Unedited draft of chapter forthcoming in Philosophical Aesthetics and Aesthetic Psychology, Elisabeth Schellekens and Peter Goldie (eds), Oxford: Oxford University Press, forthcoming in 2010.

Imagination Unblocked

“Resistance is futile.”

--The Borg

1. The puzzle(s) of imaginative resistance

Some things are very hard to imagine—round squares, prime numbers evenly divisible by eight, and five and seven not adding to twelve are pretty plausible mathematical cases of this. Most famously, our imagination seems especially constrained in the moral realm (Walton 1994, Moran 1994). For although it appears fairly easy to imagine a world in which people falsely believe that the torture and murder of innocents is required, it is rather difficult (arguably impossible) to imagine a world in which the torture of innocents is morally required. Relatedly, it seems hard to make sense of a fiction in which it is true in that fiction that the torture and murder of innocents is morally required.

But it is a noticeable feature of artistic practice that talented authors can turn the unimaginable into the stuff of fiction. Graham Priest arguably succeeds in making it true in his story “Sylvan’s Box” that there is an absolutely empty box with something in it (Priest 1997). Tamar Gendler, whose essay “The Puzzle of Imaginative Resistance” (Gendler 2000) initiated a flurry of interest in a range of aesthetic issues having to do with the limits of the imagination, composed a clever story in which it certainly seems to be true—and reported as imagined by at least some readers—that five and seven do not add up to twelve. And a range of philosophers skeptical of the puzzle Gendler presented have pointed to cases in which writers do seem able to get us go along with (i.e., accept as true-in-the-fiction and perhaps even imagine) moral impossibilities—Icelandic sagas, mafia movies, the hard-boiled detective genre, and Hamlet have all been suggested as fictions which do not generate imaginative resistance in response to the morally problematic worlds they portray (Landy 2008, Todd 2008, Kieran 2010). Even (fairly) untalented authors can do the trick. Later in this paper, we construct a story that is designed (and we think succeeds) in turning what initially appears to be unimaginable—Brian Weatherson’s “Wiggins World” example—into the imaginable by adding a few details (Weatherson 2004).

Work by Gendler and Weatherson, as well as recent discussion by others such as Gregory Currie (2002), Derek Matravers (2003), Shaun Nichols (2006b), Kathleen Stock (2003, 2005), Dustin Stokes (2005), Cain Todd (2008), Kendall Walton (2006), and Stephen Yablo (2003) has enriched our understanding of the first set of phenomena—the various forms imaginative resistance including imaginative refusal (an unwillingness to imagine) and imaginative blockage (difficulty or full-fledged inability in imagining). But the phenomenon of unblockage—the way in which various imaginative impediments may be avoided or overcome—has gone largely unnoticed and untheorized. We think this is a serious gap in the discussion. For it is a condition on a successful theory of imaginative resistance that it be able to explain the various ways that we can get round such resistance. That is what we propose to focus on in this paper.

In an earlier work on this topic (Weinberg and Meskin 2006), we argued that philosophers’ almost exclusive reliance on metaphysics and folk psychology (see Currie 2002 and Nichols 2006b for noticeable exceptions) has meant that many relevant phenomena are left unexplained. We remain convinced that the psychologically and (cognitive) architecturally informed theory we presented in that work—with some modification—can do a better job than any of the other contenders of explaining various aspects of imaginative resistance. In particular, we shall argue that our theory offers natural explanations of the various ways that imaginative blocks (or would-be blocks) can be circumvented. In the next section of this paper we lay out that theory briefly. We then explain why it predicts the phenomenon of imaginative blockage. In the later sections of the paper we turn our attention to two ways in which we may be able to avoid blockage—either getting around it temporarily or getting rid of it altogether.

Our central interest is in the phenomenon of imaginative blockage. That is, we are primarily interested in a psychological phenomenon rather than in artistic concerns. But like many other writers on this subject, we assume that there is an intimate connection between fiction and the imagination. In broad strokes, fiction may be characterized as a tool to direct the (cognitive) imagination (Currie 1995, Walton 1990). So fiction, and facts about our engagement with them, will provide us with rich source of data about the cognitive imagination. But what is the cognitive imagination? To get a handle on that, we need to turn our attention to issues of cognitive architecture.

2. The Cognitive Architecture of the Imagination

Two important empirically-supported results about the nature of the imagination are key to our account: (1) the functional similarity between believing and imagining and (2) the existence of a distinct cognitive system that underwrites the workings of the imagination. While we do not have room to present the empirical evidence for these results here, we can say a bit about how these two characteristics of imaginings are to be understood

Beliefs and imaginings are functionally similar insofar as they interact with (largely) the same mental mechanisms. By and large, if the belief system takes input from or produces output to a cognitive mechanism, then the imagination system does as well (and vice versa). For example, there is plenty of evidence that various inferential mechanisms operate on both beliefs and imaginings (Leslie 1994). In addition, the cognitive mechanisms that do interact with both systems (e.g., our emotional systems) treat representations from either system in roughly the same way (Lang 1984). In Shaun Nichols’ terms, imaginative states and beliefs states are in a ‘single code’ (Nichols 2004b).[1]

But despite the functional similarity between beliefs and imaginings, a separate system—distinct from the belief system—must be posited. To take an obvious point, imagining does not appear to drive the action system in the same way that belief does. Since systems are individuated by their functional role, this suggests that there really are distinct systems that subserve imagining and believing.

