Imagination and Belief in Action

Abstract

Imagination and belief are obviously different. Imagining that you have won the lottery is not quite the same as believing that you have won. But what is the difference? According to a standard view in the contemporary debate, they differ in two key functional respects. First, with respect to the cognitive inputs to which they respond: imaginings do not respond to real-world evidence as beliefs do. Second, with respect to the behavioural outputs that they produce: imaginings do not motivate us to act as beliefs do. I argue that this view is mistaken in one important respect. The distinction between imagination and belief does lie at the functional level; but the relevant functional difference does not concern behavioural outputs – since, in spite of appearances, imaginings and beliefs motivate us to act (and react) in the same ways. To see the difference, we need to focus on the inputs side – and, relatedly, on the sorts of inferential relations that imaginings and beliefs bear to each other. I show that this view does not have the absurd consequences that it may prima facie seem to have; on the contrary, it has important implications for our understanding of how the mind works.

Imagination and belief are obviously different. Imagining that you have won the lottery is not the same as believing that you have won. But what precisely is the difference? According to a rather standard view, they differ in two key functional respects: (i) with respect to the cognitive inputs to which they respond – imaginings do not respond to evidence and reasons as beliefs do; and (ii) with respect to the behavioural outputs that they produce – imaginings do not motivate us to act as beliefs do. I argue that this view is mistaken in one important respect. The distinction between imagination and belief does lie at the functional level, but the relevant functional difference does not concern behavioural outputs – since, in spite of appearances, imaginings and beliefs motivate us to act (and react) in the same ways. To see the difference, we need to focus on the input level – and, relatedly, on the sorts of inferential relations that imaginings and beliefs bear to each other.

In §1, I introduce the standard view, and I argue that this view fails to deal adequately with a range of cases where imaginings seem to motivate action jointly with our desires, in the same way in which beliefs motivate. In §§2–3, I turn to cases where imaginings do not motivate – such as cases of daydreaming and engagement with fiction – which are alleged to show a motivational difference with respect to beliefs. But I argue that such cases do not reveal any difference, since the same factors that prevent imaginings from motivating in these cases are factors that, all else being equal, would also prevent beliefs with the same contents from motivating. On this basis, in §4 I conclude that imaginings and beliefs dispose us to act under the same conditions: they may contingently differ with respect to the satisfaction of such conditions, but they do not differ in motivating power. The critical difference between them should be sought at the inputs level.

1 The Standard View (and Where it Goes Wrong)

1.1 Classic Imaginative Explananda

In asking how imagination differs from belief, it is natural to start with cases of daydreaming, engagement with fiction, pretence, and modal thinking. These are indeed what we may call ‘classic imaginative explananda’: the phenomena that most obviously require us to postulate an imaginative capacity somehow distinct from belief in order to be properly understood. According to the dominant view in the contemporary debate, the relevant distinction here lies at the functional level. The cognitive states underlying our engagement with daydreams, fiction, and pretence, may have precisely the same contents as our beliefs, but differ from them in two key functional respects: (1) with respect to the cognitive inputs to which they respond: they do not respond to evidence and reasons as beliefs do; and (2) with respect to the behavioural outputs that they are able to produce: they do not motivate us to act as beliefs do.¹ The view that imagination and belief differ in both these respects is widely accepted, almost as a matter of fact – Nichols (2006) described these two differences as “central facts about the propositional imagination that have shaped almost all theorizing in the recent literature” (Nichols 2006: 6). Let’s then call this ‘the Standard View’ of imagination and belief (for influential endorsements, see e.g. Nichols and Stich 2000; Currie and Ravenscroft 2002; Van Leeuwen 2009, 2016; Sinhababu 2012; Liao and Doggett 2014; Kind 2016a, b²).

Some controversies concerning the extent of the motivational difference between imagination and belief arise among advocates of the Standard View in relation to cases of pretence, where imaginings are associated with relevant action outputs. Friends of the traditional Humean Theory of Motivation observe that ‘associated with’ does not mean ‘causally responsible for’: belief is the only cognitive state that, jointly with desire, can cause and rationalize action, and it does so also in cases of pretence, where the role of imaginings is merely indirect (see e.g. Nichols and Stich 2000). Friends of the so-called ‘Imagination-as-Motivation Theory’, by contrast, contend that a proper understanding of pretence requires us to credit imagination with an autonomous motivating power that can be exerted with no mediation of beliefs: most typically, on these views pretence is explained as driven by pairs of belief-like and desire-like imaginings (see e.g. Currie and Ravenscroft 2002; Gendler 2006; Doggett and Egan 2007³).

