2. Clarifying the explanandum

What kind of imaginative failure are we talking about when we are talking about IR? I agree with Brian Weatherson that we should distinguish between alethic, phenomenological, imaginative and aesthetic puzzles (Weatherson Reference Weatherson2004). The alethic puzzle is this: why cannot propositions like ˂Giselda did the right thing [by killing her baby]˃ (KB) and ˂Craig did the right thing [by shooting Jack and Jill]˃ (DF)) be true in fiction (assuming that they cannot be, of course)? It seems that the authority of authors with respect to what is true in fiction breaks down in the case of particular moral propositions.Footnote ⁴ The phenomenological puzzle is about explaining the feeling of resistance that we have when reading propositions like KB or DF. The imaginative puzzle concerns our inability to imagine that such propositions could be true. Finally, the aesthetic puzzle is about the question of why such propositions cause negative evaluation of the fictions in which they figure.

In this paper, I am primarily concerned with the imaginative puzzle, explanation of which should be psychological in nature, appealing to the limitations of our imaginative and representational capacities. Naturally, the answer to the imaginative puzzle should also say something in response to the phenomenological puzzle. It is plausible that the feeling of resistance arises from the fact that our psychological faculties face an obstacle to their normal functioning. It is unclear whether the solution to the imaginative puzzle has any implications for how to solve alethic or aesthetic puzzles, and I can be agnostic about this.

One might also ask what is it about resistance-causing propositions that makes them different from propositions that do not cause resistance. Thus far we have only considered two examples, KB and DF. Is there any common characteristic to propositions that cause IR? At this point I want to leave this question open and assume that we can fix the phenomenon of interest by simply pointing out paradigm cases. A claim that there are cases of IR seems to be less contentious than any characterization of the common core of propositions that cause it. For instance, while one could ask whether propositions that cause IR are such that we believe them to be false only in the actual world or such that we believe them to be necessarily false, I find it doubtful whether we can give a definite answer to this. What is more, since I want to leave open the possibility of moral anti-realism, I am not even sure if we should give an answer in the present context.

Having thus stated that the primary puzzle concerns the question of why we are unable to imagine propositions like KB and DF, a more basic question arises: what is this imagination we are talking about when we talk about IR? The first suggestion is that imagining that p is an activity that involves taking p to be true in fiction, or that it involves taking the situation represented by p to obtain in the story-world. An uncontroversial case of imagining could then include taking Sherlock Holmes to live on the Baker street when reading Conan Doyleʼs stories, for instance. That being said, in the literature on IR and on imagination more generally, people distinguish between (engaged) imagination and supposition (Arcangeli Reference Arcangeli2014; Currie and Ravenscroft Reference Currie and Ravenscroft2002; Doggett and Egan Reference Doggett and Egan2007). What both of these mental activities have in common is that they involve taking certain situations to obtain in non-actual world. There are differences, however, when it comes to IR. We can presumable consider any comprehensible proposition as true (i.e. suppose that it is true) in order to see what follows from it. Supposition cannot be the type of activity that is involved in causing IR, because we can suppose for the sake of the argument, for instance, that a proposition is true (Gendler Reference Gendler and Nichols2006). There must be a thicker notion of imagination which seems to fail in the case of IR. This intuition has been expressed by Gendler by saying that when we imagine we participate in the imagined world (Gendler Reference Gendler2000, 80). Another way to put the difference between engaged imagination and suppositionFootnote ⁵ is that while IR arises only during engaged reading of fiction, there is no problem with assuming the truth of moral propositions when we detach ourselves from the narrative (Stueber Reference Stueber2011, 161). Fleshing out the notion of engaged imagination has proven to be difficult, however.

