A decline in memory efficiency is one of the principal complaints associated with normal aging, and the validity of this complaint is generally confirmed when older adults are assessed on traditional memory tests that require direct recollection of specific events (e.g., recall and recognition tests). However, when memory is assessed via indirect memory tests that do not involve direct or intentional recollection, older adults often perform at comparable levels to young adults. This pattern of "preserved versus impaired" memory functioning (i.e., an age-related dissociation) has intrigued cognitive aging researchers and stimulated a great deal of theoretical debate.
There are currently two major theoretical interpretations of age-related dissociations. The first approach, systems-theory, proposes that direct-indirect test dissociations reflect the operation of structurally different memory systems in the brain. Two relevant systems are identified: an episodic system that processes contextual and associative information, and a perceptual-representation system (PRS) that processes modality-specific information about the structure of incoming stimuli (Schacter & Tulving, 1994). Test instructions determine the type of information (contextual vs. a-contextual), and memory system (episodic s. PRS) accessed during memory performance. The alternative, processing-based approach argues against the notion of different memory systems - hypothesising instead that dissociations arise because direct and indirect tests employ different types of retrieval cues (e.g., word-fragments versus semantic cues), which subsequently invoke different types of processing at test (conceptually driven vs. data-driven) (Roediger & Blaxton, 1987).
While both of these approaches have their proponents in the cognitive aging field, it is our contention that such simple, single-factor models are unable to account for all patterns of age effects on direct and indirect tests. Multi-factor models that conjointly emphasise (a) test instructions and (b) processing are advocated instead (Cohen, 1997; Humphreys, Bain & Pike, 1989). Such models have not been previously utilised in the cognitive aging field. Thus, a central goal of this thesis is to examine the validity of single-factor vs. multi-factor models by determining the minimum number of test parameters (one vs. two) necessary to explain patterns of age-related performance on direct and indirect tests. Three task parameters were assessed in the thesis, namely (a) the nature of the test instructions (i.e., direct vs. indirect), (b) the type of processing invoked by the test cues (data-driven vs. conceptually driven); and (c) the type of target stimuli (i.e., verbal vs. iconic).
At an empirical level, the thesis also sought to introduce a number of novel methodological improvements to the field. These improvements comprised (a) the use of multi-task methodology (Experiment 1) to provide convergent validity for memory constructs and task parameters, (b) the application of the "retrieval intentionality criteria" to control for potential contamination on indirect tests, (c) the use of word norms to control for response competition confounds at test, and (d) the development of more rigorous testing procedures in the iconic, nonverbal domain (i.e., brief visual presentations and abstract, geometric stimuli).
In terms of the structure of the thesis. Chapter 1 provides an introduction to the field of direct and indirect memory research. It describes the standard methodology used to assess direct and indirect memory, and then reviews the extant literature on aging and direct and indirect memory. Chapter 2 provides an overview and critique of the primary theoretical accounts that have been developed to explain direct and indirect memory phenomena. The chapter then concludes with a statement of research goals. These goals are implemented over the course of three experiments in the thesis.
The first experiment is described in Chapter 3. This experiment sought to determine which task parameter(s) govern age-related performance on direct and indirect tests in the verbal domain. To accomplish this goal, the experiment compared the performance of young and older adults on a number of memory tests that were varied according to (a) type of test instruction (direct vs. indirect), and (b) type of process (data-driven vs. conceptually driven). Results indicated that age effects were primarily mediated by test instructions (i.e., age effects only emerged on direct memory tests), although some additional variance was explained by the processing parameter, such that age effects on the direct memory tests were more pronounced following semantic (conceptual) encoding than nonsemantic (data-driven) encoding. This pattern of results supports the concept of a multi-factor theory, rather than either of the single-factor theories currently advocated by aging researchers.
The second experiment (Chapter 4) sought to examine age effects on direct-indirect tests in the iconic (visuo-spatial) domain. The primary goal was to determine whether test instructions would continue to mediate patterns of age-related performance when abstract, geometric stimuli were employed in lieu of traditional verbal stimuli. Results reconfirmed the importance of the instructional parameter (i.e., direct vs. indirect test instructions), with age-related decrements only emerging on the direct memory tests.
The final experiment (Chapter 5) sought to examine the link between test instructions and age-related memory dissociations. Multi-factor theorists, such as Cohen (1997) and Humphreys et al. (1989) have hypothesised that direct memory performance is dependent upon contextual/relational processing. Under direct test conditions, participants need to access two linked things: the target items linked with a contextual representation of the learning episode. In order for this to occur participants must first bind feature representations of the study context with item representation acquired during the study phase. Later at test, they must mentally reinstate the appropriate context for use as a retrieval cue to help access the relevant (associated) item information. Extrapolating from these theories, it was hypothesised that age-related deficits on direct memory tests occur because older adults are less efficient at processing relational-contextual information than young adults. To test this hypothesis. Experiment 3 compared young and old adults on a list-specific recognition paradigm that manipulated temporal (list-specific) context across four successive learning trials. Focus was placed upon temporal context, for two reasons. First, little gerontological research has been conducted on temporal context, especially in relation to direct memory tests. Second, direct test instructions often make explicit reference to specific temporal or list contexts for example the time of day, or date from which target information must be retrieved. Thus, success on these tests is predicated not just on the ability to recollect details about individual target items, but on the ability to recollect temporal or list information as well. Results confirmed the contextual- processing hypothesis: older adults were indeed less adept than young adults at using temporal context information to guide their performance on direct memory tests.
Finally, the discussion chapter (Chapter 6) (a) summarises and integrates the empirical findings from the three experiments, (b) evaluates the single-factor vs. multi-factor theories in the light of the obtained data, and (c) discusses research limitations and ideas for future research. Overall, results support the view that multifactor models are necessary to account for age differences in performance on direct and indirect memory tests.