2. Two metaphysical approaches to interpreting transcendental idealism

First, Intentional Object Phenomenalism (IOP). In general, phenomenalist interpretations hold that the existence of empirical objects is metaphysically dependent on certain complex representational states in the minds of human subjects, and that these objects are entirely ontologically distinctFootnote ⁷ from the mind-independent entities that causally groundFootnote ⁸ those representational states. Put simply, the mind-dependent empirical object is one thing, the non-spatiotemporal ground of our representations is something else. According to what we might call ‘traditional’ or ‘ontological’ or ‘Berkeleyan’ phenomenalism, empirical objects are identified with mental entities in the mind of the perceiver, usually organized collections of sensations.Footnote ⁹ On this view, physical bodies are, in the final analysis, composed of mental stuff. This sort of interpretation is out of favour for a variety of reasons,Footnote ¹⁰ and I will not be concerned with it in this paper. However, a different sort of phenomenalism remains a viable option for many (myself included). According to IOP, empirical objects are identified not with representational states themselves, but rather with the intentional objects of those representational states. That is, they are the contents of our cognitions. They are what our experience refers to and purports to be about. Now, obviously, no parties to the debate will deny that our experience is about and refers to empirical objects. What makes IOP a phenomenalist view is the additional claim that empirical objects are mere intentional objects: they do not have any self-standing ontological status in their own right, and they are numerically distinct from whatever non-spatiotemporal, mind-independent thing caused the representations of those objects. As Sellars puts it, ‘physical objects and events exist only “in” certain actual and obtainable conceptual representings’ (Reference Sellars1968: 48). Van Cleve puts the point in terms of empirical objects as ‘virtual objects’ (Reference Van Cleve1999: 8). And Jaurenig describes them in terms of ‘presentational contents’ (Reference Jauernig2021: 35), which are ‘representation-immanent’ (p. 33).Footnote ¹¹

Next, Metaphysical Two-Aspect readings (MTA). These readings agree with the phenomenalist that the distinction between appearance and thing in itself is metaphysical (not merely epistemological or methodologicalFootnote ¹²). But they deny that it is an ontological distinction between two numerically distinct entities. Instead, they claim that there are things that causally impinge on the senses and provoke sensations and other representations in us, that the mind-independent nature of these things is not spatiotemporal (and is otherwise not cognizable by us), but that these same things also have relational, mind-dependent, spatiotemporal properties that we can cognize. Rosefeldt (Reference Rosefeldt, Schafer and Stang2022 and Reference Rosefeldt and Stolzenberg2007) describes them as ‘response-dependent properties’; Marshall (Reference Marshall2013) calls them properties of ‘things-qua-appearances’; Allais (Reference Allais2015) calls them ‘essentially manifest qualities’; and Chignell (Reference Chignell, Schafer and Stang2022) calls them ‘phenomenal features’. While there are important differences between these readings (some of which I will discuss below), they all share a commitment to the claim that the objects of experience are ultimately the mind-independent causes of our experience.

On the face of it, the difference between these views seems quite drastic. However, if we step back, we see that the two approaches share a great deal in common. Specifically, they both agree on the following three points:

1. (1) Kant’s thesis about the non-spatiality of things in themselves is intended as a substantive metaphysical claim: there are things that exist independent of us and our cognition, and these things (a) do not have spatiotemporality as mind-independent properties, but (b) do have other mind-independent properties and causal powers which are unknowable by minds like ours.

2. (2) Due to the causal influenceFootnote ¹³ of these things on the faculty of sensibility, sensations and empirical intuitionsFootnote ¹⁴ occur in the mind.

3. (3) These sensations and intuitions lead, after various stages of cognitive processing, to representations of bodies with determinate spatiotemporal properties and locations, and other properties which depend on spatiotemporality, e.g., colour, mass, or motive force.

Put together, we have a story wherein non-spatiotemporal things provoke the cognitive machinery into action, the end result of which is experience that represents spatiotemporal bodies. Where IOP and MTA part ways is when we ask the further question: are the represented bodies mentioned in (3) numerically identical to (some aspect of) the mind-independent things mentioned in (1) and (2)? IOP says no, MTA says yes. Notice, then, that there is a natural way to state the disagreement between these views in terms of a question about intentionality: At what are our representations aimed? What is their intentional object? Where IOP is happy to end the story with (1) – (3), and claim that the (mere) intentional object is one thing, the cause of the representation of it is something else, MTA makes an additional claim:

1. (4) The representation of a body is intentionally directed at the mind-independent thing(s) that causally ground(s) the representation.

