|
Ashby, F.G., Noble, S., Filoteo, J.V., Waldron, E.M., & Ell, S.W. (2003) Category Learning Deficits in Parkinson’s Disease. Neuropsychology, 17, 1, 115-124. Purpose A wide variety of evidence suggests that human category learning relies on multiple systems. This belief is grounded in empirical evidence. One type of category learning is explicit reasoning (rule-based), while another type of category learning is based on implicit reasoning. It appears that the rule-based system is mediated by frontal-cortical structures and by the head of the caudate nucleus, while the implicit system is mediated by the tail of the caudate. Dopamine facilitates the operations of both systems and Parkinson’s disease (PD) patients have a decrease in dopamine level. It is assumed that the damage resulting from PD first affects, and is the most severe in, the head of the caudate nucleus. Because of the location and the severity of the damage, PD patients may be impaired in categorizing via explicit rules. It has also been shown that PD patients suffer impairment when having to categorize by using implicit reasoning. Even though both types of impairments occur in PD patients, they have only been shown in separate studies. The purpose of this study was to look at both types of category learning deficits within the same PD patients to see if one type of categorization was more impaired. Experimental work The experiment involved 16 PD patients and 15 older, normal controls (ONC). Older, normal controls were used because dopamine levels decrease by 7 to 8% for every decade that we age. The two groups were compared to each other and also to a group of younger, normal controls (YNC) who were tested in a separate experiment. For the experiment, card-like stimuli that varied on four binary-valued dimensions were utilized. The four dimensions were background color (blue or yellow), symbol color (red or green), the number of symbols (1 or 2), and symbol shape (square or circle). Two different tasks were used; a rule-based task and an information-integration task. Rule-based tasks look at how well/poorly someone is at explicit category learning. The rule needed to maximize accuracy is one that is both easy to test and to describe verbally. In this experiment, the rule necessary for accuracy involved only one dimension (i.e., respond A if number of symbols is 1, and respond B if number of symbols is 2). The information-integration task requires an implicit understanding (as opposed to an explicit understanding) of the rules in order to maximize accuracy. In this type of task, it is almost impossible to verbalize the rule. In the information-integration task, one dimension was considered irrelevant and each level of the other three relevant dimensions was assigned a value (either 1 or 0). Subjects then had to categorize based on the dimensions summing greater than 1.5 (category A) or less than 1.5 (category B). The PD patients and ONC participants were exposed to two explicit rules of sorting and two information-integration rules of sorting. Learning was spaced over one week and the subjects were exposed to explicit rules in the first session and the implicit rules in the second session. For each trial, the subjects were shown one of the stimuli and had to respond by assigning the stimulus to category A or B. Each subject was given feedback of ‘correct’ or ‘incorrect’ with distinct tones. Learning trials ended when the subject either got 10 responses correct in a row or when 200 stimuli had been presented without the subject getting 10 correct responses in a row. A new trial started after criterion had been met. Although the subjects knew they would start a new trial during the experiment, they were not informed exactly when the new rule was in effect. The YNCs participated in the same type of experiment, the only difference being that each YNC subject completed only the rule-based tasks or the information-integration tasks. Subjects were classified as either “learners” or “nonlearners” for each type of task. A participant was considered a learner in a particular task if he/she was able to learn both rules in that task. Failing to learn both rules for a given type of task qualified the subject as a nonlearner. In general, the PD patients and YNCs failed equally as often in both types of tasks. However, the ONCs failed more frequently in the information-integration tasks than in the rule-based tasks. With respect to the rule-based categories, there were more PD patient nonlearners than there were YNC nonlearners or ONC nonlearners. Half of the PD group failed to learn at least one of the rules in the task while only 14% of the ONC group failed to learn one of the two rules. In the information-integration tasks, both the ONCs and the PD patients did worse than the younger adults, but they did not differ significantly from each other. There was also a difference between the groups in their averaged trials-to criterion (for the learners) performance. It took the ONC learners longer than the YNCs in the rule-based tasks to reach criterion, but the PD patients did not differ from either of the groups. Also, all groups were not significantly different in the information-integration tasks. Conclusion Compared to the ONCs, the PD patients performed much worse on the simple rule-based tasks, but not on the more complex information-integration tasks. Compared to the YNCs, the ONCs behaved similarly on the rule-based tasks, but worse on the information-integration tasks. Perhaps only a distinct subgroup of PD patients did worse than the ONCs. When level of PD was correlated with ‘learning’ the task, the PD patients rated at the higher level were more likely to fail to learn the explicit rule. The authors give two reasons as to why an information-integration category –learning deficit occurred: 1) previous studies have shown it to occur, and 2) the brain area responsible for the information-integration learning is thought to be damaged in PD patients. Other experimental evidence is used to argue that multiple systems mediate different types of category learning. It has been shown that, without feedback, subjects can learn rule-based categories, but they cannot learn information-integration categories. With delayed feedback (over 5 seconds), people are able to learn explicit rules but not implicit rules. Also, if the experiment requires subjects to observe someone else learning the rule instead of actually doing the task themselves, the subjects can still accurately state the rule in the explicit task, but they are impaired in the implicit task. The authors argue that this is not simply because one task is more difficult than the other. For example, the PD patients actually did worse on the relatively easy rule-based task (as compared to the ONCs). The authors argue that this and other studies support the theory that rule-based and information-integration category learning are mediated by separate systems. However, there are two other possible interpretations of the results. One interpretation is that PD does not cause a specific impairment, but rather a general impairment in cognition. This cannot be the case because a general impairment would make the difficult task more difficult, and PD patients did significantly worse on the easy task and the same as the ONCs on the hard task. A second possibility is that the subjects memorized the 16 exemplars along with the feedback. However, it has been shown that when the task is made difficult to the point of overextending working memory, subjects do the same as those who are able to use working memory. Discussion questions: Is there a problem with the fact that the younger adults were tested on only one type of task? Perhaps if they had also been tested on both tasks, we would see them also doing worse on the second, harder task. Notice that the harder task was always tested after the easier one in both the ONCs and the PD patients. Could this be a reason why no difference between the two groups was found? Would the results be different if the order of the IVs were random? Participants were considered “learners” of a particular type of task if they learned both rules for that task. Participants were considered “nonlearners” of a rule type if they could not learn at least one of the rules for that particular type. What happened to the subjects who did learn one rule? Do you think that their data should have been looked at? Why or why not? |
|
This page was last updated: |