2.1 The basic argument

Here’s an argument, adapted from Bradford (Reference Bradford2023), against phenomenalism_WS:

1. 1. To be a welfare subject just is to have the capacity to acquire basic welfare goods and bads (i.e., welfare constituents, what makes one better or worse off).

2. 2. There are “extra-experiential” welfare goods and bads (Bradford Reference Bradford2023, 907) – basic goods and bads, which can be obtained without the capacity for consciousness.

3. ∴ Being a welfare subject doesn’t require the capacity for consciousness.

While (1) isn’t uncontroversial (e.g., Lee Reference Lee2025), we focus on (2). Why believe in extra-experiential goods? A central reason is that there is a wealth of empirical evidence from the cognitive sciences that many, if not all, types of mental states often occur nonconsciously. Such states play roughly and sometimes exactly the same roles as their conscious counterparts and fit nicely with a range of theories of mentality that explain them independently of phenomenology, such as varieties of physicalism and functionalism.

One example of nonconscious mentality is perceptual states (e.g., Block Reference Block2016; Berger and Mylopoulos Reference Berger and Mylopoulos2019; Shepherd and Mylopoulos Reference Shepherd and Mylopoulos2021). In blindsight, for example, subjects report not seeing items presented to them, though they are able to distinguish them in forced-choice scenarios (e.g., Weiskrantz Reference Weiskrantz1997). The natural explanation is that they perceive (they see) but do not consciously perceive those items.Footnote ⁸

Now, it’s not obvious that perceptual states will count as welfare goods on any theory of welfare, so that they can occur nonconsciously doesn’t tell us much. (Perhaps if veridical perceptual states constituted a form of knowledge, they could be candidates for welfare goods on an objective list theory; however, the objective list theorist will likely say that not every instance of knowledge is a basic good.) Yet as Bradford (Reference Bradford2023, 907–909) herself points out, desires can be nonconscious too, and (satisfied/frustrated) desires are plausibly one of the things that make our lives go better or worse. So, there are at least some extra-experiential goods.

How, then, might one object to (2)? According to Bradford, you just have to be an experientialist about welfare – most obviously – a hedonist:

So, while perhaps many mental states often occur nonconsciously, the thought is that the ones that constitute basic goods or bads (on hedonism) are an exception: pleasant and unpleasant states just are conscious states.

And that’s what we deny. Contra Bradford, hedonism need not be an experientialist theory. To make our case, we consider pain and assume that what we say extends to pleasant states.

Although many simply assume that pains are necessarily conscious, an increasing number of theorists contend that sensory states, and pains in particular, do occur without being conscious (e.g., Reuter and Sytsma Reference Reuter and Sytsma2020; Aguilera Reference Aguilera2020; Pereplyotchik, Reference Pereplyotchik and Corns2020). Consider Rosenthal’s (Reference Rosenthal2005, 39) example mentioned above regarding the daylong headache that subjectively comes and goes. The hypothesis that the pain is present the entire day but isn’t always conscious could be distinguished from the alternative that invariably conscious pain simply came and went. Pain plays various functional roles in the mind, such as inhibiting cognition. If we were less able to concentrate on various tasks throughout the day, then it’s reasonable to think the pain was always occurring, though at times nonconsciously.

There’s also a wealth of third-personal neuroscientific and cognitive-psychological evidence for nonconscious pains. Nociception certainly occurs without consciousness (e.g., Garland Reference Garland2012), and there is evidence that we can perceive painful stimuli while asleep (e.g., Pontén et al. Reference Pontén, Fust and D’Onofrio2019) and even while suffering from unresponsive wakefulness syndrome, or being in a persistent vegetative state (e.g., Riganello et al. Reference Riganello, Macrì, Alleva, Petrini, Soddu, Leòn-Carriòn and Dolce2016).

But now the (experiential) hedonist might protest: a neural representation of noxious stimuli is surely related to pain, and maybe that can be nonconscious. But pain, in the welfare-relevant sense, isn’t just that. That pains represent noxious stimuli isn’t what makes pains basics bads on hedonism – they must also hurt.

This yields a different way to reject (2). For now, even if there are nonconscious pains, perhaps only their conscious variants count as welfare constituents. Extra-experiential states, if there are any, have no direct bearing on welfare. Recall Pleasant/Unpleasant: if pains are basic bads, it’s because they are unpleasant. Supposing there are nonconscious pains, are they or even could they be unpleasant in this way?

