Memory impairment is a debilitating disorder that impacts on all aspects of daily life. The neuropsychological assessment of memory dysfunction that forms the basis of remediation programs is a long and arduous procedure. Hence, it is important that the processing underlying performance on these tests be fully understood. Currently, the processing underlying many neuropsychological tests has not been clearly explicated. The overall concern of this thesis was an evaluation of the Verbal Paired Associates (VPA) subtest of the Wechsler Memory Scale-Revised Edition (WMS-R) and the Wechsler Memory Scale-Third Edition (WMS-111).
Previous research, which has examined performance on the VPA in terms of episodic and semantic access, has not proved to be decisive. Thus, in this thesis memory processing was examined from a different perspective in that processing was examined by using a task analysis to identify the processing components of the VPA (WMS-R) and the VPA (WMS-111). Then performance on the VPA was compared to performance on verbal-learning tasks in order to isolate and elucidate each identified component process. The comparison-verbal-learning tasks were chosen for their similarities and differences in processing when compared to the VPA. That is, tasks were chosen that involved similar but not identical processing. Thus, instead of considering if memory impairment was a reflection of dysfunction in episodic or semantic memory, or in terms of dysfunction in a unified memory system, performance was examined in terms of processing required by the task itself.
The first experiment examined processing underlying performance on the VPA (WMS-R). The VPA (WMS-R) is a cued recall task comprising of eight word pairs, four of which were related, that is had a preexisting semantic association and the remaining word pairs were unrelated. The aim of the first experiment was to discover: first, if there was a broad episodic memory component present; second, the role of having word pairs with a preexisting semantic association. Third, was problematic performance a result of poor inter-item learning; and, fourth, did poor response production contribute to poor performance?
The comparison tasks were two combined pair and item recognition tasks, one consisting of related word pairs and as such was comparable with the easy pairs, and one consisting of unrelated word pairs and was comparable to the hard pairs of the VPA. There was also a cued recall with an extra-list associate recall task. As well as comparing performance across tasks, performance was also compared across memory-impaired and non-impaired groups. Thus, participants were 20 memory-impaired individuals who had been referred for neuropsychological assessment for memory problems, 20 individuals matched for age, gender and education and 20 first-year undergraduate students from the University of Queensland.
The second experiment replicated the first with the following changes, the combined recognition task was separated into pair and item recognition tasks, a cued recall with a part-word cue task was introduced and the comparison with the non-impaired groups was dropped. Participants were 40 memory-impaired individuals who had been referred for neuropsychological assessment.
Results from the two studies showed first, for the VPA (WMS-R) that there was no broad episodic component evident. Second, in terms of pairs with a preexisting semantic association, superior recall of the easy pairs of the VPA was a result of guessing not facilitation of learning that the word occurred together on the study list. Third, in contrast to expectations, a stronger positive correlation was demonstrated between the VPA and cued recall tasks rather than with pair recognition tasks.
The third study examined processing underlying performance on the VPA (WMS-111). The VPA (WMS-111) was a cued recall task comprising of eight unrelated word pairs. Thus, according to a task analysis, the processing components were episodic memory access, inter-item learning and response production. The aim of the experiment was first, to evaluate the processing underlying performance on the VPA (WMS-111). Second, to examine if processing underlying performance on the VPA (WMS-111) was the same as processing underlying performance on the VPA (WMS-R). Third, the effect on processing of having word pairs with and without a preexisting semantic association in one task was examined. Again performance was compared across verbal-learning tasks, which were yes/no pair and item recognition tasks, forced choice pair and item recognition tasks and cued recall with an extra-list associate and with a part-word cue tasks. In addition, a crossed-associates cued recall task was introduced.
Results from the third experiment demonstrated, in terms of processing underlying performance on the VPA (WMS-111), that first, in contrast to the VPA (WMS-R) there was a broad episodic memory component present. Second, the VPA (WMS-111), in contrast to the hard pairs of the VPA (WMS-R), demonstrated a stronger correlation with pair recognition and a weaker correlation with cued recall tasks. Thus, the VPA (WMS-111) was an episodic task with some but not overwhelming contribution from inter-item learning.
Overall, the results also appeared to suggest that the task analysis was useful in examining processing components underlying performance on memory tasks. Conclusions were that first, there was a difference in processing underlying performance between the VPA (WMS-R) and the VPA (WMS-111). Thus, comparison of performance on the hard pairs of the VPA (WMS-R) and the VPA (WMS-111) should be treated very cautiously. Second that mixing episodic and semantic information had both a direct and indirect effect on processing. Third, that memory-impaired participants, in particular those with more severe impairment, demonstrated a heavy reliance on semantic memory in the presence of semantic associations.
Treatment implications, in terms of the VPA (WMS-111), were: first, as a precaution or to eliminate a possible confusion, the pairs which have related information should be removed; second, the Recognition subtest did not appear to have any clinical utility; and third , for the VPA to be a measure of memory rather than cognitive impairment, the instructions should be simplified. In terms of rehabilitation programs, results supported the use of programs that capitalize on semantic memory such as errorless learning programs, instead of trial and error programs. In addition, results indicate that programs that utilize premorbid strengths or abilities would be more useful than those that ignored these abilities or tried to override those abilities.