Seminar: University Seminary on Cognitive and Behavioral Neuroscience (#603)

Date: September 19, 2002

Title: Memory in Macaque Monkeys: Representation, Retention, and Self-Reflection

Speaker: Robert Hampton, Ph.D., National Institute of Mental Health

Attendees: Herbert S. Terrace (Chair), Psychology Department, Columbia University

Jon Horvitz, Psychology Department, Columbia University

Jennifer Mangels, Psychology Department, Columbia University

Len Matin, Psychology Department, Columbia University

Dustin Merritt, Psychology Department, Columbia University

Robert L. Thompson, Psychology Department, Hunter College

Nate Kornell, Psychology Department, Columbia University

Lisa Son, Psychology Department, Barnard College

Bridget Finn, Psychology Department, Columbia University

Kate Lynch, Psychology Department, Columbia University

Emily Stern, Psychology Department, Columbia University

Chris Sommerfield, Psychology Department, Columbia University

Tammy Moscrip, Psychology Department, Columbia University

Jay Edelman, Biology Department, City College

Josh Wallman, Biology Department, City College

Jonathan Levitt, Biology Department, City College

Nancy Dallal, New York, New York

Rapporteur: Michael R. Drew

Summary:

Dr. Hampton’s talk described experiments with old world monkeys (rhesus macaques), which have less developed brains than humans. The most recent common ancestor of humans and rhesus macaques lived 20 million years ago.

The experiments concern memory. Memory has many uses in primates: For instance, it allows an animal to recognize relatives, to learn which foods are safe and which are dangerous, and to navigate to food and water. In humans memory gives us our personal history and sense of identity.

There are 2 requirements of memory: first, information must get into the brain (this is the perception or representation stage); second, the information must stay in the brain. Dr. Hampton’s first experiments concerned this distinction between perception and retention.

By way of background, Dr. Hampton described the visual system of the macaque brain. Initial processing occurs in V1, and processing becomes more complex as it proceeds down the ventral path to the temporal lobe and the perirhinal cortex (PRh). As an example, cells in VI represent simple characteristics such as line orientation, whereas cells farther down the ventral stream represent whole, complex objects.

Dr. Hampton’s first experiment asked: does PRh play a role in perception (initial representation of visual stimuli) or memory? PRh is a small strip of cortex, but lesions to it have profound effects on cognition. In this experiment 6 animals received a bilateral PRh lesion and 4 animals received sham suregery.

After surgery the animals were given an object discrimination task. On a touch sensitive screen 2 stimuli appeared. Touching one stimulus produced reward; touching the other produced nothing. There were many trials. The task was conducted three times with stimuli that varied in complexity. The stimuli were 2D in one color; 2D in multiple colors; or 3D in multiple colors (but in any given task, both stimuli were of the same complexity level).

PRh lesions impaired all these discriminations, in terms of errors to criterion. This could be caused by deficits in learning, memory, or perception. The subsequent experiments attempted to determine which of these deficits is the cause. Jon Horvitz asked how much damage to PRh was caused by the lesions. Dr. Hampton answered about 90%.

Experiment 2 explored the role of perceptual demands. The strategy was to (1) ensure that monkeys have mastered the discrimination problems, then (2) administer single, perceptually challenging probe trials, free of any requirement for new learning.

The task was a discrimination (as above) with complex stimuli. On probe trials the stimuli were masked by cutting out parts of the stimulus image (e.g., a waffle pattern was removed). All animals were impaired on the probe trials, and the PRh-lesioned animals were no more impaired than controls. This suggests that deficits in perception are not the cause of the discrimination deficits in the PRh-lesioned animals.

Experiment 3 examined the role of memory, using reversal learning. The animals were trained on the discrimination task with complex stimuli (e.g., S1+, S2-), then the outcomes were reversed (S2+, S1-). After the reversal, the animals had to learn to suppress responding to the initially-rewarded stimulus (S1) and commence responding to the initially-non-rewarded stimulus (S2). The PRh-lesioned animals were slightly impaired on acquisition of the reversal. Dr. Terrace asked how many reversals. Dr. Hampton replied that just one was required. Dr. Terrace then pointed out that the initial response extinguished at about the same rate in both groups. The PRh animals were impaired only in acquiring the response to S2. That is, both groups reached chance performance after the reversal at about the same rate. Dr. Hampton replied that this pattern is anomalous and that usually the PRh animals are impaired from the start of the reversal.

Experiment 4 further examined the role of memory by looking at retention over a delay in a visual delayed-matching-to-sample (DMTS) task. If the PRh lesions cause only a perceptual deficit, then increasing the delay will effect both groups equally. On the other hand, if the lesions cause a memory deficit, then increasing the delay will have more severe effects on the PRh animals. The data were consistent with the latter scenario, suggesting a memory function of the PRh.

The next experiment looked at the role of the hippocampus (Hp). There is a lot of evidence that it is necessary for spatial memory. But recent studies in Dr. Hampton’s lab have failed to find that the Hp is necessary for non-spatial DMTS tasks. In this experiment, bilateral Hp lesions were created by injecting NMDA at toxic levels into the Hp. This procedure produces very specific Hp lesions that spare surrounding areas of the temporal lobe. Animals then performed a delayed-matching-to-position task in open field. The animals needed to find food, then wait a delay, and then return to the same location to find more food (a win-stay task). Both groups of animals were able to perform the task at a short delay – so Hp is not necessary for spatial representation. At longer delays the Hp animals were more impaired than controls.

