2. Varieties of Phenomenalism

The purpose of this section is to clarify and distinguish a cluster of philosophical theses that fall under the umbrella of ‘phenomenalism’.Footnote ⁶ I begin with phenomenalism stated in the most general terms.

Although this should sound like a familiar and uncontroversial characterization of phenomenalism, depending on how one takes the important terms – ‘objects’, ‘experience’, ‘representations’ – this thesis could be construed in several different ways.

Perhaps the most natural reading takes ‘objects of experience’ to refer to physical bodies (tables and mountains, etc.) and ‘the subject’s own representations’ to refer to collections of sensations (or perhaps some other kind of mental state, though I will focus on sensations) in the mind of the perceiver. This would yield a strong metaphysical thesis regarding the ontological status of bodies:

Ontological Phenomenalism may well be the first thing many philosophers think of as ‘phenomenalism’. However, when one surveys the history of modern philosophy through the twentieth century, one frequently finds phenomenalism articulated and defended with semantic or epistemic claims about the nature of experience. If we construe ‘experience’ in the original characterization of phenomenalism as either a semantic notion (experience as intentional-directedness towards objects) or an epistemic notion (experience as knowledge of objects), we get the following:

Semantic Phenomenalism is a claim about the proper analysis of what we are (and take ourselves to be) thinking about when we think about the objects encountered in experience: a proper analysis of the content of our thoughts about objects reveals those thoughts to refer only internally to the mind’s own states. What I really mean when I refer to ‘this mug’ is something like ‘this grey, cylindrical-looking collection of sensations’. Epistemic Phenomenalism is a claim about the epistemic status of experience of objects: to have knowledge of an object is to have knowledge of mental states in the mind of the knower.

The relation of Semantic and Epistemic Phenomenalisms to Ontological Phenomenalism is natural in many contexts. If one thinks that the mind can make reference only to its own mental states (Semantic Phenomenalism), and one accepts the common-sense assumption that we do make successful reference to physical bodies, the consequence is a commitment to Ontological Phenomenalism. (We will see in the next section that Berkeley makes this move.) Likewise with Epistemic Phenomenalism and the assumption that we do have knowledge of physical bodies.

Weaker versions of Semantic and Epistemic Phenomenalism were endemic throughout the modern period. Without insisting that reference to or knowledge of physical bodies consists in reference to or knowledge of mental states, many claimed that reference to or knowledge of bodies depends on reference to or knowledge of mental states. On these views, only mental states can be the immediate objects of reference or knowledge.

According to Weak Semantic Phenomenalism, reference to physical bodies is never immediate, requiring instead mental states functioning as semantic intermediaries. Likewise, Weak Epistemic Phenomenalism posits that knowledge of physical bodies is mediate (inferred) knowledge, depending on an inference from immediately known mental states.

Where Ontological Phenomenalism is a thesis about the status of objects (what kinds of things they are), Semantic and Epistemic Phenomenalism (and their weaker cousins) are theses about the subject and what it has access to in experience. Another once popular thesis dealing with the subject’s access to its objects would construe phenomenalism as a thesis about what is directly present to consciousness, or ‘before the mind’s eye’ in experience. If one holds that only sensations and their phenomenal qualities are presented to conscious awareness in the experience of objects, one gets:

Presence Phenomenalism represents a long tradition in the history of philosophy. It has appeared in the guise of the ‘Cartesian theatre’ or the ‘veil of ideas’, and it is associated with various forms of representationalism. In claiming that only sensations are present to consciousness in experience, Presence Phenomenalism denies that bodies outside the mind can be present to consciousness. Hence it has typically been taken to entail a denial of direct realism regarding our knowledge of the external world (but I will argue below that Kant can avoid this supposed entailment). Most of the early moderns accepted some version of Presence Phenomenalism. Some were led to infer versions of Epistemic or Semantic Phenomenalism (or their weaker variants) on the basis of it.Footnote ⁸ (Berkeley, but not Kant, I will argue, numbers among them.)