We think so, although this wouldn’t show up in introspection, as nonconscious pains are not introspectable. Yet looking at theories of what makes pain unpleasant will help.

To start, consider evaluativism. Here, being in pain consists in being in an interoceptive state that represents a certain bodily condition C, and the pain’s being unpleasant consists in it representing C as having some evaluative property F, like being bad or being harmful (e.g., Bain Reference Bain and Corns2017, 40–41; Cutter and Tye Reference Cutter and Tye2011; Carruthers Reference Carruthers2018). We urge that, on evaluativism, even unpleasant pains can be nonconscious.

This is for two reasons. First, nonconscious representation of value explains results concerning subliminal motivation and instrumental conditioning. Pessiglione et al. (Reference Pessiglione2008), for example, demonstrated that subjects find larger quantities of money more strongly motivating than smaller quantities when those quantities are represented nonconsciously, suggesting that participants nonconsciously represent quantities as good or bad.Footnote ⁹

Second, it is doubtful that there’s any property that, if it can be represented at all, can only be represented consciously. We visually represent healthy grass in normal lighting as green. Yet there isn’t anything about greenness that demands conscious representation. Likewise, it’s reasonable to think that there isn’t anything about harmfulness or badness that demands conscious representation. Put differently, if there can be nonconscious mental states in the first place, then there isn’t obviously anything special about unpleasant mental states that would explain why they and they alone must be conscious.

The overall line of thought, then, comes to this:

1. 3. Pains contribute to our well-being when they are, and in virtue of, being unpleasant [Pleasant/Unpleasant].

2. 4. Pains are unpleasant in virtue of representing evaluative properties.

3. 5. Pains do not represent negative evaluative properties (e.g., badness, harmfulness) in virtue of being conscious and can represent such properties without being conscious.

4. ∴ Pains need not be conscious to contribute to our well-being.Footnote ¹⁰

The conclusion is just what you’d expect on mentalism. Does it show that phenomenalism_WS is false?

Close, but not quite. Recall (2):

1. 2. There are “extra-experiential” welfare goods and bads – basic goods and bads, which can be obtained without the capacity for consciousness.

One thing the phenomenalist_WS might say – which doubles as an objection to mentalism – is although token mental states, including welfare-relevant pains, can occur nonconsciously, some background capacity for consciousness is necessary for a creature to exhibit mentality of any type in the first place.Footnote ¹¹ That’s enough to refute (2) and cut against mentalism in one fell swoop, since the mentalist wants to hold that welfare goods/bads can be had in subjects without any capacity for consciousness.

This style of view stems from assuming something akin to Searle’s (Reference Searle1992) connection principle, which says that what makes a state mental is that it is potentially conscious. But as many have argued (e.g., Rosenthal Reference Rosenthal1990; Fodor and Lepore Reference Fodor and Lepore1994), there is no formulation of this principle that renders it even prima facie plausible. The notion of a state that is not conscious but that could “in principle be conscious” is at best obscure.

There is, however, a way of rejecting (2) that cuts against mentalism alone. This is to claim that the evidence of nonconscious mentality of the sort we cite above fails to show such states are really not conscious or, at best, establishes there are nonconscious, but nonmental, information-carrying states, akin to states of the retina that carry information about edges in the early visual system or states of the stomach tracking acidity levels (e.g., Phillips Reference Phillips2018). If that were so, one might wonder how such a thing could be good or bad for the subject (e.g., Birch Reference Birch2022, 5).

This is what the mentalist will say: there is a space between mere nonmental information processing and consciousness – and that such states are welfare-relevant. That is, states may be nonconscious, and yet also good or bad for the subject insofar as they comprise the subject’s mind.

A full story – which we can’t supply here – would need an account of the mark of the mental. But since our main goal is to counter phenomenalism, not defend mentalism per se, we’re largely neutral about the nature of mentality – and there would seem to be many accounts of mentality to which we could help ourselves (see, e.g., the essays in Voltolini Reference Voltolini2025). These include views, for example, on which mental states are personal level, as opposed to subpersonal, or states that exhibit genuine intentionality, as opposed to mere information processing. The only mark of the mental that we definitely deny is that the mental is either necessarily or potentially conscious.Footnote ¹²

For our part, we’re drawn to the view that the mental corresponds to the personal level. Here, mental states are roughly attributable to whole persons, while nonmental states are attributable only to subsystems of a person. For example, we think it is right to say that it is the person with blindsight that sees, but does not consciously see, their environment – and not simply that their visual system detects it.

