1 Introduction

In The Imaginary, Sartre claims that what he calls “having an image”, by which he means a conscious exercise of imagination or an imagining, is transparent: whenever we are imagining, we are aware of imagining. It follows that certain kinds of confusions or mistakes are excluded. For instance, we cannot have a visual imagining that we are wrongly aware of as a visual or memory experience. In his own words:

[S]omeone who, in an act of reflection, becomes conscious of ‘having an image’ cannot be mistaken. [...] In fact, confusion is impossible: what is conventionally called an ‘image’ gives itself immediately as such to reflection. But this is not a matter of a metaphysical and ineffable revelation. If these consciousnesses are immediately distinguishable from all others, it is because they present themselves to reflection with certain marks, certain characteristics that immediately determine the judgement ‘I have an image’. (Sartre, 2005, p. 4)

Sartre makes analogous claims about memory. Like imaginings, memories are transparent: whenever we are remembering, we are aware of remembering. In particular, we cannot have a memory that we are wrongly aware of as an imagining, since “there is [...] an essential difference between the thesis of the memory and that of the image” (Sartre, 2005, p. 181).

Other authors have made claims similar to Sartre’s. For example, Ricoeur (2004) emphasizes the difference between memories, which belong to the “world of experience” and imaginings, which inhabit the “world of fantasy”, and claims that “they cannot be confused or mistaken one for the other” (Ricoeur, 2004, p. 49).

In this essay, we are concerned with clarifying two impossibility claims that both Sartre and Ricoeur seem to have accepted, namely:¹

1. It is impossible to confuse a memory with an imagining.

2. It is impossible to confuse an imagining with a memory.

As formulated, these claims do not sound very plausible. The notion of confusion they involve is too general. The passage quoted from Sartre suggests two specific interpretations of this notion. On the one hand, the relevant confusion can concern self-ascriptions of imaginings or memories, i.e., explicit judgements or beliefs of the form “I am imagining” (or as Sartre puts it, “I have an image”) or “I am remembering”. The question here is whether it is possible to make false self-ascriptions of imaginings or memories. On the other hand, the confusion can concern more primitive forms of introspection or inner awareness, which do not obviously involve self-ascription or the deployment of explicit concepts of imagination or memory. The question here is whether it is possible for memories to be subjectively felt as imaginings, or imaginings as memories. Can we be wrong about what it is like to have a conscious memory or imagining?

Our contention is that these interpretations are not equivalent.² It is best to distinguish two notions of confusions, which we will call respectively “reflective” and “phenomenological”.³ In principle, then, four impossibility claims should be considered:

1a. It is impossible to reflectively confuse a memory with an imagining.

1b. It is impossible to phenomenologically confuse a memory with an imagining.

2a. It is impossible to reflectively confuse an imagining with a memory.

2b. It is impossible to phenomenologically confuse an imagining with a memory.

We take it that these impossibility claims concern the episodic forms of memory and imagination, and thus specific sensory-like conscious experiences of remembering and imagining. We are not concerned, for instance, with semantic memory or cognitive imagination.⁴

In what follows, we critically examine these impossibility claims, with the aim of clarifying the relationship between imagination and memory. An investigation into the relevant notions of confusion can also improve our understanding of the relevance of empirical studies on memory. If we consider the extensive literature in psychology on memory errors and confabulation, we might think that the debate is already settled: we do indeed confuse memories with imaginings, and imaginings with memories.⁵ However, things are more complicated, since the phenomenological and the reflective notions of confusion are not always carefully distinguished, which can lead to ambiguity about the scope of the empirical hypotheses. Only a theoretical analysis can bring important conceptual knots to the fore and offer a more precise assessment of the kind of errors in question.

We shall start by providing several hypothetical scenarios that contradict each of the four impossibility claims. These scenarios are prima facie coherent, and some, if not all of them, may have an initial plausibility. The role of these scenarios is to clarify the distinction between the two notions of confusion. It does not follow that they are genuinely (i.e., metaphysically) possible, but if they are, the impossibility claims will be wrong whatever level, reflective (§2) or phenomenological (§3), is concerned.

We then spell out the idea of being confused at both the reflective and the phenomenological levels (§4). Moreover, we explicitly relate these levels to the ontological one, which determines whether a particular subject really remembers or imagines something. The addition of the ontological level helps to better understand our scenarios and shows a further type of cases.

It might be wondered whether adding the ontological level makes the impossibility claims bounded to an ontological difference between memory and imagination. Sartre and Ricoeur arguably defended a “discontinuist” view according to which the latter are different types of mental states, or at least different “manners of positing their objects” – in Sartre’s terminology.⁶ Nowadays, the idea that memory is a specific use of the imagination (Hopkins, 2018; Michaelian, 2016) is gaining increasing popularity, due to empirical research indicating a common system, or process, subserving both our capacity to mentally re-live an episode of one’s own past and our capacity to mentally pre-live an episode of one’s future (locus classicus is Schacter & Addis, 2007). However, proving the falsity of the impossibility claims does not necessarily bring water to the continuist’s mill. At least some impossibility claims are compatible with continuist views. For instance, to anticipate, Hopkins, who claims that memory is imagination “controlled by the past” (Hopkins, 2018, p. 47), seems to suggest that reflective confusion is possible (i.e., 1a. and 2a. are false), whereas phenomenological confusion is not (i.e., 1b. and 2b. are correct).