How might the proponent of MTA go about justifying (4)? One approach would be simply to take the exegetical route and point to all of the passages where Kant seems to endorse a one-object view. But this would be unsatisfying for two reasons. First, as noted above, Kant is not univocal on this issue, and the many one-object passages are counterbalanced by at least as many two-object passages; the phenomenalist could justly claim that the exegetical contest comes out a draw. Second, and more to the point, simply noting that Kant occasionally endorses (4) would still leave open the question why he would be justified in affirming it, given his claims about the radical differences between things in themselves and the objects of experience. Thus, the proponent of MTA owes us an account of why an affirmation of (4) as either an addition to or consequence of (1) – (3) would be plausible within Kant’s system. Since (4) is a claim about the intentionality of experience, giving such a justification will require spelling out what exactly establishes the (supposed) intentional connection between the representation and its mind-independent cause.

In sections 4 and 5, I will consider two different ways – a representationalist approach and a direct realist approach – that MTA might go about justifying (4). First though, in the next section, I want to make the problem more vivid and put MTA squarely on the defensive by considering how the situation looks from IOP’s perspective.

3. The mismatch argument

Here is a blunt way to put the question at issue. Consider some object, call it x, which possesses properties i ₁, i ₂, etc. as mind-independent, intrinsic, ‘in itself’ properties, but does not possess properties a ₁, a ₂, etc. as mind-independent properties.Footnote ¹⁵ In virtue of x’s causal influence on the subject, a representation is formed that represents some y as possessing properties a ₁, a ₂, etc., but not i ₁, i ₂, etc. Under what conditions, if any, is it reasonable to say that y is numerically identical to x (or at least some aspect of x), and thus that the intentional object of the representation is x, and perhaps even that x possesses a ₁, a ₂, etc. as ‘response-dependent’ or ‘essentially manifest’ properties? Note that one should not hope to get such intentional directedness too cheaply (unless one accepts a very watered-down theory of intentionality; more on this in section 4). Imagine (abstracting momentarily from Kantian metaphysics) that x is a hallucinogenic mushroom, and the y that I subsequently represent is a talking cat. It would be highly counterintuitive to claim that my experience of the cat is really an experience of the mushroom, such that the mushroom is my intentional object, and I am representing it as a talking cat. Rather, it seems much more natural to claim that the talking cat is a mere intentional object. Claiming that ‘talking-catness’ is a response dependent, essentially manifest, qua-object, or phenomenal feature of the mushroom (à la Rosefeldt, Allais, Marshall, or Chignell, respectively) would not do much to change this impression. But from IOP’s perspective, an equally implausible claim is being made when MTA declares that my experience of a body in space and time is really an experience of something whose mind-independent, non-spatiotemporal nature bears no resemblance whatsoever to what my representation purports to be about.

At issue here is the intuition that intentional contact between representation and object requires that the representation’s content achieve some minimal threshold of accuracy with respect to that object. As Michelle Montague expresses (and endorses) the idea, ‘for a perceptual experience to be about an object, there must be a certain degree of match between the properties an object has and the properties the experience represents the object as having’ (Reference Montague and Kriegel2013: 29). In the context of a more traditional, common-sense realism, this accuracy is easy to come by. I may misperceive the wineberry as a raspberry, but my representation is still correct with respect to its colour, size, position relative to me, and even my approximate expectation of its taste. And even if the colour and taste, as secondary qualities, are not features of the berry ‘in itself’, its size and position are. The fact that some non-trivial, mind-independent features of the object are accurately represented makes it seem natural that we should identify the berry as the intentional object of the representation, even though the representation is not entirely accurate. But in the context of Kant’s idealism, where none of the non-trivial, mind-independent features of the object are represented, the situation starts to look closer to the mushroom/talking cat case than the wineberry/raspberry case. We can formulate this point as an argument for IOP and against MTA.

I think this argument is basically sound. The proponent of MTA will not be so easily convinced, however, and there are at least two possible responses to the argument that are worth close consideration. One option is to deny M2 and claim that our representations do represent some non-trivial features that the object has in itself. Chignell (Reference Chignell, Schafer and Stang2022) takes this route, with his appeal to what he calls ‘straddlers’ of the phenomenal/noumenal divide. I will address this response in section 6. First though, I want to take some time considering what I take to be the more likely response: that M1 is false because it sets too high a standard for intentional directedness.Footnote ¹⁷

According to MTA, although our representations of things do not reflect their mind-independent natures, their mind-dependent or subject-relational nature can accurately be reflected in the representation. For again, according to these views, response dependent, essentially manifest, etc. properties may not be properties of things considered independent of their relation to perceivers, but they are (a) properties that things really have and (b) what our representations represent their objects as having. Thus, MTA will want to replace M1 above with something like:

With M1* instead of M1, the Mismatch Argument does not go through.

