The idea that imagination should be thought of as a skill has in recent years become increasingly common in philosophy of imagination (see, e.g., (Kind, 2020); (Blomkvist, 2022)). If we accept this idea, then two claims seem to follow. First, some people will be better at imagining than other people, and second, imagining is something at which we can get better. And this in turn yields an important question: What exactly are we evaluating when making a judgment on either of these scores? In other words, when we are judging that someone is a better imaginer than someone else, or judging that someone’s imagination has improved, what are the factors that underwrite the difference in imaginative skill?

There are undoubtedly many dimensions along which imaginings differ from one another and along which one imagining might be said to be better than another. When it comes to escapist fantasies, for example, as when we imagine tropical vacations while stuck in boring meetings at work, one imagining might be judged better than another in virtue of being more diverting. When it comes to revenge fantasies, as when we imagine how we might get back at someone who has wronged us, one imagining might be judged better than another in virtue of being more creative. More generally, and this list is by no means meant to be comprehensive, imaginings might be judged better or worse along dimensions of level of detail, evocativeness, comprehensiveness, and accuracy. It’s this last factor that will occupy our attention in this paper. What does it mean for one imagining to be more accurate than another?

Judgments about the accuracy of imagining would seem to be important not just in determining how imaginings fare with respect to skillfulness but also in determining how they fare with respect to epistemic value. That accuracy would be important to the epistemology of imagination might seem especially plausible in light of the role that accuracy plays in epistemology in general. Indeed, in recent years, the topic of accuracy has been of increasing interest in epistemology, particularly among those in formal epistemology who take what’s known as an accuracy-first approach (see, e.g., (Pettigrew, 2016)). That said, my exploration of accuracy with respect to imagination should not be taken to imply any commitment to an accuracy-first epistemology of imagination, whatever such an epistemology would be. In exploring what it means for imaginings to be accurate, and for one imagining to be more accurate than another, I am not taking any stance on whether accuracy is the highest epistemic good. We can accept that accuracy has epistemic value without claiming that it is the principal epistemic value.

Philosophers have not previously paid much attention to imaginative accuracy, perhaps because of a sense that imaginings cannot be meaningfully assessed for accuracy, i.e., that the notion of accuracy is inapplicable in the context of imagination. I thus begin in Part 1 by addressing two worries that might naturally arise about the very coherence of the investigation I’m here undertaking – what I call the peculiarity worry and the mismatch worry. With those worries addressed, I turn in Part 2 to an exploration of what is meant by imaginative accuracy. My discussion relies on two key points: first, that accuracy is best understood in terms of aim; and second, that imaginings aim at the representation of fictional states of affairs. I call this line of thought the fictionality approach. In Part 3, I look more closely at six different types of imaginings in an effort to develop and clarify the fictionality approach. Finally, in Part 4, I turn to what I call the calibration objection. Given the nature of imagination, there seems to be no way to calibrate one’s judgments of imaginative accuracy, which may seem to render moot much of the discussion of this paper. After showing how much of the force of the calibration objection can be defused, in the paper’s final section I offer some brief concluding remarks.

1 Can imaginings be accurate?

In order to address the question of whether considerations of accuracy are genuinely applicable in the context of imagination, it will be helpful first to make clear what’s meant by imagination. One particularly useful characterization has been offered by Shen-yi Liao and Tamar Gendler in the Stanford Encyclopedia of Philosophy:

To imagine is to represent without aiming at things as they actually, presently, and subjectively are. One can use imagination to represent possibilities other than the actual, to represent times other than the present, and to represent perspectives other than one’s own. Unlike perceiving and believing, imagining something does not require one to consider that something to be the case. (Liao & Gendler, 2020)

As philosophers working on imagination have often noted, these imaginative representations come in many different forms. Consider three different imaginative activities: imagining that Taylor Swift is performing a surprise concert in the philosophy department lounge, imagining the concert itself, and imagining one’s exhilaration at the performance. In line with a taxonomy I’ve previously developed (Kind, 2016b, 2022), the first of these imaginings is categorized as an instance of propositional imagination, the second as an instance of imagistic imagination, and the third as an instance of experiential imagination.¹ In propositional imagination, the content of one’s imagining is a proposition – in this case, the proposition that Taylor Swift is performing a surprise concert in the philosophy department lounge. Propositional imagination is thus best understood on the model of other propositional attitudes like believing, hoping, or fearing. In contrast, imagistic imagination is best understood on the model of perceptual states like seeing and hearing. When imagining the concert itself, one might naturally do so by way of visual or auditory mental imagery – or perhaps a combination of the two. Other instances of imagistic imagining might proceed by way of different perceptual modalities, perhaps using gustatory mental imagery or olfactory mental imagery. Finally, experiential imagination is best understood on the model of experience. Just as one might experience exhilaration, one might imagine one’s experience of exhilaration while attending the performance.

Though we can distinguish these three types of imaginings, I find it useful to group experiential imagining together with imagistic imagination (Kind, 2022, pp. 9–10). This grouping can be motivated partly on the grounds that neither of these forms of imagining takes a propositional form, and partly on the grounds that we can stretch the notion of imagery not just beyond the visual case (as we do when we talk of auditory imagery) but even further to include pain imagery, emotional imagery, and so on. By treating imagistic imagination and experiential imagination together under a heading of sensory imagination, we thereby collapse the three-fold distinction into a two-fold distinction. My exploration of imaginative accuracy concerns imagining in both its propositional and sensory forms.

1.1 The peculiarity worry

On first thought, however, the examples I have used to differentiate different types of imaginings might seem to give rise to a sense that we shouldn’t be talking about accuracy in connection with imagination at all. After all, much to the disappointment of any Swifties among department members, Taylor Swift isn’t actually performing in the philosophy department lounge, nor has she done so in the past. So it might seem peculiar to be talking about accurately imagining that state of affairs. Likewise, it might also seem peculiar to talk of accurately imagining the concert itself or accurately imagining one’s feelings of exhilaration during the concert. There is no actual concert or feelings of exhilaration against which these imaginings can be compared.

