1. Introduction

In work on the imagination, it is a commonplace that imagination is heterogeneous, and that we must draw numerous important distinctions in order even to discuss it—conceding at least temporarily that there is an “it” to discuss.¹ Besides a basic distinction between sensory and propositional or attitudinal imagination, we also find cross-cutting distinctions between creative and re-creative imagination, hypothetical and dramatic imagination, imagining as mental representation or as a constructive process, as well as distinctions between different kinds of imaginative use.² We must also acknowledge, independently of these distinctions, that the range of activities taken to involve imagination is simply very broad. Standard examples in the literature run from basic quasi-sensory experiences such as imagining a red patch, to evaluating remote counterfactuals, engaging with narrative art, “mindreading,” empathy, planning one’s future, make-believe, fantasy, and more.

Despite broad agreement about the fact of heterogeneity, however, its nature is underdiscussed, and philosophers differ as to its implications. For some, the heterogeneity is one of several considerations that raise doubts about whether the imagination constitutes a unified subject of investigation at all. Some, for example, take the heterogeneity to bolster the case for a reductive approach, understood as aiming to show that the various kinds of imagination should be analyzed in terms of some other type or types of mental state or activity.³ Other philosophers treat the heterogeneity as an indication that our ordinary categories related to the imagination are scientifically or theoretically unsound, and that, accordingly, we should treat the concept of imagination as a mere “folk” concept, to be replaced (if possible) by a scientifically recognized mental kind.⁴ Sometimes the heterogeneity is taken to indicate divergent conceptions of imagination, or senses of the term, rather than types or kinds of imagination; if this is correct, it may be that philosophers have incorrectly assumed they were using the same term with the same meaning, and so failed to realize they were talking past one another.⁵

Much work, on the other hand, nevertheless proceeds under the assumption that there is enough consensus for us to treat the imagination as a unified subject or domain of inquiry. Walton’s (1990) summary of the problem is often cited with approval: “What is it to imagine? We have examined a number of dimensions along which imaginings can vary; shouldn’t we now spell out what they have in common? Yes, if we can. But I can’t. Fortunately an intuitive understanding of what it is to imagine, sharpened somewhat by the observations of this chapter, is sufficient for us to proceed with our investigation”(19). Kind 2013 seems to fall in between these two views: she maintains that there is a single activity to be picked out as imagination proper (142n.), but argues that it is unable to “play all of the explanatory roles that have been assigned to it”(157), since some of the heterogeneous types have jointly incompatible properties.⁶

Thus, while most philosophers acknowledge the heterogeneity of the imagination, they disagree about whether it is substantive or superficial, and its consequences are likewise unclear. Both types of response have a reasonable claim. An unwieldy proliferation of cross-cutting distinctions naturally invites skepticism about whether there is a real kind with a natural unity. Nevertheless, many objects of investigation admit of heterogeneity in an unproblematic way, and there do appear to be genuine commonalities among the activities generally considered imaginative; so we cannot infer without some further argument that this situation presents special difficulties for imagination.

I think that we can do justice to both responses—that the heterogeneity is philosophically robust, but that there is nevertheless a natural unity despite the diversity—if we look at two prominent historical approaches. Both Aristotle and, in a different way, Hegel, describe a series of activities that substantially overlaps with the range of activities discussed by contemporary philosophers under the heading of imagination—including voluntary and involuntary experiences of mental images, certain kinds of anticipation, deliberation, fantasy, and others. For both thinkers, these activities are genuinely heterogeneous in ways I will discuss below, but they nonetheless form a series connected by one or more unifying principles. These principles are of interest both because they indicate a middle path between two reasonable but conflicting assessments of the heterogeneity, and because they hold promise for understanding how this range of activities hangs together. The bearing of historical discussions on contemporary problems is rarely straightforward, of course, and imagination is no exception: while we cannot assume that these approaches are live options, we also cannot assume that they are not.⁷ Examining them, therefore, will also allow us to assess the extent to which current thinking about imagination is continuous or discontinuous with that of past figures.

After clarifying what I take to be the main worries raised by the heterogeneity of imagination (§ 2), I present three unifying strategies based on principles we find in Aristotle’s and Hegel’s treatments of imagination (phantasia) and representation (Vorstellung), respectively (§ 3). I then argue that we can pursue a fourth strategy that uses modified versions of these principles, while bracketing some potentially controversial assumptions made in their original frameworks (§ 4).