Granted, the personal-subpersonal distinction is vexed (e.g., Dennett Reference Dennett1969; Drayson Reference Drayson2020), and without a clear delineation between the personal and the subpersonal, much less the mental and nonmental, one might wonder whether there could be subpersonal welfare goods/bads – perhaps subpersonal bodily damage is an example – and object that mentalism is not compatible with them.Footnote ¹³ Surely having a welfare level is a personal-level state, but it might not immediately follow that all welfare constituents (goods and bads) are likewise personal level.

Yet here the pendulum begins to swing back against phenomenalism. For if there are subpersonal welfare goods/bads, then that would be all the worse for that view. After all, few would deny that the subpersonal is nonconscious, or that there can subpersonal states in entities wholly devoid of the capacity for consciousness. Conditional on hedonism, subpersonal welfare constituents are unlikely.Footnote ¹⁴ But assuming there are any, if these subpersonal states are mental, mentalism can accommodate them. The view would then simply require an account of the mental independent of not only consciousness, but the personal level, such as an account which grounds mentality in nonconscious intentionality. In any case, what we are mostly challenging currently is the contention that, simply because a state is mental, it is thereby conscious, and that only conscious mental states are welfare constituents.

2.2 The argument generalized

Our initial argument began with neutral territory, premise (3), that is, Pleasant/Unpleasant, and then offered an account of pleasantness and unpleasantness in (4) with evaluativism. So, a natural question at this juncture is whether analogs of (5) remain true given various adjustments to (4). That is, does the argument generalize beyond evaluativism?

Yes. Consider this adjustment to (4):

Premise (4*) invokes an attitudinal view of unpleasantness (e.g., Feldman Reference Feldman2002).Footnote ¹⁵ But it too is consistent with mentalism insofar as arguably all attitudes can be and often are nonconscious. Consider, for example, Nisbett and Wilson’s (Reference Nisbett and Wilson1977) classic work on confabulation. Participants selected consumer items from an array and reported that they did so because of the intrinsic qualities of the items when in fact the order of the items in the array explained their choices. The natural interpretation is that they confabulated their preferences, reporting that they (consciously) desired an item for its look when, in fact, they (nonconsciously) desired it for its location.Footnote ¹⁶ And while “being displeased that p” may not exactly be a desire, the point remains the same.Footnote ¹⁷ So, while the case for nonconscious desires and attitudes more generally is certainly not settled (see, e.g., Smithies and Weiss Reference Smithies and Weiss2019), it’s plausible that the relevant analog of (5) remains true even on (4*).

Or consider this version of (4):

In (4**), we have imperativism. On arguably the most developed version of this view (Klein Reference Klein2015), ordinary pains have two components: a command directed at the damaged body part, and a higher-order command directed at the first-order command. The latter gives pains their unpleasantness (Ibid., 183–186). This is different from evaluativism, since imperatives aren’t indicative representations – they don’t attribute properties (such as evaluative properties) to things. But that doesn’t change things. For neither must the higher-order command be directed at conscious pains, nor must the higher-order command be itself conscious. So, the analog of (5) holds once more.

A final option:

Premise (4***) employs the distinctive feeling theory (e.g., Bramble Reference Bramble2013). This view may seem at first to be the most obviously incompatible with mentalism. But it’s not. The problem here is that expressions such as “feeling” are ambiguous. We sometimes use such terms to pick out the conscious variants of pains, but pain doesn’t cease to be a feeling simply because it isn’t conscious. And as we have seen, pains can be unpleasant without being conscious. Just as there is nothing more to dancing a dance than simply dancing, there is plausibly nothing more to feeling a feeling than having that feeling (cf. Block Reference Block2007, 484). Crucially, this is regardless of whether the individual can report on how that feeling feels if it isn’t conscious. Nonconscious pains are feelings.

In sum: none of the (4)s, including even (4***), obviously support phenomenalism_WS, and mentalism is neutral between (4), (4*), (4**), and perhaps even (4***). Thus, the argument against phenomenalism_WS, and the companion case for mentalism, generalizes.