Let’s call this concern the peculiarity worry. To my mind, the worry stems in large part from the fact that talk of accuracy in imagination is importantly disanalogous from talk of accuracy in connection with other mental states. As we’ve just noted, propositional imagining is best understood on the model of belief. The accuracy of a belief depends on its truthfulness. But the accuracy of a propositional imagining does not seem to depend on its truthfulness. The fact that Taylor Swift isn’t performing a surprise concert in the department lounge, and thus that the proposition the Swiftie is imagining is false, does not settle the question of the imagining’s accuracy.² Matters are similar with respect to sensory imagining, which can be best understood on the model of sensory states. A sensory state is accurate if and only if it is veridical.³ But just as the accuracy of a propositional imagining does not seem to depend on its truthfulness, the accuracy of a sensory imagining does not seem to depend on its veridicality. The lack of a concert means that a Swiftie’s imagining of the concert taking place is non-veridical, but that does not settle the question of the imagining’s accuracy. So, even though we can think of propositional imagining on the model of belief and sensory imagining on the model of sensory states, we cannot think of accuracy of imagining on either the model of truth/falsity or of veridicality/non-veridicality.

To defuse the peculiarity worry, then, we need to broaden our notion of accuracy. As a first step, let’s note that we talk of accuracy in all kinds of contexts where truth or veridicality seems irrelevant: someone who knits a sweater may have done so accurately or inaccurately, and likewise for someone baking a cake. What exactly accurate knitting or accurate baking amounts to is a question we’ll return to below; for now, let’s simply observe that it might naturally be understood in terms of fidelity to a given pattern or recipe. But knitting and baking are not representational activities. More important for defusing the peculiarity worry is the fact that we can broaden the notion of accuracy even within the representational context. The reason that truth matters for the assessment of accuracy with respect to belief, and that veridicality matters for the assessment of accuracy with respect to perception, is that they set the terms of representational success. It seems plausible, however, that there might be other ways of achieving representational success. Consider maps, for example. The representational success of a map does not seem aptly captured in terms of truth and falsity nor in terms of veridicality and non-veridicality (Silins, 2021), and we might think something similar with respect to photographs and portraits. Insofar as we are accustomed to thinking of representational success in terms of truth and veridicality, the fact that the representational success of an imagining does not consist in these terms makes talk of imaginative accuracy seem peculiar. So, what does the representational success of an imagining consist in?⁴ We will return to this question in Part II, below.

1.2 The mismatch worry

Suppose that we accept that there’s nothing peculiar in principle about talk of imaginative accuracy. Still, one might think that there is still something peculiar about this talk in practice. In many contexts of imaginings, particularly in fantastical ones, it seems that there’s no particular value in being accurate. When a parent is making up tall tales to amuse their children, for example, accuracy is entirely beside the point. Indeed, in this enterprise, inaccuracy might very well be more valuable than accuracy, since certain kinds of fantastical inaccuracies will likely be better at amusing the children than stories that are too close to the truth.⁵ Besides these contexts in which accuracy lacks value or relevance, there are other contexts in which accuracy might be entirely inapplicable. It’s hard to know how we would judge accuracy when someone imagines getting revenge on someone who has wronged them, or when children imagine themselves to be scary monsters, or when an author who is starting work on a novel imagines what their protagonist looks like. This general set of concerns gives us a second worry about the notion of accuracy when it comes to imagination, what I will call the mismatch worry.

Underlying the mismatch worry seems to be an overly narrow conception of when and why we engage in imaginative activities. In particular, the proponent of the mismatch worry seems to be focusing solely on what I call the transcendent use of imagining (Kind & Kung, 2016). When we put imagining to transcendent use, we are in some way trying to transcend or escape the world in which we live. This typically happens in activities like daydreaming, fantasizing, and playing games of pretend. For now, let’s suppose that the proponent of the mismatch worry is correct that accuracy is irrelevant or inapplicable in transcendent uses of imagining. Even so, there is a second important use of imagination, what I and Kung call the instructive use. When we put imagining to instructive use, we are in some way trying to learn from the world in which we live. This typically happens in activities like decision-making, planning, thought experimentation, and mindreading. In these activities, accuracy is not only relevant and applicable but essential.

Consider mindreading, an imaginative activity in which we are attempting to understand the mental states of another, often in an attempt to explain and predict their behavior.⁶ When we engage in this imaginative activity, we are trying to get things right, i.e., we want our imaginings to be accurate. Relatedly, when we engage in empathetic imagination, when we try to put ourself in the same kind of experiential state that we take someone else to be in, we also want our imaginings to be accurate.⁷

Accuracy also seems to matter in other instructive uses of imagination such as decision-making and planning. When we’re out shopping for a birthday present for a friend, and we’re trying to decide between two hats by imagining which would better suit them, it’s important for the task at hand that our imaginings of the friend’s face shape be accurate. Likewise, when we’re imaginatively exploring a change of career or a move to a new city, or when we’re imagining how the stew we’re making for dinner would taste if we added more cumin, we want our imagining to represent the given scenario accurately.

Finally, consider the fact that imagination is put to instructive use while doing philosophy, and here too accuracy matters. In the 18^th century, in a text that analyzes the key traits underlying genius, the Scottish minister William Duff emphasizes the importance of imaginative accuracy for success at philosophy:

Imagination receives a very different modification of form in the mind of a Philosopher, from what it takes in that of a Poet. In the one it extends to all the possible relations of things; in the other it admits only those that are probable, in order to determine such as are real. Hence it should seem, that in the first instance it ought to possess greater compass, and in the last, greater accuracy. Here then we have one characteristical indication of a Genius for philosophical Science; and that is, accuracy of imagination. (Duff, 1767)

As this passage from Duff suggests, we shouldn’t expect it to be the case that all of the various evaluative factors mentioned come into play in every context of imagining. Though creativity and pleasure have considerable relevance for poetic imaginings, they have less relevance for imaginings about whether a piece of furniture we’re contemplating buying will fit in the trunk of our car. The fact that accuracy isn’t always relevant or applicable to our judgments of imaginative success does not mean that we can’t talk of accuracy in imagination.

