From jamais to déjà vu: The respective roles of semantic and episodic memory in novelty monitoring and involuntary memory retrieval

Abstract

Barzykowski and Moulin's model proposes that déjà vu and involuntary autobiographical memories are the result of a continuously active memory system that tracks the novelty of situations. Déjà vu would only have episodic content and concern interpretation of prior experiences. We argue that these aspects of the model would gain to be clarified and explored further and we suggest possible directions.

The model of Barzykowski and Moulin (B&M) proposes an interesting comparison of involuntary autobiographical memories and déjà vu experience, whereby déjà vu would be the product of an underdeveloped form of involuntary memory retrieval. The model includes the attractive notion that familiarity/novelty detection mechanisms might need to be permanently active, sometimes leading to déjà vu experience and to false detections. Moreover, it is proposed that déjà vu would only have episodic content and concern interpretation of prior experiences, rather than semantic knowledge. These interesting aspects of their proposal are not elaborated upon, unfortunately. However, they are consistent with the finding that semantic and episodic memory appear to have different sensitivity to novelty processing. Repeated semantic processing has often been associated with the phenomenon of semantic satiation, a temporary loss of meaning attached to stimuli. Explicit attention to meaning is necessary to counteract these effects (Renoult, Wang, Mortimer, & Debruille, 2012). Once an object has been identified and categorized, the conceptual system may thus no longer be oriented to this particular object (Murphy, 2002). As shown in another work by the authors (Moulin, Bell, Turunen, Baharin, & O'Connor, 2021), repeated semantic processing may actually lead to the opposite experience, the so-called, jamais-vu (never seen).

In contrast, episodic memory allows to encode unique events (even if it is just a repetition of the same item, Tulving, 1972, 1983; or an event similar to a previous experience, Rubin & Umanath, 2015), while related neural processes may create distinct representations even for seemingly similar events (Nadel & Moscovitch, 1997; Norman, 2010; O'Reilly, Bhattacharyya, Howard, & Ketz, 2014). Therefore, if déjà vu experiences depend on constantly active novelty detection mechanisms, episodic memory would thus be more likely to be associated with these experiences, consistent with B&M's current proposal. Nonetheless, a better characterisation of these novelty detection mechanisms would help to better understand how the experiences of involuntary autobiographical memories and déjà vu are generated.

As a matter of fact, the points made in the article that déjà vu would “concern interpretation of prior experiences” and would typically happen for “familiar places and people” suggest a role for semantic knowledge. Consistent with the proposal that episodic memory operations typically require semantic memory (Tulving, 2002), one may ask, for example, whether déjà vu could be related to the retrieval of schema knowledge about a similar past experience, which would act as a cue for a (failed) episodic retrieval attempt. In other words, it could be that it is this activation of semantic knowledge about similar prior experiences that would trigger a retrieval mode: A “tonically maintained state” needed for episodic retrieval (Rugg & Wilding, 2000; Tulving, 1983). Accessing conceptual information can indeed prime the retrieval of specific memories (Mace, McQueen, Hayslett, Staley, & Welch, 2019; Reiser, Black, & Abelson, 1985) and semantic processing of cues was proposed to be essential for episodic retrieval (Tulving, 1983). However, in the case of déjà vu, this retrieval attempt may fail, or conflict with the awareness that the experience is novel. Presumably, the experience of jamais vu (Moulin et al., 2021; O'Connor, Wells, & Moulin, 2021) would similarly not be associated with the retrieval of specific memories. However, in that case, it would be due to disrupted semantic retrieval (semantic satiation), which may also deprive episodic memory from a relevant cue. Whether these proposals are valid or not, further exploration of the notion that déjà vu would only have episodic content appears important, as well as clarifying the role of interpretative process (and of semantic knowledge) in generating these experiences.

Finally, as noted by B&M, the role of personal semantics (e.g., the type of personal semantics involved and the timing of its involvement) would be worth exploring further in the context of evaluating how it may influence or prime, as discussed in the article, the content of retrieval. For instance, activating context-dependent types of personal semantics (e.g., I visit this café every Monday) might more likely lead to the retrieval of involuntary autobiographical memories. More abstracted form of personal semantics (e.g., I take milk in my coffee) would more likely be restricted to feeling of familiarity or déjà vu (Sheldon, Peters, & Renoult, 2020). This would be consistent with the findings of Berntsen, Staugaard, and Sørensen (2013) that involuntary episodic memories are retrieved more often in response to specific compared to more generic cues.

We thoroughly enjoyed the proposal of B&M and hope that further research, based on their model, will help clarify these issues.