The Mirror Account of Hope and Fear

Abstract

I provide a unified account of hope and fear as propositional attitudes. This “mirror account” is based on the historical idea that the only difference between hope and fear is the conative attitude involved, positive for hope and negative for fear. My analysis builds on a qualified version of the standard account of hope. The epistemic condition is formulated in terms of live possibility and the conative according to a non-reductive view on desire and aversion. The account demonstrates the theoretical fruitfulness of accepting Jack M. C. Kwong’s distinction between hope and fear as propositional attitudes and experiential states.

1. Introduction

Contemporary philosophical literature tends to handle hope and fear as two distinct topics. ¹ Such approach contrasts sharply with a strong historical tradition, going back at least to Seneca, of treating hope and fear together. This was the approach taken by philosophers such as Aquinas, Spinoza, and Hume: ²

[Fear and hope] are bound up with one another… the two of them march in unison like a prisoner and the escort he is handcuffed to. (Seneca, 1969, 38)

Fear and hope are principal passions… in respect of good, movement begins in love, goes forward to desire, and ends in hope; while in respect of evil, it begins in hatred, goes on to aversion, and ends in fear. (Aquinas, 1967, 1585)

There is no fear without hope and no hope without fear. (Spinoza, 2004, 78)

The passions of fear and hope may arise [together] when the chances are equal on both sides, and no superiority can be discovered in one above the other. (Hume, 1969, 490)

In what follows, I resurrect this historical practice and suggest a unified account of hope and fear, one which is informed by the contemporary philosophical literature on the analysis of hope. This expanding literature is centered on the insight that the standard account of hope in terms of desire and epistemic possibility is somewhat deficient, and there have been many suggestions regarding how the analysis should be improved (Bovens, 1999; Calhoun, 2018; Chignell, 2023; Kwong, 2019; Martin, 2014; Meirav, 2009; Palmqvist, 2021; Pettit, 2004). ³

There is not much corresponding current literature that directly and exclusively concerns the analysis of fear, which is rather surprising given the fact that fear remains a common topic in political philosophy ⁴ and the philosophy of emotions. ⁵ However, going back to the late twentieth century, we find some substantial analyses that can complement the hope literature as points of departure for the present account (Davies, 1987; Day, 1970, 1998; Gordon, 1980).

Several interesting ideas concerning the connection between hope and fear are suggested or implied by the historical accounts. I want to highlight two such ideas. ⁶ The perhaps most important is the assumption that the only difference between hope and fear is that one attitude is positive and the other is negative. This assumption suggests what I call the mirror analysis thesis ⁷ (MAT):

Mirror analysis thesis: Fear and hope are to be analyzed symmetrically, and the only difference between them should be the conative attitude (positive/negative).

Another important observation from the historical accounts (see especially the quotes from Seneca and Spinoza) is that hope and fear tend to appear in tandem, so that when we hope that p we typically also fear that −p, or at least we are in a position where such fear is possible. We can call this the co-variation thesis (CVT):

Co-variation thesis: In every situation where hope is possible, so is fear, and in every situation where fear is possible, so is hope.

The unified account of hope and fear I develop in what follows is based on MAT. While my analysis will suggest that CVT holds in most circumstances, I ultimately reject its claim at generality (see Section 5).

MAT might seem unlikely to anyone familiar with the current literature on hope. As the analysis of hope grows and new and more elaborate conditions are added to the standard account, it might seem unrealistic to suppose that everything except the positive stance will also hold for fear. I will sidestep this problem by going back to the standard account. This move might seem surprising, but it is motivated by following Kwong and others in making a distinction between hope and fear as propositional attitudes, and hope and fear as experiential states (Davies, 1987; Gordon, 1980; Kwong, 2020, 2022a, 2022b). With such a distinction in place, it becomes apparent that the philosophical puzzles fueling the search for a new analysis of hope are tied to the experiential states, while the mirror account concerns the propositional attitudes.

Although I find Kwong’s distinction essential for understanding hope and fear, it is somewhat controversial. Providing a proper defense is unfortunately outside the scope of this text (for such defense, I refer the reader to Kwong, 2020, 2022a, 2022b). However, one important way of supporting a philosophical position is to demonstrate its theoretical fruitfulness, that is, its ability to solve puzzles, dispel troubling counterexamples, and enable new theoretical moves. That strategy is pursued in what follows. In other words, rather than providing external arguments for Kwong’s distinction, I demonstrate its theoretical fruitfulness by showing how it can explain puzzling cases and proposed counterexamples, and how it enables us to reconnect with the historical praxis of treating hope and fear in unison.

Also, I will not defend the standard account in its basic form. Even restricted to propositional attitudes, it still requires substantial elaboration before becoming satisfying, especially if it is to cover fear as well as hope. My point is that we can elaborate it without adding any third condition, since a satisfying analysis can be reached simply by qualifying the two conditions of the standard account.