To motivate the idea that it is possible to accurately represent a mind-independent object by way of representing its mind-dependent properties, appeal is sometimes made to an analogy with secondary qualities like colour. On some theories, colours are treated as properties that mind-independent objects really possess, albeit not in themselves, and only in relation to a certain class of perceiver. Rosefeldt (Reference Rosefeldt, Schafer and Stang2022: 19) discusses this analogy with colour in dispostionalist terms, and Allais puts it in terms of her direct realism (Reference Allais2015: 103ff.).Footnote ¹⁸ I will have more to say about the important differences between these two approaches shortly. But in both cases, their goal is to motivate the idea that there is nothing terribly unusual about accurately representing an object as having a property that it does not have in itself. According to those theories of colour, the wineberry may not be red in itself, but it is – really is, not just apparently – red in relation to perceivers like me. Likewise, according to MTA’s picture of Kant’s metaphysics, the cause of my representation may not be spatiotemporal in itself, but it is – really – spatiotemporal in relation to perceivers like me. Thus, there is no mismatch after all. The object is more or less just as my experience represents it to be.

I think that this analogy with colour is misleading, since it ignores a major difference between the two cases. For as I hinted above in the wineberry example, I never represent an object by way of colour alone. Rather, whenever I represent something’s colour in experience, I also represent (at least) the shape of the coloured surface and its location. In the context of non-Kantian, common-sense realism, the latter are properties that the object has in itself. And one might reasonably think that my representation of colour qualities is intentionally directed at the object only because that representation is bound up with a representation of properties that the object has in itself. But this is not what is going on in MTA’s version of transcendental idealism. For at the core of these interpretations is the claim that we represent objects only by way of their mind-dependent properties (see Allais (Reference Allais2015: 125) on this point).

MTA’s appeal to the analogy with colour makes it seem like all that MTA needs to justify M1* is this:

With caveats, I have no problem with this claim. However, what MTA really needs is this much stronger claim:

There seem to be two possible ways that MTA might go about justifying I2.

First, one could be a representationalist who adopts a causal theory of intentionality. According to this view, experience can be exhaustively characterized in terms of the representational contents that constitute the experience, and the intentional object of these contents is whatever object causally provoked them.Footnote ¹⁹ Although he does not thematize it as explicitly as I am about to, this would seem to be Rosefeldt’s approach. He develops the idea that ‘Objects appear to us by causally affecting our mind and bringing about certain representational effects in us’ (Reference Rosefeldt, Schafer and Stang2022: 26). These ‘effects’ are representations of various spatiotemporal properties construed as ‘response-dependent properties’ of the mind-independent causes of these representations (pp. 25ff.). On this view, intentional directedness to the mind-independent object is secured by its causal connection to the representation. This would justify I2 because this kind of intentional contact does not require that the ‘representational effects’ represent properties that their cause has in itself.

Second, one could be a direct realist who appeals to an ‘acquaintance’ relation as the ground of intentional directedness. According to this view, experience (specifically, empirical intuition) involves a direct and immediate awareness of the mind-independent objects that causally affect us. This primitive acquaintance relation is not itself representational (it has no content), but it secures the conscious link with the object, such that experience as a whole refers to the mind-independent object presented in intuition. This is Allais’s stance. She holds that the ‘essentially manifest qualities’ of the mind-independent things that causally affect us are immediately ‘present to consciousness’ in intuition (Reference Allais2015: 143-4).Footnote ²⁰ This approach would then justify I2 by claiming that the intuition’s acquaintance relation does not require that we be acquainted with properties that the object of acquaintance has in itself.

The next two sections critically appraise these two options. My goal is not so much to prove them wrong as to take the teeth out of the motivation for accepting them. Regarding the representationalist approach (section 4), I will argue that, when properly unpacked, the view ends up indistinguishable in all but description from the core claims of IOP. And regarding the direct realist approach (section 5), I will argue that it stretches the notion of ‘acquaintance’ far beyond how contemporary (non-transcendental idealist) philosophers use it, and perhaps so far as to render the notion either too thin or too mysterious.

