What We Do

How do neurons in neocortex work in concert to encode the natural world, construct a representation of its behaviorally relevant content, and coordinate behavior based on those representations? This is a complex problem that we are addressing in multiple ways, with an interdisciplinary and comparative approach. The perspective that multineural activity is shaped by interconnectivity and recent dynamics is central to our approach.

Multineuronal Patterns as a Computational Substrate

Understanding how local circuitry relates to systems-level processing requires evaluating what features of multineuronal activity are accessible to neurons downstream. To evaluate how information is represented within local interconnected circuits we are quantifying the decodable elements of circuit activity using machine learning algorithms. We generate large-scale recordings of circuit dynamics in primary visual or auditory cortex while a locomoting mouse views natural movies or listens to natural auditory scenes, respectively.

Short and long term stability of the circuit representation of visual input 

Neurons in visual cortex work in concert to encode a visual scene and construct a representation of its behaviorally relevant content. The brain’s code for the visual world is accurate and robust across the lifespan of an organism, despite the fact that the response of individual neurons can be highly variable. We are evaluating how stable multiple neuron are responses are from day-to-day in a live animal, and are testing what readout schemes the brain might use to maintain a good representation of the visual world. Functional tuning properties are a powerful time-averaged summary of these dynamic responses. In describing mean responses, they encompass trial-to-trial variations in spike-timing related to excitability, correlation-structure, and other factors. We are quantifying these trial-to-trial differences explicitly in order to understand how stable perception can be achieved with variable population responses in real time. By analyzing downstream decoding performance, our quantitative approach will further our understanding of how visual representations at the circuit level are sensitive to or robust against input instability.

Short and long term stability of the circuit representation of motor output and motor learning

Motor cortex dynamics are linked to movement onset and refinement of behavior. Neurons in motor cortex cooperate as a population and integrate diverse contextual inputs. We are studying population activity in motor cortex during naturalistic exploration of an enriched arena. To complement these unconstrained observations, we are studying reaching: the freely-moving mice initiate trials at their own volition to obtain millet seeds, while we record limb kinematics using a computer-vision approach.

Network Science – the neocortex as a complex system

To analyze multineuronal activity patterns we describe statistical dependence in the firing of pairs of neurons. We have developed multiple algorithms to summarize activity patterns in related but differently specialized ways. These functional “wiring diagrams” (based on links derived from firing patterns rather than anatomy) can be used in two ways depending on how spiking data are collected and analyzed: (1) to produce mathematically tractable, statistical descriptions of cortical circuit dynamics; (2) to identify excitatory synaptic connections between specific neurons. Most recently we have moved beyond descriptions of topology in functional wiring diagrams and have begun to mechanistically describe the origins of specific motifs by accounting for the non-linear integrative properties of neurons.

Network Models

We also design neuronal network models capable of reproducing realistic activity to test our circuit-centric hypotheses and generate new ones. Simulation is a valuable complement to experiments in living tissue because it reveals the consequence of design assumptions and provides omniscient access to simulated synaptic connectivity patterns. By moving back and forth rapidly between experiment and theory, our group is working to achieve a better understanding of the fundamental principles that govern neocortical computation.

Learn more about our tools.