Imagistic Imagining

I begin with imagistic imagining. Everything said here about imagistic imagining can equally be said of sensory or perceptual imagining, as those terms tend to be used (as in, for example, Peacocke Reference Peacocke, Foster and Robinson1985; Noordhof Reference Noordhof2002; Martin Reference Martin2002; Byrne Reference Byrne2007). These remarks also apply to ‘perception-like imagining’ as Gregory Currie and Ian Ravenscroft (Reference Currie and Ravenscroft2002: 24–27) use that term. (For Currie and Ravenscroft, perception-like imagining is just one species of a broader class of ‘recreative’ imaginings [Reference Currie and Ravenscroft2002: 9]. Others in that class are ‘belief-like imaginings’ [Reference Currie and Ravenscroft2002: 18], which are captured by the discussion of attitudinal imagining, below.) Imagistic imagining requires, as a necessary feature, the use of mental imagery—imagery keyed to some sense modality or other. This much is uncontroversial among theorists working on imagination. Less settled are the questions of how to define mental imagery, and whether using mental imagery is sufficient for imagistic imagining or, instead, merely necessary. Mental imagery is generally thought of as a kind of endogenously caused perception-like mental state. Bence Nanay usefully characterizes it as ‘perceptual processing that is not triggered by corresponding sensory stimulation in a given sense modality’ (Reference Nanay2018: 127).

As noted, some hold that any active use of mental imagery is an instance of imagistic imagining (Currie and Ravenscroft Reference Currie and Ravenscroft2002: 24–27; Langland-Hassan Reference Langland-Hassan2015, Reference Langland-Hassan2020; Van Leeuwen Reference Van Leeuwen2013), whereas others insist that only some uses of mental imagery qualify (Arcangeli Reference Arcangeli2020; Kind Reference Kind2001). Typically, the reason given for thinking that only some uses of imagery are imagistic imaginings is that only some involve a person's taking the relevant (imaginative) attitude toward a content represented with mental imagery. To the extent that those concerns are relevant to the (dis)continuism debate, I confront them when discussing imagining in the attitudinal sense (below). For now I will focus on the still common view that any use of mental imagery is a case of imagistic (or sensory, or perceptual) imagining.

What becomes of the (dis)continuism debate if we take kind C to be imagistic imagining? It looks as though we have no debate at all. Discontinuists accept that episodic rememberings and episodic future thoughts/episodic counterfactual thoughts, often, or even always, make use of mental imagery. (See, for example, Debus Reference Debus2014: 333–334). Nor do continuists make their case by arguing that episodic remembering, like episodic future thought/episodic counterfactual thought, involves mental imagery. All sides assume as much. So, if there is a substantive debate afoot concerning whether episodic remembering and episodic future thought/episodic counterfactual thought are continuous in being fundamentally the same kind of mental state, the relevant kind is not imagistic imagining.

Attitudinal Imagining

To imagine, in the attitudinal sense, is to take a certain attitude—the attitude of imagining—toward a content (Kind Reference Kind and Kind2016; Langland-Hassan Reference Langland-Hassan2020; Liao and Doggett Reference Liao and Doggett2014; Van Leeuwen Reference Van Leeuwen2013). What are we saying about someone when we say that they are imagining a content? A popular proposal is that the attitude of imagining is more like the attitude of belief than like desire, insofar as it is a cognitive (and not conative) state. Cognitive states are often distinguished from desire and other conative states by their ‘mind-to-world’ direction of fit (Searle Reference Searle1983). To have a mind-to-world direction of fit is to represent the world as being a particular way, and, thus, to be in epistemic need of revision if the world is not that way. Conative states, like wishes and desires, are held to contrast in having a world-to-mind direction of fit in the sense that the role of the state is to motivate its possessor to bring the world into a condition that matches the content of the state. So, for instance, a belief that there is cake mix in the pantry is false and in need of revision when there is no cake mix in the pantry; whereas, a desire that cake mix is in the pantry plays the role of motivating the person who has it to bring there to be cake mix in the pantry and is not false or in need of revision when there is no cake mix in the pantry.

