Two primary processes that are critical for normal higher level visual performance are memories for items we have seen recently, and, second, the ability to distinguish one from another while recognizing that that the items are similar, e.g. distinguishing 2 cats is different than distinguishing cats and dogs irrespective of which cat or dog. Short term visual memory seems to have two modes, one in which every image viewed passively creates a short-term memory trace, and one in which image of particular interest are actively held in memory for a short time. To distinguish these two types of short term, we train rhesus monkeys to perform two visual short-term memory tasks with a similar trial structure. In each trial a sequence of two to eight images is presented in sequence with a second interval between images. The monkey must indicate whether the last image matched either any image in the preceding sequence (match any) or whether the last image matched the first in the sequence (match first). In both tasks, the test image might match an image that occurred early or late in the sequence, otherwise it almost certainly matched an image in some previous trial. All eight monkeys learned both permutations of the task with overall performance ranging from 70 to 90% correct. In the match any task there was a strong linear relation between performance and sequence position of the remembered image, that is, the likelihood of a correct response went down linearly as the sample image position was further back in time. The linear model also extrapolated to predict the likelihood of mistakenly trying to match the test image when its match did not occur in the current trial, but had been in the previous trial. We suggest that the monkeys do not explicitly try to memorize the image sequence but that a memory trace is initiated for every image presented on the screen. It is this trace that decays linearly with time. In the match first task, all monkeys passively remembered images in the sequence, as shown by the success of the model in predicting mistaken responses errors when the test image matched non-first images from the sequence. Early in training several monkeys achieved scores of 60 to 70% correct on the active task while exclusively using match any mechanism. After learning the active task, extrapolation of the linear model failed to predict the responses when the test matched the first image in the sequence. The linear extrapolation underestimated percent correct by 10 to 40% across monkeys. This is evidence that the monkey invoke some other mechanism to remember the first image memory. This task provides a means to study both passive and active short term memories. In the past it has been somewhat difficult to demonstrate that monkeys learn to generalize stimuli across exemplars of a class of visual stimuli. We group has shown that normal monkeys quickly (in one testing session or less) learn to categorize different classes of visual stimuli (e.g., dogs vs cats;cars vs trucks). Results from classical neuropsychological studies on pattern discrimination were generalized to imply that temporal lobe area TE might be crucial to visual-perceptual generalization. However, we have recently found that monkeys with bilateral TE lesions learn these discriminations quickly (in a single session or less);they are seemingly unimpaired on a simplified visual categorization task in which they were asked to judge if an image was a dog or a cat. Although this result demonstrates that area TE is not required for visual categorization per se, we hypothesized that a more challenging version of the task would reveal impairments. We made the task more challenging by including computer-generated images that are morphs between pairs of dogs and cats. We trained seven monkeys to report whether the morphed image was more dog-like or more cat-like. All seven correctly categorized images when they were >65% dog or cat. However, the monkeys with bilateral TE lesions are impaired when the image categories are more ambiguous (51-60% dog or cat). This result demonstrates that although area TE is not required for categorization that can be carried out with large-scale features, when fine discrimination between ambiguous stimuli is needed to carry out the categorization, area TE is essential.

Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
U.S. National Institute of Mental Health
Zip Code
La Camera, Giancarlo; Bouret, Sebastien; Richmond, Barry J (2018) Contributions of Lateral and Orbital Frontal Regions to Abstract Rule Acquisition and Reversal in Monkeys. Front Neurosci 12:165
Eldridge, Mark Ag; Matsumoto, Narihisa; Wittig Jnr, John H et al. (2018) Perceptual processing in the ventral visual stream requires area TE but not rhinal cortex. Elife 7:
Kuboki, Ryosuke; Sugase-Miyamoto, Yasuko; Matsumoto, Narihisa et al. (2016) Information Accumulation over Time in Monkey Inferior Temporal Cortex Neurons Explains Pattern Recognition Reaction Time under Visual Noise. Front Integr Neurosci 10:43
Mochizuki, Yasuhiro; Onaga, Tomokatsu; Shimazaki, Hideaki et al. (2016) Similarity in Neuronal Firing Regimes across Mammalian Species. J Neurosci 36:5736-47
Wittig Jr, John H; Morgan, Barak; Masseau, Evan et al. (2016) Humans and monkeys use different strategies to solve the same short-term memory tasks. Learn Mem 23:644-647
Matsumoto, Narihisa; Eldridge, Mark A G; Saunders, Richard C et al. (2016) Mild Perceptual Categorization Deficits Follow Bilateral Removal of Anterior Inferior Temporal Cortex in Rhesus Monkeys. J Neurosci 36:43-53
Wittig Jr, John H; Richmond, Barry J (2014) Monkeys rely on recency of stimulus repetition when solving short-term memory tasks. Learn Mem 21:325-33
Lerchner, W; Corgiat, B; Der Minassian, V et al. (2014) Injection parameters and virus dependent choice of promoters to improve neuron targeting in the nonhuman primate brain. Gene Ther 21:233-41
Clark, Andrew M; Bouret, Sebastien; Young, Adrienne M et al. (2013) Interaction between orbital prefrontal and rhinal cortex is required for normal estimates of expected value. J Neurosci 33:1833-45
Kim, Hideaki; Richmond, Barry J; Shinomoto, Shigeru (2012) Neurons as ideal change-point detectors. J Comput Neurosci 32:137-46

Showing the most recent 10 out of 14 publications