The Contents of Imagination

Abstract Our imaginings seem to be similar to our perceiving and remembering episodes in that they all represent something. They all seem to have content. But what exactly is the structure and the source of the content of our imaginings? In this paper, I put forward an account of imaginative content. The main tenet of this account is that, when a subject tries to imagine a state of affairs by having some experience, their imagining has a counterfactual content. What the subject imagines is that perceiving the state of affairs would be, for them, like having that experience. I discuss three alternative views of imaginative content, and argue that none of them can account for two types of error in imagination. The proposed view, I suggest, can account for both types of error while, at the same time, preserving some intuitions which seem to motivate the alternative views.


Introduction
We use our faculty of imagination frequently, and we do it for various purposes. In some cases, our imaginings consist in bringing to mind perceivable states of affairs (as opposed to merely supposing, or assuming, that they are the case). Such imaginings typically involve quasi-perceptual experiences wherein those states of affairs are presented to us. They involve, more specifically, mental images. 1 If I have, for example, an imagining that I would report by saying that I imagine that I am celebrating the end of the COVID-19 pandemic by drinking with a friend in the pub, then, typically, I have an experience wherein I visualise that my friend and I are celebrating in the pub; a mental image that allows me to picture the scene in my mind. Let us call imaginings of this kind 'sensory imaginings. ' Sensory imaginings seem to be similar to both perceiving and remembering episodes in that sensory imaginings have content. Sensory imaginings represent, or are about, something. But what exactly is the content of our sensory imaginings? Do those imaginings represent states of affairs out there, in the world, or do they represent our perceptual experiences of those states of affairs? And what is the source of the content of our sensory imaginings? Is their content determined by the mental images which are involved in those imaginings, or is it determined by what we are trying to imagine when we have those mental images? The aim of this discussion will be to answer these questions about the nature and the source of imaginative content, thus clarifying the intentionality of imagination.
Essentially, my contention will be that the content of sensory imaginings is counterfactual. When a subject tries to imagine a state of affairs by having some experience, what the subject imagines, strictly speaking, is that perceiving the state of affairs would be, for them, like having that experience. To put it differently, what the subject imagines is that if they perceived the state of affairs, then they would have the type of experience that they are currently having. I will refer to this view as 'the counterfactual view. ' The case for the counterfactual view will be built as follows. In section 2, I will specify the sort of mental state and the sort of mental content which will concern us in this discussion. In sections 3 through 5, I will consider three candidate accounts of imaginative content in the literature. I will argue that the three accounts fail to accommodate our intuitions about two kinds of cases involving error in imagination. Nevertheless, our discussion in sections 3 through 5 will not be fruitless. For it will serve to highlight two valuable intuitions which seem to motivate the three accounts of imaginative content. One of those intuitions is the intuition that our sensory imaginings do not need to answer to what goes on in the actual world. The other one is the intuition that we can use a single mental image to imagine different things at different times. In section 6, I will put forward the counterfactual view, and argue that it accommodates the two kinds of cases which are not easily accommodated by the alternative accounts. Furthermore, I will argue that the counterfactual view captures the two intuitions about imagination which seem to motivate those accounts. Accordingly, I will conclude that the counterfactual view is the correct account of imaginative content.

Scope of the project
Sensory imaginings can be voluntary or involuntary. If I try to imagine, for example, that my friend and I are celebrating the end of the pandemic and, as a result, I visualise the scene of the celebration in the pub, then my sensory imagining is voluntary. But sensory imaginings can also be unbidden, or involuntary. They can even be unwelcome. If I am, for instance, trying to concentrate on marking a philosophy paper and, out of the blue, I find myself visualising the scene in the pub, then my sensory imagining is involuntary.
The focus of our discussion here will be on voluntary sensory imaginings. Furthermore, we speak of imaginings as if they concerned states of affairs, properties, events, actions, or objects. Thus, we self-ascribe imaginings with expressions of the form "I imagine that p" (where p stands for a proposition, as in "I imagine that I am drinking with my friend"), "I imagine e" (where e stands for an event, as in "I imagine the celebration"), "I imagine Q" (where Q stands for a property, as in "I imagine our joy in the pub"), "I imagine Φ-ing" (where Φ-ing stands for an action, as in "I imagine drinking with my friend"), and "I imagine x" (where x stands for an object, as in "I imagine my friend"). I will take propositional imagination to be the central kind of voluntary sensory imagination. 2 The issue that will concern us in this discussion, then, is what the contents of voluntary sensory imaginings are, and how those contents are determined, when the imaginings at issue are propositional. Hereafter, I will use the term 'imagining' to refer to voluntary sensory imaginings of that sort. 3 In what follows, I will also be operating with a relatively undemanding notion of mental content. The claim that, according to some candidate theory of imaginative content, the content of an imagining is the proposition that p will be understood simply as the claim that, according to the theory, what it takes for the imagining to be successful is that p or, to put it differently, the imagining is successful just in case p. This is a modest claim. Notice that, by attributing to a subject's imagining the content that p, in this sense, we are allowing for the possibility that the subject themselves may not be able to spell out what they imagine as the proposition that p. (This may happen, for example, if the subject lacks the necessary concepts to frame the proposition.) Attributing to a subject's imagining the content that p is only meant to imply that the proposition that p captures the success conditions for the subject's imagining. 4 Thinking of the content of an imagining as its success conditions, however, has the potential to generate some confusion. For there are at least two other kinds of mental states, typically associated with imaginings, which also enjoy success conditions. In the interest of clarity, then, let us distinguish the success conditions of imaginings from those of the other two kinds of mental states.
