Imagination Between Bats and Cats: Commentary by Margherita Arcangeli on Explaining Imagination (with reply)

By Margherita Arcangelia

Imagination is clearly “a dense and tangled piece of country” (Furlong 1961: 15). The last decade, however, saw considerable philosophical work aimed at mapping this terrain of the mind. Peter Langland-Hassan’s book is a sophisticated and thought provoking atlas, whose purpose is to show that where other explorers have found mysterious creatures, in fact there are ordinary animals. It seems that I belong to those explorers: contrary to Peter, who believes that imagination can be reduced to basic folk psychological states, I claim that imagination is a sui generis mental state.

More precisely, I take imagination to be the capacity to “recreate” non-imaginative kinds of mental state. “X-like imagining”, or “recreating X in imagination” (where X is a type of non-imaginative state like perception or belief) means that the relevant imagining is similar, from a phenomenological and/or functional point of view, to the counterpart it recreates. Imagination is a family, within which different genera and species can be identified, that recreate non-imaginative mental states, but in a partial way that makes them different in nature to their counterparts (it is an open question how many varieties of recreative imagination there are, that is, which mental states imagination recreates). This view is often conflated with the simulationist approach – notoriously held by Currie & Ravenscroft and Goldman, but it has older roots, which can be traced back to Husserl’s philosophy.

Despite our divergent ontological commitments, Peter and I agree in thinking that imagination exists. Indeed, his reductive proposal is not eliminativist. According to him, imagination is ontologically solid like water and yogurt cakes are: analyzing it in terms of the right combination of more basic entities does not make it disappear from the shelves of the mind’s market. Moreover, Peter offers a taxonomy of the “imagination” aisle and distinguishes between two senses of imagining: Imagistic Imagining (I-imagining) and Attitude Imagining (A-imagining). In a nutshell, I-imagining refers to the use of mental images, while A-imagining “is rich or elaborated thought about the possible, unreal, or fantastical that is epistemically safe” (p. 61).

Such a distinction might easily remind us of two varieties of recreative imagination, namely Sensory Imagination (SI) and Cognitive Imagination (CI) – different labels have been used to capture such a distinction, here I am using my preferred terminology. While the former is the recreation of perception, the latter is the recreation of belief.

However, Peter stresses that we should resist conflating I-imagining/A-imagining with SI/CI for at least two interrelated reasons. First, although A-imagining and CI roughly refer to the same phenomenon (p. 6), the same does not hold for I-imagining and SI. Some uses of mental imagery are not recreative imaginings (p. 72), thus hinting at a different phenomenon.

Second, SI and CI are different genera within the family of recreative imagination, whereas I-imagining and A-imagining are categorically distinct. Although both are called “imagining”, Peter suggests that what we have here is mostly a case of homonymy and homography: I-imagining stands to A-imagining, as the object bat stands to the animal bat – though according to him there is an overlap between I-imagining and A-imagining (maybe vampire bats can be used to hit baseballs, after all!).

Strangely it might seem, I agree with Peter. In my own taxonomy (see Arcangeli 2018 and 2020) I have identified two different classes of mental phenomena, which (misleadingly, I think) are lumped together under the umbrella “imagination”: imagination as a type of content – perhaps a format (i.e., mental imagery), and imagination as an attitude (i.e., recreative imagination). My suggestion is to broaden the notion of A-imagining – which in my view is not necessarily propositional nor only similar to belief – to encompass, at least, both SI and CI (in Arcangeli 2018 I argue for the inclusion of supposition as a third variety). Here stark divergences between our views start emerging.

Peter would undermine my taxonomy by arguing that SI and CI are not recreative in the same sense, thus failing to belong to the same family. He claims that while SI is recreative in a format sense (i.e., sensory imaginings “recreate the (presumably pictorial, or iconic) format of different perceptual states” – p. 73), IC is recreative in a functional sense (i.e., cognitive imaginings recreate aspects of the functional role of beliefs). I reject this view, which is based on the questionable equivalence of I-imagining and SI: in my view, the former is recreative in the format sense, whereas the latter is recreative in the functional sense (roughly objectual recreation and mental recreation, respectively, in my terminology – see Arcangeli 2020).

Peter would press me on clarifying in which sense SI functionally recreates perception, given their crucial functional differences (e.g., endogenous vs. exogenous causes, different epistemic relations to beliefs). My answer is that recreative imaginings, be them perception-like, belief-like or whatever, mimic only part of their counterparts’ cognitive profiles – which include their functional roles, as well as their phenomenological aspects. Recreative imaginings also show a proper phenomenology and have their own functional features, which suggests that recreative imagination is a primitive psychological attitude.

