1 Introductory Remarks

There are two fundamentally different ways of conceiving perceptual experience. The first one is this. Perception has a representational content. When I see a moth on a tree, I represent the world as being a certain way, that is, my perceptual states have conditions of satisfaction. ^{Footnote 1} When there is a match between the way the world is and the way that our perceptual experiences represent the world as being, their representational content is veridical; when there is a mismatch the content is falsidical. According to John Campbell, we may call this the ‘content view’ of perceptual experience. ^{Footnote 2} Of course, Campbell’s ‘content view’ is an umbrella term that covers quite different views of he alleged content of perceptual experience. For those who are not acquainted with the recent literature, it is worth mapping the fundamental distinction (without being exhaustive).

The first divide is between general and singular content views. According to the general content view, the content of experience is best modelled as a canonic existential proposition. ^{Footnote 3} For example, my perceptual experience is of a moth on the tree. In contrast, according to the singular content view, my perceptual experience represents this or that moth on the tree. This singular content can be modelled either as a Russellian proposition, or as a Fregean proposition, or set-theoretically à la Stalnaker. If we assume the Russellian model, particulars are seen as constituents of the content itself. In contrast, if we assume the Fregean model, particulars are not constituents of the content, but rather their modes of presentation. As perceptions rely on the relation between the subject and particulars, such modes of presentation are usually conceived as being de re. ^{Footnote 4} Finally, if we assume the set-theoretical model, the representational content of perceptual experience is modelled as a set of the worlds containing the particulars and their properties. Such content is true if and only if (iff) the actual world is a member of the relevant set. ^{Footnote 5}

A second divide opposes reductionist and anti-reductionist accounts of the content of perceptual experience. According to Fred Dretske and Michael Tye, among others, it is possible to account for the representational content of perceptual experience in non-semantic and non-psychological terms. ^{Footnote 6} In contrast, according to David Chalmers and Tyler Burge, among others, the notion of the representational content of experience is primitive and irreducible. ^{Footnote 7} A third divide opposes those who claim that the representational content of perception should be modelled as a complete proposition ^{Footnote 8} to those who claim that such content should instead be modelled as the content of complex predicates: properties. ^{Footnote 9} As I hope to make clear in the paper, I reject the singular content view, I leave the debate about reductionism open, and I reject the complete content view.

The second view is this. Perceptual experience is a matter of putting the agent in contact with particulars. In traditional modern philosophy, this relationship was mainly conceived of as indirect: one is never perceptually acquainted with things directly as they are in themselves, but only indirectly, via direct acquaintance with their mental proxies: ideas, Vorstellungen, sense-impressions, etc. Yet, in contemporary philosophy of perception this relation is conceived of as direct; this is what people today usually call ‘naïve’ or ‘direct’ realism. In accordance with Campbell, we may label this the ‘relational view’ of perception. ^{Footnote 10} Versions of this view were popular amongst the early 20^th century Oxford Realists such as Bertrand Russell, ^{Footnote 11} but the recent work of Campbell, Charles Travis, Mark Johnston, Bill Brewer, William Fish, and Michael Martin have brought the proposal back into discussion. ^{Footnote 12} Martin calls his position ‘naïve realism’ ^{Footnote 13} ; while Brewer calls his own the ‘object view.’ ^{Footnote 14} But I prefer Campbell’s label: the ‘relational view.’ ^{Footnote 15} Again, Campbell’s ‘relational view’ is also an umbrella term. However, since the differences are minor, they are irrelevant for the claim I support in this paper. All we need to keep in mind is that according to the relational view the function of perception is to put us in contact with particulars and that particulars are constituents of our perception.