Not surprisingly, though, none of these theorists take pretence in itself to prove that imagination has the same motivating power as belief. Indeed, they say, beliefs typically motivate in conjunction with real desires, not with imaginative ones; moreover, they motivate action across many different contexts. If imaginings lead to action only in conjunction with imaginative desires and in the context of pretence, their motivating power would surely be different from that of beliefs: if anything, more limited. Cases of pretence can at most challenge the Humean Theory according to which motivating power is an exclusive prerogative of belief, but they do not ultimately challenge the Standard View according to which belief has a distinctive motivating power that sets it apart from imagination.

What is more surprising, is that the general consensus on the Standard View remains unshaken also in the face of various other phenomena that have recently been suggested to be best understood in imaginative terms: phenomena that are pervasive in many contexts of our lives and that include, most importantly for us here, also a large number of (non-pretence) actions. Recently, there has been indeed a trend towards recognizing that the role of imagination in our lives is not limited to episodes of daydream, engagement with fiction, pretence, and modal thinking, but is manifested also in a variety of other domains, where imaginings combine with our (real) desires, motivating us to act in relevant ways. I will call these allegedly imagination-driven actions ‘novel imaginative explananda’ – and argue that insofar as at least some of them are indeed imagination-driven, they challenge the standard view that imagination and belief differ in motivating power.

1.2 Novel Imaginative Explananda

Many examples of the imagination-driven actions I am referring to can be found in the recent literature on ‘belief-behaviour mismatches’. Philosophers have been devoting increasing attention to cases of actions that are not easily explained in terms of what the agents believe, and for many such cases imagination has been suggested to be the state that plays the relevant motivating role. Cases of this sort include: superstitious actions, so-called ‘expressive actions’ (i.e. actions expressive of an emotion), ritual and symbolic actions, actions performed out of self-deception, out of delusion, and out of implicit bias. What all these otherwise disparate actions have in common, is the fact that the cognitive states which, jointly with desires, motivate them, seem to lack such key features of belief as sensitivity to evidence and coherent inferential integration with the rest of subject’s doxastic states: cognitive states like superstitions, self-deceptions, delusions, etc. typically do not respond to evidence, reasons, and coherence constraints in the ways and to the extents to which genuine beliefs do.⁴

This is why for all the actions in question non-doxastic explanations in terms of imagination-desire pairs have been suggested, according to which, for instance: the superstitious agent who knocks on wood does not really believe that by so doing she will bring about good luck – she just imagines that, and this imagining (together with a real desire for good luck) is what motivates her action (Currie and Jureidini 2004); the enthusiastic football fan who loudly encourages her favourite team while watching a match on TV does not really believe the players to hear her, she just imagines that – and this imagining (together with a real desire to encourage the players) is what motivates her (Velleman 2000); the self-deceived agent who obstinately denies her partner’s infidelity does not really believe that her partner is faithful, she just imagines that, and this imagining (together with a real desire to tell the truth) is what motivates her denial (Gendler 2007); the Capgras patient who calls the police to denounce the replacement of her family by impostors does not actually believe that her family has been replaced by impostors, she just imagines that – and this imagining (together with a real desire to denounce impostors), is what motivates her call (Currie 2000); the implicitly biased employer who discriminates women’s CVs does not actually believe women to be worse candidates than men, he just imagines that – and this imagining, together with a real desire to select the best candidates, is what motivates him (Sullivan-Bissett 2018); and so on and so forth, for the other examples I mentioned.

Although admittedly these sorts of explanations in imagination-desire terms are not uncontroversial, for at least some cases they are pretty intuitive, and indeed widely accepted. So, for instance, no one would question that the enthusiastic football fan just imagines, and does not believe, that the players she’s watching on TV can hear her cheering; and many authors have endorsed Velleman’s view that, in this and in similar cases of expressive actions, such imagining is (the cognitive part of) what motivates action (see e.g. Currie and Ravenscroft 2002; Gendler 2006; Van Leeuwen 2009, 2016; Currie and Ichino 2012; Gerrans 2014; Ichino 2018).

Moreover, even when imagination-desire explanations are less intuitively appealing, I take them to enjoy abductive advantages over the alternative explanations that have been offered in the recent literature, which have significant theoretical and ontological costs.⁵ Whilst granting that more would need to be said to properly establish this, anyway, for the sake of my argument I do not need to defend imagination-desire explanations for all the cases of belief-behaviour mismatches I mentioned. Here it is enough for me to assume the existence of a number cases where imaginings motivate action jointly with real desires, as beliefs do – which I have shown to be something that, on some or some other grounds, many authors nowadays accept.

What is not typically accepted, on the other hand, is the view that I defend on the basis of those cases: the view according to which those cases reveal imagination’s motivating power to be the same as belief’s. Indeed, it is commonly argued, even acknowledging a variety of cases where imaginings motivate in conjunction with our desires, as beliefs do, there still remain many cases where imaginings do not motivate in such ways – daydreams and responses to fiction being the most obvious examples.

