What’s (episodic) memory good for anyway?

What makes episodic memory stand out from the motley array of memory types, running the gamut from working memory through to procedural memory (with many more waiting in the wings!)?

Following Muhammad Ali Khalidi, we can cleave episodic memory from its family of related capacities thanks, in part, to its phenomenology.[1] When I recall the first videogame I played on my Super Nintendo, that memory of playing A Link to the Past on a rainy November day seated on the carpet just a few feet from the TV, is very much mine. Further, it’s this personal aspect which helps one avoid confusing a memory for another cognitive state, such as a daydream. There is a causal chain linking my having been there in 1994 and it actually having rained (I checked for the first time when writing up this post and surprisingly (?) it turns out it did!) and my recollection of that event now. That externally anchored etiological dimension—the causal chain leading up to my recollection—is the second core aspect of episodic memory that Khalidi highlights; tying it to Martin and Deutscher’s earlier (1966) theory.[2]

Khalidi picks out the memory trace as the functional vehicle which bears the right informational relation to the progenerative content and which also correctly traces the causal history linking that initial experience to the present. Importantly, the trace need not correspond to a univocal neural or mechanistic realizer, which Khalidi identifies with the engram, but rather is located with the more rarefied members of Marr’s computational level. Tying everything together, it’s those causal and functional  properties engendered by episodic memory’s dual phenomenal and etiological aspects that license us referring to it as a bona fide cognitive kind.

However, why should we think that episodic memory is cut at the right explanatory size to merit so much of our collective attention? Should we even be trying to save the (recent) appearances?

Charting the history of episodic memory, we arrive at Tulving’s work in the 1970s, where it branches off from other, less personally inflected, forms of declarative (e.g., semantic) memory. Even here we might pause to complicate the picture: after all, we can reach further back to William James to see an objection to this very partition, as he argues that there need not be a bespoke memory faculty, but rather the same ends could be achieved through a synthesis of other abilities, such as imagination, perception, judgment, and belief (1890, 614). The key difference between say a daydream and memory is that the later retains a similar personal, phenomenological aspect highlighted by Khalidi, or as James metaphorically describes it “the electric current, so to speak, between it and our present self” (ibid). James’s view would predict much the same results as Tulving’s albeit without positing a new capacity to handle the process, while at the same time—given its mashup roots—offering less firm ground for any claim of kindhood. I’m not suggesting that James is necessarily correct but merely pointing out that we’d be so lucky if, at the end of the day, Tulving’s view turned out to be both the right one and the one that happens to give us good grounds to think episodic memory is a real kind.

Of course, Khalidi and others will point to the empirical work that’s been carried out, and the pathological cases he reviews present compelling evidence for differential memory deficits that would force James to complicate his account or risk an ad hoc charge. However, these “top-down” cases and the review of “bottom-up” neuroimaging work that precedes them within Khalidi’s chapter motivate an alternative view in which episodic memory is one aspect of a more ecumenical faculty of past-and-forward thinking and imagining. Khalidi dismisses this move by suggesting that the neuroimaging results could be instead interpreted as evidence of a suite of faculties—including episodic memory—all sharing in the same neural resources. That may be the case, but then we’re owed an explanation as to just what would count as decisive evidence against episodic memory’s standing as a proper faculty. Otherwise, we risk a slippery and hard-to-falsify view of episodic memory less amenable to scientific investigation.

A final thought: Russ Poldrack (2016) conducted a corpus analysis where he compared the number of terms in the Cognitive Ontology project that could be traced back to James’s original Principles (about a quarter) and compared it to the number of terms that could be traced back to a 19th century biology textbook from a comparable ontology of biology. The number of traceable terms was orders of magnitude higher in the psychology case. Poldrack has used this example as an intuition pump of sorts: Is it that we got lucky with our (perhaps phenomenologically inflected) initial batch of psychological concepts, or is psychology in need of a conceptual revolution? If the former, then maybe Khalidi (and Tulving) are right that episodic memory is a good and proper line to draw in the cognitive arena, but if it’s the later, then—perhaps in a strange turn of events—we should take lesson from James’s own intuition and lessen the importance of memory for a future cognitive ontology.

Cheng, Sen & Werning, Markus (2016). What is episodic memory if it is a natural kind? Synthese 193 (5):1345-1385.