4. The representationalist approach

To recap, the challenge to the proponent of MTA is to tell us a story about why we should identify the intentional objects of our representations with their mind-independent causes, given that there is a severe mismatch between the ‘in itself’ nature of the causes on the one hand, and the properties represented in empirical objects on the other. The representationalist can attempt to justify this identification by appeal to a causal theory of intentionality, according to which, in general, our representations refer to their mind-independent causes. To see how exactly this works, let us consider Rosefeldt’s version of MTA more closely.

On Rosefeldt’s reading, Kant’s distinction between appearances and things in themselves turns on a distinction between ‘response-dependent’ and ‘response-independent’ properties (Reference Rosefeldt, Schafer and Stang2022: 25ff.). A response-independent property is a property that an object has ‘in itself’, while a response-dependent property is a property that an object has only in virtue of its effect on something else (in this case, the human subject). Rosefeldt’s proposal is that we should understand the ideality of spatiotemporal properties in such terms: they are response-dependent properties of mind-independent things; thus, cognitions of spatiotemporal things are simultaneously cognitions of mind-independent things (just not of those things as they are in themselves).

Let us consider a simple example to see how this works. Say I am veridically perceiving a red cube (this is Rosefeldt’s stock example). On his analysis, there is some mind-independent, in-itself-non-spatiotemporal thing that, under the right circumstances, causes me to have sensations of a certain type (call them r-sensations), which are given a certain spatiotemporal orderingFootnote ²¹ in intuition, and which lead to a mental content that represents something as a red cube.Footnote ²² Rosefeldt argues that we should identify the red-cubicness of the represented object with a response-dependent property of the mind-independent thing (viz., the property of being the kind of thing that has the response-independent causal power to cause in perceivers like me representations of ‘redness’ and ‘cubicness’). Thus, on this view, even though the mind-independent thing that causes the representation is neither red nor cubic in itself, it nevertheless has redness and cubicness as response-dependent properties, which is what allows Rosefeldt’s Kant to say that my representation of something red and cubic is simultaneously a representation of that mind-independent cause.

A couple of observations about this analysis: First, note that on this reading the response-dependent properties are not doing any of the causal work. The causal powers of the thing’s unknowable, response-independent properties – the thing as it is in itself – are what provoke the sensations that fuel experience. The object only has its response-dependent spatiotemporal properties at all because of the causal work performed by the response-independent properties. Thus, the ‘cubicness’ of the object is in no way causally responsible for my representation of it as cubic.Footnote ²³

Second, and more importantly, note that Rosefeldt’s analysis involves two distinct moves (though they tend to get blurred in his presentation of the view): (1) pointing out that mind-independent things have these causally inert, response-dependent properties and then (2) claiming that we should identify the spatiotemporal properties that we experience with those response-dependent properties. Rosefeldt directs most of his attention to articulating the idea behind (1), but this obscures the fact that the identification made in (2) is the more important and more contentious claim. In fact, point (1) seems to me completely innocuous, and I take it that it follows directly from the points that IOP and MTA both explicitly agree on. They both claim that some unknowable features of mind-independent things cause us to have sensory representations of various sorts, which entails that those mind-independent things have the response-dependent property of being the kinds of things that cause those kinds of representations. Thus, the proponent of IOP should see nothing objectionable about granting that mind-independent things have these response-dependent properties. The second move, however, seems open to scepticism, and this is where MTA and IOP part ways.

Let us consider what’s going on here in more detail. Let x be the mind-independent thing(s) that are causally responsible for my representation of the red cube. Here are the facts about this case that both parties accept:

1. 1. x has a response-independent causal power (call it P) which can provoke r-sensations in certain cubic spatiotemporal orderingsFootnote ²⁴ in perceivers like me.

2. 2. In virtue of having P, x also has the response-dependent property (call it R) of ‘being-the-kind-of-thing-that-causes-r-sensations-in-such-orderings-in-perceivers-like-me’.

3. 3. In virtue of having those sensory states, I represent something (call it y for nowFootnote ²⁵) as a red cube.

As we saw in Section 2, IOP is content to end the story with these first three claims, but Rosefeldt’s MTA will want to claim further that:

1. 4. (i) The represented y mentioned in 3 is numerically identical to the x mentioned in 1 and 2 (or at least some aspect of x), and (ii) the red-cubicness of y is identical to R.

As indicated, I take it that these identifications are meant to follow from an implicit assumption of a causal theory of intentionality, according to which, in general:

4(i) follows immediately from 1, 3, and CTI. And 4(ii) follows from 4(i) and 3: if I represent y as a red cube, and y is x, then I am also representing x as a red cube; and if x cannot have red-cubicness as a response-independent property (per Kant’s idealism), then it must be a response-dependent property. I take it that this sort of inference is what underlies Rosefeldt’s general claim that:

This endorsement of CTI and RD is what make Rosefeldt’s MTA a one-object reading. These claims thus seem to mark a major disagreement with IOP. However, I think that the proper response from IOP is not so much that CTI and RD are wrong. Rather, when properly unpacked, they are innocuous and mark no substantive (that is, metaphysical) divergence from the commitments that IOP is happy to embrace.