Semantic Phenomenalism and Presence Phenomenalism both have to do with intentionality in representation, but they deal with different kinds of intentionality. The intentionality in Semantic Phenomenalism has to do with representational content, i.e. the purport of representations, what the subject takes itself to be representing. Hence the intentionality in Semantic Phenomenalism always involves representing an object as being an object of a certain kind. The intentionality in Presence Phenomenalism, by contrast, has to do with what is present to consciousness in a real relation to it, and is independent of whether the object is represented as anything. I emphasize that Presence Phenomenalism involves a ‘real relation’ to contrast it with Semantic Phenomenalism, which will not always be relational, e.g. in representations of non-existent objects, since there can be no real relation to non-existent things. The real relation at work in Presence Phenomenalism is the relation of phenomenal awareness obtaining between the conscious subject and the sensory states existing in the subject’s mind.

How one understands the relation between Presence and Semantic Phenomenalism turns out to be of crucial importance. We will see below that because Berkeley thinks that one can conceive of something only if that thing can be present to consciousness, he thinks that Presence Phenomenalism entails Semantic Phenomenalism (and subsequently Ontological Phenomenalism as well). Because Kant has an entirely different theory of intentionality, he is able to reject such an entailment relation.

There is an additional version of phenomenalism that is found primarily in discussions of Kant’s idealism. According to ‘intentionalist’ interpretations of Kant, empirical objects are identified with representations, but not in the sense that they are mental entities (things literally in the mind), but rather in the sense that they are representeds, i.e. intentional objects. But importantly, according to these interpretations, they are mere intentional objects, having no ontological status of their own. I will call this:

According to Intentional Object Phenomenalism, physical bodies are what we take ourselves to be representing in normal perception, but these objects have no reality independent of our representation of them. As an interpretation of Kant, Intentional Object Phenomenalism is meant as an explanation of his thesis that physical objects have no transcendental reality. Nevertheless, so the interpretation goes, they are represented and cognized as being fully empirically real.

Like Ontological Phenomenalism, Intentional Object Phenomenalism is a thesis about the status of objects, not just a thesis about the subject’s access to objects. But where Ontological Phenomenalism grants a positive ontological status to objects by virtue of identifying them with things that really exist (collections of sensations), Intentional Object Phenomenalism denies them any such status. Thus Van Cleve will describe Kantian appearances as ‘virtual objects’, which are ‘not to be conceived as having [their] own special kind of being’ (Van Cleve Reference Van Cleve1999: 8). Likewise, Aquila writes that ‘perceptual appearances would … not need to comprise a distinct set of mental or perceptual entities’; they should instead ‘be construed … as ordinary objects of experience … as they happen to “inexist” in instances of perception’ (Aquila Reference Aquila2003: 232). I will discuss Intentional Object Phenomenalism as an interpretation of Kant in greater detail in section 6.

I do not claim that the above taxonomy is exhaustive,Footnote ⁹ but it will be more than enough for present purposes. With this menu of phenomenalist options in full view, I turn now to an examination of Berkeley and Kant to show how their idealisms can be characterized and distinguished in terms of the different phenomenalist theses.

3. Berkeley’s Phenomenalism

In this section I consider Berkeley’s idealism in terms of his commitment to some of the above phenomenalist theses. Although I will not attempt to reconstruct every aspect of his idealism (and I will not touch on the immaterialist side of the Berkeleyan coin at all), we will see that one of the central lines in Berkeley’s reasoning can be cashed out in terms of his commitment to most of the phenomenalist theses laid out above.

First, it is clear that Berkeley would not have accepted Intentional Object Phenomenalism. Berkeleyan ideas may well be intentional objects in the sense that they are what the mind is directed towards in experience. But they are not mere intentional objects. They really exist; their existence just happens to be a mental existence. Berkeley’s slogan is that esse est percipi, not esse est intendi nor esse est inesse (see Winkler Reference Winkler1989: 7 on this point). Second, I think it is uncontroversial to say that Berkeley’s idealism involves commitments to Presence, Epistemic, Semantic and Ontological Phenomenalism. My concern, however, is not to argue that Berkeley did accept each of these (again, I take it to be uncontroversial that he did), but rather to examine the relations between some of them in his system. I want to suggest that Berkeley thinks that Semantic Phenomenalism entails Ontological Phenomenalism, and that Presence Phenomenalism entails Semantic Phenomenalism. Hence Berkeley thinks that once one has accepted the claim that only mental states are present to consciousness (Presence Phenomenalism), one has already begun an irreversible slide into full empirical idealism (i.e. Ontological Phenomenalism).