Now suppose that someone imagines that cake mix is in the pantry. Most philosophers hold that such an imagining, like a belief with the same content, represents cake mix as being in the pantry—that it has a mind-to-word direction of fit. However, they also deny that such an imagining is false, inaccurate, or in epistemic need of revision, in cases where there is no cake mix in the pantry. This creates an instability in the definition of attitudinal imagining, as it is not clear what it is to represent the world as being some way (like a belief) while, at the same time, not being epistemically defective when the world is not that way (like a desire). If it is not the role of imaginings to represent the world as it really is (or was, or will be), what are we saying about attitudinal imaginings when we say that, like beliefs, they (really) represent the world as being some way?

The notion of fictionality is often introduced as part of the answer. Neil Van Leeuwen proposes that attitudinal imaginings occur when someone ‘takes a cognitive attitude toward [the proposition] c that nevertheless treats c as somehow fictional’ (Reference Van Leeuwen2013: 221). And Amy Kind characterizes attitudinal imaginings as having a ‘mind-to-world direction of fit’, where the relevant world is ‘best understood to be a make-believe or fictional world rather than the actual world’ (Reference Kind and Kind2016: 5). Questions persist: What is it to represent a proposition as ‘somehow fictional’? Which fictional world is an arbitrary daydream or fantastical imagining supposed to fit? Can imaginings fail to match their designated fictional world? How would we know if that occurred?

This is not the place to explore these questions fully. It suffices to see that the best we can do in characterizing attitudinal imaginings, without becoming submerged in deep theorizing, is to describe them as mental events where we represent a certain content without thereby taking the world to be as we represent. This meshes with the most general definition of imagining provided by Liao and Gendler (Reference Liao, Gendler and Zalta2019) in the Stanford Encyclopedia of Philosophy: ‘To imagine is to represent without aiming at things as they actually, presently, and subjectively are’. Such characterizations are neutral on the question of whether we must make use of mental imagery in the process of representing such contents; and, indeed, philosophers are divided on that question (Kind Reference Kind2001; Langland-Hassan Reference Langland-Hassan2015, Reference Langland-Hassan2020; Van Leeuwen Reference Van Leeuwen2013; Williamson Reference Williamson, Kind and Kung2016). We can therefore leave open whether all attitudinal imaginings, only some, or none at all are also imagistic imaginings, as defined above. Assuming—as many explicitly allow (Langland-Hassan Reference Langland-Hassan2015, Reference Langland-Hassan2020; Van Leeuwen Reference Van Leeuwen2013; Kind Reference Kind2001)—that at least some imagery-involving mental states are attitudinal imaginings, we can now return to the main question to ask if the (dis)continuism debate concerns whether episodic rememberings, episodic future thoughts, and episodic counterfactual thoughts are continuous in the sense that they are all attitudinal imaginings.

Here the answer is a clear no. Episodic rememberings are events where we represent the world as having been a certain way. Like beliefs about the past, they are considered defective, or in need of revision (or, on factive views of remembering, not memories at all), when the world was not the way they represent it as having been. This much is granted by continuists and discontinuists alike. For instance, Michaelian explains that his continuist theory ‘discriminates between memory and episodic counterfactual thought by the requirement that [in the case of episodic remembering] the episodic construction system must aim at simulating an episode from the [actual] personal past’ (Reference Michaelian2016c: 108). To say that the episodic construction system ‘aims at’ an episode from one's actual personal past is to say that its products are in epistemic need of revision when that aim is not met—viz., when the episodic memory does not accurately represent an episode from one's actual personal past. Even for the continuist, the role of episodic remembering is not to represent a content in a way that is ‘somehow fictional’, but, rather, to represent things as they really were. Thus, both continuists and discontinuists will, without any arm-twisting, allow that episodic rememberings are not attitudinal imaginings, and thus not continuous with other states (such as episodic future thoughts and episodic counterfactual thoughts) in virtue of being attitudinal imaginings. We can reasonably conclude that attitudinal imagining is not kind C.