In §5, we examine several lines of objection to the interpretation of our scenarios. All of them turn on the claim that the alleged dissociations do not occur in real life, or only in special circumstances involving psychologically unstable internal incoherences. We respond to the objections by relating at least some of our scenarios to empirically known conditions affecting memory or imagination. The existence of these conditions supports the view that the presented dissociations are genuinely possible, thus opening the door to a systematic exploration of the implications of the falsity of the impossibility claims for the adequate account of the relationship between memory and imagination.

2 Reflective confusion

Let us begin with reflective confusion (i.e., claims a in the introduction), which concerns the level of explicit self-ascription of memories or imaginings, at which the subject forms a judgement or belief about their own mental state. It can be defined as follows:

A subject S reflectively confuses a memory with an imagining, or an imagining with a memory, when

(M-ref) S has an occurrent memory that S falsely self-ascribes as an imagining, or

(I-ref) S has an imagining that S falsely self-ascribes as an occurrent memory.

Two clarifications are in order. First, self-ascriptions of memories can have the form “I remember” or “I seem to remember”/“I have an apparent memory”. Thus, they can be true even if the ascribed memories themselves are false, for instance if they are cases of misremembering. Second, one might worry that while some imaginings could be falsely self-ascribed as memories, a memory cannot be falsely self-ascribed as an imagining, because it is or essentially involves an imagining. However, here and in the following, we use the notion of imagining narrowly to refer to mere imaginings, where a mere imagining is not a memory. Our use is in accordance with ordinary language: when we say “she is imagining it”, we usually mean that someone is merely imagining, not remembering, something. Any account of memory should acknowledge that there are mere imaginings, i.e., imaginings that cannot be identified with memories.

As an illustration of the first type of reflective confusion, consider the famous painter case by Martin and Deutscher.

The Classic Painter

“Suppose that someone asks a painter to paint an imaginary scene. The painter agrees to do this and, taking himself to be painting some purely imaginary scene, paints a detailed picture of a farmyard, including a certain colored and shaped house, various people with detailed features, particular items of clothing, and so on. His parents then recognize the picture as a very accurate representation of a scene which the painter saw just once in his childhood. [...] Although the painter sincerely believes that his work is purely imaginary, and represents no real scene, the amazed observers have all the evidence needed to establish that in fact he is remembering a scene from childhood.” (Martin & Deutscher, 1966, pp. 167–168)

On the intended interpretation of the case, the painter self-ascribes a mere imagining of a house, but his self-ascription is false. In fact, he is remembering the house.⁷

As an example of the second type of reflective confusion, consider the following variation on Martin and Deutscher’s original scenario.

The Misinformed Painter

A painter is painting a beautiful house in the countryside, beeches and birches surround the building. “This is aunt Emma’s house!”, says the painter’s sister. The painter has vague memories of aunt Emma and none of her house. He claims that what he is painting is a mere figment of his imagination. The painter’s sister tries to make her brother realizing that the scene is, in fact, a product of his memory, by providing him with a lot of the details about the house and the games they played when they were kids. The painter gets convinced, sincerely believes that his work comes from his memory, and represents aunt Emma’s house, although in fact he has imagined the house and his sister is wrong.

Contrary to The Classic Painter, in this scenario there is a dynamic. The painter starts by self-ascribing the belief that he is imagining a certain house and indeed the content of his painting is a product of his imagination. There is no reflective confusion here. Then the painter’s sister convinces him that the painted house is exactly like aunt Emma’s. This may sound fancy, but how many times have we needed others to jog our memory?⁸ So, it shouldn’t be difficult to imagine the scene: the persuasion of the painter’s sister, the details about the house and the games they played when they were kids she might give, the impact of her words on the painter. In The Misinformed Painter, however, the painter’s sister is wrong (she may have been mistaken about the house, the aunt, etc. – we leave it to the reader to imagine how and why she was mistaken) and, thus, she is unintentionally implanting false past information in her brother’s head, rather than triggering accurate memories. Here comes the reflective confusion: in the end the painter genuinely self-ascribes a memory of aunt Emma’s house, but his self-ascription is false. In fact, he is merely imagining a house.

The classic painter reflectively confuses memory with imagination, while the misinformed painter reflectively confuses imagination with memory. Are these scenarios possible? The answer depends of course on one’s ontological perspective on the mental states of the painters. The classic painter remembers the house, and the misinformed painter merely imagines it, only if their self-ascriptions are not constitutive of what attitudes they actually have. We will come back to this, especially in the light of empirical studies, but let us note that these scenarios are at least initially coherent, if not plausible.