To my mind, this is enough to show that the mismatch worry can be dismissed. If talk of accuracy is relevant and appropriate when it comes to instructive uses of imagination, then even if (as I have been supposing) it is not relevant and appropriate when it comes to transcendent uses of imagination, we will have undercut the suggestion that there is a mismatch between imagination and accuracy. Interestingly, however, it turns out that reflection on accuracy in the context of instructive uses of imagination will help us to see how accuracy can also be important in the context of transcendent uses of imagination. I will return to this point below, where I show how we can appropriately make judgments of accuracy in several different types of transcendent imaginings. In the meantime, with the peculiarity and mismatch worries laid to rest, it’s now time to take up more directly the main issue: What does it mean for an imagining to be accurate? Exploring this issue will be the task of Part 2.

2 What is imaginative accuracy?

To start, recall that while addressing the peculiarity worry, we noted that we saw that we can talk about accuracy in both representational and non-representational contexts. Let’s return to our earlier examples where accuracy is assessed in non-representational contexts, as when one is knitting or baking. As we noted, accuracy in these activities might naturally be understood in terms of fidelity to a given pattern or recipe, respectively. This appears to suggest that accuracy is determined against a fixed standard, a suggestion further supported by other (non-representational) activities in which accuracy is assessed. Consider darts, for example. The accuracy of dart throwing seems to be measured against the fixed standard of the center bullseye. A shot is accurate when it hits the bullseye; otherwise, it’s inaccurate.

But this suggestion is too quick. In many games of darts, such as 501, there are times when hitting the bullseye will result in one’s losing the game.⁸ A shot may well hit the bullseye yet be inaccurate. Why? Because the bullseye wasn’t the target at which the player was aiming. An analogous point can be made in relation to other games and sports. In soccer, when a player kicks the ball, whether they’ve made an accurate pass depends on which teammate they were aiming to pass the ball to. A pass that initially looks accurate – because it ends up very close to a teammate’s feet – might in fact not be accurate, if the player was aiming to pass to a different teammate. And of course, an analogous point also applies in the knitting case. To determine whether a knitting project is done accurately, we need to know which pattern the knitter was aiming to match. Yes, the accuracy of these performances is judged relative to a standard, but which standard is the relevant one will depend on the performer’s aim. Accuracy in these contexts is a matter of aim.

In fact, this point about the connection between aim and accuracy is often exploited for the furtherance of a plot, or for humor value, in fiction and film. One famous example comes from John Sturges’ 1960 Western, The Magnificent Seven. After the character Britt shoots a fleeing bandit off his horse, his fellow gunfighter Chico is in awe: “That was the greatest shot I’ve ever seen.” As Britt responds, “The worst! I was aiming at the horse.”⁹

In non-representational contexts, then, accuracy is not typically measured with respect to a fixed standard but rather with respect to the standard taken as one’s aim. Does this same point apply in representational contexts as well? At first it might seem that the answer is no. Consider belief. As we saw in Part 1, judgments about accuracy with respect in belief concern truth or falsity. Truth is a fixed standard. How can aim be relevant? But this line of reasoning is mistaken. Consider the fact that we talk of beliefs as being truth-directed. Truth-directedness is naturally understood in terms of being directed at, or aiming at, truth. In short, the very reason that truth is relevant to judgments of the accuracy of beliefs is that truth is the aim of belief ((Williams, 1973, p. 136); see also (Velleman, 2000)). To focus on the fixed standard of truth with respect to the accuracy of belief is to focus on aim.

It seems reasonable to expect that matters would be likewise when it comes to imagination, i.e., that accuracy in imagination should likewise be measured with respect to aim.¹⁰ In the remainder of this section, I will explore how this suggestion can be developed. To do so, it will be helpful to distinguish two different levels where aim comes into play.¹¹ One level concerns the aim of a particular imagining qua imagining. An account of imaginative aim at this level would be analogous to the kind of account we give when we say that belief aims at the true. As such, an exploration of imaginative accuracy at this level gives us a fairly coarse-grained picture, but assessments of imaginative accuracy typically operate in a more fine-grained manner. This is where the second level comes in. At this level, we are concerned with what the aim of a particular imagining is qua that particular imagining.¹² An account of imaginative aim at this level would be analogous to the kind of account we give when we explicate what particular truth a given belief aims at.

2.1 Three first-level approaches

Several different options are available to us at the first level. The first stems from the long tradition in philosophy that connects imagination with possibility. In line with this tradition, it is sometimes said that imagination aims at the possible the way that belief aims at the true. That would tie imaginative accuracy to the representation of possible states of affairs. Call this the possibility approach.

A related connection is sometimes drawn between imagination and fiction. Kendall Walton has famously claimed that “imagining aims at the fictional as belief aims at the true” (Walton, 1990, p. 41). This way of casting things ties imaginative accuracy to the representation of fictional state of affairs. Call this the fictionality approach.

Beyond these two approaches, a third possible approach grows out of recreationist or simulationist theories of imagination (see., e.g., (Currie & Ravenscroft, 2002)). Someone who adopts this kind of theory takes imagination to be primarily simulative in nature. For example, a shopper who imagines their friend wearing the fancy blue hat they’re considering purchasing is best understood as engaging in a simulation of seeing their friend wearing that fancy blue hat, and a child who imagines that there is a scary monster under their bed is best understood as engaging in a simulation of believing that there is a scary monster under their bed. For the simulation theorist, then, it would be natural to construe the aim of imagination in terms of the simulation of other mental states; in the first example (a sensory imagining), the shopper is aiming to simulate a particular perceptual state, and in the second example (a propositional imagining), the child is aiming to simulate a particular cognitive state. Call this way of explicating the first-level aim of imagining the simulative approach.

