Over the past half century, it has been shown that damage to the lateral inferior temporal cortex, areas TE and TEO, interferes with the ability of monkeys to integrate visual features into identifiable objects. Given this history, when we removed either TE or TEO, the architectonically defined inferior temporal regions, we were surprised that the monkeys had only moderate impairments in placing visual stimuli into categories based on perceptual similarity. Furthermore, when the monkeys gained just a little familiarity with the specific sets, the deficit became substantially milder. Now we have tested whether combining damage of both TE and TEO would result in a severe deficit of the type we had originally expected with TE removal. As in the earlier studies, we used stimuli made up from morphed (blended and warped) images of 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 monkeys had to touch a bar to initiate trial. We compared the behavioral data collected 10 days before the TE+TEO removals, with the data collected 10 days after recovery from surgery. Combined TE+TEO removal produced a severe impairment in visual categorization that showed only a small amount of recovery across the 10 post-surgical testing days. Despite this profound deficit in categorization of the morphed stimuli, these monkeys did not show any impairment in either discriminating two black and white checkerboard patterns, or in a test of visual acuity where the monkeys had to indicate whether a faint striped pattern was present, that is, a test for contrast sensitivity. Thus, combined TE+TEO removal does not affect low-level visual function but does lead to a profound loss of high-level visual categorization in monkeys. Another aspect of vision that plays an important role in daily life and is studied in old-world monkeys is visual short-term memory. Recently we showed that monkeys and humans seem to use different strategies for short-term memory tasks. Rhesus monkeys rely heavily on recency of stimulus repetition to solve visual short-term memory tasks, whereas humans rely heavily on specific memorization. In the past we have also shown that monkeys can integrate the identity of images across visual noise, although as the amount of noise increases the performance decreases. In five old-world monkeys (two Rhesus monkeys and three Japanese monkeys) we studied whether the presence of visual noise interacted with short-term memory. A sequence of visual stimuli (2-5) was presented. The monkeys were required to report whether the test (last) stimulus matched the first stimulus in the sequence by releasing a bar in the primate chair if they matched. We superimposed random dot visual noise on the test stimuli to degrade the stimuli patterns. We previously found that the behavioral performance of Rhesus monkeys was affected by the amount of visual noise. Both Rhesus and Japanese monkeys were affected by the noise, with the three Japanese monkeys performing better than either of the two Rhesus monkeys. In line with the earlier studies, the performance of both species was affected by the number of stimuli in the trial. In the assessment of working memory, all of the monkeys performed better in shorter trials, suggesting that they were relying on recency to perform. In addition, the error rate increased significantly when the matched stimulus in the previous trial was presented as the test stimulus in the current trial. These results seem to show a strong recency effect in a visual recognition task in both species of monkey, an effect unaffected by visual noise on the test stimulus. Manipulating the activity of neurons during experiments is an important technical approach to studying the relation between neural structures and behavior. Ideally a technique would be specific, and reversible. New tools coming from molecular genetics offer the opportunity to develop such tools for use in primates. We have pioneered the development of one agent, the DREADD (designer receptor exclusively activated by designer drug), to silence neurons. It would be helpful to have other agents. Artificially introduced ligand-gated ion channels (LGICs) have been used in the mouse to inactivate selected brain regions repeatedly and non-invasively. When the anthelmintic drug ivermectin (IVM) is applied, cells expressing GluCl become hyperpolarized because the cell membrane has a functional leak in a type of chloride channel. In monkeys, GluCl can be expressed in an area of interest using replication-deficient viral vectors to carry the gene into the neurons. After infecting neurons with this viral construct, systemic injection of IVM causes neurons expressing GluCl to become hyperpolarized, thereby silencing the target region. We injected lentivirus vectors expressing GluCl/IVM into primary visual cortex of two rhesus monkeys. We also removed a region of contralateral visual cortex symmetrical to the AAV2 injection site by aspiration as a positive control. After recovery from surgery, the monkeys performed a signal detection task. On 80% of trials, a high-contrast target was presented in one hemifield and a lower-contrast distractor was presented symmetrically about the vertical meridian. The monkeys were supposed to saccade to the brighter cue to obtain a liquid reward. The monkeys failed to saccade to the target when it was presented in the scotoma corresponding to the aspiration removal. This was accompanied by an increase in the number of saccades to the distractor presented in the symmetrically opposite site. When we administered injected IVM (7.5 mg/kg) 24 hours prior to testing, the monkeys also failed to detect the distractor in the opposite hemifield (i.e. in the receptive field corresponding to the region of V1 expressing GluCl). Thus, we have shown that the technology for Glu/IVM channels can be used to silence neurons reversibly and repeatedly in the primate visual system.
Showing the most recent 10 out of 14 publications