I will argue that this line of reasoning is flawed. Once we recognize the need to credit imagination with some motivating power on its own, cases where imaginings fail to motivate do not in themselves reveal such power to be different from that of beliefs, given that there is plenty of cases where beliefs, too, fail to motivate. A proper comparison between imagination and belief’s motivating powers requires us to look closely at the conditions under which such powers are – or fail to be – manifested; and once we do that, we discover that the conditions in question are precisely the same for belief and for imagination. Differences in belief and imagination’s behavioural manifestations are due to contingent differences in the ways in which those conditions are satisfied; not to intrinsic differences in their underlying behavioural dispositions.⁶

2 Velleman on the Motivational Equivalence of Imaginings and Beliefs

The line of argument I just outlined for the motivational equivalence of imagination and belief was famously put forward by Velleman (2000) – who was arguably the first to introduce in the recent debate a number of novel imaginative explananda involving motivation via imagination-desire pairs:

I have now introduced several categories of examples that feature motivation by imagining. These examples show that imagining that p and believing that p are alike in disposing the subject to do what would satisfy his conations if p were true, other things being equal. Admittedly, the examples have also suggested that other things are rarely equal between cases of imagining and believing, and hence that the actual manifestations or these states are often different. But these differences do not undermine my thesis. After all, belief itself cannot be characterized in terms of the behaviour that it actually causes, since most beliefs cause no behaviour at all, and the same belief will cause different behaviour in different psychological contexts. Belief can be characterized only in terms of its disposition to produce behaviour under various conditions, such as the presence of a relevant conation and the requisite motor skills, and the absence of conflicting motives and inhibitions. The examples suggest that imaginings can be characterized as having the same conditional disposition as belief; the only differences have to do with the satisfaction of the associated conditions. (Velleman 2000: 271-272).

Velleman’s argument, however, is incomplete in one important respect, and requires some development.

To begin, let’s unpack two importantly different claims that are at stake in it:

• (i) There are a number of cases where an agent who imagines that p is motivated to act in the same ways in which she would act if, all else being equal, she believed that p.

• (ii) In any possible case, an agent who imagines that p is motivated to act in the same ways in which she would act if, all else being equal, she believed that p.

Novel imaginative explananda like those I discussed in §1.2 support claim (i): insofar as (at least some of) the relevant imagination-desire explanations are right, those are cases where an agent’s imagining that p motivates her to act in ways that would promote the satisfaction of her desires if p were true, which is the way in which a belief that p would typically motivate. But, of course, (i) does not entail (ii), which is what ought to be true if imaginings really had the same motivating power of beliefs.

Velleman’s critics blame him for switching too easily from (i) to (ii). As Van Leeuwen (2009) puts it: “Velleman mistakenly generalizes from cases in which imagining that p and believing that p happen to yield similar behaviour, to the conclusion that this holds for all cases” – a conclusion which Van Leeuwen, in line with the Standard View, considers “indefensible in virtue of its absurd consequences” (Van Leeuwen 2009: 232; see also Currie and Ravenscroft 2002: 117–119). Indeed, we have already seen how the standard argument goes: even granting the existence of cases where imaginings motivate in conjunction with our desires, as beliefs do, there still remain many cases where imaginings do not motivate in such ways – daydreams and responses to fiction being the most obvious examples.

However, if it is true that Velleman jumps too quickly from cases where imaginings motivate as beliefs to the conclusion that their powers are the same, the sort of standard criticism I just outlined does not do better: it jumps too quickly from the consideration of cases where imaginings do not motivate to the conclusion that their motivating power differs from that of belief – overlooking the fact that there are plenty of cases where beliefs do not motivate, either.

To see this point more clearly, consider a paradigmatic case of non-motivating imagining: a case of daydreaming. I am working at my desk and, in the midst of procrastination, I start hanging around on on-line newspapers. An article on the Royal Wedding prompts me to imagine that I am the Duchess of Cambridge, alias Kate Middleton, just about to get married to Prince Williams… We all agree that my imagining in a case like this does not motivate me to act. I do not wear an elegant white dress, I do not look for my chauffeur in order to be taken to Westminster Abbey, I do not prepare for the honeymoon; indeed, I do not do anything but sit here, staring into space. However, it is one thing to agree on this; it is quite another thing to conclude that my imagining in this case does not motivate me to act as I would if I believed what I imagine to be true. This conclusion is not as obvious as it prima facie seems to be.

