This research seeks to illuminate the mechanisms underlying human episodic memory through both computational modeling and experimental studies. Episodic memory is the ability to link the information that we experience with its temporal and situational context. The ability to do so places us within our memories, making them autobiographical. Failures of episodic memory are a hallmark of normal aging and neurodegenerative disease. The first aim of the empirical studies is to assess the influence of prior knowledge and memories of past events on people's ability to encode and retrieve newly learned information. A second aim is to examine how repetition influences memory at a mechanistic level and to explain why repetitions are most beneficial for memories that are widely distributed in time. In addressing both aims, the investigators will use and assess the context maintenance and retrieval model, using neural network models of how temporal context is represented in memory, how it evolves through experience, and how it interacts with semantic context and source context in the formation and retrieval of associative information.
Advancing the understanding of human learning and memory has implications for the diagnosis and eventual treatment of disease-related memory impairments such as Alzheimer's disease and other dementias. The work may also impact instructional technology and educational theory and practice.
Modern models of human memory assume that as we experience new events they become associated with an internal context representation (Lohnas, Polyn, and Kahana, 2015; Polyn, Norman, & Kahana, 2009; Healey & Kahana, 2015). The context representation drifts through a high-dimensional space driven by new stimuli in the environment. Because the drift is gradual, items presented in temporal proximity become associated with similar contextual states. The current state of this internal context allows us to recall events that were associated with similar contexts. Due to the slow drift, we tend to remember sequences of events that occurred close together in time (Howard & Kahana, 2002). These models have been extremely successful in accounting for the details of how people recall recently learned information. However, to date, these models have been tested using computer simulations, which assume memory extends only a few minutes into the past. This is because the computational demand of simulating a larger repository of memory has been prohibitive. However, by contrast, human memory is influenced by many past events. Therefore, we have developed an extension of a leading model—the Context Maintenance and Retrieval (CMR) model—that can simulate the influence of past knowledge on the recall of more recent events (Lohnas, Polyn, and Kahana, 2015). This new model, called the Context Maintenance and Retrieval Model Version 2 (CMR2), can simulate an entire session of free recall trials within a common associative network. CMR2 describes how memories accumulate over longer periods of time and provides a rigorous account of important findings in the science of memory, such as how existing memories can interfere with recall of new memories. A second goal was to examine CMR2's ability to account for a wide-range of learning and repetition effects. The core insight of CMR2 is that repetition effects arise from the storage of multiple item-to-context and context-to-item associations (Lohnas & Kahana, 2014a). Findings confirming this prediction were reported in the Journal of Experimental Psychology: Learning, Memory, and Cognition (Lohnas & Kahana, 2014a). In a separate paper, we found evidence for another model prediction: that the cue for recall of an item is a weighted sum of recently activated cognitive states. Confirming this prediction, we found that temporal contiguity effect is greater when the two most recently recalled items were studied in contiguous list positions (Lohnas & Kahana, 2014b). The third key goal of the grant was to develop a better understanding of why memory ability varies among individuals. It is well known that variation among individuals in overall success on memory tasks is highly predictive of general intelligence, but it is not known which specific memory processes underlie this correlation. Previous work, both in our lab and that of other researchers, has used the free recall task to provide a window on the processes that underlie memory search by measuring how participants use long-standing and newly formed associations to guide memory search. In several publications, we examined individual differences in these measures. We found that individuals show qualitatively similar patterns of memory search with individual differences taking the form of quantitative variation (Healey & Kahana, 2014). This quantitative variation, particularly in the tendency to use new associations between events and their temporal context, is highly predictive of both overall memory ability and general intellectual ability as measured by IQ tests (Healey, Crutchley, & Kahana, 2014). To facilitate further research on these issues, we have created a data archive (http://memory.psych.upenn.edu/Data_Archive) through which all data from our publication are freely available to other scientists. Already, several independent researchers have published analyses of data from the archive. References: Healey, M. K., Crutchley, P., and Kahana, M. J. (2014). Individual differences in memory search and their relation to intelligence. Journal of Experimental Psychology: General, 143(4), 1553–1569. Healey, M. K. and Kahana, M. J. (Under revision). A new approach to understanding age–related memory impairments. Healey, M. K. and Kahana, M. J. (2014). Is memory search governed by universal principles or idiosyncratic strategies? Journal of Experimental Psychology: General, 143, 575–596. Howard, M. W. and Kahana, M. J. (2002). When does semantic similarity help episodic retrieval? Journal of Memory and Language, 46, 85–98. Lohnas, L. J. and Kahana, M. J. (2014a). Compound cuing in free recall. Journal of Experimental Psychology: Learning, Memory and Cognition, 40(1), 12-24. Lohnas, L. J. and Kahana, M. J. (2014b). A retrieved context account of spacing and repetition effects in free recall. Journal of Experimental Psychology: Learning Memory and Cognition, 40(3), 755-764. Lohnas, L. J., Polyn, S. M., and Kahana, M. J. (Under review). Expanding the scope of memory search: Intralist and interlist effects in free recall. Polyn, S. M., Norman, K. A., and Kahana, M. J. (2009). A context maintenance and retrieval model of organizational processes in free recall. Psychological Review, 116, 129-156.