What is wrong with this view? I am not stipulating a mysterious “faculty of mind or collection of sui generis mental states, quarantined from our actual beliefs, desires, and intentions”, that “could be carved off the mind while leaving our self-defining commitments and inclinations intact” (p. 1). On the contrary, I am treating imagination on a par with other primitive psychological attitudes, like perceptions, beliefs, desires, emotions.

I agree that in ordinary language “‘Imagine’ is a lot like ‘bat’” (p. 4), but I think that philosophical analysis should (also) seek an “exploratory theory” (Stock 2017: 6) of imagination. That it is to say, philosophers should not simply analyze the folk concept of imagination, but try to improve it, which might lead to a transformation of our usual way of understanding imagination. This endeavor should result in sorting out misuses of imagination-related vocabulary, identifying those occurrences which truly refer to mental phenomena and, if necessary, separating them into different categories. Within the terminological regimentation of imagination-related vocabulary in philosophical discourse I defended, imagination is not heterogeneous like bats, but rather like cats (which I have gladly gone on to study!).

Reply from Peter Langland-Hassan

Margherita Arcangeli and I are something like the Scylla and Charybdis of the philosophy of imagination.  Drift too close to me and imagination is ground to mental dust, a heterogenous cluster of other familiar mental states.  Shift sails in Margherita’s direction, and imagination fractures into myriad sui generis pieces—perhaps as many different kinds as there are “ordinary” sorts of states for imaginings to recreate. 

Most will prefer an intermediate position.  But the way between us is narrow.   As Liao and Doggett (2014) observed, what reasons there might be for positing one kind of sui generis imaginative state (imaginative beliefs, say) will likely reappear as reasons for positing many other kinds of sui generis imaginative states, such as imaginative desires, imaginative intentions, imaginative rememberings, and even imaginative imaginings.  It seems we can either accept the general form of argument and enter Margherita’s imaginative menagerie, or reject it and dwell in my desert landscape.

But the question today is not whether Margherita and I can indeed crush the foolhardy who sail between us, but what we should say to each other.  According to Margherita’s view, the most fundamental similarity had by each type of imaginative mental state is that it “recreates” some other type of (non-imaginative) mental state.  For one kind of state to recreate another, Margherita explains, it must recreate at least “part of” its counterpart’s cognitive profile, which includes its functional role and “phenomenological aspects.”  In addition to recreating part of the cognitive profile of its counterpart, she adds, “recreative imaginings also show a proper phenomenology and have their own functional features,” which “suggests that recreative imagination is a primitive psychological attitude.” 

So, on Margherita’s view, each kind of imaginative state will have two parts—an A side and B side, as it were.  The A side will be the recreative aspects of the state, which mimic the functional role and/or phenomenology of some other kind of state.  The B side will contain the “proper phenomenology” and other functional features distinctive of recreative imaginings alone, and in virtue of which the state qualifies as a recreative imagining.

“What is wrong with this view?” Margherita asks. An argument I offered in the book was that a single candidate for being a recreative imagining could have multiple A sides, recreating aspects of the functional role or phenomenology of multiple different mental states.  This would create indeterminacy in which state it in fact serves to recreate, and thus in what kind of (recreative) state it is.  Consider, for instance, the experience of episodically remembering a friend’s birthday party.  A strong case can be made that this state recreates the phenomenal and functional aspects of a past perceptual experience and is therefore a perception-like recreative imagining.  Its A side recreates the cognitive role of a perceptual experience.  Yet a case can also be made that the remembering recreates aspects of belief.  Like a belief, it is stimulus-independent, associated with taking things to have been a certain way, and (plausibly) enters with other beliefs into reasoning sequences.  Which type of recreative imagining then is this?  The general problem is that there are many ways in which one type of mental state might mirror the cognitive role of another and no barrier to its mirroring that of many.  Without clearer guidelines in place concerning which functional and phenomenological aspects will count as the relevantly recreated aspects (and why), we lack means for determining which type of recreative imagining we have before us in any instance.

A second problem stems from a lack of clarity about the nature of the B side.  What are the functional and phenomenological features that all recreative imaginings have in common?  Currie and Ravenscroft—whose (2002) innovated talk of “recreative imaginings”—did not say.  Yet knowing something substantive about the B side is critical.  Note that whenever the A side of a recreative imagining mirrors aspects of the cognitive role of some other state (such as a belief), the state that is mirrored (e.g. the belief) likewise duplicates aspects of the cognitive role of the recreative imagining.  Mirroring is transitive, after all.  Thus, mirroring aspects of the cognitive role another type of state is not sufficient for being a recreative imagining.  This means that we cannot know whether a state is a recreative imagining—or, instead, the kind of state that might be recreated—until we clearly grasp the distinctive B side of all recreative imaginings.  I have not yet seen a satisfying characterization of this shared B-side.  (My apologies to Margherita if she has explained this elsewhere and I have overlooked it.)