The fact is that it is not so obvious how I would act if I believed that I was the Duchess of Cambridge on my wedding day. To be honest, if I really believed that, I would probably try to escape before it was too late: the idea of marrying Prince William does not sound really appealing to me! And even granting that I were happy with this wedding… Would I wear a white dress? I probably would, if I wanted to conform to traditional practices and I believed that brides traditionally dress in white. Though I wouldn’t if, instead, I wanted to surprise everyone with a bohemian style, or if I believed that brides traditionally dress in pink. Would I look for my chauffeur to be taken to Westminster Abbey? I probably would, if I believed that the ceremony will take place there. Though I wouldn’t if, instead, I believed it to take place in a London registry office. And what if I got up in the morning with one of my awful migraines? In such a case, I might well have to remain in the darkness of my room for a while, hoping that it passes and it doesn’t completely ruin my day, as only migraines can do.

What these hypothetical scenarios highlight is a basic feature of belief’s motivating power, with which we are all familiar: whether beliefs motivate us to act and what sort of actions they motivate, depends on a number of factors concerning our internal and external environment – factors which include, but are not exhausted by, the presence of relevant desires. Belief’s connection to action is indeed a dispositional connection, which is expected to manifest itself: (a) in the presence of the relevant desires, and (b) if a number of other conditions are also satisfied – while failures to satisfy such conditions can excuse lacks of behavioural manifestations, without thereby compromising the holding of the disposition itself.

This is crucial to keep in mind when we compare belief’s and imagination’s motivating powers. It is true that the existence of cases where imaginings motivate in ways that would promote the satisfaction of our desires is not enough to conclude that imagination’s motivating power is the same of belief. But in order to support this conclusion we do not need to show that imaginings always motivate us to act in ways that would promote the satisfaction of our desires. We need to show that they motivate us to act in such ways under the same conditions under which beliefs do so.

Admittedly, Velleman does not do enough to show this. Most of his discussion focuses on cases where imaginings’ motivating power is actually manifested, while to show that their conditional disposition to action is precisely the same as beliefs’ he should have paid more attention to cases of lack of behavioural manifestation. More precisely, he should have shown that whenever an imagining that p does not motivate actions that would promote the satisfaction of the agent’s desires, this lack of behavioural manifestation is excused by some condition C which (all else being equal) would also excuse the lack of behavioural manifestation of a belief that p. This is what I aim to show in the next section, thereby providing the missing piece necessary to support Velleman’s conclusion.

Of course, I can’t possibly consider every possible case where imaginings do not motivate. But I will identify some basic conditions that excuse the cases of non-motivating imaginings typically pointed out by Velleman’s opponents (i.e., notably, cases of daydreams and responses to fiction), as well as all other cases of non-motivating imaginings I can think of. Hence, I will conclude by shifting the onus of proof onto our opponents.

3 Excusing Conditions

There are at least two kinds of factors that are typically responsible for the lack of behavioural manifestation of our imaginings in paradigmatic cases of daydreams and fiction consumption. First, factors concerning the meta-cognitions that typically accompany our imaginings in such cases: the beliefs and desires that, in such cases, we typically have about our imaginings. Second, factors concerning the kinds of propositional contents that our imaginings typically have in such cases, and the inferential networks into which they are integrated. I shall discuss these two kinds of factors in turn, arguing that, all else being equal, they would also excuse the lack of behavioural manifestation of beliefs with the same contents.⁷

3.1 Meta-Cognitive Factors: The ‘Meta-Cognition Condition’

Why, when I daydream that I am the Duchess of Cambridge, I do not act upon this imagining? If you asked me on the spot, I would probably tell you that this is because I believe that it is just a daydream, and I do not see any point in acting upon it. Arguably, there would be some truth in this response. When we have a daydream with content p, we are often aware that p is something that we merely imagine and do not believe to be true. And – partly because of this – we desire not to act upon p, since we know that the best chances to satisfy our desires rest on our representing reality accurately, and we want our actions to be realistically purposive. Simplifying a bit, let’s then say that a standard case of daydreaming is one where the agent imagines that p, and:

• (a) Believes that she does not really believe, but just imagines, that p;

• (b) Desires not to act upon her imagining that p.

These meta-beliefs and meta-desires concerning the agent’s imagining are both factors that prevent the imagining itself from motivating⁸ – for the reasons just seen: the meta-belief ‘tags’ the imagining as ‘disconnected from evidence/reality, hence unsuitable for effective action’, and the meta-desire explicitly sets ‘not acting upon the imagining’ as a goal. What I contend is that either of these two meta-cognitive factors would also, all else being equal, prevent a belief that p from motivating.

Admittedly, if we stick to the Royal Wedding case, the scenario we get by keeping everything the same as the daydream scenario apart from the fact that I believe, and not imagine, that I am the Duchess of Cambridge, is somewhat improbable. This would be a scenario where I believe that I am the Duchess of Cambridge, though I have also for some reason come to believe that I don’t believe but just imagine that, so I don’t want to act upon this imagining, and I’m rather sat here, working on my paper. No doubt, improbable. In cases of ‘exotic’ daydreams like this, indeed, the circumstances in which we imagine are not likely to be equal to the circumstances in which we would find ourselves if we believed what we imagine to be true. But improbable does not mean impossible. What matters to my argument is that you grant the following counterfactual claim: if that scenario occurred, the behavioural consequences of my belief that I am the Duchess of Cambridge would not be different from the behavioural consequences of the corresponding daydream (i.e.: if that scenario occurred, I would fail to act upon my belief as I do fail to act upon my imagining in the daydream scenario). This is plausible: all else being equal, if one believes that p but for some reason comes to believe that she does not really believe that p, and/or comes to desire not to act upon p, then we may well expect her to fail to act upon her belief that p.