Now, in philosophy of perception, the particularity thesis is the general claim that perceptual states are individuated, at least in part, by the particulars that brought them about. Of course, this view is not indisputable. For instance, Frank Jackson, David Lewis, Gilbert Harman, Alan Millar, Collin McGinn, Martin Davies, Charles Siewert, Alex Byrne, and Christopher Hill, among others, reject the view that particulars play a role in individuating perceptual states. According to them, perceptual experience is constituted only by general elements. ^{Footnote 16} However, there is indirect evidence that supports the particularity thesis: (i) the fact that perceptual experience grounds demonstrative reference, (ii) the fact that perceptual experience yields singular thoughts, and (iii) the fact that perception provides us with the evidence that justifies singular thoughts. If we assume David Kaplan’s account of demonstratives, according to which the particular demonstrated is the value of a function that maps the context of demonstration onto the particular demonstrated in that context, it is natural to assume that the perception of particulars in each context is what fixes the reference of the demonstrative in that context. Likewise, if we assume that there are singular or direct thoughts whose reference is determined not by satisfaction of any identifying conditions, but rather by perception or perception-like relations, it is natural to assume that perceptions are of particulars, etc. However, I will take the particularity thesis for granted here rather than argue for it.

Now, even for those like me who endorse the particularity thesis, the relation between particulars and perceptual states is an open question. And this is the question that I wish to address in this paper. Based on the main divide between the ‘content view’ and the ‘relational view,’ we can envisage at least two different versions of the relation between particulars and perceptual states: the epistemic particularity thesis and the singular content thesis. According to the first, the function of perception is to put us in direct contact with particulars, regardless of the ability to discriminate them from other particulars and from the background. For example, I see a moth perfectly camouflaged on a tree regardless of whether I am able to discriminate it from its background and from other particulars.

According to the second, particulars are represented as constituents of the content of perceptual experience. The idea here hinges on the assumption that we cannot perceive anything without being able to discriminate it from other particulars and from its background. I cannot perceive the moth camouflaged on the tree if I am unable to discriminate it from its background. Intuitively, what my perceptual experience represents is not a particular as being a moth, but rather this particular as being a moth. Likewise, if the condition of satisfaction of my content is not met by the state of the world, what I misrepresent is this thing as being a moth rather than some particular that fails to be a moth.

Interestingly, the singular content view and the singular relational views are attractive and unattractive for the same reasons. Insofar as both endorse the particularity thesis, the most obvious attraction of both of them is the fact that they satisfy the constraint that Susanna Schellenberg some years ago called the “particularity desideratum”: ^{Footnote 17} regardless of the way that perceptual experiences are type-individuated, they are also certainly token-individuated by the particulars that bring them about. ^{Footnote 18} In the case of the singular content view, this constraint is supposedly met by assuming that my perceptual experience represents or misrepresents that particular on the tree as being a moth. In the case of the singular relational view, the constraint is supposedly met by assuming that my perception puts me in direct contact with that moth on the tree.

The unattractive character of both views is the fact that they must embrace different sorts of disjunctivism to account for generic hallucinations. Let us consider first the relational view. Since in the case of generic hallucinations there is no perceived particular, there is no perceptual ‘relation’ in the first place (because any relation, say, between x and y requires the existence of both x and y). Thus, generic hallucinations are not perceptual experiences in the light of the relational view. Perceptual experience and hallucinations do not belong to the same psychological kind. There is no shared mental state common to veridical and hallucinatory visual experiences, regardless of whether they introspectively look the same. Perceptual experience and hallucinations are no more closely related than a lemon and a bar of soap that just looks like a lemon. ^{Footnote 19} When I hallucinate a yellow cube straight ahead of me, I perceptually experience nothing.

The singular relational view must embrace a content disjunctivism. Perceptual experiences (illusory or not) represent a singular content. In contrast, hallucinatory experiences represent a gappy singular content. Now, as long as there is no common content or no common experience shared by perception and hallucination, both the singular content view and the singular relational view fail to meet the constraint that Schellenberg called the “indistinguishability desideratum.” ^{Footnote 20} Neither of them can account for the intuitive fact that, if I am staring at a moth on the tree, and unbeknownst to me another qualitatively identical particular replaces it, or if I start to hallucinate a qualitatively identical particular at the same place, I cannot notice any phenomenological difference by introspection.

