Constructive Memory and Conscious Experience

Abstract

Episodic memory relies on constructive processes that support simulating novel future events by flexibly recombining elements of past experiences, and that can also give rise to memory errors. In recent studies, we have developed methods to characterize the cognitive and neural processes that support conscious experiences linked to this process of episodic recombination, both when people simulate novel future events and commit recombination-related memory errors. In this Perspective, we summarize recent studies that illustrate these phenomena, and discuss broader implications for characterizing the basis of conscious experiences associated with constructive memory from a cognitive neuroscience perspective.

INTRODUCTION

Memory researchers have expended considerable effort attempting to characterize the nature of conscious experiences associated with various forms of memory. This issue became a focal point of research during the 1980s and 1990s that attempted to characterize the nature of and relation between explicit or declarative and implicit or nondeclarative forms of memory (cf. Squire, 1992; Schacter, 1987; Graf & Schacter, 1985), including various experimental approaches to separating conscious from nonconscious influences of memory on task performance (e.g., Jacoby, 1991; Roediger, 1990; Schacter, Bowers, & Booker, 1989). At around the same time, Tulving (1985a) distinguished among distinct types of memory-related consciousness (anoetic, noetic, and autonoetic) and developed the widely used “Remember/Know” technique in an attempt to measure distinct features of the conscious experience of memory (Tulving, 1985b). In closely related research, Mandler (1980) and others (e.g., Yonelinas, 1994) delineated the relationship between recollection and familiarity. More recently, other aspects of conscious experience and memory have come to the fore, including debates regarding how best to measure memory confidence and characterize its relation to memory accuracy (e.g., Wixted & Wells, 2017), as well as growing numbers of attempts to delineate the neural correlates of conscious experiences of remembering (for a review, see Simons, Ritchey, & Fernyhough, 2022).

Although these and related efforts have yielded important methods, findings, and theoretical ideas, a notable and increasingly prominent theme during the past decade and more is that the reach of memory, specifically episodic memory, extends beyond simply remembering past experiences. For example, a growing body of evidence has revealed striking cognitive and neural similarities when people remember past experiences and imagine future experiences, and explored the ways in which individuals use past experiences to simulate novel future events (for reviews, see Mulally & Maguire, 2014; Schacter et al., 2012). According to the constructive episodic simulation hypothesis (Schacter & Addis, 2007, 2020), episodic memory is a constructive process that supports our ability to simulate future events by flexibly recombining elements of past experiences into simulations of novel experiences. Although the process of episodic recombination is adaptive for purposes of simulating future and other hypothetical experiences, by this view, episodic recombination can also give rise to memory errors.

Approaching the issue of conscious experience and memory through the lens of the constructive episodic simulation hypothesis, we have recently developed new methods in an attempt to characterize the cognitive and neural processes that support conscious experiences linked to episodic recombination during both future simulation and when people commit recombination-related memory errors. In this article, we review several of our recent studies in both domains, and then discuss broader implications for understanding conscious experiences linked to constructive memory from a cognitive neuroscience perspective.

CONSCIOUS REINSTATEMENT OF EPISODIC DETAILS DURING AUTOBIOGRAPHICAL REMEMBERING AND FUTURE IMAGINING

As noted earlier, over the past two decades, numerous studies have demonstrated cognitive and neural similarities when people remember past experiences and imagine future or other hypothetical experiences. One of the most robust findings in this literature is that a core network of brain regions that largely overlaps with the well-known default network, comprising regions in the medial temporal lobe including hippocampus, posterior cingulate including retrosplenial cortex, medial pFC, and lateral temporal and parietal regions, shows similarly increased activity when people remember past experiences and imagine future experiences (Benoit & Schacter, 2015; Schacter, Addis, & Buckner, 2007). Given the prominence of the hippocampus in neuroscience-based memory research, it is hardly surprising that there have been numerous empirical and theoretical attempts to specify its role in episodic simulation of future experiences (for a review, see Schacter, Addis, & Szpunar, 2017). Critically, the hippocampus has been linked with the reinstatement of recently encoded episodic information during various retrieval tasks in studies that have used multivoxel pattern analysis (MVPA) to examine reinstatement of event features (e.g., Wing, Ritchey, & Cabeza, 2015). According to the constructive episodic simulation hypothesis, recombined event details are critical constituents of imagined future episodes, so it should be possible to find evidence for hippocampally mediated reinstatement of these details.

