The epistemic import of phenomenal consciousness

ABSTRACT

This paper controverts the ability of intentionalism about perception to account for unique epistemic significance of phenomenal consciousness. More specifically, the intentionalist cannot explain the latter without denying two well-founded claims: the transparency of experience, and the possibility of unconscious perception. If they are true, intentionality of perception entails that phenomenal consciousness has no special epistemic role to play. Although some intentionalists are ready to bite this bullet, by doing so they effectively undermine one of the standard motivations of their view, i.e. the claim that perceptual experiences justify beliefs. Consequently, whatever reason might there be to think that phenomenal consciousness has unique epistemic import, it is also a reason to reject intentionalism. I recommend replacing the latter with an unorthodox formulation of relationalism about perception.

KEYWORDS:

• Phenomenal consciousness
• epistemic justification
• transparency of experience
• unconscious perception
• phenomenal overflow
• understanding

The term ‘phenomenal consciousness’ (from now on, P-consciousness) refers to the qualitative aspects of conscious experience. If a mental state is phenomenally conscious, there is something it is like to be in that state. P-consciousness contrasts with access consciousness (A-consciousness), i.e. the availability of mental content for reasoning and rationally guiding speech and action (Block 1995).

On the face of it, P-consciousness seems to play an indispensable role in acquiring knowledge. Part of the reason why rocks, viruses and laptops are not considered knowers is that there is no good reason to think that they can have P-conscious mental states. Part of the reason why it makes sense to ascribe knowledge to cats, horses and octopuses is that there are some good reasons to think they have P-conscious mental states. And yet attempts at elucidating the link between knowledge and P-consciousness tend to be so obscure that they led many thinkers to believe that the latter is epistemically insignificant. I will argue that this verdict is precipitate. The difficulty of explaining the epistemic role of P-consciousness is a theoretical artefact.

Since the paradigmatic example of P-consciousness is perceptual experience, ascribing unique epistemic significance to P-consciousness usually takes form of phenomenalism, i.e. the view that ‘perceptual experiences provide justification for beliefs at least partly in virtue of their conscious phenomenal characters’ (Berger, Nanay, and Quilty-Dunn 2018, 573).

This paper argues for two claims, one negative and one positive. The negative claim is that phenomenalism is untenable if the intentional theory of perception is true. For intentionalism cannot explain the epistemic significance of P-consciousness without denying two well-founded claims: the transparency of experience, and the possibility of unconscious perception. The positive claim is that unique epistemic significance of P-consciousness can be explained in terms of an unorthodox formulation of the relational theory of perception. In contrast to intentionalist phenomenalism, relationalist phenomenalism (at least my version of it) is compatible with transparency of experience and the possibility of unconscious perception.

If both claims are correct, whatever reason one might have to think that P-consciousness is uniquely epistemically significant, it is also a reason to prefer relationalism over intentionalism. But if there is no good reason to think P-consciousness is epistemically special, one of the standard motivations of intentionalism is defunct, i.e. it is not true that perceptual experiences justify beliefs.

Section 1 outlines a comprehensive case for intentionalist phenomenalism offered recently by Smithies (Smithies 2019). Section 2 argues that intentionalist phenomenalism fails if perceptual experience is transparent. Section 3 argues that intentionalist phenomenalism fails if unconscious perception is possible. Section 4 explains the epistemic role of P-consciousness in terms of an unorthodox formulation of the relational theory of perception. Section 5 concludes the paper with a dilemma: either perception is intentional and P-consciousness is epistemically insignificant, or perception is not intentional and P-consciousness can play some special epistemic role.

1. Intentionalist phenomenalism

As I understand it in this paper, intentionalist phenomenalism consists of two claims: (a) intentionalism, i.e. the view that perception has intentional content that represents the world as being a certain way; (b) phenomenalism, i.e. the idea that a perceptual experience (i.e. a conscious perception) plays its epistemic role at least partially in virtue of its phenomenal character (i.e. a set of properties that determine what it is like to undergo that experience from the first-person perspective).²

The aim of this section is to elucidate intentionalist phenomenalism. I do this by taking a closer look at one of the most recent and most developed versions of intentionalist phenomenalism offered by Smithies (Smithies 2019). Nevertheless, the target of the negative part of the paper (Sections 2 and 3) is not just Smithies’ view, but intentionalist phenomenalism in general.

To say that P-consciousness has unique epistemic significance is to say that there is an epistemic status E such that being P-conscious is necessary for having E. Smithies (Smithies 2019, 36) identifies E with playing a role in epistemic justification. When it comes to perceptual experience, that role consists in providing epistemic justification that is (i) immediate (i.e. it does not depend on a posteriori justification to believe any other proposition), (ii) defeasible (i.e. it can be defeated by a posteriori justification to believe other propositions), and (iii) propositional (i.e. having it does not require making any use of it³) (Smithies 2019, 74–75).

Smithies also endorses representationalism, a version of intentionalism according to which the phenomenal character of perceptual experience is a kind of mental representation (Smithies 2019, 90–91). The representationalist believes that the phenomenal character of a perceptual experience is exhausted by representational properties of that experience (Smithies 2019, 37).

But it does not follow that every perceptual representation has a phenomenal character. If having a phenomenal character is an indicator of being conscious, representationalism allows for unconscious perceptual representation. Smithies mentions two examples of such representations: (i) an a priori example of a perception had by a P-unconscious philosophical zombie, and (ii) an a posteriori example of type-1 blindsight.⁴ He claims that in both cases unconscious perception does not deliver perceptual justification because it lacks phenomenal character.

Consider a P-unconscious zombie who has perceptual justification for believing the same propositions as its P-conscious twin. Call it ‘epistemic zombie’. According to Smithies (Smithies 2019, 9, 14–15), epistemic zombies are inconceivable because epistemic capacities are defined in terms of their connections with P-consciousness (i.e. they cannot be functionally defined in terms of their causal roles, see also Section 3.1).

In Section 4.1, I argue that epistemic significance of P-consciousness cannot be determined by considerations about the conceivability of philosophical zombies. But Smithies does not base his case on a zombie argument. Instead, he compares ordinary conscious seeing to blindsight in order to demonstrate that the phenomenal character of perceptual experience is not only necessary (N), but also sufficient (S) for perceptual justification. N alone attributes unique epistemic significance to P-consciousness. S adds that perceptual experience delivers perceptual justification solely in virtue of being P-conscious.

Regarding N, Smithies (Smithies 2019, 81–82) emphasizes that, in contrast to ordinary conscious perception, blindsight does not dispose the subjects to form beliefs about the stimuli in their blind fields. Instead, the subjects tend to withhold judgment. If blindsight provided perceptual reasons for beliefs about the stimuli in the blind field, withholding judgment about those stimuli would be irrational. But blindsight does not cause any deficit in rationality. Therefore, blindsight is not a source of perceptual justification.

Blindsight research involves a large amount of experiments on a very small amount of subjects. Typically, the more experiments a blindseer partakes in, the more confident they get about their guesses concerning what is shown in their blind field. They start to form beliefs (or at least make conjectures) about the stimuli as they acquire background knowledge about the reliability of their blindsight (cf. Overgaard and Mogensen 2015). Crucially, however, that knowledge is gained through conscious perception. Hence blindsight in itself is insufficient for perceptual justification.

However, these remarks only suggest that consciousness in general is necessary for perceptual justification. If blindsight is unconscious perception, it lacks not only P-consciousness, but also A-consciousness. Why is P-consciousness necessary for perceptual justification? Why mere A-consciousness would not be sufficient?

Smithies (Smithies 2019, 83–86) responds that, while A-consciousness is necessary for doxastic justification, it is neither necessary nor sufficient for propositional justification. It is not necessary because one can possess a perceptual reason without being able to use it in thinking or guiding action. It is not sufficient because there is no relevant difference between blindsight and a hypothetical case of ‘super-blindsight’ (i.e. A-conscious P-unconscious perception) that would explain why justification is absent in the former case but present in the latter. The super-blindseer, just like the ordinary blindseer, is merely disposed to reliably form beliefs about certain stimuli. The only difference between blindsight and super-blindsight is that the super-blindseer does not need any encouragement from the experimenters to do so.⁵

As to S, Smithies (Smithies 2019, 91) claims that conscious perceptual experience is a source of perceptual justification because its phenomenal character has presentational force (PF). Since mental states such as unconscious perception and imaginative experience lack this type of phenomenal character, they cannot provide perceptual justification. This seems to contradict representationalism, as the latter entails that the phenomenal character of perceptual experience is reducible to its intentional content, and that for every perception there is a possible imaginative experience with the same content. However, Smithies (Smithies 2019, 92–93) endorses an ‘impure’ form of representationalism, according to which a perception of X and an imaginative experience of X are phenomenally different despite having the same intentional content. The difference is PF, which is present in conscious perception but absent from sensory imagination.

Smithies’ case for S plays a fundamental role in his account. Whatever it is about perceptual phenomenology that substantiates S is also crucial for maintaining N. If the phenomenal character of perceptual experience was not epistemically loaded in any way, it would not make the relevant difference between P-conscious perception and blindsight. If that was the case, both S and N would be false. Thus, the whole account hinges on PF.

2. The epistemic impotence of presentational force

The explanation of the epistemic role of P-consciousness in terms of PF is one of the main targets of this paper. Section 2.1 briefly explains what PF is supposed to be. Section 2.2 argues that attributing the epistemic import of P-consciousness to PF is inconsistent with the independently plausible thesis that experience is transparent.

2.1. What is presentational force?

If S is true, perceptual experience provides the perceiver with propositional justification in virtue of its phenomenal character alone. S presupposes that the phenomenal character of perceptual experience has PF. But what exactly is PF?

As already mentioned, PF differentiates perceptual experience from imaginative experience. Pace Hume, this difference cannot be explained by simply pointing out that the former is phenomenologically clearer and more stable than the latter, since a blurry perceptual experience could be more unstable and unclear than an imaginary experience. More importantly, even the blurriest perceptual experience has stronger epistemic import than the clearest imaginary experience, at least as far as justifying beliefs about the mind-independent world is concerned. The purpose of positing PF is to explain why it is so.

Here are some representative elucidations of PF: (i) an assertoric force that presents experience as veridical even if the subject believes that it is not (Heck Jr. 2000, 508), (ii) something that makes it seem to the subject that an object represented by the content of their perceptual experience actually exists (Huemer 2001, 79), (iii) distinctive phenomenology of seeming to ascertain that a given proposition is true (Pryor 2004, 357), (iv) something that not only makes it seem to the subject that p, but also makes it seem to them as if their experience makes them aware of a truth-maker for p (Chudnoff 2013, 32–40).

Smithies (Smithies 2019, 94) stresses that PF is not a supposition accompanying perceptual experience to the effect that the content of that experience seems to be true. For one thing, such suppositions are known to occur even when the subject is not undergoing the target experience (vide Anton's syndrome). More importantly, if PF was a cognitive phenomenon, it could not substantiate the claim that P-consciousness has unique epistemic import.⁶ Hence it is unsurprising that Smithies views PF as an experiential phenomenon. He describes it is an introspectively accessible phenomenal aspect of the phenomenal character of perceptual experience that is absent from the phenomenal character of sensory imagination. This is supposed to explain why perceptual experience does but imaginative experience does not provide perceptual justification.

PF-theorists usually endorse intentionalism, which compels them to hold that PF is present in the phenomenal character of hallucination. Since representing X is consistent with X's inexistence, intentionalism entails that every genuine perception has a possible hallucinatory duplicate, i.e. an objectless experience with qualitatively identical phenomenal character. Call it ‘perfect hallucination’ (P-hallucination). The possibility of P-hallucination motivates the view that perceptual justification is defeasible (see also Section 4.1).

2.2. Presentational force and the transparency of experience

To explain why P-consciousness has unique epistemic significance, PF must be experiential in nature. But if it is experiential, it is either inconsistent with the transparency of experience, or its epistemic import is negligible.

According to the transparency thesis, introspection of a perceptual experience does not provide direct awareness of any properties of the experience itself. When one introspectively focuses on one’s perceptual experience, one becomes aware of nothing but the perceived objects, their properties and relations. This is not to say that one cannot become aware of one’s perceptual experience through introspection at all. The point is only that introspective awareness of one’s perceptual experience does not reveal any structural features of the experience. Introspection reveals perceptual experience indirectly, by revealing the perceived objects, their properties and relations.

There is currently a consensus (or something close to it) among philosophers of perception that the object of perception is mind-independent. Formulations of transparency usually reflect that consensus by mentioning ‘mind-independent’ (Martin 2002, 378) or ‘externally located’ (Gow 2019) objects (along with their properties and relations) as the only things that are revealed by the introspection of perceptual experience. Importantly, this should not be taken as implying that mind-independence is one of the properties that perceptual experience makes one perceptually aware of. After all, mind-independence is not something that can be seen or heard. If it could be, making sense of the notion of mind-independence would not be as difficult as it actually is (cf. Rosen 1994). This caveat will become relevant shortly.

While some intentionalists endorse transparency (see e.g. Tye 2002, 2014), the claim also figures among the main motivations of relationalism (a.k.a. naïve realism), the main competitor of intentionalism in metaphysics of perception. Relationalism explains the phenomenal character of conscious perception in terms of a relation between the subject and the mind-independent object.⁷ While the basic idea of transparency is the same in both cases, each side unpacks it differently. The relationalist believes that the perceived features of the environment are presented to the subject, whereas those of intentionalists who also endorse transparency think that these features are represented to the subject.

Transparency can be understood as either a phenomenological or a metaphysical claim (Gow 2016). According to phenomenological transparency, experience introspectively seems to be transparent, which leaves open the possibility that it is not in fact transparent. Metaphysical transparency is a stronger claim that perceptual experience does in fact make the subject aware of nothing but the perceived objects, their properties and relations.

Still, even if only phenomenological transparency is true, PF turns out to be explanatorily idle. For PF to make the kind of impact described in Section 2.1, experience must appear to have PF. But if experience is phenomenologically transparent, attending to the phenomenal character of perceptual experience only reveals the perceived objects, their properties and relations, and PF is clearly none of those things. Of course, phenomenological transparency allows that introspective training could enable one to attend to mind-dependent features of experience, and PF could be one of those features. Nonetheless, characterizing the impact of PF as dependent on phenomenological training defeats the purpose for which it was posited, since it renders PF dependent upon the background knowledge of the subject, which in turn compromises the immediacy of perceptual justification.

The detractor could try to rebut this argument by simply rejecting transparency (cf. Crane 2000). But this move is not easy to make given the cogent arguments that have been recently put forward in favour of transparency. As it turns out, not only is transparency immune to phenomenological considerations about after-images, phosphenes, blur, and the like (Gow 2019), but also denial of transparency seems to be at odds with the current state of art in neuroscience of attention (Weksler, Jacobson, and Bronfman 2021). And even if it were assumed that considerations for and against transparency cancel each other out, my argument would still stand because it is a conditional argument. Its point is not so much that intentionalist phenomenalism fails because transparency is true; it is rather that intentionalist phenomenalism fails if transparency is true. If correct, my argument is relevant to the current debate in metaphysics of perception even if the status of transparency is regarded as uncertain. To see why, suppose that you have some good reasons to believe that phenomenalism is true. If my argument is correct, any point made in favour of transparency will be a reason for you to reject intentionalism.

Another possible objection is to deny that there is any clash between transparency and PF. For instance, Dorsch (Dorsch 2018) uses transparency in support of the view that perceptual experience provides non-inferential and non-conceptual access to perceptual reasons. According to him, transparency means that attending to one's experience reveals not only the perceived objects, their properties and relations, but also their being externally located, their existence, and their being a part of one's environment. As a result, perceptual experience creates an ‘impression of relationality’, which makes it seem that experience ‘nomologically depends on reality':

It is in this sense that perceptual reasons are phenomenally present to us: from the inside, veridical perceptual experiences seem to possess a property (i.e. the property of being determined by reality) which is a determinant of the determinable property of being reason-providing. (Dorsch 2018, 217)

Dorsch (Dorsch 2018, 219) claims that this impression of relationality is a phenomenologically accessible generic structural feature of perceptual experience. Each perceptual reason combines this feature with some specific perceptual content.

The problem with this proposal is that it misconstrues transparency. Recall the caveat I have made when introducing transparency at the outset of this section, i.e. that being perceptually aware of mind-independent objects does not amount to being perceptually aware of their mind-independence. If my visual experience of my cat is transparent, all that I am visually aware of is the cat. Introspection of this experience reveals only those of the cat’s properties that are visible from my current visual perspective. That is all that transparency means. I am not visually aware of my cat's mind-independence, its existence, let alone the nomological dependence of my experience on reality. If transparency is true, none of those things is part of the phenomenology of my experience.

Phenomenological transparency supports, and is explained by, the claim that the objects of perception are mind-independent, and that perception depends nomologically on reality. But phenomenological transparency does not entail that visual experience makes one visually aware of the mind-independence of its objects, nor of its nomological dependence on reality. Dorsch conflates the features that explain why experience is transparent (e.g. my cat's mind-independence) with the features that experience makes one visually aware of in virtue of being transparent (e.g. the colour of my cat's fur). When the conflation is removed, Dorsch's view turns out incompatible with phenomenological transparency.⁸ If experience reveals only the perceived objects, their properties and relations, it cannot reveal any of its own structural properties. If transparency is true, the claim that ‘what we have access to from the inside goes beyond the sensory (e.g. visual) presentation of objects and their features’ (Dorsch 2018, 222) is false.

The intentionalist may respond by insisting that mind-independence is reflected in the mode of perceptual representation, not in its content (Crane 2000; Smithies 2019, 44–46). In other words, mind-independence features in the phenomenal character of perceptual experience as a way in which the cat is represented, not as a property the cat is represented as having.

It is disputable whether the content/mode distinction explains anything at all about the phenomenal character of experience. According to Bourget (Bourget 2017b, 683), it merely restates the explanandum in some technical jargon:

we have no real grasp of how the combination of intentional modes and contents is supposed to generate phenomenal character, or of what intentional modes are. […] We understand the schematic idea that phenomenal character is jointly determined by content and mode, but we have no idea how the combination is supposed to occur. This obscurity has its roots in the fact that the phenomenal character of an experience does not, as far we can tell in introspection, divide into an intentional mode and a content. […] As a result, we have no ability to observe the combination process that is posited, or of contemplating intentional modes in isolation. This makes it hard to see how modes and contents are supposed to combine to yield phenomenal characters. (Bourget 2017b, 683)

The content/mode distinction is not an introspective given; it can’t be ‘read off’ of a conscious experience. Neither does it add anything to what we already know from introspection. The purpose of drawing the distinction is to account for a difference in phenomenal character between experiences with the same content (e.g. perceiving X vs. imagining X, or perceiving X vs. thinking about X). As Bourget (Bourget 2017b, 683) points out, the reason why the distinction might seem to explain a phenomenal difference between the two experiences with which one is familiar is that one already knows from introspection how they differ phenomenologically. But once one considers an experience one has never had before (e.g. a perceptual experience of something one has only experienced in thought), it turns out that the content/mode view’s prediction about the phenomenal character of that experience is unknown.

The burden of proof is on the intentionalist phenomenalist to show that the content/mode distinction is not virtus dormitiva. But even if the content/mode distinction does provide a genuine explanation, note that the mode is supposed to be a structural feature of experience. If the mode is reflected in the phenomenal character of experience along with the content, it follows that a structural property of a perceptual experience can be revealed by introspection of that experience. Hence the proposal in question is inconsistent with transparency, which brings us back to the problems of the previous objection.

To preserve transparency, the intentionalist can attribute PF to intellectual seemings that accompany perceptual experiences (Gow 2019). Intellectual seeming can be defined as either an inclination to believe or a kind of experience (characterized by a specific sort of cognitive phenomenology and propositional content) that co-occurs with perceptual experience. But this will not save Dorsch’s account either. Because intellectual seeming is distinct from perceptual experience, it does not lend itself to account for the epistemic significance of the latter. Even though it may have its own specific phenomenology, its epistemic import is essentially cognitive. This is because the content of an intellectual seeming depends on the beliefs the subject has about their current experiential situation.

One should not read too much into transparency, but the same can be said about PF. Am I expecting too much from PF? Maybe the concept of PF simply captures the fact that people are naturally inclined to let their perceptual experiences guide their action and belief formation unless they are dissuaded from doing so. Thus understood, PF seems consistent with transparency. Call it ‘transparent presentational force’ (T-PF).

Consider Dorsch's ‘experiential intentionalism’ (Dorsch 2010, 192–195). On this view, the phenomenal character of perceptual experience is by default perceptual (i.e. committal with respect to existence of the experienced object) even though P-hallucination is possible. This is because perception and P-hallucination both present themselves as committal regarding the existence of the experienced objects, albeit in cases of the latter sort this non-neutrality is merely apparent. Experiential intentionalism can be interpreted as an example of T-PF.

The problem with T-PF is that its epistemic significance is so negligible that it opens its proponent to the charge of supporting unique epistemic significance of P-consciousness with lip service only.

First, note that Dorsch (Dorsch 2010) introduced experiential intentionalism in response to Martin's argument against intentionalist account of visual imagination (Martin 2002). The objective was to show that intentionalism can accommodate Martin's observation that visualising consists in imagining a visual experience that is just as transparent as ordinary seeing. So the cost of assimilating PF to Dorsch's account is that T-PF is the same no matter whether the experience is genuinely perceptual, hallucinatory, or imaginative. Hence T-PF is not truth-conducive.

Second, because the impact of T-PF boils down to actualizing some doxastic propensity of the subject, its effect can be described as an automatic reaction, akin to conditioned response. As a natural inclination of the subject, it exercises its influence irrespective of the extent to which the subject conforms their beliefs to what their evidence supports. So T-PF does not play any role in epistemic justification, at least insofar as being justified amounts to conforming to the standards of epistemic rationality (cf. Smithies 2019, 74). On the contrary, its epistemic relevance is so marginal that attributing any rational role to it can be accused of mistaking causation for justification.

By the same token, it might be argued that unconscious perception is no different from conscious perception as far as T-PF is concerned. After all, if unconscious perception exists, it also systematically produces certain cognitive and behavioural effects (recall the super-blindsight case discussed in Section 1). And if unconscious perception has T-PF, the latter is not a distinctive feature of P-consciousness. Generally speaking, it seems that T-PF can only provide justification in the externalist sense, which makes it explanatorily useless for epistemic internalists such as Smithies and other intentionalist phenomenalists.⁹

To sum up, PF is either incompatible with transparency, or unable to secure any epistemic role for P-consciousness. If transparency is true, PF fails to support S and N.¹⁰ Perhaps there is some way to defend intentionalist phenomenalism without attributing PF to the phenomenal character of perceptual experience, but here the burden of proof is on the intentionalist phenomentalist. Whether or not they can show that such an account is available, PF-based intentionalist phenomenalism is incompatible with transparency.

3. The epistemic import of unconscious perception

To explain why P-consciousness has unique epistemic significance, it is not enough to indicate some phenomenological feature that plays some special epistemic role. A satisfactory account has to (i) explain what makes conscious perception epistemically superior to unconscious perception, and (ii) acknowledge what conscious and unconscious perception have epistemically in common. I will argue that intentionalist phenomenalism cannot meet these desiderata. Section 3.1 sets the stage. Section 3.2 presents the argument. Section 3.3 summarizes the negative part of the paper.

3.1. Internalism vs. externalism

While the nature of perception is a matter of many controversies, it is commonly accepted that perception guides action. If a mental episode cannot guide action, it is not a genuinely perceptual episode. For this reason, one of the key issues in the current debate about unconscious perception is whether the presumptive instances of unconscious perception can guide action. Both the enthusiasts and the sceptics about unconscious perception agree that this is a proper criterion for determining whether perception can occur unconsciously (Block and Phillips 2017).

If the potential to guide action is a good measure of whether an episode is perceptual, so is epistemic import. If an episode has no epistemic import, it is not a perceptual episode. For example, blindsight is perception as long as it has some epistemic import, even if the latter is vastly inferior to the epistemic import of conscious perception. Otherwise, it would be doubtful whether blindsight is perception at all. This requirement applies to all presumptive instances of unconscious perception.

In fact, the presumptive action-guiding role of unconscious perception is arguably conditional on its epistemic import. Suppose that unconscious perception of an object O is accompanied by unconscious recognition that O is F, and that the former provides epistemic warrant (i.e. ‘justification’ in the externalist sense of the term) for the latter. If so, it can be argued that unconscious perception-based unconscious recognition that F-ness is instantiated in the environment causes (and rationalizes) unconscious intention or volition to behave in a certain way. On this view, the behavioural effect of unconscious perception is rationally explicable from the third-person perspective, which constitutes a substantive reason to regard such behaviour as an instance of action. But if behavioural effects of unconscious perceptual representation are never preceded by unconscious recognition and unconscious intention/volition, they’re arguably better described as automatic impulses or reflexes that fall short of genuine action.

If this is correct, there is clearly a tension between the possibility of unconscious perception and intentionalist phenomenalism. For the latter is closely associated with epistemic internalism, which, in most of its forms¹¹, confines the reach of epistemic rationality to consciously accessible contents. Hence the intentionalist phenomenalist will either ignore unconscious perception as epistemically irrelevant or provide some reasons for thinking that it is epistemically irrelevant. Here again Smithies’ version of intentionalist phenomenalism provides a representative example. For Smithies endorses the following claims:

Necessarily, which propositions you have epistemic justification to believe at any given time is determined solely by your phenomenally individuated mental states at that time. (Smithies 2019, 25)

Necessarily, perception justifies belief about the external world if and only if it has some phenomenal character. (Smithies 2019, 82)¹²

Although Smithies believes that unconscious mental representation plays an indispensable explanatory role in psychology and cognitive science (Smithies 2019, 35, 48), he regards unconscious perceptual representations as ‘subdoxastic states’, which makes them ‘consciously inaccessible: if I subdoxastically represent that p, then I’m not thereby disposed to judge that p when I entertain the question whether p’. (Smithies 2019, 124). According to Smithies, ‘your “subdoxastic” mental representations, unlike your beliefs, cannot affect which propositions you have epistemic justification to believe’ (Smithies 2019, 26). It follows that ‘unconscious perceptual information in blindsight doesn’t provide epistemic justification for beliefs about the external world’ (Smithies 2019, 25).

Given the foregoing, Smithies’ intentionalist phenomenalism is committed to the kind of scepticism about unconscious perception that can be found in the works of Phillips (Phillips 2016, 2018b). Relatedly, Smithies believes that ‘your zombie twin doesn’t have the same mental states that you do’ because ‘consciousness is a unique source of reasons for belief and action’ (Smithies 2019, 21–22). Hence Smithies says that we can’t ‘explain the zombie’s behaviour in a way that shows it to be rational in light of the zombie’s own reasons for belief or action […] because the zombie has no conscious experience’ (Smithies 2019, 21). On this view, a perceptual representation has to be phenomenally conscious in order to play any epistemic role, which is why Smithies doesn’t offer any positive account of the epistemic import of unconscious perception.

Insofar as phenomenalist accounts leave the epistemic significance of unconscious perception unexplained, they are vulnerable to the objection that unconscious perception provides perceptual justification (see e.g. Baergen 1992; Berger 2014; Berger, Nanay, and Quilty-Dunn 2018). The objection is grounded on evidence from perceptual psychology. According to Baergen (Baergen 1992, 107, 111), for example, because the formation of perceptual judgments can be influenced by numerous unconscious factors (e.g. blindsight, subliminal priming, cognitive penetration), confining the evidential basis of such judgments to conscious experience is mistaken.

One might respond that this reasoning presupposes epistemic externalism, which makes it unlikely to persuade the epistemic internalist. Indeed, externalists and internalists attach quite different meanings to ‘epistemic justification’. The opponent of unconscious perceptual justification emphasizes the difference between (a) evaluating the rational stance of a subject by simulating their first-person perspective, and (b) determining whether a subject's behaviour is rationally explicable from the third-person perspective (cf. Ginsborg 2006). If the subject is unable to consciously maintain coherence between their perceptual states and beliefs, the first-personal evaluation of their beliefs and/or behaviour in terms of epistemic justification is impossible (Hellie 2011, 2014a, 2014b). In contrast, the advocates of unconscious perceptual justification maintain that third-personal rational explicability of the subject's beliefs and/or behaviour is sufficient for epistemic assessment (Berger 2014; Berger, Nanay, and Quilty-Dunn 2018).

Of course, everyone is free to use the concept of justification in whatever sense they want. However, the proponents of unconscious perceptual justification argue that (i) empirical evidence for unconscious perception supports externalism about perceptual justification (Baergen 1992, 117), and that (ii) interpreting that evidence through the lens of internalism would amount to begging the question against their view (Berger, Nanay, and Quilty-Dunn 2018, 5 footnote 6, 16). This merely illustrates how.

the parties to at least the most radical of the disputes about epistemic justification are using ‘justified’ to pick out different properties of beliefs, different epistemic desiderata or collections thereof. Instead of having persistent disagreements about a common target, they are arguing past each other. (Alston 2005, 26)

Following Alston's advice, we should draw a sharp distinction between warrant and justification. The term ‘warrant’ stands for justification as understood by epistemic externalists. Warrant confers positive epistemic status on one's belief that p irrespective of whether one can come to know that the belief is warranted. For example, it can be realized by the fulfilment of a reliable belief-forming procedure. Because reliable belief-forming procedures are not always truth-conducive, some externalists think that warrant can be overridden by defeaters (see e.g. Alston 1988). Others understand warrant as a placeholder for whatever it is that transforms true belief into knowledge (see e.g. Plantinga 1993).

The present distinction restricts the meaning of ‘justification’ to the way epistemic internalists use this term. Thus understood, justification confers positive epistemic status on belief, albeit not without conscious and deliberate involvement on the part of the subject. One's belief is justified if it was formed without violating any epistemic norms and obligations, to which one is able to adhere. Whether a belief is justified turns on its being formed via rational deliberation, or at least on its being defendable in this way.¹³ Whether justification is defeasible or indefeasible turns on whether justifying reasons are factive (whether they entail the truth of the propositions they support).

Warrant and justification are distinct yet complementary tools of epistemic evaluation. The phenomenalist can use them to meet the desiderata mentioned at the outset of this section. They can argue that both conscious and unconscious perception suffice for epistemic warrant, whereas only P-conscious perception can deliver justification. However, as I argue in the next section, this asymmetry is unmotivated if intentionalism is true.

3.2. Two dilemmas for phenomenalism

Berger and colleagues (Berger, Nanay, and Quilty-Dunn 2018, 576–577) present a dilemma for the phenomenalist. On the first horn, no functionalist theory of consciousness is true, which makes it difficult to explain why should P-consciousness play any special epistemic role. Because the remaining theories attempt to explain consciousness in a non-causal fashion, they make it arbitrary to ascribe any special epistemic import to consciousness. The phenomenalist might postulate a primitive link between P-consciousness and some epistemic status, but this is not going to persuade the competition (see Section 3.1). On the second horn, some functionalist theory of consciousness is true, which renders phenomenalism implausible because no such theory supports it. Since functionalist theories characterize consciousness in causal terms, they focus on A-consciousness. No important role is attributed specifically to P-consciousness. I am sceptical about the possibility of resolving this dilemma in keeping with intentionalist phenomenalism, no matter which horn is embraced.

Of course, epistemic import does not have to be functional or causal. Consciousness may have epistemic value even if it is a by-product of other psychological processes (cf. Rosenthal 2008, 837). Indeed, the account I put forward in Section 4.2 explains the epistemic role of consciousness primarily in terms of its relation to truth and understanding, and only secondarily in terms of its function and its causal relations. But the problem with intentionalist phenomenalism is not that the explanation it offers is non-functional or non-causal. The problem is that intentionalist phenomenalism is undermined by transparency and, as we shall see below, also by the possibility of unconscious perception.

As we have seen in Section 2.2, PF fails to explain what is epistemically special about P-consciousness if perceptual experience is transparent. Second, Berger and colleagues are right that intentionalist phenomenalism is not supported by functionalist theories of consciousness. On top of that, the phenomenalist faces yet another dilemma. While the only way to overcome the challenge set forth by the proponents of unconscious perceptual justification is to sharpen the epistemic contrast between P-conscious and P-unconscious perception, there is good reason to think that any such attempt will either (i) backfire on the view that P-consciousness has unique epistemic significance, or (ii) underestimate the epistemic import of unconscious perception.

The former problem occurs when one tries to rise the bar for propositional justification so as to render it conditional on P-consciousness. For example, one might insist that propositional justification is present only if some corresponding doxastic justification is obtainable. On this view, a mental state provides propositional justification if and only if its content is available for being used in a conscious inference to the conclusion that some belief is true. This marks a substantial epistemic difference between conscious and unconscious perception, as it entails that only consciously available contents can serve as propositional justification. However, doxastic justification requires A-consciousness. So if propositional justification required the availability of doxastic justification, it would require A-consciousness too. This, in turn, undermines S. Even if P-consciousness has PF, the latter does not suffice for propositional justification. N could still be true, but only if transparency were false (see Section 2.2).

The second problem arises when one tries to draw an epistemic contrast between P-conscious A-unconscious perception and unconscious perception. The possibility of P-conscious A-unconscious perception is postulated by the proponents of phenomenal overflow hypothesis. According to it, perceptual phenomenology is so rich that one cannot become A-conscious of all of it at once, which means that one can be P-conscious of what one is not A-conscious (see e.g. Block 2011). The problem at hand ensues from the fact that the plausibility of phenomenal overflow is inversely proportional to the plausibility of unconscious perception hypothesis. According to the latter, ‘episodes of the same fundamental kind as episodes of conscious perception can occur unconsciously’ (Block and Phillips 2017, 165).

Both hypotheses are supported by very similar empirical evidence. In both cases, the subjects are presented with some visual stimuli for a very short amount of time, and then asked to solve certain stimuli-related tasks. In both cases, the observed performance suggests that the subjects have seen more than they are able to report. The main difference is that the stimulus presentation time is longer in overflow studies. In effect, the subjects report consciously seeing the stimulus for a brief amount of time, whereas in unconscious perception studies no conscious experience of it is reported (subjects who report seeing it are typically excluded from the study).

The overflow theorists often emphasize that the subjects report having an impression that they have seen all of the details of the stimulus, even though they cannot report those details. According to overflow hypothesis, this is because they are P-conscious but not A-conscious of what is presented to them. However, the evidence for that impression is questionable (Cova, Gaillard, and Kammerer 2021), and even if the impression is real, it might just be an illusion caused by the fact that one's working memory is constantly updated by shifts of attention (see e.g. Kouider et al. 2010; Schlicht 2012). Consequently, the performance in overflow studies may in fact be due to unconscious perception (Phillips 2018a). On the other hand, it is also possible that performance in unconscious perception experiments is due to residual P-consciousness of the stimuli (Phillips 2018b), which is basically the P-conscious A-unconscious perception that is hypothesized to occur in overflow. In short, it is a genuine possibility that (A) the putative cases of overflow are in fact cases of unconscious perception, and it is also a genuine possibility that (B) the putative cases of unconscious perception are in fact cases of overflow. However, A and B are incompatible. While A validates unconscious perception at the cost of overflow, B controverts unconscious perception by vindicating overflow. Therefore, the plausibility of unconscious perception is inversely proportional to the plausibility of overflow.

Now, I am not suggesting that overflow and unconscious perception are mutually exclusive. They are not. What I want to emphasize instead is how small the difference is between P-conscious A-unconscious perception and unconscious perception. Since their respective cognitive effects are so similar that it cannot be ruled out that what we are dealing with is a single phenomenon rather than two distinct ones, positing any considerable epistemic difference between them is unwarranted. Insofar as there is no clear epistemic difference between P-conscious A-unconscious perception and unconscious perception, it is unclear why should P-consciousness, in and of itself, make any epistemic contribution.

The foregoing leads to the conclusion that unique epistemic significance of P-consciousness is tenable only on the assumption that all putative cases of unconscious perception are in fact cases of overflow (i.e. that unconscious perception is impossible). For only then the epistemic import of P-consciousness cannot be intercepted by unconscious perception. If all putative cases of overflow are in fact cases of unconscious perception, there is no such thing as P-conscious A-unconscious perception, in which case P-consciousness lacks any unique epistemic significance. P-consciousness is epistemically impotent also if both overflow and unconscious perception are possible, at least insofar as there is no clear epistemic difference between them.

The similarity between unconscious perception and P-conscious A-unconscious perception amplifies the challenge set forth by Berger and colleagues, which only deepens the worry that PF does not pick anything epistemically significant.

If having epistemic import is as appropriate a criterion of perceptuality as guiding action, the phenomenalist could turn the table on Berger and colleagues by insisting that the putative cases of unconscious perception are not really genuine instances of perception because they are insufficient for perceptual justification. Since the status of unconscious perception hypothesis is currently an open question, scepticism about it is a perfectly viable position. Nevertheless, given the large amount of empirical evidence supporting the hypothesis, denying it on purely theoretical grounds would be highly controversial.

What the phenomenalist needs instead is an account that consistently combines the following claims:

1. the unique epistemic significance of P-consciousness;

2. the epistemic significance of unconscious perception;

3. the epistemic superiority of 1 over 2.

As we have seen, however, 1 and 3 are in tension with 2. In particular, the similarity between P-conscious A-unconscious perception and unconscious perception is going to hinder any attempt at reconciling 2 with 3. This suggests that 1 is tenable only if 2 is rejected (i.e. only if unconscious perception is impossible), exactly as predicted by Berger and colleagues. The same can be said about N, since N is just a specific formulation of 1. Given all this, intentionalist phenomenalism is doomed to fail.

3.3. A diagnosis

Intentionalism entails that the phenomenal character of perceptual experience is consistent with inexistence of what it represents. Whether an experience is veridical or not turns on its aetiology, not phenomenology. Whether one should take one's experience at face value or reject it as misleading turns on one's beliefs about its cause, not on what it is like for one to have it. This leaves perceptual phenomenology with no epistemic role to play, which is precisely why PF is ill-suited to explain the epistemic superiority of P-conscious A-unconscious perception over unconscious perception. If the phenomenal character of perceptual experience is transparent and consistent with inexistence of what it represents, so is PF, which means that there is nothing epistemically significant about the latter.

4. Relationalist phenomenalism

The failure of intentionalist phenomenalism does not entail that P-consciousness lacks unique epistemic significance. For it is far from obvious that the phenomenal character of perceptual experience is consistent with inexistence of the experienced object, and as soon as that core assumption of intentionalism is rejected, a possibility opens that Berger and colleagues have overlooked. Section 4.1 shows that the assumption in question lacks sufficient motivation. Section 4.2 presents an alternative account of the epistemic role of P-consciousness that becomes available when the assumption is rejected.

4.1. Imagining objectless experiences and philosophical zombies

Many philosophers believe that the intentionalist assumption I have just blamed for rendering P-consciousness epistemically impotent is indispensable. In this Section, I explain why that assumption is at best optional.

Recall Dorsch’ response to Martin's transparency argument I have mentioned in Section 2.2. Martin (Martin 2002, 417) claims that imagining a P-hallucination consists in imagining a perception from which that P-hallucination is indistinguishable, and then making a non-imagistic supposition that the imagined experience is not a perception. Dorsch (Dorsch 2010, 195–196) disagrees. If relationalism is true, the phenomenal character of the imagined experience is inconsistent with inexistence of the experienced object, and the added supposition cannot change that. In effect, relationalism makes it impossible to imagine a P-hallucination. For Dorsch, this is an untoward result because ‘we seem […] to be able to experientially imagine having such a hallucination’ (Dorsch 2010, 196). He concludes that his experiential intentionalism is preferable over relationalism.

But it is far from obvious that one can imagine what it would be like to have an objectless experience. On the contrary, there are good reasons to believe that the phenomenal character of P-hallucination is actually unimaginable (Ali 2018; Raleigh 2014). And the impossibility to imagine a P-hallucination is only to be expected if the idea of such an experience is incoherent and merely prima facie conceivable.

This brings us back, perhaps a bit unexpectedly, to the issue of epistemic zombies. For the reason why (i) the alleged conceivability of a P-hallucination does not undermine the relationalist conception of phenomenal character is precisely the same as the reason why (ii) the presumptive conceivability of epistemic zombies does not undermine unique epistemic significance of P-consciousness. The reason is that arguments from conceivability are by nature question-begging. As Brown (Brown 2010) has shown, for any such argument a reverse argument can be formulated that leads to the opposite conclusion.

For example, some philosophers (see e.g. Chalmers 1996) argue that the conceivability of a phenomenal zombie (i.e. a P-unconscious physical duplicate of a P-conscious human being) undermines physicalism (i.e. the view that everything is physical). To see why this is a non sequitur, consider zoombie, i.e. a nonphysical duplicate of a P-conscious human being that lacks any nonphysical P-consciousness. Since there is nothing obviously incoherent in this description, zoombie seems no less conceivable than phenomenal zombie. But if the conceivability of a phenomenal zombie undermines physicalism, the conceivability of zoombie undermines dualism (i.e. the view that the phenomenal is nonphysical). The dualist is going to complain that the zoombie argument begs the question against dualism, in that it presupposes that nonphysical properties are insufficient for P-consciousness. While this is correct, exactly the same complaint can be raised against the phenomenal zombie argument because the latter presupposes that physical properties are insufficient for P-consciousness.

Brown (Brown 2010, 68) concludes that both zombies and zoombies are merely prima facie conceivable. In order to establish that zombies are ideally conceivable¹⁴ (i.e. conceivable in a strong sense that renders zoombies ideally inconceivable and thereby makes the phenomenal zombie argument go through), one would have to first show that the phenomenal does not reduce to the physical. Therefore, the zombie argument presupposes its own conclusion.

The same diagnosis applies to the conceivability of epistemic zombies. To show that they are more than prima facie conceivable, one has to first establish on some independent grounds that P-consciousness lacks unique epistemic significance.

According to Lee (Lee 2014), simply assuming that P-consciousness is reducible to the physical suffices to render epistemic zombies ideally conceivable. If reductive materialism in the philosophy of mind is true, P-consciousness lacks unique epistemic significance. The main premise in Lee's argument is the conceivability of a functional zombie such that (i) it is not P-conscious because its cognitive architecture is different from that which gives rise to P-consciousness, but (ii) its epistemic standing is identical to that of a P-conscious being because it has internal states that play the relevant functional role. The conceivability of this zombie follows from the claim that P-consciousness lacks special natural significance (i.e. the distinction between the P-conscious and the P-unconscious does not carve nature at its joints). This, in turn, is a consequence of reductive materialism (Lee 2014, 224–225).

But the zombie argument backfires on Lee's view. Consider an android whose conscious experience is exactly like that of a conscious human being, yet its cognitive architecture is nothing like that of a conscious human being. If the conceivability of Lee's functional zombie did pose a genuine problem for the view that P-consciousness has a special epistemic role to play, then the conceivability of our P-conscious android would pose a genuine problem for the view that there is nothing epistemically special about P-consciousness. This demonstrates that Lee's argument has inherited the question-begging nature of the original zombie argument.

Now, conceiving of a P-hallucination is no different from conceiving of a philosophical zombie. P-hallucination is prima facie conceivable, but for all we know, it may not be any more conceivable than that.¹⁵ Thus Martin's take on imagining such experiences may well be correct.

This is relevant to epistemology, as the conceivability of P-hallucination is the cornerstone of some radical sceptical scenarios (e.g. Evil Demon, Brain in a Vat). Given that P-hallucination might be merely prima facie conceivable, it is questionable whether (and if so, to what extent) such scenarios should impinge on our theorizing about perceptual knowledge.

For example, Smithies (Smithies 2019, 99–100) argues that justification delivered by a genuine perception cannot be superior to that ensuing from a corresponding P-hallucination:

1. A genuine perceiver S and their P-hallucinating counterpart S* have experiences with qualitatively identical phenomenal characters.

2. Rationality requires conforming one's beliefs to one's perceptual experience.

3. Drawing a justificatory contrast between genuine perception and P-hallucination entails that S* is less rational than S just because of being a victim of a radical sceptical scenario, which is not true.

If the way things phenomenally appear to S* is merely prima facie conceivable, premise 1 is questionable. Since it is unclear what the subjective perspective of S* is like, it is also unclear what epistemic import does it have, if any. For this reason, Smithies’ argument is insufficient to undermine the view that genuinely perceptual reasons are factive.

4.2. Perceiving truthmakers, consciously and unconsciously

P-consciousness has unique epistemic import if the phenomenal character of conscious perception is inconsistent with inexistence of the perceived object. Consequently, whatever reason one might have for thinking that P-consciousness is uniquely epistemically significant, it is also a reason to replace intentionalism with relationalism. For the latter entails that.

[s]ome of the objects of perception – the concrete individuals, their properties, the events these partake in – are constituents of the experience. No experience like this, no experience of fundamentally the same kind, could have occurred had no appropriate candidate for awareness existed (Martin 2004, 39, emphasis added).

To illustrate, suppose that you hallucinate a river ahead on the road. Apart from the river, everything you visually experience is really there, which means that your (total) experience is an hallucination only partially, and you can still learn a lot from it (e.g. that it is sunny).¹⁶

Under intentionalism, your experience is such that you could undergo exactly the same experience even if there were no road or sun around, and even if there actually were a river ahead of you. Whether (and if so, to what extent) a given experience is genuinely perceptual is in no way reflected in its phenomenal character. Whether you should consider an experience as a genuine perception depends on your beliefs about the reliability of your senses, not on what it’s like for you to have that experience. You might believe that the river is hallucinatory and the rest of what you visually experience is real, but the phenomenal character of your current experience is, in and of itself, neutral about this. It doesn’t give you any justification for that belief. It has to be supplemented with some beliefs about, say, how probable it is that there is a river in front of you, or how likely it is that you are hallucinating, and so on.

Things are different under relationalism, which construes perception and hallucination as completely different. Since your experience of the road and the sun is genuinely perceptual, you wouldn’t have the same experience had the road and the sun not been around. The phenomenal character of your experience of the road would have not been the same had the road and the sun not been there in front of you. This is why your experience of the road and the sun buys you indefeasible warrant for the belief that it is sunny.

According to relationalism, conscious perception is direct (i.e. objectually and psychologically unmediated¹⁷) awareness of mind-independent objects. This means that a mental state M of a subject S is a perception of an object O in virtue of the fact that O is a specific proper part of M, in the sense that some mind-independent properties of O participate in shaping the phenomenal character of M.

Epistemic import of thus understood perceptual experience consists in the subject's being directly presented with truthmakers for potential perceptual judgments. According to Brewer, ‘acquaintance in perception provides the evident ground for concept application in judgment’ (Brewer 2011, 144). Kalderon (Kalderon 2011, 225–228) takes a similar approach. The object of perception, a mind-independent particular, constitutes a truthmaker for a number of propositions. Necessarily, if it exists, they are true. In virtue of this alethic connection, the object of perception can itself serve as a reason for holding a number of perceptual beliefs. Conscious perception makes this reason accessible to the subject. By making the subject aware of the object, it provides the subject with an opportunity to acquire knowledge.¹⁸ This proposal can be summarized as the claim that the object of conscious perception is an objectual reason that warrants perceptual belief (cf. section 3.1).

Since this view does not appeal to PF, it accommodates the unique epistemic import of P-consciousness without violating transparency. What about unconscious perception? Relationalism was originally introduced as a theory of conscious perception, and relationalist accounts of perceptual reasons follow suit. Nonetheless, I believe that the relationalist analysis can and should be extended to unconscious perception. If the objects of perception (i.e. objectual perceptual reasons) are mind-independent, they are consciousness-independent too. Therefore, unconsciously perceived object can be regarded as an objectual reason that warrants unconscious perceptual belief, i.e. the belief that causes the kind of behaviour that prompted empirical researchers to hypothesize that perception can occur unconsciously.

Two doubts immediately arise. First, applying the relationalist analysis to unconscious perception may seem incoherent. If conscious perception is a ‘modification of consciousness’ by conscious acquaintance with a mind-independent object (Brewer 2011, 92), how could such acquaintance be sometimes conscious and sometimes not? How could unconscious perception involve acquaintance with a mind-independent object, if that object makes ‘no contribution to the subject's conscious perspective on the world’ (Phillips 2018b, 472)?

My response is twofold. First, notice that the perceived object is not the sole constituent of perceptual relation. The latter is also constituted by the subject, and determined by the circumstances of perception (see e.g. French 2018). When the latter two conditions are suboptimal, it is expectable that the perceived object's constitutive contribution to perceptual relation will be suboptimal as well. For example, when the subject has lesions in their visual cortex, or a stimulus is presented to them for a fraction of a second, they are unlikely to perceive the stimulus consciously. If, in spite of such obstacles, the resulting mental state is similar enough to ordinary conscious seeing (i.e. if it can guide action and provide epistemic support for belief), it is reasonable to regard it as unconscious acquaintance. On this view, the object of unconscious perception is an objectual reason that warrants unconscious perceptual recognition. Unconscious acquaintance is an opportunity to acquire unconscious knowledge in virtue of the possession of such unconscious perceptual warrant.

Second, the possibility of unconscious perceptual relation is a natural consequence of the fact that relationalism construes the phenomenal character of perceptual experience as at least partially constituted by the mind-independent object. If phenomenal qualities (i.e. the qualities that determine what it is like to have a perceptual experience) are for the most part¹⁹ mind-independent, they are also consciousness-independent. They are phenomenal not in the sense of being inherently conscious, but in the sense of determining what it is like to be conscious of them when they are perceived consciously. This is in line not only with the transparency of experience, but also with Rosenthal's observation that the existence of mental qualities does not depend on their being consciously experienced (Rosenthal 2010).

If this is correct, perceptual P-consciousness can be construed as a specific kind of A-consciousness, namely A-consciousness of phenomenal qualities. On this view, P-conscious perception is A-conscious perception of phenomenal qualities, whereas unconscious perception is A-unconscious perception of phenomenal qualities. Simply put, both conscious and unconscious perception have a phenomenal character, but only in the former case the phenomenal character is conscious (cf. Marvan and Polák 2017).

An anonymous referee has raised a worry that construing P-consciousness as A-consciousness of phenomenal qualities might amount to redefining P-consciousness as being exactly what the proponents of the A-consciounsess vs. P-consciousness distinction claim it is not. As I see it, however, my proposal does not deviate from the canonical definitions of P-consciousness and A-consciousness introduced at the outset of this paper. P-consciousness and A-consciousness are two distinctive aspects of consciousness, and the purpose of a theory of consciousness is to explain both these aspects and the relation between them. There are many different attempts at explaining them in the literature, and the claim that perceptual P-consciousness is A-consciousness of phenomenal qualities is neither more nor less than just another such attempt. It is not a redefinition of P-consciousness, but an explanation of what P-consciousness consists in, i.e. what constitutes the what-it-is-like-ness of conscious perceptual experience. It is an explanans, not a redefinition of an explanandum.

The claim that perceptual P-consciousness is A-consciousness of phenomenal qualities is particularly plausible on the assumption that the phenomenal character of perceptual experience is entirely constituted by the perceived items. The latter claim, known as ‘diaphaneity’, has been recently defended by Zięba (Zięba 2021). To illustrate how the combination of these two claims might work, suppose again that you see a road. There is something it is like for you to consciously perceive the road because your conscious perceptual experience of the road consists in being A-conscious of the visible qualities that the road has independently of being perceived. The visible qualities of the road aren’t inherently conscious. But when you A-consciously perceive them, they make it like something for you to consciously experience them, and this is what makes you P-conscious of them.

To my best knowledge, the notion of P-consciousness doesn’t presuppose that what-it-is-like-ness of conscious perceptual experience is necessarily constituted by something in the nervous system. Neither does it rule out that perceptual P-consciousness is enabled by A-consciousness.²⁰ If so, locating the constituents of perceptual what-it-is-like-ness out there in the environment does not amount to redefining P-consciousness. What is unorthodox about my proposal is not how it defines P-consciousness and A-consciousness, but how it explains the nature of perceptual P-consciousness and its relation to A-consciousness. No doubt this account is unorthodox and controversial, but I don’t see anything in the canonical definitions of P-consciousness and A-consciousness that would rule it out.

What motivates my proposal is that it provides the foundation for a version of phenomenalism that is not undermined by transparency and the possibility of unconscious perception. The claim that both conscious and unconscious perception have a phenomenal character explains what they have epistemically in common, and why unconscious perception is so similar to overflow. The present proposal is also supported by considerations motivating diaphaneity (Zięba 2021) and unconscious phenomenal character (Marvan and Polák 2017; Zięba 2022). Of course, much more would have to be said to render this account of P-consciousness preferable over those already present in the literature, but that is a task for another paper. The purpose of introducing this view here is only to indicate a so far overlooked alternative to intentionalist phenomenalism. Since this alternative is neither unintelligible, nor obviously false, nor unmotivated, it deserves to be taken seriously, and its availability shows that rejecting intentionalist phenomenalism is not equivalent to rejecting phenomenalism in general.

Let us move on to the second worry. To guide action, unconscious perception must be followed by recognition and belief formation. The object of unconscious perception, i.e. unconscious perceptual reason, warrants unconscious beliefs that result from its recognition, thereby enabling purely externalist and non-reflective unconscious perceptual knowledge (cf. Berger 2020; Mandelbaum 2016; Rosenthal 2008). When the same object is perceived P-consciously, it can be consciously recognized. If it is, it warrants the resulting conscious perceptual belief, thereby enabling the acquisition of conscious perceptual knowledge. But if both conscious and unconscious perception deliver the same type of perceptual reason, it seems that consciousness makes no epistemic difference.

In response, I argue that even though the same type of reason is operative in conscious and unconscious perception, its epistemic import is stronger in the conscious case. My argument rests on the knowledge-first idea that knowledge justifies beliefs, not vice versa (Williamson 2000). In particular, I assume that seeing that p (which is a specific way of knowing that p) justifies believing that p (see e.g. Millar 2010, 139).

Only consciously known truths can be justifiers. When the subject consciously perceives an object O, and consciously recognizes it as an instance of F, they are in a position to cite their seeing that O is F (i.e. their knowledge that O is F) as a reason for their belief that O is F, and thereby to justify their belief that O is F. Since the acquisition of conscious perceptual knowledge that O is F is enabled by conscious perception of O, the latter effectively expands one's inventory of justifiers (i.e. propositional reasons). By contrast, unconsciously known truths cannot be justifiers. After all, unconscious mental states are not reportable. When the subject unconsciously perceives an object O, and unconsciously recognizes that O is F, they are not in a position to cite their seeing that O is F as a reason for their belief that O is F.

This already indicates an epistemically relevant contrast between conscious and unconscious perception. Crucially, however, the difference in question does not come down to reporting seeing and citing it as a reason for beliefs. Justification is a manifestation of understanding; the more one understands, the more beliefs one can justify. This indicates that conscious perception not only warrants perceptual beliefs, but also enables one to reflect on how the contents of those beliefs are related to one's background knowledge. In other words, the unique epistemic import of P-consciousness consists in creating an opportunity for the subject to increase their understanding of some subject matter. Justification is merely a reflection of that.

The epistemic asymmetry between conscious and unconscious perception is explained by the claim that the phenomenal character of unconscious perception is A-unconscious. The explanation assumes that A-consciousness (‘consciousness’ as understood by functionalism, i.e. the availability of mental content for reasoning and rationally guiding speech and action), is necessary for understanding, i.e. ‘grasping of explanatory and other coherence-making relationships in a large and comprehensive body of information’ (Kvanvig 2003, 192). If this is correct, perceptual P-consciousness qua A-consciousness of phenomenal qualities is uniquely epistemically significant because it enables expanding such bodies of information by filling them with new truths that one comes to know via P-conscious perception.²¹

One advantage of this proposal is that it does not fall into any of the two dogmas of empirical justification identified by Lyons (Lyons 2020). The first dogma is that perceptual experiences justify beliefs, the second is that perceptual beliefs are justified by perceptual experiences in virtue of being based on the latter.

Lyons argues that both dogmas are unwarranted and probably false. The problem with the first dogma can be represented as a dilemma: either (i) perceptual experience has no conceptual content, in which case it is ill-suited to justify beliefs, or (ii) it has some conceptual features, which makes it mysterious what experience is supposed to be. The second dogma is controverted by empirical evidence. First, studies on unconscious perception suggest that the transfer of information from perception to belief does not require consciousness. Second, empirical results concerning the time course of perception indicates that perceptual judgment precedes perceptual experience. If so, the former cannot be based on the latter. Third, studies on perception of abstract categories suggest that the perceiver gains cognitive access to generic features of the perceived scene before they get access to its specific features (e.g. one is able to tell that what one sees is an animal before one is able to tell what kind of animal it is). If beliefs are based on experiences, however, we should expect the opposite result.

As a knowledge-first account, my proposal holds that perceptual belief is justified by the totality of what one knows, not by perceptual experience. Neither does my account entail that perceptual belief is based on perceptual experience, as it recognizes the possibility of unconscious belief and unconscious perceptual knowledge. Since my proposal is immune to Lyons’ criticism, (i) the latter effectively supports the former by making it preferable over the competing views that follow the dogmas; (ii) insofar as my account is a viable option, rejecting the dogmas does not suffice to show that perceptual consciousness plays no epistemic role.

Another advantage of my view is that it maintains the unique epistemic significance of P-consciousness even on the assumption that propositional justification renders doxastic justification obtainable (i.e. that propositional justification requires A-consciousness). This is because relationalism understands perceptual P-consciousness as objectually and psychologically unmediated awareness of truthmakers for propositions. As a result, the epistemic import of conscious perception is both factive and fully reflectively accessible to the subject. Because the factivity is specifically due to P-consciousness qua A-consciousness of phenomenal qualities, this epistemic import cannot be reduced to those of its features that are due to A-consciousness simpliciter.

For the same reason, my proposal cannot be replicated within the intentionalist framework. The latter construes the phenomenal character of perceptual experience as consistent with inexistence of what it represents. On that view, phenomenal qualities are produced in the subject, and can occur even if nothing in the environment corresponds to them. The presence or absence of justification is a matter of intentional content, not phenomenal character. Consequently, the epistemic role of P-consciousness qua A-consciousness of phenomenal qualities is realized by those of its features that make it an instance of A-consciousness simpliciter, whereas those of its features that are characteristic of P-consciousness simpliciter have no special epistemic role to play. Therefore, any intentionalist adaptation of my proposal would undermine the epistemic significance of P-consciousness.

To summarise, relationalist phenomenalism construes perceptual P-consciousness as A-conscious perception of phenomenal qualities. These qualities are mind-independent perceptible features of the environment. They’re phenomenal not in the sense of being inherently conscious, but in the sense that they determine what it is like to perceive them when they’re A-consciously perceived. On this view, perceptual phenomenal character is (i) transparent, (ii) incompatible with inexistence of the perceived items, and (iii) consciousness-independent. Given (i), perceptual experience doesn’t have PF. Given (ii), perception provides indefeasible warrant for perceptual recognition despite being transparent. Given (iii), perception provides this warrant even if it is unconscious. When a perception is conscious, however, the warrant it provides is conscious as well. Having a conscious perceptual warrant puts one in a position to consciously recognize, and thereby to know, that O is F. Consciously knowing that O is F puts one in a position to understand more, and thereby to be able to justify more of one’s beliefs. This is how relationalist phenomenalism explains the epistemic import of P-consciousness in a way that is compatible with both transparency and the possibility of unconscious perception.

5. Conclusions

I have argued that intentionalism about perception cannot account for unique epistemic significance of P-consciousness without denying the transparency of experience and the possibility of unconscious perception. For if intentionalism is combined with the latter two claims, the epistemic import of P-consciousness is at best derivative or negligible. This leaves us with a dilemma: either perception is intentional and P-consciousness is epistemically insignificant, or perception is not intentional and P-consciousness can play some special epistemic role. Consequently, whatever reason might there be to believe that P-consciousness has unique epistemic import, it is also a reason to prefer relationalism over intentionalism.

The intentionalist who denies that P-consciousness has unique epistemic significance will most likely welcome this conclusion as a corroboration of their view. However, denying that P-consciousness is epistemically significant undermines one of the original reasons for attributing intentional content to perceptual experience, namely the need to explain how the latter can justify beliefs.

Being in a perceptual state involves having a perceptual experience if the former has a conscious phenomenal character. Therefore, to say that perceptual experiences justify beliefs amounts to saying that perceptual states justify beliefs in virtue of their conscious phenomenal characters. But if P-consciousness is epistemically impotent, so is having a conscious phenomenal character. It follows that perceptual states do not justify beliefs in virtue of their phenomenal characters, i.e. that perceptual experiences do not justify beliefs. Although perceptual states that happen to have a conscious phenomenal character may still be regarded as sources of justification, the fact that being in such states involves having a perceptual experience has nothing to do with their epistemic import, just as the fact that some cooks have moustaches does not entail that moustache plays any role in cooking.²²

Positing perceptual content to account for the epistemic import of perceptual experience makes no sense if the latter has no epistemic significance. Somewhat ironically, what was originally introduced to explain how perceptual experiences can justify beliefs turns out to be incompatible with the claim that perceptual experiences justify beliefs.

Acknowledgments

Thanks to Błażej Skrzypulec and an anonymous referee for this journal for their comments on earlier versions of this paper.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the National Science Centre of Poland under grant number 2019/32/C/HS1/00113.

Notes

2 Intentionalists disagree about the relation between perceptual content and phenomenal character. But they agree that the phenomenal character of a perceptual experience is consistent with inexistence of what that experience represents, irrespective of whether the experience is a genuine perception, or an hallucination.

3 Propositional justification contrasts with doxastic justification. Having propositional justification amounts to having a good reason to believe that p. To be doxastically justified in believing that p, one not only has to possess the relevant reasons, but also base the belief that p on them. Simply having good reasons for believing that p is insufficient for doxastic justification because one might have excellent reasons for p, yet believe that p due to wishful thinking.

4 Blindsight occurs in subjects who are not consciously aware of visual stimuli presented in a part of their visual field because they have lesions in their primary visual cortex. It consists in the ability to respond to those stimuli in spite of the lack of awareness of them. There are two types of blindsight. In type-1 blindsight, the subject is completely unaware of stimuli in their blind field, whereas type-2 blindsight involves some residual awareness of them.

5 As an epistemic internalist, Smithies holds that the manifestation of a reliable doxastic disposition does not suffice for epistemic justification. The epistemic externalist might disagree (see Section 3.1).

6 For this reason, PF would not support the unique epistemic import of P-consciousness if it was a beneficial effect of cognitive penetration (cf. Lyons 2011). The latter occurs when higher-order mental states (e.g. beliefs, desires, emotions) modify the phenomenal character of perceptual experience. If PF was assimilated to cognitive penetration, its epistemic import would be essentially cognitive, and only derivatively experiential.

7 I set aside moderate views that combine intentionality and relationality of perception (see e.g. Schellenberg 2014). This is because such views always attempt to explain relationality in terms of intentional content, never in terms of phenomenal character. For this reason, if intentionality of perception is incompatible with unique epistemic significance of P-consciousness, the intentionalist cannot save the day by adopting a moderate view.

8 Characterizing Dorsch's view in terms of metaphysical rather than phenomenological transparency would not help because metaphysical transparency cannot be established on purely phenomenological grounds (Gow 2016).

9 The debate between epistemic internalism and epistemic externalism is further discussed in Section 3.1.

10 For further reasons to be sceptical about explaining the epistemic role of P-consciousness in terms of presentational force, see (Byrne 2016; Farkas 2014; Teng 2018).

11 There are some exceptions (see e.g. Conee and Feldman 2001).

12 Note that for Smithies having phenomenal character is equivalent to being phenomenally conscious.

13 In my view, having propositional justification entails that the justifying reason is available for being used in rational deliberation whether or not it is actually deployed for such purpose. Smithies disagrees because he thinks that having propositional justification does not require A-consciousness.

14 For more about various dimensions of conceivability, see (Chalmers 2002).

15 A slightly different argument to this conclusion can be found in (Masrour 2020).

16 I thank an anonymous referee for this counterexample.

17 According to some versions of the sense datum theory, perception is mediated objectually. On this view, a mental state M of a subject S is a perception of an object O in virtue of the fact that S perceives (or directly apprehends) a sense datum D, where D is understood as something different from O. According to intentionalism and traditional direct realism, perception is unmediated objectually, yet mediated psychologically. On this view, a mental state M of a subject S is a perception of an object O in virtue of the fact that O features as an element in a specific causal chain that resulted in M's coming into existence (cf. Foster 2000, 4–14; Millar 2007, 182–183).

18 The intentionalist might object that the role of perceptual experience in rationalizing beliefs cannot be explained without an appeal to perceptual representation, but this argument is not persuasive (see e.g. French 2020; Travis 2013).

19 Depending on what account they give of cases such as cognitive penetration and blurry vision, the relationalist considers either all phenomenal qualities or just some of them as mind-independent.

20 If the proponent of my proposal wanted to accommodate the possibility of phenomenal overflow, they would have to associate three different types of brain activity with A-conscious P-conscious perception (type-A activity); A-unconscious P-conscious perception (type-B activity), and completely unconscious perception (type-C activity), respectively. On the resulting view, overflow occurs when type-A activity is not instantiated, but type-B activity is instantiated. One could consider this type-B activity as a realizer of ‘limited-A-consciousness’, since the activity in question would enable the cognitive and behavioural effects attributed to A-unconscious P-conscious perception, be insufficient for the effects attributed to A-conscious P-conscious perception, and enable more than what is attributed to completely unconscious perception. In short, the idea would be that perceptual P-consciousness is a limited form of A-consciousness of phenomenal qualities. Of course, a question would then arise why shouldn’t unconscious perception be considered as an even-more-limited-A-consciousness of phenomenal qualities. Again, this merely illustrates the tension between unconscious perception and overflow.

21 One might object that grasping is completely independent of one’s epistemic state because one can grasp a proposition without any change in one’s beliefs, and it is possible to grasp propositions that are false or unjustified (Bourget 2017a, 292). This might be taken as suggesting that understanding is not factive. However, the notion of understanding that is operative in my proposal is objective understanding. As Grimm (Grimm 2011, 91) puts it, 'subjective understanding [is] the kind of understanding one achieves by grasping a representation of the world (a model, perhaps, or an explanatory story of some kind) that fits or coheres with one’s ‘world picture.’ On the other hand, […] objective understanding [is] the kind of understanding that comes not just from grasping a representation of the world that fits with one’s world picture, but also from grasping a (more or less) correct representation of the world. Objective understanding therefore entails subjective understanding but goes beyond it, requiring that the grasped representation in fact obtains'.

22 In this connection, it is mystifying why would someone 'skeptical that consciousness plays any role at all in justification' (Berger, Nanay, and Quilty-Dunn 2018, 573, footnote 6) declare that 'perceptual experiences justify beliefs' (Berger, Nanay, and Quilty-Dunn 2018, 569).