MIT Reports to the President 19992000
The human brain is the most complex, sophisticated, and powerful information-processing device known. To study its complexities, the Department of Brain and Cognitive Sciences at the Massachusetts Institute of Technology combines the experimental technologies of neurobiology, neuroscience, and psychology, with the theoretical power of computational neuroscience and cognitive science.
Edward Adelsons lab is continuing its research on the perception of lightness and transparency. They have developed a new understanding of the interaction of statistical and configural constraints in the human visual system's analysis of surface color. In addition, they have started a new project on the perception of material properties, such as shininess or roughness. Unlike object recognition, few have looked at the perception of materials. Since this is an important topic for machine vision as well as human vision, they are devising machine vision algorithms that will allow material recognition.
A primary focus in Bart Andersons lab is stereopsis. Many of the cells that exhibit disparity tuning in V1, MT and MST are also tuned to the direction and speed of a stimulus, as well as to its orientation content. They have developed a method that has generated the first psychophysical findings that demonstrate the role of motion and orientation in stereoscopic matching. These results suggest that one of the functional roles of cortical cells that multiplex the stimulus dimensions of speed, motion direction, and contour orientation is to help solve the binocular matching problem.
Nancy Kanwisher's lab has developed a new "event-related fMRI adaptation" technique that allows the characterization of neural representations in discrete visual cortical regions. Her lab has also recently discovered the strongest neural evidence yet for object-based visual attention.
Research in Peter Schillers lab investigates the neural control of vision and eye movement using physiological and behavioral methods. Recent work on vision concentrates on the neural mechanisms of depth perception, particularly stereopsis and motion parallax, but studying neuronal properties in the visual cortex of rhesus monkeys. Work on eye movement control examines how saccadic eye movements are produced by the numerous cortical areas involved that reside in the occipital, parietal and frontal cortices.
Since starting at MIT in September, 1999, Pawan Sinha has set up his laboratory for research in high-level vision. He seeks to understand the computational principles underlying the brain's ability to recognize objects, scenes and sequences. His approach involves examining human recognition performance with highly impoverished images. Computational analyses of these simplified stimuli provide valuable clues about the nature of information the brain uses for visual pattern recognition.
Emilio Bizzis lab seeks to understand motor control and motor learning. With respect to motor learning, they have described the neural changes observed in the primary motor cortex, the pre-motor cortex and the supplementary motor areas in monkeys while they learn a new motor skill. In all these areas they found a sizable population of cells that changed their tuning properties. During learning, these cells took on the properties of neurons that are involved in the control of movement.
Ann Graybiels lab studies brain mechanisms related to how we acquire habits. In one study reported in the past year, they recorded the activity of neurons in the basal ganglia while rats learned how to navigate a T-maze in order to receive chocolate rewards at the end of the maze. As the animals learned how to carry out this procedural task or "habit," there were large-scale changes in the activity patterns of neurons in the basal ganglia. These changes were long lasting, suggesting that circuits involving the basal ganglia have their activity rewired as a consequence of learning.
In a second study, the lab studied brain changes that occur in rats exposed to repeated doses of drugs such as amphetamine and cocaine, which lead to addiction and drug-dependent habits. The researchers found that this treatment induced highly distinctive changes in the patterns of expression of early response genes in the basal ganglia. There was a very high correlation between these changes in pattern and the changes in behavior exhibited by the animals.
Earl Millers lab found a neural correlate of the concepts "match" and "nonmatch" in monkeys that could apply them to any image. They also found prefrontal neurons that conveyed which of 3 complex rules a monkey was currently using. Finally, they discovered a neural correlate of perceptual learning. After 5 days practice with a set of objects, monkeys were better at recognizing them when they were degraded with noise. Practice also resulted in fewer neurons being activated in their prefrontal cortices, but these neurons communicated more information and were better at discriminating the degraded objects compared to when they were novel.
At CBCL, Tomaso Poggios group continues to work on the problem of learning in a) theory; b) engineering applications; c) neuroscience. In the theory domain, they are extending a mathematical framework for statistical learning. They are applying learning techniques to bioinformatics (DNA chips data), to information extraction from multimedia data, to financial markets, to trainable man-machines interfaces and to computer vision for object detection. In the neurosciences, they have continued to work on models of cortical circuitry underlying object recognition and object categorization and begun collaborations to test experimentally its predictions, including one with Earl Miller on object categorization in prefrontal cortex.
Molly Potters lab continued three lines of research and initiated one new line: short-term memory for naturalistic pictures, showing that the immediate representation is of gist, not just visual features; competition for attention between two near-simultaneous written words, showing intense competition over at least 500 ms; immediate visual memory for visual colors, shapes, and orientations; and interaction between pictured objects and words in rapidly presented sequences.
Through functional brain imaging, Anthony Wagners lab examines how human memory is organized and supported by the mind and brain. Over the past year, they have explored the executive control of memorythat is, how attention mechanisms subserved by the frontal lobes interact with other brain structures to guide memory formation and retrieval. Recent advances include characterization of multiple frontal-temporal neural circuits that support memory formation, and delineation of the functional organization of the frontal lobes.
Research in Matt Wilsons lab addresses the question of how memories are formed and maintained within the mammalian nervous system. Of particular interest is the possible role of sleep in the long-term establishment of
memory. By studying the interactions between brain areas using simultaneous neural recording techniques, they are pursuing the flow of mnemonic information during awake and sleep states between brain areas involved in memory formation and areas involved in higher-level cognition and decision making. They have recently found direct evidence of dreaming in rodents by identifying the reactivation during REM sleep of memory patterns established during recent awake experience.