To evaluate this hypothesis, Thakral, Madore, Addis, and Schacter (2020) adapted the MVPA approach used in the aforementioned studies of episodic reinstatement to the domain of autobiographical memory and future imagining. The study consisted of two separate sessions, and it made use of a procedure developed earlier in our laboratory called the experimental recombination paradigm (Addis, Pan, Vu, Laiser, & Schacter, 2009), which we created to promote the controlled use of recombined episodic details when people imagine future events. Participants first completed a stimulus collection session in which they recalled specific events from the past 5 years that occurred in a specific time and place, and that lasted no longer than a few minutes to an hour. They provided a title for the event, and they also specified a critical person and location involved in the event. For example, the event titled Picture with Seagulls featured Janine W. and Ocean Beach Pier, whereas the event titled Dog Training Sunday featured Otis W. and Kitchen in Dallas. The experimenter recombined the person and location for a subsequent episodic simulation task so that, for example, a participant would be instructed to imagine a novel future event featuring Otis W. and Ocean Beach Pier. In the subsequent experimental fMRI session, participants were scanned while performing memory, simulation, and control tasks. For the memory task, participants were first given a task cue (“recall memory”) followed by an event title (e.g., Picture with Seagulls), and 10 sec to recall the appropriate memory, focusing on the person and location. For the episodic simulation task, participants were first given the task cue (“novel simulation”) followed by a person and location (e.g., Otis W., Ocean Beach Pier), and 10 sec to imagine a novel event that could plausibly occur in the next 5 years involving the specified person and location. Memory cues in the recall task did not include the person or location names used as cues in the simulation task to avoid across-task perceptual overlap, so that any similarity across memory and simulation reflects information retrieved from memory and not cue-related similarity (see Wing et al., 2015; Kuhl & Chun, 2014). Participants also performed a control task used in previous related studies that involved constructing a sentence from specified cue words (e.g., Addis et al., 2009). For each task, participants made several ratings, the most important of which was a vividness rating, ranging from low (1) to high (5). For the memory and simulation tasks, vividness ratings were made separately for the person and location details. Vividness ratings provide one index of the conscious experience associated with remembering and imagining, and we were interested in assessing behaviorally whether a detail that is vivid during remembering is also vivid when it is part of an episodic simulation, and neurally whether a potential link between vividness and hippocampal activity can provide some insight into whether this activity is modulated by subjective experience during memory and simulation.

Similarity between memory and simulation was assessed by adapting the MVPA analytic approach used in previous memory reinstatement studies (e.g., Wing et al., 2015). Specifically, the multivoxel pattern associated with the memory containing a specific detail (e.g., the location detail “Ocean beach pier”) was correlated with the simulation containing that detail to yield a matching correlation. In addition, the same detail from that memory was correlated with all the other episodic simulations that did not contain that detail to yield a mismatching correlation. The two correlations were compared, with the logic that a significantly larger matching than mismatching correlation provides evidence of reinstatement of that specific detail from memory during an episodic simulation. In addition to the hippocampus, we also examined matching versus mismatching correlations in the angular gyrus, a core network region that has been linked to both episodic memory and simulation (e.g., Thakral, Madore, & Schacter, 2017).

Analysis of behavioral data revealed that the vividness of both location and person details was significantly positively correlated, thus confirming that they rely on similar episodic content. The MVPA analyses showed that for the hippocampus, the matching correlations for person and locations details were significantly higher than the mismatching correlation, but the difference was apparent only for highly vivid details (Figure 1A, right). By contrast, in the angular gyrus, the matching correlation was higher than the mismatching correlation, and the effect did not depend on the vividness of the details (see Figure 1A, left). Thus, although there was evidence of episodic reinstatement in both regions, the effect in the hippocampus appeared to be modulated by subjective experience in a way that it was not in the angular gyrus. We will return to this observation later in the article. The key point for now is that the results implicated the hippocampus in conscious reinstatement of episodic details during simulation of a future experience, and in a way that differed from the angular gyrus.

