Temporal structure in associative retrieval.

Zeb Kurth-Nelson,Dino Sejdinovic,Peter Dayan,Ray Dolan,Gareth Barnes

doi:10.7554/elife.04919

Zeb Kurth-Nelson, Dino Sejdinovic + Show 3 more

Open Access

https://doi.org/10.7554/elife.04919

Copy DOI

Abstract

Electrophysiological data disclose rich dynamics in patterns of neural activity evoked by sensory objects. Retrieving objects from memory reinstates components of this activity. In humans, the temporal structure of this retrieved activity remains largely unexplored, and here we address this gap using the spatiotemporal precision of magnetoencephalography (MEG). In a sensory preconditioning paradigm, 'indirect' objects were paired with 'direct' objects to form associative links, and the latter were then paired with rewards. Using multivariate analysis methods we examined the short-time evolution of neural representations of indirect objects retrieved during reward-learning about direct objects. We found two components of the evoked representation of the indirect stimulus, 200 ms apart. The strength of retrieval of one, but not the other, representational component correlated with generalization of reward learning from direct to indirect stimuli. We suggest the temporal structure within retrieved neural representations may be key to their function.

Highlights

Associative memory in animals and humans provides a model of the environment
We show that the neural representation of the indirect stimulus can be decomposed into at least two temporal components with distinct properties, and these are retrieved at different times during the Reward-learning phase
Using one-way ANOVA, we found that the raw amplitude, in single time bins, of the event-related field (ERF) at many individual sensors was significantly related to the stimuli alternated between photographs (Si) category (Figure 2). (The significance threshold was set to 95% of peak-level over space and time from 100 random category label shuffles, to correct conservatively for multiple comparisons.)

Summary

Introduction

Associative memory in animals and humans provides a model of the environment. Retrieval of such memories, driven by cues or occurring autonomously, is suggested as central to a wide variety of processes and functions, including online and offline planning and model-learning (Sutton, 1991; Moore and Atkeson, 1993; Foster and Wilson, 2006; Johnson and Redish, 2007; Hasselmo, 2008; Lisman and Redish, 2009; Gupta et al, 2010; van der Meer et al, 2010; Jadhav et al, 2012; Wimmer and Shohamy, 2012; Pfeiffer and Foster, 2013; Singer et al, 2013), cognitive search (Kurth-Nelson et al, 2012; Todd et al, 2012; Morton et al, 2013), mental time travel (Hopfield, 2010; Schacter et al, 2012), memory maintenance and consolidation (Marr, 1971; Nádasdy et al, 1999; Káli and Dayan, 2004; Kuhl et al, 2012; Deuker et al, 2013) as well as temporal expectation (Sakai and Miyashita, 1991; Rainer et al, 1999).Retrieval is classically linked to reinstantiation of a particular distributed spatial pattern of neural activity mirroring that evoked by the original experience of the object or context being retrieved (Tulving and Thomson, 1973; Nyberg et al, 2000; Hoffman and McNaughton, 2002; Polyn et al, 2005; Johnson and Rugg, 2007; Gelbard-Sagiv et al, 2008; Danker and Anderson, 2010; Rissman and Wagner, 2012; Miller et al, 2013; Kuhl and Chun, 2014). Electrophysiology experiments robustly demonstrate that when an object is directly experienced, the evoked pattern of neural activity evolves rapidly over tens to hundreds of milliseconds (Makeig et al, 1997; Schmolesky et al, 1998, 1998; Näätänen and Winkler, 1999; VanRullen and Thorpe, 2001; Rossion and Jacques, 2008; Schneider et al, 2008; Cichy et al, 2014). This implies that direct experience of an object evokes multiple distinct spatial patterns of neural activity in sequence. These distinct spatial patterns have never been identified independently at retrieval

Methods

Results

Conclusion