Examples of the first condition – i.e. of a case where one believes that p, but comes to believe that she does not believe that p, hence fails to act upon p – may look a bit tricky to find, since we tend to assume that ‘we know what we believe’, crediting ourselves with privileged introspective access to our own minds. However, the question whether we do actually have such a privileged access is controversial; and anyway – even granting that we do – few would deny that mistakes are possible, or even common.

Take for instance Gloria, who believes that Ben is cheating on her, having plenty of evidence about this, and feeling indeed very upset. As Gloria expresses this belief, Ben seeks to persuade her that it is false, arguing that she is just paranoid and that, being blinded by obsessive jealousy, she mistakes her fantasy for truth. Let’s also assume that, given his excellent rhetorical skills, Ben manages to brainwash Gloria into thinking that what she takes to be a belief is actually a mere imagining. Of course, you may think that in so doing Ben is simply changing Gloria’s first-order belief: his ‘brainwashing’ means that, for a little while at least, she is no more believing that Ben is cheating on her. But this is not the only possible description of what goes on here. Another possibility is that, given the abundant evidence she has about Ben’s infidelity, Gloria keeps believing that he is cheating on her (and she does indeed keep feeling upset); although she mistakenly tells herself that she doesn’t – believing of her belief that it is just a paranoid imagining (as Ben insisted). This is not an incoherent scenario; nor, arguably, a psychologically implausible one.⁹ And my claim is that in a scenario like this – where one believes that p, but mistakenly believes that she doesn’t believe so – one may well fail to act upon her first-order belief in various ways. If you asked Gloria about her relationship, for instance, she may sincerely deny any infidelity on Ben’s part.

Examples of the second condition – i.e. of cases where a lack of behavioural manifestation of a belief that p is excused by the desire not to act upon p – are much easier to find. Take cases of pretence. The child who pretends that she is a cat, and responds to your questions only by meowing, obviously does still believe that she is a child and that she can talk properly. But, for the duration of the game, she desires not to act upon these beliefs (and to act upon her imaginings, instead).

Eventually, then, it seems that we can formulate the following excusing condition that holds both for belief’s and imagination’s lack of behavioural manifestation. The meta-cognition condition: both a belief that p and an imagining that p motivate the agent to act in ways that would promote the satisfaction of her desires if p were true, unless the agent: (1) believes that she doesn’t really believe that p, or (2) desires not to act upon her representation of p. If either (1) or (2) are the case, this may excuse a lack of behavioural manifestation of the belief/imagining in question, without thereby compromising the holding of its behavioural disposition. Since (1) and (2) are typically the case when we daydream, we can say that the lack of behavioural manifestation of our imaginings in such cases does not, in itself, reveal the difference between imagination’s and belief’s motivating powers that Velleman’s opponents point out.

Of course, this excusing condition – as indeed the other one I will introduce below – is itself defeasible: whilst the holding of either (1) or (2) can excuse lack of behavioural manifestation of a given belief or imagining, it can also happen that the inhibiting force of (1) and (2) is itself inhibited by the presence of other factors. So, for instance, competent pretenders do typically believe that they do not believe, but only imagine, the contents of their pretence imaginings – thereby satisfying disjunct (1); but they do nonetheless act upon such imaginings. This may be due to a variety of reasons (including the fact that they have an explicit desire to act upon what they merely imagine, rather than upon what they believe¹⁰), which, however, I won’t discuss here. For the sake of my argument, it is enough to have shown that the presence of relevant meta-cognitions can be what prevents imaginings from motivating action; and that, all else being equal, it can do the same with beliefs – as my meta-cognition condition suggests.

One here might worry that this meta-cognition condition is in one important respect vacuous, since in cases of imaginings it is necessarily satisfied by virtue of its first disjunct. This would be in line with what some meta-representational accounts of imagination and pretence suggest: imagining necessarily involves (if not even amounts to) meta-representing one’s representations as being imaginative – more precisely, meta-believing that one’s own representations are imaginative (cf. Leslie 1994; Nichols et al. 1996). If this were indeed the case, then we would seem to have an intrinsic motivational difference between imagination and belief, since imaginings would necessarily involve something – i.e. the meta-belief that one is imagining – which undercuts motivational force.¹¹

But why endorse this sort of meta-representational account of imagination? The default position should rather be to reject it, because postulating unmotivated necessary connections between mental states is undesirable (see Currie 1998: 41–42). And the main motivations adduced by advocates of that account – based on developmental evidence about the co-emergence of pretence behaviour and some kinds of mindreading capacities – have been widely criticised, both on empirical and on conceptual grounds (see e.g. Harris 1995; Currie 1998; Nichols and Stich 2000). Without postulating any necessary connection between imagining and meta-believing that one is imagining, on the other hand, my account simply points out a number of cases in which such connection is likely to hold, noting that imaginings in cases of daydreaming are often accompanied by relevant meta-beliefs.