Figure 1. .

(A) MVPA results from Thakral, Madore, Addis, and Schacter (2020) within the angular gyrus (left) and hippocampus (right) as a function of correlation type (match and mismatch) and vividness (high and low). (B) Univariate results from Thakral, Madore, and Schacter (2020) within the angular gyrus (left) and hippocampus (right) as a function of vividness (high and low) and number of episodic details (high and low). (C) Univariate results from Thakral and colleagues (2022) within the angular gyrus (left) and hippocampus (right) as a function of source accuracy (accurate and inaccurate) and remember-know.

EPISODIC REINSTATEMENT AND MEMORY DISTORTION

A key tenet of the constructive episodic simulation hypothesis noted earlier is that the same episodic recombination processes that serve the adaptive function of constructing novel event representations that support episodic simulation and related functions can also contribute to memory errors when elements of different experiences are miscombined. Previous studies have shown that memory errors can result from miscombining elements of different experiences (e.g., Devitt, Monk-Fromont, Schacter, & Addis, 2016; Odegard & Lampinen, 2004), and there is also evidence for a link between memory errors and cognitive processes that support adaptive functions, including semantic representation and memory updating (e.g., Howe, 2011; Schacter, Guerin, & St. Jacques, 2011; Hardt, Einarsson, & Nader, 2010), However, until recently, there were no direct tests of the central claim of the constructive episodic simulation hypothesis that the same flexible episodic recombination processes that support the adaptive function of constructing novel event representations also produce memory errors. In a recent series of studies, Carpenter and Schacter (2017) provided evidence to support this hypothesis by showing that when people succeed in making associative inferences—combining related information acquired in distinct episodes to make novel connections that have not been directly experienced—they showed increased susceptibility to false memories that result from mistakenly combining contextual elements of the associated episodes. On the basis of these behavioral findings, we applied the fMRI-MVPA reinstatement approach described in the preceding section to elucidate the underlying neural basis for the mistaken conscious experiences that accompany false memories that are linked to successful associative inference (Carpenter, Thakral, Preston, & Schacter, 2021). As we have argued elsewhere (Schacter, Carpenter, Devitt, & Thakral, in press), it is important more generally to understand the underlying basis for these kinds of mistaken subjective experiences for both theoretical and practical reasons (e.g., mistaken eyewitness memories). We first briefly outline the behavioral paradigm developed by Carpenter and Schacter (2017) and then summarize the subsequent fMRI approach and findings from Carpenter and colleagues (2021).

The behavioral paradigm involves two separate sessions separated by a 24-hr delay. In the first session, participants encoded a series of scenes that included a person (e.g., a man), an object (e.g., a yellow toy), and background setting (e.g., a living room). Participants were instructed to remember each scene and also to infer the association between people in different background settings who were linked to one another because each is paired with the same object (e.g., a man holding a yellow toy in a living room with a white couch, and a boy holding the same yellow toy in a different living room with a brown couch). The next day, participants were tested regarding memory for the scene context of half the studied items (e.g., they indicated whether the man holding the yellow toy appeared in a room with a white couch or a brown couch), followed by an associative inference test for all the items (e.g., they indicated which of two boys the man was associated with via a common object), and finally by a test of memory for the scene context of the other half of the studied items.

Carpenter and Schacter (2017) found that participants made more contextual memory errors (e.g., mistakenly recalling that the man was in a room with a brown couch) when they had made correct inferences about the relations between the people in these scenes than when they had made incorrect inferences (e.g., when they correctly inferred that the man and boy were linked via the toy vs. when they did not). Critically, however, the observed boost in memory errors for correct inferences occurred only when contextual details were tested after the associative inference test; there was no difference in memory errors for correct versus incorrect inferences when memory for contextual details was probed before the associative inference test. This result is important theoretically because Carpenter and Schacter (2017) hypothesized that the associative inference test elicits the episodic recombination processes that link the two scenes and allow an individual to make the correct inference.

