Suffering is bad: experiential understanding and the impossibility of intrinsically valuing suffering

Abstract

Suffering, I argue, is bad. This paper supports that claim by defending a somewhat bolder-sounding one: namely that if anyone—even a sadistic ‘amoralist’—fully understands the fact that someone else is suffering, then the only evaluative attitude they can possibly form towards the person’s suffering as such is that of being intrinsically against it. I first argue that, necessarily, everyone is disposed to be intrinsically against their own suffering experiences, holding fixed their specific overall degree of emotional aversiveness, because any evaluative attitude other than ‘being against’—including mere indifference—would in certain key circumstances make our suffering less emotionally aversive and thus different from the suffering experience (stipulatively) at issue. Second, fully understanding that someone else is having a given experience—Mary’s experiencing a vividly blue sky, say, or Job’s experiencing heart-rending grief—requires that we represent experientially their very instance of that experience-type (it requires, in other words, token phenomenal concepts). The result is that what goes for our own suffering goes for others’, too: maintaining an accurate experiential representation of the fact that someone else is having a suffering experience with a specific degree of overall emotional aversiveness is only compatible with coming to be intrinsically against their suffering. So suffering is—‘objectively’—bad: it’s only possible to respond with indifference towards anyone’s suffering if we don’t fully understand that they are suffering in the first place.

1 Introduction

The claim in my title might not strike you as the discovery of the decade: suffering’s badness is likely to seem more obvious than any other normative fact, and it might even be a basic fact, making it a fool’s errand to try to argue for it in the first place.

Even so, the objective badness of suffering at least seems to be in tension with another widely-accepted view: namely that it is possible to understand full well that someone else is suffering without being bothered in the slightest. Some people might even fully understand but be positively happy about the other person’s suffering, just because that person is suffering. An epistemically idealized version of the reputedly sadistic Roman emperor Caligula, for instance (cf. Gibbard, 1999; Street, 2016), might be exceptionally good at understanding exactly how it feels for his victims to suffer, even in arbitrarily specific ways. But that might be precisely what he enjoys: he might delight in their suffering in proportion to how well he understands, in a vividly experiential way, what he’s putting them through.

Such cases challenge the idea that it is irrational or ignorant—necessarily irrational or ignorant, for everyone, that is—to fail to care about someone else’s suffering. Our imagined Caligula is epistemically perfect, after all (if only by supposition), so maximizing his sadistic joy might also be fully practically rational. But if there were nothing practically or epistemically objectionable about Caligula’s sadism, then it would be difficult to make sense of the idea that suffering was objectively bad. What would its badness amount to, if it had no objective practical or epistemic ‘reason-giving force’—if caring about it and not caring about it could be on equal normative footing?

My goal in this paper is to defend the normative force of suffering’s objective badness by arguing directly against the possibility at issue: that is, by arguing that failing to care about someone else’s suffering (in contrast to sadistically valuing it or even just being evaluatively neutral¹) is only possible if we don’t fully understand that they are suffering. In other words:

If anyone fully understands the fact that someone else is suffering, it will be impossible for them—if they form an evaluative attitude towards it at all—to form any evaluative attitude other than being intrinsically against (intrinsically disvaluing²) that person’s suffering.

If my argument for that claim is sound, then an epistemically idealized, all-understanding, but sadistic Caligula is strictly impossible: practically and epistemically rational or not, if he fully understands the fact that his victims are suffering and forms an evaluative attitude towards that fact, then he will necessarily be disposed to disvalue their suffering.³ And that means suffering meets a very plausible sufficient condition for being (‘objectively’) bad (cf. Lewis, 1989).

To be clear, I won’t be offering any serious argument in favor of that sufficient condition (although see Sect. 4.2), and some people might well reject it. But the possibility of an all-understanding sadist is arguably the core reason for moral skeptics’ denial that anything is objectively good or bad. So showing that suffering meets this condition would in effect clear away the main metaethically salient obstacle to accepting that something—namely, suffering—is bad.⁴

My argument draws on two core claims. The first is that fully understanding the fact that someone else is having an experience of a given type—say, a redness experience, or the taste of cold, fresh-squeezed lemonade—requires that we represent their very instance of that experience-type experientially (cf. Jackson, 1986), specifically by means of a token phenomenal concept.⁵ The second claim is that it’s impossible to form any evaluative attitude other than being intrinsically against any experience we represent experientially as suffering. Together, those two premises entail that, with respect to the fact that someone else is suffering, it’s impossible to come to any evaluative attitude other than being intrinsically against it—as long as we fully understand that they are suffering.

Here is how the paper proceeds. Section 2 goes into more detail on the key concepts of and ⁶ and defends the claim that, for any experience we ‘represent experientially as suffering,’ it’s impossible to form any intrinsic evaluative attitude towards it other than being intrinsically against it, by arguing that it’s impossible not to disvalue our own suffering (if we form an intrinsic evaluative attitude towards it at all). Section 3 defends the claim that full understanding requires that we represent other people’s suffering experientially, by means of token phenomenal concepts, such that we relate to their suffering in a way that is importantly similar to the way we would relate to it if we were experiencing it directly for ourselves. Section 4 puts those claims together to argue that suffering is—objectively, agent-neutrally—bad. I close with a brief discussion of possible implications for explaining suffering’s badness and its ‘reason-giving force.’

2 On being against our own suffering

My goal in Sects. 2.1 and 2.2 is to draw out a few intuitive and relatively theory-independent constraints on suffering and disvaluing. The short version of my two proposals is that, first, a given suffering experience is at least partly individuated by the specific degree of overall emotional painfulness or aversiveness (negative affect) of our experience as a whole; and second, if we value or disvalue something, then we are disposed to be, at least in certain privileged contexts, motivationally and affectively in favor of or against it, respectively.

2.1 On suffering

I’ll be arguing, in Sect. 2.3, that if we form an evaluative attitude towards our suffering at all, it will be impossible not to be intrinsically against it. A first reason to doubt that claim is that it seems quite possible not to disvalue many experiences we at least wouldn’t normally hesitate to call ‘suffering.’ Some people claim to enjoy physical pain of certain kinds, for instance, and seemingly intrinsically: just in light of the way such pains feel.⁷ And in everyday speech, all or most physical pains could also be described as forms of physical suffering.