Guosong Lius lab studies the mechanisms that control the levels of NMDA receptor activation at single synapses. This is a critical issue for the understanding of molecular mechanisms of synaptic plasticity. They found that the levels of NMDA receptor activation during synaptic transmission are determined by the concentration of transmitters in the synaptic cleft and can be enhanced by genetic modification of the NMDA gene. These findings shed new light on the mechanisms controlling NMDA receptor activation during synaptic transmission.
At the cellular level little is known about mechanisms underlying activity-evoked synaptic remodeling during visual system development. In Elly Nedivis lab, screening a pool of candidate plasticity genes (cpgs) that they previously identified revealed a subset that are expressed in the developing visual cortex and are activated by light, suggesting they may be involved in activity-dependent aspects of visual system development as well as everyday efficient function. They are now using cpgs as molecular tools to probe activity-dependent plasticity in the visual system.
Mriganka Surs lab demonstrated principles underlying plasticity of circuitry in the developing and mature cortex. By routing visual projections to the auditory thalamus and cortex in developing animals, they showed that the auditory cortex could develop local and long-range networks that were typical of the visual cortex. By probing the function of visual cortex networks underlying orientation tuning in adult animals, they showed that these networks undergo profound plasticity as a consequence of responses to previous stimuli.
Richard Wurtmans lab discovered that Nerve Growth Factor causes parallel increases in both the phospholipids and the transmembrane protein APP in membranes. (APP is involved in Alzheimers Disease, but probably is also involved in neurite outgrowth). Moreover, exposing cells to the membrane precursors choline and cytidine also increases the secretion of the neurite-expanding fragment of APP, i.e., "soluble APP."
Research in Ted Gibsons language comprehension lab has demonstrated that reading times in English sentences are strongly affected by the distance between key words in the sentence structure (e.g., a subject noun and its verb that depend on each other for interpretation). This work has important implications for understanding what makes language easy or difficult to understand.
With Stan Dehaene, Elizabeth Spelkes lab explored two systems of representation that underlie adults' numerical reasoning. Combined behavioral and neuroimaging techniques provide evidence that reasoning about exact numerosities depends in part on a frontal brain system tied to language and verbal memory, whereas reasoning about approximate numerosities depends in part on a parietal brain system tied to spatial memory.
Edward Gibson was granted tenure and Barton Anderson was promoted to Associate Professor without tenure. In the last year, the department began several new initiatives including a weekly colloquium series. An international group of distinguished speakers were featured. In addition, two Hans-Lukas Memorial lectures and Margaret Roche Donlan Bidwell lecture were held. For the first time, the department hosted an event at the annual Society for Neuroscience Meeting. The response was overwhelmingly positive and will take place again next year.
Ten new graduate students enrolled in the Fall 1999. One had received and NSF Fellowship, two were supported by fellowships made possible by a generous grant from Walter A. Rosenblith, and the remainder were funded by department NIH training grants. Of our six students receiving Ph.D.s, four were recipients of postdoctoral fellowships, one has a faculty appointment at the University of Illinois, Urbana-Champaign, and the sixth is working on manuscripts with her former advisor. Two new or completely revised graduate courses are Human Memory and Learning, taught by Anthony Wagner, and Introduction to Neural Networks, taught by Sebastian Seung.
Of the 93 undergraduate students enrolled in the program, 22 were graduating seniors. The curriculum was reviewed by an ad hoc education committee and will be streamlined this year by reducing the four tracks to two areas of specialization: Systems and Computational Neuroscience, and Cognitive Science. In addition, the same core subjects will be require for all majors and a greater variety of other courses will be available. New courses to be offered include: Introduction to Computational Neuroscience (Sebastian Seung), Foundations of Human Learning and Memory (Anthony Wagner), Cognitive and Behavioral Genetics (Elly Nedivi with Steve Pinker and David Housman), and Object and Face Recognition (Pawan Sinha). The later two will meet with graduate sections as well.
Earl Miller Troland Research Award from the National Academy of Sciences
Elly Nedivi Fred and Carole Middleton Career Development Chair
Steven Pinker MIT MacVicar Faculty Fellow for 2000 for devotion to undergraduate education Peter de Florez Professor of Psychology. His book, The Language Instinct, was named one of the 100 best science books of the century by American Scientist magazine.
Tomaso Poggio Laurea Honoris Causa from the University of Pavia for Voltas Bicentennial
Sebastian Seung Howard Hughes Medical Institute Assistant Investigator Robert A. Swanson Career Development Professorship in the Life Sciences McKnight Scholars Award
Pawan Sinha Alfred P. Sloan Research Fellow for 2000 NEC Award for 200001 Award from the Defense Advanced Research Projects Agency for 200002.
Elizabeth Spelke American Psychological Associations Distinguished Scientific Contribution Award American Psychological Societys William James Prize for Scientific Contributions
Mriganka Sur MIT Teaching Prize for Excellence in Graduate Education by the School of Science Australian Neuroscience Society Distinguished Lecturer for 2000 Sigma Xi Distinguished Lecturer for 20012003
Anthony Wagner Surdna Foundation Research Award, 2000 Ellison Medical Foundation New Scholars Award, 2000
More information about the Department of Brain and Cognitive Sciences can be found on the World Wide Web at http://web-bcs.mit.edu/.
Mriganka SurMIT Reports to the President 19992000