These results raise some crucial questions. Which cognitive mechanisms interact with both imaginings and beliefs, and which ones interact with just one of the systems in questions?[2]

We have already mentioned that various inferential mechanisms operate on both beliefs and imaginings. A particularly important piece of the cognitive architecture for our purposes is the mechanism (or mechanisms) that Nichols and Stich (2000) have termed the ‘UpDater’. The UpDater handles the crucial task of adding and deleting beliefs in response to the receipt of new information. And this mechanism is clearly required to make sense of imaginative engagement just as much as belief revision. For example, we regularly update the contents of our fictively-generated imaginings in light of new fictive input.

Other largely automatic or ‘modular’ systems will plausibly interact with both imaginings and believings. For example, most fiction relies heavily on audience use of folk psychology to make sense of characters and their actions, so it is reasonable to suppose that the systems that underwrite those capacities are able to interact with the imagination in much the same way that they interact with belief. In the context of our discussion of imaginative resistance, we are particularly interested in mechanisms that underwrite our moral capacities. For it is crucial to our engagement with ordinary narrative fiction that some moral mechanisms be engaged by the imagination. We could not make sense of the moral emotions of fictional characters and respond appropriately to them were we not able to make moral judgments about the fictive (and, hence, imagined) events which they face.

While the aforementioned mechanisms interact with both beliefs and imaginings, there is at least one important special-purpose mechanisms that must be posited to make sense of the workings of the imagination. Although we cannot generally make ourselves believe whatever we decide to, some mechanism must allow us to imagine just about whatever we decide to – modulo cases of blockage, of course. We will call the mechanism which subserves our capacity to do this the ‘InPutter’.

With this rough sketch of the architecture of the imagination in place, we can turn to our account of the phenomenon of imaginative blockage.

3. Imaginative Blockage: A Diagnosis

Many of the usually automatic systems that interact with the imagination can either add representations to, or remove representations from, that system. In addition to the InPutter, various modular reasoning systems that add or subtract from our store of beliefs also add or subtract from our store of imaginings. And because of this, it is possible to get a conflict between these various systems – in particular, a situation may arise in which one system (most typically the InPutter) is trying to insert a representation even while another (most typically the UpDater) is trying to remove it. On our account, this is exactly what happens in cases of imaginative blockage.

Consider moralimaginative resistance. Suppose in the course of some exercise of the imagination we are confronted with an invitation to imagine some morally abhorrent propositionpand that this results in blockage. Our explanation is that two systems have been put into conflict: as we attempt to comply with that invitation, the InPutter is accordingly trying to insert p into the imagination, but at the same time, the UpDater registers a conflict between p and some output of our moral judgment systems (not-p, since it detects the abhorrence of p). Importantly, since the moral judgment system works automatically -- and is outside of the imagination -- removing not-p will not be effective. The moral system will automatically function to reinsert not-p into the imagination in response to morally salient features of the imagined situation. So the UpDater's only way to resolve the conflict is to remove or reject the offending representation p – even while the InPutter is trying to add p in. At that point, we can no longer proceed smoothly and automatically. And there is no obvious non-automatic way to proceed, either. We can cast about for a way to imagine p without engaging the moral judgment system, but none will prove easily forthcoming. So, we are stuck: we are instructed to do something that we are simply unable to do.[3]

Our account easily generalizes to other cases of blockage reported in the literature. Most typically, one of the systems involved in the conflict will be the InPutter itself, although we do not believe that this is a necessary condition on blockage. (In fact, our theory raises the very real possibility of blockages generated by the inconsistent outputs of distinct automatic systems.)And note that conflict need not generated by an initially imagined proposition, but may be generated by some other proposition that we automatically derive from it.

Under some circumstances, when we run into such a conflict in the imagination, we simply would end up with one of the relevant propositions removed and then be able to move on. Something similar would happen with a similar conflict in our beliefs, and the single-code approach considers imaginings and beliefs both to be generally subject to the same sorts of coherence-driven revision processes. But this is not always possible: some conflicts involve contents produced by automatic and modular processes and, hence, cannot be revised away. Nor, in many cases, can contents that the InPutter has already inserted be easily removed. For example, we tend to treat many of those contents as (almost) sacrosanct—only allowing the contents we input from engagement with fiction to be removed if we come to believe that we were dealing with an unreliable narrator or a dream sequences or some other “epistemological twist” (Wilson 2006).[4]

So far, we have presented a general picture of the cognitive underpinnings of imaginative blockage. The account rests on an independently well-supported theory of cognitive architecture (see Nichols and Stich 2000 and Weinberg and Meskin 2006), and that theory of cognitive architecture makes its own predictions as to when we will experience blocks[5]. We turn now to a consideration of how the account and architecture of blockage are relevant to two phenomena exhibited in our engagement with fiction: imaginative deferral and blockage removal.