Still, there is an everyday way of understanding suffering that makes those apparent counterexamples seem like they are missing the point. One way to get into that frame of mind is to note that another natural description of such cases is as instances of pain that don’t end up causing us to (‘really’) suffer—or at least not as much as they would have if we didn’t enjoy them.

Hopefully that way of talking about suffering is intuitive. It may also be worth noting that virtually all philosophical accounts of suffering make the distinction between pain and suffering (Bain et al., 2020), with suffering usually taken to involve specifically emotional pain or ‘felt aversion’ (cf. Kauppinen, 2020): you only suffer, when experiencing physical pain—or experiencing a loss, annoyance, or anything else—if you ‘mind’ it, affectively, in the right way.

For that reason, it’s probably better not to focus on cases of physical pain, as you read through this paper, but on emotional experiences of, say, grief or sadness. While physical pain may usually be emotionally aversive, it also seems to be the sort of thing we can in principle enjoy—or at least experience in an affectively neutral way as a result of, say, meditation practice (Grant, 2014) or brain injury (Grahek, 2007). And that could conceivably make it less emotionally aversive⁸ without making it any less physically painful.

As I hinted above, then, the minimal feature of suffering I’ll be drawing on—a feature compatible with virtually all accounts of experiential suffering—is that each suffering experience comes with a specific degree of overall emotional aversiveness (‘overall affective negativeness’). The most paradigmatic, thoroughgoing instances of suffering, for instance, seem to involve a kind of unmitigated overall emotional aversiveness.⁹

An important consequence, for the sections to follow, is that when I talk about suffering experiences, what I have in mind are whole experiences—almost always with some emotionally neutral or positive components in addition to the negative ones—rather than just the experience’s emotionally negative parts.¹⁰ The reason is just that that’s how actual suffering experiences are. We don’t experience the negative and positive components in isolation; we experience them together, making every given suffering experience come with a specific degree of overall emotional aversiveness.

2.2 On ‘being against’

Turning to the attitude of ‘being against’ or disvaluing, the main question is whether the kind of attitudes at issue—valuing, disvaluing, and evaluative neutrality—are cognitive, non-cognitive, something in between, or some combination of the two.

All that matters for the argument to come, though, is that whatever cognitive features they may have, attitudes of valuing, disvaluing, and evaluative neutrality seemingly necessarily involve, at least under certain specific conditions, some corresponding affect and motivation. If someone claims to be against something—racism, say—but never feels or would feel even the slightest negative affect or motivation of any kind against it,¹¹ then it’s natural to think they don’t really disvalue racism.

It might still be possible (though it also might not be: cf. Rosati, 2016) for someone who is ‘left cold’ by racism in that way to make a related sincere judgment or form a related normative belief, such as . But we would be justified in characterizing such a person—someone who truly had no accompanying motivating or affective attitude in favor of combating racism—as someone who didn’t care about combating racism in the way required for it to count among their values, properly understood.

In short, it seems safe to endorse a very weak version of affective and motivational internalism about valuing and disvaluing: valuing something, for instance, entails that at least under certain specific conditions (see Sect. 2.3.1) and to some subtle degree, we would feel affectively positive towards, and motivated in favor of, what we valued.¹²

Before I move on, note that on its own, Sect. 2.1’s minimal conceptual constraint on suffering—that each suffering experience comes with a specific degree of overall emotional aversiveness—has no direct entailments for the attitude of disvaluing as I’m understanding it here; my argument below is not a defense of the tautology that we’re necessarily disposed to disvalue the kind of suffering that we’re necessarily disposed to disvalue. Even if there turns out to be some tight relationship between them, overall emotional aversiveness is neither, say, constituted by nor conceptually inseparable from disvaluing,¹³ which may well include (for all I’ve said here) much more complicated cognitive attitudes. Overall emotional aversiveness is an experiential quality with which we are all familiar and which we can understand independently of disvaluing.

2.3 Evaluative attitudes towards our own suffering

Together, those two minimal constraints on suffering and valuing provide an argument for my first main claim: necessarily, everyone is disposed to be intrinsically against their own suffering—or any experience they ‘represent experientially as suffering.’

Here is the core idea, without further ado: if someone comes to value or be evaluatively neutral towards their suffering, then their overall experience will—at least in certain key moments—thereby become less overall emotionally aversive. In slogan form, if you’re into it, then you’re not suffering as much. The result is that it’s impossible to come to value (positively or neutrally) the specific ‘pre-evaluative-attitude’ suffering at issue while maintaining its specific degree of overall emotional aversiveness.

2.3.1 The ‘test conditions’ for intrinsically valuing our suffering

It’s worth addressing one potential worry right away. Remember that on my proposed way of understanding valuing (Sect. 2.2), it only necessarily involves feeling affect and motivation under certain specific conditions. So it might sound possible, despite the argument I just introduced, to come to value a suffering experience intrinsically without thereby reducing its overall emotional aversiveness. In particular, as long as the ‘certain specific conditions’ in which we have the relevant affective attitude (associated with the evaluative attitude) only come about when we’re not actively experiencing our suffering, the argument sketched above won’t secure the impossibility of coming to value, intrinsically, our suffering: there would never be any conflict between our suffering experience’s degree of overall emotional aversiveness and the positive affective contribution of our valuing it.

It’s true that Sect. 2.2’s proposal—that valuing requires that we would feel the corresponding affective attitude, at least under certain key conditions—left that possibility open. The thought was just that having no positive or negative affective disposition whatsoever about something would entail that we neither valued nor disvalued it. Valuing and disvaluing (again, possibly in contrast to attitudes like simply believing something is valuable) get us ‘affectively involved.’ So I now want to make two specific points in connection with the conditions in which our evaluative attitudes would ‘contribute affectively,’ to help show why coming to value our suffering experiences as such—holding fixed their specific degrees of overall emotional aversiveness—is impossible.

