CORTICAL PROCESSING OF VISUAL MOTION INFORMATION 
J. Anthony Movshon, Ph.D., Investigator 
The primate visual cortex consists of more than 
two dozen separate areas, many of which are known 
to be specialized for particular forms of visual analy- 
sis. One of the best-understood segments of this 
pathway contains a series of connected areas 
thought to be involved in visual motion processing. 
The lowest levels in this subpathway contain neu- 
rons that make local measurements of image speed 
and direction; higher levels of the pathway combine 
these local signals to perform more complex analy- 
ses and to make motion signals available to the mo- 
tor system for the control of behavior. 
Dr. Movshon's laboratory has been active for a 
number of years in the study of neuronal signals in 
this part of the visual cortex, concentrating on a par- 
ticular visual area called MT or V5, which occupies 
a pivotal position in the processing sequence. Re- 
cently, Dr. Movshon and his colleagues have con- 
centrated on two main questions: How are the sig- 
nals that are carried by MT neurons used to support 
perceptual judgments of motion? How are such sig- 
nals used to generate visuomotor behavior? 
MT Signals and Visual Motion Perception 
In collaboration with Dr. William T. Newsome's 
laboratory (Stanford University), Dr. Movshon has 
used statistical techniques based on the theory of 
signal detection to analyze the information carried 
by individual MT neurons, recorded while trained 
monkeys performed a psychophysical task tailored 
to make maximum use of the signals carried by the 
neuron under study. This combined behavioral and 
electrophysiological technique makes it possible to 
measure neuronal and psychophysical performance 
in parallel under strictly controlled conditions. 
Most recent models of visual cortex have stressed 
the distributed nature of neuronal processing, lead- 
ing to the expectation that the signals of individual 
neurons might not be particularly informative. 
These experiments revealed, however, that most MT 
neurons, in fact, carry enough information to sup- 
port the psychophysical performance actually ob- 
served in trained monkeys and humans. Although 
this result does not rule out the idea that combina- 
tions of signals across large numbers of neurons may 
be useful for other tasks, it suggests that for appro- 
priately constructed circumstances, perceptual 
judgments may be based on signals carried by a 
handful of neurons. 
MT Signals and Visuomotor Behavior 
In addition to their role in supporting the percep- 
tion of motion, signals carried by MT neurons have 
rather direct access to the portions of the motor sys- 
tem involved in the generation and control of 
smooth pursuit eye movements. These movements 
are used by primates to stabilize the image of a mov- 
ing target on the retina so that it may be properly 
seen. The signals that drive pursuit have long been 
known to carry information about the speed and di- 
rection of target motion. Recent studies of pursuit 
have provided much information about the dynamic 
properties of the visual signals needed to support 
these precise eye movements; and in collaboration 
with Dr. Stephen G. Lisberger (University of Califor- 
nia, San Francisco), Dr. Movshon has been study- 
ing the visual response properties of MT cells to 
establish whether their signals carry the requisite 
information. 
These experiments have particularly explored the 
dimension of stimulus acceleration, since theoreti- 
cal analysis has shown that target acceleration sig- 
nals are likely to be needed to generate fast and accu- 
rate pursuit. The results have revealed that the 
dynamics of MT responses are well suited to the task 
of providing inputs to the motor system and that at 
least half of the neurons also carry the necessary sig- 
nals about target acceleration. This outcome 
strengthens the idea that MT signals are used rather 
directly to support pursuit. 
A Fly in the Ointment 
Before concluding that pursuit is inextricably 
linked to signals carried by MT neurons, it is neces- 
sary to take account of another set of results from Dr. 
Movshon's laboratory. In these experiments (in col- 
laboration with Dr. Lisberger and with Dr. Michael J. 
Hawken at New York University) , pursuit eye move- 
ments were measured in human observers viewing 
specially constructed targets designed to isolate par- 
ticular components of the visual pathway. One par- 
ticular target was created by modulating the color 
but not the luminance of the display. The visual mo- 
tion pathway in general, and MT cells in particular, 
are known to be relatively insensitive to such isolu- 
minant color targets, leading to the expectation that 
these targets would elicit inferior pursuit. In fact, 
isoluminant targets are precisely as effective as con- 
ventional luminance targets in driving pursuit. This 
NEUROSCIENCE 421 