3 Phenomenological confusion

As suggested in the introduction (claims b), the impossibility claims can rely on another notion of confusion, based on the fact that we feel mental states to be a certain way. This is phenomenological confusion, which can be defined as follows:

A subject S phenomenologically confuses a memory with an imagining, or an imagining with a memory, when

(M-phen) S has an occurrent memory that S wrongly feels as an imagining, or

(I-phen) S has an imagining that S wrongly feels as an occurrent memory.

Take the classic painter again: not only does he self-ascribe the belief that he is imagining, but he is also likely to feel that he is imagining, although he is remembering. Therefore, the Classic Painter illustrates both reflective and phenomenological confusion of the first type (i.e., M-ref and M-phen).

The Misinformed Painter is a trickier case. Arguably, before being persuaded by his sister, the misinformed painter feels to be imagining, but what about after his sister’s speech? Does he feel remembering? Does he cease to feel imagining? Different intuitions arise, depending on how the example is framed. The fact that The Misinformed Painter does not straightforwardly exemplify the second type of phenomenological confusion suggests an interesting possibility: mental states can involve specific feelings independently of whether we eventually self-ascribe them. Unlike self-ascriptions, feelings are not judgements but inner experiences of our own mental life. This leaves room for potential conflicts between them. For instance, the question of whether an experience can be felt as an imagining or a memory even if we believe otherwise at least makes sense. Following this suggestion, we will take the Misinformed Painter to be an example of reflective confusion (I-ref, specifically) only: the misinformed painter feels that he is imagining, while falsely self-ascribing his imagining as a memory.

Some questions become pressing: Can we offer a case of I-phen, possibly not involving I-ref? Differently from The Classic Painter, is it possible to have M-phen without M-ref? Consider the following scenarios:

The Puzzled Painter

A painter is painting a beautiful house in the countryside, beeches and birches surround the building. He strongly feels that what he is painting comes from his imagination. “This is aunt Emma’s house!”, says the painter’s sister and tries to make her brother realizing that the scene is, in fact, a product of his memory – as it is the case. All of a sudden, the painter starts remembering: the aunt, the house, the games he played with his sister when they were kids. However, the painter is puzzled: although he now sincerely believes that his work comes from his memory, he cannot get rid of the feeling of imagining.

The Informed Painter

A painter is painting a beautiful house in the countryside, beeches and birches surround the building. “This is aunt Emma’s house”, the painter says to his sister. He strongly feels that what he is painting comes from his memory. The painter’s sister tries to make her brother realizing that the scene is, in fact, a mere figment of his imagination, by providing him with a lot of details about the real house. Although he cannot get rid of his feeling of remembering, the painter is finally convinced and sincerely believes that his work is purely imaginary, and does not represent aunt Emma’s house. Indeed, he has imagined the house and his sister is right.

While the puzzled painter has genuine memories, which he truly self-ascribes as memories, the informed painter has genuine imaginings, which he truly self-ascribes as imaginings. In our terminology, neither painter suffers from reflective confusion. However, they make phenomenological confusions, about which they are perfectly lucid. The puzzled painter’s memories are wrongly felt as imaginings, though they probably used to be felt as memories. In contrast, the informed painter’s imaginings are wrongly felt as memories. If these scenarios are genuinely possible, the two impossibility claims cannot be maintained at the phenomenological level. Indeed, phenomenological confusion of the first type (i.e., M-phen) is illustrated by The Puzzled Painter and phenomenological confusion of the second type (i.e., I-phen) by The Informed Painter.⁹

Of course, the claim that phenomenological confusions are possible must be supplemented by a theory of the correctness conditions of feelings or what it is like to have an imagining or a memory. For instance, if feelings are construed on the model of inner perceptual experiences, the latter could be illusory or hallucinatory in some contexts, just as outer perceptual experiences. Alternatively, feelings can be construed as metacognitive experiences arising from subpersonal source-monitoring (for which see Simons et al., 2017). Source-monitoring can go wrong, and an imagining might be monitored as a memory, or the other way around, giving rise to a confusion at the phenomenological level (see §5.2 below).

To conclude this section, let us note that at least one author, namely Robert Hopkins, has explicitly acknowledged some degree of autonomy between the phenomenological and the reflective levels. On his view, memories and imaginings have the same phenomenology.¹⁰ It follows that phenomenological confusions are impossible: memories cannot be wrongly felt as imaginings, or imaginings as memories. Still, Hopkins makes clear that they can give rise to different self-ascriptions. He suggests that our spontaneous self-ascriptions can be “brute”, i.e., not grounded on intrinsic phenomenological differences:

[W]e might think there must be some difference in phenomenology between memory and imagining. For how otherwise are we able to recognize the two? Sometimes we are uncertain which state we are in, but often we are not. How do we tell the states apart, if not by exploiting some consciously accessible difference between them? [...] Perhaps [...] it is simply brute that some memories and imaginings bring with them accurate beliefs about their status. (Hopkins, 2018, p. 55)

In the light of our distinction between phenomenological and reflective confusion, what Hopkins suggests here is that the impossibility of phenomenological confusion is compatible with the possibility of reflective confusion. Depending on how brute self-ascriptions are conceived (for instance, what underlying mechanisms they involve), we might have an imagining that we wrongly self-ascribe as a memory, even though we get the phenomenology right (as in The Misinformed Painter).

