Are involuntary autobiographical memory and déjà vu cognitive failures?

Abstract

This commentary supports Barzykowski and Moulin's model, but departs from it on the question of functionality, where IAMs and déjà vu fractionate. The authors seem to say that IAMs are functional, while déjà vu is not. As there is no hard evidence supporting the idea that IAMs are functional, I argue that both phenomena should be viewed as cognitive failures.

Barzykowski and Moulin (B&M) present an interesting model for understanding involuntary autobiographical memories (IAMs) and déjà vu, two mysterious cognitive phenomena. They argue that déjà vu is the likely non-functional by-product (or “side effect”) of other cognitive processes, and while they make similar arguments for IAMs, they seem to indicate that IAMs have, in contrast, many functions (see Table 1 in their article). It is at this point, the question of adaptive functionality (e.g., Baddeley, 1988; Bluck & Alea, 2002), that I depart from their model. It is not clear to me how one (déjà vu) is seen as a cognitive failure and the other (IAM) is not. There is plenty of evidence that IAMs are by-products of other processes (e.g., unique cuing and priming; Ball, 2015; Berntsen, Staugaard, & Sørensen, 2013; Johannessen & Berntsen, 2010; Mace, 2005; Mace & Kruchten, 2022; Mace, McQueen, Hayslett, Staley, & Welch, 2019; Mace & Unlu, 2020, see target article). It is also reasonable to assume, even without this evidence, that IAMs, like déjà vu, may merely be cognitive failures. Thus, my commentary focuses on the idea that IAMs, like déjà vu and other similar phenomena, may be cognitive failures.

Proponents of the view that IAMs are functional have a number of challenges to overcome. For example, questionnaire studies on IAM functions report functional hit rates ranging from one- to two-thirds (e.g., Mace & Atkinson, 2009; Rasmussen & Berntsen, 2011; Rasmussen, Ramsgaard, & Berntsen, 2015). One challenge is to explain how an evolved adaptive mechanism functions with such a low hit rate, presuming these numbers can be accepted. Relatedly, many of the reported functions in these studies can be seen as dubious (e.g., an IAM was entertaining or it made one feel better). Function theorists will also need to show how responses like these are functional, and not merely the function of the means used to assess the question, the survey method. Another challenge is to explain how an evolved adaptive mechanism, one which presumably reads the environment in some way to produce needed memories, interfaces with the cuing and priming variables to accomplish the same task. Indeed, aligning the function hypothesis with variables that seem to independently produce IAMs may prove challenging.

One additional explanatory variable for IAMs is the concept of retrieval mode (Tulving, 1983), a phenomenon which the target article addresses. Though originally discussed by Tulving (1983) as a tentative proposition, there is now good evidence that when the cognitive system is focused on intentional retrieval that stimuli are treated as episodic cues and memories of the past come to mind automatically and effortlessly (e.g., Herron, 2018; see also Rugg & Wilding, 2000). This process has been shown to be a routine, and often dominant, part of voluntary recall in autobiographical memory (e.g., Uzer, Lee, & Brown, 2012, known as direct retrieval, see also Conway, 2005, and the literature on direct retrieval). It is certainly possible that some IAMs may be the product of retrieval mode, a state which may be initiated inadvertently by aspects of the internal and external environment (e.g., Mace, 2010). If some IAMs are the products of retrieval mode gone awry, then this variable, too, would need to be aligned with the functional mechanism approach. However, this could be the more promising case if one could show that all IAMs were a function of the retrieval mode state, and such a state was not functioning erratically, but purposefully. This would raise a set of additional questions (e.g., the role of intent, or covert intent), and it clearly would be difficult to establish (but see Herron, 2018, and the retrieval mode and orientation literature).

Another problem concerns individual perceptions of spontaneous processes. Perceptions of IAMs are likely to differ greatly from déjà vu and other spontaneous processes, like action slips (e.g., Norman & Shallice, 1986; Reason, 1979). A déjà vu is likely to be seen as a false impression, an impossibility, and the circumstance may often reveal that. The experience of IAMs, though, can be quite different. They can be perceived as pleasant, circumstantially congruent relivings of one's past (e.g., Berntsen, 1998). As snippets of one's past, individuals may be more likely to imbue them with meaning, perceive them as functional, or imagine that they must be, even if they cannot see how. Such perceptions are likely to taint any self-report measure, and the design of such measures may only reinforce them. Given that experimental means are currently unavailable to answer questions of function (e.g., see interesting discussions in Downes, 2015), researchers have been forced to rely on self-report measures, like surveys. This presents another challenge for function theorists, as they also need to show that the data from such approaches are largely free of bias, or they can distinguish biased from unbiased data. A good example of potentially biased data are the dubious responses mentioned above, which do account for a substantial portion of the data (see Rasmussen et al., 2015). Researchers need to be cognizant of how personal biases might affect survey data, and guard against their own.

Thus, there are many challenges to the position that IAMs are functional, or sometimes can be. Because reliable means do not exist to adequately answer this question, the default, interim position, that IAMs are cognitive failures like déjà vu, appears to be the logical and prudent choice. Involuntary remembering can be explained in absence of a functional account. A number of variables have been identified as sources of IAMs (e.g., cuing, priming, and perhaps retrieval mode), variables which appear to independently cause them. B&M's model of IAM and déjà vu may work better if they presume the default position for both phenomena. Their model, and others, should treat IAM and déjà vu similarly if and until the function question can be answered. Alternative use accounts are possible (i.e., how memories might get used, Bluck, Alea, Habermas, & Rubin, 2005), but they should not be taken for function accounts.