I am more concerned with the ‘Presence Phenomenalism entails Semantic Phenomenalism’ claim, and so my discussion of the ‘Semantic Phenomenalism entails Ontological Phenomenalism’ claim will be brief. The basic idea with the latter, as many have noticed, is that some of Berkeley’s core arguments for idealism turn on semantic considerations about what can or cannot be conceived (e.g. Rickless Reference Rickless2013: 93–7). Simplified, his reasoning runs something like this:

1. 1. The only things to which the subject can make content-laden reference in experience are the subject’s own mental states. [Semantic Phenomenalism]

2. 2. We successfully refer to physical bodies.

3. 3. Therefore, physical bodies must be mental states. [Ontological Phenomenalism]

Premise 2 is supposed to be a common-sense assumption not open to serious doubt, and hence the move from Semantic Phenomenalism to Ontological Phenomenalism is fairly direct.

This line of reasoning is most apparent in PHK §3, where Berkeley gives an account of the semantics for ‘exists’:

The claim here is that the representational content of my thought about a body involves the thought that such objects are essentially perceptibles and nothing else besides this. His inference is hasty, but it is clear that he thinks that, since it is possible to conceive bodies only as perceptibles, i.e. as collections of ideas (Semantic Phenomenalism), they must therefore be only perceptibles (Ontological Phenomenalism). The move from Semantic Phenomenalism to Ontological Phenomenalism is also at work in the so-called ‘Master Argument’ at PHK §§22–3. There Berkeley argues that since we cannot even form a coherent conception of something existing independent of the mind, the bodies we do successfully conceive must therefore exist only in the mind.

Clearly Semantic Phenomenalism is doing important work in getting Berkeley to his full empirical idealism. Accordingly, one wants to know how he arrives at Semantic Phenomenalism. I argue that Berkeley derives Semantic Phenomenalism from Presence Phenomenalism, and that this derivation depends on his additional commitment to his empiricist theory of intentionality.

The core of Berkeley’s theory of intentionality has to do with what Gallois has called the ‘imagistic criterion’ as a condition on concept possession, i.e. on the ability to conceive of or have a thought about something: ‘In order for someone to have the concept of things of a certain kind, he must be able to [conjure an] image [of] things of that kind’ (Gallois Reference Gallois1974: 59; see also Winkler Reference Winkler1989: 3; PHK §§5, 14, 33 and DHP 203). A corollary of the imagistic criterion is a thesis we might call the ‘resemblance criterion’. This is the claim that an idea cannot represent an object unless the idea resembles (‘is like’) its object (see PHK §§8–9, 27 and DHP 204).Footnote ¹⁰ Winkler describes the position as follows:

Accordingly, Berkeley’s theory of intentionality places considerable restrictions on what can be conceived in thought. If it is not possible to form an image of a certain kind, then I cannot have a representation of an object of that kind.

So what delineates the range of possible image-formation? According to Berkeley, it is possible to form an image of something only if it is possible to have an immediate sensory perception of that thing:

Every time I successfully conjure an image of an object of a certain kind, I am relying on memory of an earlier immediate perception of an object of that kind. And for Berkeley the only objects that can be perceived immediately are ‘ideas or sensations’ (DHP 215; see also PHK §§1, 3, 46 and DHP 174).

Putting it all together, we get the following argument:

1. 1. It is possible to conceive of an object of a certain kind only if it is possible to form an image of an object of that kind.

2. 2. It is possible to form an image of an object of a certain kind only if objects of that kind can be immediately perceived.

3. 3. The only things that can be immediately perceived are ideas.

4. 4. Therefore, it is possible to conceive only of ideas.