These historical approaches allow us to give plausible responses to the contemporary challenge from heterogeneity, although there are ways in which applying them in the contemporary context is not straightforward (§ 5). I also argue, however, that there is a sharp conflict between these principles of unification and a common modern claim about the epistemic status of imagination (§ 6). It is commonly held that, unlike some other types of mental representation such as belief and perception, imagination is not “constitutively constrained by truth.”⁸ That is, even if some uses of the imagination are “instructive,” it is typically assumed that the default case is “transcendent”—instructive uses arise as special cases when the normally transcendent imagination is appropriately constrained.⁹ On any of the approaches I consider, however, it turns out that this is not the right way to draw the contrast. If I am correct, then, Aristotle and Hegel offer attractive ways of reconciling plausible contemporary claims about imagination that are in tension, but we cannot simply accept their principles without reassessing some important assumptions of our own.

2. Heterogeneity and Its Implications

Despite its prevalence, it is not always clear what exactly is being denied to imagination by the claim that it is heterogeneous, or why this heterogeneity is more problematic than what we find in other cases. To summarize a wide-ranging and complex debate, here are four main ways to make the heterogeneity challenge more precise:

(1) There is no single faculty or capacity at work in the main types of activity we count as imagination, even within the so-called sensory imagination.¹⁰ [Heterogeneous capacities]

(2) There is no single type of mental state or activity corresponding to the main types of activity we count as imaginative—imagination is not a sui generis mental kind.¹¹ [Heterogeneous states or activities]

(3) The term ‘imagination’ is ambiguous in ways that go beyond the obvious cases (‘I imagine there’ll be a recession in the next two years’), and there is no good way to separate out core or authentic senses of ‘imagine’ from extended or metaphorical ones, at least while staying close to ordinary use.¹² [Intractable ambiguity or conceptual instability]

(4) To the extent that we can agree that paradigmatic cases of imagination involve some sort of representation that is independent of present actuality, this description applies to a whole range of phenomena from pretense to supposition to, under some accounts, perception itself; and so the notion of imagination as such dissolves across the range.¹³ [Explosive intension]

(3) and (4) seem closely related, but we should distinguish them. (3) argues that ‘imagination’ does not carve mental activity at a natural joint because the elements in its extension are too varied and conceptually non-uniform, but it still allows that we might agree on the term’s extension, and it may even have sharp boundaries. (4), by contrast, states that ‘imagination’ does not carve at a joint because its extension seamlessly blends into other activities from which we wrongly supposed we could distinguish it—and so the term, which may even have something like a generally agreed-upon intension, turns out to have no clear limits on its extension.

Underlying the four challenges above, I think, is an assumption that if imagination is a unified type or kind, picked out unambiguously by a given term, then we should be able to treat it the way we treat paradigmatic natural kinds—e.g., by specifying the essential feature or cluster of features that determine the kind, and which all instances of it share in the same way.¹⁴ Or, at least, we ought to be able to specify something of this sort, such that everything in the term’s extension essentially involves it. According to any of the four types of challenge raised above, however, imagination does not fit standard models for treating natural kinds: there is no one specification that can answer to all relevant uses—nothing, for example, plays the role for ‘imagination’ that H₂O plays in relation to ‘water’—nor is there a further natural kind agreed to be involved in all cases.¹⁵

We should notice, however, that these challenges point to very different sorts of failures to constitute a natural kind. (1) and (2) assert that, for the different activities we generally agree to be imaginative, there is in fact no underlying unity that can be framed in terms of established categories in psychology, cognitive science, and the philosophy of mind, regardless of whether we have a unified conception of what we intend to be discussing when we talk about imagination. (3), by contrast, attacks something more conceptually basic: it asserts that, upon reflection, the kinds of activity we are considering under a common term do not really exhibit conceptual unity at all, and we do not need information from psychology or cognitive science to see this—ordinary linguistic or conceptual competence is sufficient. Challenge (4) may be posed either way, depending on the reasons offered for thinking that the intension fails to establish a boundary.

For reasons that will become apparent, I think that these differences indicate an important aspect of the problem, but it is not clear how to frame them. One salient feature is that they appear to correspond to different epistemic stages: challenges (1) and (2) depend on commitments from comparatively advanced specialist investigations in psychology, cognitive science, or philosophy of mind, whereas challenge (3), as I have described it, can be formulated and argued independently of such commitments. Because I mean to be neutral as to whether these more specialized commitments are based on empirical or philosophical investigation (or both), I will use the labels ‘pre-theoretical’ and ‘theoretical’ to indicate the contrast between earlier and later stages of investigation.

Challenges at these different stages can be made independently of one another. Thus, we might argue that we do not have a unified pre-theoretical notion of imagination, while acknowledging that we might still discover an underlying unity to these various activities at a more advanced stage of investigation. Conversely, we might affirm or leave it open whether we have a unified pre-theoretical notion of imagination, but argue that, in any case, there is no underlying unity to these kinds of activity once we understand them more accurately. (This latter might be the diagnosis where a term turns out to be a mere “folk” concept, but it is compatible with other situations as well.)

