Retrieval orientation alters neural activity during autobiographical memory recollection
Introduction
Autobiographical memories are mental representations of past experiences that are reconstructed at retrieval (Schacter and Addis, 2007; Schacter, 2012; Sheldon and Levine, 2016). These memories are richly detailed and contain a diverse array of information – including sensory perceptual and contextual event (episodic) details as well as factual knowledge associated with a particular event (semantic details; Tulving, 1983). Autobiographical memory construction relies upon a distributed network of brain regions that includes the hippocampus, medial prefrontal, middle and lateral parietal as well as temporal regions to access this content (Svoboda et al., 2006). While it is known that the hippocampus is critical for associating together the various details of an experience into a memory representation (Moscovitch, 1992; Winocur and Moscovitch, 2011), considerably less is known about how these details (i.e., memory content) are represented within this greater neural network.
One theory is that there are distributed functional neural subsystems prioritized to process specific episodic components of a remembered event, such as subsystems that support accessing the general concept of what happened to the contextualized details of the remembered environment (Cabeza and Moscovitch, 2013; Sheldon et al., 2019). Support for these subsystems comes from research examining neural support during internally-directed cognitive tasks – those that involve directing cognition towards self-generated thoughts and memories rather than information present in one's external environment (Dixon et al., 2014). These tasks typically activate the default network (DN) – a resting-state network that overlaps considerably with the autobiographical memory network (Buckner et al., 2008; Spreng and Grady, 2009). When internally-directed tasks are directed towards processing conceptual and contextual information, there is evidence that dissociable subsystems within the DN will be active (n.b., other resting state subsystems have been proposed, see Lee et al., 2012). The dorsal medial subsystem comprised of the dorsal medial prefrontal cortex (dmPFC), temporal pole, lateral temporal cortex, and temporoparietal junction processes self-referential and conceptual information and is engaged during evaluative decision-making. The medial temporal subsystem comprised of the hippocampal formation, parahippocampal cortex, retrosplenial cortex, posterior inferior parietal lobule, and ventrolateral prefrontal cortex (vmPFC) processes contextualized information that critically supports episodic memory (Andrews-Hanna et al., 2010; Andrews-Hanna et al., 2014).
The above-reported division within the DN parallels research that has reported distinct medial temporal lobe (MTL) subsystems for accessing conceptual and situational (i.e., contextual) elements of learned stimuli (Ranganath and Ritchey, 2012; Reagh and Ranganath, 2018). Specifically, there is a memory system that is activated for recalling the conceptual and semantic content of a past episode (an anterior subsystem including the perirhinal, temporopolar, and lateral orbitofrontal cortices) and one for recounting the specific situational context associated with an event (a posterior medial subsystem including the parahippocampal cortex, retrosplenial cortex, vmPFC, precuneus, and angular gyrus; Ranganath and Ritchey, 2012).
We recently proposed a framework (Sheldon et al., 2019) positing that these dissociable systems are engaged during different forms of episodic autobiographical remembering. This framework was based on a prominent theory for autobiographical memory organization in which the conceptual and the contextualized aspects of an event are organized at different ‘hierarchical levels’ within one's autobiographical knowledge base (Conway, 2001, 2009; Conway and Pleydell-Pearce, 2000). However, with our framework, we suggest that these different aspects of an event are stored at the same level and can co-exist rather than being nested within one another as described in the original model. With a non-hierarchical structure, distinct episodic memory representations of the same autobiographical experience can be built as a function of the relative weight assigned to conceptual (i.e., event-based) versus contextual (i.e., perception-based) event-specific knowledge accessed from one's autobiographical knowledge base (see Fig. 1). This weighting will thus determine how dissociable conceptual and contextual neural networks are recruited during remembering.
