Jeffrey D Schall
Center for Integrative and Cognitive Neuroscience, Vanderbilt Vision Research Center, Department of Psychology, Vanderbilt University,
Recent studies aim to explain the duration and variability of behavioral reaction time in terms of neural processes. The time taken to make choices is occupied by at least two processes. Neurons in sensorimotor structures accumulate evidence that leads to alternative categorizations, while other neurons within these structures prepare and initiate overt responses. These distinct stages of stimulus encoding and response preparation support variable but flexible behavior.
Introduction
The neural correlates of decision making have been reviewed so often that one might review the reviews. Instead, I focus here on new studies that relate neural processes to reaction time (RT). An enduring problem in psychology is that of explaining the duration and variability of response times [7]. However, most behavioral studies conducted with monkeys treat RT as an experimental confound to be avoided by imposing arbitrary delays in tasks in an attempt to separate the sensory-evoked from the movement-related modulation of discharge rate. Often this procedure is used without recognizing that insertion of the delay invokes additional processes of anticipation and readiness [8] that result in neural modulation [9]. In spite of the pivotal role of RT in theories of cognition, neurophysiological studies aimed at explaining the duration and variability of RT have only recently been conducted [10].
Accumulation of evidence
Lateral intraparietal area
A recent study by Roitman and Shadlen [11] extends a well-known line of research on the neural basis of visual discrimination. Monkeys discriminated the net direction of motion of a field of moving dots, with variable amounts of random noise, by shifting their gaze to one of two targets. Performance on this task is based on the representation of the motion stimulus in the middle temporal (MT) area [12]. However, the signals in MT are not sufficient to produce the saccade by which the discrimination is reported. To understand this transformation, activity has been recorded in sensorimotor parts of the brain that are innervated by MT, such as the superior colliculus (SC) [13], the lateral intraparietal area (LIP) [14] and the dorsolateral prefrontal cortex including the frontal eye field (FEF) [15]. The recent study by Roitman and Shadlen [11] permitted monkeys to report the direction of motion as quickly as they could. When most of the dots moved in the same direction, the monkeys produced a high fraction of correct responses with short response times. When a small fraction of dots moved coherently, the monkeys required more time and made more errors. The evolution of activity in the LIP was related to the quality of the stimulus and the time of the saccade. If the motion of the dots signaled a saccade to the target in the receptive field, the average activation of LIP neurons increased gradually following appearance of the motion stimulus. The increase was more rapid in trials with stronger motion and shorter response times. In response to a given stimulus, variability in response time was correlated with the variability in the rate of growth of an average of activity of LIP neurons. The results are interpreted in a framework that supposes that the activity of neurons in LIP (and by extension SC and FEF) corresponds to the accumulation of the difference in responses between pools of motion-sensitive neurons in area MT that represent the alternative directions of motion, which is an optimal quantity for decision making [4].
[optinform]