In what follows, I’ll focus on the fictionality approach. Though all three approaches have explanatory promise, the fictionality approach has the further benefit of being importantly neutral on several key issues. In undertaking my exploration into imaginative accuracy, I’d like to refrain from making theoretical commitments about imagination where this can be avoided. Whether imagination is fundamentally simulative in nature is a substantive matter, and one about which there is considerable disagreement among contemporary philosophers of imagination.¹³ Other things equal, it seems prudent to avoid tying the notion of imaginative aim to a particular theory of imagination.¹⁴ Because both the fictionality and the possibility approaches remain neutral on the nature of imagination, they thus strike me as preferable to the simulative approach.

So why choose the fictionality approach over the possibility approach? Unlike the possibility approach, the fictionality approach allows us also to remain neutral on the question of whether we can imagine the impossible. Though most philosophers think that imagination is limited to the possible, some reject this claim. Moreover, even someone who thinks that we can’t succeed in imagining the impossible might still think that we can try to do so, i.e., that an imagining can aim at an impossible state of affairs. Insofar as fictions can depict impossibilities, and thus that fictional worlds can contain impossible states of affairs, these problems are avoided by taking the aim of imagining to be the fictional rather than the possible.

Note also how the fictionality approach matches up with Liao and Gendler’s definition of imagination that I quoted above: that what it is to imagine is “to represent without aiming at things as they actually, presently, and subjectively are.” A state of affairs that is different from how things actually, presently, and subjectively are for a given imaginer is a state of affairs that is fictional for that imaginer. As this suggests, for a state of affairs to count as fictional it is not required that it be a state of affairs described in an existing work of fiction; rather, it only needs to be a state of affairs describable by a work of fiction. As this also suggests, some fictional states of affairs will be actual states of affairs – e.g., in the fictional world of Harry Potter, King’s Cross Station is located in London, a fictional state of affairs that corresponds to the actual state of affairs.¹⁵

2.2 The fictionality approach

So let’s think look more closely about what the fictionality approach tells us about imaginative aims. With respect to the first level of imaginative aim, the answer has already been given, i.e., imagining aims at the fictional. But what happens when we extend this approach to the second level? What does it say about the aim of a particular imagining qua that particular imagining. That we need to address this further level should be clear. Suppose that I am aiming to imagine Harry Potter arriving at King’s Cross Station but I instead imagine Sherlock Holmes arriving at King’s Cross Station. Here I have indeed imagined a fictional state of affairs, but I haven’t imagined the one that I was aiming to imagine. My imagining was inaccurate. For a particular imagining to be accurate, it matters which fictional state of affairs the imaginer is aiming at with that particular imagining – just as in darts, for a particular throw to be accurate, it matters which part of the dartboard the player is aiming at with that particular throw.

Unfortunately, it often proves hard to specify which fictional state of affairs an imaginer is aiming at – or at least, it is often hard to do this in a substantive way that does not seem to be question-begging. Take the imagining just mentioned, the one where I am aiming to imagine Harry Potter arriving at King’s Cross Station. In specifying which fictional state of affairs I’m aiming at, it’s tempting simply to answer: the one where Harry Potter is arriving at King’s Cross Station. But this does not seem to be an edifying answer. More worryingly, however, on certain views of imaginative content, it seems to lead to an unintuitive conclusion, namely, that our imaginings must always be accurate. If we are intentionalists about imaginative content – if we believe that the content of our intention about what to imagine fixes the content of what we imagine – then we would always meet our imaginative aim by default.

At least since the middle of the twentieth century, intentionalism has seemed to be the dominant view about imaginative content. We can trace it at least as far back as some remarks by Ludwig Wittgenstein in the late 1940s. In talking about acts of imagistic imagining, Wittgenstein noted that “I am not imagining whoever is like my image; no, I am imagining whoever it is I mean to imagine” (Wittgenstein, 1980, sec. 22e). More recently, we see an especially explicit endorsement of intentionalism by Colin McGinn, who claims that “the identity of my imagined object is fixed by my imaginative intentions” (McGinn, 2004, p. 31). As he goes on to explain:

The identity of the object is epistemically prior to its appearance in the image: the imaginer starts with the object and then constructs an image of it. He does not have to figure out the identity of the object from the way his image represents it. The object is given, not inferred. I know that my image is of my mother because I intended it to be. (McGinn, 2004, p. 31)

Not all philosophers accept intentionalism. Consider Kripke’s discussion of imagination in Naming and Necessity (Kripke, 1980, pp. 113–114). According to Kripke, though I may intend to be imagining the wooden table in front of me as being made of ice, given that it’s impossible for that table to have been made of ice, I must not be imagining what I thought (that is, what I intended); I must not be imagining this very table. Kripke thinks that my imagining misses its intended target. More recently, Daniel Munro and Margot Strohminger (Munro & Strohminger, 2021) present a series of arguments against intentionalism and suggest that the causal history of mental imagery plays a key role in fixing the content of imagining.

Rejecting intentionalism solves the problem just discussed, that is, it avoids the unintuitive conclusion that our imaginings must always be accurate. But we might be able to avoid this problem without rejecting intentionalism or, at least, without wholly rejecting its spirit. Suppose we distinguish between imaginings that miss their targets and imaginings that mischaracterize their target (Kind, 2016c). Intentionalism does not allow for imaginings of the first sort, but there’s no reason that it can’t allow for imaginings of the second sort. Insofar as our imaginings can mischaracterize the targets that we are intending to imagine, then, our imaginings can still be inaccurate even if (as intentionalism suggests) we cannot miss them entirely.

Even given these various options for warding off the unintuitive conclusion, however, we still are faced with the issue of saying something substantive about how to specify which fictional state of affairs an imaginer is aiming at. I take up this question in Part 3.

3 At which fictional state of affairs does an imagining aim?

Our discussion thus far has foreshadowed the difficulty in specifying which fictional state of affairs an imagining is aiming at. To my mind, the task is made even more difficult when we try to tackle everything at once, i.e., when we try to produce an explanation straight off the bat that applies to all imaginings across the board. We can best make progress by proceeding in a stepwise fashion. In particular, I’ll proceed by starting with some of the more straightforward types of cases – cases where it is easier to get some traction on the relevant fictional state of affairs – and then I’ll work my way up to the less straightforward cases.

