1 Introduction

In many cases, our intentions determine what we are imagining. When I imagine the flowing rivers of a distant planet, the fact that I am imagining those rivers—and not, say, a nearby stretch of the Ohio river—is determined by my intention to imagine them. It would be odd for someone to interrupt my musings to claim that, despite my best efforts, I’ve been imagining the Ohio all along. Even if my image rather resembles the Ohio river, and even if the image causally derives from my past viewings of the Ohio, my intention to imagine a merely possible extraterrestrial river is sufficient for such a river to be what I am, in fact, imagining. Or so it seems.

Interestingly, other cases point in the opposite direction. Here is an example from a recent paper by Daniel Munro and Margot Strohminger:

You’re visiting Cambridge for the first time, and a friend is showing you around. She points to a certain college and says, ‘That’s King’s College.’ In fact, the college she was pointing to was Trinity College. From this experience, you store an image in memory that exactly resembles Trinity College, which you take to be a memory of King’s College. Later, you form an intention to imagine King’s College on fire and act on that intention. Your mental image resembles Trinity College, and its causal history traces back to the experience in which your friend introduced Trinity College to you as King’s, via your memory from that experience. (Munro & Strohminger, 2021, p. 11851)

What college did you imagine engulfed in flames? There is a good case to be made that you imagined Trinity, despite your intention to imagine King’s. The fact that you intended to imagine x is in this case insufficient to render x the true object of your imagining. Or so it seems.

We face a decision. We can take the two kinds of case at face value and conclude that a person’s intentions only sometimes fix the object of their imagining. This is the verdict Munro and Strohminger (“M&S,” hereafter) recommend. Going this route, one owes an account of the conditions under which someone’s intentions will and will not suffice to fix the object of their imagining. We will want to know why, if our intentions fix the object of our imaginings in some cases, they do not in others.

A second option is to deny the appearances—or to explain them away—by holding that, in fact, our intentions always suffice to determine the object of our imaginings (or, alternatively, never do). Going this route, one owes an account of why the appearances are mere appearances. That is the strategy and position I will defend: short of a neural blip, or a bout of insanity, an intention to imagine x suffices to make x the object of the resulting imagining. This is, historically, the default position—one that, as M&S rightly note, tends to be asserted without much argument (as in, e.g.; Langland-Hassan (2016); McGinn (2006); Noordhof (2002); Kind (2019); and Dorsch (2012)). Roused from my dogmatic slumber by Munro and Strohminger’s provocative article, I now aim to put the default view—which they term “intentionalism about imagination”—on better footing. However, while what follows is pitched as a dispute with M&S’s account of things, it is equally a conversation with myself. I have felt pulled in both directions on this issue, and this paper is my attempt to understand why.

Here is a road map to what follows: Section 2 argues that M&S fail to explain how we ever manage to imagine objects that are not the causal sources of our images. I then show that, in order to remove the impression that one has imagined what one didn’t intend to imagine, we need to remove certain false beliefs in the imaginer. This helps to undermine the key analogy that M&S propose between sensory imagination and non-propositional seeing and, in turn, the intuition that we sometimes imagine what we intended not to imagine. Section 3 fills in the details of a positive analysis of the error that occurs when we’re tempted to say someone hasn’t imagined what they intended. I propose that, while such individuals imagine the object they intend, they do not do so in the manner they intend. Section 4 builds on that idea by showing how it meshes with a plausible account of what it is to imagistically imagine—and, by extension, of what it is to intend to imagistically imagine. This sets the stage for a further argument against anti-intentionalism: namely, that there is no plausible way to specify what it is to intend to imagine that coheres with anti-intentionalism. Section 5 addresses the issue of unintentional imaginings. And Section 6 concludes by explaining how intentions may fix the objects of episodic rememberings in much the same way they do for other imagistic imaginings, thereby bolstering the argument—already familiar within debates about memory—that remembering is a kind of imagining (Addis, 2020; Langland-Hassan, 2021; Michaelian, 2016; Michaelian et al., 2020).

2 Removing misapprehensions: How the (supposed) default really gets waived

No one, to my knowledge, holds that a person’s intentions never suffice to determine what they are imagining. Perhaps the reason for this is that the best candidate for an object-determining factor for an imagining, other than one’s intentions, is a particular kind of causal relationship the imagining bears to a past act of perception. And, in many cases, we will want to hold that the object of an imagining is nothing the imaginer has ever perceived. Much imagining occurs in the context of consuming fictions, after all. The psychology of fiction consumption—and aesthetics generally—would be upended if we could not imagine fictional characters we’ve never perceived. And it is certainly counterintuitive to hold that, in the most whimsical fantasies, we are only imagining certain objects and scenes we have seen before—and that the imagination of mere possibilities is strictly-speaking impossible. Of course, one could plausibly hold that all sensory imagining draws on information and abilities acquired through past episodes of perception, and in that way causally derives from past perception. But that is a far cry from the thesis that the only things we can imagine are the very things we have perceived.