My aim in this paper is to defend a peculiar epistemic version of the particularity thesis, which in the absence of a better name I call the ‘singular relational plus relativistic content view.’ This peculiar view results from a sui generis combination of what I call the singular relational view and what I call the ‘relativistic content view.’ From the singular relational view, I take the idea that particulars are not represented as part of putative singular content. Instead, we are perceptually acquainted with them in the relevant sense that experience puts us in direct perceptual contact with them. In that sense, particulars token-individuate perceptual states by being causally responsible for the actual token experience.

From the relativistic content view, I take the idea that the content of perceptual experience is less than a complete proposition by Fregean lights (something that is absolutely true or false). Rather, the content of perceptual experience is best modelled as a propositional function, that is, the content of a complex predicate that is true or false only relative to some circumstances of evaluation, namely the so-called ‘centred worlds,’ that is, worlds with a certain subject, a certain time, and certain particulars that are causally responsible for the relevant type of experience under normal conditions designated at the centre. The content of perception is veridical iff the complex predicate is true only of the particular that is causally responsible for the relevant type experience under normal conditions.

My epistemic particularism takes the following form. Particulars enter into the picture twice: first as the cause of the actual token experience, and second as the indexes of evaluation of the relative content of experience, namely, as the cause of the relevant type experience under normal conditions. Thus, while perceptual states are token-individuated by the particulars that are causally responsible for the actual token experience, they are also type-individuated by the content of experience, that is, by the representation of locational properties (relativistic content view) that are veridical or falsidical only of those particulars that under normal conditions would be responsible for the token experience (singular relational view).

My defence is an additional case of the argument to the best inference. In Section 2, I argue that the singular content view cannot meet the particularity desideratum just like the old general content view: in Grice-like cases, there is a mismatch between the veridicality of the putative singular content of experience and the illusory character of those experiences. In Section 3, I present my own singular relational plus relativistic content view, a hybrid view that results from a peculiar combination of the relational with the content view.

In Section 4, I argue that my hybrid view has all the virtues and none of the problems of the rival singular content view. It can meet both the particularity and the indistinguishability desiderata. It can easily meet the particularity desideratum by assuming the particulars that token-individuate perceptual experience are those that perceptual experience puts us in epistemic relation to and that are actually causally responsible for the relevant token experience. However, neither the causal relation nor the particulars are represented as components of the content of experience. My account can easily meet the indistinguishability desideratum by assuming that veridical, falsidical, and hallucinatory experiences have the same content: a locational property. This is veridical or falsidical not of the particulars that are actually causally responsible for the relevant token experience. Instead, this property is veridical or falsidical of the particulars that would normally be causally responsible for the token experience. The case of generic hallucination is just a case of falsidical perception: whatever is actually causing the relevant token experience are not the particulars that would normally be causally responsible for the relevant type experience.

2 Singular Content View

Let us now assume that the content of experience is singular. Schellenberg presents what she calls the ‘singular content’ argument in favour of this claim. If subject S perceives particular a, then S discriminates and singles out a.

1. V. If a subject S perceives a particular a, then S discriminates and single out a.

2. VI. If S discriminates and singles out a, then S’s perceptual state M is constituted by discriminating and singling out a.