Firstly, if a subject has an imagining in virtue of forming some mental image, then there is usually something that the subject was intending, or trying, to imagine when they formed their mental image. We may refer to what the subject was intending to imagine as their 'imaginative project.' Suppose that I intend to imagine that my friend and I are celebrating the end of the pandemic and, as a result, I visualise the scene in the pub. Then, my imaginative project is that my friend and I are celebrating the end of the pandemic. Secondly, if a subject has an imagining in virtue of forming some mental image, then there is usually some broader activity that the subject was carrying out with their imagining; some purpose that their imagining was supposed to serve. We may refer to the broader activity that the subject is carrying out with their imagining as their 'imaginative exercise.' For example, suppose that the reason why I try to imagine that my friend and I are celebrating in the pub is that I am in the process of deciding whether celebrating in the pub would be more fun than, let us say, celebrating at home. Then, my imaginative exercise is a decision-making exercise, namely, making up my mind on how to celebrate the end of the pandemic.
On the face of it, both our imaginative projects and our imaginative exercises have success conditions. After all, the notion of imaginative project introduced above is that of an intention; the intention to have a certain imagining. For each of our imaginative projects, then, we can think of its success conditions thus: our imaginative project is successful just in case we are having whatever imagining we intended to have in that project. Similarly, the notion of imaginative exercise introduced above is that of an action; the action of using an imagining for the purposes of achieving a certain goal. For each of our imaginative exercises, then, we can think of its success conditions as follows: our imaginative exercise is successful just in case, by using the corresponding imagining, we have achieved the goal of our imaginative exercise.
Notice, now, that the success conditions of our imaginings are different from those of our imaginative projects. The former concern the issue of whether our imaginings are correct, or accurate, whereas the latter concern the issue of whether our imaginings are fulfilling the imaginative projects in which they originate. 5 Thus, the former kind of conditions do not need to be satisfied in order for the latter kind to be satisfied. I may try to imagine, for example, that some painting which has been consumed by an accidental fire is looking beautiful before the accident. And I may succeed in doing so even if, as a matter of fact, the painting never looked beautiful and, for that reason, my imagining is 4 The fact that the content of an imagining, in the sense which will concern us in this discussion, is identical with its success conditions explains why it may sometimes be appropriate to assign to a subject's imagining the content that p, even though the subject themselves would not report their imagining by saying that they imagine that p. The reason is that the success conditions of a subject's mental state do not need to be available to the subject through introspection. This is not a new idea. It is a common idea, for example, within the externalist literature on mental content (Putnam 1975). 5 If, when a subject has an imaginative project and they imagine something as a result, what the subject imagines needs to coincide with their imaginative project, then, trivially, our imaginings always fulfill the imaginative projects in which they originate. But the view that such a coincidence must obtain seems debatable. See, for example, Williams (1973, 33) for an endorsement of this view and Munro and Strohminger (2021) for a rejection of it. incorrect. 6 The success conditions of our imaginings are also different from those of our imaginative exercises. The former concern the issue of whether our imaginings are correct, or accurate, whereas the latter concern the issue of whether our imaginings have served the purpose for which we had them or not. Thus, the former kind of conditions do not need to be satisfied in order for the latter kind to be satisfied. For example, I may, once again, imagine that the painting which was consumed by the fire is looking beautiful before the accident. I may do so for the purposes of delighting myself in the beauty of the imagery involved. And I may succeed in obtaining some aesthetic pleasure from this fantasy even if, as a matter of fact, the painting never looked beautiful and, for that reason, my imagining is incorrect. Neither do the success conditions of our imaginative exercises need to be satisfied for those of our corresponding imaginings to be satisfied. Suppose, for example, that I imagine that the painting is looking beautiful before the accident for the purposes of delighting myself in the beauty of the imagery involved, and it so happens that the painting did look beautiful before the accident. But my imagining fails to provide me with any aesthetic pleasure because, let us say, I am deeply depressed. 7 There seems to be a broad sense, then, in which, not only our imaginings, but also our imaginative projects and our imaginative exercises have content, since they all seem to have success conditions. But the relevant conditions are importantly different in each case. The focus of our inquiry here will be on the content of our imaginings, as opposed to the content of our associated imaginative projects and exercises. Let us consider, then, three proposals regarding the nature, and the source, of this kind of imaginative content.

The possibility view
Perhaps the most natural view of imaginative content is that, just like perception and memory represent actuality, imagination represents possibility. If I try to imagine, for example, that I am celebrating with my friend in the pub, and I have some mental image as a result of that imaginative project, then my imagining represents the possibility of what is depicted by my mental image. If my mental image represents, for example, that my friend and I are celebrating in the pub, then the content of my imagining is that it is possible that I am celebrating with my friend in the pub. Let us call this view, the 'possibility view' (or, for short, 'PV'). We may formulate the view more precisely as follows:

Possibility View
For every subject S, proposition p, and mental image i: If S has an imaginative project which causes S to have i, and i represents that p, then, in virtue of having i, S imagines that it is possible that p.
