Andrea Rivadulla-Duró is a postdoc at the Centre for Philosophical Psychology at Antwerp University, working on topics related to mental imagery, emotion, and belief formation.
A post by Andrea Rivadulla-Duró
“But although I now know, intellectually, that this memory was ‘false,’ it still seems to me as real, as intensely my own, as before.”
Oliver Sacks
“That's what the world is, after all: an endless battle of contrasting memories.”
Haruki Murakami
The other day, while going over childhood memories with my sisters, I found out that the summer beach house I had always remembered as blue was actually yellow. The evidence was overwhelming: not only did my sisters attest to it, but a photo album also proved unequivocally that the house was indeed yellow. However, although I immediately surrendered to the evidence, the inaccurate memory of the blue house still remains shrouded in a halo of reality when I evoke it. Although I now believe that the house was yellow, the episodic representation of the house painted in blue is still accompanied by the feeling of pastness and familiarity that usually accompanies episodic memories (Russell, 1921:163). Complementarily, although I can represent the real yellow house concordantly with my updated belief, this representation lacks the phenomenological texture of recollection.
This phenomenon is known as non-believed memories (Mazzoni, Scoboria and Harvey, 2010; Otgaar, Scoboria, Mazzoni, 2014). This label refers to episodic representations that one used to take as memories and that still come accompanied by the phenomenology of memory now, even though it has been clarified that the memory is inaccurate (or even entirely false). Philosophers might claim that, from a factive conception of memory, some of these episodic representations are not indeed memories, but fictions, since the events represented never happened. However, non-believed memories can be only partially inaccurate (e.g., I never summered in a blue house). Following the empirical literature, I will refer to the phenomenon by this label (using italics to note that what is primarily meant by the term memory is that they have the phenomenology of recollection and not that they are entirely accurate).
The phenomenon of non-believed memories allows us to observe in broad daylight how episodic memory functions when new information about an event is obtained. And, as the initial example shows, even when we give full credibility to testimonial information and mistrust our initial recollection, the new information tends not to be smoothly integrated into the episodic memory. The system in charge of episodic memory seems resistant or impenetrable to information being incorporated voluntarily. I claim that the incorrigibility of non-believed memories challenges two important current tenets in the memory literature. First, it casts doubts on the claim that episodic memory is a hypothesis machine about our past with no qualms about integrating information from diverse sources. Second, it compromises theories that see memory as another kind of representation produced by an Episodic Constructive System devoted not only to the simulation of past episodes but also to a wide range of imagined episodes (De Brigard 2014, Michaelian 2016).
Non-believed memories show that episodic memory does not behave as a hypothesis machine about what happened in the past, taking by default information from diverse sources and blending it into an episodic representation. In the case of episodic memories, constraints are in place in blending new information, even when we consider this information true. I suggest that these constraints can be described as a resistance to episodicization: the system in charge of episodic memory exhibits resistance to translating propositional information into the episodic format in the case of episodic memory.
The concept of episodicization tracks well the development of our conception of memory. From a conception of episodic memory as a form of imagination, it has now become a trend to contend that memories are created by an Episodic Constructive System that produces all episodic representations: among them, episodic memory, but also episodic anticipation and counterfactual thinking (De Brigard 2014, Michaelian 2016a, 2016b). To put it briefly, in the case of memory formation, the Episodic Constructive System constructs the best hypothesis about what happened in our past. To this end, the System has no qualms in incorporating information from various sources: information obtained from other events, background knowledge, testimonial information, etc. When bringing an episodic memory to mind, if the Episodic Constructive System could talk, it would say, “after consulting all the available sources, this is my best hypothesis about what happened!”. This view has the advantage of accommodating evidence showing that, when remembering an episode, we often combine information obtained from other episodes (e.g., Brainerd & Reyna, 2005; Loftus, 2005) and sources (e.g., testimonial information; Meade & Roedinger, 2002). Furthermore, it goes in line with neural similarities observed between episodic imagination and episodic memory (Addis et al., 2007; Szpunar et al., 2007, Mullally et al., 2014)).
However, the non-believed memories phenomenon strongly compromises the unitary nature of the Episodic Constructive System. This is because, while episodic anticipation and counterfactual thought allow for their immediate correction when new information is obtained, non-believed memories illustrate that episodic memories, that are in principle generated by the same constructive system, are insensitive to such corrections. Take the following case. Suppose I am invited to what I expect to be a crowded party. After imagining what the party will be like, I feel discouraged by the idea of it being too crowded and decide to stay home. However, when I call the hostess to inform her of my decision, she says that we will only be four. Right after this, I can automatically change my episodic imagining of the party, picturing fewer people and a smaller table. This example illustrates that episodic anticipations are easily updated when we obtain new information about the event represented. In contrast, non-believed memories show that the system in charge of episodic memory does not exhibit episodicization through testimony or nonsocial external evidence (e.g., photographs or videos). This happens even when we take the obtained information to be accurate and agree that what we seem to remember did not occur. Suppose I have a memory of my tenth birthday party in a swimming pool with my cousins. This memory has accompanied me for years. When I evoke it, I seem to re-experience the party at the swimming pool (Tulving 2002: 6), and images of my cousins come with “feelings of warmth and intimacy” (James 1980: 650). Then one day, in conversations with reliable witnesses and after seeing photos, I learn that those who were at the party were my classmates and not my cousins. The evidence is undeniable, and it leads me to firmly believe that my classmates were the ones at the party. However, after learning this, the non-believed memory of my cousins being at the party preserves the qualitative character of memory. When I evoke it, I still seem to reexperience the party with my cousins. On the other hand, if I attempt to represent the party with my classmates, this representation is far from having the phenomenological features episodic memories are accompanied by. It is deprived of feelings of familiarity, warmth, and intimacy, and I do not feel like reexperiencing the episode. While episodic memories are doomed to remain unchanged, episodic representations of the future are easily edited, showing a parallel between the properties of time that the episodes evoke.