First, note that if I’m right and anyone who values something positively, negatively, or neutrally is necessarily disposed to have the corresponding affective attitude under some conditions, then the test of whether they intrinsically value something—friendship, say—concerns the moments in which they are or would be experiencing the relevant affective attitude characteristic of valuing (an affectively positive attitude, at minimum) towards friendship. They only intrinsically value friendship, in other words, if there are or would be such moments of being affectively involved in the way associated with valuing it.

Second, note that we are specifically interested in what follows from valuing something intrinsically. As I use that expression in this paper, to value something intrinsically is to value it just in light of that ‘something’ itself: just in light of its essential or constitutive features, or in light of understanding it for what it is.¹⁴ But suffering, as I’m understanding it (see Sect. 2.1), is essentially experiential: one of its essential features is its feeling the way it feels, experientially.¹⁵ Coming to an evaluative attitude towards it just in light of its essential features thus requires that we be in conditions of experientially appreciating ‘what it’s like’ to have the relevant suffering experience.

The result is that another part of the ‘test’ of whether someone values their own suffering intrinsically is whether they value their suffering when confronted with it in its full experiential glory—that is, while experiencing it and focusing just on its precise constitutive experiential qualities (cf. Sect. 3).

Putting those observations together gives the following result for the conditions of the ‘test’ of whether someone intrinsically positively, neutrally, or negatively values their suffering. In particular, if a person values their suffering, then there must be circumstances in which they would be both (i) experiencing that suffering, with its same specific degree of overall emotional aversiveness, while also (ii) coming to experience the positive, negative, or neutral affect (towards that suffering) that characteristically accompanies valuing, disvaluing, or being evaluatively neutral, respectively. That is, someone only intrinsically values their own suffering if they could come to feel affectively positive towards their suffering experience as a whole while actively experiencing it.

It follows that coming to value (positively or neutrally) our suffering experience as a whole, with its specific degree of overall emotional aversiveness—coming to value it intrinsically, that is—is impossible: it would only be possible if it were possible under those two conditions: (i) actively experiencing the suffering and (ii) experiencing the new valuing-specific affect; and it isn’t. The reason, as I introduced above, is that adding positive or neutral affect to our suffering experience reduces its degree of overall emotional aversiveness, thereby constitutively chasing away the specific degree of overall emotional aversiveness that was in question in the first place.

2.3.2 Other ways of valuing our suffering

It’s worth emphasizing that this argument says nothing about the possibility of coming to value our suffering experience non-intrinsically—as, say, a necessary step towards some larger goal, or even just as a sign of progress towards it. First, the ‘test’ of non-intrinsic valuing doesn’t require that there be moments of confronting suffering ‘in its full experiential glory’ at all. What makes us count as (say) instrumentally valuing something is just our attitude towards the relevant instrumental features. And second, even if there were moments in which instrumentally valuing our suffering made our overall experience less emotionally aversive,¹⁶ we might still count as valuing our suffering—just as long as by “our suffering” we don’t mean our specific suffering experience as a whole, with its specific previous degree of overall emotional aversiveness. In fact, the same goes for coming to value our suffering intrinsically: if we use “our suffering” that way—to pick out, say, a specific affectively negative experience-component (see Sect. 2.1)—then it’s perfectly possible to come to value ‘our suffering’ intrinsically.

For parallel reasons, on yet another way of understanding the expression “our suffering,” my argument doesn’t even rule out the possibility of coming to value our suffering intrinsically while holding fixed the degree of emotional aversiveness we are experiencing over time. In particular, that will be possible if by “our suffering” we mean not some specific ‘pre-evaluative attitude’ suffering experience, but any experience with the same degree of overall emotional aversiveness.

Here’s an example. Suppose that, at t$_1$, we are experiencing (i) the emotional aversiveness of a stinging insult about our choice of hairstyles but also (ii) the emotional pleasingness of a tasty lollipop in our mouth, with a resulting overall affective aversiveness-to-pleasingness ratio of (say) 3:1. Now suppose that, at t$_2$, we have managed to maintain exactly the same 3:1 aversiveness-to-pleasingness ratio but have replaced the emotional pleasingness contributed by the lollipop with the emotional pleasingness contributed by valuing this (new) suffering experience—that is, the experience constituted by the aversiveness of the insult and the pleasingness associated with valuing the new but still 3:1 aversive-to-pleasing experience as a whole. We would then have come to value a suffering experience while maintaining the same degree of overall emotional aversiveness.

Note, though, that the main question relevant to generalizing from the case of our own suffering to the suffering of others is what new evaluative attitudes we can form towards a given pre-existing, specific suffering experience as a whole, with its specific degree of overall emotional aversiveness. The ‘valuing-replaces-lollipop’ case does involve the formation of a new evaluative attitude,¹⁷ but that new attitude is also towards a new and importantly different suffering experience. It is partly constituted by our now liking our experience qua suffering experience, after all, and has no lollipop-based pleasingness. So we can ask my question with respect to either suffering experience in the ‘valuing-replaces-lollipop’ case: What new evaluative attitudes can we form towards the insult-plus-lollipop suffering experience? What about towards the insult-plus-valuing one? In both cases, the argument in Sect. 2.3.1. entails that we can only come to disvalue them. The mere existence of the insult-plus-valuing experience, then, doesn’t constitute an answer to the question of what attitudes we can form towards the insult-plus-lollipop experience, despite their sharing the same degree of overall emotional aversiveness, and despite, in the insult-plus-valuing case, our valuing our suffering experience in light of its degree of overall emotional aversiveness.

More generally, the point of the argument in this section is just that given any particular suffering experience—a specific experience of suffering, with a specific degree of overall emotional aversiveness—coming to value it positively or neutrally would make the overall suffering experience we were experiencing less emotionally aversive, in the relevant ‘test’ conditions, and thus an importantly different experience.

It would be importantly different, that is, with respect to the standard of having exactly the same degree of overall emotional aversiveness—even if it were in all other respects the same. That standard may not, in general, be particularly practically relevant when it comes to our own suffering: why not make our suffering less emotionally aversive? But again, and as I will be arguing in the sections to come, it turns out to be a central epistemic requirement for understanding someone else’s specific suffering experience as a whole, which (unlike our own suffering) doesn’t change, experientially—let alone for the better—just in virtue of our coming to value it.