This conclusion is at odds with a recent defense of discontinuism from Sarah Robins (Reference Robins2020), who defends discontinuism on the grounds that, unlike episodic counterfactual thoughts and episodic future thoughts, episodic rememberings are not attitudinal imaginings. She proposes, instead, that episodic remembering involves taking an attitude of ‘seeming to remember’ toward a content, which has stark differences with the attitude of imagining. Robins explains: ‘if I seem to remember receiving a piano for my birthday, then I take it to be the case that I did in fact receive a piano’ (Reference Robins2020: 481). By contrast, ‘when I imagine that I received a piano instead of a bicycle, I do not believe that I received the piano’ (Reference Robins2020: 481). In connecting the attitude of seeming to remember to that of taking it to be the case, Robins is proposing that seeming to remember, unlike attitudinal imagining, is subject to belief-like norms of correctness. If seeming to remember is indeed defined, in part, by appeal to such norms, this is good grounds for thinking that episodic rememberings are not attitudinal imaginings.

The problem with this defense of discontinuism is that the continuist already accepts it. As shown above, in Michaelian's discussion of the ‘aims’ of episodic memory, continuists agree that different accuracy norms apply to episodic rememberings than to episodic future thoughts and episodic counterfactual thoughts. Nor is there reason to think they would object to understanding those differences in norms as entailing differences in attitudes. Thus, when continuists claim that episodic remembering and episodic future thoughts/episodic counterfactual thoughts are all imaginings, it must not be the attitudinal sense of imagining they have in mind.

We could instead interpret the debate between Robins and continuists as concerning which kind of psychological kind is most fundamental—attitudinal kinds or, say, neurocognitive kinds—with Robins (Reference Robins2020) arguing that a difference in attitudinal kinds is the one that should matter. But this is to take a deflationary perspective on the (dis)continuism debate. Continuists and discontinuists are then no longer disagreeing over whether episodic rememberings are cases of imagining; they are, instead, talking past each other, using imagining in different senses—and disagreeing, if implicitly, over which sense corresponds to a more fundamental kind. One could argue that this is all there is to the (dis)continuism debate. But I make the case in the next section that there is more to it than that.

Constructive Imagining

If the class of imagistic imaginings is too big, and the class of attitudinal imaginings too small, perhaps the class of constructive imaginings will be just right. In a Philosophy Compass article distinguishing three distinct uses of imagines active in philosophy (the other two being imagistic and attitudinal imagining), Van Leeuwen (Reference Van Leeuwen2013: 221) characterizes constructive imagining as ‘a constructive process of assembling mental representations’: when we say that X imagines c, in the constructive sense, we ‘express that X is engaged in a process of coming up with mental representations that have c content’ (224); when we speak of constructive imagination as a capacity or faculty, we refer to ‘the capacity to form novel representations’ (224).

One might worry that this notion of constructive imagining is too broad to be of interest. Reasoning, and indeed thinking in general, would appear to involve, as Van Leeuwen describes it, ‘a constructive process of assembling mental representations.’ It is not immediately clear which episodes of stimulus-independent thought will not qualify as constructive imaginings.

And yet, from the perspective of the (dis)continuism debate, the notion of constructive imagining gains some traction. For there is at least one salient mental process that Van Leeuwen's characterization clearly rules out as an instance of constructive imagining: perception. When we perceive the world, we enjoy a sequence of mental representations. But we do not construct a sequence of mental representations. Aside from the effects of moving our eyes and bodies, and various involuntary top-down effects on how perceptual inputs are processed, the sequence of representations we experience in perception is determined by how the world is outside of us. Yes, we are active agents in perceiving the world; and, yes, our cognitive systems make creative contributions to how that input is processed. But it would go too far to say that perceiving involves, in Van Leeuwen's words, ‘coming up with’, ‘constructing’, or freely ‘assembling’ representations. Perceiving remains stimulus-dependent (Beck Reference Beck2018) and is thus not a kind of constructive imagining.

One might see episodic remembering as similar to perception in this regard. Even if episodic remembering is not determined by what is now perceptually before us, it might seem that, in order to be a case of episodic remembering, a mental process must be tightly constrained by one's past perceptual experiences. Suppose, for instance, that we accepted the intuitive (if now roundly rejected) view that episodic remembering is a kind of replay of previously recorded and dutifully stored perceptual experience. On that sort of view, we are no more ‘assembling’ or ‘constructing’ a sequence of representations when replaying memory episodes than we are when having perceptual experiences. We experience the sequence in the very form in which it first impinged upon our senses. If activating this kind of replay is what it is to remember, we have a clear contrast with cases of episodic future thought and episodic counterfactual thought, which, by their nature, cannot be mere replays of past experience. They must instead involve the active construction or assembly of representations into some novel form. They must involve imagining in the constructive sense!

