2. Kant’s Target: the Cartesian Sceptic

Kant tells us who the specific target of the Refutation is: ‘the theory that declares the existence of objects in space outside us to be … merely doubtful … [This] is the problematic idealism of Descartes, who declares only one empirical assertion, namely I am, to be indubitable’ (B274).Footnote ⁹ Kant apparently takes the narrator of the Meditations to be describing Descartes’s own views. Since that may be a mistake, I describe Kant’s argumentative opponent as the Cartesian meditator or the Cartesian sceptic. Many types of scepticism have been dubbed ‘Cartesian’, so we should clarify what Kant has in mind. B274 provides two points of guidance. First: the entities whose existence is in question are ‘objects in space’, not mind-independent objects in general. Kant repeats this in stating the theorem to be proved: ‘The mere, but empirically determined, consciousness of my own existence proves the existence of objects in space outside me’ (B275, see also Bxxxix n.). Kant never suggests that the Refutation is concerned with sceptics who doubt the existence of everything outside their mind.

Second: Kant states that the Cartesian sceptic allows only one indubitable empirical assertion, namely, ‘I am.’Footnote ¹⁰ This leaves open the possibility that the Cartesian sceptic accepts some non-empirical assertions. In the first note to the Refutation, Kant gives one example: ‘Idealism assumed that the only immediate experience is inner experience’ (B276). Presumably, Kant therefore thinks that the question of which experiences are immediate is not an empirical issue for the Cartesian sceptic.Footnote ¹¹

These considerations show which part of Descartes’s Meditations Kant is focused on. By the end of the Meditations, Descartes’s narrator no longer doubts spatial objects’ existence and makes many empirical assertions. So Kant’s conception of problematic idealism is based on Meditations I and II. There, the narrator focuses his doubts on the existence of spatial objects, though the doubts also extend to God. The culmination of the doubts is expressed as follows:

This, I claim, is the type of scepticism Kant is concerned with: the doubting of the existence of all spatial things. We can now consider how well the Refutation fares against its target.

3. How Kant’s Argument Seems to Beg the Question

Here is the core text of the Refutation, with the amendment Kant suggests at Bxxxix n. added in brackets:

Kant’s proof is naturally divided into two parts, marked here with a double slash. The first half is primarily metaphysical, concluding with ‘the existence of actual things that I perceive outside myself’. The second half is more epistemological, concluding that ‘the consciousness of my own existence is at the same time an immediate consciousness of the existence of other things outside me’. Like other commentators, I focus on defending the first half, even if Kant’s ultimate aim is achieved only in the second.Footnote ¹³

From its first step on, Kant’s proof raises difficult questions. Even the relation of the first premise to the Theorem is obscure. What exactly are ‘determinations’ in this context? Is ‘consciousness of my existence as determined in time’ supposed to be equivalent to ‘[t]he mere, but empirically determined, consciousness of my own existence’? Most of the questions the proof raises do not directly bear on whether it begs the question against the Cartesian sceptic, however, since that sceptic is not particularly concerned with issues like the temporal determination of consciousness. The main question that does directly bear on that issue is the following. The proof seems to begin with only one ontological commitment, namely, a commitment to my (temporally determined) existence. The proof ends, however, with an ontological commitment to ‘actual things that I perceive outside myself’. Given what Kant has told us beforehand, we know that ‘outside’ is meant spatially. Yet it is hard to see how any valid argument could start with just the former ontological commitment and end with the latter, unless it begged the question at some point. We can easily imagine the grim scenario Mark Twain describes:

Proving that such a scenario is impossible would seem to require philosophical magic.

If Kant’s proof contains magic, it is well-hidden. The proof appeals to ‘something persistent in perception’, which can be read as being either a real temporal thing or a mere perception as of a temporal thing. The perception of that persisting thing is said to be possible ‘only through a thing outside me and not through the mere representation of a thing outside me’. Kant says little about how this is supposed to work. Some commentators (e.g. Abela Reference Abela2002: 186) have suggested that he is drawing on the Analogies of Experience here (A176/B218–A218/B265), but the Analogies do not have any obvious anti-Cartesian magic either. It is therefore hard to see how Kant’s proof can reach its ontologically robust conclusion from its ontologically modest premise without begging the question against its target.Footnote ¹⁴

Hence, some commentators have concluded that Kant’s conclusion takes spatial objects to be mere representations, despite his insistence that he is concerned with ‘a thing outside me’, and not ‘the mere representation of a thing outside me’. The Cartesian sceptic accepts that we have representations of spatial objects, after all. This would bring Kant’s view close to Berkeley’s, however, which he explicitly rejects in the introduction to the Refutation. Of course, Kant could be either misunderstanding or misrepresenting Berkeley’s view. It would be desirable, however, if we could find a non-Berkeleyan reading of Kant’s argument that did not beg the question against its target. My approach below is to identify implicit premises we can attribute to Kant that involve and exploit an additional (but dialectically safe) ontological commitment. Before doing so, however, I turn to the question of when an argument begs the question.