3. VII. If M is constituted by discriminating and singling out a, then S represents a (under a mode of presentation).

4. VIII. If S represents a (under a mode of presentation), then S’s perceptual state M is constituted by a.

5. IX. If a subject S perceives particular a, then S’s perceptual state M is constituted by a. ^{Footnote 21}

The singular content thesis is a specification of the particularity thesis in semantic terms, which Schellenberg favoured in her last paper. ^{Footnote 22} According to her, the epistemic version of the particularity thesis is best explained via semantic content, otherwise, “it is not clear what it would be to be brutely acquainted with a particular in this way.” ^{Footnote 23} The key idea is that the ability to discriminate particulars is our fundamental perceptual ability:

However, the claim that perceptual acquaintance can only be comprehensible by means of the notion of perceptual discrimination is controversial, and is disputed by several psychologists and philosophers of perception in the so-called ‘cases of change blindness. Dretske, for example, notoriously claims:

Notoriously, there is barely any agreement on how to understand the notion of acquaintance. The best we can say is that acquaintance is a perception-like relation of contact so that, when I am acquainted with something, I am directly aware of it. However, assuming that perceptual discrimination is the fundamental component of perception amounts to embracing the controversial claim that attention and recognition are necessary conditions for perception of particulars. Yet, we can see particulars without noticing or recognizing them. I do not have to notice, recognize, identify, or single out every book I see on the shelves as I scan them in search of a particular book. In this search, I have probably seen a hundred books or more, while I have noticed only a few of them. I may have seen the book I was searching for more than once, without noticing it. So, unless something was blocking the woman’s vision of her husband, there is no reason to deny that she saw him if he was standing right in front of her, regardless of whether she was able to discriminate him from other people in the crowd. Schellenberg must be wrong when she claims that: “being acquainted with a particular at minimum requires the employment of perceptual discriminatory capacities whereby one singles out the particular from its surround.” ^{Footnote 26} The ability to single out particulars is not a necessary condition for perception of particulars.

The other question is whether discriminatory ability is a sufficient condition. Let us retake the visual experience of a yellow cube straight ahead. According to Tye, for example: “Intuitively, I misperceive that cube. My experience misrepresents it.” ^{Footnote 27} Moreover, as we have seen, it is counterintuitive to claim that my perceptual experience of a yellow cube straight ahead represents a or some particular that instantiates or not the property of being a yellow cube straight ahead. Again, my visual experience represents or misrepresents that cube over there.

However, Campbell complains:

Indeed, semantic particularism also faces problems in Grice-like scenarios. The first scenario is what in the literature is called a case of ‘veridical misperception.’ Let us suppose that I am looking directly ahead to where there is a yellow cube. Yet, unbeknownst to me, someone has projected a hologram of a yellow cube straight ahead of me, covering up a real yellow cube. In this simple case, my visual experience is clearly illusory because I do not see the real yellow cube straight ahead of me. What I see is the hologram of a yellow cube that covers the real one. Yet, the content of my visual experience is veridical. First, since the real yellow cube and the hologram of a yellow cube occupy exactly the same place straight ahead of me, by singling out the hologram (‘this thing straight ahead of me’) I am also singling out the real thing. Second, that thing straight ahead possesses the property of being a yellow cube. Moreover, considering that, if the veridicality of the putative singular content of my visual experience depends on my ability to discriminate the right particular and on my ability to attribute the right property to it, it is hard to see why the content of my visual experience could not be veridical while the experience itself is illusory. To use Tye’s own words, the singular content view also “yields an unequivocal result of content-veridicality” when the perceptual experience is clearly illusory. ^{Footnote 29}

Let us suppose a case of veridical hallucination. There is a yellow cube in front of S. However, unbeknownst to S, this information is reflected in the light from the cube and reaches S’s retina, but is processed no further. An evil neuroscientist has blocked the signals from S’s retina to her optic nerve, while simultaneously activating her visual cortex by means of electrical probes that work in the same way as neurological signals. ^{Footnote 30} Under these abnormal circumstances, the right thing to say is that the demonstrative successfully singles out the particular straight-ahead of S. Moreover, we cannot deny that the singular content is veridical since the particular straight-ahead of S is a yellow cube. However, the experience is hallucinatory. It follows that the semantic particularism should be rejected. The ability to single out particulars is not a sufficient condition for perception of particulars. I can discriminate the hologram of a yellow cube without seeing it. The fundamental element in perception is the ‘brute’ contact with those particulars that under normal conditions are causally responsible for the relevant experience.