My opponent may insist that this is still enough to defeat my view: if imaginings are typically accompanied by beliefs that defeat their motivating force, doesn’t this prove a motivational difference between them and beliefs? But here note that I never said that imaginings are typically accompanied by meta-beliefs that defeat their motivational force; what I said is that some sorts of imaginings – such as daydream imaginings – often are. This is compatible with there being many instances of imaginings that are not accompanied by such meta-beliefs, instead. The ‘novel imaginative explananda’ that I discussed in §1.2, for example, typically involve imaginings of this latter sort – which do, indeed, motivate us to act.¹²

A different sort of opponent may challenge my view from an almost opposite angle: rather than questioning the claim that some instances of imaginings may not be accompanied by the relevant meta-beliefs, she may observe that I didn’t provide enough reason to support the claim that daydream imaginings themselves are typically accompanied by such beliefs. In other words: far from being necessarily present, meta-beliefs about our own imaginings may not be even common. So, the problem would not be that my meta-cognition condition is necessarily met by all imaginings as such. But, on the contrary, that I haven’t provided sufficient argument for the claim that the kinds of imaginings I’m concerned with – such as daydream imaginings – do indeed meet such condition, being thereby excused for failing to motivate, as I claim them to be.

My reply to this is twofold. First, I agree that the claim that our daydream imaginings are often accompanied by the meta-belief that we are merely imagining is an empirical claim that, as such, would require systematic data to be decisively established. Whilst I’m not aware of any such data, however, I take some solid support to that claim to come from the observation that in cases of daydream like my Royal Wedding case – which are (to some extents at least) deliberate, conscious, and ‘exotic’ in content – daydreamers themselves have plenty of evidence about the fact that they are (merely) imagining and not believing the content of their daydreams. And the fact that a subject has plenty of evidence for a given proposition p, is a good reason to ascribe her the belief that p (in this case, a meta-belief).¹³

That said, I am also happy to grant the existence of daydreams where my meta-cognition condition does not hold. We can, and surely sometimes do, daydream without believing that we are daydreaming (as well as, indeed, without desiring not to act on our daydreams). But when this happens – and if nothing else prevents us – then we are indeed motivated to perform the relevant actions. So, for instance, when we get carried away by our daydreams and ‘temporarily forget’ that we are just imagining, or when, being alone, we lack social inhibitions towards unrealistically motivated behaviours, we may well end up acting out our imaginings in a number of ways, moving around and even talking aloud accordingly.¹⁴

The specification that we will do that only insofar as nothing else prevents us is important, though. Indeed, beyond our inhibiting meta-cognitions, there are also other factors that may prevent what we imagine (in cases of daydream, as well as in other cases) from motivating us: factors that do not directly concern the beliefs and desires that we have about our imaginings, but have more to do with the propositional contents that such imaginings themselves typically have, and the inferential network of cognitions into which they are integrated.

3.2 Contents and Inferential Network: The ‘Practical Indeterminacy Condition’

A further reason why, when I daydream that I am the Duchess of Cambridge, I do not act upon this imagining, is this: nothing in its content tells me what I should do in order to act in the relevant ways (i.e. in ways that, if this imagining were true, would promote the satisfaction of my desires). Even if wanted to act, I would – so to say – lack the relevant instructions.

A comparison with paradigmatic cases of belief-desire motivation will make this point clear. A notable feature of beliefs is that they are inferentially integrated with each other within a (to some extents at least) holistically coherent system. And the whole system of beliefs to which a belief that p belongs is crucial in determining what sort of actions, if any, p will motivate in conjunction with the relevant desires. In particular, there seem to be one specific type of beliefs which must be present in the system in order to allow the other beliefs to manifest their motivating potential: beliefs about what sort of actions we should perform in order to satisfy our desires – i.e. beliefs of the form [I can obtain ϕ if I perform action A] (where ϕ is something I desire). If the belief that [today is my wedding] motivates any action at all, this is (also) because I have a number of wedding-related desires – e.g. the desire [to dress properly], [to be on time for the ceremony], etc. – and a number of beliefs about how to satisfy them – e.g. the belief that [I can wear my white dress if I take it from the wardrobe], that [I can be on time if I book a cab for 11 am], etc... Without beliefs of this form, my belief that today is my wedding would fail to motivate me, even in the presence of relevant desires.