4. Question-Begging

Some philosophers (e.g. Walton Reference Walton1994) see question-begging as a pragmatic phenomenon, such as a violation of conversational rules. Others (e.g. Sgaravatti Reference Sgaravatti2013) see it as purely epistemic, such as using a premise one could justifiably accept only if one already had a justified belief in the conclusion.Footnote ¹⁵ Partly for these reasons, it is often hard to evaluate whether a given argument is question-begging.Footnote ¹⁶

Question-begging comes in both relational and non-relational forms. An argument can beg the question against a particular thinker, or it can beg the question simpliciter. These may or may not be inter-definable.Footnote ¹⁷ My concern here is only with sufficient conditions for avoiding relational question-begging at the level of individual premises. I propose the following disjunctive principle:

Let’s consider each disjunct separately.

On (a): if I accept P, and you appeal to P in arguing against me, then I should not complain that you are begging the question against me. That holds even if your argument exploits an inconsistency in my views. If I have asserted both P and not-P while denying Q, then there is nothing question-begging about you appealing to P in arguing for Q against me.

On (b): consider Kant’s assumption (e.g. at A39–41/B56–8, Prolegomena, 4: 282), in his arguments-by-elimination against transcendental realists, that space and time might be forms of intuition. That assumption crucially involves the notion of a form of intuition. By ‘crucially involves’, I mean that the premise could not contribute to the argument if it were reformulated without that notion.Footnote ¹⁸ Kant’s assumption about forms of intuition does not beg the question against the transcendental realists because the realists Kant is concerned with never considered this notion and because the assumption itself does not conflict with transcendental realism.Footnote ¹⁹ For those reasons, Kant’s assumption does not seem question-begging against the realists. This is what disjunct (b) of No Begging is meant to capture.

No Begging applies only to individual premises. However, some arguments, as a whole, beg the question even though no individual premise does. Consider, for example, an argument of the form (directed at someone who denies P): P or R, P or not-R, therefore P (cf. Sgaravatti Reference Sgaravatti2013: 4). Because it focuses on individual premises, No Begging does not tell us why such arguments are question-begging. Fortunately, my purposes do not require any general principle. Instead, I take a two-step intuitive approach: to determine whether Kant’s argument is question-begging, we (1) clear its individual premises using No Begging, and then (2) make an intuitive holistic assessment of whether the premises in combination beg the question against the relevant thinker (e.g. via logical tricks).

5. Implicit Assumption: Affection

I now turn to identifying assumptions that (i) would not beg the question against the Cartesian sceptic and that (ii) we have textual reasons to ascribe to Kant. This section presents one assumption, and sections 6–7 present three more. In section 8, I argue that these assumptions can make the Refutation valid without begging the question against the Cartesian sceptic. In section 9, I briefly consider how plausible these assumptions are.

The first assumption is:

This assumption could be formulated in terms of ‘cognitions’ or ‘experiences’. Perception, however, is Kant’s focus in the Refutation. He consistently describes perceptions (Wahrnehmungen) as involving sensation (Empfindung).Footnote ²¹ Whatever else perceptions may involve, Kant thinks sensations arise through affection. Kant endorses Affection in various places. For instance:

Similarly, Kant defines sensation in the Transcendental Aesthetic as ‘[t]he effect of an object on the capacity for representation, insofar as we are affected by it’ (A19–20/B34). The natural reading of these passages takes the object affecting the mind to have a perception-independent existence. Some commentators (e.g. Kitcher Reference Kitcher1999) reject that natural reading. I do not enter into that debate here, though I think a larger consideration of Kant’s views supports the natural reading.Footnote ²²