In an attempt to elucidate the underlying neural basis for the inference-related contextual memory error, Carpenter and colleagues (2021) combined this behavioral paradigm with the fMRI-MVPA approach from Thakral, Madore, Addis, and colleagues (2020), using the matching versus mismatching correlation analysis approach described earlier. In this experiment, there were also two sessions separated by a 24-hr delay. The first session included an encoding phase conducted outside the fMRI scanner that was similar to the encoding phase described previously. In addition, before the encoding phase, there was a pre-exposure phase in the fMRI scanner where participants viewed the scenes that would be presented later during the encoding phase, such as the living room with a white couch and the living room with a brown couch, but without the person and object who appeared in those scenes during the encoding phase. The purpose of the pre-exposure phase was to obtain a measurement of the multivoxel pattern associated with each scene. A day later, participants were scanned again as they completed the same sequence of tests described earlier: memory for the scene context of half the studied items, followed by an associative inference test for all the items, and finally a test of memory for the scene context of the other half of the studied items.

The critical fMRI analyses focused on the correlation between the multivoxel patterns elicited by the “empty” scenes during the pre-exposure phase and the patterns elicited during the two contextual memory tests, when participants were trying to recall the specific scene (e.g., living room with a white couch or a brown couch) associated with an individual (e.g., the man who was holding a toy). Here, the matching correlation was the correlation between the pattern elicited during the pre-exposure phase by the scene containing the incorrect contextual detail (e.g., the living room with a brown couch) and the pattern observed during the scene memory test; the mismatching correlation was the correlation between the pattern observed during the scene memory test and all the other patterns elicited during the pre-exposure phase by scenes that did not contain the incorrect contextual detail. Using the matching versus mismatching correlation comparison described earlier, Carpenter and colleagues (2021) focused on three critical ROIs: anterior hippocampus and posterior medial pFC, which have been implicated by previous studies in successful associative inference (e.g., Schlichting, Mumford, & Preston, 2015) as well as episodic recombination (e.g., Benoit & Schacter, 2015), and the left inferior temporal gyrus, which has been implicated in reinstatement of contextual details similar to those at play in the present paradigm (e.g., Ranganath, Cohen, Dam, & D’Esposito, 2004). In each of these three critical ROIs, Carpenter and colleagues (2021) showed that the neural patterns during retrieval of contextual details following successful associative inference became more similar to the overlapping yet incorrect context compared after unsuccessful inference, as reflected by a larger difference between matching correlation and mismatching correlation after successful than unsuccessful inference. In other words, there was more neural evidence for the incorrect “brown couch pattern” on the contextual memory test following the correct inference that the man and boy were linked than following the incorrect inference that they were not linked. No such evidence was observed in control regions. This observation converges with the behavioral data to suggest that participants' subjective experiences during the contextual memory test involved a representation of the “false” neural pattern containing the mistakenly combined elements, which was boosted by prior successful associative inference. Further cognitive/behavioral studies will be needed to characterize more precisely the conscious experience of remembering that is associated with neural reinstatement of such “false” neural patterns, a point to which we return in the next section.