The distinction between pre-theoretical and theoretical cannot be perfectly sharp, of course, and it is a relative distinction rather than an absolute one: that is, a stage Y may be pre-theoretical relative to stage Z, but theoretical relative to stage X (this point will be relevant in § 5 below). With these caveats, however, I think that the distinction is useful for characterizing important differences between the challenges and the available responses.

Thus, we can distinguish four separate challenges that may be raised on the basis of heterogeneity, and we must be careful about whether they derive from claims at more advanced or less advanced stages of investigation.

3. Historical Precedents for Unification under Heterogeneity

I have suggested that we should take seriously the thought that the heterogeneity of imagination is substantive rather than superficial, but that this result need not threaten its unity as a subject of investigation. If so, a good account should find a unifying principle without making the heterogeneity disappear altogether.

The first three strategies I wish to consider can be seen especially clearly by looking at the work of Aristotle and Hegel. Both thinkers discuss and group together arrays of distinct mental activities, and their categorizations overlap substantially with contemporary notions of imaginative activity, as I have noted. And they do so without taking them all to issue from a distinct faculty of imagination, or to involve a sui generis mental state. There are important differences in the ranges of activity they describe, but they include, as central cases, paradigmatically imaginative activities such as the quasi-perceptual representation of non-present or unreal particulars, and representations of possible or future actions and states of affairs. Thus, even though they are not attempting to solve a heterogeneity problem, or to present a theory of imagination in the contemporary mode, they still approach this domain in a way that answers to my desiderata. For present purposes, I will therefore set aside questions about whether and to what extent their terms or concepts line up with ours, although I will return to this question below (§ 5).¹⁶ Indeed, I take it to be an encouraging sign that both thinkers consider this range of mental activities to be unified, but not as instances under a common term, since this suggests that the hope for unifying them can be dissociated from contingent features of terminology.

Common to both thinkers is a guiding idea that we should note at the outset: both take it that perception (or, for Hegel, “intuition”) involves a cognitive relationship with present particulars, but that the content of such experience can be retained and redeployed. Both likewise think that, in human beings, this retained content is transformed through a series of stages, until it results in an ability to think universally and abstractly. Thus, both take it that there is a stretch of cognitive terrain beyond occurrent perception of particulars, but falling short of discursive thought with universals or concepts, and that there is a series of distinct activities to bridge this gap. The activities in this series constitute the domain of phantasia for Aristotle and Vorstellung for Hegel.¹⁷ Both thinkers, however, also attribute further, more specific types of unity-in-diversity that connect the elements of that series to each other, and we should distinguish these.

Aristotle draws two main conclusions about phantasia in De Anima III 3, the primary chapter in which he discusses it directly.¹⁸ He argues first that phantasia is not identical to one of the main “critical” (i.e., judging or discerning) capacities that distinguish what is true or false: it is not opinion (doxa), knowledge (epistêmê), perception (aisthêsis), or thought/intellect (nous) (428a5–b10).¹⁹ Instead, he argues, phantasia is a type of representation constituted by a secondary motion or change (kinêsis) in our perceptual faculties, and which depends on the primary activity of the perceptual faculty both causally and for its content (428b10–429a2). There are three main types of perceptual content, he claims, and accordingly, three main types of content for phantasia (428b17–30): (1) “proper” or per se perceptibles to which the individual sense organs are sensitive, such as color for vision (roughly, secondary qualities in later traditions), (2) “common” sensibles such as magnitude or shape, which are perceived by multiple sense modalities (roughly, primary qualities in later traditions), and (3) so-called “co-incidental” perceptibles. The latter category is broad and somewhat unclear, but it seems to include much of what ordinary perception actually does: Aristotle thinks that we perceive things such as ants and houses and people, which are not per se perceptibles (like colors), but are nonetheless genuine perceptibles in that such kinds and properties are included in the contents of our ordinary perceptual states themselves.²⁰ Both perceptual and imaginative states therefore have relatively robust content of three different types.

Thus, on Aristotle’s approach, imagination lacks unity in ways corresponding to the challenges listed above: it is not a single faculty in its own right, nor a single type of activity or state, nor do imaginings have a single type of content. Nor, however, are imaginings reducible to or identifiable with some other type of state, such as belief. Although not itself a faculty, phantasia is parasitic on a single faculty that is unified—the perceptual faculty (aisthêtikon)—and on its activities in two ways: it exhibits (1) causal dependence, as being a type of secondary change following upon perceptual activity, and (2) content dependence, since phantasiai derive their three content types from those available to perception.²¹ Thus, despite all these sources of multiplicity, it makes sense to treat phantasia as a unified series of activities—distinct from both perception, on which it depends, and judgment, which depends on it.