6 The distinction between the success conditions of our imaginative projects and those of the imaginings which originate in them will be deliberately blurred in section 5, when we discuss the possibility of imagination being a kind of action. 7 The distinction between the success conditions of our imaginings and those of the imaginative exercises in which they are involved sets this project apart from other similar projects in the literature. Peter Langland-Hassan, for example, offers a pluralistic account of the success conditions of imaginings according to which our imaginings have different kinds of success conditions depending on the helpfulness of those imaginings in guiding our actions and achieving our goals (2015, 665). Langland-Hassan regards the fact that, on his pluralistic account, there is a link between the conditions under which an imagining aids human action and inference, and the conditions under which it is veridical, as a virtue of the account. The present project, by contrast, starts by assuming that the two kinds of conditions need to be differentiated, and that no such link needs to exist between them, as the cases involving the burnt painting seem to illustrate. The methodological assumption that I will be making, then, is that the contents of our imaginings do not need to depend on what we are trying to achieve by having those imaginings. This is, after all, an assumption which is often made in discussions of other types of mental content, such as doxastic content, perceptual content, or mnemonic content. It seems that approaching imaginative content differently would be ad hoc. Naturally, we might be pushed toward the pluralistic position if we reach the conclusion that a unified account of imaginative content, one which applies to our sensory imaginings irrespective of the purpose for which we are having those imaginings, cannot be found. But, as I see it, we should initially aim for a unified account, on grounds of theoretical simplicity.
Stephen Yablo, for example, seems to endorse a version of PV when he claims that conceiving "involves the appearance of possibility" (1993,5). An important virtue of PV is that it succeeds in capturing a certain intuition about imagination. This is the intuition that our imaginings do not need to answer to what goes on in the actual world. 8 This idea is worth preserving. It squares, for example, with the fact that our imagination is the main faculty involved in our engagement with fiction. PV captures this intuition by positing that our imaginings represent states of affairs which do not need to obtain in the actual world. Instead, they may obtain in other possible worlds. 9 If PV is correct, then it is no wonder that we have the sense that our imaginings do not need to answer to what goes on in the actual world.
There are, however, two difficulties that the PV advocate will need to face. The first difficulty is that PV seems to have trouble accommodating our intuitions about a particular type of sensory imagining. Consider the following case:

Well-Intentioned Parent
Paul's daughter, Mia, is about to turn six years old. Paul wants to buy her a birthday present that she is going to be thrilled by, and decides to buy her a doll house as big as a bicycle. Mia, however, is really hoping for a new bicycle. On the eve of Mia's birthday, Paul wraps the dollhouse up in red paper, leaves it in the living room, and tries to imagine that Mia is opening her present in the morning. He can visualise Mia ripping up the red paper and looking thrilled as she discovers the giant dollhouse inside. As it happens, in the morning, Mia sees a present the size of a bicycle, and quickly unwraps her present. But she discovers a dollhouse instead, and she cannot hide her profound disappointment. Thus, Paul thought, on the eve of Mia's birthday, that he was accurately imagining that Mia is opening her birthday present. But he was wrong, since Mia is not happy at all when she discovers her present.
Well-Intentioned Parent illustrates one type of mistake that we can make while using our imagination (Kind 2016). It is the type of mistake that we make when, by having some mental image, we imagine some state of affairs that we expect to be the case, but our imagining is not correct because the state of affairs in question does not eventually obtain. We may call this kind of error, a 'prediction error' in imagination. It seems that PV will have trouble accounting for the possibility of prediction errors in imagination (Langland-Hassan 2015, 672). Take Paul in Well-Intentioned Parent, for example. The intuition is that, when Paul visualises that Mia is happy while she discovers her birthday present, Paul is not accurately imagining the scene that will take place in their living room in the morning. PV seems to conflict with this intuition. For if our imaginings have the kind of content that PV attributes to them, then Paul's imagining should be correct: as a result of trying to imagine that Mia opens her birthday present, Paul has a mental image which represents that Mia is happy while she opens it. And, on PV, this means that, in virtue of having his mental image, Paul is imagining that it is possible that Mia is happy when she opens her birthday present. But it is indeed possible for Mia to be happy when she opens her birthday present (even if, as a matter of fact, she is not happy at all). Thus, Paul's imagining should be correct if it enjoys the content attributed to it by PV. And yet, intuitively enough, Paul's imagining is not correct. It seems, therefore, that the possibility of prediction errors in imagination poses a challenge to PV.
A further difficulty for PV concerns a certain idea within the view; an idea about the sources of imaginative content. This is the idea that, when a subject imagines something by having a mental image, what the subject imagines is a function of what is depicted by the mental image that they are having. According to PV, the content of Paul's imagining, for example, is not determined by what he is trying to imagine when he visualises that Mia is happy while she opens her present. Instead, it is determined by the mental image that Paul has actually formed. For, on PV, what Paul is imagining, in virtue of having his mental image, is the possibility of the state of affairs which is depicted by his mental image. The difficulty for PV, now, is the following. It seems possible for a subject to imagine two different things by having, at different times and in conjunction with different imaginative projects, the very same mental image. Intuitively enough, one may imagine, for example, that there is a suitcase in a room by having some mental image, and one may also imagine that a cat is hiding behind a suitcase in a room by having the same mental image. 10 But if the content of one's imagining is a function of what is depicted by the mental image involved in that episode, as the PV advocate suggests, then it should not be possible for one to do this. How can I sometimes imagine that it is possible that there is a suitcase in a room, and sometimes imagine that it is possible that a cat is hiding behind a suitcase in a room, by forming the very same mental image if, by assumption, the content of my imagining is fixed by my mental image? 11 The possibility that one may imagine two different things by having, at different times and in conjunction with different imaginative projects, the same mental image seems, therefore, to pose a challenge to PV as well.