Now, if semantic particularism must be rejected, how can we intuitively explain that I see this yellow cube or that I have misperceived that yellow cube? Schellenberg provides us with the answer. In those cases, we can best talk of phenomenological particularity in the following terms: “A mental state manifests phenomenological particularity if and only if it seems to the subject that there is a particular present.” ^{Footnote 31}

On reflection, the main problem with the singular content view is the fact that demonstratives and other indexical devices cannot capture the main difference between the particulars that are causally responsible for the relevant actual token experience and the particulars that under normal conditions would be causally responsible for the relevant token experience. In the first Grice-like scenario, the particular that is actually responsible for the relevant token experience is not the real yellow cube covered by the hologram—which under normal conditions would be causally responsible for the relevant token experience—but the hologram itself. In the second Grice-like scenario, the particular that is responsible for the relevant actual token experience is not the real yellow cube straight ahead of me—which under normal conditions would be causally responsible for the relevant token experience—but rather the projected image of a yellow cube.

Particulars play two essential roles in perceptual experience. First, they token-individuate perceptual states insofar as they are causally responsible for the relevant actual token experience. Second, they constitute the circumstance of the evaluation of the content insofar as under normal conditions they are causally responsible for the relevant token experience. That said, the particular that matters for settling the veridicality of the content of experience is not the one that I am able to discriminate, but rather the one that under normal conditions would be causally responsible for the relevant token experience. In Grice-like cases, I am not even aware of them: in the first scenario, it is covered by the hologram. Demonstratives and the corresponding ability to discriminate or single out particulars cannot avoid the mismatch between the content and the experience. Therefore, we have lost the main reason to prefer singular content thesis to the epistemic particularity thesis, namely the assumption that the ability of discrimination is the fundamental element of perception.

The key question is how to conceive of the causal relation between the particular and the relevant type experience in relation to the content. According to John Searle, Chalmers, and François Recanati, ^{Footnote 32} those causal relations belong to a further layer of content. Nonetheless, if the causal relation belongs to the content of perceptual experience, we have to face the old accusation of over-intellectualizing the content of experience; after all, how can a subject, without any concepts whatsoever, represent the particular she sees straight ahead of her, and the same particular be causally responsible for the relevant token experience in her mind? According to Burge, the causal relation between the particular and the relevant token experience in the subject’s mind required by Searle’s account is simply “too complex or too sophisticated.” ^{Footnote 33} It is cognitively too demanding to assume that every perceiver has the resources required to refer to her own token experience as caused by particulars. ^{Footnote 34} According to Tye, the proposal is also counterintuitive: “When I see a tomato, for example, my visual experience is directed upon the tomato. It is not about itself in addition to the tomato.” ^{Footnote 35}

Chalmers addresses this objection as follows:

However, Chalmers’s explanation is not fully convincing. Any attribution of content (even nonconceptual!) must capture the way things appear to the subject or how the subject represents the world. Let us call this ‘Bermúdez’s constraint of content attribution.’ ^{Footnote 37} Now, it is beyond any doubt that not every creature that is able to represent a yellow cube straight ahead is also able to represent that particular as causally responsible for the relevant token experience in the subject’s mind at the time the experience takes place. It seems much more appropriate to think of the causal relationship between the particular and the token experience as an acquaintance relation that lies outside the content rather than a further layer of content represented by the creature. In other words, the causal relationship between particulars that are normally causally responsible for the relevant token experience determines the accuracy conditions of experience, but they are not aspects of these conditions. They are not conditions that must be satisfied for the content of experience to be veridical. Instead, they are preconditions that must be satisfied for the experience in question to have the veridical conditions it actually has, and in order to be counted as a perception.