BROADER IMPLICATIONS AND FUTURE DIRECTIONS

Hippocampus, Angular Gyrus, and the Conscious Experience of Remembering

The distinction we have previously observed between hippocampal and angular gyrus contributions to episodic simulation using MVPA (Thakral, Madore, Addis, et al., 2020) is conceptually related to another fMRI study we conducted to identify the specific contributions of the individual regions comprising the core network to memory and simulation. In Thakral, Madore, and Schacter (2020), we assessed whether members of the core network are differentially associated with two common indices of episodic processing: subjective episodic experience (i.e., vividness) versus objective amount of episodic content comprising those events (i.e., the number of details). During scanning, participants imagined future events in response to object cues. On each trial, participants rated the subjective vividness associated with each future event on a 5-point scale. Participants later completed a postscan interview where they viewed each object cue from the scanner and verbally reported whatever they had thought about. We quantified the amount of episodic details in accordance with the Autobiographical Interview (AI; Levine, Svoboda, Hay, Winocur, & Moscovitch, 2002). In the AI, the details that participants produce are categorized as either “episodic/internal” or “non-episodic/external” details. Episodic details reflect the who, what, when, and where details of each central event, and non-episodic details mainly reflect related facts, commentary, and off-topic information. Our fMRI analysis revealed that regions that tracked the level of objective detail included the angular gyrus (Figure 1B, left; for a replication using fMRI, see Thakral, Benoit, & Schacter, 2017, and for TMS evidence, see Thakral, Madore, et al., 2017). Regions that varied with the subjective experience/vividness of simulated episodic content included the hippocampus (Figure 1B, right).

Quantifying the Conscious Experience

The question remains as to what process contributes specifically to the conscious experience of episodic events. An apparent answer that follows from the aforementioned findings would be a hippocampal-mediated process given that the hippocampus was consistently sensitive to self-rated vividness (see also, Addis & Schacter, 2008). However, there are instances where the hippocampus has been found to not be sensitive to the subjective episodic experience. Consider a recently reported study where participants made both objective (source memory for spatial context) and subjective (Remember-Know) judgments for each memory test item (Thakral, Yu, & Rugg, 2022; see also, Slotnick, 2010). Results indicated that the hippocampus was exclusively sensitive to the amount of contextual information retrieved (i.e., objective source accuracy but not subjective “Remembering”; Figure 1C, right), whereas the left angular gyrus was exclusively sensitive to Remembering (Figure 1C, left). These findings contrast with those discussed earlier and suggest that the mapping of subjective indices of episodic processing (either vividness ratings or Remember responses) onto specific neural regions may be context dependent.

On the basis of the overall pattern of results, we suggest that there are at least two distinct mechanisms that are involved in supporting the conscious experiences associated with episodic constructive processes (see Figure 2 for our proposed functional–anatomic dissociation between the angular gyrus and hippocampus). One hippocampal process (Figure 2, right) supports the retrieval of specific episodic information. In our prior studies, hippocampal activity was observed in situations that necessitated the retrieval of specific information, such as when a vividness rating indexed the retrieval of detail-specific (person or location) memorial information (Thakral, Madore, Adis, et al., 2020), the vividness of imagined future events (Thakral, Madore, & Schacter, 2020), or when a task required the retrieval of specific spatial context information during a source memory task (Thakral et al., 2022). This interpretation aligns with popular models of hippocampal function stipulating its role in the relational binding of episode-specific details (e.g., Diana, Yonelinas, & Ranganath, 2007) and reinstatement of contextual information unique to a given episode (Renoult et al., 2019). We believe that our findings suggest that vividness ratings can be used as a way to quantify or index the availability of specific episodic details (e.g., D’Angiulli et al., 2013), likely taxing hippocampal processes associated with the initial reinstatement and/or binding of episode-specific details.

Figure 2. .

Proposed functional–anatomic dissociation between the angular gyrus and hippocampus. Color coding is intended to represent the gradient of episodic process and/or content associated with the angular gyrus and hippocampus, respectively (i.e., episode-general to episode-specific).

In contrast, the angular gyrus serves as a convergence zone to support the representation of more general event information (i.e., a schema) that is a necessary component of event-based functions such as episodic remembering and imagining (Figure 2, left; Ramanan, Piquet, & Irish, 2018; Gilboa & Marlatte, 2017). In our prior studies, neural activity in the angular gyrus scaled with the number of internal details comprising imagined future and remembered past episodes as assessed with fMRI (Thakral, Madore, & Schacter, 2020; Thakral, Benoit, et al., 2017) or TMS (Thakral, Madore, et al., 2017). We believe that these effects reflect the retrieval of detail-rich yet episode-general information (i.e., who, what, where, and when event information that is shared across events). This idea is conceptually similar to proposals suggesting that the angular gyrus plays a role in other schema-related processes: the retrieval of event concepts (i.e., familiar concepts that imply an unfolding over time and space; Binder & Desai, 2011), the reinstatement of conceptual processing that covaries with episodic retrieval (Renoult et al., 2019), or personal semantics (i.e., the generalized facts that define personally relevant stimuli; Renoult et al., 2012).

