Understanding how the brain transforms sensory stimuli into perceptions,
memories, and actions requires not only knowledge of its molecular
and cellular components, but also an understanding of the principles
by which the these components interact.
Faculty in the Center use modern techniques such as two-photon fluorescence
microscopy, multi-electrode recording, patch clamp recording and pharmacological
manipulation. Function is studied at the level of synapses and dendrites,
as well as neural circuits in the retina, cerebellum, brainstem, and
cerebral cortex.
Two-Photon Fluorescence Microscopy and Uncaging
Two-photon fluorescence microscopy allows
structure and activity to be observed at the level of single
synapses both in vitro and in vivo.
Caged compounds are molecules rendered biologically inactive
by the covalent addition of a light-sensitive caging group that
blocks the action of the molecule with its receptor. A flash
of light removes the group, thereby producing a biologically
active molecule. Two-photon uncaging
allows neurotransmitters such as glutamate to be produced in
femtoliter volumes (1 µm3) within a millisecond. With more widespread
activation, action potentials can be evoked in the target neuron.
|
|
Two-photon microscopy is being used by Professor
Sam Wang to
look for coincidence detection in the cerebellum, a likely site
for motor learning. This learning has been suggested to be mediated
in part by long-term depression (LTD), a calcium- dependent
process in which coincident activity of parallel fiber (PF)
and climbing fiber (CF) synapses causes a long-lasting decrease
in PF synaptic strength onto Purkinje cells. In the Marr/Albus/Ito
hypothesis for motor learning, the effectiveness of conditional
stimulus information arriving over the PF pathway undergoes
LTD in response to instruction by unconditional stimulus information
arriving over the CF pathway. |
|
Short Term Memory
In brain regions as diverse as prefrontal cortex
and brainstem, a key correlate of short-term memory is the presence
of persistent neural activity: spike activity that corresponds to
the memory of a stimulus and continues even after the stimulus is
removed. A promising system in which to study short-term memory is
the brainstem, where in lower vertebrates as few as 100 neurons are
capable of storing eye position. These neurons can maintain a fixed
firing rate corresponding to a given eye position for many seconds
without visual feedback, and this brain region performs well even
after sharp temperature changes or partial inactivation. Such robust
performance suggests the presence of rapid compensatory mechanisms.
By manipulating second messenger signaling or activity-dependent synaptic
learning rules, candidate mechanisms for short-term memory can be
studied.
Multi-electrode Recording
Our perceptions and actions result from
the activity of many neurons. Multi-electrode recording arrays
allow simultaneous experimental access to populations of neurons
in one recording. The array consists of 61 platinum electrodes
on a glass cover slide placed closely against neural tissue.
Recordings have been made from up to 80 retinal ganglion cells
for periods of up to 12 hours (see figure 5).
Retinal Encoding and Processing
Vision begins when light strikes the retina, a thin sheet of
neural tissue in the back of the eye. The optics of the eye
form a focused image in the plane of the retina, where light
is absorbed by a layer of photoreceptor cells and converted
into a neural signal. This signal flows through a layer of bipolar
neurons to the final layer of ganglion cells, which send brief
electrical pulses, known as action potentials or spikes, down
the optic nerve to the central brain. In addition, classes of
interneurons called horizontal cells and amacrine cells spread
signals laterally within the plane of the retina.
All visual information is encoded in the time of occurrence
of the spikes produced by ganglion cells. The multi-electrode
recording array allows spikes from many ganglion cells to be
recorded while complex and realistic visual stimuli are presented
to the retina with a computer monitor. Together, these techniques
enable precise control of the input to the retinal circuit and
measurement of the output relevant to the brain.
How does the activity of retinal neurons represent a visual
scene? At the level of the primary light sensors of the retina,
this representation is quite straightforward. One can think
of the layer of photoreceptors as the biological equivalent
of a CCD camera, where each photoreceptor is a pixel in the
retinal image and the voltage across its cell membrane encodes
the intensity of light falling on it at each moment. However,
the layer of ganglion cells is quite different: each cell receives
input from an overlapping set of many photoreceptors and delivers
its output in the form of a sparse sequence of spikes. Furthermore,
the human retina contains 100 million photoreceptors but only
1 million ganglion cells, so it must perform enormous data compression
without degrading the content of the visual input. Research on retinal encoding
is being conducted by Professor
Michael Berry.
|
![]() |
|
|