In Section 2, I present the standard account of hope and fear, along with Kwong’s distinction between propositional and experiential hope, and some other theoretical remarks. In Section 3, I defend the epistemic condition from recent attacks and elaborate on it using the notion of a live possibility. Section 4 concerns the conative condition, and I argue that we need non-reductive notions of desire and aversion to properly separate hope from fear. In Section 5, I return to CVT and the connection between hope and far. I reject CVT as being too radical by pointing to circumstances where hope and fear do not co-vary. The text is concluded by some concluding remarks (Section 6).

2. The Standard Account of Hope and Fear

In a series of recent papers, Kwong (2020, 2022a, 2022b) has suggested that “hope” is a polysemic word, and that we must make a distinction between “hope” as a mental act or propositional attitude (“hoping” in Kwong’s terms) and “hope” as a positive emotional state (“hopefulness” in Kwong’s terms).

Hoping in this sense is merely a mental act, on par with other propositional attitudes such as believing, desiring, planning, and expecting. (Kwong, 2022b, 321)

A person is hopeful about x when… he has, on balance, positive thoughts or feelings about the likelihood that the desired outcome will be realized. (Kwong, 2020, 839, italics added)

While this distinction is new in the literature on hope, the corresponding distinction for fear was made already in the 1980s by Gordon (1980) and Davies (1987). For clarity, I will use Davies’s terminology, which distinguishes between propositional attitudes and experiential states, for both hope and fear in what follows.

It is fully possible to have propositional hope or fear without being in the corresponding experiential state. For example, I might have propositional fear that my uncareful son might break my cell phone while borrowing it, but under ordinary circumstances, I will not actually be afraid when lending it to him. Likewise, I might hope to catch an early train home, without experiencing any positive feelings about my chances of doing so.

Whether one can be in the experiential state without having the propositional attitude is a trickier issue. Kwong (2022a, 1429) suggests that hopefulness always builds on propositional hope, but Gordon (1980, 565) points out that higher animals and perhaps all mammals seem capable of experiencing fear, even though it is doubtful that such creatures have propositional attitudes. I will therefore rest content with noticing that in humans capable of forming propositional attitudes, in standard cases experiential hope and fear seem to appear when there is a corresponding propositional attitude.

My mirror account of hope and fear concerns hope and fear as propositional attitudes. ⁸ I have at least two good reasons for limiting the account in this way. First, once we make the distinction between propositional attitude and experiential state, it becomes obvious that the traditional analysis, to a large extent, captures the propositional attitude, while it seems dubious whether it captures anything at all concerning the experiential state. By limiting the account to propositional attitudes, we have a good point of departure for the analysis in the standard account of hope (more on this shortly). ⁹ Second, the historical philosophers who proposed that there is a strong connection between hope and fear must have had the propositional attitude in mind. CVT makes little sense for experiential states. It is obviously possible to be hopeful or afraid when the other state is not available. It seems equally clear that, if MAT holds, it must be for propositional hope and fear, as it would make little sense to claim that the experiential state of hope should be analyzed as a mirror of the experiential state of fear. ¹⁰

Some readers might feel skeptical toward my intention to return to the standard account of hope, even if only to use it as a point of departure in the mirror analysis. Does not the current literature on hope strongly suggest that the standard account is false? However, according to Kwong (2022a, 1431–1434), the inability to make the distinction between propositional and experiential hope has severely obfuscated the debate on hope. He suggests that many key issues and counterexamples to the standard account of hope can be easily dissolved with the distinction in place. While it is unfortunately outside the scope of this paper to demonstrate this in detail, I rest content with relating Kwong’s treatment of the most common kind of counterexample to the standard account of hope (for other issues, I direct the reader to Kwong’s work, see Kwong, 2020, 2022a, 2022b).

In the common counterexamples to the standard account, two subjects assign the same probability to a desired outcome, yet only one of them is considered to be hoping, because only one is upbeat and feels positive about the outcome. Kwong begins by challenging the central assumption in these thought experiments, namely that only one of the subjects have hope. He claims that a correct understanding of propositional hope makes it obvious that both subjects in these scenarios do have such hope:

If we understand [propositional] hope to play only a desire-registration role, and interpret the question of what hope is in terms of what it is to [have propositional] hope, we see that these cases actually mischaracterize the two people when they describe one person as hoping and the other as not… both people in the above examples should be characterized as hoping for the outcome. (Kwong, 2020, 835–836)

Turning to Martin’s (2014, 14–15) famous version of the counterexample, where two cancer patients go through an experimental treatment, Kwong goes on to argue that it is hopefulness (i.e., the experiential state) rather than (propositional) hope which separates the subjects, and that it is the former rather than the latter which is in need of further philosophical consideration:

We are now in a position to locate differences between the two patients who hope to be cured: Why is Bess hopeful about being cured when the odds are so low? Why does Alan not feel good about the chances of being cured when he faces the same odds as Bess? Keeping hope and hopefulness distinct helps us to highlight that… the crucial question is really one about their hopefulness (or lack thereof) and not about whether they hope or not. (Kwong, 2020, 837).