Lastly, with respect to the dissociation we observed during laboratory measures of episodic retrieval, relative to the hippocampus, we observed a unique relationship between subjective remembering and the angular gyrus (Thakral et al., 2022). These data suggest that Remember responses do not reflect the retrieval of the same kind of specific episodic content as that supported by the hippocampus. Although we can be confident that Remember responses reflect autonoetic consciousness (Tulving, 1985a), one limitation of the Remember-Know task is that participants can make a Remember response when retrieval is accompanied by any information associated with a studied event. Thus, it is unknown what type of content is associated with “remembering.” One possibility is that both Remember responses and the production of internal details capture similar detailed yet episode-general content, with this type of content being quintessential to autonoetic consciousness. Although more data are needed to directly test this idea, the present data do align with patient and brain stimulation evidence showing that lesions to the left angular gyrus lead to reductions in the number of episodic/internal details during remembering and imagining and to reduced Remember responses, together with preserved memory for specific contextual details (e.g., as in tests of source memory; see Thakral, Madore, & Schacter, 2017; Davidson et al., 2008). The angular gyrus may support the common autonoetic conscious experience during remembering, the generation of autobiographical past and future episodes, and those also engaged during other complex adaptive functions (such as divergent creative thinking, see below; and for a related proposal, see Humphreys, Jung, & Lambon Ralph, 2022).

At a broader level, the current data highlight that future studies aimed at identifying the neural correlates of conscious episodic processing should not assume that any given measure of episodic processing reflects a single cognitive function, but instead likely depends on the specific requirements of the task. For example, objective measures of episodic content (e.g., source memory or the episodic detail metric in the AI) may index different processes depending on the specificity of the content (e.g., specific information or more event-based information, respectively). The same point holds for self-report vividness ratings, which cannot be assumed to be a process-pure measure of subjective experience because these ratings are influenced by objective performance/content. It should also be noted that although we have proposed that the angular gyrus and hippocampus are sensitive to different types of episodic processes (Figure 2), that does not preclude the idea that these regions are interactive and play a necessary role in either type of episodic process (e.g., detail-specific reinstatement supported by the hippocampus may be necessary during subjective Remembering). In fact, in our studies, we have shown that angular gyrus-to-hippocampus functional connectivity is necessary for the generation of objective episodic detail during future imagination (Thakral, Madore, Kalinowski, & Schacter, 2020). Our findings also highlight the need to characterize the conscious episodic experience in multiple ways. Employing different ways to operationalize the conscious experience within a given experiment provides the unique opportunity to fractionate the conscious experience into dissociable processes.

