Visual short-term memory tasks depend upon both the inferior temporal cortex (ITC) and the prefrontal cortex (PFC). Activity in some neurons persists after the first (sample) stimulus is shown. This delay-period activity has been proposed as an important mechanism for working memory. In ITC neurons, intervening (nonmatching) stimuli wipe out the delay-period activity; hence, the role of ITC in memory must depend upon a different mechanism. Here, we look for a possible mechanism by contrasting memory effects in two architectonically different parts of ITC: area TE and the perirhinal cortex. We found that a large proportion (80%) of stimulus-selective neurons in area TE of macaque ITCs exhibit a memory effect during the stimulus interval. During a sequential delayed matching-to-sample task (DMS), the noise in the neuronal response to the test image was correlated with the noise in the neuronal response to the sample image. Neurons in perirhinal cortex did not show this correlation. These results led us to hypothesize that area TE contributes to short-term memory by acting as a matched filter. When the sample image appears, each TE neuron captures a static copy of its inputs by rapidly adjusting its synaptic weights to match the strength of their individual inputs. Input signals from subsequent images are multiplied by those synaptic weights, thereby computing a measure of the correlation between the past and present inputs. The total activity in area TE is sufficient to quantify the similarity between the two images. This matched filter theory provides an explanation of what is remembered, where the trace is stored, and how comparison is done across time, all without requiring delay period activity. Simulations of a matched filter model match the experimental results, suggesting that area TE neurons store a synaptic memory trace during short-term visual memory? ? Another visual higher function of great interest is category learning. Category learning in monkeys is often studied with behavioral tasks in which one observes monkeys category judgment through their choice of action and reinforces action indicating the correct choice. We have shown that monkeys can, in just a day or two, learn visual categories when performing a reward postponement task in which a choice between actions is not required. We tested 3 monkeys with categorical cue sets (20 stimuli for each of the 3 categories, dogs, cats, and rats) as predictors of postponement duration. On the third day, all monkeys differentiated among categories. The error rates did not change over 20 days. As a control for memorization of stimuli, we presented only novel stimuli for each category in a single session. The monkeys accurately differentiated the categories. Thus, monkeys can quickly learn to categorize visual stimuli and generalize to novel members of a set without instruction. ? ? Following-up the finding monkeys learn categories easily, we found that monkeys also dynamically regrouping of. The stimulus set was created from linear morphs between two given shapes (along the feature dimension shape) and two given colors (color dimension). Separation of the experimenter-defined classes was by shape or color, with unannounced switches during each session. The stimulus meaning is determined by its current class assignment. In addition we have tested a widely held hypothesis that this type of dynamic regrouping of stimuli might depend on lateral prefrontal cortex. Monkeys with bilateral lateral prefrontal ablations appear to learn and carry out this dynamic regrouping of stimuli as quickly as intact monkeys.? ? Humans have an unmatched ability to form abstract concepts and use them in complex rule-based behavior. To a lesser extent, this ability is observed in monkeys performing e.g. a delayed-match-to-sample (DMS) task. The classical DMS task requires differential action in response to different situations (e.g., GO if the test matches the sample, NO-GO otherwise), and in our laboratory, takes a few months to learn. We devised a DMS task where match and non-match trials are associated to alternative reward contingencies but do not instruct a differential behavioral response. The monkeys sit in a chair facing a computer screen and start each trial by holding a touch-bar. They are required to release the bar when a red spot (which appears overlaid to the sample image, in the center of the screen) changes its color into green (occurring after a short random interval from the appearance of the test image). Failing to do so results in the immediate abortion of the trial and a commencement of a new trial after the inter-trial interval. N=50 stimuli and 1s delay were used from the first day of testing, and delay duration was increased gradually up to 9s after the task had been acquired (see below). Rhesus monkeys (n=2) abstracted the concept of trial type across the delay and learned in 3-6 days to use it to predict the forthcoming contingency, as inferred from the highly significant different reaction times and percentage of aborted trials in different trial types (3-8 days if a measure analogous to trials-to-criterion was used). Thus, learning to form abstract concepts can occur quickly in monkeys if it emerges spontaneously from exploratory behavior.? ? Monkeys with bilateral lesions of the lateral prefrontal cortex (LPFC) were not impaired in the task, suggesting that the involvement of the LPFC may not be critical when the expression of a differential motor response in different trial types is not involved. There was, however, an impairment in learning the reversed rule-contingency, i.e., to switch the behavior after the predictive meanings of the trial types were swapped. This latter finding points to a crucial involvement of LPFC in reversing a rule-based behavior which requires short-term memory and abstract concept formation, but not in the in the initial acquisition of the same behavior.? ? The number of spikes emitted by a neuron varies with experimental condition, and all that we know about how neurons work indicates that these spike trains are responsible for carrying information quickly over distances from several millimeters to many decimeters (e.g., into the spinal cord). To describe a neuronal response completely we would need to keep track of the arrival time of each spike. We have shown that all of the information carried by neuronal spike trains requires specifying only the spike count distribution (which is approximately truncated Gaussian), the variation in firing rate with a bandwidth of less than 30 Hz (equivalent of measuring spike counts in 30 ms wide bins), and the peristimulus-time histogram, that is, how the firing rate changes over time. The spike count distribution (that is, knowing how many times each spike count occurs) is important because our assessment of temporal coding depends on the spike count. Intuitively this seems clear when we realize that the more spikes that are present, the richer the potential temporal code. Thus, the influence of variation in the number of spikes that occurs with successive presentations of a stimulus must be taken into account properly when estimates of neuronal coding are made. If the responses arise from a random process with a certain overall pattern in time, the responses must follow well-known statistical rules that are described by order statistics.? ? The order statistic representation allows exact knowledge of the amount of information carried by neuronal responses if the spike count distribution and the average variation of firing rate can be measured. Using a straightforward reformulation of the basic formula of order statistics, we derived a decoder that decodes neuronal responses millisecond-by-millisecond as the response evolves. This algorithm can form the basis of an instant-by-instant neuronal controller.
Showing the most recent 10 out of 13 publications