3 The Singular Relational plus Relativistic Content View

As we have seen, there are at least two main different ways of understanding the particularity thesis: the epistemic particularity thesis and singular content thesis. As I have argued in the previous section, the singular content view has little to recommend it. Discrimination ability is neither a necessary nor a sufficient condition for perception of particulars. Moreover, indexicals cannot capture the main difference between the particulars that are causally responsible for the actual relevant token experience and the particulars that under normal conditions would be causally responsible for the relevant token experience.

I now want to present my version of the epistemic particularity thesis. My view results from a combination of what I call the ‘singular relational view’ with what I call the ‘relativistic content view,’ which in the absence of a better name I call the ‘singular relational plus relativistic content view.’ My fundamental claim is that perceptual states are token-individuated by particulars that are causally responsible for the relevant actual token experience (singular relational view) and type-individuated by what Lewis called ‘locational properties’ that those perceptual states represent (relativistic content view) and that are veridical or falsidical of those particulars that would normally be responsible for the relevant token experience. To be sure, we see and directly experience particulars in our perceptual field. Yet, we do not necessarily represent them as part of the representational content of our experience. ^{Footnote 38} Instead, we directly perceive them by being perceptually acquainted with them. And, by being perceptually acquainted with them, we represent complex propositional functions that under normal conditions are either true or false of them.

The framework we need is Lewis’s. According to him:

The idea is that the truth-value of an utterance depends upon circumstantial features such as a world, a time, a place, or whatever. And what fixes the relevant world, time, place, or whatever and makes evaluation possible is the so-called ‘context of utterance.’ The world, time, place, or whatever is the world, time, place, or whatever, of the context, that is, the world, time, place, or whatever in which the sentence is uttered. That is what Lewis above calls “contingency” or indexicality in the broad sense. However, as Lewis recognizes, some features of context of utterance are “shiftable.” ^{Footnote 40} In his own words:

Modal and temporal operators, for example, shift the point of evaluation: they shift the default point of evaluation originally fixed by the context of utterance. The sentence in the scope of a temporal operator is evaluated with respect to a time distinct from that of its utterance. For example, the sentence ‘there are dogs’ is true iff there are dogs right now at the time that the utterance is made. But the sentence ‘there have been dogs’ is true iff there were dogs at some time before that of the utterance. ^{Footnote 42} When the point of evaluation is shifted because some feature of the context has been shifted by an operator, the sentence in the scope of the operator must be evaluated with respect to what Lewis calls an ‘index’ (an n-tuple of features) that is different from the original context. ‘Index’ is what Lewis calls indexicality in the strict sense.

Thus, Lewis’s content-relativism has it that the content of a sentence is neither absolutely true nor false in all possible worlds, and hence is incomplete in light of Fregean standards. Instead, it is best understood as relative content, true or false in contexts and indices, modelled mathematically as complex functions from context-index pairs to truth-values. Let me provide a few examples. Sentences such as ‘there is a yellow cube straight ahead’ will be true with respect to a context c and an index i iff there is a yellow cube 1 in the place, time, and world of i. In contrast, the sentence with the temporal operator ‘there is a cup of tea yesterday’ will be true with respect to a context of utterance c and an index i iff there is a yellow cube 2 in the place of c one day before the utterance in the world of i. Likewise, the sentence ‘there is actually a yellow cube’ will be true in a context c and an index i iff there is a yellow cube 3 of tea at the place and time of i in the world of c (actual world).

Therefore, my proposal is as follows. First, in an analogy with natural language, we may think of c as what Tye calls “the context of experiential contact.” ^{Footnote 43} As before, the context of experiential contact is what fixes the default circumstances against which the relativistic content of experience should be evaluated. However, besides the world and the time, these include the subject S and the particular a that under normal conditions would be causally responsible for the relevant type experience e in subject S. For example, the content of token experience e of a yellow cube can be modelled as the function from context c to truth-values that takes the form of the content of a predicate, roughly a yellow cube. It will be true with respect to the default context of experiential contact c iff the particular straight ahead of S in the place, time, and actual world is yellow in colour and cubic in shape and is causally responsible for the token experience e. But following Lewis, we must also recognize that the default circumstances of evaluation can be shifted by means of operators, in our case the conditional ‘would be,’ Thus, the content of experience is best modelled as a complex function from context-index pairs to truth-values. It will be true in a context c and an index i iff the particular straight-ahead of S at the place and time of i in the possible world of c where that particular is normally causally responsible for the relevant token experience e is cubic in shape and yellow in colour.

