The DMTS model employed in our laboratory is an adaptation and elaboration of that developed by Bartus and colleagues.21 Regarding the mnemonic functions that may be assessed using this model, some uncertainty likely exists. Clearly, standard DMTS or delayed matching-to-location tasks require the function of working memory, defined as retention of information needed to complete a given task at a given time. A large body of work conducted by Goldman-Rakic and others has identified many of the anatomical and neurochemical underpinnings of working memory12-14 and performance on these tasks is, indeed, altered by manipulation of these systems. However, the distinction between working and other complex forms of memory in completing these tasks is unclear. The complexity of most delayed recall task rules, as well as the extensive baseline training regimen in both rodents22 and non-human primates19,23 implies the involvement of reference memory in completion of each individual task. This is reflected in the requirement that, to complete a given trial, animals must retain information concerning the format of the task for periods encompassing years. Thus, this suggests then that the tasks described in this section may be characterized as more complex mixed-memory tasks.24
Animal Housing and Training — All monkeys (macaca mulatta and macaca nemestrima) used in our studies are housed at the Medical College of Georgia Animal Behavior Center in stainless steel cages composed of multiple 130 by 75 by 75 cm units. Toys and foraging tubes are provided and all monkeys are allowed to observe television programs each afternoon after testing to promote psychological well-being. During a test week, monkeys are maintained on a feeding schedule that allows approximately 15% of their normal daily food intake to be derived from banana-flavored reinforcement pellets awarded for correct responses. Testing is typically conducted on five days of each week. Standard laboratory monkey chow, fresh fruits, and vegetables compose the remainder of an animal's daily food intake following testing each day. Water is available ad libitum.
It is important to note that even with the continuous daily testing that monkeys are subjected to over the course of several months or years, no reinforcer devaluation has been observed in our animals. Previous animal studies have documented rein-forcer devaluation in rodents performing various instrumental tasks following over-training,25 however, it is likely that the dietary modification employed with our monkeys aids in maintaining reinforcer salience.
For initial training in use of the test apparatus and to develop competence in performing the delayed matching task, all monkeys are gradually presented with a program of acclimation, habituation, and basic rule learning in the test environment, their home cages. Other investigators have suggested that behavioral testing of nonhuman primates in home cages may prove distracting.18 Indeed, others typically perform delayed recall testing in non-human primates only following removal from the home cage, in a dedicated test chamber. However, we have not observed evidence that animals fail to attend to test panels. Further, monkeys can be trained to a very high degree of accuracy on shorter delay trials and recall accuracy (following different delays) across months and even years is quite stable. Perhaps most significant in this regard are our observations that latencies to respond to sample and choice stimuli are stable and typically less than 3 to 4 seconds in all animals.
Test panels were attached to the front of home cages such that animals had free access to the entire face plate of the panel. After initially placing panels on the front of home cages, animals were trained by shaping to approach the panel for a banana pellet reward delivered by a remotely operated feeder bin attached to the panel exterior. Following habituation to the panel and mechanical feeding apparatus, which typically requires several short sessions over the course of several days, monkeys were exposed to the first of several fully automated training programs incorporating the test stimuli. Stimuli used during all tasks were 2.54 cm diameter colored disks (red, yellow, or green) presented by light-emitting diodes located behind clear plastic push-keys positioned in a pyramidal shape on the face plate of the test panel. We have found the duration of training with each program to vary markedly between animals and some acceleration or reversion between programs may be necessary for individual animals. For each of the initial training programs described, no delay intervals between sample and choice push-key illumination are included. At each stage of training, it is essential to monitor patterns of responding for the presence of several different forms of strategic or reflexive behaviors that may be elicited by animals, particularly during transition to more demanding cognitive tasks. During the early stages of training, prior to attaining competence in the DMTS paradigm, we have observed limited numbers of monkeys, particularly aged monkeys, to employ strategies of side or color perseveration, or perseverative behavior with regard to sample key pressing. In many instances, the behavior appears to be a trial and error attempt to apply a basic, assumed rule to the task to obtain reward (i.e., red = reward). However, we have found that with counterbalancing for all side and color combinations for each delay interval, these forms of experimentation by the monkey can be quickly extinguished. Representative training programs and the order in which they may be employed are described below.
Matching with Delays — Completion of each of the training programs above, in succession, is followed by gradual, titrated imposition of variable delay intervals. Four possible delay intervals between a monkey's response to the sample light and the presentation of the two choice lights are employed: Zero seconds delay or a Short, Medium, and Long delays. The duration of these delay intervals should be gradually titrated to ensure continued reinforcement at a level above chance levels for delays other than long delays. Short, medium, and long delay intervals are individually adjusted to produce stable performance levels approximating the following levels of accuracy: short (75 to 85% correct); medium (65 to 75% correct); and long (55 to 65% correct). Monkeys performance for zero seconds delay trials typically averages 85 to 100% correct. Monkeys complete 96 trials on each day of
A. Stimulus push-key (top) is illuminated by one of the three colored lights in random order. Animals receive a food reward for depression of the stimulus push-key after illumination and another trial begins.
B. Stimulus push-key is illuminated. After depressing this key, no reward is given and the top light remains illuminated. One of two bottom push-keys is illuminated with the matching color. The other is not illuminated. Monkeys are rewarded for depressing; the error results in the start of a new trial.
C. Simultaneous Matching-to-Sample: Same as
Program B, with both choice lights illuminated. Monkey is rewarded for depressing the matching key. Press of the non-matching key extinguishes the trial and a new trial begins.
Correction: Same as above with the exception that incorrect choices result in presentation of the same color problem until it is completed correctly. Colors are varied for each trial.
E. Matching-to-Sample without Delay: Sample key is extinguished after pressed and choice keys are both lit. Colors are varied for each trial.
100 presses of sample key for reward each day (during one hour) for four consecutive days
80% stable accuracy in depressing correct, illuminated choice key for reward over 1 to 2 weeks *Color of sample is constant until monkey reaches 80% accuracy over several days. One of the other three colors is then substituted using the same criterion (color titration)
80% stable accuracy over 3 to 4 weeks * color titration
80% stable accuracy over 3 to 4 weeks
85% stable accuracy over 3 to 4 weeks testing with trials of each delay interval (including zero delay) presented an equal number of times. Typical delay intervals for aged and non-aged macaques, and their accuracy of recall following these intervals are illustrated in Figure 8.2.
The progression of a standard DMTS trial is as follows:
1. A trial begins with the illumination of the sample key by one of the colored disks. The sample light remained lit until the sample key was depressed by an animal, initiating one of four pre-programmed delay intervals, during which no disks were illuminated (i.e. no distractors present).
2. Following the delay interval, the two choice lights located below the sample key are illuminated. One of the choice lights matches the color of the sample light. These disks remain illuminated until a monkey presses one of the two lit keys.
3. Key-presses of choice stimuli that matched the color of the sample stimulus are rewarded by a 300 mg banana-flavored pellet. Non-matching choices are neither rewarded nor punished. A new trial is initiated 5 seconds after the second key-press on a preceding trial.
As may be seen in Figure 8.2, use of this method yields age-dependent differences in not only baseline DMTS performance (i.e., recall without delays), but also in the ability to accurately recall target stimuli after increasingly longer delay periods.
Was this article helpful?