This asymmetry between episodic memory and episodic anticipations casts doubts on the claim that the Episodic Constructive System is the one in charge of both. If the Episodic Constructive System is at play in the case of memory, it should allow for the same inputs that it allows in the simulation of other episodes (e.g., in episodic future thought). Given episodic anticipations allow for the immediate and voluntary incorporation of information we take to be accurate, this should also be the case of episodic memory if it is operated by the Episodic Constructive System (Rivadulla-Duró 2022). However, non-believed memories show this does not occur. New information about an event does not equally blend in with the episodic representation in memory as in anticipation. The opacity episodic memory exhibits to the conscious incorporation of information goes against the postulation of a single Episodic Constructive System.
The impenetrability episodic memory exhibits to the voluntary incorporation of information might be due to the relevance of first-hand evidence in the case of memories. Episodic memory conveys not only data (e.g., that my classmates attended the party), but it also conveys that this information was obtained first-personally (e.g., that I saw the classmates and that I remember they were there based on seeing them). In the case of memory, the episodic format not only provides information about events, but about the fact that we obtained this information through experience. Now, suppose we could easily integrate testimonial information we believe into our episodic memories. In that case, episodic memory will mislead us into thinking that we retain first-hand evidence of this information (e.g., that we saw the classmates at the party). The imagistic features of episodic memory have a function; they inform us that what is episodically represented was acquired first-hand. If episodic memory was operated by a system integrating information from all sources and translating it into the episodic format, this would have negative epistemic consequences. We would be led into thinking that information obtained through testimony (second-hand evidence) has indeed been obtained first-personally. In the case of memory, the episodic format has, so to speak, a different epistemic pedigree and the system bringing episodic memories seems to be well aware of this. The summer house we used to go for vacation will remain, in all its forms, blue.
References
Addis, D., Wong, A. A. T., & Schacter, D. L. (2007). Remembering the past and imagining the future: common and distinct neural substrates during event construction and elaboration. Neuropsychologia, 45(7), 1363–1377.
Brainerd, C. J., & Reyna, V. F. (2005). The science of false memory. Oxford University Press.
De Brigard, F. (2014). Is memory for remembering? Recollection as a form of episodic hypothetical thinking. Synthese, 191, 155-185.
James, W. (1890). The principles of psychology (1st ed.). Macmillan.
Loftus, E. F. (2005). Planting misinformation in the human mind: a 30-year investigation of the malle- ability of memory. Learning & Memory, 12(4), 361–366.
Otgaar, H., Scoboria, A., & Mazzoni, G. (2014). On the existence and implications of nonbelieved memories. Current Directions in Psychological Science, 23(5), 349-354.
Mazzoni, G., Scoboria, A., & Harvey, L. (2010). Nonbelieved memories. Psychological Science, 21(9), 1334-1340.
Meade, M. L., & Roediger, H. L. (2002). Explorations in the social contagion of memory. Memory & Cognition, 30(7), 995–1009.
Michaelian, K. (2016a). Mental time travel: episodic memory and our knowledge of the personal past. MIT Press.
Michaelian, K. (2016b). Against discontinuism: Mental time travel and our knowledge of past and future events. In K. Michaelian, S. Klein, and K. Szpunar (eds) Seeing the Future: Theoretical Perspectives of Future Oriented Mental Time Travel (pp. 62–92).
Mullally, S. L., & Maguire, E. A. (2014). Memory, imagination, and predicting the future: a common brain mechanism? The Neuroscientist, 20(3), 220–234. https:// doi. org/ 10. 1177/ 10738 58413 495091
Rivadulla-Duró, A. (2022). The Simulation Theory of Memory and the phenomenology of remembering. Phenomenology and the Cognitive Sciences, 1-21.
Russell, B. (1921). The analysis of mind. Routledge.
Sacks, O. (2013). Speak, memory. The New York Review. Retrieved from https://www.nybooks.com/articles/2013/02/21/speak-memory/?lp_txn_id=1422203
Szpunar, K. K., Watson, J. M., & McDermott, K. B. (2007). Neural substrates of envisioning the future. PNAS, 104, 642–647.