It seems, then, that PV succeeds in capturing one of our intuitions about imagination, namely, the intuition that, normally, our imaginings are not answerable to what goes on in the actual world. However, it also seems that PV has trouble accommodating a different intuition about imagination, that is, the intuition that prediction errors are possible in imagination. Furthermore, PV locates the source of the content of our imaginings in the mental images involved in those imaginings, and this feature of PV does not sit easily with the possibility of using a single mental image for imagining different things; something that we seem to be able to do. Let us keep in mind, then, those three features of PV while we turn our attention to an alternative view of imaginative content.

The world view
In light of the challenges for the possibility view, one avenue worth exploring is to emphasise a certain similarity between imagination and perception. The relevant similarity is the following. Our perceiving episodes are, in a sense, aimed at the actual world. In what sense? Our perceiving 10 Christopher Peacocke (1985, 19) brings up this possibility. While this possibility is widely accepted in the literature (White 1990, 92;Martin 2002, 403;Kung 2010, 624), it is not accepted unanimously. For a dissenting view, see Wiltsher (2016). 11 The PV advocate could argue that, in the two suitcase scenarios, I am having different, though intrinsically indistinguishable, mental images. After all, in one case, the mental image I am having is an image of a suitcase in a room and, in the other case, it is an image of a cat hiding behind a suitcase in a room. To make this response persuasive enough, though, the PV advocate will need to produce an independent account of how the content of our mental images is determined. The account of imaginative content to be proposed in section 6, by contrast, is not committed to any particular account of the intentionality of mental imagery. The reason is that, on the proposed account of imaginative content, the mental image that a subject is having when they imagine something sensorily makes a contribution to the content of their imagining. But the subject's mental image makes such a contribution, not in virtue of its content, but in virtue of its phenomenal features. episodes are correct just in case the states of affairs that we seem to perceive when we undergo those episodes obtain in the actual world. Similarly, one might argue, our imaginings are aimed at some world, but not the actual world. Every imaginative exercise targets some possible world and, by imagining something, we have a mental episode which is correct just in case the state of affairs that we are trying to imagine when we undergo that episode obtains in the relevant world. Let us call this view, the 'world view' (or, for short, 'WV'). We may formulate the view more precisely as follows:

World View
For every possible world W, subject S, proposition p, and mental image i: If S has the imaginative project that p, S's imaginative project causes S to have i, and i represents W, then, in virtue of having i, S imagines that, in W, it is the case that p.
The main idea in WV is that imagining something is representing it to be the case in a certain world; a world that the subject visualises through a mental image. When I try to imagine, for example, that I am celebrating the end of the COVID-19 pandemic with my friend by having a mental image of the scene in the pub, my resulting episode of imagination has a content with two different sources. On the one hand, my imaginative project makes a contribution to it. I am trying to imagine that I am celebrating with my friend, and this is why what I imagine is that, in a certain world, I am celebrating with my friend. On the other hand, my mental image makes a contribution to what I imagine as well. I am visualising some possible world by having that mental image, and this is why what I imagine is that, in that particular world, I am celebrating with my friend. The combination of the two factors yields the content outlined above: what I imagine, in virtue of having my mental image, is that what I am trying to imagine is the case in the world depicted by my mental image. Alon Chasid (2021), for example, subscribes a version of WV. Kendall Walton, too, seems to subscribe WV when he claims that "imagining aims at the fictional as belief aims at the true" (1990,41). 12 The world view enjoys two important virtues. Like PV, WV preserves the intuition that, normally, our imaginings do not need to answer to what goes on in the actual world. It preserves this intuition by positing that our imaginings concern, not the actual world, but possible worlds that we visualise through the mental images involved in our imaginings. Unlike PV, however, WV differentiates between what we imagine when, as a result of trying to imagine something, we have some mental image, and what we visualise through that mental image. This allows the WV advocate to accommodate the intuition that one can, at different times and in conjunction with different imaginative projects, imagine different things by having the same mental image. Consider, for example, the possibility that, by having the same mental image, one may imagine either that there is a suitcase in a room or that a cat is hiding behind a suitcase in a room. The issue of whether one imagines one thing or the other, the WV advocate will argue, depends on whether one's mental image is the result of trying to imagine that there is a suitcase in a room or, by contrast, it is the result of trying to imagine that a cat is hiding behind a suitcase in a room. In the former case, what one imagines is that, in the world visualised through one's mental image, there is a suitcase in a room. In the latter case, what one imagines is that, in the world visualised through one's mental image, there is a cat hiding behind a suitcase in a room. In this way, the WV advocate can accommodate the possibility of using the same mental image in imaginings with different contents.
It seems, however, that making room for the possibility of using the same mental image in imaginings with different contents will come at a cost for the WV advocate. Notice that the way in which this is achieved is by making the subject's imaginative project a prominent source of the content of their imagining. This, I suggest, will make it difficult for WV to account for our intuitions regarding a different kind of imagining. Consider the following case:

Junior Fireman
John has taken a new job at the Oxfordshire fire brigade, and he decides to take a tour of Oxford University to familiarise himself with the location of its various buildings. During the tour, the tour guide stops at Catte Street and points at Hertford College. John cannot hear what the guide is saying. He overhears, however, another man in the group mentioning New College, and he believes, incorrectly, that they are in front of New College. A week later, John's boss asks the team where they would go if the fire station received a call alerting the team that New College is on fire. John visualises the building at which the tour guide was pointing on fire, and he answers that he would drive the fire engine to Catte Street. John thinks that he is accurately imagining that New College is on fire. But John's boss tells him that this cannot be right since New College is not on Catte Street.