Take CTI first: ‘The thing that caused a representation is the intentional object of the representation’. There is a weaker and a stronger way to read this claim, depending on whether we take the intentional relation established by the causal relation to be something over and above the causal relation. According to a stronger, ‘non-reductivist’ reading, the intentional relation is something above and beyond the causal relation, a new item in the metaphysical architecture, an ‘intentional laser beam’ shooting back at the object, a relation parallel to and dependent on – but nevertheless metaphysically distinct from – the causal relation. And according to a weaker, ‘reductivist’ reading, the intentional relation just reduces to the causal relation, such that, metaphysically speaking, no more is being said by ‘x is the intentional object of r’ than ‘x caused r’.Footnote ²⁶ On this weaker version, there are no intentional laser beams, only causal relations.Footnote ²⁷

Do we have reason to attribute either of these positions to Kant? In one sense, not really, since he says very few things that could be taken as direct endorsements of a causal theory of intentionality in the first place. More often he seems to describe (what we call) intentionality in terms of a certain kind of relationship that obtains between representational contents: representations relate to objects when they ‘necessarily agree with each other’ (A105); the relation to an object ‘is nothing other than the necessary unity of consciousness [in the] synthesis of the manifold’ (A109); ‘the unity of consciousness is that which alone constitutes the relation of representations to an object’ (B137); ‘relation to an object [results from] making the combination of representations necessary in a certain way, and subjecting them to a rule’ (A197/B242). These remarks strongly suggest that Kant does not think of intentionality in primarily causal terms.

That being said, there are a couple of places where Kant does seem to flirt with the idea that the notion of ‘relation to an object’ (he never uses the term ‘intentionality’ in our sense) can be cashed out in causal terms. In the famous 1772 letter to Herz, he asks about ‘the relation of that in us which we call “representation” to the object’, and grants that,

The same point is put in terms of the object ‘making the representation possible’ in the lead-up to the Transcendental Deduction:

Now, in both passages, Kant goes on to make clear that this causal relation is not the kind of ‘relation to an object’ that he takes to be most philosophically interesting, and this alone should call into serious question an attribution of a causal theory of intentionality to Kant. However, let us give the benefit of the doubt and grant for argument’s sake that Kant is open to treating the causal relation between a representation and its cause as an intentional relation. Would it then be more plausible to attribute the reductivist or the non-reductivist version to Kant? I think it is clear that there is nowhere near enough evidence to attribute the stronger claim to him. For the stronger (non-reductivst) claim involves affirming the existence of a new metaphysical relation between representation and its mind-independent cause: in addition to the causal relation, there is this other thing, the ‘intentional laser beam’. If Kant actually meant to affirm such a claim, we would surely expect him to thematize it and explain it explicitly (in at least as much detail as he articulates the more internalist conception of intentionality discussed in the previous paragraph). But we don’t see that. Instead, in the letter to Herz, he says that a representation ‘is in conformity with its object … as an effect accords with its cause’ (Br, 10: 130), which is a weaker claim than that the conformity depends on or arises out of this ‘accord’, as something distinct from it. And in the Deduction passage, the claim is that the object’s ‘mak[ing] the representation possible’ is how the two ‘come together’, ‘relate to each other’, and ‘meet’ (A92/B124), but Kant does not suggest that this ‘coming together’ is anything above and beyond the ‘making possible’ (i.e., ‘causal’ or ‘grounding’) relation. In both passages then, the claim is that the representation relates to the object in the sense that the latter causes the former.

If these considerations are on the right track, and if it is reasonable to attribute a causal theory of intentionality to Kant at all, then only the weaker, reductivist variant would be warranted. And once again, on that view, the claim that ‘x is the intentional object of r’ says nothing more than that ‘x caused r’. And this is not a claim that IOP is bothered by! IOP’s proponent may find it unnatural and counterintuitive to describe the causal relation as an intentional relation. But arguing over this would be arguing over the right to use a term, and that does not seem to me a war worth waging.