4. Imaginative Deferral

In many fictions, the putative necessary cognitive conditions for blockage obtain, and yet we are not blocked. One way this may happen is if we are able configure our imaginations so that various subpersonal and automatic systems that typically interact with it (and with belief) are temporarily disconnected (Weinberg 2008). That is, we have a capacity to adjust our imagination, often aided substantially by the construction of the work of fiction, so that it does not take inputs from various systems (e.g., some of the moral systems); and this capacity is plausibly what enables us to imaginatively engage with works in the various countermoral genres mentioned above.

In some other cases the fiction is so structured as to enable us to defer the conflict between our cognitive systems. Such deferrals are usually paid off by the narrative later demonstrating that the conflict was spurious: what initially struck us as impossible is later revealed as possible, albeit in a surprising way. Apparently unfulfillable prophecies are perhaps the most famous version of this phenomenon. In Macbeth, for example, we seem to imagine all of the following: that the witches speak truly in their claim that“no man of woman born” can harm him, nor can harm befall him “until great Birnham Wood onto high Dunsinane hill shall come"; that these conditions are impossible to satisfy; yet also, that Macbeth will ultimately be overthrown. This is, after all, a Shakespearean tragedy, and we know in advance that matters cannot go well for the title character, and the proper appreciation of works in that genre requires that such knowledge be deployed in our readings and viewings.

We know that these cannot all be true together, yet we seem to imagine them nonetheless. It is clear that all of these propositions are in some way active in our engagement with the play, after all, or else we would not experience the admixture of foreboding and mystery that we do. Hence, cases like Macbeth cannot be explained away in the simple terms of our either just not noticing the conflict between the propositions, or ignoring it. Nor do we allow ourselves to become blocked – it is hardly the case that we throw the text across the room, decrying what a hack this Shakespeare fellow is! So we seem to have a case of an imagined acknowledged impossibility that nonetheless does not create blockage, and we must address the question of how this could be so.

Our suggestion is to take the following propositions to be the ones actually entertained in the imagination: (i) That the witches’ prophecy will be true; (ii) That Macbeth will nonetheless suffer a downfall; and (iii) it does not seem possible that (i) and (ii) can both be true. It is important that this last proposition not be (iii’): it is not possible for (i) and (ii) to both be true, or else we'd still just have the contradiction.

But one might wonder why that is not indeed exactly what we imagine. For it does seem that, on our first reading or viewing, we typically do not find ourselves able to imagine just how it might be that the witches’ prophecy could be fulfilled, consistent with other aspects of this particular imaginative project (e.g., that human reproductive biology is the same in the world of Macbeth as it is in the actual world). Or, at least, we cannot do so – until Shakespeare shows us how. He does this by ultimately presenting us with Malcolm’s wooden tactics and Macduff’s claim to be from his mother’s womb untimely ripped. But we are not supposed to make use of those rather improbable ways of satisfying the witches’ words until near the end of the play. So for most of the play, we are not to imagine any such way. Nonetheless, under normal circumstances, we quite readily make inferences along the following lines:

I cannot imagine how it could be possible that p.

Therefore, not-p.

So it might seem odd that we are so epistemically reticent with regard to propositions like “the witches’ prophecy will come true”.

The close resemblance between belief and imagination in a single code architecture can play a useful role here, however. For we can consider the existence of epistemically unusual, though far from extraordinary, circumstances in which we do not find ourselves making this inference from apparent unimaginability to the outright falsity of what cannot be imagined – namely, those in which one has independent strong reasons for p. (Bits of contemporary physics which we accept on the basis of testimony may be like this for many of us.) Compare to a typical transaction in perception, in which a vision system tokens p, and this leads to the tokening of a belief with content p. Under normal circumstances, this transition happens automatically, even unconsciously. But when we take ourselves to have very good evidence against p – suppose that p is the proposition “a large purple elephant has just appeared in the room” – then we do not seem typically to token p itself in our beliefs, but rather only “it seems that p” or “it seems visually that p”. We take it that this representation-downgrading typically happens fairly automatically when it occurs. For example, when we see the Mueller-Lyer figure while we know that the two lines are the same length, we unconsciously and effortlessly find ourselves in a state of the two lines merely seeming to be of different lengths. So our hypothesis is that the same mechanism of epistemic deflation that avoids cognitive meltdown in cases like perceptual illusions, also serves to help defer blockage in some cases that prima facie should put us into a state of imaginative blockage.