Abstract: A prevailing notion in memory research is that the persistence of a memory depends on the stability of the neural codes established during learning. However, recent studies have shown that hippocampal place codes and cortical sensory codes gradually change (or drift) over time, challenging the idea that stable codes are required for stable memories. I will present results from experiments using longitudinal optical imaging to study hippocampal and entorhinal codes for long-term spatial memory. Our findings indicate that spatial memory content can be reliably preserved over weeks, even without stable hippocampal representations. Like hippocampal place codes, entorhinal grid codes also drift over time. Unlike place cells, however, grid cells (within a module) drift coherently, reflecting shifts in their anchoring to the external reference frame. I will discuss how these findings inform current theories of memory and spatial cognition.
Vision provides animals with detailed information about their surroundings, conveying diverse features such as color, form, and movement. Computing these parallel spatial features requires a large and diverse network of neurons, such that in animals as distant as flies and humans, visual regions comprise half the brain’s volume. These visual brain regions often reveal remarkable structure-function relationships, with neurons organized in spatial maps with shapes that directly relate to their roles in visual processing. Our group has used the Drosophila motion pathway to map the connected biological computations that estimate optic flow—the pattern of changes that self-motion induces in the visual scene. We identified, targeted with genetic tools, and functionally characterized cell types across six layers of the brain—from photoreceptors through directionally selective neurons to central circuits that disambiguate global optic flow patterns. By combining computational neuroanatomy with functional measurements, we traced how each step-by-step transformation contributes to perception and behavior. To explain long-standing puzzles in fly visual responses, we expanded our mapping to the sensory organs themselves and discovered that the global organization of directionally selective neurons’ preferred directions is determined mainly by the fly’s compound eye, revealing intimate connections between eye structure, functional properties of neurons, and locomotion control. Time and again, high-resolution anatomy has provided crucial functional insights, motivating us to extend these methods to an entire visual system. We recently completed a comprehensive Drosophila optic lobe connectome, sorting ~53,000 neurons into ~700 cell types, and paired this with genetic driver lines matched to connectome-defined cell types and accessible analytical tools for exploration. I will present examples of how working between single-cell characterization and large connectomic approaches accelerates discovery in the fly visual system, where convergent anatomical, functional, and genetic approaches reveal both the hierarchical logic of biological computations and the stunning organization of a complex visual system.
Abstract: As we interact with the world around us, we experience a constant stream of sensory inputs, and must generate a constant stream of behavioral actions. What makes brains more than simple input-output machines is their capacity to integrate sensory inputs with an animal’s own internal motivational state to produce behavior that is flexible and adaptive. Working with neural recordings from subcortical structures involved in regulation of survival behaviors, we show how the dynamical properties of neural populations give rise to motivational states that change animal behavior on a timescale of minutes, while neuromodulation can alter these dynamics to change behavior on timescales of hours to days. Using methods from control theory and reinforcement learning, we demonstrate that different sites of modulation within a neural circuit produce different resulting effects on behavior and neural activity. We then show how theoretical models can reveal unexpected effects of neuromodulation on the dynamic regimes of recurrent neural networks, illuminating the ways in which the brain might use small molecules to reshape its activity and thus modify behavior.
