Readout and control of spatiotemporal neuronal codes for behavior
To survive, organisms must first accurately represent their environment and use that representation to generate appropriate actions. Historically, these two processes – the mapping from stimuli to neural responses (“neural coding”) and the mapping from neural activity to behavior (“readout”)– have largely been treated separately. However, there is still no consensus about the neural code for stimuli in most brain areas because it is unclear which signals the brain uses (“reads out”) to perform particular behaviors. We have developed a theoretical framework to approach this problem. Experimentally informing this framework requires manipulating patterns of neuronal activity at the same spatiotemporal scale as natural firing patterns during stimulus-guided behaviors. This work must be done in behaving animals because it is essential to know which neural codes can guide behavioral decisions.
With support from the NIH BRAIN Initiative, we have formed a consortium of 11 laboratories spanning 6 institutions to develope the technology necessary to realize this goal. We are extending patterned neuronal stimulation technology and applying it to answer long-standing questions about neural coding and readout in the visual, olfactory, and auditory systems. We are identifying which neurons within a network encode behaviorally relevant information for particular behaviors, and how the temporal patterns of their activity are used to guide behavior. Additionally, we study these neural coding principles across dynamical contexts and learning to determine how internal state and past experience shape coding and readout.
Through this work, we aim to provide the research community with the tools needed to test theories of how neural populations encode and decode information throughout the brain. Also, we will reveal fundamental principles of spatiotemporal neural coding and readout in the visual, olfactory, and auditory systems of behaving animals. Finally, our unifying theoretical framework for cracking neural codes will allow the broader neuroscience community to resolve ongoing debates regarding neural coding that have been stalled by considering only half of the coding/readout problem.
Areas of Focus
The Administrative Core will coordinate all activities to ensure that the benefits of the multidisciplinary approach are realized. Administrators support the work of each of the science projects and resource cores by providing oversight and organization, coordinating team meetings to serve the Program’s needs.
Determining which neurons contribute to a particular behaviorally distinguishable percept. Paramount to studying neural codes is knowing which neurons are doing the encoding. The onset of any stimulus activates an extensive population of neurons, but without cell-specific causal interventions, it is impossible to know which subset contributes to a behavioral response. Here we will develop, validate, and apply our approach to determine the relative behavioral weighting given to individual neurons with different stimulus selectivities, laminar positions, genetic identities, and axonal projection targets.
The Technology Core will tackle a set of technically advanced yet feasible dissemination and development steps that will enable two-photon optogenetic stimulation to achieve robust and precise distributed neural control in behaving animals across brain areas and cell types.
Determining the contribution of relative timing of activities across neurons in coding behaviorally distinguishable percepts. Typical behaviors involve sensory cell responses that span tens or hundreds of milliseconds, but the behavioral significance of fluctuations in spike rate over these periods remains largely unknown. Project 2 will measure how behavioral choices integrate spikes over time and will determine the temporal resolution of readout mechanisms.
data science core
The Data Science Resource Core will support and enable the three Research Projects and the Technology Core by developing and implementing a data science framework that will make it possible to effectively compare model-based experimental studies across experimental sites and sensory regions.
Determining how neural coding and readout depend on internal state and past experience. The context in which sensory signals are processed is constantly changing, but the mechanisms and principles that neuronal circuits use to accommodate or exploit those changes remain
poorly understood. Project 3 will measure how sensory coding and readout change with states of arousal and how they respond to spontaneous fluctuations in neuronal excitability. We will also measure how encoding and readout change across states of arousal and when new perceptual skills are learned.