2.3 Further concerns

Before we press on to phenomenalism_CW and its mentalist counterpart, here are three further concerns.

First, one might object that the view proposed here is less unified than its phenomenalist counterpart.Footnote ¹⁸ As noted at the outset, different types of states – pains, fear, depression, and so forth – may all be unpleasant and thus welfare bads. We must consequently distinguish the properties that type mental states from the properties in virtue of which such states are unpleasant. A form of phenomenalism (or a form of experiential hedonism) offers a straightforward and unified account of the latter: unpleasant states are all unpleasant in virtue of their conscious unpleasant phenomenal character.

That would indeed be a unified, simple view, and arguably supportive of some form of phenomenalism. But not all phenomenalisms. For the view just sketched effectively rejects the variants of (4) and offers a nonreductive account of unpleasantness. By contrast, any reductive phenomenalism that seeks to explain unpleasantness in nonphenomenal terms will face similar questions about unity.

To see this, suppose we paired experientialism with evaluativism. Again, that conjunction holds that a state’s unpleasantness consists in its representation of negative evaluative properties, and such representation is conscious. The problem is that there are arguably states that represent such properties, which are not unpleasant and are thus not welfare bads. Consider simply thinking that something has a negative evaluative property. There must therefore be something that explains why certain representations of negative evaluative properties, such as pains, are unpleasant, whereas other representations, such as thoughts about those properties, are not. Since many think that thoughts themselves are frequently phenomenally conscious (e.g., Pitt Reference Pitt2004), even the phenomenalist will have to appeal to some other feature to explain why thoughts with negative evaluative content aren’t basic bads, but pains with such contents are.

Evaluativists have acknowledged this problem; Bain (Reference Bain and Corns2017, 49–50), for example, urges that states are unpleasant just in case they represent evaluative properties in a distinctively sensory way that “impresses” their badness on the subject. In that way, a state’s unpleasantness consists not only in a certain type of representational content, but also in its being represented by a particular type of attitude toward or mode of that content. Such modes thereby play the role of both typing states and explaining their unpleasantness.

One might worry that, if pains have such modes, then it is incumbent upon the mentalist to show not only that evaluative properties can be represented nonconsciously, but that modes can occur nonconsciously as well. Yet in the relevant sense, a state’s mode just is the way that state represents its content. After all, everyone must agree there is some difference between the belief that p and the desire that p, or a pain that represents an evaluative property and a thought that does so, even if those pairs of states have the same contents. Whatever those differences consist in, to the extent we have shown those contents may occur nonconsciously, we ipso facto show those modes can occur nonconsciously as well. So, there is no additional burden on mentalism. And to emphasize, our point isn’t that the (4)s, including evaluativism, are incompatible with phenomenalism_WC. It’s that they provide no motivation for phenomenalism_WC.

Second, one might protest that we have treated nonconscious pains (and nonconscious mentality generally) as a straightforwardly empirical matter. Isn’t our concept of pain a phenomenal concept, one that picks it out in respect of its conscious character?

We’re not denying that we may have a phenomenal concept of pain. Yet even if we have a concept of X that picks it out in respect of feature Y, Y needn’t be a necessary feature of X. Our ordinary concept of water arguably picks it out in respect of its being clear and potable, but water needn’t be clear and potable. So, we can instead pick out water by, for example, its chemical concept. Likewise, we may often conceptualize pain as conscious, but pain is not essentially conscious, and we have a physical or functional concept of pain according to which pains needn’t be conscious.

Third, one might worry more generally that part of what it is to have a mind is to have a perspective on the world, and nonconscious states can’t “coalesce” in the right way to form one. Chalmers (Reference Chalmers2022, 341) captures the point this way: “[p]hilosophical zombies lack consciousness… there is arguably no one home to mistreat.” But while zombies may lack a conscious perspective, according to mentalism, someone is still home. Nonconscious perceptual states present a perspective insofar as they modally encode information about the environment from a particular egocentric origin. And pains only have motivational force if they have de se-like cognitive significance (cf. de Vignemont Reference Vignemont de2015), and so similarly can comprise perspectival facts. Zombies are subjects with a point of view on the world.

So much for phenomenalism_WS. What of phenomenalism_CW?