2.3.3 ‘Intrinsic meta-affect’ and the possibility of disvaluing our suffering

A last question concerns the affective contribution of disvaluing. All I have said so far is that, in the relevant (‘test’) conditions, valuing and evaluative neutrality would lessen our experience’s overall emotional aversiveness and thus make it an importantly different suffering experience from the one we were supposedly forming the attitude towards. But you might think, for parallel reasons, that disvaluing our suffering would increase its emotional aversiveness. If that were so, it would be impossible to come to any evaluative attitude whatsoever with respect to our own suffering’s specific pre-evaluative-attitude degree of overall emotional aversiveness—disvaluing included.

My answer is that, as others have observed (e.g. Klein, 2015, p. 55; Korsgaard, 1996, p. 154), it seems to be an intrinsic feature of suffering that we are averse not only to the subjective source of our suffering but to our suffering itself—at least when focusing on it, as we’re here supposing we are. When we’re suffering, then, we already have an affectively negative attitude towards our suffering: ‘intrinsic meta-aversion’ is an existing part of our suffering experience’s degree of overall emotional aversiveness.¹⁸

The result is that we can come to disvalue our suffering intrinsically without increasing its degree of overall emotional aversiveness. Suffering’s existing intrinsic meta-aversion can be (or be ‘recruited’ or ‘co-opted’ as) the affective component of our disvaluing. Intrinsic meta-aversion is directed at our suffering, after all, and in all likelihood our intrinsically generated meta-aversion to suffering is in fact part of the unpleasantness of many unpleasant everyday experiences.

In sum, the only evaluative attitude compatible with sustaining a given suffering experience as a whole—with its specific degree of overall emotional aversiveness—is disvaluing it. So, necessarily, everyone is disposed to be intrinsically against their own suffering.¹⁹

3 Full understanding and token phenomenal concepts

It may already have sounded intuitive to you that necessarily everyone is disposed to disvalue their own suffering (if they form an evaluative attitude towards it at all)—and on its own, that fact doesn’t entail anything about what attitudes we can form towards other people’s suffering. So you might take what I’ve said so far to establish that suffering is bad ‘for the sufferer’; but Caligula, for instance, might happily agree, and just take it as further fuel for his sadistic delight: he likes that suffering is bad for his victims.

So this section moves on to defend the second main premise of the paper: namely, that fully understanding the fact that someone else is suffering requires that we represent their suffering experientially. If so, then (by the argument in Sect. 2.3), necessarily, anyone—Caligula included—who fully understood the fact that someone else was suffering could only come to be intrinsically against it.

A few important clarifications are in order before diving in. A first point to acknowledge is that some things are no doubt just impossible to understand, fully or otherwise, and for all I’ve said so far, that could turn out to be the case for understanding that someone else is suffering. If so, that might make it hard to interpret my claim—that anyone who fully understood that someone was suffering could only come to be intrinsically against that fact—as establishing a practical-rationality-relevant epistemic deficiency in everyone who intrinsically valued or was evaluatively indifferent about someone else’s suffering. In particular, how could it establish a necessary epistemic deficiency if someone like Caligula, say, could go on intrinsically valuing someone’s suffering even when understanding as fully as humanly possible that the person was suffering?²⁰

Now, the most straightforward barriers to full understanding are the finite time, storage, and ‘processing power’ available to our minds: some things might just be too complex, or require knowledge of too much information, to afford full understanding by a given individual, or even by humans in general. It’s also arguably impossible for any mind—finite or not—to understand fully something like what it’s like to be devoting 100% of one’s attention to (say) the taste of their coffee while at the same exact time also fully understanding what it’s like to be devoting just half of one’s attention to the taste of their coffee.²¹

In order to see whether such worries apply to our case, the main thing to note is that the question my argument is concerned with is (i) which intrinsic evaluative attitudes we can form (ii) towards a fact constituted by a very minimal set of features: namely, the fact that someone other than ourselves is suffering. Next, recall that on my understanding of intrinsic valuing, it involves valuing something ‘just in light of its essential or constitutive properties’ (Sect. 2.3.1). The result is that counting as having a given attitude intrinsically towards the fact that someone else is suffering requires fully understanding just (i) that there is someone who isn’t us and (ii) that they (whoever they are) are experiencing an instance of the experience-type suffering. Finally, if my observations in the sections to come are right, then it should in fact be possible to understand—even fully—those much more limited features of the suffering-situation at issue.

One last point to note is that even though I’ll be defending in passing the idea that fully understanding those minimal features is possible, my argument actually doesn’t hinge on whether that’s true. The reason is that the core relevant premise of my argument really only concerns a specific necessary condition for fully understanding that someone is suffering—namely, again, that we represent the other person’s suffering experientially, with an experiential mode of presentation or ‘token phenomenal concept.’

Strictly speaking, then, as long as my argument for the possibility of satisfying that necessary condition works, my central claim could be restated in terms of the “fullest (humanly) possible understanding”: necessarily, anyone who understood as fully as (humanly) possible the bare fact that someone else was suffering, to a given specific degree of emotional aversiveness, could only come to disvalue, intrinsically, that person’s suffering.

3.1 On ‘understanding’

Quite a few historical and contemporary metaethical projects have tried to establish something like that claim: namely, that having a certain epistemic good would make it impossible not to care about other people or their well-being (e.g. Kant, 2002; Nagel, 1970; possibly even Hume: see Railton, 1999). But the relevant ‘epistemic goods’ vary. The projects closest to mine, for instance, are Marshall’s (2018) Compassionate Moral Realism and Atiq’s (2021) “Acquaintance, knowledge, and value” (cf. Atiq & Duncan, in press; Sinhababu, 2022). Marshall’s target epistemic good is being in touch, and Atiq’s (acquaintance) knowledge.²²

There are no doubt important differences between the target goods, but my ‘full understanding’ is meant to be maximally inclusive: my concern is with epistemic improvements of any kind, such that fully understanding something entails that no epistemic situation with respect to it would be (epistemically) better. My actual target is thus the broad idea corresponding to—to use a colloquial expression—(as fully as possible) knowing what we’re talking about.