Granting the natural reading, Kant’s confidence about Affection is striking. J. N. Findlay writes that, for Kant, it is ‘a primordial certainty which stands in no need of justification, even if its lack of empirical content means that it can never be ranked as knowledge’ (Findlay Reference Findlay1981: 2). Findlay implies that Kant regards Affection as non-empirical. This is what we should expect, if it concerns what happens prior to the most basic elements of experience (sensations). Against Findlay’s suggestion that it is ‘primordial’ for Kant, though, we might think Affection has a priori grounds, such as the principle that objects can change only by being affected by other objects (a principle Kant accepts in some pre-Critical work).Footnote ²³ Or we might think that it is part of the concept of experience, arising through mere analysis.Footnote ²⁴ In any case, Kant regarded it as non-empirical. He therefore could rely on it while holding that the Cartesian sceptic only granted one empirical claim (‘I am’).

Note that Affection does not say that perceptions represent the affecting objects that helped produce them. It is compatible with the perceptions being pure hallucinations that do not represent anything real. At the same time, it does not rule out the view that perceptions represent their causes (as e.g. Spinoza thought) – a view that Descartes himself might have accepted.Footnote ²⁵ I return to this issue in section 7.

On my interpretation, Affection is key to avoiding the ontological leap I described in section 3. Would this beg the question against the Cartesian sceptic? It says nothing about the nature of the objects that affect us. It does not say they are spatial. Nor does it say whether the affecting object might be us or something in us. For this reason, Kant could use it in an anti-Cartesian argument without begging the question. For while Descartes’s meditator doubts whether his experiences arise from spatial objects, he never considers whether they might come from nowhere.Footnote ²⁶ As we saw in section 2, the most radical alternatives the meditator considers are that his thoughts come from ‘God, or whatever I may call him’, or from himself. This was so despite the meditator’s aim of doubting everything that could be doubted. The Cartesian sceptic therefore seems to implicitly grant that there is ‘no doubt whatever’ that Affection is true. If so, then No Begging allows Kant to use it without begging the question against his target.

6. Implicit Assumptions: Spatial and Temporal Perceptions

The second and third implicit assumptions I ascribe to Kant are:

There is more than one way to motivate attributing these claims to Kant, but my approach starts with a certain metaphysical reading of Kant’s idealism.

On the reading in question, Kant holds that each object of experience is, in some important sense, the same thing as some mind-independent thing in itself.Footnote ²⁸ This sameness holds despite only the former entities being spatiotemporal. In contrast to some non-metaphysical one-world readings, this reading holds that the appearance/thing-in-itself distinction is a distinction between properties of things, as opposed to merely between ways of representing things. On this view (unlike Berkeley’s idealism), spatiotemporal properties are mind-dependent but belong to objects that have some kind of mind-independent existence.Footnote ²⁹ The properties are not mere representations, but are instead rooted in both the mind-independent objects and in the minds. This puts them halfway between purely mind-dependent properties and purely mind-independent properties.Footnote ³⁰ An object’s spatiotemporality consists in being perceived (or perceivable) as spatiotemporal, in virtue of that object affecting the mind in a certain way. Suitably elaborated, this type of idealism would imply Perceived Time and Perceived Space.

What are the criteria of experience that the premises evoke? There is no easy summary of Kant’s theory of the conditions of experience. But these conditions clearly concern the unity of experience, something Kant devotes far more attention to than his predecessors. Hence, Kant writes that:

Kant’s criteria of experience hinge on the idea of a synthetically unified consciousness. He suggests that this unity of consciousness is the keystone to the whole of his transcendental philosophy (B134 n.).

The relevance of this to the Refutation is clear from Note 3 to the Refutation, where Kant writes, ‘Whether this or that putative experience is not mere imagination must be ascertained according to its particular determinations and through its coherence with the criteria of all experience’ (B279). On the reading of Kant’s idealism I have been discussing, it is when and only when we properly connect (or could properly connect) our perceptions that an object that is perceived as having some spatiotemporal property really has that property. What this ‘proper connection’ ultimately amounts to is one of the central interpretative questions about the Critique. For now, it is enough to think of it as some relation among representations that Kant thinks his predecessors have overlooked.

