Research areas: Systems and Computational Neuroscience
The focus of research in my laboratory is on understanding the structure and function of the visual system. Vision is the most important sense for humans, and damage to vision through disease or stroke is a serious problem affecting millions of people. The human visual system is quite complicated, consisting of several dozen distinct modules (visual areas) arranged in a highly interconnected, hierarchical and parallel network. The visual system is also tightly integrated with other sensory subsystems and systems for memory and language. Because of this interconnectivity, and because the brain is build on modular principles, vision research also has important implications for understanding other brain systems.
The goal of the Gallant Lab is to understand the structure and function of the human visual system at a quantitative, computational level, and to build models that accurately predict how the brain will respond during natural vision. Predictive models of brain activity are the gold standard of computational neuroscience, and are critical for the long-term advancement of neuroscience and medicine.
The research program in my lab reflects a tight integration of three distinct approaches: neuroscience experiments involving both classical electrophysiology and functional neuroimaging (fMRI); statistical analysis using methods adapted from nonlinear system identification and nonlinear regression; and theoretical modeling. Much of our research uses modern statistical tools to fit quantitative computational models that describe how visual stimuli elicit brain activity. Statistical tools drawn from classical and Bayesian statistics and machine learning are used to fit appropriate computational models to these data. The resulting models describe how each element of the visual system (e.g., a neuron, a voxel or an entire visual area) encodes information about the visual world. Models are evaluated both by statistical significance and by their ability to predict brain responses under new conditions. This second criterion, accurate prediction, is the gold standard of science and is fairly unique to our approach.
Computational encoding models that accurately predict brain activity have many practical uses. First, they provide a critical foundation for other work aimed at rehabilitation of visual function; after all, one needs to understand how a system functions before one can hope to repair it. Second, these models provide a new tool for neurological evaluation and diagnosis. Third, the models can be inverted in order to decode brain activity, providing a direct and principled way to do "brain reading", and to build brain-machine interfaces (BMI) and neural prosthetics.
Last spring we published an article in Nature demonstrating that our computational models be used to interpret human brain activity with unprecedented accuracy. The success of this effort suggests that these models may soon be good enough for practical application as clinical and BMI tools. This research was picked up in many newspapers and magazines worldwide, for example here, here, here, and here.
Other graduate programs at Berkeley
The research program in my laboratory integrates core knowledge from several different fields: neuroscience, statistics, computer science, physics and psychology. For this reason, we take graduate students from many different departments. I am affiliated with several different graduate programs at Berkeley, including Psychology, Bioengineering, Biophysics and Vision Science.
My lab runs the Neural Prediction Challenge, an informal contest meant to further development of quantitative neural models. The site provides data that can be used to fit models, and a procedure for automatic evaluation of model predictions on separate validation data sets. We also release our data under the auspices of a new NSF data sharing initiative, Collaborative Research in Computational Neuroscience. This program is intended to make primary neuroscience data available to theorists and modelers (and to other experimentalists), in order to speed scientific discovery and reduce needless duplication of neuroscience experiments.
Open source software tools
My lab also helps to develop and maintain the STRFPak system identification software toolbox. This software package can be used to estimate quantitative encoding models for any sensory system, based on data collected using neurophysiology, EEG or fMRI.
Kay, K.N., Naselaris, T., Prenger, R.J., & Gallant, J.L. (2008). Identifying natural images from human brain activity. Nature, 452, 352-355.
David, S.V., Hayden, B.Y., Mazer, J.A., & Gallant, J.L. (2008). Attention to stimulus features shifts spectral tuning of V4 neurons during natural vision. Neuron, 59, 509-521.
Hansen, K.A., Kay, K.N. & Gallant, J.L. (2007). Topographic organization in and near human visual area V4. Journal of Neuroscience, 27, 11896-11911.
David, S.V., Hayden, B.Y. & Gallant, J.L. (2006). Spectral receptive field properties explain shape selectivity in area V4. Journal of Neurophysiology, 96, 3492-3505.
Wu, M.C.K., David, S. V., & Gallant, J.L. (2006). Complete functional characterization of sensory neurons by system identification. Annual Review of Neuroscience. 29, 477-505.