So much for Aristotle’s most general characterization of phantasia in De Anima III 3. The work of phantasia, however, is broad: elsewhere, when Aristotle describes our cognitive progress, he starts with perception and ends with intellectual achievements such as craft (technê), knowledge (epistêmê), and intellect (nous). In Metaphysics I 1, the middle stages include especially memory (mnêmê) and what he calls “experience” (empeiria), the final cognitive stage prior to grasping universals. The word ‘experience’ here does not really mean subjective experience: Aristotle’s example is of a worker who is able to relate present situations to comparable ones from past encounters and so act appropriately, but who is nonetheless not an expert in the full sense.²² Memory and experience, as we learn elsewhere, are the work of phantasia, since they involve reactivating and comparing or combining retained movements from the perception of particulars, building up to the capacity for grasping and deploying a universal, which is the work of thought.²³ Indeed, everything lying beyond perception but still relating to particulars—thus falling short of intellectual, discursive thinking by means of universals—is, he thinks, the domain of phantasia. This includes dreaming, expectation, and a mode of presentation of the particular objects and aims of action, which both generates the kind of affective engagement that leads one to pursue or to avoid a certain course of action, and also aids the cognitive work of figuring out how to execute a plan or satisfy ultimate ends.²⁴ Notoriously, although the exact point is controversial, he also thinks that discursive thought itself requires images.²⁵

The functions of phantasia are therefore diverse, but are crucial for building up from perception to anything that counts as full rationality in human action and thought. Nevertheless, Aristotle thinks that many animals with no capacity for thought nonetheless have phantasia, and so he cannot say that serving these higher cognitive activities is the essential function of phantasia across all the animals possessing it. So while Aristotle’s approach yields an interesting type of cluster-unity in terms of phantasia’s causal origin and the source of its content, phantasia is not unified with respect to any particular function it subserves.

Hegel’s treatment of the stages of “representation” (Vorstellung) in the Philosophy of Mind has essentially the same external boundaries as Aristotle’s operations of phantasia—it lies between “intuition” and thought (Anschauung and Denken)—as well as some other important similarities. On his account, though, the stages do exhibit a further sort of functional unity. Representation has three stages: “recollection” (Errinerung), “imagination” (Einbildungskraft), and “memory”(Gedächtnis). Although he uses ‘imagination’ (Einbildungskraft) for the middle stage only, the whole span of “representation” is a good match for the range of activities Aristotle takes to be the work of phantasia, and again, to a large extent, for the activities contemporary theorists include under ‘imagination’.²⁶ ‘Recollection’, the first stage of representation, is “the involuntary arousal of a content that is already ours”(§ 451, 185), the content of which is the same as in intuition. The last stage, memory, culminates in an explicitly symbolic stage in which one can retrieve mental contents using only words. It is a stage we reach when the imagistic content of what began in perception has been stripped out.²⁷ It is the last stage before passing over into the first stage of thought (Verstand, understanding).

Imagination proper, the middle stage, has three stages of its own: “reproductive imagination,” “associative imagination,” and “symbolizing and sign-making fantasy”(Phantasie). The first stage involves retrieval of the content of intuition, but now under voluntary control, unlike recollection; the second creates subjective links between images; and the third brings out the universal content present in the subjectively-related images.²⁸

The details of this account are complex and rich, and at certain points obscure. What matters most here is that these stages are ordered in such a way as to bring about the components underlying abstract thought—“signs” that have been progressively freed from perceptual or “intuitive” content by the work of intelligence. Given that Hegel takes these stages as in some manner developments along the way to conceptual thought, and ultimately to self-conscious rationality, it would, minimally, be wrong to say that the stages merely happen to result in discursive thinking stripped of imagistic content. They are rather parts of a whole process of development, and so are intrinsically connected to what results.

This is therefore a teleological order, in the sense that these activities proceed as they do for the sake of the end that results, and are thus partially explained by that end, even if, in various contexts, they can be engaged in independently. Furthermore, I take it that part of what is essential to the proposal is that the explanatory end is external to the process of development: none of these stages of Representation is a thought in Hegel’s sense, but they—or the sequences as a whole—are the way they are at least partly for the sake of thought. Thus, where Aristotle presents phantasia as a causal result of perception without implying that it is essentially for the sake of any specific further function, Hegel, while conceiving the process itself somewhat differently, also adds to this array of imaginative activities a much stronger teleological unity.²⁹

4. An Extension

Aristotle and Hegel thus give us three ways to unify imaginative activities along two different axes. First, along a source-based axis, they both hold that these activities ultimately derive both (1) causally and (2) for their content from a distinct capacity (perception) which is itself unified. On Hegel’s view, there is also (3) an extrinsic end, such that some forms of imagination occur the way they do because of the contribution they make to that further activity. These approaches do not appear to be mutually inconsistent, and Hegel indeed embraces all three.