Next, RD: ‘Spatiotemporal properties are response-dependent properties of mind-independent things’. On the one hand, Rosefeldt’s MTA claims explicitly that mind-independent things have spatiotemporal properties as response-dependent properties. The mind-independent cause of my representation may not be red and cubic in itself, but it is red and cubic in relation to perceivers like me. Once more, this claim appears to mark a serious divergence from IOP’s take on things. But it is important to be clear about what exactly this claim is and is not saying. For it is not as though bringing human beings onto the scene somehow literally inflates a heretofore non-spatial reality. This would make us into some strange kind of gods. Rather, the response-dependent property is just the property of being the kind of thing that has the (response-independent) power to cause certain representational effects in me. Thus, the claim that spatiotemporal properties are response-dependent properties is just the claim that mind-independent things are the kind of thing that have the power to cause representations whose contents describe objects with various three-dimensional features. And so, the claim that the mind-independent cause is a red cube is just the claim that the object is the kind of thing that has the power to cause representations of redness and cubicness in me.Footnote ²⁸ But once again, this is not a claim that IOP needs to dispute, since we cede no ground by agreeing that mind-independent things have these response-dependent properties. We may think it is rather odd to identify spatiotemporal properties with response-dependent properties (as in 4(ii) above). But we are happy to grant that the latter properties are real.

Summarizing, the challenge posed to MTA asked how it is possible for our representations to refer to their mind-independent causes, given the severe mismatch between those causes and what our representations seem to represent. The representationalist proponent of MTA appeals to the causal relationship to argue that our representations refer to their causes, and that we represent these causes as the bearers of spatiotemporal properties. I have argued that, when fully unpacked, this view is coherent, but relatively toothless, at least from IOP’s perspective. In the next section, we will consider how the direct realist attempts to secure the intentional relation to the mind-independent thing.

5. The direct realist approach

Lucy Allais has recently argued for a non-conceptualist reading of empirical intuition. According to her (Reference Allais2015: 147ff.), the ‘singularity’ and ‘immediacy’ that Kant attributes to intuition (A320/B377) are best explained by a relationalist or direct realist model, where an intuition involves a direct ‘presence to consciousness’ of or ‘acquaintance’ with the mind-independent object. Allais also takes the text of the first Critique to describe a one-object/two-aspect metaphysics. Thus on her view, the objects with which intuition acquaints us are the mind-independent causes of our sensations. Given her commitment to a metaphysical reading of the non-spatiotemporality of things in themselves (see Ch. 4), she thus needs to reconcile her commitment to direct realism about intuitions with Kant’s doctrine of our noumenal ignorance of things in themselves. She threads this needle by appeal to what she calls ‘essentially manifest qualities’ (Ch. 5), which are properties that things have only in their perceptual relation to minds like ours.

Clearly, essentially manifest qualities play a role analogous to Rosefeldt’s response-dependent properties. Where the views differ is in our cognitive access to them.Footnote ²⁹ While Rosefeldt thinks that our representations are intentionally directed at mind-independent things, he does not think those things enter into any kind of phenomenological contact with the mind. Moreover, as I argued in the previous section, the intentional-directedness in that account does not amount to much more than a redescription of the causal relation between mind-independent things and our representations. Does Allais’s appeal to an acquaintance relation provide a more substantive kind of intentional directedness (of a sort that would mark a more meaningful difference between IOP and MTA than we’ve seen so far)?

To answer this question, we need to get more precise on what exactly this acquaintance relation is supposed to be. I take it as obvious that acquaintance with empirical objectsFootnote ³⁰ depends on a causal relation: one must be affected by the object in order to be acquainted with it. However, unlike the theory of intentionality discussed in the previous section, it is clear that Allais does not want to reduce acquaintance to a mere causal relation: ‘acquaintance with objects (not merely being affected by them) is an essential ingredient in cognition’ (Reference Allais2015: 146). Consequently, it seems like the direct realist does want to introduce the acquaintance relation as an additional piece of the metaphysical architecture. So what exactly is this ‘presence to consciousness’ relation, such that it depends on but does not reduce to causal contact with the object?

In the literature on contemporary theories of acquaintance, this is a notoriously tricky question. Most acquaintance theorists take acquaintance to be a brute, primitive, unanalysable notion.Footnote ³¹ It is easy enough to gesture at what they’re getting at by appealing to the sense of ‘right-thereness’ manifest in our experience of objects. But describing this sense of immediacy as ‘acquaintance’ or ‘presence to consciousness’ remains rather elliptical and even metaphorical. Moreover, as others have pointed out, the sense of immediacy in experience can be accounted for in purely phenomenological terms as a feature of all experiences that have objective purport, irrespective of whether they are actually related to an external object (see Frey Reference Frey and Kriegel2013 on this point). The direct realist, however, needs a notion of acquaintance that is not just phenomenological but also metaphysical; that is, their ‘presence’ must involve a real contact between mind and object, not just the phenomenological purport of one. And again, contemporary acquaintance theorists typically find themselves at a loss for words when pressed on what this ‘contact’ ultimately amounts to.