3.1 Type 1: Other places, other people

In some cases of imagining, we are aiming at things as they presently and actually are, but not how they subjectively are. Sometimes this is because we are imagining how things are somewhere else. Suppose I have just sent my son off to his first day of kindergarten, and I am now trying to imagine what he is doing right now. In this imagining it is clear what state of affairs I am aiming at, and there is a clear answer to the question of whether my imagining meets its aim and is thus accurate. If I’m imagining that my son is fingerpainting, but in fact he is spending the morning playing with trucks in the sandbox, then I haven’t imagined what I am aiming to imagine; my imagining is inaccurate.

Other times, we aim at things different from how they subjectively are in virtue of imagining how things are with someone else. Consider a case where imagination is employed in service of mindreading, i.e., trying to understand the mind of another person. When I am talking with my son at the end of his first day of kindergarten, and he sadly tells me he doesn’t ever want to go back, I might try to imagine what he’s feeling in an effort to figure out how best to respond. Here again it is clear what state of affairs I am aiming at, and there is again a clear answer to the question of whether my imagining meets its aim and is thus accurate. Of course, it might not be easy for me to determine the answer. But our interest in this discussion concerns the nature of imaginative accuracy – it concerns what accuracy of imagining consists in – and not how an imaginer can determine whether their imagining is accurate.

There are many other cases that will fall under Type 1 cases. Consider, for example, the kinds of imaginings that Munro (Munro, 2021) calls actuality-oriented imaginings. When asked how many windows are on the outside of my home, I may visually imagine the house, rotating the mental image as needed, to determine the answer. As Munro notes, in this type of case we are using imagination “to represent things as they are in the actual world”; in these kinds of cases, “one uses one’s imagination in a way that’s aimed at representing the actual” (Munro, 2021, p. 1).

3.2 Type 2: Other times

Now consider cases of imagining where we are aiming not at things as they actually and subjectively are at present but at things as they actually and subjectively were or will be. Having sent my son off to his first day of kindergarten, I might try to imagine dropping him off for his first day of college. And let’s suppose that this isn’t a mere flight of fancy; I am aiming at the actual future. Or suppose I try to imagine my mother’s first day of college – an event that previously took place but that I wasn’t present for. Or perhaps I try to imagine my own first day of kindergarten – an event that previously took place and that I was present for, but that I don’t really remember. In all these cases, past and future, there are clear answers about whether my imagining meets its aim and is thus accurate.

3.3 Type 3: Almost-actual scenarios

In the first two types of cases, it was relatively straightforward to specify the state of affairs at which an imaginer is aiming because that state of affairs is actual. But now consider cases where an imaginer, while not aiming at some actual state of affairs, aims at a state of affairs that is like the actual state of affairs in almost all respects. Take an example often discussed in the imagination literature: Someone is trying to determine whether a given piece of furniture – a desk, say – will fit through a doorway or whether they will have to disassemble it first. To make this determination, they need to imaginatively rotate the desk, so that it will have a different spatial position in imagining from the one it actually has, but in other relevant respects they need their imagining to match the actual state of affairs. The desk must be the same size as it actually is, the doorway the same height and width as it actually is, and so on. Here too it looks like there is a clear answer about what state of affairs they are aiming at and thus a clear answer about whether their imagining is accurate.

We might naturally think of this type of case as one of almost-actual imagining. Though I have specified imaginative aim in terms of fictional states of affairs rather than possible states of affairs, this notion of the near-actual might be helpfully understood in terms of a metaphor often used in discussions of possible worlds, namely, the metaphor of distance. It’s often said that some possible worlds are more distant from the actual world than others. Likewise, we might think that some fictional states of affairs are more distant from the current, actual state of affairs than others. We can use this to make our point about accuracy of imagination a little more precise: In some cases of imagining, the imaginer aims at more distant fictional states of affairs, while in some cases of imagining, the imaginer aims at very close fictional states of affairs – so close as to be almost-actual.

Many of the imaginings that are put to work in problem-solving or decision-making contexts – like the desk rotation case – will fall into this almost-actual category – and likewise for instructive uses of imagining more generally. When putting imagination to instructive use, the fictional state of affairs at which we’re aiming is usually fairly close to the current, actual state of affairs. It generally would have to be in order for imagination to be instructive, i.e., for it to teach us about the world in which we live. In contrast, in transcendent use, the fictional state of affairs at which we’re aiming is usually considerably further from the current, actual state of affairs.¹⁶

This won’t always be the case, of course. Suppose a group of scientists is exploring ways to combat climate change, and one of their proposals involves unleashing mass amounts of a chemical that would radically affect all organisms currently living on the planet. This might involve so many changes that the relevant fictional state of affairs is quite far from the current, actual state of affairs – much farther, say, than the fictional state of affairs contemplated in usual daydreams and fantasies. Nonetheless, we can take it as a rough, imperfect principle that transcendent imaginings involve more distance from current actuality than instructive imaginings do.

In philosophical discussion, the distinction between transcendent and instructive uses of imagining is often accompanied by a consideration of imaginative constraints (Badura & Kind, 2021; Kind, 2016a). Instructive imaginings are typically more constrained than transcendent imaginings. While instructive imagining involves the imposition of constraints; transcendent imagining involves the release of constraints. This fact helps to show why instructive imaginings are more likely to have epistemic relevance than transcendent imaginings. It is generally thought that imaginings must be appropriately constrained in order for them to have epistemic relevance. For example, in previous work (Kind, 2016a) I have distinguished two constraints – the reality constraint and the change constraint – which play a key role in ensuring that imaginings are epistemically relevant. To satisfy the reality constraint, the imagining needs to conform in relevant respects to the world as it is; as the imagining unfolds, one satisfies the change constraint when the changes made in imagination conform to the real-world consequences of that change. It’s perhaps obvious how this connects to imaginative accuracy. In the same way that constraining our imagination appropriately enables us to produce imaginings that are epistemically relevant, it also enables us to produce imaginings with a higher degree of accuracy. In fact, it’s precisely in virtue of doing the latter (achieving accuracy) that we are able to do the former (achieving epistemic relevance). In this way, we can see how accuracy has epistemic value with respect to imagination.