And now, were someone to come along and reject the claim that episodic remembering is a replay of past perceptual experience—holding that episodic remembering, too, is a kind of constructive simulation of perception, where we actively assemble representations in the service of representing the past, without any need for a causal connection to a past perception (Michaelian, Reference Michaelian, Michaelian, Klein and Szpunar2016a, Reference Michaelian2016c)—the (dis)continuism debate would be up and running between the replay theory and this new simulationist perspective. Moreover, the debate would be appropriately conceived as concerning whether episodic remembering, episodic future thought, and episodic counterfactual thought are all forms of (constructive) imagining.

While I think this is indeed the right way to think about the (dis)continuism debate, two possible concerns should be addressed. First, it is true that, even on a simulationist view such as Michaelian's, episodic remembering is subject to constraints of a kind. Recall his provision that episodic remembering ‘aims’ to represent the actual past and is in that sense constrained by how things actually were. For that reason, it might seem that even the simulationist cannot hold that episodic remembering is constructive imagining. However, the form of constraint just noted remains compatible with constructive imagining. After all, episodic future thoughts and episodic counterfactual thoughts are also constrained by how things were in the past. This is not because they are replays of the past, but because how we represent the future, and how we represent what would have been, must, to serve as reliable guides to action, be informed by how things have gone in the past. We are not flying free of all constraints when we aim to represent our futures, or how things would have gone had some counterfactual premise held. Instead, our a posteriori knowledge of how things typically go is relied upon, and serves to constrain, such ponderings. There remains an important contrast between these broadly rational constraints, inherent in simply doing one's best to construct an accurate representation of some situation—be it the past, future, or counterfactual—and the arational constraints at work when simply replaying an earlier-recorded stimulus-dependent experience. The latter is clearly at odds with the notion of constructive imagining, while the former is not.

Second, as remarked, few if any contemporary memory researchers hold that episodic remembering involves a simple replay of past perceptual experience. Instead, most—including most discontinuists—allow that episodic remembering involves a degree of construction, a kind of active assembly of representations in the service of representing the personal past that is influenced by factors over and above whatever perceptual experiences occurred at the time. This might suggest that my characterization of the (dis)continuism debate is inaccurate, in attributing to discontinuists a view they reject. And yet, strong traces of the replay view remain common among discontinuists, insofar as they defend transmissionist accounts of episodic remembering (Michaelian and Robins Reference Michaelian, Robins, Michaelian, Debus and Perrin2018), according to which a successful episodic remembering must recreate at least some aspects of the content of perceptual experiences that were had during the remembered episode—with the original experience effectively transmitting some of its content to the act of remembering. (Such transmissionist theories are closely related to constructive casual theories of memory (Cheng and Werning Reference Cheng and Markus Werning2016; Michaelian Reference Michaelian2011; Sutton Reference Sutton1998).)Footnote ¹ The transmitted aspect of an episodic remembering's content can be viewed as the non-constructed part of the episodic remembering. For the content of an experience to be transmitted, in the relevant sense, an appropriate causal relation must hold between the original experience and the would-be remembering—one which serves to transmit the former's content. Typically, such content is said to be transmitted through the work of a memory trace (Bernecker Reference Bernecker2010; Martin and Deutscher Reference Martin and Max Deutscher1966). While the precise analysis of a memory trace remains a matter of dispute, the general idea is of a mental state that is caused by an act of perception, encodes information about the perceived event, and continues to store that information until some later moment when, by suitably causing an episodic remembering, it allows the event that originally caused it and about which it carries information, to be remembered (De Brigard Reference De Brigard2014; Robins Reference Robins2016). The event is remembered just because the original act of perception transmits some of its content to the episodic remembering via the work of the memory trace.