The first line embodies Berkeley’s belief that an idea can represent an object only insofar as that idea functions as a resembling image of its object. The second line specifies Berkeley’s claim about the limitations on image formation: I can form images only of things that could be present to consciousness (i.e. be ‘immediately perceived’). Lines 3 and 4 should be familiar: they are Presence Phenomenalism and Semantic Phenomenalism stated in Berkeleyan terminology. Regarding 3, Berkeley makes it clear that his notion of immediate perception involves a real relation between the perceiver and an idea which is directly present to consciousness; to immediately perceive an idea is for the whole idea to be given directly to conscious awareness (on this point, see Pappas Reference Pappas1982: 7; Park Reference Park1982: 35; Rickless Reference Rickless2013: 42). This amounts to a commitment to Presence Phenomenalism. And regarding line 4, Berkeley’s notion of conception, or rather the range of possible conception, has to do with the possibility of what I have been calling content-laden reference. Accordingly, the claim that only ideas can be conceived amounts to the claim that they are the only possible objects of content-laden reference.Footnote ¹¹

We can see clearly now that by accepting a Berkeleyan theory of intentionality, together with the claim that only ideas are present to consciousness, we arrive first at the semantic claim that only ideas can be referred to in experience, and subsequently that the physical objects we refer to must themselves be no more than collections of ideas. Thus for Berkeley, Presence Phenomenalism entails Semantic Phenomenalism, which itself entails Ontological Phenomenalism. I will argue in the next section that, because Kant adopts a radically different theory of intentionality, he could accept Presence Phenomenalism without being forced down the slide through Semantic Phenomenalism to Ontological Phenomenalism.

4. Kant’s Empirical Realism

The primary complaint from anti-phenomenalist critics has to do with a perceived incompatibility between phenomenalism and Kant’s empirical realism. I will be arguing in the following two sections that it is possible to attribute Presence Phenomenalism and Intentional Object Phenomenalism to Kant while maintaining a significant distance between his theory and Berkeley’s, and while staying true to the central tenets of empirical realism. In this section I will briefly explain what those tenets are and how they place interpretative constraints on our understanding of transcendental idealism.

The core of empirical realism, as I understand it, comprises the following metaphysical, semantic and epistemological theses.

In other words, the empirical world exists more or less as physical science and common sense describe it (outside the mind!) and our knowledge of and reference to objects in the external, empirical world is immediate.

I take these components of empirical realism to be non-negotiable aspects of Kant’s theory. Any interpretation that is inconsistent with any of these cannot be viable. When anti-phenomenalist critics reject phenomenalist interpretations as too Berkeleyan, their complaints typically involve the claim that phenomenalism would contradict one or more of these features of empirical realism. For instance, Collins and Abela focus on Semantic and Epistemic Immediacy:

Langton worries primarily about Scientific Realism:

Allais is concerned with Epistemic Immediacy and Scientific Realism:

While I whole-heartedly agree that any version of phenomenalism that contradicted the metaphysical, semantic and epistemic aspects of empirical realism should be rejected, I just as whole-heartedly believe that not all phenomenalisms present such a threat.

It is clear that the ontological, semantic and epistemic variants of phenomenalism are incompatible with empirical realism. Ontological Phenomenalism is incompatible with Kant’s realism about the meta-physical status of physical bodies, and Semantic and Epistemic Phenomenalism are incompatible with the belief that we refer to and know those bodies. Further, because of empirical realism’s semantic and epistemic immediacy, even the weaker versions of Semantic and Epistemic Phenomenalism are ruled out.

What should we make of Presence Phenomenalism and Intentional Object Phenomenalism? In the remaining two sections, I will make the case that each of these can be construed in a way that is consistent with empirical realism. Consequently, the anti-phenomenalist claim that all forms of phenomenalism collapse into Berkeleyan empirical idealism and undermine empirical realism is unfounded.