4 Three levels of explanation

So far we have mainly examined scenarios that illustrate either reflective or phenomenological confusions. We have also suggested that it is possible to be confused at both the reflective and the phenomenological levels. On one reading, the classic painter feels that he is imagining, so he is also confused at the phenomenological level (M-phen, specifically). What about the other case: can genuine imaginings be felt as memories and be wrongly self-ascribed as memories? Consider the following scenario.

The Confabulating Painter

Suppose that someone asks a painter to paint a scene from his childhood. The painter agrees to do this and, feeling to be painting some remembered scene of aunt Emma’s house, paints a detailed picture of a farmyard, including a certain coloured and shaped house, various people with detailed features, particular items of clothing, and so on. His parents then recognize the picture as a very inaccurate representation of Emma’s house. Although the painter sincerely believes that his work comes from his memory, and represents a real scene, the amazed observers have all the evidence needed to establish that in fact he is imagining.

In this scenario, the painter is confabulating, on the hypothesis that confabulation involves imaginings that are felt as memories and are self-ascribed as memories.¹¹ Thus, the confabulating painter is confused at both the reflective and the phenomenological levels (more precisely, I-ref and I-phen).

The cases examined so far suggest that any account of our experiences of memory and imagination should recognize, at least notionally, three levels of explanation, even if one level is eventually considered to be fully determined by another. First, there is the ontological level, where it is determined whether we really have a memory or an imagining. Second, there is the phenomenological level, where it is determined whether we feel that we are remembering or imagining. Finally, there is the reflective level, which deals with our self-ascriptions of the form “I am imagining/remembering”.

Table 1 summarizes the relevant cases, where M and I mark the relevant states as being either about memory or imagination respectively. For instance, the Classic Painter involves a genuine memory (M), which is felt as (I) and is ascribed (I) as an imagining by the painter.

If we set aside the cases in which all three levels are aligned (either M/M/M or I/I/I), there is one remaining line in the table. It corresponds to a case in which the subject has a genuine memory that is felt as a memory but is wrongly self-ascribed as an imagining (i.e., a case of M-ref without M-phen). Here is a possible illustration.

The Mirrored Misinformed Painter

A painter is painting a beautiful house in the countryside, beeches and birches surround the building. “This is aunt Emma’s house”, the painter says to his sister. He strongly feels that what he is painting comes from his memory. The painter’s sister tries to make her brother realizing that the scene is a mere figment of his imagination, by providing him with a lot of details about the real house. Although he cannot get rid of his feeling of remembering, the painter is finally convinced by his sister and sincerely believes that his work is purely imaginary. However, his sister is wrong and in fact the painter is accurately remembering and the work represents aunt Emma’s house.

We have tried to construct scenarios which show dissociations between the nature of our experiences, how they are felt and how they are self-ascribed. Of course, the plausibility of the scenarios does not entail that they are genuinely possible. Perhaps some of our intuitions are misguided, and the phenomenology of memory and imagination inherently shapes the reflective level, and vice versa. Yet we think that the scenarios force opponents to the dissociations to explain why the scenarios are prima facie coherent if not plausible. This suggests a methodology whereby different dissociation scenarios are presented and explicit arguments are given to exclude some or all of the rows in the table above. In the following section we will consider, and try to undermine, three of these arguments.

5 Three lines of objection

5.1 Spontaneous and deliberate self-ascriptions

Consider the relationship between the phenomenological and the reflective levels. As with any judgement (see, e.g., Evans & Frankish, 2009), we can distinguish between spontaneous and deliberate self-ascriptions. The former are made without much thinking, while the latter result from explicit reasoning. The two could come apart. For instance, we may be convinced by our guru, or a psychologist, that we are merely imagining something, while in fact we are at least inclined to self-ascribe a genuine memory. Our deliberate self-ascription is about imagination, while our spontaneous self-ascription is about memory.

Now an opponent to the dissociations between the phenomenological and the reflective levels might argue that they can occur only with deliberate self-judgements. Indeed, all our examples showing a divorce between these two levels (i.e., The Misinformed Painter – both normal and mirrored –, The Puzzled Painter and The Informed Painter) involve deliberate self-judgements based on testimony. These cases do not show that spontaneous self-ascriptions can be dissociated from the way our experiences are felt. In accordance with their feelings, both the misinformed painter and the puzzled painter might be spontaneously led to self-ascribe imaginings. Likewise for the informed painter and the mirrored misinformed painter: they might be inclined to self-ascribe memories. It is still possible that spontaneous self-ascriptions always reflect the phenomenological profile of our experiences. For instance, it could be argued that felt inclinations to believe that we are imagining or remembering are constitutive of the phenomenology of imagination or memory.

