The impact of long-term visual representations on consolidation in visual working memory

Visual working memory (VWM) is the cognitive mechanism that encodes, maintains, manipulates, and retrieves visual and spatial information over the short-term. In order to prevent visual representations in VWM from being overwritten by subsequent visual information, some process must stabilize the activated representations so that they are less susceptible to interference. This process has been termed short-term consolidation (Jolicoeur & Dell'Acqua, 1998; Vogel, Woodman, & Luck, 2006). The current dissertation examines this short-term consolidation process, specifically whether or not long-term visual representations facilitate the stabilization of visual representations in working memory.;Four experiments were conducted that compared performance between familiar and unfamiliar stimuli in a backward masking paradigm that provided information on the time course of consolidation as well as how the process varied depending on the number of to-be-remembered items in the memory array. In the backward masking paradigm, a set of visual masks are presented at varying intervals following a memory array (17, 134, 250, 367, and 484 ms) before the test array is shown and participants make a same/ different judgement (i.e., did one of the items change between study and test). Varying the presentation of visual masks provides insight into the consolidation process: masks shown soon after the memory array will decrease performance because the activated representations in VWM have not been stabilized to reduce interference from subsequent visual information. Additionally, the set size of the visual array was manipulated (1, 2, 3, or 4 items) to examine how the impact of the visual masks varied depending on how many items there were in the array.;In Experiment 1, simple, colored squares were compared to complex, random polygons as an extreme comparison between stimuli that vary in terms of familiarity. This experiment showed an advantage for squares over polygons in both accuracy and capacity. Additionally, familiar squares were consolidated faster into VWM than random polygons. Experiments 2a and 2b equated for the type of stimulus by training on a subset of random polygons and comparing those trained items with novel items. In Experiment 2a, participants trained using a four alternative forced choice task (in which participants view an item then pick that item out of a set of four) and an advantage was demonstrated for trained items in both accuracy and capacity as well as faster consolidation for trained items. Additionally, there was no interaction between set size and delay for trained items, suggesting that participants were able to consolidate the array as a single unit instead of requiring more time for more items. In Experiment 2b, participants trained using a change detection task and no differences were demonstrated between trained and untrained items, including no interaction between set size and delay for either stimulus type, demonstrating transfer of change detection training to novel stimuli and evidence that participants created a unitary configuration of the items with increased training. To examine the impact of creating a configuration of items, Experiment 4 used an incidental learning paradigm in which the same four shapes were shown in the same spatial locations on every third trial and those repeating items were compared with novel items. With this repeating context, familiar items were remembered better than novel items and the impact of masks did not vary depending on the number of items for familiar items only. This supports the implication from Experiments 2a and 2b that spatial relations between items is part of the consolidation process.;Overall, these results support both the Unitary and Workspace Models of WM. Additionally, there are three primary conclusions from the current set of experiments. First, long-term memory plays an early role in VWM by facilitating the process involved in stabilizing WM representations. Second, short-term consolidation is a multidimensional process that strengthens both visual and spatial relational information to create more stable representations of perceptual input. Finally, training of visual stimuli can transfer to novel stimuli depending on the type of training task used and how similar it is to the criterion task. Taken together, these conclusions provide strong evidence that WM and LTM are highly interactive and provide unique insights as to how these processes work together to facilitate everyday, complex cognition. (Abstract shortened by UMI.).