5. Kant as Presence Phenomenalist

According to Presence Phenomenalism, only internal, mental states are directly present to consciousness in experience. Kant makes several remarks that could easily be taken as expressions of Presence Phenomenalism.Footnote ¹⁵ For instance, throughout the first Critique, he repeatedly describes the idealistic aspect of his transcendental idealism with the claim that ‘we have only to do with our representations; how things in themselves may be (without regard to representations through which they affect us) is entirely beyond our cognitive sphere’ (A190/B235; see also A105, A129, A390). This expression is admittedly vague, but one natural way to take it is as the claim that only the sensory effects resulting from the transcendental affection of things in themselves are immediately present to consciousness in experience (we have only ‘to do’ with them). Although sensations could be present to consciousness only in the context of their organization into intuitive form, as the matter of what is presented in intuition, it would be only them with which we have ‘to do’. Further indication of Presence Phenomenalism as a part of Kant’s theory of experience appears early in the Aesthetic in his account of the relationship between intuition, sensation and space. He claims that empirical intuitions involve sensations being ‘ordered and placed’ in space (A20/B34), and that we ‘represent them as outside and next to one another’ (A23/B38). The model Kant seems to have in mind is one in which intuitions enable the representation of objects by way of presenting spatially arrayed collectionsFootnote ¹⁶ of sensations to the subject’s conscious awareness.Footnote ¹⁷

Several commentators have adopted this line of interpretation over the years. For instance, Vaihinger (Reference Vaihinger1892: 32) argued that Kantian appearances are originally constituted by an Empfindugsmaterial arrayed spatiotemporally. More recently, George (Reference George1981: 240) defends a ‘sensationist’ interpretation of Kant according to which reference to external objects depends on ‘an interpretation placed upon certain sequences of sensations’, and that ‘initially only the succession of sensation is present’. Sellars argued that Kantian intuition involves the construction of mental ‘image models’ that stand in as ‘proxies’ for the external objects we come to represent (Reference Sellars2007: 460; for similar interpretations, see Haag Reference Haag2007 and Jankowiak Reference Jankowiak2016).

One’s first reaction to this sort of reading might be to exclaim that it is just unabashed Cartesianism and hence marks a lapse back into a pre-Kantian theory of mind. For if only sensations are present to consciousness, then only they can be the immediate objects of reference or knowledge (in violation of empirical realism’s semantic and epistemic immediacy). And this leads directly either to the Berkeleyan empirical idealism or at least the Cartesian scepticism that Kant took himself to have refuted in the Refutation of Idealism.

This reaction would be premature. It only follows that the objects of immediate reference and knowledge are the same sensations that are present to consciousness if one follows Berkeley in the inference from Presence Phenomenalism to Semantic Phenomenalism. The claim that we can only immediately refer to objects that are present to consciousness is by no means tautological and accordingly the inference depends on a substantive theory of the conditions on the possibility of reference to objects. In this regard, we saw above that Berkeley held an imagistic theory of representation according to which the range of what can be conceived is coextensive with the range of what can be present to consciousness.

Now at a certain level of description, Kant is in agreement with Berkeley regarding the possibility of intentional reference in experience. Both hold that something can be referred to in cognition only if such a thing could be represented sensibly. That being said, no one would attribute the Berkeleyan theory of intentionality to Kant. The Berkeleyan theory entails that I can form and employ a concept only if the concept in question can be presented in an image of an object to which the concept applies. Thus I can use the concept ‘dog’ only if I can form an image of a dog. A consequence of this is that unless I can literally see (or otherwise sense) a property in an image, I cannot form a concept of it. This leads Berkeley to banish several core philosophical and scientific concepts from his picture of reality. Thus the concept of physical bodies as causally efficacious is ruled out on the Berkeleyan model because the images of physical bodies we encounter are passive, with no efficacy discernible in them (see PHK §25). He gets rid of any metaphysical concept of substantiality as well: since a persisting substratum, distinct from the perceived sensible qualities in an object, cannot be perceived in a ‘perpetually fleeting and variable image’ (DHP 205), it follows that the only concept of physical substance available to Berkeley denotes no more than the continued conjunction of several sensible ideas (see also PHK §§1, 37 and DHP 197–9). And since scientific posits such as gravitational or magnetic fields can never be perceived directly, their concepts can serve only to refer to certain sensible regularities (iron filings lining up near iron bars, objects falling towards the earth, etc.), but never to an actual invisible field (see PHK §§103–7).

