Skip to main content
. Author manuscript; available in PMC: 2015 Aug 12.
Published in final edited form as: Annu Rev Psychol. 2011 Sep 13;63:101–128. doi: 10.1146/annurev-psych-120710-100344

Figure 4.

Figure 4

Selective cortical reinstatement of content-specific activity patterns scales with the specificity of episodic retrieval, diminishes with interference from competing memories, and predicts subsequent memory outcomes. (A) Receiver operating characteristic curves depict the ability of an MVPA classifier, trained to discriminate face-vs. scene-related activity patterns measured from ventral occipitotemporal cortex during event encoding, to index the reinstatement of these patterns during cued associative retrieval. The degree of neural reinstatement tracked participants’ phenomenological retrieval experience, such that decoding performance was most robust when participants reported recalling the specific face or scene associated with a given cue word (Specific Recollection; AUC = 0.83), significantly lower when they only were able to recall the generic category (General Recollection; AUC = 0.75), and no better than chance when participants reported that they could not recall whether the associate was a face or scene (Don’t Know; AUC = 0.54). (B) Neural evidence for selective reactivation of the target category was diminished during competitive retrieval (AC trials) relative to non-competitive retrieval (AB trials). Histograms depict the mean distribution of trial-specific estimates of target category reinstatement and illustrate that the classifier’s predictions were less heavily skewed towards the target category when interference was present from an overlapping association. (C) Weaker reactivation of the C term (i.e., the target associate) during AC retrieval was linked to an increased likelihood that the competing AB associate would later be remembered in a post-scan memory test. This subsequent memory effect, which was observed regardless of whether AC retrieval yielded Specific or General Recollection, suggests that lower fidelity (i.e., less selective) reactivation during AC retrieval may in fact reflect the coactivation of both the target (C term) and competing (B term) associations. Adapted from Kuhl et al. (2011).