Let’s take stock. The three types of cases considered thus far help to show what imaginative accuracy amounts to in cases of instructive imagination. But what does accuracy amount to with respect to transcendent uses of imagination? Is accuracy ever relevant there, or are entirely different standards in place when judging transcendent uses of imagination? In fact, I think our discussion thus far has planted the seeds needed to see that accuracy can indeed be relevant to transcendent uses. Even in transcendent uses of imagination, there will sometimes be a fictional state of affairs at which we are aiming, and there will sometimes be constraints in place. To see this, it will be helpful to consider three more types of cases.

3.4 Type 4: Scripted scenarios

Consider cases where I’m engaging with an actual work of fiction, like a novel or a short story. In that case, it likely matters to my imaginative aims that I’m imagining the characters as the author has described them. Accuracy will here be measured not against the actual world but against the fictional world that the author has been created, and I want to constrain my imagining appropriately so as to match that fictional world. Though the reality constraint is not in play, there is an analogous constraint – perhaps one that can be thought of as a fictionality constraint – that needs to be met in order for my imagining to be accurate. For example, suppose I’m reading Lee Child’s The Killing Floor and I want to be engaging with the story and the characters as the novel depicts them. If, while reading, I imagine the main character Reacher to look like Tom Cruise, then my imagining will not be accurate. Though that’s what Reacher looks like in a different fictional state of affairs – the fictional state of affairs depicted by the 2012 film Jack Reacher – it’s not at all what Reacher looks like in the fictional state of affairs of the novel; there, he’s six foot five and around 250 pounds.

When engaging with novels and short stories and other works of fiction like movies and plays, the fictional state of affairs that my imagining aims at has been independently specified. But there are other cases of transcendent imaginings in which this happens as well. Suppose my sons are playing a game of pretend, and I decide to join in. They are imagining the sofa to be a pirate ship and the rug to be shark-infested ocean. In joining the game, I want to be imagining what they’re imagining; I am aiming at the same fictional state of affairs. If that fictional state of affairs involve three sharks, and I imagine only two, then my imagining is inaccurate.

3.5 Type 5: Unscripted scenarios, part 1

In many other cases of transcendent imagining, however, I won’t be starting with a pre-existing script or set of instructions. Yet these imaginings need not be aimless. If I’m fantasizing about winning the lottery and then quitting my job, it matters to my imaginative intentions that I’m imagining the windfall coming to me as I more or less am – at a certain job, of a certain age, with a certain family. It likely also matters that I’m a human being and not a mermaid, or a Martian, or a philosophical zombie. This all helps to fix which fictional state of affairs at which I am aiming and shows how my imagining needs to be constrained in order to be successful. This success consists at least partly in terms of accuracy. My aims, and the corresponding constraints, help to set a standard against which the accuracy of my imagining can be assessed.

Note, however, that these standards might be different from the ones we saw in the almost-actual cases above. Compare a Type 3 case with a Type 5 case:

In an effort to make a decision about what to do going forward, I’m imaginatively exploring a future in which I quit my job.

Just for fun, I’m fantasizing about winning the lottery and then quitting my job.

Given that these two imaginings have different purposes, they aim at two different fictional states of affairs. Though both of these fictional states of affairs involve my quitting my job, the first is constrained in a way that the second is not. In the first, various other facts about my current situation are held fixed in a way they are not in the second imagining. As a result, and in line with the aim-based approach to imaginative accuracy that I have been exploring, what it takes for the first imagining to be accurate will be very different from what it takes for the second to be accurate. In the second imagining, the lack of constraints makes it likely that I am not really aiming at a particular fictional state of affairs but rather at a range of such states. In fact, perhaps I am aiming at a range in the first imagining too, but the range will be much smaller in the first case than in the second. When my aim is a more generalized range of fictional states of affairs rather than a particular state of affairs, accuracy is assessed relative to the range.¹⁷

3.6 Type 6: Unscripted scenarios, part 2

In some cases of daydreaming, however, I might not even have this more generalized aim. Sometimes, my daydreams are much less deliberate – as when I’m engaged in imaginative mind-wandering. In such cases, not only will I not be aiming at one specific fictional state of affairs rather than another, I won’t even be aiming at one specific range of fictional states of affairs rather than another. My imagining is aimless. That leaves the fictionality approach with no way to assess the imagining for accuracy.

Importantly, however, this shouldn’t be seen as a shortcoming of the fictionality approach. The fictionality approach has given us exactly the right result. If accuracy is assessed relative to aim, and my imagining lacks an aim, then this imagining simply can’t be assessed for accuracy. Should we be troubled by this outcome? To my mind the answer is no. Let’s recall how our discussion of accuracy began. At the start of the paper, I noted that there are various different dimensions along which imaginings vary from one another and, correspondingly, in terms of which one imagining might be said to be better than another. Such dimensions are important for judging imaginative skill. Importantly, however, not all of these dimensions need to come into play in each case of imagining. Here it might help to compare imagining to a different skilled activity, namely, painting. In some cases of painting, like portraiture, accuracy is a relevant factor of skilled performance. In other cases of painting, like abstract art, accuracy is not a relevant factor of skilled performance. In such paintings, accuracy cannot even be meaningfully assessed. Likewise, the fact that there are some imaginings in which accuracy cannot be meaningfully assessed should not be seen to call into question the notion of imaginative accuracy.

Again, let’s take stock. As the discussion of these final three cases has shown, imaginative accuracy is not just relevant in instructive uses of imagination; there are some transcendent uses where it is relevant as well. Because our aim in some of these transcendent uses is more diffuse, the conditions of accuracy will also be more diffuse – specified in terms of a range of fictional states of affairs rather than simply a single fictional states of affairs. But the basic fictionality approach is still applicable.