The next experiments pursued the distinction between explicit and implicit memory and specifically asked the question, do monkey’s have explicit memories? In nonhuman animals this dichotomy has not been demonstrated. Could it be the case that all animal memory is implicit? The task was a visual DMTS task in which animals were given a choice, on each trial, about whether or not to take the recognition test after the delay. That is, after seeing the target stimulus and waiting the delay, the animals were presented with two stimuli that were constant across trials. Pressing one of the stimuli (S-test) caused the recognition test to be presented, and the recognition test offered the possibility of earning a preferred reward. Pressing the other stimulus (S-no-test) produced a less preferred reward but no recognition test. On this task animals tended to choose both options some of the time. In another version “forced choice” of the task the animals were required to take the test on 1/3 of trials (only the S-test choice was presented during the choice phase). On the remaining trials the animal could either take the test or avoid it as before. On forced tests the animals performed more poorly than on the tests they chose to take, indicating that animals tended to choose the recognition test only when they had a good chance at success. Jon Horvitz then asked if when the animals are forced to choose the test, whether the animals were faster to press S-test on those trials when they get the subsequent recognition test correct (as compared to those trials when they get the recognition test incorrect)? Dr. Hampton replied that he had not yet been able to examine those latency data. Herb Terrace then asked what is the duration of the delay. Dr. Hampton replied that the delay was 34 to 38s, titrated until the monkey is 70% correct.

Why do monkeys decline the memory test? There are probably many variables that are correlated with forgetting (and thus with the choice to decline the memory test) – e.g., noises, itches, changes in motivation, and absence of memory. To address whether absence of memory is the controlling factor, the next experiment manipulated memory directly. This experiment used the same paradigm as in previous experiment, with the exception that on some trials the sample was not presented. If the decision to take the recognition test is controlled by mnemonic introspection, then monkeys should always decline the recognition test on trials when the target stimulus is not presented. In fact, animals did decline on most of the trials on which no target stimulus was presented. The performance cannot be due to trial-and-error learning because there was no learning curve. In the case of one subject, the test was declined 100% of the time from the start of the condition. Herb Terrace then asked what about if the delay is increased? Dr. Hampton replied that they avoid the test more often as the delay is increased. Dr. Hampton concluded that these experiments provide some evidence that monkeys can “introspect” about their own memories.

The next experiments use the processes dissociation paradigm pioneered by Larry Jacoby to examine the role of PRh in explicit and implicit memory. The paradigm affords a quantitative estimate of the relative contributions of implicit and explicit memory to task performance.

A DMTS task was again used, but this time the same target stimulus was used repeatedly. There were, however, also “recollection” (explicit memory) trials on which a relatively novel stimulus served as the target. There were thus two processes that could control choice performance in a given recognition test: habit, which always favored the stimulus that was presented repeatedly, and “recollection,” which was an episodic memory of the particular target stimulus presented on that trial. By manipulating which type of stimulus served as the target (novel v. repeated), Hampton could manipulate whether the implicit and explicit memories were congruent or incongruent.

The analysis focused on errors. The PRh lesioned animals made significantly more errors on incongruent (i.e., novel target stimulus) trials than did controls. On error trials, controls and lesioned animals were equally likely to choose the habit stimulus. The results indicate that PRh lesions leave habit (implicit) learning intact but impair recollection (explicit learning).

Discussion:

Chris Sommerfield pointed out that in humans PRh is thought to represent novelty, and on recollection trials the stimuli are more novel than on habit trials. He asked whether novelty could explain some of the current results? Dr. Hampton replied that the “novel” stimuli were not in fact completely novel – they were relatively new. Also novelty is a representation issue, and some of the earlier experiments exclude such a role of PRh.

Dr. Terrace asked whether monkeys introspect in the regular DMTS task. Dr. Hampton replied he thinks the animals will introspect if it increases the chance of obtaining reward. Dr. Terrace followed by asking, if so, then why doesn’t regular DMTS performance demonstrate introspection on the part of monkeys. Dr. Hampton replied that, while it may be the case that the monkeys are introspecting, there is no expression of it in the task. That is, there is no need to posit introspection to explain their performance.

Dr. Horvitz asked whether the Hp lesion will impair nonspatial DMTS. Dr. Hampton replied that selective lesions to Hp will not. The literature is quite murky, however, because most lesion studies involve lesions that are somewhat nonselective.

Dr. Wallman asked whether you have to accurately know what you remember in order for it to be an explicit task? He suggested that the introspection experiment should include trials when subjects get the recognition test after they have chosen to NOT take the recognition test (i.e., when they have selected S-no-test). Dr. Hampton replied that this is an important idea, but earlier work on directed forgetting suggests some potential problems. In directed forgetting, animals are given a DMTS task, and on some trials a stimulus is presented during the delay to indicate that the recognition test will not be administered. On a few probe trials, this stimulus is presented but subjects are given the recognition test nonetheless. The subjects perform more poorly on these trials than on normal trials. Early interpretations focused on the idea that “directed forgetting” was occurring, and this implied that animals could control retention of information in memory. But another possibility was that on forgetting trials the recognition test was surprising (and there is evidence that this in itself can impair memory). So there would be problems with interpreting the data from trials on which unexpected recognition tests are given.

Prepared by Michael R. Drew, September 20, 2002

(Revised 10/7/02)