One way to resist this argument is to question the relevance, in the present context, of the distinction between spontaneous and deliberate self-ascriptions. It is true that the self-ascription made by the informed painter is deliberate (as for the misinformed painters and the puzzled painter), but it could become spontaneous as time passes and the painter forgets about the origin of his judgement.

Anyway, even if it is relevant to distinguish between spontaneous and deliberate self-judgements, one might insist that spontaneous self-ascriptions can also diverge from the way our experiences are felt. Consider the following example.

The Painter with Déjà Vécu

Suppose that someone asks a painter to paint an imaginary scene. The painter agrees to do this and, taking himself to be painting some purely imaginary scene, paints a detailed picture of a farmyard, including a certain coloured and shaped house, various people with detailed features, particular items of clothing, and so on. All of a sudden, the painter has a sort of déjà vécu experience and strongly feels that his work comes from his memory, and represents a real scene from his past. Anyway, the painter does not pay much attention to this feeling. He continues to sincerely believe that his work is purely imaginary, and represents no real scene. Indeed, he is imagining not remembering.

The Painter with Déjà Vécu is like The Informed Painter: it illustrates a phenomenological confusion (i.e., I-phen) without reflective confusion. However, it is more open to question whether the painter with déjà vécu’s self-ascription of imagination is deliberate, since it is based neither on testimony nor on clear phenomenological grounds. It is conceivable that his self-ascription of imagination was spontaneous before the onset of his déjà vécu experience and did not change its status afterwards.

Here is another possible illustration of the dissociation between the phenomenological level and the reflective level, where the latter is not necessarily construed in terms of deliberate self-judgements. Suppose that the content of a given memory is extremely painful for the subject to accept, because it represents traumatic events that have happened in their childhood. The subject might then refuse to self-ascribe even an apparent memory and insist that they have a mere imagining about a purely hypothetical event. This case is analogous to The Mirrored Misinformed Painter: there is reflective confusion (i.e., M-ref) without phenomenological confusion. Even more than the painter with déjà vécu, the traumatized subject is likely to form their self-ascription of imagination in a spontaneous way. Even if the phenomenology naturally inclines the subject to self-ascribe a memory, they have a stronger natural inclination to self-ascribe an imagining, and the latter inclination trumps the former.

Is there empirical evidence for such a dissociation? Consider the phenomenon called “non-believed memories” (Mazzoni et al., 2010, 2014). These are episodic representations that were once considered memories, and that still evoke the phenomenology of memory, even though the subject does not believe that the event has really happened. The episodic representations themselves can be accurate or not, and thus the subject’s belief can be true or false. Neither case seems to involve a phenomenological confusion: the episodic representations are felt as memories, whether the latter are veridical or not. Is there a reflective confusion? That is, do the subjects self-ascribe anything but at least an apparent memory? Of course, the fact that they do not believe what they seem to remember is compatible with the absence of reflective confusion. However, if they spontaneously refuse to describe their experiences as memories (because of some trauma or experimental manipulation), there is indeed a gap between phenomenology and reflection: there is evidence that they have memory-like episodic representations (non-believed memories do not differ from believed memories in terms of clarity, richness and feeling of reliving) that are not ascribed as memories.

In a nutshell, spontaneous self-ascriptions arguably have more than one aetiology and need not feed from the phenomenology of the self-ascribed experiences.¹² Thus, the subpersonal mechanisms underlying them might not be the same as the mechanisms underlying our feelings of remembering or imaginings, which creates potential gaps between the phenomenological and the reflective levels. Even if such gaps do not occur most of the time, since the phenomenology is arguably an important heuristic cue for the mechanisms of self-ascription, there are cases in which they arise.

5.2 Asymmetries between memory and imagination: ontology and phenomenology

Another way to undermine the possibility of (at least some of) our cases is to insist on certain asymmetries between memory and imagination. Consider the relationship between the ontological and the phenomenological levels. One might acknowledge that the phenomenology of memory can occur in the absence of genuine memories at the ontological level but have doubts about the analogous claim about imagination.

Suppose that the feeling of remembering is metacognitive, in contrast with the feeling of imagining (Dokic, 2014, 2022). On this view, the mechanisms underlying the phenomenology of memory involve monitoring the specific source of the information currently processed, namely whether it comes from one’s past experience or not. This might create a gap between the ontological and the phenomenological levels: we can make a mistake about the origin of our experience and metacognitively confuse an imagining with a memory. For instance, take hyperphantasics, who have imaginings with highly detailed contents compared to ordinary subjects (see Zeman, 2020). If their feelings of remembering rest on monitoring cues such as the density of information (Simons et al., 2017), we can surmise that there will be more false positives than in ordinary subjects: for hyperphantasics more imaginings will be wrongly felt as memories. The existence of such a gap between the ontological and the phenomenological levels is specific to memory according to the metacognitive view, which need not say anything about an analogous gap in the case of imagination.