The reason for my maximally inclusive notion of epistemic value is that, first, when it comes to representing experiences or experiential qualities—the taste of fresh-squeezed lemonade, say—there is already quite a broad consensus (see Nida-Rümelin & O’Conaill, 2021) that someone who has never had a given kind of experience is importantly epistemically impoverished with respect to that experience-type.²³ And second, the core reasons motivating that consensus are ultimately all my argument relies on.

It’s worth noting, despite my very inclusive notion of epistemic value, that most dedicated theories of understanding as such would also support categorizing the specific epistemic good at issue in this section as (at least) a form of understanding. First, understanding a given fact is usually taken to be less compatible with epistemic deference than merely knowing the same fact (e.g. Hills, 2016).²⁴ And second, while some have argued that understanding might ultimately reduce to propositional knowledge (e.g. Sliwa, 2015; cf. Lawler, 2018), even they think that, relative to just knowing some fact, understanding it generally requires richer information and more fine-tuned cognitive abilities (Grimm, 2021; Pritchard, 2009, Sect. 2.2) or ‘know-how’ (cf. Hills, 2016).

What follows for understanding experiential qualities like suffering, given those established commitments about understanding? A natural proposal is that fully understanding an experience-type requires the specific ‘know how,’ gained from experiencing it for ourselves, of being able to recognize the target experience-type—say, the taste of cold, fresh-squeezed lemonade, again—when experiencing it (cf. Lewis, 1990; Sliwa, 2017).

We can go further, though: in particular, being actively confronted with the experience also seems to be doing some of the epistemic work in these understanding-related moments of recognition.

Consider Jackson’s (1986) famous example of Mary the color scientist who has lived in a gray-scale environment her whole life and finally sees, say, a stunningly blue sky. Now imagine that Mary returns to her gray-scale environment and completely forgets how the sky looked—despite, we can suppose, now having the ability to recognize blueness if she sees it again. It seems clear that Mary is in a worse epistemic position, blueness-wise, than when she was outside. What’s more, if she is suddenly struck by a vivid visual memory of the blue sky, it seems clear that she is back in a better epistemic position.

Arguably for the same reason, Mary also seems to be in a worse epistemic position with respect to blueness whenever she isn’t actively recalling (or seeing) something blue, even if she could summon up a vivid memory on command. We might not hesitate to say—though on my view we would be speaking somewhat loosely—that Mary, understood as a temporally extended thinker and imaginer, ‘knows full well what she’s talking about,’ now, when talking about blueness experiences. And yet, as Atiq (2021, Sect. 3) has argued at greater length, Mary remains importantly like her former, experiential-blueness-benighted self, in these specific moments: she could, for instance, coherently wonder whether she has completely forgotten what blueness looks like or whether she can still simulate a blue mental image on command—even if she in fact can. Relatedly, the moments actually constituting her remembering what a blue sky looks like are, arguably, the moments in which she is actively experiencing a visual memory of a blue sky. So there are good reasons to think that it’s only in those moments that Mary is in the best epistemic position with respect to experiential blueness.²⁵

In any case, I will hereafter be assuming that when it comes to something essentially experiential like suffering, we are always in a better epistemic position to the degree that we are actively experiencing (or vividly ‘simulating’²⁶ or recalling) the relevant experiential quality. For anyone unswayed by the above observations or related arguments from Atiq and others, my overall claim in this paper should thus be understood conditionally: if fully understanding something essentially experiential requires an ‘actively experiential’ mode of presentation, then suffering is bad.

To be sure, that is more understanding than we need, for most purposes. The point is that any evaluative attitudes—towards, say, the color blue or the taste of lemonade—that we would form if we did have this more robust kind of understanding would be more authentically about blueness or lemonade-taste as they really are.

In any case, as promised, this proposed requirement for full understanding is equivalent to one requiring that we represent things like blueness or the taste of lemonade using phenomenal concepts: here type phenomenal concepts, and specifically ones constituted by an occurrent instance²⁷—understood broadly to include vivid memories or mental imagery—of an experience of the relevant type.

There is an active debate over the nature of phenomenal concepts. Some (e.g. Levin, 2006) say they are are best understood as being like demonstratives: deploying a phenomenal concept of the taste of lemonade is like thinking , where directly refers to the remembered, simulated, or actual taste-experience or the experience-type it instantiates.²⁸ Balog’s (2012) ‘quotational’ account, in contrast, proposes a “concept-forming mechanism that operates on an experience and turns it into a phenomenal concept that refers to either the token experience, or to a type of phenomenal experience that the token exemplifies” (Balog, 2012, p. 33).

For my purposes, all that matters is that phenomenal concepts can be taken to constitute an essentially experiential representational format or mode of presentation.²⁹ Representing the taste of cold, fresh-squeezed lemonade using a phenomenal concept makes us relate to that taste-type in an experiential way, partly constituted by our currently (re)experiencing the lemonade-specific experiential character.

3.2 Token phenomenal concepts of other people’s experiences

In short, fully understanding an experience-type seems to require that we represent it experientially. Now, that might sound like it already establishes what I set out to establish: if understanding the type suffering requires that we represent it experientially, then the argument in Sect. 2.3 might entail that (if they come to an evaluative attitude about it at all) it’s impossible for anyone not to disvalue not just their own suffering but the whole experience-type of which it is an instance.

Even if that were true, though, it would be unclear what followed from it. What is it to disvalue an experience-type? If it entails disvaluing all instances of the type (‘in’ anyone, ever), then I don’t need any further argument for the conclusion that necessarily, anyone who fully understood the fact that someone else was suffering would only be able to come to be intrinsically against it. But disvaluing an experience-type might just as easily turn out not to have that entailment, and it’s not obvious how to determine which verdict is right. So I’ll now turn directly to the question of what it takes to represent someone else’s token instance of a given experience-type with full understanding.