It should be helpful to bring out characteristic features that make engaged imagination distinct from supposition. Weinberg and Meskin, for instance, bring out various markers for distinguishing between the two in our folk practice: (1) imagining allows for embellishment of its content, p, while supposing only follows the consequences of p; (2) the truth of almost any proposition can be supposed, but not imagined; (3) the typical epistemic role of supposing is in hypothetical and reductio arguments, while the main epistemic role of imagining is in demonstrations of possibility; (4) supposing is almost always used for epistemic purposes, while imaginings have various purposes (Weinberg and Meskin Reference Weinberg, Meskin and Nichols2006, 193). They argue on the basis of this list that these markers point towards the fact that imagining and supposing are activities which involve different cognitive processes. For instance, imagining, unlike supposing, engages our affect systems and allows for open elaboration (195).

Alvin Goldman maps the distinction between imagination and supposition onto enactment- and supposition-imagination (E-imagination and S-imagination), respectively: while enactment-imagination recreates or enacts a selected mental state, supposition-imagination simply entertains the truth of a proposition for the sake of hypothetical reasoning (Goldman Reference Goldman2006, 47). Margherita Arcangeli brings out in addition that imagination, as opposed to supposition, is not entirely under our voluntary control (Arcangeli Reference Arcangeli2014, 615). Finally, Brian Weatherson distinguishes imaginings from supposings by taking the content of the first to be fine-grained and detailed. For instance, if we imagine chair in an engaged manner, we imagine a specific kind of chair, not simply a chair as such (Weatherson Reference Weatherson2004, 20).Footnote ⁶

From this brief overview we can already see that the question of what is the distinction between imaginings and supposings has not received a uniform answer. It is not even clear whether imagining and supposings constitute two distinct types of mental activity. Weinberg and Meskin, for instance, are willing to admit that there are in-between cases (Weinberg and Meskin Reference Weinberg, Meskin and Nichols2006, 197); while Goldman thinks that S-imagination can in principle be reduced to E-imagination (Goldman Reference Goldman2006, 48). This does not mean, however, that the distinction cannot be useful. In fact, in order to make sense of IR, we need such a distinction because, as already noted, we can suppose that propositions like KB and DF are true and do not face any resistance.

Imagining that we are seemingly unable to perform in the case of IR-causing propositions, then, is a mental activity which involves considering the situation described by KB and DF to obtain in the story-world in an engaged (not merely suppositional) manner. As a shared intuition, we can assume that imaginings, as supposed to supposings, are in an important sense participatory. From Weinberg and Meskin, we can take on board the relatively uncontentious features of that kind of imagining: it allows for embellishments on the imagined proposition and disposes us to affective responses. From Arcangeli, we can take the idea that supposings are more fully under our voluntary control. We can also agree with Weatherson that the content of imaginings is filled with detail.Footnote ⁷ Admittedly, this is quite vague. In the next section, I will argue how to flesh out the notion of engaged imagining more fully.

The question of IR in this paper is, then: why are we seemingly unable to imagine, as opposed to suppose, that in the world of the respective stories, Giselda and Craig did the right thing, while, in contrast, we do not have any difficulties of imagining that other things happened in the story (Giselda killing her baby, Craig killing Jack and Jill)? To put it in more general terms: why cannot we imagine the truth of some moral propositions (and possibly some other types) in story-worlds, while we do not have difficulty supposing the truth of even the most ridiculous non-moral propositions? To answer this question, we have to explain the difference between engaged imagination and mere supposition, and why one is conducive to IR while another is not.

3. Engaged imagination and mental imagery

Both engaged imaginings and supposings have been taken to be activities that relate agents to propositions: by imagining or supposing one assumes the truth of a proposition in a specific manner. Under this view, the difference between imagining and supposing seems to boil down to the difference in their functional role, not content, which can be the same for both. There is also an intuition, however, brought out by Weatherson, that imaginings are more fine-grained and detailed than supposings, and this seems to be a difference in content. In this section I am going to argue that this intuition about the difference in content allows us to explain the imagination-supposition distinction and the characteristic features that engaged imaginings are taken to have.