For Kant, by contrast, we do succeed in conceptualizing the objects of experience in terms of active causal interactions between objects, we do think of bodies in terms of persistent substrata and we do incorporate unperceived entities (like magnetic fields) into our scientific representations of the world. And we form these conceptions despite the fact that causal interactions can never be seen directly, that our perceptions of objects are fleeting (never eternally persistent) and that many scientific entities are not even in principle perceivable. Clearly then, the Kantian sense in which we can conceive only of objects that can be represented sensibly is very different from the Berkeleyan sense. In contrast to Berkeley’s imagistic model of intentionality, Kant argued that an object can be conceived so long as its concept can be represented within the spatiotemporal form of intuition. Thus although I cannot perceive causality itself, I can successfully represent the temporal determination of irreversible sequences of events in intuition (see A188/B232ff.). Although I never perceive the substratum underlying sensible accidents, I can still represent intuited bodies as containing a persistent substratum (see A182/B224ff. and Bxli n.). And although I can never directly perceive magnetic fields, I can still refer to them as actual entities because positing them as the cause of perceived effects is necessary for making sense of those effects (see A226/B273).

Let us take stock. Berkeley held a theory of intentionality according to which we can conceive of something only if such a thing could be present to consciousness. This claim, together with the belief that only ideas are present to consciousness (Presence Phenomenalism), leads Berkeley to infer that we can conceive only of ideas (Semantic Phenomenalism). The Kantian theory of intentionality, by contrast, allows for the possibility of conceiving of objects even though those objects cannot be present to consciousness. Consequently, if we attribute to Kant the claim that only collections of sensations are present to consciousness (Presence Phenomenalism), we need not thereby also attribute to him the claim that only collections of sensations can be conceived in experience (Semantic Phenomenalism), since he is already on the record saying that it is possible to refer to something even if it cannot be present to consciousness. Thus a Kant who adopts Presence Phenomenalism need not follow Berkeley down the slide through Semantic Phenomenalism and Epistemic Phenomenalism all the way to Ontological Phenomenalism.

I take this to establish the claim that Presence Phenomenalism would not collapse Kant’s idealism into Berkeley’s. Anti-phenomenalist critics may not yet be satisfied however. For it could be objected that even though Presence Phenomenalism does not entail Semantic Phenomenalism on Kant’s theory of intentionality, it would still entail Weak Semantic Phenomenalism (the claim that the only immediate objects of reference in experience are mental states). According to Weak Semantic Phenomenalism, it is possible to conceptualize and refer to objects existing outside of my mind, but my intentional relation to these objects is mediated by a prior conceptual awareness of internal mental states. While not quite as objectionable as the stronger version, Weak Semantic Phenomenalism is nevertheless incompatible with empirical realism’s semantic and epistemic immediacy.

However, even this entailment relation (from Presence Phenomenalism to Weak Semantic Phenomenalism) would not hold for Kant. One would only think that Presence Phenomenalism entailed Weak Semantic Phenomenalism if one thought that all conceptual reference to objects not present to consciousness must be constructed out of a conceptualized awareness of whatever is present to consciousness. But this is simply not Kant’s theory of intentionality. As he explains in the Transcendental Deduction and Second Analogy, all objective representation is generated by the rule-governed synthetic activity of the categorially structured understanding.

I take Kant to be saying here that the ability to refer to an object as an object, and as an object of a certain kind (especially including as a spatial object outside the mind), depends on the understanding’s ability to construct the concept of the object by combining representations in accordance with rules.Footnote ¹⁸ This account of the generation of conceptual reference nowhere requires that conceptual reference to an object not present to consciousness depends on a conceptual reference to what is present.

What all this shows is that it is possible to attribute Presence Phenomenalism to Kant so long as one also insists (as one should), that having an object present to consciousness is not the same thing as (and neither entails nor is entailed by) making content-laden reference to that object. Thus Presence Phenomenalism entails neither Semantic Phenomenalism nor its weaker cousin and so does not threaten empirical realism.