To these I would add a fourth possibility: we may apply either a source-based or a teleological criterion to distinguish some uses or functions within the range as primary, and some as secondary. The cluster as a whole may then be unified around these primary imaginative functions, but without appealing to an external source or end from outside the series. This approach would allow us to develop the basic insights of these approaches, but without relying on anachronistic or at least controversial assumptions about a hierarchy of cognitive states, and about imagination’s relationship to other elements in that hierarchy.

There are a variety of ways to apply such a criterion. One natural thought is to order types of imagination by reference to their types of content, and then argue that one or another content type is more basic than the others. We might think that the most basic kind of content is simple, unimodal sensory content, for example, but we need not, if we do not think that re-experiencing sensory qualities is the most basic kind of imaginative content. Cases could instead be made for giving priority to representations of objects, situations, manifolds, or propositional contents. Alternatively, if there is empirical evidence for thinking that some mechanisms used in imagining are causally more basic than others, then that evidence would give us reason for taking the causally most basic kinds as primary.³⁰ Both approaches would amount to using a source-based criterion, but again, would take the source from among the range of imaginative activities, rather than from outside it.³¹

More interestingly, perhaps, we can also draw primary/secondary distinctions using the teleological axis, which allows us to employ teleological principles weaker than Hegel’s, while still assigning priority differently compared with the source-based axis. Indeed, one need not have a vision as grand as Hegel’s to argue that there are good reasons to give priority to reasonably sophisticated types of imagination, rather than to the one with the simplest content or mechanism.

Consider the range of the most prominent examples of imaginative activity: representing something one has perceived; creating a mental image of something one has not perceived; representing or anticipating someone else’s action or feeling (mindreading); judging a counterfactual physical situation; engaging with or creating a fictional narrative; engaging in a sustained act of make-believe. Setting aside questions about their contents and sources, these imaginative activities play vastly different roles in the lives of the creatures that engage in them.

The roles these activities play in the lives of the creatures deploying them can in turn provide a basis for distinguishing primary from secondary types of imagination. Thus, for example, if some forms of imagination are consistently used for securing basic needs, such as food or safety, or for planning simple actions, then we could arguably assign priority to these functions over other forms such as make-believe on the grounds that they relate to more fundamental requirements of the living being. From this perspective, priority might go to activities like rehearsing an action, representing an object from alternate perspectives, or consulting and adjusting a mental map one has of oneself in one’s surroundings. (By this last example I mean the way occurrent perceptions are integrated into an overall representation of one’s present situation, especially in relation to other locations: e.g., while reading a book I remain aware of the walls behind me, the children asleep in the next room, the empty downstairs, and so on.) These activities are utterly pervasive in performing the basic tasks of everyday life, including in navigating to the location of some remembered object. One could, of course, argue that other criteria besides securing basic needs deserve priority. Evaluating priority claims is a separate task in its own right, and parallel questions arise with any attempt to find an organizing structure for a range of activities.³² My only claim is that, however we distinguish functionally primary from functionally secondary activities, it is reasonable to think that we can apply similar criteria to the entire range of imaginative activities.

The teleology involved on this approach is modest, since we need only assume that if certain actions in the life of a creature depend on imaginings of a certain sort, then those actions provide objective standards by which we can evaluate and order imaginative activity. Nor is there any need to suppose that these types of activity form a neatly ordered array as they do on the robust Hegelian teleological model. Indeed, which activities turn out to be primary or secondary, and in what sense, would plausibly be a contingent matter, and in principle open to some variation across species, over time, across cultures, and even between individuals.³³

We can therefore use versions of Aristotle’s and Hegel’s principles to draw distinctions between primary and secondary functions of imagination, without committing ourselves to the whole sweep and sequence of the way they think these activities run from perception to thought and knowledge. This is not to say that the Aristotelian and Hegelian approaches themselves are non-starters, either: some version of a causal and content dependency thesis might also be defensible, and so might a claim that our human imaginative capacities are the way they are, or develop the way they do, because of what they contribute to more abstract forms of thinking and planning.³⁴

As a general matter, all four approaches share the view that what unifies all these activities is that they jointly contribute to a capacity—or rather to a series of increasingly sophisticated capacities—for integrating past experiences in ways that allow them to inform or influence other practical and cognitive abilities. Collectively, they allow us to form useful and meaningful representations of our past, a holistic grasp of our present, and detailed intentions to act a certain way in the future, along with a reasonably accurate understanding of how different courses of action would unfold. If we accept the basic principles of these approaches, the question becomes not whether this heterogeneous range of activities has a natural unity to it, but rather how to sort out the several unifying factors that all seem to apply.

