Memory Illusions and Memory Attributions : Behavioural and Electrophysiological Data

Abstract: Although memory often serves us well, it can also prove highly unreliable. The study of illusions and distortions of memory unveils important information about the nature of human memory functioning and, moreover, factors affecting memory accuracy. By taking a cognitive neuroscience perspective, the present dissertation aimed at elucidating processes promoting both true and false remembering and the neural underpinnings thereof. Paper I presents a Swedish version of the Deese-Roediger-McDermott paradigm for experimentally inducing false memories. Briefly, participants tend to falsely remember non-presented lure words (e.g., sleep) after having studied lists of the lures’ strongest associates (e.g., bed, rest, awake, etc.). The results of Experiment 1 confirmed that the developed stimulus material induces robust false recall and recognition effects at levels comparable to true memory. Furthermore, ratings of phenomenology indicated that true and false remembering were experienced similarly. Experiment 2 demonstrated that false recognition can be dramatically reduced when study conditions promote encoding of contextual information and when participants are encouraged to retrieve such information. In Paper II, a perceptual implicit memory test was employed to examine the notion that the non-presented lures may be activated as implicit associative responses during study of their associates. Participants solved anagrams constructed from previously studied lures, non-presented lures, and new words and judged their difficulty. Experiment 1 demonstrated false priming effects for the lures on both measures, suggesting an activation of the lures during study. However, the effects were eliminated in Experiment 2 that introduced an articulatory suppression task during study, but left the priming effects for studied items still reliable. Together, the results support the idea that the lures may be activated as covert verbal responses during study of the associates. Paper III used event-related potentials (ERPs) to examine the neural correlates of true and false recognition and differences in the susceptibility to false recognition. Varying levels of false remembering were examined as a function of encoding task (read vs. generate) and by dividing participants into “good” and “poor” performers. Drawing on known ERP memory effects, the results suggested that false recognition may be based on both familiarity and recollection, and that familiar lures may be rejected when contextual information is inaccessible. In addition, “good” performers may have been able to reduce their false remembering via post-retrieval evaluation. Since various memory distortions have been attributed to source-memory failures, this was examined in Paper IV. ERPs were recorded while participants made item- or source-memory judgements about previously perceived and imagined pictures. The results add to previous results suggesting prefrontal engagement in the retrieval of contextual information. Furthermore, the results support the notion that memory judgements may be based on different amount or types of information that may revive at different rates. The results are discussed in relation to theoretical accounts of memory and memory distortions—activation/monitoring theory; constructive memory framework; dual-process models; fuzzy-trace theory; source-monitoring framework. Consistent with previous proposals, it is argued that memory distortions may be considered as by-products of an adaptive memory system, rather than as flaws in design.