An Investigation of the Relation Between Remembering and Learning
Remembering requires an awareness of prior occurrence. In contrast, learning is indicated by savings on performing a task; no awareness of prior occurrence is necessary. Previous research has shown that performance on measures of learning can be functionally and statistically dissociated from performance on remembering tasks. Some investigators have concluded that these dissociations indicate that there are separate memory systems. The experiments performed in this thesis investigate the alternative explanation that dissociations between measures of memory result because of differing retrieval requirements. Whereas previous experiments employed learning and remembering tasks that were mismatched in their reliance on conceptual processes, the present experiments employ a learning task that focuses on the conceptual relations between words.
Meaningfulness of study processing was manipulated by requiring categorization of word pairs as similar or dissimilar. When the same categorization was performed at transfer, learning was of greater magnitude and of longer duration for more meaningfully related words. When repeated words were categorized by different attributes virtually no learning was observed. Therefore, in contrast to other research (e.g., Jacoby & Dallas, 1981), the meaningfulness and context of processing words were important determinants of learning.
The relation between remembering and savings on categorization was investigated by requiring a recognition decision after each categorization at transfer. Effects on recognition paralleled those on categorization. Better recognition performance was observed for more meaningfully related words and for words repeated in the same task context. Discrimination of task context also was better for more meaningfully related words. Therefore, the effect of equating processing requirements between measures was to produce a functional dependence between remembering and learning. A statistical dependence between measures also was obtained. At transfer, faster than average categorizations were associated with "old" recognition decisions in conditions in which subjects based their recognition decision on familiarity only. However, the conclusion of statistical association is tentative because the requirement to recognize after categorization interfered with categorization.
Manipulations of retrieval processes were successful in converting dissociations into associations. Therefore, this investigation supports the differing retrieval requirements explanation of dissociations between measures of memory.