One way to think about the fine-grainedness of content is to use a perceptual analogy. It seems that there is a difference between merely thinking about something and actually perceiving it. The latter is more detailed than the first.Footnote ⁸ For instance, if I think that there is an apple under a tablecloth, I token a thought about the property of being an apple and that of being a tablecloth, but when I see, taste, touch and/or smell the apple, I am presented with an abundance of fine-grained properties, unlike the properties I was merely thinking about. As it is one thing to suppose that something is the case and another to genuinely imagine it, it is one thing to think about something and another to perceive it. Can we use this analogy to throw some light on the supposition-imagination distinction? I think we can, because there is an imaginative counterpart to perception, namely, mental imagery.

Mental imagery is the redeployment of our perceptual capacities endogenously, without a perceptual stimulus being present, in order to create mental images which have some of the characteristic properties that the corresponding percepts have (Kosslyn, Thompson, and Ganis Reference Kosslyn, Thompson and Ganis2006, 4). In other words, mental imagery is simulated perception.Footnote ⁹ It is noticeable that in their discussions of the difference between imagination and supposition, at least in the context of trying to explain IR, philosophers have generally ignored mental imagery. This is not surprising, given that the puzzle is about imagining the truth of certain propositions and the latter are taken to be non-sensory and non-perceptual. However, given that we take the content of engaged imaginings to be more fine-grained than that of supposings and that mental imagery mimics perception whose fineness of grain is uncontroversial, it might be promising to consider the possibility that mental imagery is required for such imaginings. But first we have to look more closely at similarities between perception and imagery to see how the second mimics the first.

So what are the similarities? Perception divides into various sensory modalities and so does imagery. We have visual imagery, for instance, which simulates visual perception, and we also have auditory, olfactory, gustatory and motor imagery, among others, which simulate corresponding perceptual modalities. Since visual imagery is the most well-researched area, I will focus on it. First, there are phenomenological similarities. Visualizing something, for instance, is taken by many to be seeing something ‘in mindʼs eye’. There is more to the similarity than simply the metaphor, however, because we also have experimental evidence that in certain conditions, people are not able to discriminate between seeing and visualizing (Perky Reference Perky1910; Segal Reference Segal and Sheehan1972). Aside from phenomenology, there is also neuropsychological data which indicates similarities between seeing and visualizing. For instance, the visual cortex is topographically organised and shows similar pattern of activation both in the case of seeing shapes and visualizing them (Klein et al. Reference Klein, Dubois, Mangin, Kherif, Flandin, Poline, Denis, Kosslyn and Le Bihan2004) so that the activation in the visual area mirrors the configuration of points in the external world (Kosslyn, Thompson, and Ganis Reference Kosslyn, Thompson and Ganis2006, 104). Visual imagery and visual perception also have similar processing constraints. For instance, people face the oblique effect – the relative difficulty in visually processing oblique lines, as opposed to horizontal and vertical ones – both in the case of perceiving and imagining (Kosslyn, Sukel, and Bly Reference Kosslyn, Sukel and Bly1999). There is thus quite a bit of evidence in support of the idea that the two are relevantly similar (see also OʼCraven and Kanwisher Reference O’Craven and Kanwisher2000).

Most importantly for the present concerns, perception and mental imagery seem to share representational properties. It is plausible that properties that can be perceived within one sensory modality do not transcend those that can be imagined within the same type of sensory imagery. Given that imagery within one modality is about redeploying the perceptual capacities within that same modality, the representational scope of the first is constrained by the latter. Just as perception attributes various properties to the perceived scene, so does imagery attribute various properties to the imagined scene, and the types of properties that the first attributes is limited to those that are attributed by the latter (for further argument in support of this assumption, see Nanay Reference Nanay2015). It is true that we can imagistically imagine unicorns and other fantastic creatures which presumably cannot be perceived in reality, but the features from which these creatures are ‘built up’ have to be perceptible features such as shapes and colours.Footnote ¹⁰