Discontinuists—including the constructive causal theorists mentioned above—typically hold that successful rememberings must be caused, in part, by a suitable memory trace, while allowing that the content of a remembering can omit aspects of—or even slightly expand upon—the content contributed by the memory trace (and so are also constructed, to a degree) (De Brigard, Reference De Brigard2014). This leaves a clean dispute between discontinuist transmissionists, on the one hand, and simulationists (such as Michaelian), on the other, who do not think that successful episodic remembering requires any such causal condition to be met and who therefore can propose that episodic remembering is entirely constructed, in the constructive imagining sense. And, indeed, this is precisely where the (dis)continuism debate typically occurs: between simulationists, like Michaelian, and causal theorists such as Perrin (Reference Perrin, Michaelian, Klein and Szpunar2016) and Dorothea Debus (Reference Debus2014), who hold that at least some content preservation-via-causation must occur in genuine episodic remembering. Hence Perrin and Michaelian's conclusion, in an overview of the (dis)continuism debate, that ‘the continuist–discontinuist debate may bottom out in a clash of intuitions over the necessity of causation for remembering’ (Reference Perrin, Kourken, Bernecker and Michaelian2017: 236).

At last, we look to have a satisfying answer to question number 3: to be of kind C is to be a constructive imagining.

Episodic Memory from a Functionalist Point of View

Consider again the two success conditions of accuracy and causal provenance. If accuracy is a necessary feature of episodic memory, then episodic remembering is factive. On factive views of episodic remembering, there is no such thing as an inaccurate episodic remembering of some feature x of an event y; one's episodic remembering, to be an episodic remembering, must accurately enough represent things as they were. (How accurate is accurate enough can be left open.) So, factive views of episodic remembering treat episodically remembers as a success term, at least insofar as accuracy is required for success. Yet, on most views of episodic remembering, accurately representing things as they were is not all that success requires. A proper causal provenance is needed in addition. After all, I might accurately represent what it was like to be on the beaches of Normandy during D-Day, on the basis of films and news reports, without successfully remembering those events. Among transmissionists and causal theorists, those who see episodically remembers as a success term will hold that all episodic remembering, in order to be episodic remembering, must not only be accurate but must also bear appropriate, content-preserving causal relationships to the past episodes they serve to recall; they must be caused by memory traces encoded during the episodes remembered and, in that sense, bear proper causal provenance from one's prior experience.

However, it is important to see that there can be transmissionist views of episodic remembering that do not treat episodically remembers as a success term—views that allow for misrepresentational, non-transmissive instances of episodic remembering, while nevertheless maintaining that all successful episodic rememberings will be both accurate and involve transmitted content through causation-by-a-memory-trace. (See Fernández [Reference Fernández2019] for a non-factive view along these lines, but which lacks any explicit appeal to memory traces.)Footnote ² Yet, holding that episodically remembers is not a success term raises questions. For one thing, it presumes a means for typing states as episodic remembering other than by the fact that they satisfy accuracy and causal provenance requirements (where being caused by a memory trace is, for the transmissionist, essential to having proper causal provenance). Doing so might seem at odds with transmissionist views of episodic remembering, as it might seem to suggest that representing transmitted content is not essential to the notion of being an episodic remembering after all. However, that is in fact a misunderstanding, as can be revealed by considering how a functionalist should analyze episodic remembering.

On a functionalist understanding of the nature of mental states, accuracy and proper causal provenance conditions can come into play in how episodic remembering is defined even if episodically remembers is not a success term. For episodic remembering (considered as a type of state) will be understood in terms of its typical causes and effects (or as the kind of state apt to be caused by such and such and apt to cause such and such). As David Lewis explains in one of functionalism's founding documents, ‘the definitive characteristic of any (sort of) experience as such is its causal role, its syndrome of most typical causes and effects’ (Reference Lewis1966: 17, my emphasis). David Armstrong makes similar remarks in A Materialist Theory of the Mind: ‘The concept of a mental state is primarily the concept of a state of the person apt for bringing about a certain sort of behavior’, and ‘apt for being brought about by a certain sort of stimulus’ (Reference Armstrong1968: 82). Like a tuxedo at a children's birthday party, a state can be ‘apt’ for a situation that does not obtain. Accordingly, a standard functionalist definition of pain—taken from the Stanford Encyclopedia of Philosophy's entry on functionalism—is ‘[t]he state that tends to be caused by bodily injury, to produce the belief that something is wrong with the body and the desire to be out of that state, to produce anxiety, and, in the absence of any stronger, conflicting desires, to cause wincing or moaning’ (Levin Reference Levin and Zalta2018).

