 |
Neural Codes for Perceptual Discrimination of the Temporal Structure of Vibrotactile Stimuli
Summary: Ranulfo Romo's lab combines psychophysical and neurophysiological experiments to investigate the neural codes for perceptual discrimination. Current research in his lab aims to understand how sensory experiences arise from activity of brain circuits.
The problem of neural coding has stimulated a large amount of research in neuroscience. The underlying belief is that unraveling the neural representations of sensory stimuli from the periphery to early stages of cortical processing is the key to addressing brain function, be it local or distributed. Recent studies in my laboratory that combined psychophysical and neurophysiological experiments have provided new insights into this problem.
In monkeys trained to discriminate the difference in frequency between two mechanical vibrations applied to the skin of one fingertip, we established that the rapidly adapting neurons of primary somatosensory cortex (S1) are directly involved in frequency discrimination, and that a firing rate/count code covaries with the animals' discrimination performance. This firing rate/count code is used by central areas of S1 to encode the stimulus frequency during working memory and decision making. An important concept that can be derived from these results is that S1 drives higher cortical areas where past and current sensory information are combined, such that a comparison of the two evolves into a behavioral decision. These and other observations in visual tasks indicate that perceptual decisions emerge from highly distributed processes in which the details of a scheduled motor plan are gradually specified by sensory information.
Although we have rejected all corresponding codes except the firing rate/spike count code as the basis of vibrotactile discrimination, there is nevertheless the possibility that each of the rejected codes (i.e., periodicity, burst count) might be useful for different purposes during the vibrotactile task. Given that our analysis is based on single units, it is possible that a temporal code based on interactions between multiple neurons (e.g., one based on spike synchrony), for either or both periodic or aperiodic stimuli, has escaped our scrutiny. Because the monkeys discriminate the difference in frequency between two vibrotactile stimuli, it seems that an “intense” rate code (firing rate/count) is sufficient to solve the task; subjects do not need to take temporal order or structure of the stimuli into account. Our results show that they may simply count whether one stimulus is higher than the other.
It is well known, however, that in everyday life we make perceptual decisions based on evaluating the temporal structure of the current sensory information available. Over the past years, we have implemented a new sensory discrimination task to address the role of temporal encoding in perception and decision making. The task is based on the premise that subjects must pay attention to the precision of the temporal structure of vibrotactile stimuli for successful discrimination. The task mirrors the process of language understanding, the meaning of a word, for example: here it is necessary to pay attention to the whole temporal sequence of the stimuli to understand it. In the task, the two vibrotactile stimuli are of equal frequency (3, 5, 7, 10, and 15 sinusoidal cycles per second), amplitude, and duration, but they differ in their temporal structure. In a trial, subjects must pay attention to the first stimulus, keep it in working memory for two seconds, and then compare the memory trace of the first stimulus to the current temporal structure of the second stimulus. The comparison must be retained during a delay period of two seconds, until a signal instructs the subject to press one of two push-buttons to indicate whether the temporal structure of the second stimulus is similar to or different from the temporal structure of first stimulus. Because the temporal structure of the stimuli varies from trial to trial, monkeys cannot discriminate between them by simply using one of the two stimuli. They cannot use an “intense” code, as in the vibrotactile frequency discrimination task, because in the new task the two stimuli contain the same number of mechanical pulses. Subjects are forced to compare all the intervals of the stimuli. This new experimental condition therefore opens the possibility to investigating the role of the precision of temporal codes in perception (i.e., spike timing versus “intense” codes of firing rates).
We examined the psychophysics of this new task and found that it is difficult for both humans and monkeys, but with sufficient training, monkeys are able to perform similarly to human subjects. We then recorded the activity from neurons of several cortical areas while the monkeys performed the task. Our database comes from recordings in the primary somatosensory cortex (areas 3b, 1, and 2), secondary somatosensory cortex (S2), areas 5 and 7b of the parietal lobe, premotor cortices (ventral premotor cortex, dorsal premotor cortex and medial premotor cortex), prefrontal cortex (several subfields) and from primary motor cortex (M1). All these areas are anatomically connected, and the question is whether, in this task, perception arises from the interaction between these cortical areas, and if so, how. Today, we have a fairly rough panorama of the neuronal activation in several cortical areas during this task.
There are important questions that can be investigated in this new sensory discrimination task. The first is concerns the sensory representation of the temporal structure during stimulus presentation in the somatosensory cortices and their central areas; the second addresses where and how the temporal structure of stimulus is stored in working memory; and the third concerns how the current temporal structure of the stimulus is compared with the memory trace of the first. The fourth question is how the comparison process evolves and is incorporated into the motor plan, and the last, most important question concerns the link between the different components of the task. We expect to answer each of these questions in the near future.
Last updated September 2008
|
|
|
|
