakes on easy trials, demonstrating that so-called "lapses" often arise from sub-optimal weighting of task covariates. Finally, I will describe recent work on adaptive optimal training, which combines ideas from reinforcement learning and adaptive experimental design to formulate methods for inferring animal learning rules from behavior, and using these rules to speed up animal training.

Perception is an active process, requiring the integration of both proprioceptive and exteroceptive information. In the rat’s vibrissal system, a classical model for the study of active sensing, previous works explored the relative contribution of the two information streams at the peripheral, thalamic and cortical levels. However, this issue was never addressed in the brainstem, and was only indirectly inferred from its thalamic projections. The current work addressed this gap in our knowledge by performing the first comparative study of the encoding of whisking and touch signals in the oralis, interpolaris and paratrigeminal nuclei. We also examined possible mechanisms for TIPs (Touch Induced Pumps) generation, a touch reflexive behavior mediated by the brainstem. By combining induction of artificial TIPs in anesthetized animals and computational models, we showed that passive retraction of the whisker pad is the most plausible mechanism for TIPs generation. Overall, our findings bridge a critical gap in our knowledge of the vibrissal system, providing the first characterization of responses to active touch in the trigeminal brainstem, the first evidence that the paratrigeminal nucleus is involved in the processing of vibrissal information and a novel mechanism for TIPs generation.

A fundamental problem in neuroscience is to elucidate the relationship between neuronal network anatomy and its functional dynamics. The nematode worm C. elegans is an ideal model to study these problems. Its nervous system has just 302 neurons and all synaptic connections between them have been fully mapped. Using a large-scale Ca2+-imaging approach, we previously discovered nervous system wide neuronal population dynamics in the worm that encode action commands. These dynamics feature various network attractor states during which neurons coordinate and synchronize their activities, thereby providing functional interaction maps. In this talk, I will discuss unpublished work where we combine graph-theoretical and experimental approaches to understand which anatomical features in network connectivity relate to these functional dynamics and interactions between neurons.

Denise Cai: Linking memories across time The compilation of memories, collected and aggregated across a lifetime defines our human experience. My lab is interested in dissecting how memories are stored, updated, integrated and retrieved across a lifetime. Recent studies suggest that a shared neural ensemble may link distinct memories encoded close in time. Using in vivo calcium imaging (with open-source Miniscopes in freely behaving mice), TetTag transgenic system, chemogenetics, electrophysiology and novel behavioral designs, we tested how hippocampal networks temporally link memories. Multiple convergent findings suggest that contextual memories encoded close in time are linked by directing storage into overlapping hippocampal ensembles, such that the recall of one memory can trigger the recall of another temporally-related memory. Alteration of this process (e.g. during aging, PTSD, etc) affect the temporal structure of memories, thus impairing efficient recall of related information. Tristan Shuman: Breakdown of spatial coding and interneuron synchronization in epileptic mice Temporal lobe epilepsy causes severe cognitive deficits yet the circuit mechanisms that alter cognition remain unknown. We hypothesized that the death and reorganization of inhibitory connections during epileptogenesis may disrupt synchrony of hippocampal inhibition. To test this, we simultaneously recorded from CA1 and dentate gyrus (DG) in pilocarpine-treated epileptic mice with silicon probes during head-fixed virtual navigation. We found desynchronized interneuron firing between CA1 and DG in epileptic mice. Since hippocampal interneurons control information processing, we tested whether CA1 spatial coding was altered in this desynchronized circuit using a novel wire-free Miniscope. We found that CA1 place cells in epileptic mice were unstable and completely remapped across a week. This place cell instability emerged ~6 weeks after status epilepticus, well after the onset of chronic spontaneous seizures and interneuron death. Finally, our CA1 network model showed that desynchronized inputs can impair information content and stability of CA1 place cells. Together, these results demonstrate that temporally precise intra-hippocampal communication is critical for spatial processing and hippocampal desynchronization contributes to spatial coding deficits in epileptic mice.