Kwong’s point is that it is not propositional hope, understood according to the standard account, which is the problem here. It is the nature of experiential hope and its relation to propositional hope, which is the source of the puzzle.

After these rather lengthy preliminaries, which I believe are necessary given the current debate on hope, it is time to take a closer look at the standard account of hope, and how it can be expanded into a mirror account of hope and fear. According to the standard account, hope requires belief in possibility and desire. Given MAT, it seems reasonable to suggest that also fear should be analyzed as requiring belief in possibility, but in combination with aversion (understood as the negative opposite of desire).

Such an account can be found in the work of Day (1970, 1998). To my knowledge, Day is the last philosopher to offer an explicit mirror account of hope and fear, and he is also sometimes credited for having given the standard account of hope its current form (Kwong, 2020, 833). According to Day, hope and fear are to be analyzed as follows:

Then ‘a hopes that p’ is true if and only if ‘a desires in some degree, however small, that p and a believes that it is probable in some degree, however small (e.g. 1/1000), that p’ is true. Similarly, ‘a fears that p’ is true if and only if ‘a is averse in some degree, however small, that p and a believes that it is probable in some degree, however small (e.g. 1/1000), that p’ is true. (Day, 1970, 121)

Day talks about belief in probability, later clarified as subjective probability (Day, 1998). To keep close to the literature on hope, I suggest that we rephrase this as belief in possibility, and for clarity, I suggest we make explicit that we are talking about epistemic possibility. Since we only tend to use the term epistemic possibility when we are uncertain regarding p, I will take epistemic possibility to exclude certainty (even though it does not do so in a strict, logic sense). That gives us the following epistemic condition for both hope and fear:

Epistemic condition: S believes that p is epistemically possible.

Since the conative condition is supposed to keep hope and fear apart, the attitudes require separate conative conditions. While I discuss them in detail later (see Section 4), for now we can simply import them from Day’s account:

Conative condition _HOPE : S desires that p.

Conative condition _FEAR : S is averse that p.

Of course, if hope and fear concern the same proposition, p should be negated in one of the conative conditions, so that we get “S is averse that −p.”

In Sections 3 and 4, I take a closer look on these conditions. I suggest some clarifications and elaborations required to bring Day’s account up to date with the contemporary debate. However, before doing so, I begin the next section, which is on the epistemic condition, by defending its necessity from two recent attacks.

3. The Epistemic Condition

The epistemic condition has recently been questioned by Blöser (2019) and Milona (2022). They argue that belief in the possibility of p is not a necessary condition for hope. Their arguments are equally relevant for the analysis of fear, and they need to be met to warrant the inclusion of the epistemic condition in the analysis.

Blöser’s argument builds on linguistic data indicating that it is possible to hope while being in doubt (rather than believing) that p is possible:

It seems perfectly conceivable that a person could say “I hope that p, but I’m in doubt about whether p is possible. Evidence for the possibility of p seems roughly on par with the evidence against the possibility of p. On these grounds, I believe neither that p is possible nor that it is impossible.” This is a case of suspension of judgement about possibility… In cases like this, it is felicitous to attribute hope. (Blöser, 2019, 209)

Blöser is right that one can suspend judgment over whether p is possible and hope that p, and it seems equally obvious that one can suspend judgment over whether p is possible while fearing that p. Like when your car breaks down and you are unsure whether it will be possible to repair it, and you fear that it will not.

The problem with Blöser’s argument is that she conflates objective possibility (i.e., the actual possibility of p being true) with epistemic possibility (i.e., the subject’s notion of whether p is possible). As far as I am aware, no one has ever claimed that hope or fear requires the subject to believe that p is possible in the objective sense; the only plausible interpretation of the epistemic condition is in terms of epistemic possibility. If in doubt, consider that if p is objectively possible in a scenario where you fail to realize this and take p to be impossible, you cannot hope for p. In the opposite scenario, where p is in fact impossible while you take p to be possible, nothing prevents you from hoping. If you incorrectly think your brother has died in the war, you cannot hope for his safe return, but there is nothing preventing you from hoping for his safe return if he is dead but you incorrectly believe him to be alive.

Blöser’s argument relies on the idea that hope is compatible with suspension of judgment. However, suspending one’s judgment over p simply means treating p as an epistemic possibility. It is only when you are uncertain regarding the truth of p, that is, when p is an epistemic possibility, you can keep your judgment regarding p suspended. This is the case even if p concerns the objective possibility. The person in Blöser’s argument must therefore be understood as reporting that she believes the objective possibility of p to be an epistemic possibility. And that is certainly not an argument against the epistemic condition.