6. Kant as Intentional Object Phenomenalist

Intentional Object PhenomenalismFootnote ¹⁹ (also referred to as ‘intentionalism’ in the literature: see Aquila Reference Aquila2003) is a thesis about the ontological status of objects. Physical bodies are identified merely with the intentional content of certain complex representational states. They are what our representations purport to be about, but beyond the representational purport of ‘intentional inexistence’ (as it is sometimes called), such objects have no positive ontological status.

To function as an interpretation of transcendental idealism, Intentional Object Phenomenalism must be qualified in a few important respects. Most importantly, it must be noted that physical bodies (qua intentional objects) are ontologically distinct from things in themselves (which makes Intentional Object Phenomenalism incompatible with traditional ‘two-aspect’ readings). Things in themselves exist in the full sense and thus have a radically different ontological status than the physical bodies (‘appearances’) to which they correspond. So it is not as though all reality is reduced to mere intentional inexistence on such an interpretation. Secondly, Intentional Object Phenomenalism would have to be understood as a transcendental claim about the status of objects, not an empirical one. As I will spell out in more detail below, physical bodies are necessarily represented as the sorts of things that exist in space, outside the mind; this precludes the possibility (on pain of incoherence) of representing them simultaneously as mind-dependent mere intentional objects. Thus while it is transcendentally true that physical bodies are mere intentional objects, it is not empirically true. Or to put it differently, a textbook on transcendental philosophy would correctly describe physical bodies as mere intentional objects, but a textbook on physics would not.

Although usually in a minority, there have been at least a few prominent commentators who have recognized the appeal of Intentional Object Phenomenalism as a potentially fruitful interpretative approach to Kant (see Sellars Reference Sellars1967; Aquila Reference Aquila1983, Reference Aquila2003; Van Cleve Reference Van Cleve1999; Haag Reference Haag2007; George Reference George1981; Pereboom Reference Pereboom1988). However, anti-phenomenalist commentators have been quick to group Intentional Object Phenomenalism with other forms of phenomenalism and dismiss it with the rest. For instance, Allais argues that ‘to compare Kantian appearances with merely intentional objects is to keep appearances in the mind: we simply have a more sophisticated version of a phenomenalist interpretation’ (2015: 42). Sophisticated or not, she rejects this approach with all other interpretations which ‘mentalize’ Kantian appearances (2015: 37–56). Similarly, Hall argues that Intentional Object Phenomenalism ‘simply returns us to a Berkeleyan phenomenalism in a different guise, in so far as the phenomenal object would still depend entirely upon the mental activity of the subject to generate the object through acts of construction … or mental directedness’ (Hall Reference Hall2010: 47). And Ameriks describes Van Cleve’s intentionalist reading as a form of ‘neo-Berkeleyanism’ (Ameriks Reference Ameriks2006: 77). These commentators are being fair in characterizing Intentional Object Phenomenalism in terms of the dependence of physical bodies on certain kinds of mental activity. The question, though, is whether it is also fair to dismiss Intentional Object Phenomenalism out of hand along with all other phenomenalisms as not significantly different from Berkeleyan idealism. I will argue that it is not. I will first spell out more explicitly how radically different Intentional Object Phenomenalism is from Berkeley’s empirical idealism. The resulting contrast will point the way towards demonstrating the compatibility of Intentional Object Phenomenalism with the scientific realism at the heart of empirical realism.

The central contrast between a Kantian Intentional Object Phenomenalism and Berkeley’s empirical idealism has to do with the metaphysical status of the objects of experience, i.e. ordinary physical bodies. For Berkeley, physical bodies are identified with collections of ideas which exist in the mind. Specifying the metaphysical status of an object for Kant is complicated by his empirical/transcendental distinction. How we describe the metaphysical status of an object qua object of experience will differ from how we describe it qua explicandum of transcendental philosophy. In either case, though, Kant’s account will differ significantly from Berkeley’s.

