This is the web home of Professor Jack Gallant's laboratory at the University of California, Berkeley. You can find out about our research program on this page and in Questions. Radio and television interviews, magazine articles and news stories about our work can be found in Press. Some published manuscripts can be found in Publications. Our amazing new visualization tool can be found in Brain Viewer .
Our laboratory studies how information is represented and processed in the human brain. Our goal is to formulate quantitative computational models that accurately predict how the brain responds under natural conditions. Most of our work has focused on the visual system, because vision is more approachable than the cognitive systems that mediate complex thought. However, we also have several other active research programs that investigate other aspects of cognition: how the brain represents auditory and linguistic information under natural conditions; how the brain represents information while performing complex tasks such as playing video games; and how these representations are modulated by attention and intention.
Human cognition is mediated by various hierarchically organized, tightly interconnected networks of brain areas. Each area can be viewed as a computational module that represents different sorts of information. For example, the visual system consists of several dozen distinct areas. Some visual areas represent the simple structural features of a scene, such as edge orientation, local motion and texture. Other visual areas represent complex semantic information such as whether a scene contains faces, animals, vehicles and so on. Our laboratory aims to discover how each brain area represents information about perceptual, cognitive and/or motor states, and to determine how these representations are modulated by attention, learning and memory.
One way to think about the brain is as a complex system that transforms input (stimuli) into output (behavior) through a series of intermediate, non-linear transformations. Each brain area performs a non-linear transformation on the information that it receives from upstream areas, and integrates this with information received by feed-back connections from downstream areas. As a result, each brain area forms a unique, explicit representation of information that is only represented implicitly in other brain areas. Our goal is to discover these various explicit representations, and to describe them in quantitative computational models. These models describe how information is encoded in each brain areas. However, once an accurate encoding model has been developed, it is fairly straightforward to convert it into a decoding model that can be used to read out brain activity, in order to classify, identify or reconstruct mental events. In the popular press this is often called “brain reading”. Thus, our work on encoding models has (as a side effect) produced many interesting brain reading results, which you can read about elsewhere on this site.
Most of the work in our laboratory involves functional magnetic resonance imaging (fMRI), a rapidly developing technique for making non-invasive measurements of brain activity. (Specifically, fMRI measures changes in blood oxygenation, flow and volume that are indirectly coupled to neural responses in the brain.) Because the accuracy of encoding and decoding models will inevitably depend on the quality of brain activity measurements, we are actively developing new methods for collecting and processing large, high-quality fMRI datasets. Our computational models exploit many different statistical and machine learning tools, including nonlinear system identification, Bayesian estimation theory and information theory. Therefore, we also perform fundamental research on statistical methods and estimation theory.
Our laboratory is located in the Department of Psychology, University of California at Berkeley. We are also associated with the Programs in Neuroscience, Vision Science, Bioengineering and Biophysics.