Michaelian, K. (2011). Is memory a natural kind? Memory Studies, 4(2), 170–189. https://doi.org/10.1177/1750698010374287

Michaelian, K. (2015), Opening the doors of memory: is declarative memory a natural kind?. WIREs Cogn Sci, 6: 475-482. https://doi.org/10.1002/wcs.1364

Poldrack, R. (2016). “How folksy is psychology? The linguistic history of cognitive ontologies.” Russell Poldrack’s Blog. 18 Apr. 2016. https://russpoldrack.blogspot.com/2016/04/how-folksy-is-psychology-linguistic.html

Robins, S.K. (2016). Representing the Past: Memory Traces and the Causal Theory of MemoryPhilosophical Studies, 173, 2993–3013.

Robins, S. K. (2016). MisrememberingPhilosophical Psychology, 29, 432–447.

Robins, S.K. (2017). Contiguity and the Causal Theory of MemoryCanadian Journal of Philosophy, 47, 1–19.

Robins, S.K. (2022). Episodic memory is not for the future. Current Controversies in the Philosophy of Memory, A. Sant’Anna, C. McCarroll, and K. Michaelian (eds.). Routledge (pp. 166–184). PDF


[1] Thanks to Trey for inviting me to write up this post. Also, for folk curious about the ins and outs of the view, I’d recommend that they start with Khalidi’s excellent post summing up his view.

[2] For those curious to dive into the literature, I can recommend Robins’s suite of articles on the topic (2016a, 2016b, 2017, 2022), Michaelian (2011, 2015) and Cheng & Werning (2016).

One comment

  1. Javier Gomez-Lavin raises a formidable and provocative challenge to the view of episodic memory as a cognitive kind, as well as to the validity of many of our current psychological constructs. Starting with the first challenge, taking a leaf from William James, Gomez-Lavin argues that there need not be a dedicated capacity of episodic memory, “but rather the same ends could be achieved through a synthesis of other abilities, such as imagination, perception, judgment, and belief.” He also alludes to “an alternative view in which episodic memory is one aspect of a more ecumenical faculty of past-and-forward thinking and imagining.” He rightly notes that one piece of evidence that I cite in support of the claim that episodic memory is a cognitive kind in its own right has to do with double dissociations with other cognitive capacities. So there’s that. But also, I take it to be a striking coincidence that states that turn out, as a matter of fact, to rightly represent episodes from the past are also by and large those that have a distinctive phenomenology. These properties are not invariably linked. Sometimes we have the phenomenal experience of remembering having done something or participating in some event, but discover, for example, that we were only told about it (and in some such cases we may be remembering the episode of being told about it). At other times, we don’t have the experiential sense of having participated in an event but can reproduce it accurately in a way that suggests that we retain a trace of the experience. Still, it’s a notable fact about us that states with a distinctive phenomenal quality (which Tulving referred to as “autonoetic”) coincide by and large with those that can be shown to retain traces of the past. This suggests that there’s a common cause behind the coincidence: a psychological capacity dedicated to producing such states. Additionally, there’s some indirect evidence that such a capacity would have adaptive value and a number of theories of why it would do so. Moreover, if creatures are equipped with a capacity whose function is partly to represent past episodes it would be highly useful for those representational states to be easily identifiable from the first-person perspective, and a distinctive phenomenology is a good way of ensuring that. This provides an explanation of the striking coincidence.

    Another important challenge raised by Gomez-Lavin builds on Poldrack’s comparison between theoretical constructs in psychology and those in biology: why has biological ontology experienced a revolution since the 19th century, but not psychological ontology? To put it crudely, why are we still talking about concepts and memory, but not gemmules and miasma? There is lots to say here, but as I indicate at a couple of points in the book, psychology is not quite like biology for a couple of related reasons. First, we have a different type of access to our minds than we do to our bodies (including our brains). That’s not to say that our minds are totally transparent to us or that we have first-person authority, but investigating concepts or episodic memory is importantly different from investigating genes or cells. We are creatures who can reflect on our mental states and think about our thinking processes, and we have had millennia to do so, even though many of the sophisticated empirical techniques for doing so are of recent vintage. That may be one reason why there is less of a conceptual revolution in store for us in psychology. Second, a science of psychology that does not make contact with our familiar pre-theoretical categories fails to explain what we take to be important about our mental lives. So a scientific study of the mind cannot afford not to engage with our pre-scientific categories and relate their investigations back to them.

Comments are closed.

Back to Top