To be sure, all of these particulars are concerned. Moreover, there is no doubt that S sees or directly experiences the particular a, rather than some a-like sense impression that stands as an intermediary between a and S. Still, I am suggesting that both these particulars and the causal relation between a and the particular token experience e are part of the circumstance in which the content is evaluated (determined by the context and index), rather than part of the representational content itself. S sees a by being perceptually acquainted with a at the time and location of i, as the relational view correctly predicts.

4 The Particularity and the Indistinguishability Desiderata

According to semantic particularism or the singular content view, in the case of veridical or falsidical singular contents, the particular singled out by the demonstrative (that particular straight ahead) instantiates or fails to instantiate the attributed properties of being cubic in shape and of being yellow in colour. While genuine perceptual experiences represent singular contents, hallucinations represent what both Tye and Schellenberg call ‘singular gappy contents.’ ^{Footnote 44} There are two different versions of gappy contents: a Russellian and a Fregean one. According to the Russellian version, a singular gappy content is a sequence consisting of a property and a hole where a particular should go. What perception and hallucinations have in common is the fact that they represent content schemas. ^{Footnote 45}

However, it is hard to see how a content schema could be considered as a genuine content that represents the world in a precise way. In this regard, Tye completely changed his mind a few years latter:

Now, someone could claim that Schellenberg’s Fregean gappy content fares better. According to her view, rather than the hole where the particular should go, what the gappy content displays is a de re mode of presentation. In her own words:

Now, de re modes of presentation present objects relationally rather than satisfactionally: the object is presented as the object that has a certain relation to the agent rather than one that meets identifying conditions. ^{Footnote 48} Yet, in the case of a hallucination where there is no RES (thing), there cannot be prima facie a de re mode of presentation. In perceptual cases, I am representing or misrepresenting that particular straight ahead of me as being a yellow cube while in hallucinatory cases I am representing nothing because I am not standing in relation to anything in the first place. To this Schellenberg replies:

Schellenberg’s position about de re modes of presentation traces back to Kent Bach’s famous position on the matter. According to Recanati’s testimony:

Given this, Schellenberg’s singular content view faces a dilemma. One horn is this. She assumes Bach’s view according to which de re modes of presentation belong to the singular content of experience. Thus, if I perceive yellow cube A or yellow cube B, or if I hallucinate a yellow cube in the same place, my experiences are representing the same Fregean content consisting of a de re concept of that particular straight ahead and of the mode of presentation of the property of being yellow in colour and of the property of being cubic in shape. However, the price to pay is high: giving up the semantic particularity, that is, the claim particulars as such constitute perceptual states by being represented by experience. For one thing, according to Bach, if we assume that all those perceptual experiences have the same content, we must also assume that this de re mode of presentation is “individuated narrowly, without mention of their objects.” ^{Footnote 51}

Schellenberg wishes to leave aside Evans and McDowell’s view of de re modes of presentation that makes these modes object-dependent. Instead, she wishes to endorse Bach’s alternative view. However, as Recanati remarks:

The only way to save semantic particularism is to assume Recanati’s view that this de re mode of presentation is truth-conditional irrelevant:

If what I said in the previous section is correct, it is a general fact about non-descriptive representations that they involve a relational, extrinsic condition of reference distinct from the intrinsic condition of satisfaction of the representation.

