The Center for Adaptive Systems (CAS) is an
interdisciplinary research and training center whose interests intersect
the areas of biology, computer science, engineering, mathematics, and
psychology. The Center performs interdisciplinary research aimed at
discovering and developing principled theories of brain and behavior,
notably concerning how individual humans and animals adapt so well on
their own to rapidly changing environments that may include rare, ambiguous,
and unexpected events. The Center also develops technological applications
that are inspired by its biological models. Research and training are
carried out both individually and through close collaborative relationships
between faculty, students, and postdoctoral fellows. Research projects
encompass a broad range of areas concerning cognitive and neural systems,
including vision and image processing; audition, speech and language
understanding; adaptive pattern recognition; cognitive information processing;
self-organization and development; associative learning and long-term
memory; reinforcement and motivation; attention; adaptive sensory-motor
planning, control and robotics; navigation and spatial orientation;
biological rhythms; consciousness; and the mathematical and computational
methods needed to support advanced modeling research and applications.
Both normal and abnormal behaviors are analyzed, including Parkinsons
disease, attention deficit disorder, schizophrenia, and depression.
These
investigations lead to neural network models that clarify the functional
architecture of different brain regions. Recent models characterize
the functional organization of such brain areas as the visual cortex,
auditory cortex, temporal cortex, parietal cortex, motor cortex, prefrontal
cortex, hippocampus, hypothalamus, cerebellum, superior colliculus,
basal ganglia, reticular formation, thalamus, retina, and spinal cord.
General neural designs that realize specialized
functional roles in distinct brain regions are clarified through such
models. Different levels of organization are analyzed, ranging from
neural systems and architectures to neural modules, local circuits,
and cellular, biophysical, and biochemical mechanisms. For example,
CAS and CNS have led the way in modeling how and why the architecture
of all sensory and cognitive neocortex is organized into layered circuits.
This research clarifies how laminar computing contributes
to biological intelligence. Such cortical laminar cortical architectures
are under investigation in vision, recognition learning and categorization,
short-term memory, cognitive information processing, and sensory-motor
planning. A typical example on the module level is opponent processing
circuits by on-cells and off-cells. Specialized versions of this module
play a key role in vision, biological rhythms, reinforcement learning,
motor control, and cognitive information processing. Such a comparative
analysis clarifies how a single modular design may be adapted to many
different behavioral functions. A typical example on the mechanism level
is associative learning, which plays a key role in such varied behaviors
as recognition, spatial orientation, and sensory-motor control. Contributions
of the specialized electrical and chemical dynamics of individual cells
are analyzed in every model. The models also provide explanations and
predictions of data that link the several levels of behavior, evoked
potentials, neurophysiology, anatomy, biophysics, and biochemistry.
These neural models are typically naturally expressed
as nonlinear dynamical systems. Numerical and analytical investigations
of these systems lead to new mathematical results and problems, as well
as to formal bridges to other biological and physical systems, notably
dissipative systems that describe aspects of self-organization and nonequilibrium
behavior. These formal investigations suggest new designs for computer
vision, adaptive pattern recognition machines, autonomous robots, and
massively parallel computers, thereby integrating basic science with
the design of novel technologies. Faculty and students also interact
with working engineers in companies and government laboratories to implement
neural network designs in new hardware for technological applications.
As a part of Bostons large academic community,
the Center has facilitated active collaborations among scientists at
neighboring universities and research laboratories. In addition, Bostons
prime location leads to a steady stream of national and international
visitors.
back
to top