There is an additional difficulty for the view that we can only imagine what we’ve perceived. Imagine a yellow pencil writing on a white piece of paper. This imagining likely draws on past perceptions of many pencils and papers. It would be implausible to hold that, therefore, you in fact imagined all the past-perceived pencils and papers that in some way causally contributed to the imagining. Very many humdrum imaginings of this sort will have no single causal source. Thus, the single object of such imaginings cannot be explained in terms of a single causal source. Another account is needed, at least in some cases.

For these reasons, Munro and Strohmingher need a way to put the breaks on their own argument—to articulate a principled reason why the causal source of an image does not always determine its object. To that end, they propose a distinction between a “default” mode of imagining, on the one hand, and situations where a certain sort of intention suffices to override that default, on the other. Here is what they say:

According to our account, mental images—and consequently the sensory imaginings to which they belong—by default represent whatever objects to which they causally trace back in experience, unless a subject intends or stipulates that the image be used to represent some other object. (Munro & Strohminger, 2021, p. 11855)

On their telling, when you intend to imagine King’s College but in fact imagine Trinity, this is because you are imagining in a “default mode,” where the object of the imagining is the historical casual source of the mental image. Your intention to imagine King’s College is not sufficient to override this default. Nevertheless, there is a way of overriding it. To do so, one must intend or stipulate “that the image be used to represent some other object.”

To assess this proposal, let’s take ourselves out of the story and suppose that Sam is the person who once saw Trinity yet was told it was King’s. Sam later intends to imagine Kings College on fire but, it seems, nevertheless imagines Trinity, the causal source of the mental image used. For Sam to have overridden the default, he would have needed to intend to use the image “to represent some other object.”

Now, this is a bit puzzling because, in the first telling of the story, Sam did intend to represent “some other object” than Trinity. He intended to imagine Kings. Yet, it seemed, this was not sufficient for him to have overridden the default.

M&S anticipate this reaction and reply as follows:

Our subjects [in the first case] simply intend to imagine King’s College and act on this intention by bringing to mind an image which they already take to be of King’s College. So, their intention is not to stipulatively use their mental image to represent King’s College. The mere intention to imagine King’s College isn’t an intention to use a particular mental image to represent some particular object, which is an intention with a more complex content than the one on which our subjects act. (Munro & Strohminger, 2021, p. 11857, emphasis in original).

M&S’s idea is that, in the default mode, one doesn’t form intentions about one’s image at all. One simply intends to imagine King’s College and proceeds in the normal way. This default mode is only suspended through a metacognitive act of noticing one’s own mental image and stipulating or intending of it that it be used to represent a certain particular—one that need not be its causal source.

So, perhaps the key to overriding the default object of an image is to direct a metacognitive thought at the image itself, stipulating of it that it be used in a certain way. This is not unreasonable on its face. However, on reflection, such a metacognitive intention doesn’t seem to change matters. Suppose that Sam remains convinced that his mental image causally derives from King’s, when in fact it derives from Trinity. Now he brings that image before his mind and thinks: “I intend to use this image that derives from King’s College to represent King’s College on fire.” He has now had a metacognitive intention about the image itself that stipulates its intended use. Yet, intuitively, the metacognitive intention does not change anything. There is just as much temptation as before to say that he unwittingly imagines Trinity—that he doesn’t imagine what he intends. The upshot: M&S have not, in fact, provided an account of how the “default” object of an imagining gets waived. This is a real problem, as the plausibility of their account depends on an explanation of how the supposed default sometimes gets waived.

However, there is a possibility in the vicinity of M&S’s proposal worth considering. Suppose that Sam again calls to mind a mental image that derives from his having seen Trinity and thinks: “I will now use this image, whatever its causal source may be, to imagine King’s College on fire.” Does he succeed? In this case, I think the answer is yes. The intuition that he still unintentionally imagines Trinity has dissipated. Why? What has changed? The difference is not a matter of his now intending to use the image to represent King’s. He did that before. Nor is there a difference in the causal source of the image. It is still Trinity. Nor is it a matter of his now having a metacognitive intention about the image itself and its intended use. That, too, was in place in the previous example. The key difference seems to be that Sam is no longer laboring under a misconception. The specification of his intention reveals that he harbors no assumption that his image causally derives from King’s.