There is, however, one important hallmark of acquaintance that many contemporary direct realists want to highlight. I will follow Brewer (Reference Brewer, Knowles and Raleigh2019: 279) in calling it ‘revelation’, which is meant to capture the idea that, ‘at least in the most basic cases, a visual experience whose conscious character is specified by its looking to the subject as though there is an F before her is revelatory of what being F is’. In other words, acquaintance (if there really is such a thing) is important and interesting because it reveals the mind-independent nature of the perceived object. This seems to be a fairly mainstream view in the literature. For instance, Balog (Reference Balog, Gozzano and Hill2012: 17) argues that acquaintance gives us ‘direct, unmediated, substantial insight into [the] nature’ of things. Logue argues that an acquaintance theory allows one to say that ‘the phenomenal character of veridical experience gives its subject insight into what things in one’s environment are like independently of one’s experience of them’ (Logue Reference Logue2012: 227) and that acquaintance ‘reveals quite a bit about [the object’s] intrinsic nature’ (p. 229) to the point that ‘the phenomenal character of experience can literally resemble properties of the object of experience’ (ibid.). Raleigh (Reference Raleigh, Knowles and Raleigh2019: 24) describes acquaintance as a ‘source of knowledge about the intrinsic/categorical nature of mind-independent properties’. Goff (Reference Goff, Coates and Coleman2015: 122) argues that ‘real acquaintance … affords the acquainted an understanding of the real nature of the thing with which she is acquainted’ such that, through acquaintance, the nature of the qualities of experience are ‘laid bare’. And Coleman (Reference Coleman, Knowles and Raleigh2019: 54) defends a position according to which acquaintance does not give us complete access to all of an object’s intrinsic properties, but it does give us some of them.

In other words, contemporary direct realists are transcendental realists, not transcendental idealists, and they think that knowledge of ‘things in themselves’ is not as hard to come by as Kant believed. Thus, although they have trouble spelling out what exactly the acquaintance relation amounts to, metaphysically speaking, they can at least say that, if acquaintance is real at all, it is able to explain our access to some of the intrinsic, mind-independent, ‘in itself’ natures of the objects of experience. But this is precisely what the Kantian direct realist must exclude from the story, since they must claim that we have no access to how things are in themselves. What, then, is left to give substance and meaning to the notion of acquaintance in the Kantian direct realist picture?

Perhaps the Kantian direct realist would respond by suggesting a weakened version of revelation in the Kantian picture: acquaintance does not reveal how things are in themselves, but it does reveal their ‘essentially manifest qualities’, and these qualities really are properties possessed by mind-independent things. However, if this claim is meant to retain the same spirit as the original notion of revelation, then the Kantian direct realist would need to claim that acquaintance does not simply present us with essentially manifest qualities; rather, acquaintance must reveal the true, intrinsic nature of these essentially manifest qualities. But clearly that does not happen. Perceptual phenomenology does not present spatiotemporal properties as ‘essentially manifest’. When I perceive the spatiotemporal properties of objects, my experience does not present them as mind-dependent properties, which are grounded in some deeper, cognitively inaccessible, in-itself-non-spatiotemporal, intrinsic reality. Rather, common-sense, quotidian, human phenomenology naively presents spatiotemporal properties (and even many secondary qualities, e.g., colour) as mind-independent features that objects have in themselves, even when no one is looking at them. The fact that it took a book like Critique of Pure Reason to draw humanity’s attention to the non-spatiotemporality of things in themselves is all the proof we need of this.