Coming now back to the distinction between supposition and engaged imagining, my suggestion is that the second can be distinguished from the first by the fact that engaged imagination requires mental imagery while supposition does not. Although both types of imagination are about accepting p in fiction, engaged imaginings also require mental imagery which is used to represent a fictional situation, thus endowing these imaginings with more fine-grained and detailed content than supposition. This way of drawing the distinction between supposition and engaged imagination should also capture our pretheoretical intuition that genuine imagining is participatory: there is a sensuous involvement with the imagined events which is lacking in the case of supposition (Gendler Reference Gendler2000, 80). Given that engaged imagination is more perception-like than supposition, we feel that we are experiencing the imagined happenings.

What do I exactly mean by engaged imagining requiring imagery? I draw here inspiration from a dual-component view of imagination, according to which our imaginings combine assigned and qualitative content (Kung Reference Kung2010; see also Langland-Hassan Reference Langland-Hassan2015). Qualitative content is provided by a mental image which is a proper part of the imagining. According to Kung, assigned content is the conceptual component of the imagining which consists of labels and stipulations. Labels identify particular objects and stipulations are propositional contents which provide more information about the imagined situation than the image can (Kung Reference Kung2010, 625). The need for non-image content to specify the content of sensory imaginings was already noted by Christopher Peacocke: in order to distinguish between imagining a suitcase and imagining a cat behind the suitcase, the image itself is not sufficient (Peacocke Reference Peacocke, Foster and Robinson1985, 19). Since the content of mental images is limited by what can be perceived in an imagined world, one usually needs assigned content to disambiguate what is represented by the image. On the present picture, I take assigned content to be what is contributed by the suppositional component of engaged imaginings.

It should be noted that the dual-component view is intended simply as a theory of imagination, not as an account of how to distinguish engaged imagination from supposition. I do not think that this creates any problem for the present use of it though. Instead, the present application should show an additional theoretical use for the dual-component account. What may seem more problematic for my account is that Kung actually distinguishes stipulation from supposition by claiming that the first is subject to imaginative resistance because we are not able to stipulate the truth of propositions the falsity of which we find certain (Kung Reference Kung2010, 629). He does not really explain, however, why stipulations are like that and suppositions are not. Since he does not have a substantive explanation of why stipulations are constrained by certainty while suppositions are not, there is no strong reason to draw a strict distinction between them.

Given that assigned content and qualitative content are of different formats, how do they exactly combine together in an act of engaged imagining? Saying that the image is a proper part of such imagining might not satisfy everyone as a sufficient explanation. Here is my proposal. Letʼs begin with the idea that a mental image (with qualitative content) presents us with an array of spatiotemporally ordered sensible properties, ranging across different modalities (see Matthen Reference Matthen and Brogaard2014). Images can also be dynamic and represent an imagined perceptual scene unfolding in time (Gauker Reference Gauker2011, 159). I leave it open whether the qualitative content also presents us with particular objects or whether it is up to the assigned content also to identify objects on the basis of sensible properties. Moving now to the heart of the matter, when we imagine the truth of a proposition in an engaged manner, we also imagine a fine-grained sensory array or various arrays that are meant to correspond to how the state of affairs represented by the proposition would appear to us. For instance, if I am invited to imagine that there will be a war tomorrow, I suppose the truth of the respective proposition, but if I am to imagine it in an engaged manner, I also form mental images of perceivable properties which would correspond to the appearance of tanks crossing the border and/or to the president making an announcement. Such images provide much more detail to the rather general proposition, but the suppositional element is still needed to specify what is happening.Footnote ¹¹ The supposition that there will be war tomorrow allows us to form images of various events that would be unfolding if that supposition were true and the supposition together with these images constitutes an activity of engaged imagining.