On this view, what makes an experience a case of remembering is not, or at least not fully, determined by how the experience is subjectively felt by the subject. In contrast, what makes an experience a case of imagining might be fully determined by how the experience subjectively feels to the subject: the phenomenology of imagination entails that there are genuine imaginings at the ontological level. Such a view lends support to the existence of cases like The Confabulating Painter and The Informed Painter (and alike, such as The Painter with Déjà Vécu), where feelings of remembering are severed from the ontological level. It casts doubts, however, on the genuine possibility of The Classic Painter and The Puzzled Painter, insofar as they involve feelings of imagining severed from the ontological level.

Both The Classic Painter and The Puzzled Painter are mere hypothetical cases, so – as stressed above – they might be purely theoretical scenarios based on misguided intuitions. What about whether real-life cases involving the same types of confusions could be offered?

McCarroll and Sant’Anna (2023) have recently drawn attention on a psychological phenomenon underestimated in the philosophical literature, namely cryptomnesia. Cryptomnesics are subjects who take themselves to be imagining a given content, which, in fact, comes from their memory. For instance, Maar (2017) contends that Nabokov’s Lolita is a case of cryptomnesia, because of tight connections between this novel and a 1916 short story by von Lichberg called “Lolita”. Cryptomnesics are not only prone to self-ascribe imaginings, but they also feel to be imagining, failing to recognise the retention of past information. The authors note that cases of cryptomnesia are “structurally analogous to” the painter case by Martin and Deutscher (1966, p. 7). Indeed, in our view, cryptomnesia falls in the same category as The Classic Painter, insofar as it exemplifies a double confusion, both reflective (M-ref) and phenomenological (M-phen).¹³

Now, consider a real-life case which, like The Puzzled Painter, intuitively involves a phenomenological confusion (i.e., M-phen) without reflective confusion. Klein & Nichols (2012) introduce R.B., a patient whose phenomenology of memory has been significantly altered. In his own words:

I can picture the scene perfectly clearly [...] studying with my friends in our study lounge. I can ‘relive’ it in the sense of re-running the experience of being there. But it has the feeling of imagining [...]. (Klein & Nichols, 2012, p. 686)

These authors argue that R.B. has genuine episodic memories, which he truly self-ascribes as memories. In our terminology, R.B. does not suffer from any reflective confusion. In contrast, R.B. present situation results from a phenomenological alteration of his mental life. The best account of this alteration is that his memories used to be felt as memories but are now wrongly felt as imaginings. Analogously to the hypothetical puzzled painter, R.B. does make a phenomenological confusion (of the first type), one which he is perfectly lucid about.

If real cases can exemplify the dissociations among levels we find in The Classic Painter and The Puzzled Painter, should we abandon the idea that there are asymmetries between memory and imagination? Not necessarily. One might acknowledge both the dissociations and the asymmetries. The idea to be further developed is that, compared to cases like The Confabulating Painter, The Informed Painter and The Painter with Déjà Vécu, cases like The Classic Painter (e.g., cryptomnesia) and The Puzzled Painter (e.g., R.B.) would not have the same internal stability from the subject’s point of view.

To put it differently, all these dissociations are genuinely possible, but they might not show the same cognitive dynamics, because of relevant asymmetries between memory and imagination. Here is a way to illustrate such differences in cognitive dynamics. Let’s take that by default a sensory-like experience is felt as an imagining, and it is only when its source is monitored that its phenomenology possibly changes into a memory phenomenology. Source-monitoring mechanisms would be triggered only when certain memory cues are present, such as detailed content. Now it might be argued that false negatives, due to the non-activation of these mechanisms, are more easily correctable than false positives, due to the activation of the same mechanisms. For instance, it might be easier to activate these mechanisms than to stop them. If this is true, the former case (memory associated with the phenomenology of imagination) would be less stable, from the subject’s point of view, than the latter case (imagination associated with memory phenomenology).

5.3 Asymmetries between memory and imagination: phenomenology and reflection

Another asymmetry between memory and imagination, this time concerning the phenomenological and the reflective levels, might cast doubts on the possibility of some of our cases, more precisely those like The Informed Painter and The Mirrored Misinformed Painter. What does tie these sorts of cases and what would be wrong with them? They postulate the possibility to have memory at the phenomenological level and imagination at the reflective level. It might be claimed that this is not a genuine possibility.