Note, first, that the central lemma of this paper—that full understanding is compatible only with coming to disvalue someone’s suffering intrinsically—just concerns (i) someone other than ourselves (ii) experiencing an instance of suffering. In other words (as I noted above), we are abstracting away from who the sufferer specifically is, and even from how their specific instance of suffering differs from other instances of suffering—so long as (following Sect. 2.1) we represent their experience as having some specific, non-zero degree of overall emotional aversiveness. My question just concerns what evaluative attitudes we can form with respect to the bare fact that someone other than ourselves is suffering, and who the sufferer specifically is doesn’t have any direct bearing on that question.³⁰

It’s worth noting, in particular, that understanding the bare fact that they are the sufferer is very unlikely to require an experiential mode of presentation for the person as such.³¹ It is controversial, after all, whether specific ‘subjects of experience’ (our selves, in other words) are themselves direct objects of our experience in the first place (cf. “Anattalakkhaṇa Sutta: The Characteristic of Not-Self,” 2018; Nāgārjuna, 1934; Hume, 1738, I, IV, Sect. 6; Dennett, 1992; see Duncan, 2019 for an argument against such views).

In any case, I’ll hereafter be assuming that being a minimally competent user of the sufferer’s name—which, following Kripke (1980), for instance, likely requires little more than that we be properly situated in a causal chain of name-users leading back to the individual’s ‘baptism’ with that name—is (more than) enough. Fully understanding the expression “someone who isn’t me” would do just as well.

The main things we need to understand, then, are just the experience-type suffering and the fact that the relevant sufferer’s experience is an instance of that type. My claim in what follows is that fully understanding someone else’s experience as an instance of a given experience-type requires an experiential mode of presentation for their instance itself: it requires a token phenomenal concept of their experience.

At the outset, it might seem to be enough, for full understanding, just to represent the type experientially. Suppose, for instance, that the state of affairs we’re trying to understand is the fact that the long-suffering Job³² is being subjected to an instance of the mealy, unpleasantly soft mouth-feel of an unexpectedly over-ripe Red Delicious apple. It might seem to be enough to recall, vividly, an instance of that kind of experience and think,

.

The representational target, again, is Job’s experience’s being an instance of mealy apple mouth-feel. So we might think of our representation as having two argument places: one for the experiencer (Job), and one for the experience-type of which his experience is an instance (mealy apple mouth-feel).

The problem is that we might be able to understand fully both the experience-type mealy apple mouth-feel and Job’s being the experiencer but fail to appreciate fully the nature of the instantiation of an experiential type. What matters, in particular, is that an experience-type’s being instantiated is itself an essentially experiential fact.

The result is that even if we fully understand the content of both argument places—and even accurately represent the instantiation relation between them—we will still fail to understand fully the fact as a whole unless we also represent the instantiation relation itself experientially. Otherwise, with respect to the experience-type’s actually being (experientially) instantiated, we are at best like the version of Mary, above, who maybe could summon up a vivid memory of blueness on command but wasn’t currently doing so.

Full understanding, then, requires that we represent the given experience-instantiation itself experientially. But to do that—to represent experientially, say, Job’s instantiating the experience-type mealy apple mouth-feel—just is to represent his experience with a token phenomenal concept.

It turns out, then, that there is a version of the thought that can suffice for full understanding: namely, a version in which we deploy a phenomenal concept for the instantiation of the relevant type, too. And that makes that version of the thought equivalent to or , where is a token phenomenal concept of his instance of mealy apple mouth-feel—or even, if you’ll allow thoughts with less obviously propositional structure—just .³³

3.3 Two ways to represent someone else’s experience

It’s worth taking a step back to get a more intuitive grip on what we gain by representing Job’s experience in that way.

Again, for our purposes, all that matters about Job is that he is someone other than ourselves. But so far I’ve been talking as if representing ‘someone other than ourselves’ as the relevant experiencer were a straightforward matter.

It isn’t, it turns out: even leaving out everything specific to Job, fully appreciating that he is the experiencer requires fully appreciating the fact that someone other than ourselves exists, is just as truly an experiencer as ourselves—is just as much a ‘seat of experience’—and is having an experience of the relevant type. Perhaps despite initial appearances, that turns out to be a highly non-trivial task.³⁴ More to the point, it’s a task that upon closer inspection is itself impossible to accomplish without using token phenomenal concepts.

To see why, consider Job’s experiencing an instance of mealy apple mouth-feel, again, with his friends Eliphaz the Temanite and Bildad the Shuhite representing that fact in the two different ways at issue. Suppose, first, that Eliphaz represents Job’s experience by summoning up a vivid memory of mealy apple mouth-feel and thinking . Suppose Eliphaz’s experience of thinking that thought—other than the component corresponding to his type phenomenal concept—is just the experience of subvocalized versions of the relevant words. He is conceptually competent with the words he is subvocalizing, but their accompanying modes of presentation aren’t relevantly experiential.

Bildad, suppose, also has a vivid memory or simulation of the mouth-feel of a mealy apple. But in him it plays the role of a token phenomenal concept of Job’s experience: he represents Job’s instance of mealy apple mouth-feel itself with an experiential mode of presentation. Bildad’s representation is thus equivalent to .

The first thing to note is that Eliphaz’s representation is relatively abstract. We can suppose he understands the experience-type and (even arbitrarily specific) phenomenal quality of Job’s experience perfectly vividly, but his representation of Job’s actually experiencing an instance of that type is verbally mediated, centrally featuring an explicit deployment the ‘lexical’ concept . The key result is that, experientially, Eliphaz’s thought is akin to .

The problem is that failing to represent the experience-type’s instantiation experientially means that Eliphaz’s thought is compatible with having only the dimmest understanding that other people exist and have experiences. So, if this thought, experienced in the way described, is the only way Eliphaz represents Job’s situation, then he fails to appreciate fully that Job is just as much a ‘seat of experience’ as he (Eliphaz) is.³⁵

Bildad’s representation, in contrast, is like the ones we might generate when reading an engrossing biography: as we read about the person’s experiences, many of us spontaneously internally simulate those experiences, as a part of our representation of their experiencing what they experienced: we are implicitly deploying an experiential mode of presentation—a token phenomenal concept, in other words—to represent those actual events (cf. Walton, 2015).