The point is that imaginings of this form are typically absent in paradigmatic cases of daydreaming. Daydreams are often like that: little more than fleeting scenarios – which remain, to a large extent, isolated and indeterminate in their contents.¹⁵ For a moment I imagine that I am the Duchess of Cambridge, walking down the aisle, and that’s all. This propositional content is quite self-standing. I may, of course, elaborate on it, adding a number of other propositions. But the propositions I add are not typically the ones that would be needed to practically implement desire-satisfaction. This is, in a sense, a privilege of daydreaming: we can wear wonderful dresses without having to think about trying them on beforehand, and enjoy wonderful parties without having to think about how to get there on time. And the indeterminacy of our imaginings with respect to such practical details – that is, the fact that our patterns of imaginings when we daydream do not typically include imaginings whose contents specifically indicate how to satisfy our desires – is another reason why daydreams do not typically move us to act.

So, we seem to have found another excusing condition that holds both for belief’s and imagination’s lack of behavioural manifestation. Call it the practical indeterminacy condition: both a belief that p and an imagining that p motivate the agent to act in ways that would promote the satisfaction of her desires if¹⁶: (1) p is a proposition of the form [I can obtain ϕ if I perform action A] (where ϕ is the content of one of the agent’s relevant desires), or (2) p is inferentially integrated into a network of believed/imagined propositions which includes propositions of that form. If this neither (1) nor (2) are the case, this can excuse a lack of behavioural manifestation of the belief/imagining in question, without thereby compromising the holding of its behavioural disposition. Since neither (1) nor (2) are typically the case when we daydream, we can say that daydreams do not in themselves reveal the difference between imagination’s and belief’s motivating powers that they are commonly taken to reveal.

This condition highlights the crucial role that the larger network of beliefs/imaginings to which a given belief or imagining belongs has in determining its behavioural manifestations. In particular, it highlights the fact that whether or not an imagining or a belief motivate, and what sorts of actions they motivate, crucially depends on the consequences that we (do or do not) draw from them. This is important not only to understand why imaginings in cases of daydream do not motivate any action at all (while we would expect beliefs with the same contents, against the same background of desires, to motivate); but also to understand why in cases where imaginings do motivate, the ways in which they motivate look somehow sui generis (i.e. different from the ways in which we would expect beliefs with the same contents, against the same background of desires, to motivate).

Take, for instance, one of the imagination-driven kinds of actions introduced in §1.2: superstitious actions. There is little doubt that such actions are somewhat sui generis. My superstitious grandmother, who avoids travelling on Friday the 13th like the plague, does not warn me against travelling on that day – even though, presumably, she desires for me to be safe as she desires to be safe herself. If she really believed that travelling on Friday the 13th is dangerous, then we would expect her to warn me, too, against travelling.

The reason why her superstitious imagining does not motivate her to do that, arguably, is that she does not draw from it the same consequences that she would draw from a belief with the same content. If she believed that [travelling on Friday the 13th is objectively dangerous], she would also come to believe a number of propositions that obviously follow from the conjunction of this belief and the various other beliefs that she has about dangers, warnings, moral duties, etc. So, she would come to believe not only that travelling on Friday the 13th is dangerous for her (hence that she should avoid travelling that day), but also that travelling on Friday the 13th is dangerous for me (hence that I should avoid travelling that day as well, and that if I am not aware of this, then she should warn me). These are indeed obvious consequences that follow from the proposition that [travelling on Friday the 13th is objectively dangerous]; and believing that proposition would normally induce an agent to believing such consequences.

Imagining that [travelling on Friday the 13th is objectively dangerous], on the other hand, does not necessarily induce my grandma to imagine the obvious consequences of this proposition. When we imagine a proposition, we are – to some extents at least – free to choose what consequences we draw from it (and represent in further imaginings), and what consequences we ignore, instead. The fact that my grandma’s superstition is an imagining and not a belief, then, explains why she only draws from it the obvious consequences concerning her own danger, and not the equally obvious consequences concerning my own danger. And this, in turn, explains why the imagining that travelling on Friday 13th is dangerous does not motivate her to act as we would expect her to act if she held a belief with the same content. The fact that such an imagining and belief motivate different actions, as it turns out, is due to contingent differences in the inferential networks to which they belong, not to their having intrinsically different motivating powers.

4 Conclusion

I have identified two factors that seem to be responsible for the fact that, in paradigmatic cases of daydreaming, an imagining that p will typically fail to motivate actions that would promote the satisfaction of the agent’s desires if p were true. First, the presence of inhibiting meta-cognitions: when I imagine that p in the context of a daydream, often I also believe that I do not really believe, but just imagine that p, and I desire not to act upon such imagining. Second, the absence of imaginings about the actions that I should perform in order to satisfy the relevant desires: daydreams are typically indeterminate in some key practical respects. As we have seen, even just one of these factors would be enough to also prevent a belief that p from motivating in the relevant ways (all else being equal).