5. Applying These Principles to the Challenges from Heterogeneity

If one or more of these strategies is open to us, we can accept that the heterogeneity is substantive but nevertheless insist that the imagination is a reasonably unified object of investigation. To evaluate their consequences for the heterogeneity question and responses to it, such as reductionism, we can return to the four challenges from before: (1) No single faculty at work [Heterogeneous capacities]; (2) No single type of mental state [Heterogeneous states or activities]; (3) Ambiguity [Intractable ambiguity or conceptual instability]; (4) Failure to distinguish the target from other supposedly distinct kinds [Explosive intension].

On any of the forms of unity described, we would have reason to deny the force of (4): the fact that ‘imagination’ picks out a range running from perception to intellectual fantasy is no source of difficulty, as long as that range exhibits some kind of order or primary/secondary distinction.

We could accept (3) but only with qualification. Even if there is ambiguity, it is not strict (i.e., not like ‘cape’ applied to clothing and geographical features), nor are the different senses just a loose family. Even if we are not initially aware of it, the types of activity we designate by ‘imagination’ can be theoretically understood in ways that exhibit a kind of order that ultimately depends on a core use or uses, such as the ability to retain and reactivate aspects of experience.

The Aristotelian and Hegelian would also ask that (1) be qualified. On this approach there is no distinct faculty or capacity of imagination—but there is nevertheless a unified capacity on which all these states causally depend: the perceptual system. If the basic aim of reductionism is to explain paradigmatic imaginative activities in terms of other mental kinds like belief and perception, then the Aristotelian approach is compatible with this aim; but the resulting explanations are non-reductive, since imaginative activities retain their distinctive natures—they are not instances of perception.³⁵ Likewise, both would qualify (2): there is no sui generis state or event type common to all these designated activities. Yet they all ultimately depend on some one initial kind of mental state, however remote or “sublated” it becomes as we get more sophisticated.

On this approach, moreover, it makes sense that secondary uses such as daydreaming or following a fictional narrative—assuming for the moment that they are indeed secondary—would retain some properties of the primary forms of imagination while lacking others, or while acquiring new properties that the primary kinds cannot have. So, for example, basic forms of anticipation require a certain level of accuracy about the standard ways things and systems behave over time, while fantasy and make-believe do not require the same level of accuracy. Indeed, an ability to suspend some normal tendencies of things, while holding others fixed, is essential for creative thinking. Even if the secondary activity, that is, depends on initially having the ability to engage in the primary activity, it can also cancel or bracket certain of its features as well. This allows us to answer Kind’s (2013) worry about heterogeneous types with incompatible properties, mentioned above.³⁶

Nevertheless, there are ways in which applying these principles in the context of recent debates about imagination is not straightforward. First, there is a methodological difference with respect to the initial, pre-theoretical “givens” about imagination. Many modern approaches to imagination start with a mixture of ordinary usage and “folk-psychological” categories: e.g., sorting through the main uses of ‘imagine’ with paradigmatic examples, separating out metaphorical uses and outliers, and considering where we intuitively locate imagination relative to things like belief and desire.³⁷ Alternatively, phenomenological approaches aim to specify the key features characterizing the experience of the imagining subject, again as the starting point for a more sophisticated philosophical or scientific analysis.³⁸ Much contemporary work in fact combines the two, and starts with a mixture of commonsense, “folk-psychological,” and phenomenological claims about paradigm cases.

It is important to see that, for Aristotle and Hegel, the starting points are very different. For them, the topic of imagination arises within a framework for discussing the mind that already makes substantial theoretical commitments.³⁹ In Aristotle’s case, there is no real attempt to refer to “common beliefs” about imagination—he is working within the hylomorphic framework for studying the mind and according to the epistemological commitments he has already developed elsewhere, and he is responding to what look like Platonic claims in the background. Hegel’s account is also embedded in a broad theoretical sweep, and appears against a background that assumes roughly the same types of faculties Kant could assume—a range starting with “intuition” and running through understanding and reason. Thus, for Aristotle and Hegel, the pre-theoretical “appearances” about imagination largely come from prior theoretical commitments about the mind and about knowledge. They do not ignore non-expert opinions about the kinds of phenomena involving phantasia and Vorstellung, but such opinions do not constitute their starting point for considering them. As a consequence, information about common opinions, ordinary language, and subjective reports also constitutes for them a far looser constraint than for most contemporary theorists.