Here I introduce a new term, ‘MATCHING’, to characterise such a process of image-formation. An image MATCHES a supposed proposition if it represents a situation which is taken to obtain when the proposition were true. MATCHING is necessary for an act of engaged imagining to take place because image-formation has to respect the truth of the supposed proposition, relative to the imagined world. After all, mental images function to give more fine-grained content to the supposed proposition and in doing that they should be consistent with (i.e. MATCH) the proposition. In addition, without MATCHING, there would be no unitary act of engaged imagining but simply an act of supposition and a separate process of generating mental imagery which develops independently of that act. An implication of this is that an agent cannot pick whichever image she likes to MATCH with a proposition: if an image does not represent anything that could appear to be the case if the imagined proposition were true, it cannot be MATCHED with the proposition.

The postulation of MATCHING-relation, then, implies that there are constraints that the exercise of engaged imagination puts on an agent. When imagining something, one cannot just arbitrarily associate any imagistic content with a supposition. Somewhat speculatively, there is also a functional consideration why engaged imaginings have such constraints. It is plausible that imagination is used to construct at least somewhat realistic future scenarios that could be taken into account in one’s decision-making. If the imagistic content that is generated during engaged imagination is not constrained, then the imagined scene may unfold in completely unrealistic ways. But when imagery has to MATCH the supposed proposition, an agent can rely on the imagining to learn about the ways in which things could turn out to be. Engaged imagination, in order to be geared towards action, should have at least some constraints for it to do its work.Footnote ¹²

But what is the source of such constraints? What makes it the case that certain kinds of imagistic contents MATCH certain kinds of propositions and others do not? My proposal is that the application of concepts that the supposed propositions consist of is grounded in paradigmatic perceptual situations to which they are taken to apply. An imagistic content MATCHES a supposed proposition only if it is not too distant from paradigm situations that ground the concepts which make up the proposition in question.Footnote ¹³

By assuming the perceptual grounding of the relevant concepts I am not committed to empiricism about concepts which identifies concepts with a kind of perceptual representation (Barsalou Reference Barsalou1999; Prinz Reference Prinz2002). I am only appealing to the less contentious, although still controversial, idea that the application of our commonsense concepts is governed by associations with certain perceptual paradigms. In engaged imagination these associations continue playing their role by excluding perceptual scenarios which are too distant from the paradigms that are associated with the employed concept. If there is such an exclusion, there is no MATCH. Thus, although the application of concepts in imagination is more free than in belief and judgment, images that can MATCH the supposed proposition cannot be picked entirely arbitrarily.

Plausibly, the relevance of perceptual paradigms for concept application extends also to moral concepts, which are the main focus of this essay. I am here suggesting a view according to which moral judgments are a form of pattern-recognition, made possible by the encounter with moral examples which constitute paradigm situations (Sterelny Reference Sterelny2010, 287; Reference Sterelny2012, 162). If our ability to make moral judgments has developed by exposure to paradigm cases, then it is reasonable to assume the similarity to those cases determines the correctness-conditions of applying the moral concepts. Also in the context of imagining where imagistic content is involved, the imaginative process is still controlled by the similarity of imagistically represented situations to paradigm cases, associated with the concept that is applied. If an imagistic content represents a situation which is too distant from the paradigm cases, then the moral concept is not taken to be applicable even in imagination. The relevance of this for IR will become apparent in the next section.

The term ‘MATCHING’ might be somewhat misleading, however, because I do not mean to imply that an image always accurately represents what would be the case when the respective supposition were true. It suffices for MATCHING when an image seems for the imaginer, given the paradigms that she associates with a concept, to represent a situation that corresponds to the truth of a supposed proposition, even if the image is not accurate. In fact, there are numerous and varied ways in which a proposition can have MATCHING imagery, depending, among other things, on a person and her idiosyncrasies. A failure to MATCH imagery with supposed propositions, i.e. a failure to imagine in an engaged manner, should thus be relatively rare because it only occurs when a supposition that p and generated imagistic content exclude one another.