So interpreted, Kant sees spatiotemporal qualities as complex secondary qualities of objects. As Kant knew, Descartes never contemplated the possibility that space or time might be secondary qualities. In the Prolegomena, he describes the ‘generally accepted’ view of warmth and colour as predicates that ‘belong not to these things in themselves but only to their appearances’. He then claims that his view differs from this only insofar as he says the same thing about ‘extension, place, and more generally space’ (4: 289). Of course, the nature of secondary qualities is controversial (and Kant’s readers have understood it in various ways; cf. Allais Reference Allais2007 and Rosefeldt Reference Rosefeldt2007). What matters here is that there are ways of understanding Kant’s claims about secondary qualities that support attributing Perceived Time and Perceived Space to him.

There is at least one other way to motivate attributing these premises to Kant. These claims are compatible with certain forms of realism about spatiotemporal properties. A realist could take Perceived Time and Perceived Space as merely stating a sufficient condition for accurate perceptions of temporal and/or spatial objects. The criteria of experience would then be playing a purely epistemological role. These assumptions would hold because the relevant conditions are those under which spatiotemporal perceptions are veridical. Some of Kant’s readers, such as Henry Allison, have thought that Kant’s ‘idealism’ should be understood along these lines, as an unobjectionable realism combined with subtle non-metaphysical claims about epistemic conditions.Footnote ³¹

Readings like Allison’s are less popular now than they once were.Footnote ³² The important point, however, is that Perceived Time and Perceived Space are neutral about the metaphysical status of time and space. Hence, my reconstruction of the Refutation below does not hinge on the above metaphysical interpretation of Kant’s idealism (though it may not be compatible with all readings of Kant’s idealism). I now turn to showing how, according to the second disjunct of No Begging, neither of these assumptions begs the question against the Cartesian sceptic.

First, the Cartesian sceptic never considers Kant’s notion of a criterion of experience, which both Perceived Time and Perceived Space crucially involve. To be sure, in Meditation VI, the meditator claims that coherence can distinguish dreams from waking life (CSM, II. 61–2). This, however, makes no appeal to anything like the synthetic unity of consciousness, which Kant sees as essential to the criteria of experience. Nor does the meditator consider the possibility of space and time being secondary qualities, or of Allisonian non-metaphysical epistemic conditions.

Second, the two premises do not directly conflict with anything the Cartesian sceptic accepts. By themselves, they are consistent with the non-existence and unknowability of the material world and of God. They do not imply that we even have perceptions as of spatiotemporal objects or that any perceptions we do have meet the criteria of experience. That does not mean that a Cartesian sceptic would simply accept them, of course. Seeing where the argument was headed, a determined sceptic might challenge the notion of a criterion of experience. Even so, according to the second disjunct of No Begging, the claims can be used in a refutation (in Kant’s sense) without begging the question against the Cartesian sceptic. Whether that argument could itself change any determined sceptic’s mind is another question.

Two further notes. First: in the introduction to the Refutation, Kant states that the view of space ‘as a property that is to pertain to the things in themselves’ has been ‘undercut by us in the Transcendental Aesthetic’ (B274). Kant therefore had his own views about space and time firmly in mind as he wrote the Refutation. It would not be surprising, therefore, if the Refutation implicitly appealed to that view, such as by assuming Perceived Time and Perceived Space.

Second: unlike Affection, Kant probably has background arguments in support of these assumptions (primarily in the Transcendental Aesthetic). Therefore, even if these alone would somehow beg the question against the Cartesian sceptic, Kant may have further arguments for them that would not.Footnote ³³ In that case, the combined argument of the Refutation together with the Aesthetic’s arguments would not beg the question against the Cartesian.

7. Minimal Coherence and Intentionality

Neither by themselves nor in conjunction with Kant’s explicit premises do Affection and Perceived Time/Space imply that our perceptions represent any real object (accurately or otherwise), or that there are any real spatiotemporal objects. Hence, my reconstruction requires one further implicit assumption:

In effect, this says that if there are any temporal and spatial perceptions, at least some of these perceptions meet the relevant criteria.

Minimal Coherence may be hinted at in Note 3 after the Refutation, where Kant casually grants that any particular perception may be ‘the mere effect of the imagination (in dreams as well as in delusions)’ (B278). Here is one way to motivate the assumption within Kant’s framework. If Kant’s criteria concern unity, then they should pick out the most unified perceptions. That is, whichever perceptions hold together best satisfy Kant’s criteria.Footnote ³⁴ That would explain why Kant states that ‘[in] space and time … the empirical truth of appearances is satisfactorily secured, and sufficiently distinguished from its kinship with dreams, if both are correctly and thoroughly connected up according to empirical laws in one experience’ (A492/B520–1). Provided that there are spatiotemporal perceptions, Kant seems confident that at least some must satisfy the criteria of all experience.Footnote ³⁵ As a rough gloss: given any set of perceptions, at least some of them can be woven together into a coherent course of experience (while the others are given the status of dreams or hallucinations). Note that this requires some actual perception, in accordance with the second Postulate (A218/B266, A225/B372), to which the Refutation is attached.