Milona’s criticism is subtler than Blöser’s. He does not object to the epistemic condition per se, but only to spelling it out in terms of belief. According to Milona, “philosophers of hope have correctly identified hope as involving uncertainty. But we should not assume that this uncertainty takes the form of a belief [in possibility]” (Milona, 2022, 196). Milona argues against the necessity of belief by introducing a scenario where the subject’s felt hope does not match her probability assignment:

Jasmine hopes for her favorite team to finally win the championship. They have not won in her lifetime, but they have a decent team this year. The odds-makers have put the team’s chances at 5-to-1. She believes that these odds are accurate, but she nevertheless feels as if it is almost certainly will not happen. Jasmine reports that she should feel better about her team’s chances, hoping more fervently for victory. (Milona, 2022, 193)

Milona claims that if we analyze hope in terms of belief about probability, we must attribute two incompatible beliefs to Jasmine. Apart from her belief that the team’s chances of winning are about 0.17, we also need to attribute a belief to her that it is almost certain that her team will not win, since her felt hope seems to be based on such a belief. As Milona rightly points out, attributing inconsistent beliefs in this way means ascribing extreme irrationality to Jasmine, and that seems odd in this scenario. His conclusion is that the all-too faint hope Jasmine feels is not based on belief at all.

This is yet another case where adopting Kwong’s distinction between hope as a propositional attitude and experiential hope allows us to explain a problematic example. If we interpret Jasmine’s case with the distinction in mind, we realize that what is peculiar in this scenario is that the strength of Jasmine’s experiential hope does not match the strength of her propositional hope. Her propositional hope is rather strong, but her experiential hope is weak. We can only speculate on what has happened here. Perhaps, it is Jasmine’s lifelong support of this team of constant losers, which makes her hesitant, or perhaps her support of the team is somehow based on them being losers? Whatever the reason, it will have nothing to do with the epistemic condition of her propositional hope.

However, Milona’s objection seems to go beyond what is captured by his argument. On a charitable interpretation, his objection can hardly concern the conceptualization of uncertainty in terms of epistemic possibility itself, since uncertainty regarding p logically entails that p is an epistemic possibility. Rather, Milona’s point must be that it is not through beliefs, but through some other cognitive process, we primarily relate to epistemic possibilities. This is consistent with his other writings in which hope involves a “sense that p is possible” (Milona, 2020, 743) or in which hope is characterized as a “perceptual-like experience” (Milona & Stockdale, 2018). To this, I have two things to say.

First, I agree that it is psychologically unrealistic to require that a subject of hope must explicitly form the belief “p represents an epistemic possibility” through conscious deliberation. Hope only requires that it is true that p represents an epistemic possibility. However, if p represents an epistemic possibility for S, then S will have a disposition to form the belief “p represents an epistemic possibility” upon reflection (given that S grasps the concepts involved etc.). Since a dispositional reading is the only realistic reading, I take it that this is what we mean when we say that “S believes that p represents an epistemic possibility,” and this is the understanding of the epistemic condition I presuppose in this text.

Second, whether we believe epistemic possibilities or relate to them through some other, more perceptual-like process is not essential for my account. Any reader who favors the latter view is encouraged to replace “S believes that p” with “S senses that p” or “S has a perceptual-like experience that p.” ¹¹ To emphasize the inessentiality of belief, I have chosen a final formulation of the epistemic condition that leaves the question of how we relate to epistemic possibilities open (see below).

With these recent objections concerning the epistemic condition out of the way, I take it to be uncontroversial that hope and fear require epistemic possibility. But perhaps the condition needs to be further qualified?

In its basic form, the epistemic condition suggests an extremely wide concept of epistemic possibility where only certainty is excluded. The scope of what is possible in this sense is extremely wide, including very far-flung possibilities which we standardly give a subjective probability very close to zero and which we count as possibilities only because we cannot conclusively disprove them. Like the truth of a skeptical scenario where we are deceived by a Cartesian demon, or some B-movie scenario where a malevolent alien civilization exists in our solar system, conveniently placed somewhere we currently lack the technology to examine, like underground on Pluto. Even though we should most certainly be averse against being fooled by Cartesian demons or attacked by evil subterranean aliens from Pluto, these far-flung possibilities do not seem to be proper objects of fear.

I have previously addressed this issue when offering my live possibility account of hope (Palmqvist, 2021). My suggestion is that to be a proper object of hope, an epistemic possibility must be a “live possibility.” The term is inspired by James’s “live hypothesis” and is supposed to designate an epistemic possibility in the narrow senses described by James as “being among the mind’s possibilities” or being a “real possibility” (James, 2014, 9).