Conversely, then, their approaches may not fit well with some of the basic modern assumptions that do derive from common opinions, “folk-psychology,” and so on. For example, most contemporary approaches start with a version of the basic distinction between sensory and propositional imagination. The standard examples accordingly tend to fall into two clusters that neatly fit the distinction: cases with simple, unimodal sensory contents, such as imagining how a couch looks or a red patch, on the one hand; and attitudes toward evidently fictional or non-actual states of affairs, such as considering counterfactuals or acts of make-believe, on the other.

Both the basic distinction and the standard examples flowing from it are an awkward fit with the approaches I have described. On the teleological approach described above (and even some of the non-teleological ones), the functionally primary cases are activities such as rehearsing an action, anticipating someone’s experience, or consulting and adjusting the mental map one has of oneself in one’s surroundings. These examples do not fall cleanly on either side of the distinction between sensory and propositional/attitudinal imagination. Many of them, rather, are cases of what is often called “experiential imagination,” which differ in important ways from the examples that tend to illustrate the distinction most clearly. Experiential imaginings, as standardly understood, have contents corresponding to whole experiences, such as climbing a tree, swearing an oath, or being on a mountaintop.⁴⁰ Their contents are typically more complex than imaginings of unimodal perceptual contents, since they are often multi-modal, as when for example one imagines oneself performing at a concert; they tend to include one or more perspectives, such as first-person/“field” or third-person/“observer” perspectives, often both; and they are spatially and/or temporally extended—one imagines the action, with its beginning, middle, and end. These cases are also less accessible to third-person observation than many paradigm acts of propositional imagining, such as pretend play, and their subjective features are supposed to resemble or simulate the experiences they represent in roughly the way sensory imaginings do—again, by contrast with standard examples of propositional imagination.⁴¹ Furthermore, of course, they do not all involve considering their contents as fictional or non-actual.

These teleological approaches, then, yield a mismatch between the cases that are functionally primary, and the kinds of example used as paradigms of imagination in contemporary discussion. According to the teleological conception, the standard modern paradigm cases are functionally peripheral, while the teleologically central cases are at best non-paradigmatic instances of sensory or propositional imagination, if they can be rightly considered either.

A further, taxonomical worry arises on any of these approaches, with respect to the standard distinction between sensory and propositional imagination itself. It will perhaps seem obvious that these principles for unifying activities, as I have described them, could not apply to propositional imagination, but at most to only so-called sensory and experiential imagination. It is not obvious, though, and whether it is correct will depend on some of the details about how we characterize these different types of activity. It is true, however, that the guiding thesis about extrinsic unity shared by Aristotle and Hegel, and by much of the tradition in between, makes it unnatural to take the sensory/propositional distinction to mark a natural joint, if the range between perception and abstract thinking includes elements on both sides of the modern divide (as seems likely). This may be a bug or a feature, depending on how one feels about the distinction.⁴²

Thus, there are several ways in which these historical approaches cannot be straightforwardly applied to the contemporary debate. However, we should not exaggerate their importance: all of them pertain to contemporary assumptions that can be challenged on independent grounds. Indeed, it is significant that, as I have noted, there are controversies among contemporary philosophers over whether memory and supposition, specifically, should be treated as cases of imagination. These activities would fall near the boundaries of the Aristotelian-Hegelian series, and so from that perspective, such debates are unsurprising. Perhaps, then, we share some underlying assumptions about the range of this domain after all.

6. Implications for Assumptions of Epistemic Asymmetry

On the whole, the problems above do not seem insurmountable, but they do cast light on some differences between contemporary and historical assumptions about method, paradigm examples, and the basic taxonomy of imagination.

There is a sharper conflict, however, between these approaches to unity and the common claim that there is an epistemic asymmetry between imagination and other forms of representation, like perception and thinking—that imagination, unlike these other forms of representation, is not “constitutively constrained by truth.”⁴³ This contrast is often advanced as a way of getting at the basic features of the subject matter: imagination generally represents what is not present or actual, and this is usually no defect; nor does falsehood entail that someone is violating any norms of imagining. As many have argued, imagination nonetheless has epistemic value, and may even ground or justify beliefs or inferences.⁴⁴ Typically, however, this value is presented as depending on special constraints or circumstances; the default presumption is that our imagining things as being a certain way is no indication that things are or will be that way.⁴⁵

This view about the default epistemic status of imagining fits with the view that imagination is open to a loose kind of unity at best. For if all of the agreed-upon types of imagining are of equal status, and only some of them are subject to a norm of truth or accuracy, then such norms can only be incidental to the class of imaginings taken as a whole.