If we ask about the status of an object qua explicandum of transcendental philosophy, we are asking how a transcendental idealist would characterize the ultimate ontological status of empirical objects. According to Intentional Object Phenomenalism, physical bodies are mere intentional objects which have no positive ontological status of their own (this would be put as an explication of Kant’s claim that physical bodies are not transcendentally real). This differs markedly from Berkeley, who was by Kant’s lights a transcendental realist regarding physical bodies. The collections of ideas that constitute Berkeleyan physical objects really exist in the mind as more than mere intentional objects.

If we instead ask about the status of an object qua object of experience, then (according to Intentional Object Phenomenalism) we are asking for a specification of the representational content describing the object in question. After all, with mere intentional objects, what the object is will be determined by the intentional content by which it is represented. In normal cases of everyday cognition, this representational content will involve (at least implicitly) predicates such as ‘spatial’, ‘substantial’, ‘not-in-my-own-mind’ (among of course others). More specifically, the first Critique explains that all bodies must be represented as causal substances describable by extensive and intensive magnitudes; and Metaphysical Foundations of Natural Science (Kant Reference Kant2002) explains that all bodies move and interact in space and are composed of spatially diffuse attractive and repulsive forces. According to Kant’s theory of experience, it is possible to form objectively valid judgements about physical objects only when such objects are being represented and intended in accordance with these principles. Thus a physical body, qua object of experience, is exactly the sort of thing common sense and physical science take it to be: a spatially extended, causally active substance which interacts with other such substances. It is not, as with Berkeley, a collection of ideas in the mind.

The difference between Kant and Berkeley on this issue is by no means trivial, since it indicates how Kant and Berkeley can go in opposite directions with respect to the metaphysics of the physical world. To spell this out, I want to focus on the different ways that Kant (via Intentional Object Phenomenalism) and Berkeley will understand the truth-conditions on judgements about objects: under what conditions is a sentence of the form ‘o is F’ true?

Berkeley identifies physical bodies with transcendentally real collections of sensations. On his theory, a judgement of the form ‘o is F’ will be true if and only if the collection of sensations constituting o is in fact F, hence, if and only if it presents F-ness to consciousness.

Truth-conditions on claims about Kantian intentional objects will be quite different. For facts about mere intentional objects can be determined only by the representational content through which they are represented. Specifically, for Kant, a judgement of the form ‘o is F’ will be true if and only if the judgement is formed in accordance with the rules laid down by the synthetic a priori laws of the understanding. Rules such as those laid down in the Analogies of Experience determine the necessary features that objects must have (e.g. causality and substantiality) (A176/B218ff.). Rules laid down in the Postulates of Empirical Thinking determine what is actual (even if only contingently actual) in objects (A218/B266, A225–6/B272–4).

So the features of a Berkeleyan object are determined by the features displayed by collections of sensations, and the features of a Kantian object are determined by the intentional content of judgements formed in accordance with the a priori laws of the understanding. The resulting metaphysics of Berkeleyan objects differs markedly from the metaphysics of Kantian objects. I emphasize the following four differences as especially significant.

7. Conclusion

I have made the case that not all phenomenalisms are created equal. While there is a tendency in the literature to reject each while rejecting all, I hope to have shown that two forms of phenomenalism are consistent with Kant’s realism. The claim that only sensations are present to consciousness may appear to threaten Kant’s insistence on the semantic immediacy of our cognition of external objects; but this threat disappears once one takes seriously Kant’s account of intentionality, which bears little resemblance to Berkeley’s theory that an object can be intended only if it can be directly present to consciousness. And the claim that empirical objects are ultimately to be construed as mere intentional objects may appear to threaten Kant’s realism regarding the metaphysical status of physical bodies; but this threat disappears once one takes seriously what Kant thinks is required for making true claims about objects, and once one sees that these Kantian truth-conditions bear no resemblance to Berkeley’s claim that what is true of an object is determined only by what that object presents to consciousness. I do not claim to have demonstrated that Kant did indeed subscribe to these phenomenalist theses, although I do think the case can be made that he did. The purpose of this paper has simply been to show that these interpretative options remain open, viable and promising, and that it is possible to read Kant as a phenomenalist without collapsing his idealism into Berkeley’s.Footnote ²²