Junior Fireman illustrates a different mistake that we can make while using our imagination. It is the kind of mistake that we make when we are trying to imagine that some state of affairs is the case, and we entertain some mental image as a result, but our imagining is not correct because we are mistaken about the appearance of some object involved in the relevant state of affairs. We may call this kind of error, a 'misidentification error' in imagination. 13 It seems that WV will have trouble accounting for the possibility of misidentification errors in imagination. Take John in Junior Fireman, for example. The intuition is that, when John visualises the building at which the tour guide was pointing on fire, John is not accurately imagining that New College is on fire. WV has trouble accounting for this intuition, since the view fails to predict that the content of John's imagining is incorrect: as a result of trying to imagine that New College is on fire, John has a mental image of a different building (namely, Hertford College). What does this mean according to WV? It means that, in virtue of having that mental image, John is imagining that, in the world depicted by his mental image, the building targeted by his imaginative project (that is, New College) is on fire. But surely there are possible worlds in which both Hertford College and New College are on fire. And nothing in Junior Fireman seems to rule out the possibility that John's mental image is depicting one of those possible worlds. It is difficult to see, then, how WV can account for misidentification errors in imagination since, in cases such as Junior Fireman, the view leaves open the question of whether the content of the subject's imagining is correct or not. 14 Things seem to be worse for WV when it comes to Well-Intentioned Parent. For in Well-Intentioned Parent, a case could be made that WV attributes a content to the subject's imagining which is, in fact, correct. According to WV, when Paul pictures what will happen in the morning in his mind, there is a certain world that he is visualising. The question, then, is whether, in that world, what Paul tries to imagine obtains or not. Paul tries to imagine that Mia is opening her present. Paul's imaginative project, therefore, is that Mia is opening her present. Now, is Mia 13 The possibility of misidentification errors in imagination seems to be rejected by Ludwig Wittgenstein, when he claims (1958,39) that, when a person asserts that they imagine King's College on fire, it is not possible to challenge their assertion by questioning whether it is King's College, and not some other building, which they are really imagining. The possibility of misidentification errors in imagination is also discussed in (Williams 1973, 29-30) and in (Noordhof 2018, 113), from which the example of confusing Hertford College with New College is taken.
14 Can PV handle misidentification errors better than WV? Since John is having a mental image of Hertford College, the content of John's imagining, according to PV, is that it is possible that Hertford College is on fire, which certainly seems true. On the face of it, therefore, misidentification errors appear to be a challenge for PV as much as they are for WV. The PV advocate, however, may reply that the intuition that John's imagination is malfunctioning in Junior Fireman is not due to the fact that John's imagining is incorrect, but to the fact that John's imaginative project has failed. (John is not imagining what he intended to imagine.) Given that, according to WV, the content of John's imagining is almost identical to that of his imaginative project, this reply does not seem to be available to the WV advocate. opening her present in the world visualised by Paul on the eve of Mia's birthday? There seems to be no reason to think that she is not. But if Mia is opening her present in the world visualised by Paul, then the content of Paul's imagining is correct. Thus, it seems that, rather than accounting for the kind of error which takes place in Well-Intentioned Parent, the WV advocate will have to reject the idea that, in that scenario, the content of the subject's imagining is incorrect, which seems counterintuitive.
The upshot is that, while WV enjoys some significant benefits, the main tenet of WV puts it at odds with the possibility of both misidentification and prediction errors in imagination. The tenet in question is the claim that we aim to conform our imaginings to worlds other than the actual one. It seems, therefore, that dropping that claim may be an option worth considering at this point. Let us turn our attention, then, to a view according to which our imaginings are not aimed at conforming to possible worlds, or to anything else, for that matter.

The action view
To sidestep the difficulties for the Possibility View, we have considered a view which makes imaginings analogous to perceiving episodes. An alternative route around those difficulties is to emphasise, instead, a certain similarity between imagination and intention. The relevant similarity is the following. Our intentions are, in a sense, meant to be matched by the actual world. In what sense? When we intend some proposition to be the case, we take the actual world to be acted upon so as to make it the case that the proposition obtains in that world. Similarly, one might argue, our imaginings are meant to be matched by a specific part of the actual world, namely, our own minds (and, in particular, our own mental images). Every imagining consists in the act of fulfilling a certain intention; the intention which constitutes our imaginative project. Thus, when we imagine something, we take our own minds to be acted upon so as to make it the case that what we are visualising matches our imaginative project. Let us call this view the 'action view' (or, for short, 'AV'). We may formulate the view more precisely as follows:

Action View
For every subject S and proposition p: If S has an imagining by carrying out the imaginative project that p, then the goal of S's imagining is to have a mental image i such that i represents that p.
The main idea in AV is that imagining something is not a state, but an action. It is the action of forming a mental image which, when the action is successfully carried out, matches one's imaginative project. When I try to imagine, for example, that I am celebrating with my friend, my resulting act of imagining has a characteristic content, or goal. The goal of my imagining is determined by what I intend to imagine. I intend to imagine that I am celebrating with my friend. For that reason, the goal of my imagining is to have a mental image which represents that my friend and I are celebrating together. If I manage to form such a mental image, then my imagining is successful. Otherwise, it is not. Paul Noordhof, for instance, endorses a version of AV (2002AV ( , 2018.
