Brain activation during associative short-term memory maintenance is not predictive for subsequent retrieval.

Heiko C Bergmann,Mark Rijpkema,Sarah F Beul,Sander M Daselaar,Guillén Fernández,Roy P C Kessels

doi:10.3389/fnhum.2015.00479

Abstract

Performance on working memory (WM) tasks may partially be supported by long-term memory (LTM) processing. Hence, brain activation recently being implicated in WM may actually have been driven by (incidental) LTM formation. We examined which brain regions actually support successful WM processing, rather than being confounded by LTM processes, during the maintenance and probe phase of a WM task. We administered a four-pair (faces and houses) associative delayed-match-to-sample (WM) task using event-related functional MRI (fMRI) and a subsequent associative recognition LTM task, using the same stimuli. This enabled us to analyze subsequent memory effects for both the WM and the LTM test by contrasting correctly recognized pairs with incorrect pairs for either task. Critically, with respect to the subsequent WM effect, we computed this analysis exclusively for trials that were forgotten in the subsequent LTM recognition task. Hence, brain activity associated with successful WM processing was less likely to be confounded by incidental LTM formation. The subsequent LTM effect, in contrast, was analyzed exclusively for pairs that previously had been correctly recognized in the WM task, disclosing brain regions involved in successful LTM formation after successful WM processing. Results for the subsequent WM effect showed no significantly activated brain areas for WM maintenance, possibly due to an insensitivity of fMRI to mechanisms underlying active WM maintenance. In contrast, a correct decision at WM probe was linked to activation in the “retrieval success network” (anterior and posterior midline brain structures). The subsequent LTM analyses revealed greater activation in left dorsolateral prefrontal cortex and posterior parietal cortex in the early phase of the maintenance stage. No supra-threshold activation was found during the WM probe. Together, we obtained clearer insights in which brain regions support successful WM and LTM without the potential confound of the respective memory system.

Highlights

Recent years have seen renewed debate and controversy over the underlying neural substrates of working memory (WM) and itsdependence from long-term memory (LTM)
Results for the WM contrast ‘‘corrected’’ for longterm memory (LTM) performance showed that, not unexpectedly, no clusters survived the statistical threshold for the WM maintenance period
Our findings further suggest that engagement of the dorsolateral prefrontal cortex may only play a temporary role during the maintenance phase, i.e., during the early stage of WM maintenance, rather than persistently across the whole maintenance phase

Summary

Introduction

Recent years have seen renewed debate and controversy over the underlying neural substrates of working memory (WM) and its (in)dependence from long-term memory (LTM). The active manipulation and updating of information may rely on dorsolateral prefrontal activation (e.g., Fuster, 1973; Kessels et al, 2000), whereas load-related activation and retrieval may rely on the posterior parietal cortex (Berryhill, 2012; Postle, 2015) In addition to these brain regions that are typically studied in the WM literature, recent studies suggest that the hippocampus may be important in some aspects of WM processing. Due to its anatomical characteristics and extensive reciprocal connectivity with polymodal neocortical association areas (Suzuki and Amaral, 1994), the hippocampus plays a vital role in relational memory in general This involvement may be unrelated to the delay length between presentation and test (Konkel and Cohen, 2009). This view appears to be both supported by recent lesion studies (Holdstock et al, 1995; Giovanello et al, 2003; Crane and Milner, 2005; Hannula et al, 2006; Nichols et al, 2006; Olson et al, 2006a,b; Hartley et al, 2007; Piekema et al, 2007; Rose et al, 2012; van Geldorp et al, 2014, 2015), intracranial EEG and MEG (Axmacher et al, 2007, 2010a,b; Cashdollar et al, 2009) and by fMRI studies (Kirwan and Stark, 2004; Ranganath et al, 2005; Nichols et al, 2006; Piekema et al, 2006, 2009, 2010; Axmacher et al, 2008, 2009; Hannula and Ranganath, 2008; Olsen et al, 2009; Oztekin et al, 2009; Schon et al, 2009; Luck et al, 2010; Libby et al, 2014) which all demonstrated hippocampal involvement in either ‘‘typical’’ relational WM tasks or tasks most likely involving relational binding processes (often using delayed-match-to-sample tasks)

Methods

Results

Discussion

Conclusion