The important result is that token phenomenal concepts allow Bildad to represent Job’s actually experiencing an instance of the relevant experience-type with more complete understanding, insofar as his representation is experientially explicit about this essentially experiential relation. So Bildad’s token phenomenal concept puts him in a position to attribute to Job’s token experience, in a ‘cognitively direct’ way and while fully experientially understanding it, the experiential feature of instantiating mealy apple mouth-feel.

Now, that doesn’t guarantee that Bildad fully appreciates the fact that Job is just as truly an experiencer as Bildad is. Experientially, Bildad’s thought might be equivalent to , where corresponds to Bildad’s representation of the fact that the experiencer is Job. So it’s possible, even with an experiential mode of presentation for the instantiation of an experience-type, not to be particularly precise about who is having (or had) the relevant token experiences in the first place—let alone being specifically clear on the fact that we and the target experiencer both exist and are distinct individuals.³⁶

That said, Bildad gets much closer to full understanding than Eliphaz, all else being equal. By vividly representing the experiential nature of Job’s instantiation of mealy apple mouth-feel, Bildad understands, experientially, that a mealy apple mouth-feel experience is happening in an experiencer, whereas Eliphaz’s thought only experientially represents the experience-type in the abstract.

3.4 Normalizing other-regarding token phenomenal concepts

In short, then, we need token phenomenal concepts for effectively the same reason we need type phenomenal concepts: namely, to appreciate fully the experiential nature of their referents.

Even so, it’s natural to be especially skeptical about token phenomenal concepts of other people’s experiences—maybe because it can be hard to imagine what it’s like to deploy them, which might then fuel skepticism about their coherence in the first place.

The worry, first of all, seems to stem from the fact that other-regarding token phenomenal concepts represent something that is experiential (say, Job’s suffering) but which is distinct from the experience constituting the phenomenal concept itself (say, Bildad’s simulated or remembered token suffering experience).

A first response is thus to note that type phenomenal concepts have that same feature: they represent something—an experience-type—that is distinct from the token experience out of which which those concepts are formed. But there is also direct precedent for positing token phenomenal concepts that designate an experience (albeit still one happening in the representer of the experience) distinct from the experience constituting the token phenomenal concept itself. Balog, for instance, might even think all token phenomenal concepts work that way (see Balog, 2012, pp. 36–37).

Second, note that we seem to do something closely analogous to deploying other-regarding token phenomenal concepts whenever we remember our own past experiences or anticipate or ponder future experiences (or even merely possible ones) using experiential modes of presentation.

Think back to the last time you had a burrito at your favorite taquería, for instance (or any other specific food on any specific day). It’s not uncommon for vivid visual, olfactory, and gustatory ‘images’³⁷ to be a part of such memories. And most of the time, I submit, we have no separate thought like ; the images are just part of our mode of presentation of the token experience we had at the taquería (cf. Fernández, 2017; Montero, 2020).

In memories, then, the content of the time of occurrence argument-place in our representation is the time we’re remembering, while the time of occurrence of the memory itself is (say) now. That makes memory strongly parallel to token phenomenal concepts of other people’s experiences. The difference is just that in the latter, it is the content of the experiencer argument place, rather than the time argument place, that differs from the corresponding feature of the representation itself. The experiencer of Job’s suffering is Job, and the experiencer of the ‘representational’ remembered or simulated suffering—the one constituting Bildad’s phenomenal concept of Job’s suffering—is Bildad.

In both cases, then, a property of the relevant represented experience is importantly different from the corresponding property of the experience constituting the experiential representation itself. So if (as we seem to take for granted) there is nothing suspect about that feature of normal memory and prospective and counterfactual imagination, then there also shouldn’t be anything suspect about token phenomenal concepts of other people’s experiences, unfamiliar though they may seem.

Third and finally, though, other-regarding token phenomenal concepts needn’t even seem so unfamiliar: I suspect most of us already use them quite often in the context (similar to the ‘gripping biography’ context I discussed earlier) of hearing a friend describe an experience they went through.³⁸

4 Caligula and the (normative?) ‘Mary’s room’ of suffering

We now have my two central premises in place. First, if we form an evaluative attitude towards it at all, it’s impossible not to be intrinsically against any suffering we represent experientially as suffering—as exemplified by the case of our own, occurrent suffering (Sect. 2): otherwise, given the relevant ‘test’ conditions for intrinsic valuing, the attitude we end up forming can only be towards a less overall emotionally aversive experience than the suffering experience at issue. Second, fully understanding the fact that, say, Job is suffering requires that we represent his token instance of suffering experientially, by means of a token phenomenal concept (Sect. 3). It follows that anyone—sadist or saint—who fully understands the fact that someone else is suffering will care: they will necessarily be disposed to be intrinsically against the other person’s suffering.

4.1 Affect-components and impossible understanding, revisited

That might sound very counterintuitive, given that we’re no longer talking about our own suffering. Suppose Caligula fully understands the fact that Job is suffering, using a token phenomenal concept constituted by a specific degree of overall emotional aversiveness. Why should coming to value (positively or neutrally) what he thereby represents necessarily lessen that emotional aversiveness? It’s natural to think that the representation of Job’s suffering could be ‘affectively encapsulated,’ in other words, such that the rest of Caligula’s affective experience—including the affective contribution of his evaluative attitudes under the relevant ‘test’ conditions—would be free to vary.

My response is that, first, as long as two affective experience-components are happening within one genuinely unified mind and mental state, they will necessarily contribute to that mental state’s overall balance of positive, negative, and neutral affect. But, in keeping with Sect. 2.3, if Caligula is to meet the conditions for intrinsically valuing Job’s suffering, then he must be able to feel, simultaneously and within a single, unified mental state, both the positive affective contribution of valuing and the affective contribution of what would on its own be an accurate simulation of Job’s suffering experience, with its specific degree of overall emotional aversiveness. Otherwise, it isn’t Caligula as such but at most an encapsulated component of Caligula’s mind (opaque to Caligula himself) that fully understands.³⁹ So the degree of overall emotional aversiveness of Caligula’s experience—and thus of the token phenomenal concepts of overall affective tone that he can form—will necessarily be less than Job’s.