My account of live possibility is not only meant to exclude low probability cases. It aims to combine two seemingly conflicting intuitions regarding the possibility involved in hope. First, we have the intuition that a hope based on a likely possibility seems more proper than a hope based on a far-fetched possibility. It seems unproblematic to hope to get a good grade after studying hard, but we would not want to say that cancer patients should hope to be cured by the flying spaghetti monster. Second, we have the somewhat contrary intuition that in some cases, a very real and profound hope might be based on a very slim probability, the prime example being, of course, Martin’s (2014, 14–15) terminally ill cancer patient who hopes for a miraculous recovery due to some experimental treatment. Inspired by James’s idea that what constitutes a “live hypothesis” varies between contexts and subjects, I suggest that there is a probability threshold p must meet to count as a live possibility, and that this threshold can be higher or lower depending on contextual factors (like undefeated alternatives or the stakes involved) and personal factors (like optimism or pessimism). The live possibility account can therefore allow for hope based on exceptionally low probability in cases where the stakes are high and the person optimistic, like with Martin’s terminally ill cancer patient, while rejecting it in trivial and silly cases, like anything involving the flying spaghetti monster (Palmqvist, 2021, 691–694).

Since we need an account that is able to reject far-flung possibilities while allowing for the kind of profound hope based on a slim chance described by Martin and others, I think we have good reasons to incorporate the live possibility account into the mirror analysis. ¹² Therefore, I suggest that we revise the epistemic condition and rephrase it in terms of live possibility. When doing so, I take for granted that when something constitutes a live possibility for a subject, it is not at the same time an object of belief for that subject, and I also take it for granted that the subject lacks certainty regarding it. To keep the formulation simple, I leave these qualifications implicit (and for reasons mentioned above, I have also omitted “belief,” leaving the exact manner in which p represents a live possibility unspecified):

Epistemic condition*: p represents a live possibility to S.

At first glance, there seem to be several counterexamples to this version of the epistemic condition. For example, fear of flight seems to be a counterexample to the current suggestion, since the probability of getting in a plane crash is obviously too low for it to ever count as a live possibility. Corresponding examples for hope include having a hope to win the jackpot in a lottery with a million tickets, or, to mention an example from my original case for the live possibility account (Palmqvist, 2021, 695), hoping to have a chance meeting with a billionaire who has just decided to give away his fortune to the first person he meets. These hopes and fears might even come in combination with experiential hope and fear (at least when it comes to fear of flight).

I have previously suggested that we should characterize such hopes as foolish, and that the live possibility account should be restricted to rational hope (Palmqvist, 2021, 695). This suggestion seems even more plausible when it comes to fear, since fear of flight is often treated as an archetypical example of irrational fear. If the live possibility account only covers rational attitudes, the idea seems to suggest itself that rational hopes and fears require live possibility, and that the defining attribute of an irrational fear or hope is that it is only based on an epistemic possibility in the widest sense (which is wrongly taken to be live). This ability to distinguish rational hope and fear from irrational hope and fear is a significant advantage of the live possibility account. ¹³

4. The Conative Condition

With the epistemic condition in place, we now turn to the conative conditions (here repeated for ease of reference).

Conative condition _HOPE : S desires that p.

Conative condition _FEAR : S is averse that p.

According to MAT, the only difference between hope and fear should be the conative attitudes involved. It is therefore an important and nonnegotiable desideratum when conducting a mirror analysis, that the conative conditions must have the capacity to separate hope from fear.

Desideratum MAT: The conative condition must enable us to properly distinguish between hope and fear.

Satisfying desideratum MAT (D-MAT) might be trickier than it seems. To appreciate this, it might help to start the discussion by mentioning an alternative formulation proposed by Day in a later publication (Day, 1998) ¹⁴. In this version of the conative conditions, they are both formulated in terms of desire, and the only difference is that p is negated for fear:

Conative Condition _{HOPE (Day, 1998 )} : S desires that p.

Conative Condition _{FEAR (Day, 1998 )} : S desires that −p.

We have good reasons to reject Day’s later formulation, since this formulation fails to keep fear apart from hope. ¹⁵ Negating p is simply not the right way to capture fear’s negative orientation, since p rather than −p might be the bad option. If p stands for “my house burns down,” the subject will certainly fear that p and hope that −p, and not vice versa. If S regards p as a live possibility and desires −p, and both fear and hope are analyzed in terms of desire, we have no idea whether S hopes or fears −p. To be able to tell, we need to know the content of p, and in all cases where it is not obvious whether S views p positively or negatively, we also need background information to be able to tell whether the proposition in question represents a desirable state of affairs. For example, if p is “my aunt will pay a visit,” we will need to know whether the aunt’s visit would be a positive or negative occasion for S before we are to be able to tell whether S fears or hopes that −p. An analysis of hope and fear should enable us to separate the attitudes on the abstract level, without having to rely on background information.

While an analysis in terms of desire and aversion escapes the problem on the surface, these remarks suggest a deeper problem. Because what exactly does it mean to say that S desires p or is averse that p? A straightforward answer would be that S prefers p or prefers −p. But if desire and aversion can be reduced to preference, it seems that it is once again impossible to properly distinguish between the attitudes. We will get the same result as when analyzing in terms of desire, that is, without knowing the meaning of p and how the state of affairs represented by p is evaluated by the subject, it will be impossible to tell whether “S prefers that −p” is the conative condition of hope or fear. This result is unacceptable according to D-MAT.