Constructive Uses of Episodic Memory

Although we focused earlier on the role of episodic recombination in future imagining, it has become clear in recent years that episodic retrieval and recombination support related functions, including means-end social problem solving (e.g., Jing, Madore, & Schacter 2016; Madore & Schacter, 2014; Sheldon, McAndrews, & Moscovitch, 2011) and divergent creative thinking, which involves generating creative ideas by combining diverse types of information in novel ways (e.g., Madore, Addis, & Schacter, 2015). We have provided three kinds of evidence that the hippocampus plays a role divergent creative thinking. First, using a procedure we call an episodic specificity induction (ESI)—brief training in recalling episodic details (for a review, see Schacter & Madore, 2016)—we have shown that performance on a standard test of divergent thinking that involves generating alternative uses of objects (Alternate Uses Test [AUT]) increases following an ESI versus a control induction, and that the hippocampus plays a role in the ESI-related performance boost (Madore, Thakral, Beaty, Addis, & Schacter, 2019). Second, an fMRI study of episodic remembering, future imagining, and divergent creative thinking (i.e., AUT) revealed common engagement of the hippocampus in all three tasks (Beaty, Thakral, Madore, Benedek, & Schacter, 2018). Third, a study by Thakral, Madore, Kalinowski, and Schacter (2020) that used fMRI-guided TMS to the angular gyrus to target the hippocampus (see, Wang et al., 2014, for methodological basis) showed subsequent performance reductions on the AUT following inhibitory continuous θ-burst stimulation as well as on an episodic future simulation task, along with TMS-related reductions in hippocampal activity that were linked to the reductions in behavioral task performance. Although these studies provide consistent evidence for a link between the hippocampus and divergent creative thinking (for theoretical implications, see Benedek, Beaty, Schacter, & Kenett, 2023), little attention has been paid so far to the conscious experiences associated with divergent thinking, and how they are related to activity in the hippocampus and other brain regions. A start in this direction comes from studies in which participants are asked to indicate whether a novel use of an object that they generated on the AUT reflects their memory of having observed the novel use (an “old” idea) or an imagined, newly constructed use (a “new” idea; e.g., Thakral, Yang, Addis, & Schacter, 2021; Benedek et al., 2014). An important task for future research is to further develop measures of subjective experience during divergent thinking tasks that can then be related to brain activity. Here, we have focused on the role of episodic constructive processes and their neural underpinnings during episodic simulation and divergent creative thinking. However, episodic simulation and related adaptive functions such as divergent creative thinking are not simply a direct expression of episodic retrieval and recombination. It has long been recognized that semantic memory processes play a key role in successful divergent thinking (e.g., Benedek et al., 2023; Beaty et al., 2020; Hass, 2017; Abraham et al., 2012). Moreover, patients with semantic dementia, who typically exhibit severe problems with semantic memory, also demonstrate marked episodic simulation impairments with preserved episodic memory (Irish & Piolino, 2016). More recent fMRI evidence has shown that there are common and distinct neural regions engaged during episodic memory and during the retrieval of personal semantics (Tanguay et al., 2023). These findings highlight the interdependency between episodic and semantic memory processes in supporting episodic simulation and divergent creative thinking. Examining how episodic and semantic processes differentially contribute to the conscious experience is an important avenue for future research.

False Memories versus False Beliefs

Most cognitive and neuroimaging studies of false memories have used laboratory paradigms to induce these errors, such as the well-known Deese-Roediger-McDermott (DRM) paradigm (Roediger & McDermott, 1995; Deese, 1959), where people frequently claim to have studied a nonpresented lure word (e.g., sweet) that is a strong associate of previously presented words (e.g., candy, sugar, tooth, nice, honey…). Beginning with Roediger and McDermott's (1995) original report, studies have attempted to characterize the nature of participants' conscious experiences when falsely recognizing a nonpresented lure word, typically by using Remember/Know judgments and/or confidence ratings, but also by examining the nature of illusory perceptual, contextual, and associative details that accompany DRM false memories (for reviews, see Dehon, 2012; Gallo, 2010; Schacter et al., in press). Taken together, these studies support the conclusion that participants consciously experience their erroneous responses in a manner similar to the conscious experiences that accompany veridical memories. A large number of neuroimaging studies have examined brain activity during performance of the DRM and similar laboratory paradigms, revealing numerous similarities in brain activity during true and false memories, along with some important differences (for reviews, see Kurkela & Dennis, 2016; Schacter et al., in press).

In parallel to studies using word lists and related procedures to induce memory distortions, studies focused on real-world implications of memory errors have attempted to induce false memories of complex everyday experiences. This line of research began with the seminal “lost in the mall” study, where repeated suggestive procedures were used to induce a false memory of being lost in a shopping mall in an adolescent boy named Chris when he was a young child (Loftus, 1993), and then extended to a larger sample (n = 24) in which 25% of participants reported similar false memories (Loftus & Pickrell, 1995). Murphy and colleagues (2023) recently replicated and extended these findings in an even larger sample (n = 123), reporting a 35% false memory rate. These false memory rates are in line with the results of various other studies using similar procedures that have produced false memories of complex everyday experiences in roughly 25–40% of participants (e.g., Heaps & Nash, 2001; Hyman, Husband, & Billings, 1995).