The action view has three important virtues. Firstly, like both PV and WV, AV captures the intuition that our imaginings do not have to answer to what goes on in the actual world. It does so by positing that imaginings have a different direction of fit from that of perceiving episodes. According to the AV advocate, the reason why we sense that our imaginings do not need to conform to the actual world is that, in fact, they do not need to conform to anything. On AV, imaginings lack that kind of direction of fit, since they are actions. It is not surprising, then, that we feel that our imaginings do not need to answer to what goes on in the actual world. Secondly, like WV, AV makes room for the possibility that one may imagine two different things by having, at different times and in conjunction with different imaginative projects, the very same mental image. And, like WV, it does so by allowing that what one imagines, when one has a mental image, is partly determined by the imaginative project which has prompted that mental image in one's mind. Thirdly, AV makes room for misidentification errors in imagination. The AV advocate may tackle, for example, Junior Fireman as follows. The reason why the content of John's imagining is incorrect when he visualises that the building on Catte Street is on fire is that John's mental image does not match his imaginative project. John intends to imagine that New College is on fire. For that reason, the content of John's imagining, when he carries out his intention, is to visualise that New College is on fire. But John is not having a mental image of New College on fire. Instead, John is visualising that Hertford College is on fire, which means that John's mental image is not matching his imaginative project. This seems to capture the intuition that John's imagining is incorrect.
Prediction errors, however, continue to pose a challenge to AV, for similar reasons why they posed a challenge to WV. Consider Paul in Well-Intentioned Parent. A case could be made that the content of Paul's imagining turns out to be correct within AV. Paul is trying to imagine that Mia is opening her present. His imaginative project, then, is that Mia is opening her present. And, for that reason, the content of Paul's imagining, when he carries out his project, is to have a mental image which represents that Mia is opening her present. But, intuitively, when Paul has his mental image of the living room scene in the morning, he does visualise that Mia is opening her present. To be sure, Paul also visualises that Mia is happy; a state of affairs which does not eventuate. Nevertheless, the fact that Paul visualises that Mia is opening her present seems to be sufficient for satisfying Paul's imaginative project. And yet, we seem to have the intuition that Paul's imagining is incorrect. It appears, therefore, that, far from explaining how prediction errors arise in imagination, AV suggests that, in cases such as Well-Intentioned Parent, the subject's imagining is correct, which seems implausible.
The upshot is that the action view has some virtues, but also some shortcomings. Ideally, we would want a view which, like PV, accounts for the intuition that imagination does not need to answer to what happens in the actual world and, like WV and AV, accounts for the intuition that we can utilise the same mental image for imagining different things at different times. We would also want a view which, like AV, accommodates misidentification errors in imagination. And, finally, we would want a view which, unlike all of the views discussed, can account for prediction errors in imagination. I suggest that some changes in the way we think about the content of our imaginings will allow us to arrive at such a view.

The counterfactual view
The main idea in the account of imaginative content that I wish to put forward is that the contents of our imaginings are counterfactual. This idea springs from the thought that the activity of imagining sensorily is always a sort of guess, or simulation. It is a simulation of what it would be like for the subject to have some possible perceptual experience. 15 As a result, when a subject tries to imagine a state of affairs, and this leads them to have some mental image, what the subject imagines, strictly speaking, is not the state of affairs in question (though that may well be what they visualise). What they imagine is that perceiving the state of affairs would be, for them, like having the mental image that they are having. In the pandemic example, for instance, when I try to imagine that my friend and I are celebrating in the pub, and I have a certain mental image as a result, what I am doing, in fact, is simulating what it would be like for me to perceive that my friend and I are celebrating in the pub. And, for that reason, what I imagine, properly speaking, is something about my own possible perceptual experiences. I imagine that if I perceived that my friend and I are celebrating in the pub, then having that perception would be, for me, like having the mental image that I am currently having.
The proposal is, more precisely, the following. When a subject S tries to imagine that some proposition p is the case, and they have some mental image i as a result, S's imagining has a content which can be formulated counterfactually. The antecedent of the relevant counterfactual involves the perception that p, whereas its consequent involves a reference to the very mental image i that S is having. Specifically, the content of S's imagining is that if S perceived that p, then S would have an experience of a certain phenomenological kind; an experience phenomenologically similar to having i. We may call this view, the 'counterfactual view' (or, for short, 'CV'):

Counterfactual View
For every subject S, proposition p, and mental image i: If S has the imaginative project that p, and S's imaginative project causes S to have i, then, in virtue of having i, S imagines that if S perceived that p, then having that perception would be, for S, like having i.