According to a popular view, a necessary condition for having a desire that p is that one is disposed to act in ways that will help bring p about. ¹⁶ Given the assumption that aversion is the opposite of desire, this view would imply that a necessary condition for aversion is likewise a disposition to act to prevent that p. Can that solve the problem? I think not, because a disposition to act to bring about that p can easily be redescribed as a disposition to act to prevent that −p, and then we are back in a situation where one must know the meaning assigned to p (or, to be more exact, whether p represents something positive or negative to S) to be to be able to separate hope from fear. This is certainly not a strange result, since it is generally impossible to tell from a third-person perspective whether a certain disposition to act is based on hope or fear.

At this point, a pattern seems to emerge. Analyses interpreting the conative condition too reductively cannot properly distinguish fear from hope (at least not when p is negated). It is perhaps telling that what is missing from the previous suggestions is that the attempted analyses do not enable us to know how the state of affairs represented by p is evaluated by S. It seems that what we need is an account of desires and aversions which includes in itself that the object of a desire is good, and the object of an aversion is bad.

To come to terms with this problem, I suggest we turn to the account of desires offered by Stampe (1987) ¹⁷, according to which a desire that p is understood as a perceptual-like experience ¹⁸ that p seems good, rather than a belief that p would be good:

What is the difference between one who believes it would be good were it the case that p, and one who wants it to be the case that p?… In desire, one is somehow struck by, affected by, the merits of the thing wanted, or the prospect of having it, in a way one needn’t be if one merely knows it would be good… there is, in any desire that it be the case that p, a certain way in which its being the case that p seems to one: to wit, as if it would be a good thing. (Stampe, 1987, 356, 361)

“Good” must be understood in a broad sense, not restricted to the morally good (even though we might of course desire what is morally good), but also including things that seem good because they bring pleasure and so forth. Also, Stampe (1987, 355) is explicit that what seems good in this way might have bad consequences, so that we might refrain from following our desires because we believe doing so would have an overall bad result (like in the case of the recovering alcoholic who desires to have a drink). Also, it could well be the case that while p seems to be a good thing, it would not actually be a good thing (we are not always happy when we get what we desire, are we?).

The possibility of “dirty desires” requires special treatment. As noted by Stampe, we sometimes desire things which we upon reflection take to be bad, like the recovering alcoholic who desires to have a drink or as in Bovens’s (1999, 679) famous example of a man watching a motorsport race desiring to see an accident. It might be suggested that what the recovering alcoholic really hopes for is not having a drink, and that the man with the dirty desire to see a crash really hopes that there will be no accident; that is, that these persons do not hope what they desire. However, since it is obvious that we can have conflicting desires, I suggest that we should treat these examples as describing subjects with conflicting hopes. While it would be strange and perhaps even psychologically impossible to have conflicting states of experiential hope, the possibility of conflicting propositional hopes should be uncontroversial (since in my view it amounts to nothing more than “seeing” that conflicting epistemic possibilities might all be good, although in different ways).

I can see no problem with reapplying Stampe’s account of desire to aversion, by simply exchanging the experience of something good with the experience of something bad. For clarity, here is the notion of desire and aversion suggested by such an account put a bit more formally:

Desire _Stampe : S desires that p iff it seems to S that it being the case that p would be a good thing.

Aversion _Stampe : S is averse that p iff it seems to S that it being the case that p would be a bad thing.

Building this understanding of desire and aversion into our analysis, we can satisfy D-MAT, because if desire and aversion are understood this way, the conative conditions clearly let us separate hope from fear. If we have hope, we have a perceptual-like experience where p seems a good thing, and in fear, we have a corresponding experience where p seems a bad thing, and there is no room for conflation. ¹⁹

Should we not worry that including perception-like experiences in the analysis of the propositional attitudes threatens to blur the very distinction I argue we must be so careful to uphold? If the aversion and desire involved in the propositional attitudes are perception-like experiences, does that not turn these attitudes into a kind of experiential states? To this, I have two things to say. First, I want to point out that all propositional attitudes involve experiences. Even belief involves the experience of taking something to be true, and Kwong (2020, 834) acknowledges that propositional hope involves the experience of recognizing what one desires. Second, the subtle experiences involved in desire and aversion are very different from the strong experiential states of hope and fear, which, unlike the former, are heavily loaded with feelings, so there is no danger of conflating them. We can therefore be content with the following formulation of the conative conditions:

Conative Condition _HOPE*: S desires_Stampe that p.

Conative Condition _FEAR*: S is averse_Stampe that p.

Together with the live possibility version of the epistemic condition suggested in the previous section, we now have a qualified mirror account of hope and fear which fits with the requirements from MAT. It is therefore time to turn to the second thesis discerned from the historical accounts, namely the CVT.