A last thing to note is that while we can probably get close, especially with the help of sincere descriptions from the sufferer, it’s very unlikely that we or anyone can—knowingly—simulate or recall an instance of exactly the same degree of overall emotional aversiveness instantiated by someone else’s suffering experience. It is nonetheless in a clear sense ‘humanly possible’ to do so, even if we can’t know we’re pulling it off. More to the point, though, even if we’re certainly getting it wrong, intrinsic disvaluing is still the only intrinsic evaluative attitude we can form towards their suffering, as we represent it, if we represent it experientially. All other attitudes would, in the relevant ‘test’ conditions, force us to (at best) ‘change the subject’ on ourselves, by changing the overall degree of emotional aversiveness experientially available for forming our phenomenal concept of the suffering we were supposed to be considering. So the fullest possible human understanding—but also any understanding constituted by a token phenomenal concept representing a specific degree of overall emotional aversiveness (accurately or not)—makes it impossible to form any intrinsic evaluative attitude towards someone else’s suffering other than being intrinsically against it.⁴⁰

4.2 Establishing and explaining suffering’s badness

With that in place, all we need for the conclusion that suffering is bad is the following:

If some (non-‘centered’⁴¹) state of affairs is such that necessarily, everyone who fully understood it would be disposed to disvalue it intrinsically, then that state of affairs is objectively, agent-neutrally bad.

A broad swath of theorists already endorses something like this premise. Notably, denying that anything could meet its objectivity-related ‘necessary, universal valuing’ condition seems to be what makes some people error theorists (e.g. Joyce, 2001; Mackie, 1977; probably also Wittgenstein, 1965), others various kinds of anti-realist (cf. Street, 2016), and arguably others quasi-realist expressivists (Blackburn, 1984; Gibbard, 2002, e.g., p. 54).

If the arguments in the previous sections were sound, then such philosophers might happily embrace a very minimal version of moral realism—at least on the assumption that the objective, agent-neutral badness of anyone’s suffering would constitute a specifically moral fact (cf. Machery, 2012; Marshall, 2018, e.g., Chap. 9). My main goal in this paper was to give an argument that might convince global moral skeptics,⁴² so since many or most of them accept the above sufficient condition for objective, agent-neutral badness, I may now have done all I could on that score.⁴³

That said, my arguments in this paper fit especially well with one appealing story about what explains suffering’s badness.⁴⁴ First, Sect. 2’s take on suffering and valuing aligns nicely with the idea that the ‘reason-giving force’ of suffering is grounded in the fact that it feels really bad to the one suffering. It’s natural to think, in particular, that suffering’s felt badness not only makes it reasonable for the sufferer to want it to stop but also grounds all further normative facts about suffering (cf. Kahane, 2009).⁴⁵

If that’s right, then my paper can be seen as pointing to a partial explanation for one derivative (if important) class of further normative facts: in particular, it might help explain why the reason-giving force of anyone’s suffering also ‘applies’ to or generates recognizably practical-rationality-related reasons for other people. The thought is that if fully understanding the fact that someone else is suffering guarantees that we will be disposed to disvalue it, then coming to value it positively or neutrally entails that we don’t even fully ‘know what we’re talking about’ (cf. Sobel, 2009; Lewis, 1989) when considering their suffering. If Caligula values Job’s suffering, it’s not actually as ‘Job’s instance of suffering’ (as such) that Caligula values it, but at most under some importantly inaccurate or incomplete mode of presentation.

We’re arguably all committed to holding our attitudes—attitudes that we ourselves take to be about something as it in fact is⁴⁶—accountable to the consequences of fully understanding what we’re evaluating in the first place. (Think of someone pontificating on the quality of a film they haven’t seen or the taste of a food they’ve never tried or only dimly remember.) And if so, then there’s a fairly straightforward sense in which we would be going against our own, ‘non-alien’ (cf. Railton, 1986) reasons if, faced with someone else’s suffering, we failed to respond to its reason-giving force by failing to be intrinsically against it.⁴⁷

In short, requirements like the one to ‘know what we’re talking about’ might be understood as establishing a kind of normative channel, through which the normative force of more ground-floor (say, experiential) values are passed on to other agents. This is a very speculative proposal, though, and it might not be possible to identify any candidate values other than suffering that would fit the specific pattern at issue—other than, for strictly parallel reasons, happiness.⁴⁸

5 Conclusion

In conclusion, suffering is bad. More radically, but in support of that pretheoretically uncontroversial conclusion: when it comes to other people’s suffering, it’s impossible to come to any evaluative attitude other than being intrinsically against it, as long as we fully understand the fact that they are suffering.

My argument has its share of potentially controversial premises. Probably the most in need of further support is my claim that fully understanding something essentially experiential requires not just that we can vividly remember or simulate the relevant experiential qualities, but that we actually be doing so, right in the moment of understanding. Others have defended that premise in greater depth than I did in Sect. 3.1, as I noted, but my conclusion should in any case be understood as conditional on the success of those and other arguments.⁴⁹

My argument also left many questions relevant to suffering’s badness unanswered, including whether what I’ve said is compatible with retributivist justifications for punishment,⁵⁰ what follows for what we’re morally required to do about someone’s suffering, whether the argument extends to non-human suffering (though I see no reason why it wouldn’t), and in general most of the relevant normative ethical questions worth asking.

In sum, though, I hope I’ve motivated the idea that, when we consider someone else’s suffering, intrinsically disvaluing it is not just a likely or morally fitting reaction but the only evaluative attitude towards their suffering (as such) that we can form at all—that is, if we fully understand that they are suffering in the first place. Any other attitude, I’ve argued, would make it impossible to continue to represent their specific suffering experience accurately, with its specific degree of overall emotional aversiveness.

Failing that, I hope I’ve at least highlighted the difficulty—and perhaps the glimmer of the possibility of overcoming the difficulty—of establishing and explaining this seemingly obvious fact that suffering is bad.