At the highest level of vision, perceiving complex objects, and distinguishing among those that overlap in appearance, e.g., we quickly distinguish any red tomato from any red apple, regardless of the variety, is done seemingly with ease. A half-century of behavioral, anatomical and physiological research has given rise to a hierarchical model of visual processing for objects. In this model the information about objects arises in a sequence of brain regions stretching from the occipital cortex, where simple features such as oriented edges or lines are represented, to the inferior temporal cortex called area TE where neurons represent conjunctions of visual features defining whole objects. Thus, the highest level of visual perception, perceiving complete integrated objects, is thought to take place in the inferior temporal lobe. A straightforward prediction from this model would be that area TE, in the monkey inferior temporal lobe, plays a key role in the categorization of complex stimuli including those having over-lapping features. Until recently it was thought that area TE represented the final stage for this hierarchical visual processing scheme. However, in the past two decades, there has been a disagreement about which rostral temporal lobe brain region, area TE on the lateral part of the temporal cortex or rhinal cortex on the ventro-medial part of the temporal cortex, represents the final stage of visual perception supporting the ability to recognize objects. it has been suggested that the rhinal cortex, which receives strong input from area TE, is important for recognition of complex objects or objects with feature ambiguity. Proponents for the rhinal cortex view argue that after removal of perirhinal cortex, monkeys are impaired in tasks requiring discrimination among stimuli with ambiguous or overlapping features. Skeptics argue that the experiments purporting to test for perception have been confounded by a memory demand. The monkeys have been asked to report whether objects presented in sequence with or without intervening objects are the same, i.e., does the test object match the sample or not. Hence deficits following removal of perirhinal cortex might have been due to a failure of memory, not perception. Recently we reported that monkeys with bilateral removal of area TE are only mildly impaired in performing categorical discriminations based on visual properties e.g. cat vs. dog or car vs. truck. The impairment seemed less severe than would be predicted by the hierarchical model. We set out to test whether making the stimuli ambiguous would lead to a greater deficit than that seen when distinguishing dogs from cats or cars from trucks. We used stimuli with overlapping features. The stimuli were created by blending and warping cats with dogs in different proportions thus giving a range of category-ambiguous morphed images. We tested three groups of monkeys, one control group, one group in which area TE had been removed bilaterally, and one in which rhinal cortex had been removed bilaterally. We presented one image at a time. These stimuli were morphed (blended and warped) cats and dogs ranging between 0 and 100% dog, with a distribution biased around the category boundary (11 levels, 0, 25, 35, 40, 45, 50, 55, 60, 65, 75, 100% of dog). The monkey was trained to touch a bar to initiate a trial, and had to release the bar during one of two intervals; early (signaled by a red central dot) if it identified the stimulus as more cat-like, or late (signaled by a green central dot) if the stimulus was identified as more dog-like. If the monkey responded correctly, the color of the central dot turned into blue and liquid reward was given. We assessed performance on this task before and after bilateral TEO lesions. Because our task places no explicit demands on short-term memory, the monkey is comparing the current image against a stored template, any behavioral deficit produced by TE or rhinal lesions can be interpreted as a failure in categorization based on perceptual similarity. If rhinal cortex contains the neural machinery subserving the highest levels of visual perception, removal of this cortex might be expected to yield an even greater impairment with these stimuli than that observed after TE lesions. We found that the monkeys with the TE removals were able to do the categorization but they made considerably more errors than the control group. The rhinal group was indistinguishable from the control group. Thus in our testing it appears that monkeys area TE plays an important role in object discrimination, but that there are other areas of the brain that support this ability to distinguish or classify feature ambiguous stimuli, and it appears that rhinal cortex does not play a role in the perception of these objects. Given the results described above, we sought to carry out the same tests in monkeys that had area TEO, the architectonic brain region just behind (caudal to) area TE, removed. In the hierarchical model this area provides the visual information to area TE. We reasoned that this removing this tissue would lead to an impairment in classifying our feature ambiguous cats/dogs. We analyzed the behavioral data for 4 days before the lesions, and four days afterwards. The percentage of response as dog-like for each morphed level was calculated and used as indicator of categorization accuracy. Our preliminary data suggest that the TEO lesions were showed poor categorization on the first day of post-lesion testing. In particular, the discriminability in the first day after lesion was significantly degraded when the morphing range was between 25-75%. As we previously observed in TE lesioned monkeys, the accuracy score improved over the first few days of exposure to the morphed stimuli, in this case recovering to pre-lesion levels of performance. We are now testing monkeys with combined TEO-TE removals. These monkeys are severely impaired in categorizing the morphed images, although they retain some ability to categorize pure dogs and cats.
Showing the most recent 10 out of 14 publications