Because memory has a dispositional dimension, we can self-ascribe memories, for instance about our last summer vacations, without the phenomenology of memory. We know that we have such memories even if we are not presently remembering the relevant past events. Now consider the following line of argument. The phenomenology of imagination can be associated with self-ascription of memory because the relevant memory can be dispositional: only its content would be manifested in the imagining. We feel that we are merely imagining a particular event but we believe that we remember the event and that we could have the feeling that we are remembering it. There is not much tension here because the gap between the phenomenology of mere imagination and the phenomenology of imagination in a more liberal sense, which is compatible with the phenomenology of memory, is real but tenuous. In contrast, the phenomenology of memory cannot easily co-exist with the self-ascription of imagination. If our self-ascription were true, our feeling about our present attitude would entirely miss the mark, and we would be motivated to eliminate the internal incoherence: either rejecting the self-ascription of imagining or maintaining it in a way that eventually penetrates the phenomenology of our mental state.

On this line of thought, while The Misinformed Painter and The Puzzled Painter (and alike) are genuinely possible cases, The Mirrored Misinformed Painter and The Informed Painter (and alike) are not, or at least do not correspond to psychologically stable situations. Once again, offering real-life cases involving the same dissociation between the phenomenological and the reflective levels as these latter mere hypothetical examples might strengthen our view.

We have already suggested a real-life case on a par with The Mirrored Misinformed Painter, the case of rejected memories. Also in this situation the feeling of remembering is severed from the reflective level: the subject refuses to self-ascribe a memory, insisting that they have a mere imagining, while having a phenomenology of memory.

Hyperphantasia, a condition affecting imaginings mentioned beforehand, can offer another example. Here it is.

The Hyperphantasic

While daydreaming a hyperphantasic starts imagining being in a beautiful library full of shelves. She can picture the scene perfectly clearly: the wooden inlays, the colours of the books, particular items (lamps, ladders, …), various people with detailed features. Although the hyperphantasic sincerely believes that her daydream is purely imaginary, and represents no real scene, she has the strong feeling that it comes from her memory, and represents instead a real scene from her past, which is not in fact the case.

Similarly to The Mirrored Misinformed Painter and the case of rejected memories, The Hyperphantasic shows that the phenomenology of memory can come in the absence of the self-ascription of memory. However, contrary to the former, it involves phenomenological confusion (i.e., I-phen) without reflective confusion. Thus, The Hyperphantasic can be seen as a highly plausible real-life case of the category exemplified by both The Informed Painter and The Painter With Déjà Vécu.

A similar conclusion can be reached as in the previous sub-section: offering real-life cases of the dissociations among levels that we find in The Mirrored Misinformed Painter and The Informed Painter supports the idea that these scenarios are not only merely coherent or even plausible, but also genuinely possible. It does so, moreover, without dispelling the intuition that there are asymmetries between memory and imagination, as far as the phenomenological and the reflective levels are concerned. Further theoretical and empirical work is called to answer the intuition that the internal tension between the phenomenology of memory and the inclination to self-ascribe imaginings (as in The Mirrored Misinformed Painter, rejected memories cases, The Informed Painter, The Painter With Déjà Vécu and The Hyperphantasic) is greater than the internal tension between the phenomenology of imagination and the inclination to self-ascribe memories (as in The Misinformed Painter, The Puzzled Painter and R.B.). Shedding light on these tensions, such as those addressed in the previous sub-section, is of a great importance, insofar as tensions between subjective features at the ontological, phenomenological and reflective levels might also be informative of the nature of memory and imagination, the ways they are felt and self-ascribed.

6 Conclusions

In this essay, we have been concerned with four impossibility claims about memory and imagination. The claims are that it is impossible to subjectively confuse memories with imaginings, or imaginings with memories, at both the phenomenological level of feeling and the reflective level of self-ascription.

The method we have followed in this paper is to put forward merely hypothetical scenarios which illustrate various ways in which confusions between imaginings and memories can take place, instead of working with a single homogenous notion of confusion. This preliminary analysis aims at clarifying the form that such confusions could have. It is needed to understand the precise relevance of empirical studies that seem to contradict the impossibility claims. As we have seen, although some confusions are harder to maintain, and involve some amount of cognitive dissonance from the subject’s point of view, all the dissociations that we have sorted out at the conceptual level seem to be supported by real (pathological or non-pathological) cases involving imagination and memory.

Eventually, it might be that the mechanisms underlying our spontaneous self-ascriptions of memory or imagination are really distinct from the mechanisms underlying our feelings of remembering or imagining, and both are external to what makes a given conscious experience a case of remembering or imagining.

