Our Vision

At the Stanford Center for Mind, Brain and Computation, our research is based on the idea that cognitive functions arise from the collective activity of neurons organized into multi-regional systems. Such functions include the ability to think and reason, to perceive and remember, to choose among alternatives based on multiple contributing factors, and to plan and execute effective actions. Despite the vast expansion of the sciences of brain, behavior, and computation in the 20th century, we are still far from total understanding of these emergent functions, and equally far from capturing them in artificial systems. The goal of the MBC is to give future scientists the necessary tools to investigate the emergent functions of the brain and to contribute to the development of artificial systems that can emulate them.

The pursuit of this goal requires the interdisciplinary combination of mathematically sophisticated computational methods with increasingly complex experimental approaches. In the past, researchers within the traditional disciplines of psychology, neuroscience, and computer science have often worked in isolation from each other. They have used behavioral, physiological or computational approaches with little reference to the work of other types of investigators. It is now apparent that a full understanding requires an interdisciplinary synthesis of many available approaches.  

back to top

Research Goals

The nervous system solves problems that seem easy to a behaving human being, ranging from viewpoint invariant object recognition to the ability to grasp an egg and crack it into a bowl. These processes are in fact far more complex than they appear.  To better understand how such hard problems are solved by the brain, researchers in the field have recently begun to conceptualize these processes in explicitly computational terms. This requires clear quantitative characterization of the information available for problem solution, and also of the performance measures that must be optimized to obtain a solution. It also requires the use of advanced statistical and computational methods to find optimal solutions to such problems.

Research of this kind has long been carried out in close synergistic interaction with neuroscience investigations. Many important computational principles, such as the use of sparse coding for vision, and the use of temporal difference learning for reward prediction, arose in part because they simultaneously addressed computational problems and explained neurophysiological observations. Neuroscientists are more and more frequently testing computational theories, and researchers who perform experiments that investigate cognitive functions increasingly rely on advanced analytic methods.   Functional MRI, diffusion tensor imaging, the various scalp recording technologies (MEG and EEG) and the interpretation of neurophysiological data recorded simultaneously from scores of individual neurons increasingly depend on advanced methods from machine learning and statistics to allow investigators to make sense of these data sets.

It is apparent that our understanding of the emergent functions of neural systems will be dependent on mathematically sophisticated computational methodologies, both as a framework for representing our understanding of how these functions arise and as a set of tools that allows us to use experimental data to advance that understanding. In particular, the degree of mathematical and statistical expertise required to understand and apply new neuroimaging methodologies is steadily increasing. A considerable body of recent work recognizes these points, and highlights the necessity of bringing together the relevant disciplines.  The researchers leading the MBC exemplify this convergence, synergizing their efforts across disciplines in the effort to understand the inner workings of the brain.

back to top

Graduate Training Program

The goal of our program is to train the next generation of scientists to combine computational and experimental methods in order to understand the emergent functions of neural systems.  Stanford University is an exceptionally rich environment for this interdisciplinary training effort. Stanford has top-ranked departments in Psychology, Electrical Engineering, Computer Science and Neurobiology, all of which provide essential training perspectives.  Importantly, these departments are all situated within 300 yards of each other on Stanford's campus, creating unparalleled ease of access for students to laboratories in all key disciplines.

The rich intellectual environment in each department coupled with their close physical proximity creates an "incubator" that is ideal for collaborative, interdisciplinary research and education. Students will be given a structured training experience in this environment, allowing them to progress toward positions in academic and other research institutions where their combined computational and experimental skills will allow them to assume positions of scientific and engineering leadership.

back to top