The proposed view can explain the features of engaged imaginings which are taken to be characteristic to them. We have already noted how the involvement of mental imagery explains the intuition about the detailed content of engaged imaginings: the content of mental images replicates the fine-grained content of respective perceptual representations. There are other aspects of engaged imagining that the present view explains. First, it explains why engaged imaginings are open to free elaboration. Since sensory imagery is rich in content, it allows for further descriptions, not merely a limited set of inferences that can be drawn from a proposition. Take the example of imagining war taking place tomorrow: the formation of mental images to represent the situation unfolding is an open-ended process, only constrained by MATCHING, and these images can provide fodder for further propositional assumptions about what would happen in the case of war. Second, it makes it explicit why imaginings, as opposed to suppositions, are affectively more engaging. It is plausible that imagistic imagination can cause emotions automatically, without any cognitive mediation (LeDoux Reference LeDoux1996). The mental imagery which is generated in the course of engaged imagining that p can thus automatically bring about affective responses and this is why imagination feels much more affect-laden than supposition. Third, the dual-component view explains why engaged imaginings are not fully under our voluntary control: MATCH between an imagistic content and a supposed proposition is not something that the agent can simply decide. Fourth, the present view helps to make sense of the idea that engaged imaginings are a form of recreative perspective taking: by forming MATCHING images we are recreating experiences of what it would be like if supposed propositions were true. Finally, the present view enables us to understand why we can suppose the truth of almost any proposition, but why we cannot always imagine it. The representational limits of engaged imagination are the limits of the ability to find MATCHING mental images for supposed propositions. In the next section, I will elaborate on this and argue that imaginative resistance can be at least partially explained in terms of those limits.

4. Moral propositions and imagery

Given that engaged imagination requires mental imagery, what problems do moral propositions pose for our attempt to image them in an engaged manner? In this section, I will consider two possible explanations of IR. The first relies on the idea that moral propositions cannot be imagined in an engaged manner at all, thus creating IR. The second takes IR to generalise to those moral propositions which do not stand in a MATCHING-relation to already formed mental images. I will reject the first proposal and defend the second explanation, but a consideration of the faults of the first can show why we should accept the second.

The first suggestion is that moral propositions cannot be imagined in an engaged manner because no image can represent moral properties. Take KB again and consider the phenomenology of imagining the scene that is depicted there. Having read that Giselda killed her baby I (unfortunately) form a visual image of a woman killing a baby, possibly accompanied by imagery of other sensory modalities. However, when I read that she did a right thing, I face an obstacle and I am not able to continue the process of image-formation because I cannot form a mental image that depicts the moral property.

Why would one think that it is not possible to form an image of a moral property? One suggestion could be that since mental imagery inherits its representational scope from respective perceptual modalities and moral properties are not perceivable, they are not imagistically representable either. It does not matter for our purposes where the line between perceivable and non-perceivable properties exactly runs. It sufficed if one could show that moral properties are not suitable for being imagistically represented.

The fatal problem for this explanation is that it overgenerates. It implies that any fictional description which attributes imperceivable properties should face IR. For instance, since it is doubtful that we have immediate sensory access to certain types of mental property, this explanation seems to have the consequence that IR arises also when we try to imagine propositions in which mental states are ascribed to persons.Footnote ¹⁴ Yet it seems that we usually do not feel any resistance when weʼre invited to imagine someone believing, wanting or feeling something. The representational scope of engaged imaginings is broader than the present explanation of IR suggests. A satisfactory account of IR should allow for moral propositions to be imaginable in certain contexts and explain why IR is limited to moral propositions with particular contents.

