Weakening memories by half-remembering them

During retrieval, memories activate to different degrees and compete to be recalled. We suggest the following hypotheses: memories that activate highly win the competition and are strengthened; memories that activate moderately lose the competition and get weakened; the closer the competition, the greater the consequent strengthening and weakening; and memories that do not activate do not compete, and are unaffected.;In Chapter 1, we review the behavioral findings (Anderson and Spellman, 1995; Anderson and Green, 2001) that suggested such a nonmonotonic relationship between level of activation and subsequent accessibility, and three possible accounts of it (Anderson et al., 2004; Tomlinson et al., 2009; Norman et al., 2006).;In Chapter 2, we describe four behavioral experiments designed to control the degree to which a memory activates in order to cause it to be forgotten. Experiment B1 (no significant effect) used presentation duration in an RSVP task to attempt to control activation. Experiment B2 (no significant effect) introduced a novel 'watermark' suppression task with which we attempted to release the learning of new associations from proactive interference. Experiment B3 (no significant effect) tested whether competition-driven learning when forming new associations caused the old associations to be weakened. Experiment B4a (significant effect) replicated the Depue et al. (2007) think/no-think paradigm with emotional stimuli. Experiment B4b (significant effect) introduced a novel 'graduated exposure watermark task'.;In Chapter 3, we describe how the lessons learned from two fMRI pilots shaped the design of Experiment F7.;In Chapter 4, we describe Experiment F7 (significant effects), which applied MVPA and region-of-interest (ROI) methods to fMRI to provide a covert, neural measure of a memory's activation within the think/no-think paradigm, and thus to predict its subsequent accessibility.;In Chapter 5, we consider the lessons learned from this series of experiments, and propose future work.