Thus, the Kantian direct realist is appealing to a notion of acquaintance that cannot give us insight into the intrinsic natures of the mind-independent properties of mind-independent things, and it does not even give us insight into the intrinsic natures of the mind-dependent properties of mind-independent things. And thus the one hallmark that contemporary direct realists appeal to when fleshing out the notion of acquaintance is unavailable to the Kantian direct realist, and it consequently becomes very difficult to discern what exactly is being claimed when they assert that intuition acquaints the subject with essentially manifest qualities of the mind-independent object. For say we grant that, in addition to the causal ‘contact’ between subject and object, there is this second ‘contact’ that we can refer to as ‘K-acquaintance’ (to contrast it with the contemporary notion, which involves revelation of intrinsic natures). If K-acquaintance, so construed, gives no access whatsoever to the mind-independent nature of things, then it is not clear why it will have an easier time underwriting intentional relations than the causal relation did. We saw in the previous section that it is open to the representationalist version of MTA to simply stipulate a conception of intentionality according to which causal contact is sufficient for intentional reference. But IOP replies that this stipulative definition does not change the facts on the ground about the severe mismatch between representation and its cause. Likewise, a direct realist MTA could stipulate a conception of intentionality according to which intentionality is established by K-acquaintance. But if K-acquaintance does not even give us insight into the natures of the mind-dependent (‘essentially manifest’) properties of things, much less the things themselves, then once more, the severe mismatch remains. The forms and qualities that are presented in perceptual phenomenology are completely unlike the things with which we are K-acquainted. Thus, IOP contends, even if we grant that K-acquaintance is a real part of the Kantian story, it is not clear why this should lead us to infer that there is any meaningful or robust sense in which K-acquaintance establishes an intentional relation to the mind-independent object.

6. Straddlers to the rescue?

One of the premises of the Mismatch Argument (presented in section 2) is:

M2 presupposes that all of the non-trivial features that I represent in spatiotemporal bodies are either spatiotemporal features or features that depend directly on spatiotemporality (such as colour, mass, or motive force). Chignell’s MTA calls this assumption into question. On his reading, there is an important distinction between the ‘noumenal’ properties that things have in themselves and the ‘phenomenal’ properties that things only appear to have. But he thinks that there are also ‘straddlers’ of the noumenal/phenomenal gulf. These are features that ‘things both have and appear to us to have’; among these, he includes ‘existence, being a substance, being in dependence relations with other things, [and] being the ground of thus-and-such perceptions in us’ (Chignell Reference Chignell, Schafer and Stang2022: 340). Perhaps MTA can appeal to these straddlers in order to argue that the mismatch between representation and its mind-independent cause is not as drastic as M2 suggests. The mind-independent thing that causes my representation of a red cube may not be red nor cubic in itself; but it both is and is represented to be an existing substance in dependence relations with other things and which is the ground of my perception of the red cube. MTA can thus claim that there is a ‘match’ between the represented content and the mind-independent nature of the cause of my representation after all and that this match can ground the intentional link that is necessary for describing Kant’s metaphysics in terms of a ‘one-object’ picture.

The success of this reply depends on how much of a non-trivial match between content and cause is sufficient to justify claiming an intentional relation between the two. I am sceptical about whether Chignell’s straddlers are sufficient, though I grant that whether one shares my scepticism may ultimately be rooted in brute difference in philosophical intuition. That being said, in order to motivate the reader to share my scepticism, I appeal to the well-worn scenario about a brain in a vat. Say such a brain is fed experiences via electric stimulation from an attached supercomputer. Say that the computer regularly causes experiences that are neurally and phenomenologically indistinguishable from what goes on when one of us (non-envatted humans) plucks an apple from a tree and takes a bite (and let us assume that no actual apples were involved in the causal history of this story; it all happened through a cosmic accident of atoms randomly coming together in the void, ‘Boltzmann brain’ style). In this case, the vat-brain will be representing something as a round, approximately 100g, crisp, edible, sweet, etc. apple. The computer, being nothing like an apple, possesses none of these properties. However, the brain also represents the apple as an existing substance in causal relations to other things, including its perceptual system; moreover, it represents the apple as the cause of the representation of the apple. And indeed, the computer is an existing substance in causal relations to other things, including my perceptual system, and the computer does in fact cause my representation of the apple. Thus, as with Chignell’s Kant, there are straddling features: features that the cause of the representation has in itself, and that the represented object is represented as possessing. But is this sufficient for saying that the brain’s representation of the apple is really a representation of the computer? Is it natural and intuitive to describe the inedible computer as possessing ‘appleness’ as a response-dependent or essentially manifest feature? Perhaps some philosophers will be happy to bite this bullet.Footnote ³² But it seems to me to stretch the notion of intentionality far beyond recognition to claim that the representation of the apple is intentionally directed at the computer. And it seems far more natural to describe the represented apple as a mere intentional object. And thus, if the straddling features shared by the represented apple and the computer are not sufficient to underwrite an intentional link between representation and its cause, then those same kinds of straddling features shared by the represented red cube and the in-itself-non-spatiotemporal cause of that representation are not sufficient to underwrite an intentional link there either. Thus, I do not think that the appeal to straddling features gets MTA out of the mismatch problem.Footnote ³³