5. CVT and the Connection Between Fear and Hope

I have presented a mirror account of hope and fear, a symmetrical analysis in the spirit of MAT. But some historical philosophers suggested an even stronger connection, like Seneca “[Fear and hope] are bound up with one another… the two of them march in unison like a prisoner and the escort he is handcuffed to” or Spinoza “There is no fear without hope and no hope without fear.” I have tried to capture the gist of this idea in CVT (here repeated for ease of reference):

Co-variation thesis: In every situation where hope is possible, so is fear, and in every situation where fear is possible, so is hope.

Before assessing CVT, I want to discuss a recent suggestion that hope and fear might be even more intimately connected than CVT implies. Stockdale (2020) has advanced a notion of “fearful hope,” a rare contemporary account sensitive to how hope and fear can be intertwined. Stockdale questions the idea that hope is positively valanced and argues that there are “important cases in which fear constitutes hope and is not merely experienced alongside it” (Stockdale, 2020, 121, my italics). She bolsters her case by advancing several examples, most notably real-life cases where women’s fear of being assaulted is mixed up with a hope for safety. Here follows an example:

When a Black woman, Kayla, hopes that the police officer who has pulled her over for no apparent reason and is walking toward her vehicle acts without violence toward her, she experiences fearful hope. (Stockdale, 2020, 121)

Stockdale’s scenarios introduce a new kind of puzzle cases concerning hope and fear which, just like Martin’s cancer patients, every account should be required to handle. That Kwong’s distinction is able to do so is yet another demonstration of its theoretical fruitfulness.

According to Kwong’s (2022b, 315–317), we should analyze the subjects in Stockdale’s cases as having both the propositional attitude of hope and of fear, in combination with experiential fear. Kayla fears that the policeman might become violent, she hopes that he will not, and she is afraid. This interpretation has the theoretical advantage of keeping hope and fear distinct. It makes justice to the complexity involved in “fearful hopes” without resorting to the radical claim that fear can constitute hope. ²⁰

Turning to CVT, the first thing to notice is that its plausibility depends on the distinction between propositional attitudes and experiential states. CVT does not hold for experiential states. Fear as an experiential state is only possible if the state of affairs represented by p signifies a threat or danger to the subject or something the subject cares about. While it is surely possible to have propositional fear to miss the early train or that one’s favorite restaurant might not have an available table, these states of affairs are not threatening in any way, and they cannot (under normal circumstances, for a mentally healthy individual etc.) give rise to the state of fear. The experiential state of hopefulness comes with no such restrictions because any desired possibility can give rise to it. If the desire is strong enough, it is surely possible to feel hopeful that one will catch the early train or that there will be an available table at the favorite restaurant.

Another interesting difference between the experiential states is that in some particular cases, experiential fear does not seem to require the propositional attitude of fear. For example, fear of pain is also possible on propositional attitudes such as knowledge or belief. If I know that I am about to experience pain and how the pain will feel, so that there is no uncertainty regarding either the pain’s occurrence or its nature, I can still experience fear. Arguably, there is no corresponding special object of hope, where experiential hope is possible on belief or knowledge.

While CVT might not hold for experiential attitudes, it is the propositional attitudes that are my primary concern, and here CVT seems more plausible. In standard cases, a desire that p will come together with an aversion that −p, which clearly suggests that the attitudes co-vary. However, I will argue that there are nonstandard cases in which CVT does not hold.

Hume has rightfully pointed out that as the probability for something we desire or are averse against increases, it will reach a point where hope turns into joy and fear into grief (Hume, 1969, 490). I suggest that this point is when p seems probable enough to warrant outright belief. The live possibility account implies that there will be cases where −p ceases to be a live possibility before p reaches the point where hope turns to joy or fear turns to grief. Remember that −p will cease to be a live possibility when it fails to reach a threshold value. Since the threshold value depends on several contextual and personal factors, it certainly seems possible that it could be too high for −p to be live before S becomes so convinced regarding p that S starts believing p. In such a situation, only one of the attitudes will be a live possibility, and therefore CVT does not seem to hold.

Consider the following scenario. Oscar desires a relaxing evening at home with his wife, but the traffic is terrible, and it is possible that she will be unable to make it home until very late. Oscar would assign a probability of 0.8 to the outcome that the wife gets home in a timely manner and 0.2 to the outcome that she gets stuck in traffic. The stakes might not be very high, but Oscar longs for his wife, and most importantly, he is an optimistic person who has a lower threshold for positive outcomes and a higher for negative. In this situation, it seems reasonable to suggest that p, but not –p, represents a live possibility to Oscar, and that he might hope that his wife will return soon although he is not in a position to fear her delay.