However, on any of the approaches considered above, the situation is almost reversed. Before explaining why, we should note two features of the claim that imagination is not “constitutively constrained by truth.” First, I take it that the claim pertains to whether there are norms applying to these representations, such that they are generally supposed to get things right. (‘Constitutively’ cannot here mean ‘necessarily’, unless we think that false or evidence-insensitive beliefs are not really beliefs.) Second, I take it that accuracy, rather than truth, is the more appropriate concept for these norms, since if truth is a property of statements or propositions, non-discursive (e.g., sensory) imagination and perception are immediately ruled out anyway.

It is clear enough that for the teleological types of unification—types (3) and (4) above—the asymmetry does not hold. If the primary functions of imagination have to do with things like retaining information from experience, planning basic actions, keeping track of one’s surroundings, or with grasping general concepts or regularities on the basis of comparing particular experiences, then norms of accuracy would indeed apply to the primary types of imagination, even if there are perfectly good secondary uses of the same capacities for which those norms are suspended. Imagination would still arguably represent what is not present or actual or in immediate contact with the senses, whether temporally or spatially or both; but this can be accepted while also claiming that the primary imaginative activities are constitutively subject to norms of accuracy. Nor are these activities generally “quarantined” from belief—quite the opposite.

What is perhaps surprising is that it takes only a weak unifying assumption to remove the presumption of an epistemic asymmetry, even on the non-teleological strategies. Suppose we accept just a weak version of (Aristotelian) causal unification, for example, such that the various forms of imaginative activity causally depend on the basic ability to retain and reactivate the contents of perceptual experiences. This unity is compatible with the claim that there are no norms of accuracy applying to these imaginative activities—but only if that basic capacity is not itself a necessary component of other life activities to which such norms do apply, such as memory, recognition, environmental mapping, and planning simple and complex actions. If it does subserve such activities, then at least some norms of accuracy also apply derivatively to the basic capacity for retention and reactivation. There remain, of course, many paradigmatic examples of imagination in which the norms of accuracy are suspended, but under this form of unification, those norm-free modes are the special cases, relative to the core uses. Even on this weak unification, then, the epistemic value of imagination does not depend on special conditions or constraints.⁴⁶

The asymmetry disappears, of course, because a weak teleological notion of a well-functioning capacity turns out to apply indirectly to causally basic forms of imagination, in virtue of their essential role in executing other activities for which success or failure depends on truth and accuracy in representation. If it is truly a matter of indifference whether we accurately retain the content of experience, then these norms do not apply, but this is a very strong claim.

One could argue that the epistemic asymmetry still holds in a stronger way, because the connection between imagination and these other activities is not just derivative but somehow merely regular or co-incidental. But this requires argument, and we cannot say that the transcendent uses should be given default status simply because only some kinds of imagination are subject to accuracy norms, or because many common forms of imagining are not.

Perhaps one way to put the point is that, on these approaches, our basic abilities to imagine are indeed subject to norms of truth and accuracy—they may even be constitutively constrained by them—even if only some of the ways in which those abilities are exercised are likewise so subject.

7. Conclusion

As I have described them, these possible sources of unification can be embraced independently or jointly. Besides the weak form of extrinsic unity for the range of activities—the boundaries of perception and abstract thought—one might add (1) unification by causal source, if we think that most forms of imaginative activity are causally downstream from perception; and (2) unification by source of content, if they depend on it for their contents. One might further add (3) a teleological thesis, if we think (with Hegel) that some form of psychological activity beyond imagination, such as thought or intention, is ultimately subserved by it, and that this activity has explanatory value for understanding imagination itself. Finally, (4) we might use their basic principles to draw a primary/secondary distinction among imaginative activities themselves, and unify the group around the primary ones. These forms of unity, I have argued, are in principle live options. They have implications for reductive and revisionist projects, as well as for some common assumptions philosophers make in discussing imagination; furthermore, most of them conflict with the alleged epistemic asymmetry between imagination and other forms of representation.

Underlying all of these approaches is an attractive idea with a long history: that the unifying framework for thinking about imagination treats it as lying at or constituting the interface, in one or more ways, between our most basic perceptual engagements with the particulars around us, on the one hand, and our practical and cognitive abilities to act reflectively and to think discursively or universally, on the other. This framework, I suggest, remains a viable way for thinking about the unity of the various activities that we continue to associate with the imagination, and their connection to one another. There are various ways the framework might be developed, although we cannot simply adopt it without modifying other assumptions about imagination. If none of these approaches is open to us, however, it would be instructive to learn why.⁴⁷