Extended to the case of episodic remembering, this suggests a transmissionist definition of episodic remembering along the lines of the following:

Call this a functionalist transmissionist account of episodic remembering. For the functionalist transmissionist—who resists seeing episodically remembers as a success term—the fact that successful episodic remembering requires the fulfillment of a causal provenance condition (via causation by a memory trace) appears in the functional definition of episodic remembering as a typical cause of this kind of state. On such a transmissionist view, it is an essential feature of the type of state that episodic remembering is that its instances are typically caused by relevant memory traces. (By analogy, on non-factive accounts of perceptual experience, it is a typical cause of a perceptual experience as of an x that one's sensory transducers are registering a signal from an x, even if instances of that kind—such as hallucinations as of an x—can occur in the absence of such a cause.) By defining a mental state in terms of its typical causes and effects, the functionalist allows for situations—such as misremembering (or, in the case of perception, misperceptions and hallucinations)—where an instance of the kind occurs with nonstandard causes.

Why the Continuist Is Probably Already a Discontinuist

The foregoing foray into functionalism was in the service of showing how there can be a plausible transmissionist account of episodic remembering that does not treat episodically remembers as a success term. The relevance of this point to the (dis)continuism debate will be clarified soon. Going forward, I will call transmissionist theories that use episodically remembers as a success term factivist transmissionist theories; and I will call those that do not use episodically remembers as a success term functionalist transmissionist theories. Functionalist transmissionists and simulationists, such as Michaelian (Reference Michaelian2016c), agree that episodic remembering is not factive and that episodically remembers is not a success term.Footnote ³ They each allow for cases of episodic remembering—considered as a certain type of occurrent mental state or process—that are misrememberings, or merely cases of seeming to remember. And yet they disagree on whether successful episodic remembering requires transmission of content or causation by a memory trace. Simulationists (and continuists) hold that it does not.

Now, to see the (dis)continuism debate in the clearest terms possible, we should focus on perspectives that, aside of their (dis)continuism disagreement, are as like-minded as possible—views that differ only with respect to whether episodic remembering and episodic future thought/episodic counterfactual thought are continuous. Simulationists and factive transmissionists make a poor contrast group in this case because they differ both with respect to the factivity of episodic remembering and the (dis)continuity between episodic rememberings and episodic future thoughts/episodic counterfactual thoughts; whereas, functionalist transmissionists and simulationists only disagree about the latter. Thus, focusing on the debate between functionalist transmissionists and simulationists will throw the disagreement fueling the (dis)continuism debate into sharpest relief.

However, when we sharpen our focus in this way, the (dis)continuism debate begins to look insubstantial. For even though simulationists allow for instances of successful episodic remembering that do not involve transmitted content, they do not deny that instances of episodic remembering typically represent content transmitted from perceptual experiences had at the time of the episode remembered. That is, for any arbitrary act of episodic remembering—Mary remembering the time her family visited Disney World, say—the simulationist does not deny that some causal-transmission relationship likely holds between the content of her memory and experiences she had at Disney World. They simply maintain that the content's being transmitted in this way is not essential to the state's being a successful episodic remembering. Indeed, Michaelian himself notes that while ‘we should abandon the causal theory in favor of the simulation theory of remembering, the simulation theory likewise invokes [memory] traces’ (Reference Michaelian2016c: 77). And further: ‘While simulation of a given past episode presumably often draws on information originating in the agent's experience of that particular episode, it will rarely draw exclusively on such information, and in principle it need not draw on such information at all’ (Reference Michaelian2016c: 103, Michaelian's emphases).

What matters, from a functionalist point of view, is what is typically true of an episodic remembering—not what might qualify as an episodic remembering in principle. So then, for Michaelian, just how often is often? He does not explicitly say. Often certainly could stand for typically, in the quotation above. The upshot is that the simulationist need not disagree with the functionalist transmissionist's characterization of episodic remembering as that type of occurrent state that is typically caused, in part, by a memory trace, and that has thus and such typical effects on other mental states and behavior (including, for example, that it causes the person who has it to judge that they witnessed the event at which the memory trace originated).

Of course, this is not how the simulationist in fact defines episodic remembering. Comparing his simulationist view to what he calls the ‘causal theory of constructive memory’, Michaelian notes that, while the causal theory of constructive memory is more open than other causal theories in the kinds of differences it allows between the content of an initial perceptual experience and that of a subsequent successful memory, it ‘nevertheless assumes that remembering necessarily involves preservation of some content between experience and retrieval. On the simulation theory, in contrast, neither the appropriate connection condition nor the approximate content similarity condition need be satisfied’ (Reference Michaelian2016c: 111).

He moves from this claim to conclusions such as ‘Remembering is not different in kind from other episodic constructive processes’ (Reference Michaelian2016c: 103) and ‘We tend to assume that it is one thing to genuinely remember the past and another to imagine it, but that turns out not to be the case’ (120).

But notice, first, that Michaelian seems to have a factive transmissionist theory (standardly called causalism) in mind when he suggests that, on the causal theory of constructive memory, remembering necessarily (instead of typically) involves preservation of content. For remembering does not require preservation of content between experience and retrieval on a functionalist transmissionist account, as such transmissionist accounts allow misrememberings as instances of episodic remembering. Second, note that he emphasizes what is strictly necessary for memory when characterizing the simulation theory, and not simply what is typical. The causal connection condition and content similarity conditions ‘need not’ be satisfied by a successful episodic remembering, he tells us (Michaelian Reference Michaelian2016c: 103). But this is consistent with their typically being satisfied by successful instances of episodic remembering and by episodic rememberings in general.

Thus, the respective foci on factive transmissionist accounts as adversaries and on what can possibly count as a successful memory on the simluationist theory mask the fact the simulationist and functionalist transmissionist can both accept the functionalist transmissionist definition of episodic remembering sketched above. And yet, if the continuist (simulationist) and discontinuist (functionalist transmissionist) agree on the definition of episodic remembering, how can they still disagree over whether episodic remembering is continuous with imagining? Is there really a substantial debate left to be had?

It might sound as though there is when Michaelian comments, ‘what it is for a subject to remember, according to the simulation theory, is for him to imagine an episode belonging to his personal past’ (Reference Michaelian2016c: 111). Assuming that imagine here means constructively imagine—which, as shown above, is the most charitable reading—this does not sound like the kind of state or process that typically involves transmitted content. However, Michaelian is careful to add, ‘it is an oversimplification . . . to say that remembering just is [constructively] imagining the past’ (Reference Michaelian2016c: 105). For something to be a case of episodic remembering (successful or not), Michaelian adds that the state must be ‘produced by a properly functioning episodic construction system which aims to produce a representation of an episode belonging to S's personal past’ (Reference Michaelian2016c: 105). This condition is added to distinguish remembering, as a type of mental state, from ‘other forms of episodic imagination’, such as episodic future thought and episodic counterfactual thought, and from cases of merely coincidentally accurately representing some event from one's past (which Michaelian terms ‘merely imagining’ the past) (Michaelian Reference Michaelian2016c: 105). A natural thought is that one thing a properly functioning episodic construction system will do when it aims to produce a representation of an episode belonging to S's personal past is represent transmitted content, through a causal link to a memory trace. And, as shown above, Michaelian seems to allow that this is so—at least ‘often’.

Thus, it does indeed appear that even Michaelian is able to share the same definition of episodic remembering as any functionalist transmissionist who allows for a degree of construction in episodic remembering. Indeed, both sides could consistently add to the above functionalist transmissionist characterization of episodic remembering the simulationist's condition that any case of episodic remembering must be, in Michaelian's terms, ‘produced by a properly functioning episodic construction system which aims to produce a representation of an episode belonging to S's personal past’ to arrive at a comprehensive characterization that preserves the emphases of each. This would give us the following combined definition of episodic remembering:

Short of an argument from the simulationist that episodic remembering does not even typically preserve transmitted content via a memory trace, both sides seem to agree on their basic characterization of episodic remembering. Differences remain only at the outer reaches of each theory. The continuist simulationist holds that cases of successful episodic remembering that preserve no transmitted content from a prior perceptual experience are possible in principle (Michaelian Reference Michaelian2016c: 103); while the (functionalist transmissionist) discontinuist holds that such episodes should never be considered successful remembering, even if they still count as episodic rememberings in satisfying the functional syndrome definitive of such. In short, the disagreement concerns not which mental episodes are cases of episodic remembering—here they agree—nor on how episodic remembering is to be defined, but on whether or not some in a small subset of episodic rememberings are to be considered successful rememberings. It would be wrong to view that dispute as a disagreement over whether episodic remembering is, like episodic future thought and episodic counterfactual thought, a kind of constructive imagining—or, indeed, over whether episodic remembering and episodic future thought/episodic counterfactual thought are continuous with each other.

I return now to the factive transmissionist view (which coheres with causalism or the causal theory of memory as it is most commonly defended). Here there is a deeper disagreement with simulationism. The factivist does not just hold that episodic rememberings typically preserve content transmitted from a past perceptual episode, but asserts, more stringently, that there simply are no cases of episodic remembering that fail to preserve such content. Here we see a clear disagreement with the simulationist over which mental episodes count as episodic remembering. Many mental episodes that the factivist does not count as episodic remembering at all—because they are misrememberings—are still instances of episodic remembering for the simulationist. (For although the simulationist requires that episodic rememberings are the products of a reliable process, a reliable one need not be infallible.) When this difference is combined with the fact that the simulationist allows, in principle, for successful rememberings that preserve no content via a memory trace, while the factivist transmissionist does not, we look to have something more like a disagreement about the nature of memory. Yet, though there is indeed a disagreement about memory here, it has very little to do with whether memory is properly viewed as constructive imagining. For one thing, the main disagreement now is over whether memory is factive. This is a disagreement that the factive transmissionist (such as Debus Reference Debus2014) also shares with functionalist transmissionists (such as—with some amendments—Fernández Reference Fernández2019), despite their both rejecting the idea that remembering is constructive imagining. And while it remains a difference that the simulationist allows for successful episodic remembering that does not preserve transmitted content, we have seen that Michaelian's own simulationist view does not require that most, or even very many, instances of successful episodic remembering are cases of constructive imagining—only that it is possible for there to be some.

It might now seem that the entire debate has been a mirage, caused by the failure to appreciate two things: (1) the possibility of a functionalist (and non-factive) transmissionist view and (2) the fact that Michaelian's simulationism is consistent with a memory trace being a typical cause of episodic rememberings. If the combined functionalist-transmissionist and simulationist definition of episodic remembering above is correct, then episodic remembering is not, as a type, constructive imagining. For constructive imagining, as a type, is not typically caused by a memory trace, and therefore not constrained to represent content transmitted from a past perception. A-events and B-events (above, section I) are not all events of kind C. Thus, whether or not episodic future thoughts and episodic counterfactual thoughts are themselves well-viewed as constructive imaginings, episodic rememberings will not be continuous with episodic future thoughts and episodic counterfactual thoughts in falling under the kind ‘constructive imagining’. (Note that while previous defenses of discontinuism have been tethered to factive causalist theories of memory, both functionalist transmissionists and simulationists can accept discontinuism while rejecting causalism—where causalism is construed as a factive theory. In general, discontinuism does not require a factive theory of memory).

A possible objection—echoing an earlier one of my own—is that if my characterization of kind C as constructive imagining makes the debate an obvious win for discontinuism, this casts doubt on constructive imagining really being the kind at issue. My reply is that identifying the notion of constructive imagining as kind C did not leave us with an obvious victor. Instead, it left us roughly where continuists and discontinuists already saw themselves: as differing, somehow, over the role of causal connections in memory. It was only once we picked apart the question of whether any token instance of episodic remembering must, to be a case of episodic remembering, bear an appropriate causal connection to the event it represents (the factivity question) from the question of whether episodic remembering, as a type of state, should be said to have something like memory traces as a typical cause (the success condition question) that things started to tilt in favor of discontinuism.