With this in mind, we can see that there are ways to waive the supposed default without use of any metacognitive intention directed at the image. What is important is simply to clarify, in our description of Sam, that he is not wrongly assuming that his imagining causally derives from a past sighting of King’s College. We can do this by stipulating that Sam believes he has never seen King’s College. Let us suppose that he has completely forgotten about the time his friend pointed to Trinity college and incorrectly labelled it as ‘Kings’ (even if his image still derives from that sighting of Trinity). He now has the thought: “While I am certain I have never seen King’s College, I imagine that on fire it would look like so,” where ‘like so’ is a linguistic stand-in for what is in fact his Trinity-derived image. Here again, I think there is no temptation to say that he has really imagined Trinity.

There are two important conclusions to draw from this. First, while metacognitive intentions of a certain sort can appear to waive the “default” object of an imagining, it is not in fact the metacognitive nature of the intention that matters, but rather the removal of the misconception that one is causally linked (through past perception) in a certain way to what one intends to imagine. When we remove that misconception in the imaginer, the temptation to hold that they are not imagining what they intend dissipates. Second, the fact that we can modulate intuitions about whether someone imagines what they intend to imagine by modulating whether they have a misconception about their past perceptual history suggests that the lack of success they may seem to have traces not to the actual causal history of their image, but to their beliefs about its causal history. After all, if it were its actual causal history that mattered for reference, and not one’s beliefs about its causal history, then changing such beliefs without changing the causal history would have no effect on what is really imagined.

There are further ways that the observation that beliefs and intentions of the right sort can waive the supposed default object of an imagining (without changing an image’s causal history) undercuts the central motivation for M&S’s claim that intentions do not suffice to determine the object of an imagining. Much of the appeal of their anti-intentionalist view traces to an intuitive similarity between non-propositional seeing (e.g., seeing Trinity College) and non-propositional, sensory imagining (e.g. imagining Trinity College). When Jim sees Trinity College during a tour of Cambridge, it is Trinity he sees regardless of any beliefs or intentions has on the matter. Even if he is fully convinced it is King’s, and even if he intends to be visiting King’s, it’s still Trinity that he is seeing. Likewise for the squirrel in a nearby tree: it sees Trinity without believing that it does.

M&S want us to think about imagistic imagining (in its default mode) in the same terms. What it is like to imagine Trinity College is very similar to what it is like to see Trinity College, after all. And if one can see Trinity College without intending to do so—and without believing that one is doing so—it may seem natural to say that one can likewise imagine Trinity College without intending to do so, and without believing that one is doing so.

Yet there is a striking disanalogy between seeing and imagining: all sides agree that there simply is no “waiving of the default” in the case of non-propositional seeing. Staring at Trinity, Jim cannot waive the fact that he is seeing Trinity and instead see King’s by intending to do so. Changes in beliefs or intentions just aren’t the right sorts of thing to change what he is seeing. This is because perception, by its nature, requires the real-time uptake of information from the external environment through ongoing causal contact with what is perceived. That is what perception is for. There is no (non-propositional) perceiving x without ongoing causal contact with x. By contrast, it is in the nature of non-propositional (imagistic) imagination to represent things we are not in causal contact with and that we have never perceived—we imagine unicorns, Aristotle, and the dark side of the moon. So, if M&S were right that intentions and beliefs cannot always determine what we are imagining, the reason for this cannot be the same reason they cannot alter what we see. Moreover, we have seen, independently, that changes in beliefs and intentions do seem to be the right sort of thing to change appraisals about what is imagined.

So, the analogy between non-propositional seeing and non-propositional imagining does not run as deep as their phenomenological similarity might suggest. It remains possible to hold that the analogy still holds intermittently—that there are times when an intention overrides an object of an imagining that otherwise (like perception) would have been determined by the causal history of the image. However, the instability in the analogy should motivate us to consider an alternative strategy that dismisses the analogy outright and provides an alternative account of the intuitive error that occurs when we are tempted to say someone hasn’t imagined what they intended—an account compatible with intentions always determining the objects we imagine. I begin to fill in the details of that approach in the next section.

3 Imagining what you intend without imagining as you intend

It is time to give a fuller positive account of the failure of imagination that actually occurs in the cases where we may be tempted to say that intentions do not fix the imagined object. This will help to put intentionalism about imagination on better footing, by acknowledging the sense in which certain imaginings are not fully successful (even if, in undergoing them, we still imagine what we intend). Let’s return to the case where Sam brings to mind an image deriving from Trinity and thinks “I hereby use this image that derives from King’s College to represent King’s College on fire.” Intuitively, there is something infelicitous in his imagining, despite his metacognitive stipulations. The lack of full success is traceable to his beliefs and intentions: first, he falsely believes himself to have seen Kings College; second, he intends to use an image deriving from King’s College to imagine King’s College. He fails in this effort, as the image does not in fact derive from King’s. Importantly, we can see that his imaginative intentions are not fully satisfied before we even decide whether he in fact imagines King’s College. For whether or not he succeeds in imagining King’s, we know that he doesn’t do so through use of an image deriving from King’s. Leaving open what he imagined, we know he did not imagine as he intended.

This suggests the following account of the overall situation: Sam in fact imagines King’s College, but not in the manner he intended. It is the latter failure that accounts for the intuition that he did not imagine as he intended, even if he did, in fact, imagine the object he intended. As this account would predict, when Sam explicitly drops the intention to use an image deriving from King’s College—when he intends to imagine King’s through use of a certain image, whatever its causal source—his imagining is, intuitively, a complete success. More generally: whenever there is an intuitive pull to say that someone hasn’t imagined what they intended to imagine, we can ask whether this is better understood as a case where they haven’t imagined as they intended (due, often, to a false belief about their perceptual histories)—even if they did, in fact, imagine the object they intended.

Doing what we intended, but not as we intended to do it, is common in everyday life. Compare: Sally intends to sign a contract for her new home with the pen her father gave her as a graduation present. When she asks her husband to hand her the pen, he accidentally pulls a different, similar looking pen from her handbag and gives it to her. She doesn’t notice the error and uses it to sign the contract. Did she sign what she intended to sign? Yes, the contract is fully executed. Did she sign as she intended to sign? No. She didn’t use the pen she intended to use for the signing. The signing was not a complete success, even if she signed what she intended to sign.

Likewise, I propose, in the example where Sam putatively (unwittingly) imagines Trinity, he really does imagine King’s College on fire, just not as he intended (viz., not with an image deriving from King’s). The fact that he didn’t imagine it in the manner he intended explains the intuitive sense that something has gone wrong—that the imagining was not a complete success. Yet, this failure is compatible with his nevertheless having imagined King’s College and with the thesis that one’s intentions always suffice to fix the object of one’s imaginings.

Note that, in both the contract case and the Trinity/Kings case, there is an error that occurs prior to the respective acts of signing the contract and imagining Kings College that leads to corresponding false beliefs—false beliefs that infect subsequent intentions. Thanks to her husband’s mistake, Sally misidentifies the pen she is using as the one her father gave her. And, thanks to incorrect information Sam is given about his perceptual past (namely, that he was looking at King’s), he misidentifies the causal source of his mental image as being King’s college. These errors prevent the complete fulfillment of their subsequent intentions, even if Sally still signs the contract she intended, and even if Sam still imagines the college he intended.

Of course, the mere fact that this comparison can be proposed does not show that it is the correct comparison to make. One might grant the distinction between doing what one intends and doing it in the way one intends, while holding that the proper application to Sam’s case is to say that he does what he intends insofar as he imagines something or other, and that he fails to do it in the way he intends insofar as he imagines an object other than the one he intends to imagine.¹

A few responses are in order. First, the point of the pen and contract comparison is simply to clarify the sort of partial success that I hold occurs in Sam’s case, by showing that it is a kind of partial success familiar to other contexts—one rooted in false background beliefs that infect the intentions at work. The value of the analogy is its ability to make sense of the overall situation—of how it can be that intentions always determine the objects of imaginings, even if we have a sense that some imaginings are failures. Second, favoring the second comparison (where Sam only succeeds in imagining something or other, but not the specific object he intends) simply returns us to the questions of Sections 2. We then don’t know why the removal of his incorrect beliefs about the causal history of his image seems to change what he imagines (in the example from Section 2), as this doesn’t change the causal history of the image itself. By contrast, when we interpret the situation by analogy to the pen and contract example, we know that no change in fact occurs: he always imagines the object he intends (this being determined by his intentions). However, he does not do so in the way he intends, thanks to false background beliefs about the causal history of the image. This, in turn, is what generates the ultimately mistaken intuition that he did not imagine the college he intended. Third, an independent argument against anti-intentionalism is still to be levied in Section 4. If successful, it lends further credence to the analysis I’ve just proposed for vindicating intentionalism.

A second worry should be addressed. On the analysis I’ve offered, when it seems that Sam unwittingly imagines Trinity, this is because Sam intends not simply to imagine King’s College, but to imagine King’s College through use of an image deriving from King’s. Some may find it implausible to attribute Sam such a complex intention. How can we settle the issue? A good way to determine what Sam—or any person—really intends is simply to ask them. Suppose again that Sam is shown Trinity College and is told that it’s King’s. When we later ask to Sam to imagine King’s College and—unwittingly—he does so through use of an image deriving from Trinity, we can ask him: did you intend to make use of an image deriving from a past sighting of King’s? It is plausible that he’ll answer “yes.” His imagining of King’s is occurring in the context of background beliefs that he in fact saw King’s College and that the image he is using derives from that sighting. His having these beliefs rationally entails that he intends to use an image deriving from King’s when he uses the image in question, just as anyone who believes that S is y and intends to use S to φ intends to use an S that is y to φ.² (Compare: if I believe that my cup has a hole in it and intend to use my cup to drink coffee, then I intend to use a cup with a hole in it to drink coffee.)

In countless similar instances, we are guided by and hold ourselves accountable to intentions that do not announce themselves as full sentences running through our heads, but which we quickly own up to under questioning. By contrast, it is a bit of trivia, unknown to Sam, that King’s College was founded in 1441. If we ask Sam (after his imagining) whether he’d intended to make use of an image deriving from a college founded in 1441, the answer will be no. In sum: fine-grained determinations of a person’s intentions can be made after the fact through eliciting testimony, and there’s reason to think things will break in favor of my account when those reports come in.

4 What it is to intend to imagine: A further argument against anti-intentionalism

The positive view I have outlined is that an imaginer’s intentions always suffice to fix the object of their imagining. We can explain away the kinds of problem-cases M&S have described by holding that, in those cases, the imaginer does not merely intend to imagine x; they intend, in addition, to imagine x through use of an image deriving from their having seen x. It is the failure of this second part of the intention that explains the sense that something has gone wrong in the imagining, even if the imaginer has succeeded in imagining the object they intended to imagine. We have independent reason to think that imaginers have these more complex intentions (i.e. to imagine through use of an image deriving form their having seen x) because they would own up to as much under questioning. These problematic intentions trace to their false beliefs about their perceptual histories.

This view has the advantage of consistency: in every case where an imaginer intends to imagine some x—be it when enjoying a fiction, reasoning counterfactually, or in idle fantasy—their intention to imagine x is sufficient to fix the object of the imagining. Sometimes, (and especially when imagining in fantasy) there is no attached intention to use an image deriving from a certain past perception. But, often, we want our imaginings to be tethered to past experience in a certain way, and there is an attached intention reflecting as much. When such intentions are not satisfied, we have not imagined as we intended, even if we have imagined what we intended.

I am now in a place to articulate a further difficulty for M&S’s challenge to intentionalism, by focusing on the questions of what it is to imagistically imagine and, by extension, what it is to intend to imagistically imagine. That is, what exactly do people who intend to imagine intend to do? Can any more informative specification of this intention be given? The approach to the puzzle cases I have articulated fits well with a natural account of what it is to imagistically (or sensorily) imagine. Imagistic imagining, we can hold, is any occurrent use of mental imagery in cognition. (See Langland-Hassan (2020) and Van Leeuwen (2013) for similar definitions.) M&S work with a related conception of sensory imagination, glossing it as “imaginings that involve the production of mental imagery” (Munro & Strohminger, 2021, p. 11848). To imagine, in the imagistic or sensory sense, is to represent some x through the use of mental imagery. To intend to imagistically imagine, then, is to intend to represent some x through the use of mental imagery. All sides seem to agree on this. Note that this doesn’t entail that imagistic imagining requires one to have such intentions. We might imagistically imagine unintentionally. It just entails that, to the extent that someone intends to imagistically imagine, they intend to represent some X through the use of mental imagery. Importantly, this intention places no condition on the mental imagery used that it causally derives from x or even that it resembles x in any salient respects—hence the possibility of imagining things looking quite different than they do. It is only when this basic intention to imagine x is enriched by a further intention to make use of an image that fulfills various other conditions that one seriously risks a less than fully successful imagining.

Now suppose that we accept M&S’s claim that when we imagistically imagine we, “by default,” imagine the object (if any) to which the mental image causally traces back in experience. This suggests instead a view where to imagistically imagine is (by default) to represent x through the use of mental imagery that causally traces back to one’s past experience of x. In that case, to intend to imagine x is to intend to represent x through the use of mental imagery that causally traces back to one’s past experience of x. Two problems arise from this. First, if this is a correct account of the “default” intention when we intend to imagine, then we know exactly why Sam’s imagining seems not to be a success: the intention to use an image with a certain kind of causal history is not satisfied. Yet, as we have seen, this sort of failure to imagine as one intends does not entail that x wasn’t, in fact, imagined.

Second, attempts to waive such a default intention through more complex intentions will leave one wrapped up in self-contradiction. It will be self-contradictory to intend to imagistically imagine x through the use of mental imagery that does not derive from x because, properly spelled out, to do this is to intend to represent x through the use of mental imagery causally deriving from x through the use of mental imagery that does not causally derive from x. Given that M&S think it is possible to imagine things through the use of imagery that does not causally derive from past experiences of those things—and that this is possible through the use of causal-condition-waiving intentions—they cannot hold that the intention to imagine always implicates the fulfillment of a causal condition. It thus seems they will want to adopt the more minimal account of the default intention, where intending to imagistically imagine x is simply a matter of intending to use imagery to represent (or think about) x.