By contrast, Shaw and Porter (2015) used a potent combination of suggestive procedures and social pressure with a college student population that induced false memories of committing a crime as an adolescent that resulted in police contact in 70% of these individuals. Subsequently, however, Wade, Garry, and Pezdek (2018) raised questions about the nature of the subjective experiences reported by these participants. Specifically, they cited earlier studies that developed criteria for distinguishing false memories, which involve a subjective experience of remembering, from false beliefs, which involve accepting a suggestion, speculating about what might have happened, and generating mental images of the suggested event, but do not include a subjective experience of remembering the event (e.g., Lindsay, Hagen, Read, Wade, & Garry, 2004). When Wade and colleagues (2018) rescored Shaw and Porter's data using these criteria, they reported that, consistent with earlier studies, only about 25–30% of participants' reports met criteria for false memories, and the rest were classified as false beliefs.

The important point raised by these studies for the present purposes is that characterizing the nature of a constructive memory phenomenon depends critically on the nature of the conscious experiences that accompany a report of a past event; a false memory and a false belief reflect distinct subjective experiences and may have different consequences as well as different underlying substrates. For example, false memories may be based largely or entirely on constructive episodic processes, whereas false beliefs might reflect primarily the influence of personal semantics. We are not aware of any evidence that addresses this issue, but we think that it is worth examining in future studies. Moreover, although no neuroimaging evidence exists concerning false reports of everyday autobiographical events such as those elicited in the foregoing cognitive studies, it would be useful to attempt such investigations with an eye toward identifying neural markers that distinguish false memories from false beliefs. Such studies pose clear methodological challenges and would need to be interpreted in light of the points we raised earlier regarding the context-dependency of mapping subjective indices of episodic processing to specific neural regions. Thus, it may be more tractable to approach the issue by using more tightly controlled experimental paradigms such as the associative inference/contextual memory error paradigm described earlier (Carpenter et al., 2021; Carpenter & Schacter, 2017). For example, it would be of interest to determine whether the “false” neural patterns observed by Carpenter and colleagues (2021) during associative inference-related memory errors reflect false memories or false beliefs. Nonetheless, extending a neuroimaging approach to these complex phenomena has the potential to contribute importantly to our understanding of conscious experiences associated with different manifestations of constructive memory processes.

Acknowledgments

This article is based on a talk given by D. L. S. at the workshop on Next Frontiers in Consciousness Research held at the National Institutes of Health, June 26–28, 2023. We thank workshop participants for comments concerning the presentation.

Corresponding author: Daniel L. Schacter, Department of Psychology and Center for Brain Science, Harvard University, Cambridge, MA, or via e-mail: dls@wjh.harvard.edu.

Author Contributions

Daniel L. Schacter: Conceptualization; Writing—Original draft; Writing—Review & editing. Preston P. Thakral: Conceptualization; Writing—Original draft; Writing—Review & editing.

Funding Information

Preparation of this article was supported by National Institutes of Mental Health (https://dx.doi.org/10.13039/100000025), grant number: MH60941 to Daniel L. Schacter.

Diversity in Citation Practices

Retrospective analysis of the citations in every article published in this journal from 2010 to 2021 reveals a persistent pattern of gender imbalance: Although the proportions of authorship teams (categorized by estimated gender identification of first author/last author) publishing in the Journal of Cognitive Neuroscience (JoCN) during this period were M(an)/M = .407, W(oman)/M = .32, M/W = .115, and W/W = .159, the comparable proportions for the articles that these authorship teams cited were M/M = .549, W/M = .257, M/W = .109, and W/W = .085 (Postle and Fulvio, JoCN, 34:1, pp. 1–3). Consequently, JoCN encourages all authors to consider gender balance explicitly when selecting which articles to cite and gives them the opportunity to report their article's gender citation balance.