Two aspects of this view are worth highlighting. First of all, the counterfactual content that CV attributes to subject S's imagining needs to be read in a particular way. The relevant reading can be spelled out, in the terminology from the possible-worlds framework, thus: there is a possible world W in which both S perceives that p and S's experience is phenomenologically similar to having i, and any world in which S perceives that p which is as close or closer to the actual world than W is also a world in which S's experience is phenomenologically similar to having i. 16 This reading of the counterfactual content of S's imagining is strong in that it makes the counterfactual false in the situation in which S perceiving that p is impossible. This is a deliberate feature of the view. In the situation in which I try to imagine, for example, that water is not H 2 O, and I form some mental image as a result, it seems that the content of my imagining will be incorrect regardless of what mental image I have formed. And CV delivers this outcome. On CV, the content of my imagining is that if I perceived that water is not H 2 O, then my experience would be phenomenologically similar to that of having my current mental image. On the strong reading of this counterfactual, the counterfactual turns out to be false, since there is no possible world in which both I perceive that water is not H 2 O and my experience is similar to that of having my current mental image. This squares with our intuition that irrespective of what mental image I happen to be entertaining as a result of trying to imagine that water is not H 2 O, the content of my imagining is bound to be incorrect. 17 Secondly, the contributions that, according to CV, subject S's imaginative project and S's mental image i make to the content of S's imagining are also worth highlighting. S's imaginative project makes a contribution to the counterfactual content of S's imagining in that the antecedent of the relevant counterfactual is a function of S's imaginative project. By contrast, the consequent of the counterfactual is concerned not with S's imaginative project, but with the mental image i that S is having. Notice, furthermore, that i makes a contribution to the counterfactual content of S's imagining, but not in virtue of the fact that i has a certain content itself. Mental image i makes a contribution to the content of S's imagining in virtue of the fact that it belongs to a particular phenomenological kind. Now, one might worry that, according to CV, the contents of our imaginings are implausibly complex. After all, it seems natural for us to say that, in the pandemic example, I imagine that I am celebrating with my friend in the pub. This seems to be a straightforward way of describing what I imagine in that example. And yet, on CV, I can never simply imagine that I am celebrating with my friend in the pub. I can only imagine a much more complex, counterfactual proposition. So why is this not a difficulty for CV? The CV advocate may address this challenge by replying that, in the pandemic example, it is indeed appropriate for others to attribute to me an imagining by saying that I imagine that I am celebrating with my friend in the pub. It is also appropriate for me to selfattribute an imagining by making the very same claim. But those kinds of claims are to be interpreted as shorthand for a more precise description of what I imagine. 18 Strictly speaking, what I imagine is something more complex, namely, a counterfactual fact about my own possible perceptual experiences. The fact that I may not be able to spell out what I imagine in that way does not show that this is not the content of my imagining. For the success conditions of my own imagining may not be accessible to me by simply reflecting on what I am doing when I try to imagine that my friend and I are celebrating in the pub. 19 The relevant consideration, in order to evaluate CV, is whether the complex counterfactual propositions that CV posits as the contents of our imaginings capture, or not, our intuitions about the success and failure of those imaginings in different scenarios. It seems that CV fares quite well in that regard. For CV can make room for both misidentification errors and prediction errors in imagination. And it can do so while at the same time accommodating the two intuitions about imagination which seemed to motivate the alternative views of imaginative content.
Consider misidentification errors first. On CV, the reason why John fails to accurately imagine that New College is on fire, when he visualises the building on Catte street being on fire, is that the content of John's imagining is the following counterfactual: if John perceived that New College is on fire, he would have the kind of experience that he is having when he visualises that the building on Catte street is on fire. But this counterfactual, it turns out, is false. If John perceived that New College is on fire, he would experience being in front of a building which looks different from the building that he is visualising at the moment. (This is, of course, because the building on Catte Street is not New College, but Hertford College.) So he would not have the kind of experience that he is having 18 Otherwise, the imagining that I undergo when I try to imagine that my friend and I are celebrating in the pub, and I have a mental image of us celebrating as a result, would be accurate just in case, as a matter of fact, I am celebrating in the pub with my friend, which does not seem right. Intuitively enough, the fact that I am not actually celebrating with my friend in the pub does not make my imagining inaccurate in any way. 19 Interestingly, the dialectic here is similar to that in some disputes about the nature of perceptual and mnemonic content. There is a view according to which the contents of our perceptual and memory experiences amount to complex propositions which concern those experiences themselves, and the causal relations in which they enter (Searle 1983). One might be similarly worried that this view makes the contents of our perceptual and memory experiences implausibly complex (Armstrong 1991, 154). The analogous response is that the relevant contents are only meant to constitute the success conditions of our perceptual and memory experiences (Searle 1991, 184). And such conditions are not always epistemically available to us through introspection.
when he visualises that the building on Catte Street is on fire. In this way, I propose, we can account for the intuition that, in Junior Fireman, the content of the subject's imagining is incorrect.
Consider, now, prediction errors. On CV, the reason why Paul is not imagining that Mia is opening her birthday present correctly, when he visualises the morning scene in the living room, is that the content of Paul's imagining is the following counterfactual: if Paul perceived that Mia is opening her birthday present, he would have the kind of experience that he is having when he visualises the scene now. But, as it happens, this counterfactual is false. If Paul saw that Mia is opening her birthday present, he would not have the kind of experience that he is having now, when he visualises the morning scene in the living room. Why is that? The reason is that if Paul saw that Mia is opening her birthday present, Mia would not appear to be thrilled with her present. And the reason for this, in turn, is that, when Paul eventually sees that Mia is opening her present, Mia does not actually look thrilled with it. In this way, I suggest, we can account for the intuition that, in Well-Intentioned Parent, the content of the subject's imagining is also incorrect.