I contend that these same factors are also typically responsible for the lack of behavioural manifestation of other paradigmatic cases of imaginings – such as imaginings in response to fiction.¹⁷ I can, of course, think of cases where the lack of behavioural manifestation of an imagining is determined by some different factors (e.g. physical impairment); but these are always factors that, all else being equal, would also excuse lack of behavioural manifestation of corresponding beliefs. What I cannot think of, indeed, is a case in which an imagining that p fails to motivate, and this lack of manifestation is not excused by a factor that, all else being equal, would also excuse the lack of behavioural manifestation of a belief that p. As long as no one comes up with such a case, I conclude that imagination and belief dispose us to act in the same ways, and do that under the same conditions; they may contingently differ with respect to the satisfaction of such conditions, but they do not differ in motivating power.

4.1 How Do Imagination and Belief Differ, Then?

I criticised the Standard View according to which imagination and belief differ both with respect to cognitive inputs and with respect to behavioural outputs, arguing that this latter behavioural difference does not actually hold. This, as I noted, suggests that the key difference between imagination and belief lies at the inputs level. Although a thorough articulation of this positive part of my suggestion goes beyond the scope of the present project, I shall conclude by considering some implications of my view that pave the way for it.

Admittedly, the claim that imagination and belief are alike in behavioural outputs does not straightforwardly imply there aren’t outputs differences of any sort between them. Yet, what I argued in this paper seems to indicate that they do not differ in emotional and cognitive outputs, either – thereby placing the weight of the difference on the inputs side (broadly construed in the way I am about to outline).

Consider emotional outputs first. Those who argue that there is a substantial difference between imagination-driven and belief-driven emotional experiences, often appeal to an alleged difference in the motivational force associated to these two sorts of experiences. When one engages imaginatively with Psycho, for instance, her ‘fear’ for Marion does not motivate her to call the police as she would do if she feared for someone she believes to exist. And, as Walton famously put it: ‘Fear emasculated by subtracting its distinctive motivational force is not fear at all’ (Walton 1990: 202).

However, the motivational force of emotional states arguably piggybacks on the motivational force of the conative and cognitive states from which they arise: one who fears for an intended victim would not call the police unless she had a desire/desire-like state to help the victim (rather than, say, to run away in a safe place), and a belief/belief-like state according to which a good way to help the victim is to call the police (rather than, say, try to directly intervene herself). My conclusion that imaginings and beliefs have the same motivational powers upon action, then, suggests that their emotional powers are also the same.

As to imagination and belief’s cognitive outputs, it is widely agreed that both imagining and believing a given proposition dispose us to draw theoretical inferences from it, thereby coming to imagine/believe various of its consequences.¹⁸ Importantly, this does not mean that we draw the same inferences from a proposition irrespective of whether it is believed or imagined. As I noted in §3.2, when we imagine a proposition we are much freer to choose what consequences we draw from it and represent in our further imaginings, than we are when we believe it. So, for instance, if during a game I pretend to take a shot at you, we will both spontaneously infer that now you are wounded (imagining this to be true in the game), but we may choose not to imagine that you will remain infirm for a while – given that, for the sake of the game, it is rather better to ‘allow’ you to be quickly back in full form. A decision of this sort, on the other hand, would not be possible if we believed that you have been shot. Whilst factors such as interest and attention surely play a role in determining what inferences are deployed in different cases, ignoring the obvious consequences of something we believe cannot just be a matter of conscious deliberation.

But the difference between belief and imagination that we are observing here does not seem to concern primarily, nor essentially, their cognitive outputs. As it turns out, indeed, the cognitive outputs of our beliefs and imaginings are themselves new beliefs and imaginings that we form on their basis; and the difference that emerged has to do precisely with the constraints that govern this formation process – constraints governing the cognitive inputs to which imaginings and beliefs responds. The relatively larger freedom that we have in forming new imaginings with respect to new beliefs depends on the fact that imaginings are not constrained by real-world evidence (nor by reasons/coherence constraints) in the same ways in which beliefs are.

To conclude, it should be noted that my talk in terms of an ‘inputs difference’ between imaginings and beliefs is rather coarse-grained. Strictly speaking, what I pointed out is not just one functional difference, but two: one concerning the inputs responsible for the formation of imaginings and beliefs; and one concerning the inferential relations that imaginings and beliefs bear to each other. But these differences are two aspects of one and the same thing: belief’s sensitivity to evidence. And what I have argued in this paper suggests that it is this sensitivity to evidence – rather than its motivational (and other) powers – that sets belief apart from imagination.