So how is engaged imagining of properties that cannot be perceived possible? Let us consider mental properties first. We can treat observable behaviour as expressive or indicative of a mental property: although no image is formed which represents the mental property directly, it is sufficient to imagine corresponding behaviour, as long as the imagined behaviour is sensitive to changes in mental properties, i.e. as long as it MATCHES them. We can speculate that the suppositional aspect of an engaged imagining specifies what mental properties are to be (non-imagistically) imagined, while mental imagery depicts behaviour which is not too distant from the behaviours that are associated with the mental properties in question. Although no mental state is imagistically represented, images of behaviour still MATCH the supposed mental properties and there is no resistance. We can imagine propositions about mental properties in an engaged way by imagistically imagining the corresponding behaviour.Footnote ¹⁵

If propositions about mental properties can be imagined in an engaged manner by MATCHING images of behaviour with them, the same can be done with moral propositions. There are surely paradigmatic actions which are performed in situations where the properties of rightness and wrongness are applicable and these actions can be imagistically represented. As long as an imagistic content is not too distant from such paradigmatic actions associated with a concept, it can MATCH the proposition involving that concept. When incorporated to the act of engaged imagining it should be possible to imagine the truth of most propositions in an engaged manner, as long as there is a MATCH between mental imagery and the proposition imagined.

These observations allow us to reach the second, more promising explanation of IR. We should accept that we can imagine propositions about moral properties and non-perceivable properties in general by MATCHING imagistic representations with such propositions. This does not mean, however, that this process does not have its limits. There are still cases when MATCHING fails. This happens when we are invited to imagine the truth of a proposition but the mental imagery that we have formed does not MATCH the proposition in question. My second proposal, then, is that we cannot genuinely imagine a truth of some propositions because in the context in which such propositions are presented, the mental images that we have formed do not MATCH them. From this imaginative resistance follows. Although the representational scope of engaged imaginings is impressive, in certain cases we are still unable to MATCH our mental images with the imagined proposition. Invitations to imagine unacceptable moral propositions present us with such cases. If one grants that the property of rightness has a sensory profile, a scene of killing a baby (KB) or of a sudden shooting of an arguing couple (DF) are excluded from it, at least for most of us. One thus fails to imagine the truth of resistance-causing propositions in an engaged manner due to the imagery not MATCHING it. On the other hand, since one can easily MATCH scenes of killing babies and shooting people with propositions which attribute wrongness to these actions, there is no resistance in the case of imagining such propositions.

I call this proposal of how to explain IR ‘the MATCHING Hypothesis’. To put it bluntly, according to the MATCHING Hypothesis, the resistance derives from our inability to MATCH our mental imagery with some moral propositions. Unlike the first proposal, the second explanation does not assume that moral properties cannot be genuinely imagined: although they are not strictly speaking imagistically representable, they can be imagined by combining suppositional content with imagistic content. This combination fails, however, when we are invited to imagine the instantiation of moral properties in situations where imagistic content does not MATCH the assumption that such properties are instantiated.

By saying that we are not able to do this, I am relying on the idea, articulated in the previous section, that our use of moral concepts is based on paradigmatic actions and situations to which we take these concepts to apply. If moral judgments require the ability to recognise paradigmatic sensuous patterns to which moral concepts apply, it is also plausible to think that such patterns are also relevant during moral imaginings. There are certain actions and situations which are too distant from those paradigms, and the cases that cause IR involve imagining exactly such actions and situations. This generalises to propositions involving thicker moral concepts. Try to imagine, for instance, the following cases:

• (1) By running away from battle, Nick did a brave thing.

• (2) By screaming loudly in public, Sarah acted modestly.

• (3) By slaughtering the peasants, the king acted mercifully.

• (4) By peeing on the church wall, Michael acted piously.

In all these cases, something akin to IR seems to arise when we are invited to imagine these situations. The MATCHING Hypothesis explains the resistance by pointing out that the actions in question do not have an appearance which could MATCH the virtue-attributing propositions in question. Note that these cases do not present us with conceptual or metaphysical impossibilities: running away from battle could in some conditions be a brave thing, for instance. However, running away from battle is far from paradigmatic exemplars of brave action, and an image of it does not MATCH the proposition that Nick did a brave thing. It is important to point this out because one might raise a concern that the MATCHING Hypothesis actually boils down to an idea that IR arises from our inability to imagine conceptual or metaphysical impossibilities. This concern is not justified because the MATCHING Hypothesis claims IR to be about a lack of fit between sensory representations and suppositions in certain imagined situations, whether such situations are taken to be impossible or not.