However, we are now back to the Russellian singular view proposed by Tye and all the problems raised by Tye are back. That is the other horn of the dilemma. If the de re modes of presentation are truth-conditional irrelevant, particulars as such belong to the singular content of experience rather than their de re modes of presentation. Given this, Schellenberg must give up the aim of meeting the indistinguishability desideratum; that is, she must give up any attempt to account for why those experiences are introspectively indistinguishable by means of the key notion of singular content. For one thing, as we have seen, Russellian gappy singular contents are not genuine contents in the first place. The only way that Schellenberg could account for the fact that those experiences are indistinguishable by introspection in terms of her semantic particularism is by reintroducing such de re modes of presentation to the content of experience.

Now we are able to see that there is something amiss in the third premise of Schellenberg’s singular content argument, namely the premise that states that “if S represents a (under a mode of presentation), then S’s perceptual state M is constituted by a.” ^{Footnote 54} If the de re mode of presentation is truth-relevant or belongs to the content of experience, the fact that it is the particular a or the particular b that is represented makes no difference to the content of experience. Again, Bach’s de re modes of presentation are types, not tokens. In contrast, if we assume that the de re mode of presentation is truth-irrelevant or does not belong to the content of experience, then S’s perceptual state M is constituted or token-individuated by the particular a. However, there is no way to show how the indistinguishability desideratum could be met.

Since particulars do not belong to the content of experience, they play no role in the determination of its phenomenology, which is determined by the represented properties alone, and the singular relational plus relativistic content view easily meets the indistinguishability desideratum. As the phenomenal character of S’s experience of a yellow cube straight ahead is determined by the properties of being yellow and of being cubic, nothing really changes in the phenomenology of e if the particular a that is causally responsible for the token experience e is replaced by another qualitatively identical one, a, as long as the content of the complex predicate remains unaltered. Likewise, since nothing hinges on the particular a that is causing the token experience e, veridical experiences and hallucinations in which a yellow cube appears to be straight ahead also share the same phenomenal character.

The singular relational plus relativistic content view can make room for the intuitive idea that generic hallucinations is contentful and shares the same content with illusory and non-illusory perceptual experiences. To be sure, the content of perceptual experience is less than a complete proposition by Fregean lights: absolutely veridical or falsidical. Rather, it is best understood as the content of a predicate (that can be modelled as a propositional function from context-index pair to truth-values). Thus, my experience of a yellow cube straight ahead represents the properties of being cubic and of being yellow that are veridical or falsidical only of the particular that would normally be causally responsible for the relevant token experience, determined in the context and index. The experience is veridical when the particular that would normally be causally responsible for the relevant token experience instances both properties. It is falsidical when the particular does not instantiate both properties. And it is hallucinatory whenever what is causally responsible for the relevant token experience is not the particular that under normal conditions would be causally responsible for it. In other words, I am endorsing Dretske’s view that hallucinations represent uninstantiated properties. ^{Footnote 55}

The singular relational plus relativistic content view can also easily meet the particularity desideratum. To be sure, S’s token experience e only represents the predicate yellow cube, rather than the particular object a, in its particular location L within S’s visual field, at time H, in which a would be causally responsible for token experience e under normal conditions. However, if perceptual experiences are type-individuated by the properties they represent, they are also token-individuated by particulars that actually bring them about. Take two token experiences, e and e’, of a yellow cube. They are different when one of the following two conditions is met: first, when another particular, a’, is causally responsible for e’, irrespective of the time and location at which the experience takes place. Second, when the same particular a is causally responsible for e’, but it is either located in a different place L’, or in the same location L but at a different time H’.

Moreover, my view can also account for the possibility of veridical misperception and veridical hallucination in a way that the singular content view cannot: that is, by avoiding the dissociation of the veridicality of the content from the illusory character of the experience. This dissociation only happens when we overlook the difference between the particular that is singled out as the one that is causally responsible for the actual token experience and the particular that would normally be responsible for the relevant token experience.