Of course, more would need to said were we try to develop a full-fledged account of imaginative accuracy. But in going through these six types of cases, I hope to have laid the groundwork for such an account, and I hope to have shown the explanatory potential of the fictionality approach.

4 Can imagination be calibrated?

Throughout this paper, I have been concerned with the nature of imaginative accuracy. I have focused on the question of what accuracy of imagining consists in and not the question of how someone might determine whether a given imagining is accurate. That said, insofar as we’re interested in the notion of imaginative accuracy because of its relevance to imaginative skill, this latter question proves important. And thus enters what I’ll call the calibration objection. The calibration objection arises from the fact that judgments of accuracy seem to require some sort of check to ensure their acceptability. But it’s hard to see how imaginings could be calibrated. What could they be checked against? If the calibration objection is correct, then it seems that there is no way to determine the accuracy of our imaginings.

In order to explore the objection with respect to the accuracy of imagination, it will be helpful first to see how it arises in a different philosophical context, namely, the debate about the accuracy of philosophical intuition. In that context, we see a very clear expression of the importance of calibration in a passage from Robert Cummins: “an observational technique is deemed acceptable just to the extent that it can be relied upon to produce accurate representations or indicators of its targets. This is why observational procedures in general, and instruments in particular, have to be calibrated” (Cummins, 1998, p. 116). This basic point gives rise to an objection to the claim that intuitions can serve to justify philosophical theories:

What we see through a telescope justifies us in believing that the moon has mountains because we have done things like looking through telescopes at distant mountains on earth and then gone to them and discovered that the telescopes presented an accurate picture of the mountains. But what, the objection goes, can intuitions be checked against? Other intuitions? But that is like checking a crystal ball against itself. (Russell, 2020)

Let’s separate the issue about whether intuitions can serve as sources of justification from whether they can be assessed for accuracy, and let’s focus on the latter. In response to this kind of objection about accuracy with respect to intuition, defenders of philosophical intuition might point out that matters are similar with respect to perception and memory. For any given perception, we can only calibrate it against other perceptions, but for any given memory, we can only calibrate it against other memories. Yet these facts do not prevent us from assessing the accuracy of perception and memory. So why should they prevent us from assessing the accuracy of intuition?

Faced with this response, the proponent of the calibration objection will likely insist that the situation is better with respect to perception and memory than it is with respect to intuition. For example, in many cases we can use one sensory modality as a check on another. The cross-sensory calibration helps us in assessing the accuracy of perception. Likewise, in many cases, we might use pictures, videos, written transcriptions, and other historical records as a check on what’s remembered. This helps us in assessing the accuracy of memory. But with intuitions, we don’t have any of these options at our disposal. With intuition, all we have are other intuitions. Granted, some of them are other people’s intuitions, so we can do interpersonal comparisons. Even so, however, the comparison remains within a circle of intuitions and thus doesn’t count as calibration (Cummins, 1998, p. 117).

I won’t further explore the continuing dialectic with respect to intuitions. For our purposes, what’s important is that imagination – like perception and memory – seems to be in a better position with respect to calibration than intuition is. In assessing the accuracy of imaginings, we need not stay within a circle of imaginings. Consider some of the examples of imaginings that we’ve previously seen in this discussion, such as my imagining what a stew would taste like if I added more cumin or my imagining a given hat on my friend. When I engage in the first imagining, I can compare my imagining to the taste of the stew. When I engage in the second imagining, I can compare my imagining to what she looks once she puts the hat on.¹⁸ Similar points can be made with respect to other imaginings as well. When I imagine what a character described in a novel looks like, I can compare my imagining to the author’s descriptions or, in some cases, to an illustration that has been provided (an illustration that depicts the fictional state of affairs at which I am aiming). When I imagine a pirate ship while playing a game of pretend with my children, I can compare my imagining to their descriptions of the pirate ship to check whether I have accurately latched onto the pretend ship in question. And when I imagine my mother’s first day of college, or other events where I wasn’t present, I can often compare my imaginings to photos that were taken. These comparisons allow us a variety of meaningful ways that we can calibrate our imaginings – and moreover, that we can do to a significant degree.

Granted, these calibrations aren’t intersubjectively accessible. No one else can access my imaginings to check whether they match what I am imagining. But the worry wasn’t that imaginings can’t be intersubjectively calibrated. After all, it’s not the case that we can do that with respect to memories either. The worry was that it is altogether impossible to calibrate imaginings.

That’s not to say that every imagining can be checked in a manner similar to the ones I’ve proposed. If I don’t end up adding the cumin, for example, I can’t check my imagining against the taste of the stew. And if I don’t buy the hat for my friend, I’m unlikely to be able to check my imagining of her in it against what she actually looks like when she puts the hat on. But of course, not every perception or memory will be able to be checked either, and likewise when it comes to scientific instruments like telescopes. Not every telescopic image of a scene can be checked against the scene itself. Still, when our telescope has been calibrated, we are justified in believing that its images are accurate even in cases where we can’t check. Likewise for imagining. In this way, then, we can defuse the force of the calibration objection. In at least some contexts, for at least some imaginings, we do seem to have ways to calibrate our imaginings, and thus ways to assess accuracy.

5 Concluding remarks

This paper has been concerned to explore what it means for one imagining to be more accurate than another. To help address this issue, I first tried to sort out some preliminary matters that seemed to stand in the way of our even talking about accuracy with respect to imagination. In particular, I addressed two different worries: the peculiarity worry and the mismatch worry. As we saw, both worries can be defused.

With those preliminaries out of the way, I then turned to the main task of the paper, namely, exploring what’s meant by imaginative accuracy. As I argued, the accuracy of imagining needs to be understood relative to the aim of an imaginative act. Because imagination is the kind of mental act that aims at the fictional, the accuracy of a particular depends on the fictional state of affairs at which it is aiming. I then explored six different types of imaginings to show how we might go about identifying the relevant state of affairs. My discussion shows how these assessments of accuracy connect to the distinction between transcendent and instructive imagining and the notion of imaginative constraints, both of which are matters that have been frequently discussed in recent philosophical work on imagination. Although more needs to be said, it’s my hope that the exploration of imaginative accuracy done here not only shows the coherence of the notion but also helps to show the explanatory potential of the fictionality approach.