Nevertheless, these worries—perhaps surmountable—are not what really drive my resistance to the doctrine of recreative imaginings.  I share Margherita’s desire for an “explanatory theory” of imagination.  In seeking that theory—and the feeling of understanding it should bring—I want to reduce the number of primitive notions that appear in our explanations of imagination.  The problem with the recreative approach is that it requires us to understand both the states that get recreated (so we can understand the A-side aspects of recreative imaginings) and the distinctive B-side aspects of recreative imaginings.  There can be no understanding of imaginative belief, or imaginative desire, after all, without an understanding of ordinary belief and desire themselves.  Thus, if we can develop an adequate explanatory theory of imagination that only mentions the ordinary states—the A-sides, as it were—this should bring with it a deeper understanding than if we must grasp the B-sides in addition.  This is what I set out to do in Explaining Imagination, by showing how the most salient behavioral and cognitive phenomena associated with imagination can be explained by appeal to ordinary, “A-side” mental states alone.

So, to return Margherita’s serve:  what is wrong with this view?    


Currie, G., & Ravenscroft, I. (2002). Recreative Minds: Imagination in Philosophy and Psychology. Oxford: Clarendon Press.

Liao, S.-y., & Doggett, T. (2014). The Imagination Box. Journal of Philosophy, 111(5), 259-275.


  1. Thanks to both authors here for this stimulating exchange!

    I have a question for both authors regarding Peter’s criticism of what he calls the “A side” of imaginative states under Margherita’s theory (i.e., “the recreative aspects of the state, which mimic the functional role and/or phenomenology of some other kind of state”). The problem Peter raises is that the same imagining could recreate the functional role or phenomenology of multiple states, using the example of episodic memory. An episodic memory seems to recreate aspects of a past perceptual experience. But Peter argues that it also recreates aspects of belief: “it is stimulus-independent, associated with taking things to have been a certain way, and (plausibly) enters with other beliefs into reasoning sequences.” Peter argues that this seems to create a problematic indeterminacy with respect to what kind of recreative state the episodic memory is.

    Here’s the kind of analysis I’d be tempted to give in response to this argument. The belief-like features of episodic memory Peter mentions here seem to me not to suggest that it recreates belief. Rather, they seem to suggest that it is a kind of belief–i.e., that episodically remembering that P is a way of believing that P. After all, it seems plausible that what it is to be a belief is, roughly, to take things to be a certain way and to enter into the right kinds of inferential relations with other beliefs. If this is right, then we can say that an episodic memory is a state which both is a kind of belief and recreates a past perceptual experience.

    This analysis seems to me consistent with both the authors’ views here: it’s consistent with Margherita’s recreativism, according to which imagistic mental states are imaginative states which recreate perceptual experiences. But it’s also consistent with Peter’s attempt to explain imaginative states in terms of other kinds of mental states, since it posits that we can explain episodic memory in terms of belief.

    However, I’m sure there’s something I’ve said here which either or both authors would want to deny. So, it would help me get straight on both of your views to hear what this would be!

    I’m also curious whether Peter has any other examples, besides episodic memory, of the general claim that one type of imaginative state might mirror the cognitive role of multiple other kinds of states. It may be that I’m focusing in too much on the particular example of episodic memory here, and that there are other cases that illustrate this claim without being open to the kind of analysis I gave of memory here.

    • Hi Daniel,

      Thanks for these great questions. I am tempted, like you, to think of episodic memory as a kind of belief. The situation is a bit complex though. Many people think that episodic memories have more stringent correctness conditions (relating to the causal dependence of the putative remembering on a past perception of the event remembered) than ordinary beliefs about the past, which could suggest functional differences. But set that aside, and let’s assume episodic memories are beliefs. I don’t think that Margherita can say that a certain type of state is both a belief and a recreative imagining. (Though I’ll be interested to see what she says.) The reason is that, if the state really is a belief, it must have the full cognitive profile of belief. This means that it cannot have, in addition, some aspects that are not-belief-like. Now, it could be that some aspects of belief are also aspects of perceptual experience. But, in that case, those same aspects of perceptual experience are also aspects of belief. Which then—the belief or past perception—counts as the “recreated” state, and why?

      In terms of other examples where one state might seem to mirror the cognitive role of multiple others, one might imagistically imagining how things will go in the future. Such an imagining might mirror the cognitive role of a belief that, in the future, thus and such will occur. (Indeed it might be such a belief.) And one could say that it mirrors the cognitive role of a future perception, or of perceptual experience in general. For other examples, it depends on how finely you cut things. A wish might partly recreate the cognitive role of both a hope and a desire, yet not be a recreative imagining of either. The “A-side” of a belief-like imagining that X ought to be done might recreate aspects of the cognitive role of both the belief that X ought to be done and the desire to do X.