Acknowledgments

We would like to thank the editors of the current special issue, Ying-Tung Lin, Chris McCarroll, Kirk Michaelian, and Mike T. Stuart, three anonymous referees for this journal, the audience at the Successful and Unsuccessful Remembering and Imagining conference where we presented an ancestor of this article, and our students with whom we tested the plausibility of some of the scenarios discussed above. The research is supported by the ANR-17-EURE-0017 FrontCog and the ANR-10-IDEX-0001-02 PSL

References

---

1. We use the phrase “S confuses x with y” non-symmetrically, i.e., as not entailing (but as being compatible with) “S confuses y with x”.↩︎

2. Although we cannot go into the details of these philosophers’ interpretation here (which would require a substantial discussion of their respective theories of consciousness), it is fair to say that neither Sartre nor Ricoeur seem to clearly distinguish them.↩︎

3. In the present context we will not consider confusion due to the unclarity of the thing in itself (except briefly in fn 9), but we will only be concerned with confusion (whether reflective or phenomenological) in the sense of being wrong. Take a parallel with perception. I might confuse red with orange, because, say, the tie in front of me is of an indefinite colour between red and orange. But it might also be that the tie is either red or orange and confusion arises from my being wrong in identifying its specific colour.↩︎

4. Multiple forms of memory are recognized in the literature. A well-known distinction (within “declarative memory”) is that between episodic and semantic memory. As Tulving notably put it: while the latter is “memory for general facts”, the former is “memory for personally experienced events” (Tulving, 2001, p. 1506). Many philosophers have pointed out that ordinary language captures such a distinction (Byrne, 2011; Goldman, 2006). The idea is that reports of semantic memory typically come when “remember” takes a sentential complement (e.g., I remember that water is H₂O), whereas reports of episodic memory call for a gerundival complement (e.g., I remember having drunk a bottle of water in one go). It has been argued that the ordinary use of the verb “imagine” suggests a similar distinction between a propositional and a non-propositional form of imagination. Different labels have been introduced to capture these two forms (e.g., belief-like imagination, cognitive imagination, semantic imagination vs. perception-like imagination, sensory imagination, episodic imagination), and doubts have been cast on the genuine imaginative nature of the former (e.g., Balcerak Jackson, 2016; for a critique see Arcangeli, 2018).↩︎

5. Studies on memory errors are as old as studies on memory tout court. Among the best known of them are those launched by Loftus (2005) with their “lost in a shopping mall” method. On confabulation, see, e.g., Hirstein (2005) and Bortolotti (2020).↩︎

6. We use “continuism” and “discontinuism” as referring to the claims that memories are ontologically continuous, or discontinuous, with imaginings (see also Robins, 2020). In the recent literature, the same labels can refer to slightly different claims, to the effect that memories are ontologically continuous, or discontinuous, with so-called “episodic future thinking” (see Perrin & Michaelian, 2017). The issues are obviously related, as episodic future thinking is typically considered to be a case of imagination (see Michaelian et al., 2020).↩︎

7. Martin and Deutscher (1966) seem to intend this interpretation when they mention that the observers have “all the evidence needed” to establish the fact that the painter is remembering. However, they also make clear that this is a case of memory, if the “only reasonable explanation” for what he painted is that he remembers the scene from his childhood. Other views might lead to other interpretations.↩︎

8. Campbell makes this point by asking the reader to suppose that “your sister, say, is trying to remind you of the oddly shaped window in your childhood” (Campbell, 2001, p. 173).↩︎

9. For the sake of simplicity, we have considered cases in which the relevant phenomenology is sufficiently clear. There are other cases in which the subjects themselves hesitate about the phenomenology of their mental life. Depersonalized subjects, for instance, have memories, but lack the phenomenology of memory (see Sierra, 2009, p. 142; and Dokic, 2022). Moreover, they are not deluded and might well be able to truly self-ascribe memories. Their profile is similar to the puzzled painter, but unlike him they do not have a clear phenomenology of imagining: their memories are not felt as memories, but they are not felt as imaginings either.↩︎

10. To be precise, Hopkins does not claim that memories and imaginings cannot differ in phenomenology, but only that sometimes the phenomenology matches. This is all we need here to make our point. According to him, the phenomenological identity of memories and imaginings is indicated by our intuitions when confronted with cases where we move from one state (either imagination or memory) to the other (either memory or imagination). He holds, pace what some philosophers take to be obvious (e.g., Campbell, 2001), that in these cases there is no shift in phenomenology (Hopkins, 2018, p. 55).↩︎

11. This hypothesis is compatible with Robins (2016)’s claim that confabulating in general contrasts with misremembering, although Robins herself does not offer an account of confabulating as imagining.↩︎

12. This seems to be a consequence of Hopkins’s view that self-ascriptions can be “brute”.↩︎

13. McCarroll and Sant’Anna (2023)’s discussion is also interesting for us because it illustrates how commitments about the relationship between the three levels of explanation work. On the one hand, they write that one can “experience a mental state as an imagining because one believes that one is imagining” (p. 17), suggesting that the reflective level fully determines or at least penetrates the phenomenological level. On the other hand, they rate as “problematic” the hypothesis that “entertaining a content under the attitude of imagining is not necessary for experiencing a mental state as one of imagining” (p. 17), suggesting that the phenomenological level fully determines the ontological level.↩︎