However, there are problems for them in this direction as well. If M&S agree with the more minimal account of what it is to intend to imagine x—that it is nothing more than to intend to use mental imagery to represent x—then they lack any grounds for saying that an imaginer has not done what they intended to do (namely, to imagine x) in cases where the imagery used does not causally derive from x. For such imaginers did imagine intentionally, and they did not intend to make use of imagery causally deriving from x. If Sam’s intention when intending to imagine King’s College on fire is nothing other than an intention to use imagery to represent King’s college on fire, on what grounds can M&S say that his intentions weren’t fulfilled? After all, they agree that he can imagine King’s; and they agree that the specific mental image he has used can be used to do so. If they reverse course at this point and hold that Sam intends to make use of an image deriving from Kings, this fits naturally with the account I have suggested—namely, that the failure lies not in what he imagines, but in the manner in which he imagines it.

The only option for M&S at this point is to reject both proposals for what it is to imagistically imagine. This leaves open an essential question: what, exactly, do people who intend to imagine x intend to do? Without an answer, we’re in no place to ascribe Sam an intention and, thus, in no place to say any intention was left unfulfilled.

5 On the objects of unintentional imaginings

The account I have provided specifies what it is to imagistically imagine and what it is to intend to imagine in a way that preserves the capacity of our intentions to (always) determine the objects of our imaginings. The account offers a diagnosis of why some cases generate a contrary intuition—and even vindicates that intuition, to a degree—by distinguishing outright failures to imagine x from failures to imagine x in the manner one intends.

There is, however, a point that one might raise in reply. Arguably, not all of our imagistic imaginings are intentional. For instance, we may experience disturbing or distracting imagery that resists our attempts to banish it. Or we may just find ourselves imagining things without any clear purpose attached—where, if asked, “what did you intend to imagine,” we should simply say: “nothing.” Plausibly, unintentional imaginings of this sort have objects. If imaginings can get objects without getting them from our intentions, one might view this means of acquiring an object—whatever its nature—as the true “default.” And it may seem that a natural proposal for what it is that determines the object in such cases is the past perceptual experience from which the relevant mental image is derived. If that were the case, M&S could promote their view on the grounds that their account of how imaginings get their objects extends to non-intentional imaginings (and at least some intentional imaginings), while mine only applies to intentional imaginings. They may emphasize that the need for an account of how unintentional imaginings get their objects is especially keen in explaining the imaginings of animals or young children, who may not have sophisticated mentalistic concepts—such as ‘mental image’—of the sort implicated in the intentions-to-imagine earlier described.

In response, supposing that there are indeed unintentional imaginings, there is no reason to think that their objects are limited to things we have perceived. We may have unbidden imaginings of fictional characters and fantastical scenes just as well as of people and places we once saw. M&S’s account of the “default” objects of imagining will not apply to these. Thus, it is no better placed than mine to explain how the full range of unintentional imagistic imaginings get their objects. If there is an entirely general account available, none of us has offered it.

It is also worth recalling the points earlier raised concerning the imagining of types of objects (e.g. pencils and paper) and settings (e.g. a doctor’s office) that we have perceived many times and where it is doubtful that there is a single causal source for whichever image gets used. Many unintentional imaginings—equally with intentional imaginings—are likely generated through the use of imagery that has no single source in past experience, but which is itself an amalgam of similar past experiences—one shaped and refined through multiple past experiences of numerically distinct objects and events. Again, in these cases, the intuition that what goes for non-propositional seeing goes for imagistic imagining founders. Whenever we see a pencil, there is a particular pencil that we see that is the causal source of our experience. By contrast, when we imagine a pencil, there may be no particular pencil we are imagining—and, certainly, no particular past pencil we saw that is the causal source. This highlights the need for an account of how imaginings—intentional or not—get their objects that does not rely on the likely false assumption that most imaginings have a single object as their causal source. (See Langland-Hassan (2023) for an approach.)

6 Intentionalism and remembering – support for continuism

I end with some remarks on the relationship between imagistic imagining and episodic remembering, in light of the foregoing. People familiar with the philosophical literature on remembering will see affinities between M&S’s view of imagistic imagination and Martin and Deutscher’s causal theory of remembering. One of the thought experiments Martin and Deutscher offer in defense of that theory describes a painter who sets out to paint a merely imaginary farm scene, yet where the resulting painting bears an unmistakable resemblance to a farm he once visited as a child. Their (controversial) interpretation is that the painter is remembering the farm without believing that he is doing so—and, indeed, without intending to remember it. For Martin and Deutscher, this serves to establish that one’s beliefs and intentions place no constraints on the question of what one is remembering. What matters, instead, is the causal origin of the “memory trace” relied upon in the act of remembering (Martin & Deutscher, 1966).

Presumably, both Martin & Deutscher and M&S would hold in this case that the painter intends to imagine a merely possible, imaginary farm, but fails. Instead, he remembers (and also imagistically imagines) an actual farm he once visited.³ However, Martin & Deutscher never make any explicit claims about the objects of imaginings (as opposed to rememberings). Subsequent discussions of remembering focused instead on whether we can be wrong about what we are remembering, and whether remembering an event truly requires encoding, storing, and retrieving a certain kind of representation of the event.

Recently, the traditional causal theory has come under pressure from a group of theorists who emphasize similarities between episodic remembering and imagining the future. These similarities includes the neural overlap in areas governing imagining and remembering (Addis et al., 2007), the co-occurrence of deficits in imagining and remembering (Schacter et al., 2007), the likely common reconstructive principles and mechanisms underlying imagining and remembering (De Brigard, 2014; Michaelian, 2016), and what some view as the empirical implausibility of the requirements laid down by the Martin-and-Deutcher-style causal theory for genuine remembering (Langland-Hassan, 2022a; Sutton, 1998; Werning, 2020). Many of the views challenging the traditional causal theory have made it their motto that remembering just is (a kind of) imagining (Addis, 2020; Michaelian, 2016). What, exactly, it means to say that remembering is a kind of imagining—and how, precisely, this claim conflicts with the traditional casual theory—has become a subsequent focus of discussion (Langland-Hassan, 2022b, 2021; Michaelian et al., 2020). The most common proposal has been that remembering and imagining—and, in particular, imagining aimed at predicting the future—are both constructive processes that draw on multiple sources of information (and not a single, static “memory trace”) in the generation of a representation that aims at accurately depicting some state of affairs, past or present (Addis, 2020; De Brigard, 2014; Michaelian et al., 2020).

This paper’s discussion of imagining and its relation to one’s intentions suggests a further important similarity that imaging may bear to remembering: it may be that, for both remembering and imagining, one’s intentions determine the object of the mental representation. This is a further way for the simulationist to highlight the similarities of remembering to imagining, while also distinguishing their view from the causal theory. We have already reviewed reasons to accept such a view for imagistic imagining. However, in the case of remembering, we seem to face a barrier: if our intentions suffice to determine what we are remembering, then we should be able to remember events that we never witnessed and, indeed, that never even occurred. All that should be required is that we intend to remember the event. For the many who hear the verb ‘remember’ as factive in nature—such that an event can only be remembered if it in fact occurred—this may seem an unacceptable entailment and reason enough to reject the proffered similarity between remembering and imagining.

However, there is a middle way available that grants the similarity of imagining to remembering while respecting the intuition that one cannot remember an event that didn’t occur. We simply need to distinguish successful remembering from unsuccessful remembering. To unsuccessfully remember an event, it is sufficient that one mentally represents it as an event from one’s personal past; while, to successfully remember an event, one must do the same while additional conditions are met. Those additional conditions can vary according to taste. Defenders of a traditional causal theory may include: 1) that the event actually occurred, 2) that the event is represented with a suitable degree of accuracy, 3) that one experienced the event, and 4) that one’s current representation of the event is appropriately causally related to one’s past experience of the event. Successful remembering will then be factive, even if unsuccessful remembering need not be. Simulationists—who reject the appropriate causation condition (Michaelian, 2016), the factivity condition (De Brigard, 2014), and, sometimes, even the past experience condition (Michaelian, 2022)—may instead require of successful remembering that the representation is generated by a reliably functioning neuro-cognitive system of the right sort, and that the event is represented with a suitable degree of accuracy. Either way, if we allow for a distinction between successful remembering (which, for some, will be factive) and unsuccessful remembering, we can hold that our intentions always determine which object or event we are episodically remembering—just as they do for imagistic imagining in general—but not whether we remember successfully.

However, one may still object that this view entails that one can at least unsuccessfully remember whatever past event one wishes, insofar as one can act on an intention to unsuccessfully remember whichever event one wishes.⁴ And, intuitively, it may not seem that that we can even unsuccessfully remember whichever event we wish. For instance, it may seem that I cannot unsuccessfully remember events that I know did not happen, or that I know I did not witness—even if I can imagine them. My response is that the plausibility of this challenge relies on a false equation of unsuccessful remembering with what we tend to call “seeming to remember.” Let me explain.

Suppose that I decide to imagistically represent the bombing of Pearl Harbor as part of an imaginative project where I myself am the person witnessing it. It fits with my proposal to call this “intentionally unsuccessfully remembering the bombing of Pearl Harbor” (as I know the bombing happened long before I was born). However, others may wish to call this a case of (merely) intentionally imagining a past event and not “intentionally unsuccessfully remembering a past event.” Why insist on that distinction? A key reason would be to distinguish intentionally unsuccessfully remembering past events (what some others will want to call “merely imagining past events”) from unknowingly unsuccessfully remembering past events—where, in the latter sort of case, a person wrongly believes himself to be successfully remembering the event. This may seem like an important difference to mark because it is only the person who unknowingly unsuccessfully remembers a past event who seems (to himself) to remember the event. The person who intentionally unsuccessfully remembers an event does not (as we would colloquially put it) seem to remember it.