      The issue, at bottom, may be whether it is possible to articulate functional characteristics had by all and only recreative imaginings (where the function is something other than “recreating” a state!).

  2. Tommy Cleary

    I am interested in imagination in so much as it helps to compare and contrast imagination with intuition when exploring the idea of other minds and the edge of perception and temporality.
    At quantum physics conferences the scientists spend much more time trying to understand other scientists minds by imagining beyond what intuition can provide how a particular data set or theory gives insight into a particular epistemic claim.
    This is because logic can only take one so far into another’s thinking.
    Language and the ontology of thought must bend to the intuitive and counter intuitive, the spontaneously imaginary and the deliberately imaginary such as the Copenhagen explanation of quantum mechanics being imagined as scaled up to Erwin Schrodinger’s box/cat in a thought experiment which tests all parts without hurting cats.
    How do the authors, in a nutshell, help me imagine in a way that at the moment I cannot imagine, such that I, can in the future, imagine understanding Schrodinger and his claims about boxes and cats more comfortably?
    How to imagine more deliberately?

  3. Many thanks for your reply, Peter (and Daniel for your thought-provoking analysis!).

    In another blog post (see Peter’s Junkyard’s post here: you stressed that while your view on imagination epitomizes the Quinean desert, my view is more like the Meinongian rainforest. It’s true that I owe much to the phenomenological tradition (I am indebted to Kevin Mulligan for this). Husserl saw imagination as creating essential modifications of non-imaginative states. Certainly the idea of “essential modification” should be spelled out further, but what this evocative way of talking clearly suggests is that imaginings are fully fledged mental states, rather than being pale copies of other mental states (e.g., perception, belief). On this view, it’s difficult to see how an imagining could be divided neatly in two parts: the A side (what the imagining retains of its counterpart) and the B side (what the imagining has in common with all imaginings) are fused together.

    Anyway, I agree with you that a thorough analysis of recreative imagination should explain both the features that link each type of imaginative state to some non-imaginative mental state type (i.e., the A side), and the features that make a mental state an imaginative state (i.e., the B side). A satisfying characterization of the latter is yet to come, but some bases have been laid (in Arcangeli 2018 I have tried to make some progress in this sense, suggesting a common phenomenology and some core features proper to all imaginings, as well as key dimensions to compare imaginings with their counterparts).

    As for the question about the underdetermination of the recreated state by a single imagining, I think it is due to the complexity of the real, in contrast with the (over)simplification of theorizing. We might engage in an imaginative project trying to recreate only, say, the perceptual experience of a flower, but most of the time imaginings of different kinds mingle together (after all imagination tries to mimic reality, which is so complex). The case suggested by Peter might be interpreted in this way: we wouldn’t have a single mental state (yes, I don’t think that a certain type of state can be both a recreative imagining and, say, a belief), but rather a combination of perception-like and belief-like imaginings. It should be noted, however, that the example adds a further layer of complexity, namely how imagination relates to episodic memory. Daniel’s suggestion seems to highlight this: we might think that the case involves belief rather than belief-like imagination, because episodic memory is intrinsically epistemically loaded.
    Despite the complexity of reality, isolating varieties of recreative imagination is still important. It’s part of the process of breaking down phenomena that helps to understand them.

    So, why am I not tempted by a theory that goes further in the separation into parts? Why do I resist a theory that seems to do for imagination what Lavoisier and Cavendish did for water? I see two reasons. First, I don’t think that this is what the theory is really doing. Water and the chemical elements it is composed of belong to different levels, the macro and the micro respectively. Thus, explaining water in terms of its constituents means to shift from one level to another. But, imagination, beliefs, desires, and intentions are all folk psychological states. And you stress in the book that the reduction/explanation you are pursuing is to be made neither at the neuronal level, nor at the cognitive level, but at the level of folk theory posits.
    Second, I think that this kind of reduction/explanation misses something. In trying to pinpoint what this something is, I saw a parallelism. Your position on imagination reminds me of Norton’s position on thought experiments: as there is nothing special about the latter and TEs can be reconstructed (translated, or reduced) into arguments, so imagination can be reconstructed (translated, or reduced) into a combination of beliefs, desires and intentions. It has been objected that Norton’s view loses some important aspects proper to thought experimentation, especially its action-practical dimension. I’m prone to say that reducing imagination to a combination of beliefs, desires and intentions fails to capture what happens at the first personal level. To my ears it sounds as a theory of imaginings’ attribution from a third person perspective, seeking to explain behavioural dispositions. This is a worthwhile epistemic endeavour to pursue, but that tells only one part of the story.

Leave a Reply

Back to Top
%d bloggers like this: