Does Property-Perception Entail the Content View?

Abstract

Visual perception is widely taken to present properties such as redness, roundness, and so on. This in turn might be thought to give rise to accuracy conditions for experience, and so content, regardless of which metaphysical view of perception one endorses. An influential version of this argument —Susanna Siegel’s ‘Argument from Appearing’ — aims to establish the existence of content as common ground between representational and relational views of perception. This goes against proponents of ‘austere’ relationalism who deny that content plays a substantive role in philosophical explanations of conscious perceptual experience. Though Siegel’s argument purports to be neutral with respect to the metaphysics of perception, it relies upon an equivocation between the presentation of property-types and property-instances. Consequently, the argument begs the question against the austere relational view, and so fails to establish the desired conclusion. So while relationalists can and should allow that experiences have accuracy conditions, it does not follow from this that they have contents of any philosophically interesting or significant kind.

1 Introduction

According to representational views of experience (rv), to perceptually experience an object is to represent it as being some specific way. For example, representing a visible surface to instantiate some colour, texture or shape property, a sound to have a certain volume or timbre, or an odour to have a specific olfactory quality. Relational views of experience, such as Naïve Realism, on the other hand, characterise perceptual experience in terms of a psychologically primitive relation of ‘acquaintance’¹ or conscious awareness between a subject and external mind-independent particulars; e.g. objects, events, and/or the properties they instantiate. Moreover, proponents of ‘austere’ forms of relationalism² (ar) claim that this perceptual relation cannot be characterised solely in representational terms,³ or, more cautiously, that the notion of representation plays no fundamental role in philosophical explanations of conscious perceptual experience.⁴ According to ar, experiences do not afford awareness of particular objects, events or properties in virtue of representing anything to be the case, but rather furnish the perceiver with direct conscious awareness of them. This yields a distinctive account of perceptual phenomenal character as being at least partly constituted by the external particulars that are perceived, rather than being explicable in terms of representational content, or as internally grounded.

rv is the orthodox view among philosophers of perception, and commonly assumed rather than explicitly argued for. One kind of argument that has been advanced in its favour, however, draws upon a supposed phenomenological datum concerning the presentation of properties in visual experience. On this view, the visual presentation of properties entails the existence of a set of conditions under which a given experience is accurate, or veridical,⁵ i.e. its accuracy conditions. This is in turn claimed to entail the existence of representational content that is assessable for truth or falsity. If this is correct, then contrary to proponents of ar, who deny that perceptual experience is representational, both rv and ar are committed to the existence of representational content in visual experience; i.e. they are content views. This is perhaps surprising because many relationalists situate their view in opposition to representational content, arguing that perception only relates the perceiver to concrete particulars in the environment and not to abstract contents or states of affairs (Travis, 2013).

Perhaps the most influential proponent of this kind of argument for the content view is Susanna Siegel (2010a, p. 45; 2010b, p. 345), who claims that “once the role of properties in phenomenal character is acknowledged, the Content View is unavoidable” (ibid. 29). Indeed, Siegel’s “Argument from Appearing” (afa) has been and continues to be cited—one suspects approvingly —by representationalists,⁶ and remains one of the best known arguments for the view. However, afa is widely dismissed, often out of hand, by relationalists who maintain that ar does not entail the existence of accuracy conditions or content in any significant sense.⁷ The reasons for this denial vary and are often not spelled out in detail. Consequently, the debate between the two camps has remained stubbornly intractable, with each side defending their view against ever more sophisticated attacks. So, how should an austere relationalist respond to such arguments, and what (if any) implications does it have for ar?

In this paper, I set out precisely why afa, and by extension other arguments like it,⁸ fails to demonstrate that ar entails the content view. In particular, I argue that the relationalist need not accept Siegel’s formulation of the alleged phenomenological datum concerning the presentation of properties in experience. Far from being metaphysically neutral, as Siegel presents it, afa contains important ambiguities that render it either equivocal or question-begging. Nor, I argue, can these difficulties be easily fixed by reformulating the argument. afa does, however, help to shed light upon the precise metaphysical commitments of ar as well as the kind of content to which rv theorists are themselves committed, namely attributive content. Consequently, while relational views of experience do entail a form of accuracy conditions—a point denied by many relationalists—these are not of a kind that delivers any interesting or metaphysically significant form of content. Hence afa can be rejected as a means of establishing the content view. This in turn highlights the importance of property-perception for those on both sides of the debate, and the need to take the metaphysics of ar seriously in arguments for or against it.

2 The Argument from Appearing

Though I will mainly be concerned with its first three premises, for completeness, Siegel’s Argument from Appearing is as follows (Siegel, 2010a: 45; numbering altered):

P1:

All visual perceptual experiences present clusters of properties as being instantiated.

P2:

If an experience E presents a cluster of properties F as being instantiated, then:

Necessarily: things are the way E presents them only if property-cluster F is instantiated.

P3:

If necessarily: things are the way E presents them only if property-cluster F is instantiated, then:

E has a set of accuracy conditions C, conveyed to the subject of E, such that:

C is satisfied in a world only if there is something that has F in that world.

P4:

If E has a set of accuracy conditions C, conveyed to the subject of E, such that E is accurate only if C, then:

E has a set of accuracy conditions C*, conveyed to the subject of E, such that E is accurate iff C*.

CV:

All visual perceptual experiences have contents.

afa proceeds from the alleged phenomenological datum in P1 that visual experiences “present clusters of properties as being instantiated” (ibid.). This is followed in P2 through P4 by a series of steps that purport to show that this commits proponents of rv and ar alike to the existence of accuracy conditions that are “conveyed to the subject by her experience” (ibid. 28). Since Siegel takes the existence and conveying of accuracy conditions to be necessary and sufficient for content—a claim I dispute in Sect. 5—this entails the existence of representational content in visual experience, or what Siegel calls the Content View (cv). In order to make afa formally valid, we also need to add the following hidden premise based upon Siegel’s notion of content:

P5:

If E has a set of accuracy conditions C*, such that E is accurate iff C*, and C* is conveyed to the subject, then E has content.

Though afa might be taken to generalise to other sensory modalities including audition and olfaction, both of which are widely taken to present properties,⁹ I will follow Siegel in using ‘experience’ to mean visual perceptual experience throughout.

To understand why the existence of accuracy conditions by itself is insufficient to establish cv, we must consider another argument that Siegel offers in favour of cv, but ultimately rejects (ibid. 34; numbering altered):

A1:

All experiences are accurate or inaccurate.

A2:

If all experiences are accurate or inaccurate, then all experiences have accuracy conditions.

AC:

All experiences have accuracy conditions.

Though valid, this “Argument from Accuracy” falls short of establishing cv because ac could be true in virtue of experiences having accuracy conditions of a kind that does not entail content (ibid. 42–43).¹⁰

Consider the following line of reasoning. Accuracy is a property. Hence there exist conditions under which a given experience has that property. Specifically, an experience ψ that is accurate meets a trivial condition of the form:

• (A) ψ is accurate

So ψ has at least one accuracy condition and thus, according to the Argument from Accuracy, content. But (A) does not entail that ψ has content. If it did, then the existence of any arbitrary condition such as.

• (B) φ is blue

would similarly entail the existence of content. But an object’s being blue does not entail content in any interesting or substantive kind because that content is not conveyed to the subject by her experience in the relevant sense.¹¹ Hence, by parity of reasoning, neither does an experience meeting (A).¹² Given that, for all we have said so far (A) might be ψ’s only accuracy condition, this shows that the Argument from Accuracy falls short of demonstrating cv, since ac could be true and cv false. Siegel thus rejects the Argument from Accuracy, noting that a convincing argument for cv must explain both why experiences have accuracy conditions, and how they convey these conditions to the subject (ibid. 44).

afa, then, purports to proceed from metaphysically neutral premises that are compatible with both rv and ar to the conclusion that experiences have contents, i.e. cv. Proponents of rv are, however, already committed to cv, since their view straightforwardly entails it. Siegel’s real target is therefore those relationalists, i.e. austere relationalists, who deny cv. To establish that ar entails cv, then, afa must demonstrate on assumptions compatible with relationalism that experiences have accuracy conditions that are conveyed to the subject, and so content. I argue below that it fails to do this. Moreover, it fails for similar reasons to the Argument from Accuracy.

First, Siegel’s P1 begs the question in favour of cv due to ambiguities over the sense in which, under ar, experiences can be said to ‘present properties’. At best, on the most plausible relationalist reading of P1, only veridical experiences do this (Sect. 3). Second, the kind of accuracy conditions that are entailed by ar are not conveyed to the subject by experience (Sect. 4). And third, these are not in any case suitable as perceptual contents  —at least not in the way that content is understood by representationalists like Siegel (Sect. 5). These arguments stand independently, but together they block afa as a means of establishing that experiences have contents. Hence, I conclude that afa should be rejected by proponents of ar.

3 Presenting Properties ‘As Being Instantiated’

P1 of afa states that all visual experiences “present clusters of properties as being instantiated”. However, it is far from clear exactly what this is supposed to amount to. Indeed, the claim as stated is ambiguous in at least three respects, these relating to the meanings of ‘present’, ‘properties’, and ‘as being instantiated’, respectively. Many of the possible precisifications are straightforwardly incompatible with relationalism, and so can be rejected by the ar theorist as question-begging. For example, an internalist reading of P1 according to which the relevant properties are merely sensory qualities of experience that are answerable to an internal state of the subject conflicts with ar’s claim that such qualities are properties of the mind-independent objects of experience.¹³ Similarly, any view that explains experienced properties in linguistic terms — for example, by appeal to predication—begs the question in favour of cv (see Sect. 3.1). While such a claim may follow from Siegel’s argument, it cannot be assumed from the outset without rendering the argument circular, and so unpersuasive to Siegel’s intended targets.

In order to be maximally charitable to Siegel, my strategy will be to find an interpretation of P1 that is compatible with at least some variants of ar, and upon the basis of which afa can be evaluated. In this section I argue that this can be done, but only in the case of veridical experience. Attempting to generalise to non-veridical experience, however, renders subsequent premises of the argument false, and so can similarly be rejected by the ar theorist.

3.1 Property-Types versus Property-Instances

The first ambiguity in P1 concerns precisely what ‘properties’ experiences are supposed to present. It is relatively uncontroversial that among the things that experiences present (or represent)¹⁴ are sensible qualities such as colours, shapes, and spatial relations. Philosophers of perception may reasonably wish to remain neutral on the metaphysics of properties.¹⁵ However, the distinction between property-types, i.e. universals such as redness and roundness, and property-instances, i.e. particular instantiations of properties (or tropes) cannot be avoided. Indeed, part of what is at issue in the debate between representationalists and relationalists is whether experience can be characterised in terms of a relation to particulars or general types, with representationalists standardly endorsing some form of property-type or hybrid view, and relationalists standardly preferring property-instance or trope-based views (see below). Assuming a property-type or property-instance reading of P1, however, yields two different versions of the argument. Both interpretations are compatible with the claim that experiences ‘present properties’, and to the extent that pre-theoretical intuition can be taken to discriminate between the presentation of universals and their particular instantiations, one could make a case for either.

On the property-type reading, all visual experiences present clusters of property-types, or universals. If true, this relatively unproblematically entails the existence of accuracy conditions that are conveyed to the subject by those experiences, and so (according to P5) content. This version of afa will reassure proponents of rv, who think that experiences perceptually ‘present’ properties in virtue of representing or predicating property-types. This is, one suspects, the reading that Siegel has in mind, not least because property-instances cannot be instantiated in the standard sense (see Sect. 3.3). However, as previously noted, no further argument is required to show that rv entails cv, since it trivially follows from rv’s central claim that experiences are representational that they have content. The property-type reading, however, is problematic from a relationalist perspective for two reasons.

First, to assume from the outset that experiences present property-types comes dangerously close to claiming that experiences have what we might call predicational structure; i.e. they attribute properties to objects or the world. While rv theorists standardly endorse this claim, the notion of predication trivially entails the existence of representational content, namely what is predicated. Thus, if P1 were equivalent to the claim that experiences predicate properties of objects, this would trivially entail representational content and so beg the question in favour of cv. This cannot therefore be the correct reading of P1, and indeed Siegel’s argument makes no mention of objects, presumably for this reason.¹⁶ Either way, this reading of P1 can be rejected by the ar theorist. But this is not the only reading available, and so neither can afa be dismissed solely on this basis.

Second, the assumption that experiences present property-types is incompatible with many relationalists’ view that in perception we encounter only particulars, such as objects or events, and not generalities or abstracta, such as property-types.¹⁷ Instead, for the relationalist, generality is found only at the level of perceptual judgement or belief, and not in experience (Brewer, 2006, 2007, 2011; Campbell, 2002, pp. 114–115; Travis, 2013). On this view, experiences can at best be said to present an object or event’s particular instantiations of its properties; e.g. the tomato’s redness, or, equivalently, the redness of that very tomato, where this picks out a property-instance or trope.¹⁸ Insofar as the relationalist’s position is an austere one, then, they will endorse the reading of P1 according to which experiences present clusters of property-instances and not property-types. I will therefore set aside the property-type reading and proceed to evaluate the property-instance version of afa.

3.2 Extensional versus Intentional Presentation

The second ambiguity in P1 concerns the sense in which the relevant property-instances are ‘presented’. Here, we can differentiate between two different notions of presentation that I will call presentation-of and presentation-as. Presentation-of is a purely extensional notion that captures whatever entities experiences in fact present. Thus, experience E is a presentation-of a cluster of property-instances φ iff φ is among the things that E in fact presents. This is distinct from the idea that, in perception, things are presented as being some specific way; e.g. red, or instantiated. Presentation-as is an intentional notion involving some property-type or types. While experience may involve presentation-of property-types, instances, or a mixture of both, something can only be presented-as a property-type and not a property-instance (except in the sense that that it could be presented-as being of the type property-instance, though it is doubtful that experiences have this kind of theoretically loaded content—see Sect. 3.3). So, while one or both senses of ‘presentation’ may hold in individual cases, these two forms of presentation are conceptually distinct and should not be conflated.

To illustrate this difference, consider the visual presentation of the outline of an equilateral triangle —call this figure F. A visual presentation-of F need not be presented-as a triangle, polygon, or indeed anything in particular. An infant or cat, for example, might visually experience F as a mere collection of lines or marks, rather than as a triangle. Such an experience would be a presentation-of F without being presented-as, say, a triangle or polygon, since the perceiver does not register it as such, perhaps due to their lacking the relevant concept. Conversely, an assortment of lines —traced at random in the sand by ants, for example—that happens by chance to bear a striking resemblance to an equilateral triangle might, under suitable circumstances, be presented-as a drawing of a triangle in virtue of the perceiver being visually aware of its shape. This need not, however, be a presentation-of such a figure, since the shape may not actually be an equilateral triangle, but merely approximates one—for example, the lines are not all the same length, or are slightly curved. Thus, presentation-of does not entail presentation-as, nor vice versa.¹⁹

With the above distinction in hand, in what sense can visual experiences be said to present property-instances? Again, rv and ar support different responses. To the extent that rv supports the claim that property-instances are presented by experience at all, since this is not required by the view, it entails that they are presented-as falling under one or more specific types. As we have seen, this straightforwardly entails the existence of content, and so cv. ar, on the other hand, supports the claim that experience involves the presentation-of property-instances, while rejecting the claim that these need be presented-as being of any specific type. As noted above, ar theorists deny that abstract entities such as property-types are among the things that experience presents. Moreover, as we have seen, the purely extensional notion of presentation-of does not commit the ar theorist to any particular claim about how such instances are presented as being, except that they make some (potentially indiscernible) contribution to the experience’s phenomenal character.

This might seem puzzling since, one might think, perceptual experiences present us with instances of, for example, redness or roundness. That is, such properties are among the things that we visually encounter in the world. ar, however, claims that in perception we can be presented with something that is red— a ripe tomato or red trope, for example—without it being presented as being red in the sense of being presented with a property-type. While the former will, under normal circumstances, give rise to an experience in which something appears or looks red, i.e. has red phenomenal character, according to ar the latter relates to a post-perceptual state such as judgement or belief (Travis, 2004), the registration of visually relevant similarities (Brewer, 2011), or some other cognitive process ‘downstream’ of experience. Unlike rv, then, while ar grants that experiences are presentations-of property-instances in the purely extensional sense, these need not be presented-as being red, round, instantiated, or anything else. Furthermore, ar theorists deny that such a claim can be justified by reflection upon the phenomenal character of perception alone. Rather, this is a theoretically loaded judgement about the content of perceptual experiences that is inadmissible in an argument that purports to establish that experiences have contents in the first place. They will therefore reject the intentional reading of ‘presents’ in Siegel’s P1 in favour of the extensional reading.

3.3 Instantiation versus Phenomenal Presence

This leaves us with the problem of how to understand Siegel’s claim that experiences present property-instances “as being instantiated”. The idea that this identifies some way that property-instances are presented as being, i.e. instantiated, is implausible for several reasons. First, ar denies that perception constitutively involves presentation-as, which is not entailed by the purely extensional notion of presentation-of. Second, it is implausible that the theoretically loaded notion of instantiation is apparent upon the basis of visual phenomenology alone. Such a claim is highly controversial and would require further substantive argument to establish—argument that Siegel does not provide. Third, unlike property-types, which possess generality and so are multiply instantiable, property-instances are particulars and so only ‘instantiated’ in the sense that they exist, or seem to inhere in visible objects.²⁰ But, as noted above, afa makes no mention of objects. Moreover, according to the metaphysics of ar, perceptual presentation-of a property-instance straightforwardly entails its existence, since it entails that there is some such thing that is presented. So all such instances presented in perception are, by definition, ‘instantiated’.²¹ The ar theorist, then, may reject P1 on the basis that the theoretical notion of instantiation doesn’t feature in visual phenomenology, or else insist on dropping the qualification entirely by claiming it is in the nature of veridical perception to present instantiated properties in the form of property-instances or tropes.

There is, however, an alternative reading of this phrase that captures its intended spirit of without these problematic implications, which is that visual experience involves a sense of the existence or presence of visible objects and their properties. For reasons that should now be familiar, we must be careful not to construe the relevant feature in terms of experiences conveying something to the subject in the manner of testimony, since this would trivially entail the existence of content and so beg the question in favour of cv.²² Nevertheless, perceptual experience is not merely neutral as to how things in the world are, as in the case of imagining. Rather, it has prima facie import concerning the presence or existence of objects and their perceptible properties in the perceiver’s environment. The relevant import is only prima facie because it is possible to reject it —for example, by not taking experience ‘at face value’ if one suspects it to be deceptive—and may be misleading; e.g. in the case of hallucination. Though individual theories of perception may differ as to why visual experience seems this way, they can all grant that this sense of phenomenal presence, as I will call it, is a genuine feature of experience.²³ If this is what is meant by presenting properties “as being instantiated”, then the ar theorist can endorse a version of P1 without thereby presupposing the existence of intentional content in a way that would render afa circular.

Resolving each of the above ambiguities, then, we can gloss the ar-compatible reading of P1 as follows:

P1′:

All visual perceptual experiences involve (a) the presentation of clusters of property-instances, and (b) the phenomenal presence of those property-instances in the subject’s environment.

Note that my aim here is not to render the argument metaphysical committal in a way that favours ar. Rather, I merely wish to make explicit the most concessive, non-question begging interpretation of afa that is compatible with the view. As such, P1′, or something like it (see below), is what the ar theorist will take P1 to mean if they are inclined to accept it at all. Moreover, while Siegel characterises P1 as expressing a phenomenological datum about how experience seems to us, we can now see that it turns out to heavily theory-dependent. Varying the underlying metaphysics thus enables us to tease out the metaphysical assumptions built into P1 that render it question-begging against the ar theorist.

3.4 Non-Veridical Experience

There is a further problem with P1′, however, which is that it is false for non-veridical experiences such as hallucinations. Along with certain kinds of illusions, these do not involve the presentation of concrete property-instances, but merely appear to do so. Indeed, many ar theorists endorse a form of disjunctivism according to which hallucinations can only be characterised in epistemic terms as subjectively indistinguishable, or “indiscriminable” (Martin, 2004), from veridical perception rather than being states of the same psychological or metaphysical kind—a position known as negative disjunctivism (Soteriou, 2020). The negative disjunctivist will reject P1′ on the basis that it is true only of veridical perception, and not of all visual experiences as is claimed, where the latter is taken to include perceptual hallucinations and illusions.

To avoid this difficulty, we might restrict the scope of P1′ to include only veridical experience. However, Siegel’s conception of experience explicitly includes non-veridical experience, arguably rendering P1′ unsuitable for further evaluation of afa. Nor can this problem be solved by saying that hallucinations present clusters of possible property-instances. First, there are no obvious candidates for precisely which particulars these would be given that subjectively matching hallucinations are indistinguishable from the presentation of a large number of possible such clusters (see below). Second, explaining the phenomenal character of hallucinations in terms of the presentation of mere possibilia, would, given ar, require subjects to be perceptually related to entities that do not exist in the actual world. Since non-actual particulars are not spatiotemporally or causally related to the subject, this goes against the ar theorist’s motivation of explaining perception in terms of a relation to everyday objects, events and/or their properties, not to mention violating methodological naturalism, rendering the resulting view highly implausible and unattractive.

Given the disjunctivist nature of ar, a better way of reformulating P1′ to include non-veridical experiences would be to employ a disjunctive notion of presentation —call this presentation*. On this view, an experience is a presentation* of a cluster of property-instances just in case (i) it is a perceptual presentation-of a cluster of property-instances, or (ii) it is subjectively indistinguishable from an experience that is such a presentation. P1 and P2 of afa may then be reformulated as follows:

P1*:

All visual perceptual experiences involve (a) the presentation* of clusters of property-instances, and (b) the phenomenal presence of those (apparent) property-instances in the subject’s environment.

P2*:

If an experience E presents* a cluster of property-instances F and involves the phenomenal presence of those (apparent) property-instances in the subject’s environment, then:

Necessarily: things are the way E presents* them only if E is a perceptual presentation-of cluster F of property-instances.

Unlike P1′, P1* is true of both veridical and non-veridical experience, and so compatible with negative disjunctivism, avoiding the above objection. Nevertheless, problems arise later in the argument, since it does not follow from P2* that the resulting accuracy conditions will be “conveyed to the subject” as per P3. This is because the presentation* of merely apparent property-instances, i.e. it seeming that one is being perceptually presented with a cluster of particulars, does not entail the presentation of any actual property-instances, which only occurs in the veridical case. In the non-veridical case, there need be no such cluster of instances, and so no particular accuracy conditions being conveyed. Indeed, this is a recurrence of the above objection, since there are no clear candidates for precisely which property-instances (as opposed to property-types) are conveyed by hallucinations, and indeed some illusions. In such cases — hallucinating a blue rectangle, for example — the ar theorist will say that no actual instances of blueness or rectangularity are being presented. Rather, it merely seems to the subject as if they are. Thus, P3 is false on the reformulated version of the argument, and will be rejected by the ar theorist.

To the extent that ar theorists are inclined to endorse a version of P1, then, they will either (a) restrict it to veridical experiences only, i.e. those that genuinely present property-instances, or (b) reject P3. Either way afa fails. The reasons for the falsity of these premises turn on precisely what is ‘presented’ (in the case of P1) or ‘conveyed’ (in the case of P3) to the subject. According to ar, veridical perceptual experience consists in the presentation of mind-independent particulars, such as objects and their properties, to a conscious subject, viz.

P1″:

All veridical visual experiences involve (a) the presentation of clusters of property-instances, and (b) the phenomenal presence of those property-instances in the subject’s environment.

In the following section, however, I argue that even with this reformulated interpretation of P1, afa does not establish that experiences convey accuracy conditions of the kind that representationalists standardly take experiences to have, and so fails to establish cv (Sect. 5).

4 Do Experiences Convey their Accuracy Conditions?

To possess accuracy conditions, there must exist some specific (in the sense of there being one in particular) and determinate (in the sense of there being some fact of the matter) set of conditions that specifies how the world must be in order for a given experience to be accurate, or veridical. Such conditions can be thought of as a complex conjunction, each conjunct of which has the form ‘∃x φx’ (for general content) or ‘o is φ’ (for particular content), where φ is some perceptible property-type.²⁴ Alternatively, they might be thought of as proto- or non-propositional in nature (Burge, 2018). P1″, however, does not characterise visual experiences in terms of objects or property-types, but the presentation of clusters of property-instances, where this need not require the relevant property-instances to be presented-as instantiating any specific types. Rather, according to ar, experiences are presentations-of particulars in the purely extensional sense described in above (Sect. 3.2). As such, it is unclear whether presentations of property-instances have any type-based accuracy conditions, or, if they do, what those conditions would be.

The problem is not trivial, since for any given cluster of particulars that is presented, there are multiple possible decompositions into types and tokens. A red object, for example might be presented as red, reddish, vermilion, red₄₅₁ (a maximally determinate shade of red), and so on. Moreover, different decompositions carve up the world in different ways. A surface may be presented as a single property-instance, a series of adjacent colour patches, a spatially coincident combination of shape, colour, texture, etc., or any combination of these. The presentation-of property-instances thus underdetermines which, if any, property-types such instances are presented-as falling under, as noted in Sect. 3.4. In the absence of some method of determining the relevant types, P1″ is insufficient to establish the existence of specific and determinate (in the above senses) type-based accuracy conditions for experience. This threatens to render afa invalid since it does not follow from P1″ and P2 that each experience conveys any specific set of accuracy conditions to the subject, as stated in P3.

Travis (2004, 2013) presses a version of this objection against rv, concluding that perception cannot be representational. Though Siegel (2010a, p. 60) discusses Travis’s argument, she misconstrues this objection as concerning the semantics of ‘looks’, rather than the metaphysics of appearances per se. Although a detailed discussion of Travis’s argument lies beyond the scope of this paper (see Wilson, 2018), it will suffice for present purposes to note that afa does not itself contain the resources to oppose it. Rather, it simply assumes that the relevant accuracy conditions exist on the basis of whatever properties are presented in experience. This makes sense if experiences present property-types, as rv theorists standardly claim. But if, as per P1″, experiences merely present clusters of property-instances, the relevant types remain underdetermined, and so the argument does not go through.²⁵

P1″ does, however, entail the existence of a specific set of accuracy conditions of the form:

• (C) Experience ψ is accurate iff it presents cluster of property-instances φ

where φ specifies the set of property-instances that experience ψ is a presentation of. Call these an experience’s particularity conditions. Every perceptual presentation (though not presentation*) of a cluster of property-instances will have a set of particularity conditions that depends upon the exact cluster of particulars presented, and which corresponds to the obtaining of that experience’s identity conditions. Moreover, the existence of these conditions does not depend upon any further argument concerning the types that each individual property-instance is presented as instantiating, and so is consistent with P1″. The ar theorist should therefore be prepared to grant that experiences have de re accuracy conditions of this form.

In order for this to constitute content, however, experiences must also “convey” their accuracy condition to the subject, as per the hidden premise P5 of afa. According to Siegel (2010a, p. 51), this may be done in one of three ways:

• (i) “if it would be a content of explicit beliefs that are natural to form on the basis of visual experience”; for example, that there is a red tomato before one

• (ii) “if it enables the experience to guide bodily actions”, such as reaching out to grasp an object in the appropriate manner, or

• (iii) “if it is manifest to introspection that it is a content of experience”.

Note that meeting one condition alone is not sufficient to establish the existence of content since (i) and (ii) are equally compatible with views of experience, like ar, that explain the contents of belief or guidance of action in non-representational (e.g. causal) terms. Condition (iii) is also controversial since if it were apparent on the basis of introspection that experiences had content then no further argument for cv would be required. ar theorists, however, consider apparent cases of (iii) to identify the content of a perceptual judgement or belief, rather than a content of experience (Locatelli and Wilson, 2017, p. 214), thereby rendering a version of (iii) similarly compatible with ar. Nevertheless, I take Siegel’s view to be that in order to qualify as the accuracy conditions for experience, the resulting content must in some sense be available from a first-personal perspective for thought or action; i.e. it is content for the subject.

This requirement, however, places further constraints upon the factors and conditions that individuate the alleged contents. Such factors cannot, for example, lie entirely outside the subject’s perspective—for example, by requiring specialist knowledge about the functioning or evolution of the human brain or visual system—otherwise it would not be apparent to the subject what their experiences represented (Wilson, 2018, pp. 217–218). Does this mean that an experience’s particularity conditions cannot be conveyed to the subject in the relevant sense? At this point, the ar theorist might argue that the identities of the relevant property-instances are not apparent solely on the basis of their presentation in experience. If they were, then one should be able to distinguish between veridical experience and hallucinations or illusions in which no such instances are presented solely on the basis of introspection. In the case of subjectively matching hallucinations at least, this is impossible by definition, and so one might conclude that experiences do not convey their particularity conditions. This argument, however, would be too quick.

First, the situation is not so clear cut in the case of action or belief. One might find it natural to believe on the basis of the perceptual presentation of property-instance p₁ that some object is that colour, where the resulting belief refers to the colour of p₁ via a demonstrative mode of presentation (cf. McDowell, 1994, p. 57). If perception facilitates demonstrative thought about perceived qualities in this way, as many ar theorists agree it does, this arguably meets Siegel’s condition (i) for “conveying”. Similarly for condition (ii), a particular property-instance—for example, the shape of an object —might guide bodily action in a way that relates not only to a general property-type, but to the particular idiosyncratic shape that is presented. Of course, (i) and (ii) could be disputed by the ar theorist as question-begging since one might equally give non-representational explanations of these phenomena in terms of sub-personal processing or non-content-bearing states.

Even in the case of introspection, however, there is a sense in which one might take the identity of particular property-instances to be “conveyed” via the felt particularity of experience. Perceptual experiences intuitively seem to involve the presentation of independently existing objects (or the world); i.e. they exhibit phenomenal presence (Sect. 3.3). This is true regardless of whether experiences are held to possess particular or general contents, or whether, as claimed by ar, veridical experiences are partly constituted by mind-independent objects and their properties. One possible explanation of this phenomenon is that experiences convey that the very set of particulars that they present are in fact being presented. That is, (a) experiences convey de re accuracy conditions of the kind described by (C) to the subject,²⁶ and (b) this conveying explains or constitutes phenomenal presence. If this is correct then it might be thought to support the claim that, even under ar, experiences convey their particularity conditions.

An alternative hypothesis, however, is that phenomenal presence can be explained, or is constituted by, the conveying of some weaker general condition such as:

(D):

Experience ψ presents the cluster of property-instances that ψ in fact presents.

On this view, phenomenal presence does not indicate that experiences convey the set of particular property-instances which are presented. Rather, it is explicable in terms of a general, or de dicto, condition that does not differentiate between one experience and another. Indeed, in this respect (D) is analogous to the kind of trivial accuracy conditions for which Siegel (2010a, p. 42) rejected the Argument from Accuracy (Sect. 2). Since conveying conditions of the form (C) or (D) can equally explain the felt particularity of experience, insisting on the former merely to justify the conveying of particularity conditions by experience seems unnecessary and ad hoc. This is supported by the fact that experiences in which no concrete property-instances are presented, i.e. hallucinations, seem to share the same phenomenological feature. But here there is no obvious candidate for the set of particular property-instances that is being conveyed (Sect. 3.4). Hence, if phenomenal presence is to be explained by the conveying of accuracy conditions at all, (D) seems a more plausible explanation than (C), since the former is common to all perceptual experiences and not only those in which a cluster of property-instances is in fact presented, i.e. veridical perception. Hence experiences need not be thought of as conveying their particularity conditions.

In summary, the ar theorist will either

• (1) reject P1 on the basis that experiences present property-instances and not property-types

• (2) reject P1′ because hallucinations do not present clusters of property-instances, or

• (3) reject P3 because non-veridical experiences do not convey their accuracy conditions to the subject because there are no plausible such conditions.

Moreover, even in the veridical case it is doubtful that the particularity conditions (C) of an experience are conveyed since phenomenal presence can equally well be explained by the obtaining of some weaker condition, such as (D), or in non-representational terms. In the absence of some independent reason for thinking that experiences convey their particularity conditions, then, afa does not show that ar entails cv without the unnecessary and ad hoc stipulation that phenomenal presence must be explained in terms of the conveying of content. This stipulation, however, would render afa circular and so unpersuasive.

5 Does the Conveying of Accuracy Conditions Entail Content?

There is, however, a more fundamental problem with afa. Even if the ar theorist is prepared to grant that experiences convey their particularity conditions to the subject, this fails to yield any substantive or interesting form of content. As we have seen, representationalists take perceptual content to involve the attribution or instantiation of property-types: that thing is red, there is a pine tree, and so on; i.e. it is attributive content.²⁷ The accuracy conditions that are entailed by ar, however, only specify which cluster of property-instances a given experience is a presentation-of. Rather than attributing a general type to a particular, as per rv, particularity conditions attribute the existence of some specific cluster or set of particulars to the world. Or, to put it another way, they specify that some specific cluster of particulars is ‘instantiated’, i.e. that it exists. This falls well short of the kind of content that Siegel and other representationalists standardly take experiences to have. Not only does such content not contain any high-level or ‘rich’ property-types, such as being a tomato or being a pine tree (Siegel 2010a, pp. 104–8), it fails to attribute any property-types whatsoever. Rather, it merely specifies a cluster or set of property-instances without any indication of what types they fall under.

If all that Siegel meant by ‘content’ were that experiences seem to present clusters of particulars, then the relationalist should accept that experiences have content in this specific and limited sense. That is to say, according to ar, veridical perception consists in the presentation-of clusters of particulars without these being presented-as being any specific way —including “being instantiated”. But this simply restates the central claim of ar that experiences present particular properties, and does not amount to any substantive concession or commitment to cv. Thus, while ar might be taken to meet the formal conditions for the existence of content specified in P5, at least in the veridical case, this is a pyrrhic victory for Siegel since such ‘contents’ fall well short of what representationalists standardly mean by this term. Indeed, this is much less than Siegel herself takes afa to have established, as discussed throughout the rest of her book (Siegel 2010a, pp. 99–115) and elsewhere in relation to the richness of perceptual attributive content and the representation of natural kind properties (Siegel, 2006; Siegel & Byrne, 2017). Such questions simply do not arise for the kind of trivial and uninteresting ‘content’ to which ar theorists are committed, which arguably plays little or no role²⁸ in explaining perceptual phenomenal character or the subjective indistinguishability of veridical and non-veridical experiences —two of the hallmarks of most (though not all) content views.

A more direct response to afa on behalf of the ar theorist, then, is that the kind of accuracy conditions their view entails, namely particularity conditions of the kind described by (C), do not qualify as contents at all. Hence the hidden premise of Siegel’s argument, P5, is false. After all, if the mere presentation of a set of particulars were sufficient to entail representational content, then any such presentation —of a medal or a selection of hors d’oeuvres, for example—should have content. But these presentations do not entail contents in any philosophically interesting sense of the term. So, by parity of reasoning, neither does the presentation of clusters of property-instances in experience. Alternatively, if only perceptual presentations are supposed give rise to content then it remains unclear why the presentation of property-instances, as in ar, should result in the kind of content that philosophers take to be relevant to perceptual experience, i.e. attributive content. Since this was precisely what afa was itself supposed to demonstrate, the argument may be rejected as unsound due to the falsity of P5.

6 Conclusion

Though Siegel’s Argument from Appearing lends superficial plausibility to the claim that relationalism shares some common ground with representationalism—namely, being a ‘content view’—it fails to establish that ar entails the existence of attributive content, as opposed to being non-representational as its proponents claim. Thus, even if the argument is rendered valid with the addition of P5 it either: (a) begs the question against ar by illicitly assuming that property-types, rather than property-instances, are presented in experience; (b) does not apply to all visual experiences, where this is taken to include perceptual hallucinations and illusions; or (c) fails to establish that visual experience conveys any substantive or interesting form of content beyond the clusters of property-instances that, in the veridical case, it presents—something to which ar theorists are in any case already committed.

As with other arguments concerning the presentation of properties in visual experience, Siegel conflates a familiar but metaphysically neutral fact—namely, that appearances can be paradigmatic of certain kinds of objects and properties, and so may be described in predicative terms —with a contentious theoretical claim about the nature of visual experience: that it “presents clusters of properties as being instantiated”. In doing so, afa fails to take the metaphysics of austere relationalism seriously, enabling Siegel’s opponents to reject one or more of its premises. At best, afa, and by extension other arguments like it, constitute a dramatic expression of conviction²⁹ that simply asserts what representationalists already take to be the case: that visual experiences possess representational content, while lacking persuasive force against their relationalist opponents. Consequently, while there may be arguments that can adjudicate between or demonstrate the compatibility of relational and representational views, the Argument from Appearing is not one of them.