Yet we can capture this difference in “seeming to remember” by holding that the ‘seems’ in question is doxastic in nature, grounded in the person’s beliefs that they are (or are not) successfully remembering.⁵ It is because we cannot willfully change our beliefs about whether we are successfully remembering that we cannot “seem to remember” at will. If this difference in beliefs about success is the only difference between seeming to remember an event (while unsuccessfully remembering it) and merely imagining a past event (without seeming to remember it), then there is no trouble in holding that, in both cases, the person unsuccessfully remembers the event they intend to remember.⁶ For the beliefs that make the difference between one’s seeming and not seeming to remember are not proper parts of the mental state of remembering itself. In short, all personal past-directed imaginings will remain cases of remembering (whose objects are determined by one’s intentions) even if not all such rememberings will be cases we would describe as a person seeming (to herself) to remember the event.

The approach I have sketched may nevertheless remain unsatisfying to those sympathetic to the causal theory. It may seem to them that remembering, by its nature, must get its objects from past perceptual contact with the very event remembered and that “unsuccessful remembering” is no kind of remembering at all. The view has its merits. Yet it places a very stringent requirement on our neurobiology. Namely, it requires that, for each event a person can remember, there is a discrete memory trace (or “engram”) that was the result of neural encoding at the time of the event and that has been preserved until the moment it is retrieved and relied upon in an act of remembering. Any such memory trace must be “monogamous” (Langland-Hassan, 2022a), in the sense that it causally derives from exactly one past perceptual experience, which the memory trace in turn allows its possessor to remember. For if the trace is derived from multiple past experiences—if it is “promiscuous”—its use in an act of remembering cannot be what determines which event the person is remembering.

A number of theorists have argued on empirical and conceptual grounds that it is implausible that the brain stores monogamous, single-event-related memory traces of the kind required by the causal theory (De Brigard, 2014; Michaelian, 2021; Sutton, 1998). In other work, I have suggested that memory traces are less like single-event recordings and more like reusable “props” keyed to various settings and objects, where the props casually derive from multiple past perceptual encounters with those settings and objects (Langland-Hassan, 2022a). This view of traces meshes well with accounts of remembering that see it as a reconstructive, creative process, as opposed to the sort of “replay” of event-specific stored footage that the traditional causal theory seems to presume (Addis, 2020; Cochran et al., 2015; Hemmer & Steyvers, 2009; Schacter et al., 2007; Schacter & Addis, 2007).

The nature of memory traces and their relationship to theories of remembering remain active areas of research, as does the general question of how episodic rememberings gain reference to the events they concern. The point I wish to emphasize is that, if the reconstructive approach to remembering is on the right track—and if successful remembering does not require monogamous memory traces—then something other than a memory trace must determine which event it is that a person is remembering. This is good work to assign to one’s intentions and constitutes another path toward seeing remembering as a kind of imagining.

References

---

1. I thank a reviewer for encouraging me to consider this response.↩︎

2. There are exceptions to this principle when someone has temporarily forgotten that S is y and intends to use S to φ. Yet that sort of exception does not apply to the cases in question, which rely on the imaginer having present to mind a false belief about their perceptual history (as argued in Section 2). When we remove that false belief—or any memory of it—the intuition that they imagine something other than what they intend to imagine dissipates.↩︎

3. It is consistent to say that he both remembers the farm and imagistically imagines it, if we accept the definition of imagistic imagining offered above (where to imagistically imagine x is simply to use imagery to represent x). Some (e.g., Kind (2001); Arcangeli (2020)) offer richer definitions of imagistic imagining that would prevent an act of remembering from also being an act of imagining.↩︎

4. I thank a reviewer for encouraging me to discuss this tension in the account.↩︎

5. We find the doxastic, belief-related sense of ‘seems’ in sentences like: “It seems that John ate the last cookie.” This contrasts to the perceptual sense of ‘seems’ we find in a sentence like: “The top line seems longer than the bottom in this illusion, though I know they are the same length.”↩︎

6. Some may object that the ‘seems’ in question is not doxastic in nature. As support, they may point to the putative phenomenon of “non-believed memories” (Otgaar et al., 2014), where people report that they are not successfully remembering an event, yet nevertheless seem to themselves be remembering it. The possibility of such would suggest that there is a phenomenon of “seeming to remember” that is not merely doxastic in nature. I am skeptical that non-believed memories are a genuine phenomenon—for one thing, only a minority of participants report experiencing any—though I cannot address the issue adequately here.↩︎