This paper was motivated in part by the importance of accuracy in treating imagination as a skill, a claim that seems to underlie much recent work on imagination and to which I myself am committed (Kind, 2020). If imagination is a skill, and considerations of accuracy are one factor that comes into play in judging imaginative skill, then it’s important that our understanding of imaginative accuracy provides us with some way to assess when an imagination is more accurate than other. In the final section of this paper, I thus attempted to defuse the calibration objection, an objection that arises from the thought that imagination cannot be calibrated for accuracy. In line with the understanding of imaginative aim developed earlier in the paper, I addressed this objection by suggesting ways that imaginers might be able to provide some check on whether their imaginings are accurate or not. By defusing this objection, my exploration of imaginative accuracy helps clear the way for future philosophical work on imaginative skill.

Acknowledgments

Thanks to audiences at the online Bochum-Grenoble Memory Colloquium, the Workshop on Imagistic Cognition in Salzburg, the Models, Metaphors, and Simulations SLSAeu Conference in Erlangen, and the University of Glasgow (online) for helpful feedback and discussion. I am also grateful to the two referees for this journal; the paper benefitted greatly from their detailed comments.

References

---

1. Other common ways of taxonomizing imagination include Neil Van Leeuwen’s categorization in terms of constructive, attitudinal, and imagistic imagination (Van Leeuwen, 2013) and Gregory Currie and Ian Ravenscroft’s categorization in terms of creative and recreative imagination (Currie & Ravenscroft, 2002).↩︎

2. Of course, sometimes things that we imagine do come to fruition in the future. Perhaps this doesn’t happen in the case where I imagine Taylor Swift performing in the department lounge (though one can always hope!), but it does happen in other cases of everyday imagining, as when I imagine our end-of-year party in the department lounge. In a case like this, I can assess the accuracy of my prior imagining. I return to this kind of case below, in Part IV, in connection with what I call the calibration objection.↩︎

3. See, e.g., Robert Schwartz: “Perceptions are typically said to be veridical, when they appear phenomenally to be in agreement with the physical environment. Veridical experiences accurately reflect/ mirror how things are in the world.” (Schwartz, 2016, p. 384) Here, it’s worth noting that veridicality may come in degrees.↩︎

4. My discussion here assumes that imagination has representational content. Though this assumption is widely shared, there are some who reject it. See, e.g., (Hutto, 2015).↩︎

5. Thanks to a referee for this point.↩︎

6. For discussion of mindreading in the context of imagination, see e.g., (Goldman, 2006).↩︎

7. Considerable empirical work has been done regarding empathic accuracy and how it can be measured. See, e.g., the papers in (Ickes, 1997). It’s worth noting that in the empirical literature on empathic accuracy, empathy is sometimes treated as a process of inference, and this may seem different from the notion of empathetic imagination. But as I understand the discussion, it’s wholly compatible with imagination playing the key role in these inferential processes.↩︎

8. See https://www.mastersofgames.com/rules/darts-rules.htm↩︎

9. For this and other examples, see https://tvtropes.org/pmwiki/pmwiki.php/Main/AccidentalAimingSkills. Sometimes, we see the humor coming from the converse situation – a shot that looked awful and off target turned out to be perfectly accurate because it was exactly what the shooter was aiming for. For a famous example, see the quail-hunting scene in Lewis Gilbert’s 1979 film Moonraker. This and other examples are detailed at https://tvtropes.org/pmwiki/pmwiki.php/Main/ExactlyWhatIAimedAt.↩︎

10. Compare Daniel Munro’s discussion of the success conditions of imagining; as he notes: “we can determine a cognitive faculty’s success conditions by first determining its function” (Munro, 2021, p. 6).↩︎

11. With respect to propositional imagining in particular, the distinction between these two levels relates to the attitude/content distinction. For related discussion, see (Arcangeli, 2020).↩︎

12. My talk about “particular imaginings” should be understood to include particular imaginative projects. See, e.g., Chasid (2017); Dorsch (2012).↩︎

13. For example, Wiltsher (Wiltsher, 2019), Langland-Hassan (Langland-Hassan, 2020), and I (Kind, 2001) all reject the claim that imagining is fundamentally simulative in nature (though for very different reasons).↩︎

14. Note that neither the fictionality approach nor the possibility approach assumes that imagination is non-simulative. Indeed, there may be a way of construing the simulationist approach so that it’s simply a more specific version of the possibility or fictionality approach: In aiming to simulate a particular perceptual, cognitive, or other mental state, an imaginer can be taken to be aiming at a particular possible or fictional world in which they are undergoing that state.↩︎

15. In discussing the distinction between fiction and non-fiction, Stacie Friend provides numerous other examples where fictions contain claims that are true of the actual world (and thus where fictional states of affairs turn out to be actual states of affairs); see, e.g., Friend (2008).↩︎

16. As a brief aside, it’s worth noting that the distinction between transcendent and instructive uses of imagination brings to the fore an entirely different sense in which one might talk about imaginative aim. We have already distinguished two senses: the aim of an imagining qua imagining and the aim of an imagining qua the particular imagining that it is. These both concern imaginative aims related to content, what might be seen as internal aims. But in thinking about the distinction between transcendent and instructive imagining, we see that the notion of imaginative aim might be construed in terms of external aims, namely, in terms of what one is aiming to achieve with the imagining. Since this sense of aim is less relevant to accuracy, I am here setting it aside.↩︎

17. It’s worth being explicit that it is not only in transcendent cases of imagining that we might aim at a range of fictional states of affairs rather than just a particular fictional state of affairs. For example, when I am trying to imagine whether the desk fits through the doorway, there might be a range of rotations that would work rather than just a single one.↩︎

18. There are still details to work out. For example, some of these calibration efforts require us to call upon memory, and that may introduce distorting effects. Thanks to a referee for raising this worry.↩︎