Notice an important point in this explanation of an imagination error in Well-Intentioned Parent. The point is that since the actual world is the closest possible world in which Paul perceives that Mia is opening her present, the actual world becomes the world with regards to which the content of Paul's imagining needs to be evaluated. This feature of CV, however, is not likely to generalise to all of our imaginings which appear to be about the actual world. Suppose, for example, that I try to imagine, let us say, that the street on which I live was very broad one hundred years ago, and, as a result, I have a mental image of a street with a few houses, some trees, and just a handful of 1920s cars in it. Then, on CV, the content of my imagining is a counterfactual proposition. For what I imagine is that if I perceived, one hundred years ago, that the street on which I live is very broad, then perceiving it would be, for me, like having the mental image that I am currently having. Now, what matters, in order to evaluate that counterfactual, is whether there is a possible world in which, one hundred years ago, I perceive that my street is very broad, and whether, in the closest possible world in which I perceive it, my street looks like the street that I am currently visualising. But, in this case, it is doubtful that the closest possible world in which I perceive, one hundred years ago, that my street is very broad is the actual world. As far as I can see, the only scenario in which the world at issue might become the actual world is a scenario in which time travel is, not only possible, but also achieved in the actual world, and I take advantage of it to visit my street one hundred years ago. 20 Finally, let us consider the two intuitions about imagination which seemed to motivate the alternative views. Take, first of all, the intuition that our imaginings do not need to answer to what goes on in the actual world. On CV, our imaginings represent certain experiences which typically take place, not in the actual world, but in other possible worlds; worlds in which we are perceiving certain states of affairs. It is not surprising, then, that we feel that, usually, our imaginings are not answerable to what happens in the actual world. Consider next the intuition that we can use the same mental image to imagine different things, at different times, and in conjunction with different imaginative projects. On CV, when a subject imagines something by having some mental image, what the subject imagines is not fully determined by their mental image. What the subject imagines is that their mental image is the kind of experience that they would have if they perceived some state of affairs. And the state of affairs in question is fixed, not by the subject's mental image, but by their imaginative project. The reason why one may imagine, for example, either that there is a suitcase in a room or that there is a cat hiding behind a suitcase in a room by having the same mental image is the following. In the former case, one represents one's mental image to be the kind of experience that one would have if one perceived that there is a suitcase in a room. In the latter case, one represents one's mental image to be the kind of experience that one would have if one perceived that there is a cat hiding behind a suitcase in a room. In each case, then, the factor which determines what one is imagining, when one has one's mental image, is what possible perception one represents to be phenomenologically similar to that mental image. And it seems reasonable to think that, in the case of voluntary imaginings at least, this factor will vary depending on one's imaginative project. It is no wonder, then, that we have the intuition that one may use a single mental image to imagine different things, at different times, and in conjunction with different imaginative projects.
It seems, then, that a case can be made for a counterfactual account of the contents of some of our imaginings, namely, voluntary, propositional, sensory imaginings involving the sense modality of vision. But what about the contents of imaginings which do not fall into this narrow category? Extending CV to voluntary, propositional, sensory imaginings not involving vision seems quite straightforward. After all, a mental image, as conceived here, is simply an experience which allows us to bring to mind some entity, such as a perceivable state of affairs. But, importantly, even when the entity that we are bringing to mind by having a mental image is a perceivable state of affairs, the relevant type of perception does not need to be visual perception. Thus, if I try to imagine that a dog is barking outside, and I have an experience i of a particular barking sound as a result of my trying, then, the CV advocate may propose, what I imagine, strictly speaking, is that if I heard that a dog is barking outside, then hearing it would be, for me, like having i. The extension of CV to voluntary, sensory, nonpropositional imaginings is more challenging. It requires the assumption that we can experience (either perceptually, introspectively, or agentially), not only states of affairs, but also entities such as objects, properties, and events. If this assumption is made, however, then it seems that the CV advocate could extend the counterfactual view to, at least, some nonpropositional imaginings. The idea would be that if I try to imagine, for instance, an event such as the attack on the Twin Towers on 11/9/2001, and I have an experience i as a result, then what I imagine is that if I saw that event, then seeing it would be, for me, like having i. And if I try to imagine a property like, let us say, my joy while celebrating with my friend in the pub, and I have an experience i as a result, then what I imagine is that if I experienced joy while celebrating with my friend in the pub, then having that experience would be, for me, like having i. The most challenging extension of CV is to cases of involuntary imagination since, in such cases, there is no such thing as the subject's imaginative project. What should the CV advocate say about the case, for example, in which I am trying to concentrate on marking a philosophy paper and, out of the blue, I find myself visualising the fact that my friend and I are celebrating in the pub? I am inclined to say that, in involuntary cases such as this one, there is a mental state which plays an analogous role to that of a subject's intention to imagine something. This is the subject's interpretation of what they are visualising. In the example in which I am marking papers, for instance, my mental image comes to me out of the blue and not as a result of any intention to imagine anything. Nevertheless, I would suggest, I take it to be a mental image of the fact that my friend and I are celebrating in the pub. And this is why, the CV advocate may argue, what I imagine is that if I saw that my friend and I are celebrating in the pub, having that perceptual experience would be, for me, like having the involuntary mental image that I am having. It appears, therefore, that there are some prima facie possible extensions of CV to imaginings which fall outside the category of voluntary, propositional, sensory (and, specifically, visual) imaginings. Arguably, then, the scope of the countefactual view is not unreasonably narrow.

Conclusion
We have seen that an account of imaginative content should accommodate two kinds of error in imagination. We have considered three proposals about imaginative content, and found that none of them can account for the two kinds of error. We have also noted that, nevertheless, those views are motivated by two important intuitions about imagination. Then, we have turned our attention to a different view. The counterfactual view, we have seen, accounts for both kinds of error in imagination, and captures the two intuitions which seemed to motivate the alternative views. It seems, therefore, that the counterfactual view enjoys the virtues of alternative views of imaginative content while at the same time avoiding their challenges. I conclude, therefore, that the counterfactual view is the correct account of the contents of imagination.