A defender of CVT might object to the possibility of hoping that p when −p is not a live possibility. She might claim that since hope requires uncertainty, the subject of hope must recognize the possibility of p not being realized, and do you really do so if you do not consider −p a live possibility toward which fear is the proper response? The objection might seem to beg the question, since it only affirms what I deny, namely CVT. However, there is a more substantial objection lurking nearby, namely that CVT is more plausible than the live possibility account; therefore, if the live possibility account cannot be reconciled with CVT, we should abandon the former rather than the latter. ²¹ I will meet this objection by demonstrating that there are counterexamples to CVT that do not require the live possibility account.

Armand is agnostic concerning the existence of God. In his view, there are two relevant live possibilities when it comes to the deepest structure of the world. Either perfect-being theism is true, so that the world is sustained by an omnipresent, omniscient and perfectly good God, or else naturalism is true, and the world described by science is all there is. Armand thinks naturalism is much more likely than theism; he accepts it as base for action and reasoning, and he has come to terms with its personal implications (i.e., he has made his peace with death). However, at the same time he realizes that perfect-being theism would instantiate a cosmic best-case scenario. Since this most desirable scenario is also an epistemic possibility (although much less likely than naturalism), Armand hopes that God exists.

In this scenario, there is nothing Armand fears. He does not fear naturalism, nor does he fear the nonexistence of God. He only hopes for the truth of theism because this epistemic possibility represents a more desirable view of the world. However, the truth of naturalism, his “worst-case scenario” so to speak, does not strike him as bad either. Would it be accurate to say that he somehow finds both naturalism and theism desirable, but theism more so than naturalism? Or perhaps, it would be more fitting to describe it as a scenario where one alternative is desirable while the other option is “neutral,” that is, neither desirable nor object to aversion? Either way, it seems that cases like this, where the subject “got nothing to fear” from the alternatives, provide counterexamples to CVT. ²²

Curiously enough, there do not seem to be any corresponding scenarios of attitudinal fear where the subject has “got nothing to hope for.” If S is averse to an epistemic possibility p while its alternative is “neutral” in the sense mentioned above, it would seem most natural to suggest that S hopes that −p. And if the alternative −p is bad, but not quite as bad as p, it also seems natural to suggest that S hopes for the less bad alternative. Consider, for example, the case of an arrested man who is interrogated by two police officers using the famous good cop and bad cop routine. The man is averse against being interrogated by both officers, but he still hopes for the “good cop” to interrogate him because he is more averse to being interrogated by the “bad cop.”

In short, it seems that CVT holds up better for fear than it does for hope. While we always desire the opposite of what we are averse to, we are not always averse to the opposite of what we desire. However, as long as we subscribe to the live possibility account, we have theoretical reasons to reject CVT for both fear and hope.

6. Concluding remarks

Returning to the historical tradition of treating hope and fear together, I have proposed a mirror account of hope and fear as propositional attitudes, where the only difference between the attitudes is to be found in the conative condition. My account is an elaboration of the standard account of hope. Instead of improving the analysis by adding some third condition, my analysis qualifies the two conditions of the standard account. The qualifications I propose are to understand the epistemic condition in accordance with the live possibility account, and the conative condition in line with Stampe’s analysis of desire and aversion:

Epistemic condition*: p represents a live possibility to S.

Conative condition _HOPE*: S desires_Stampe that p.

Conative condition _FEAR*: S is averse_Stampe that p.

However, some historical philosophers have argued for an even stronger connection between hope and fear. The CVT claims that if fear is possible in a situation, so is hope and vice versa. I have argued that we have strong reasons to reject this idea: it is fully possible to hope while having “nothing to fear,” and the CVT also seems incompatible with the idea that an epistemic possibility must meet a certain probability threshold to count as a live possibility.

The present account demonstrates the importance of upholding the distinction between propositional attitudes and experiential states when it comes to fear and hope. Indeed, it is Kwong’s distinction and his observation that the standard problems associated with the analysis of hope are tied up with the experiential state which makes a mirror account of propositional hope and fear a viable option. The present account thereby offers a solid case for the theoretical fruitfulness of Kwong’s distinction.

While the mirror account is limited to propositional attitudes, an important next step is to bring experiential fear and its relation to both experiential hope and propositional fear properly into the theoretical framework. We have already seen that there are interesting differences between experiential hope and fear, and if it is experiential hope that gives rise to the canonical problems, how do these problems affect experiential fear? Is there anything corresponding to the “despair problem” for fear, and can fear be “profound” like hope? And how does a mirror account understanding of the propositional attitudes affect our ability to come to terms with such problems for experiential hope and fear?

The mirror account also gives us reason to revisit the old issue of whether hope is an emotion of not. Many reject the idea that hope is an emotion while accepting that fear is an emotion. Even if this issue also concerns the experiential states, the mirror account suggests that either both attitudes are emotions or none. The analysis of hope as including perceptual-like desire might seem to suggest that the former alternative is correct, but the question needs to be addressed in detail in future research.

In short, reintroducing the historical idea of a mirror analysis of hope and fear raises many interesting questions for further research. I hope to have shown such mirror account is not only possible, but also a highly feasible way forward for the philosophy of hope and fear.