We propose that the bias towards these conceptual and contextual neural networks during autobiographical remembering can be determined by a person's retrieval orientation – a retrieval goal-state that is defined by a particular cue to probe a memory (Herron and Rugg, 2003; Morcom and Rugg, 2012). Retrieval orientation effects have been well-established with laboratory-based memories that manipulate orientation by altering the cues used to reactivate memory representations and have shown how retrieval orientation changes the engaged cognitive (Rugg and Wilding, 2000) and neural processes (e.g., Herron, 2018; Herron et al., 2016; Herron and Rugg, 2003; Morcom and Rugg, 2012; Robb and Rugg, 2002; Rugg and Wilding, 2000). This line of work has shown that retrieval orientations biased towards certain modalities (words, pictures) can change the underlying cognitive and neural processes engaged (e.g., Herron and Rugg, 2003). In addition, findings have shown that retrieval orientations biased towards certain features of studied items, such as the conceptual versus perceptual aspects of those items, also relate to distinct neural correlates during retrieval (e.g., Stenberg et al., 2006). This latter body of work aligns with the conceptual and contextual aspects of autobiographical memories and the finding that a conceptual retrieval orientation relied on anterior aspects of the brain and the perceptual retrieval orientation relied on more posterior aspects of the brain (Stenberg et al., 2006) fits with the proposed distinctions within the autobiographical memory neural network.
In this study, we aimed to test if autobiographical memories retrieved with an orientation biased towards conceptual versus contextual event details will result in dissociations in brain activity. In testing for these dissociations emerging from one's retrieval orientation during autobiographical memory retrieval, we considered additional factors that influence the neural support of autobiographical memory retrieval. One factor that may influence how retrieval orientation affects brain activity is the age of a recollected autobiographical memory (Nadel et al., 2007; Squire et al., 2004). Although evidence favours the view that the same brain systems are needed to construct detailed recent and remote memories (Nadel et al., 2007; Winocur and Moscovitch, 2011), it is not clear if memories from different life periods are modified by retrieval orientation in the same way. Another factor we consider is when during the memory retrieval process retrieval orientation effects are most pronounced – i.e., when retrieval orientation specifies how a memory representation must be initially built or when retrieval orientation updates an existing memory representation. There is evidence that retrieval makes a underlying memory representation malleable and more open change (Alberini, 2011; McKenzie and Eichenbaum, 2011; Nader and Hardt, 2009; Sara, 2010). If this is the case, then retrieval orientation effects should manifest most robustly when reconstructing a memory representation for an already-recalled memory – as this representation will be more labile and susceptible to change – than when constructing a memory representation for a newly-accessed memory. Alternatively, if retrieval orientation effects primarily influence how a memory is cued (Herron and Rugg, 2003), then orientation effects should present most strongly when retrieval orientation specifies how a new memory representation should be initially constructed.
We designed a novel functional magnetic resonance imaging (fMRI) experiment to specify the effect of retrieval orientation on the neural mechanisms of episodic autobiographical memory retrieval and to determine if this effect depends on a.) the age of the autobiographical memory and b.) the stage of memory retrieval. In the scanner, participants recalled twelve pre-selected recent and remote autobiographical memories equated for a number of subjective ratings (vividness, emotionality, importance, and rehearsal) under conceptual and contextual retrieval orientation conditions that included four sequential retrieval phases (see Fig. 2): (1) recollecting the autobiographical memory without any orientation cue (full recollection); (2) re-orienting to the conceptual (the theme/purpose of the event) or contextual (the placement of people and things in the location) elements of that memory (re-oriented recollection); (3) accessing a new and related memory using these conceptual or contextual elements from the recollected memories phase (new memory access); and (4) recollecting the conceptual or contextual elements of this new memory (oriented recollection). In this way, our retrieval orientation manipulation was carried out at two separate timepoints during the retrieval process – first, this manipulation was introduced after an unbiased memory representation had already been established (re-oriented recollection) and, second, the manipulation was introduced prior to the memory representation being built (oriented recollected) – allowing us to establish when during the memory recollection process retrieval orientation exerts the strongest influence on brain activity. Thus, the overall goal of our manipulation was to shift the relative emphasis of the conceptual versus contextual details retrieved when constructing a mental representation for a past event. Specifically, our prediction was that forming a memory representation with more contextual details (i.e., with a contextual retrieval orientation) would activate regions that support visuo-spatial component processes and forming a memory representation with more conceptual details (i.e., with a conceptual retrieval orientation) would activate regions that support schematic and value-based component processes. We further predict that these dissociations will be apparent for both recent and remote memories yet be most robust when orienting within an already-accessed memory (i.e., re-oriented recollection), provided that the latter will require more episodic memory constructive processes to allow one to revise and reconstruct an updated representation specified by the retrieval orientation. Univariate and multivariate analytic approaches were taken to identify regional activation differences and network variability differences related to retrieval orientation, respectively.