By looking at the examples (1)–(4) alongside with KB and DF we can also flesh out what it means for an imagistic content to be too distant from paradigmatic cases associated with a concept, i.e. for imagistic and assigned contents to exclude one another. Concepts of rightness, bravery, modesty, mercifulness and piety are not acquired in isolation but together with concepts of wrongness, cowardice, immodesty, mercilessness and impiety, which are associated with their own paradigmatic situations. The paradigms associated with a concept and its opposite, however, are mutually exclusive. An imagistic content is too distant from paradigmatic situations associated with a concept if it is more similar to the paradigms associated with its opposite. For instance, in the case of KB, one has to MATCH the assigned content that Giselda did the right thing with an imagistic content that is paradigmatic of the concept of moral wrongness instead of rightness. And in the case of (1), what one imagistically represents is closer to the paradigms of cowardice than those of bravery.Footnote ¹⁶

Can the MATCHING Hypothesis also explain our resistance to imagining certain non-moral propositions? Some have taken IR to extend to aesthetic and epistemic evaluations, attributions of mental states and attributions of content, shape-ascribing propositions, ontological claims, etc. (Weatherson Reference Weatherson2004) If IR is such a general phenomenon, one might think that not all cases are amenable to the MATCHING Hypothesis and that I have been cherry-picking my examples.

There are examples which it handles well. For instance, an attempt to imagine a story of people taking a maple leaf to be of oval shape (Yablo Reference Yablo, Gendler and Hawthorne2002, 485) causes resistance because an image of one holding a maple leaf does not MATCH the proposition that one has found something that is of oval shape. The same can be said about the story of Quixotic Victory, for instance, in which a television and an armchair are said to have appearances of a knife and a fork (Weatherson Reference Weatherson2004, 5). There are other examples, however, which are seemingly more problematic. Take, for instance, an invitation to imagine that a knock-knock joke is hilariously funny, which also seems to cause resistance and puzzlement (Walton Reference Walton1994, 43f). Does the puzzlement really come from our inability to MATCH an image of a knock-knock joke with a supposition that the joke is funny?

I think that it is not at all ridiculous to assume that funniness has a paradigmatic perceptual shape: for instance, it may include one having internal sensations when one is amused and which are associated with funniness. One could then argue that if I imagine a knock-knock joke being told, then the imagistic content I have formed of hearing a knock-knock joke does not MATCH the supposition that the joke is funny. It does not MATCH it because the concept of being funny is associated with the imagery of feeling amused but imagining hearing a knock-knock joke fails to involve anything like it. If anything, the imagistic content which I form involves feelings of annoyance which are too distant from the paradigms of funniness. Although I may be able to imagine others having the sensations, I cannot imagine myself having them and thus I cannot apply the concept of being funny even in my imagination. This should apply to most of us who have been exposed to other, presumably more sophisticated types of humour. On the other hand, a person who tends to feel amused when exposed to simple and repetitive humour might be able to imagine that a knock-knock joke is hilariously funny. For her, hearing a knock-knock joke is close enough to, or perhaps even paradigmatic of, situations which she associates with funniness.

There is probably an indefinite amount of possible counterexamples to the MATCHING Hypothesis which I have not considered. Given that it is an empirical hypothesis, which relies on the dual-component account of imagination and on the view that concept-application is basically a form of pattern-recognition, it is also vulnerable to empirical disproof. The defense of these two assumptions will have to wait for another occasion, however. I hope that my current efforts suffice to give to the MATCHING Hypothesis at least some plausibility. Before concluding this paper, I will consider two other theories of IR, which I take to be the most promising ones in the literature, and see how they compare with the MATCHING Hypothesis.

5. Comparisons

Funding

This work was supported by the Estonian Research Council [grant number PUTJD656].