Section snippets
Participants
Twenty-eight participants were recruited to participate. Data from four participants were excluded from our analyses due to indications of anxiety while in the scanner (n = 1) or failure to comply with the task instructions (n = 3). The analyzed dataset included 24 right-handed participants (females = 17; mean age = 21.34 years, SD = 3.3; mean education = 15.26 years; SD = 2.18). All participants had normal or corrected-to-normal vision and were free of any medical conditions, neurological
Behavioural results
The average of the in-scanner ratings (vividness, date categorization) and reaction times for the oriented memories were examined with separate repeated measures ANOVA with retrieval orientation (conceptual versus contextual) and memory age (recent versus remote) as factors. None of main effects of retrieval orientation were significant (vividness: F(1, 22) = 0.795, p = .382; date categorization: F(1, 22) = 0.082, p = .778; reaction time: F(1, 23) = 1.930, p = .178). The main effect of memory
Discussion
Retrieving an episodic autobiographical memory is a constructive act that requires flexibly forming mental representations of past events (Bartlett, 1932; Tulving, 1983, 2002). This construction requires accessing separately-stored episodic event-specific information at the time of retrieval, from the conceptual meaning of the event to the sensory-perceptual contextual elements of the experience (Conway, 2001), which are predicted to be supported by distinct neural processes (e.g., Sheldon
Limitations and conclusion
Here, we provide new evidence for discrete functional brain systems for accessing two critical components of autobiographical memories – conceptual and contextual details. However, there are a few methodological issues to consider when interpreting these findings. First, the design of our study required that participants recall the same twelve pre-selected memories multiple times during the full recollection and re-oriented recollection phases (twice during the conceptual retrieval orientation
Author's contributions
SS developed and designed the research project; LG collected the data; LG and SS analyzed the data; and LG and SS wrote the manuscript.
Acknowledgements
This study was supported by a Canada Research Chair awarded to SS and Natural Sciences and Engineering Research Council of Canada awarded to SS. The authors would like to thank Sonja Chu and Elizabeth DuTemple for help with data collection and organization, and Mary Pat McAndrews for helpful conversations regarding the task design.
References (76)
- et al.
Functional-anatomic fractionation of the brain's default network
Neuron
(2010) Episodic memories
Neuropsychologia
(2009)AFNI: software for analysis and visualization of functional magnetic resonance neuroimages
Comput. Biomed. Res. Int. J.
(1996)- et al.
A framework for understanding the relationship between externally and internally directed cognition
Neuropsychologia
(2014) - et al.
Memory strength and repetition suppression: multimodal imaging of medial temporal cortical contributions to recognition
Neuron
(2005) - et al.
Repetition and the brain: neural models of stimulus-specific effects
Trends Cognit. Sci.
(2006) - et al.
Deconstructing episodic memory with construction
Trends Cognit. Sci.
(2007) - et al.
The precuneus and hippocampus contribute to individual differences in the unfolding of spatial representations during episodic autobiographical memory
Neuropsychologia
(2018) - et al.
Neural response suppression, haemodynamic repetition effects, and behavioural priming
Neuropsychologia
(2003) Direct electrophysiological evidence for the maintenance of retrieval orientations and the role of cognitive control
Neuroimage
(2018)