Though it could be challenged,Footnote ³⁶ Minimal Coherence would not beg the question against the Cartesian sceptic. It again crucially involves the notion of criteria of experience, which the sceptic never considers, and it does not directly conflict with any claims that sceptic makes. Moreover, the Cartesian meditator’s description of his experience of sitting by the fire sounds coherent (see CSM, II. 13), and that sceptic never raises the possibility that there is not even a minimally coherent set of perceptions (though other sceptics might).

We can now return to an issue raised in section 5: what does it take for a perception P (as of something spatiotemporal) to represent some actual object O? The premises I have offered imply a sufficient condition: if O produces P through affecting the mind, and P coheres with the criteria of experience, then P represents O (and, moreover, does so accurately, as spatiotemporal). For example, if a perception as of a square was produced by some mind-independent thing, and that perception coheres with the criteria of experience, then that perception accurately represents that mind-independent thing as being square (or, at least, having some spatial quality or otherFootnote ³⁷ ).

Setting aside questions about spatiotemporal qualities, why would production plus the criteria be enough for an intentional relation between a perception and an object? Since this is a mere sufficiency claim, it could be motivated in more than one way. It would follow from a simple causal view like Spinoza’s, on which whatever produces a perception is represented by it, though perhaps only confusedly (a view which, as noted above, Descartes himself might have acceptedFootnote ³⁸ ). If the criteria of experience have modal implications, then the sufficient condition might also be supported by tracking accounts of intentionality.Footnote ³⁹ Finally, if the criteria have intersubjective implications (as suggested e.g. at Prolegomena, 4: 298), then the sufficiency condition might be supported by interpretationist theories of intentionality like Donald Davidson’s (Davidson Reference Davidson1984). Kant may not have held any of these views, of course. The important point is that the view of intentionality implied by the above premises could be motivated in various ways – and motivated, moreover, without begging the question against the Cartesian sceptic.

8. Reconstructing the Refutation

There is more than one way to use the assumptions given above to reconstruct Kant’s proof in the Refutation. Here is one such reconstruction:

1. I am aware of my existence as determined in time.

2. Any awareness of existence as determined in time requires a perception as of something persisting.

3. Therefore, there is at least one perception as of something persisting. (1, 2)

4. Affection: All perceptions result from the mind being affected by some object that exists independently of them.

5. Minimal Coherence: (a) If there is at least one perception as of a persisting thing, then there is at least one perception that is as of a persisting thing and coheres with the criteria of all experience. (b) If there is at least one perception as of a spatial thing, then there is at least one perception that is as of a spatial thing and coheres with the criteria of all experience.

6. Perceived Time: If (a) an object affects the mind and produces sensations involved in a perception as of a persisting object and (b) that perception coheres with the criteria of experience then (c) that affecting object is or constitutes a real, perceived persisting object.

7. Therefore, there is at least one real, perceived persisting object. (3, 4, 5a, 6)

8. If there is at least one real, perceived persisting object, then there is at least one perception as of a spatial object.

9. Therefore, there is at least one perception as of a spatial object. (7, 8)

10. Perceived Space: If (a) an object affects the mind and produces sensations involved in a perception as of a spatial object and (b) that perception coheres with the criteria of experience, then (c) that affecting object is or constitutes a real object in space.

11. Therefore, there is at least one real object in space. (4, 5b, 9, 10)Footnote ⁴⁰

Bracket the question of whether the premises are philosophically plausible (the topic of section 9), and consider how this fares as an interpretatively plausible, non-question-begging reconstruction of Kant’s proof. On an interpretative level, there is straightforward textual motivation for ascribing premises 1, 2 and 8 to Kant.Footnote ⁴¹ Though these premises involve significant philosophical complexities, none of them would beg the question against the Cartesian sceptic. The Cartesian sceptic accepts 1, and has no independent reason to deny 2 or 8. Given No Begging, none of the other assumptions (4, 5, 6, 10) would beg the question against the Cartesian sceptic either. The individual premises are therefore cleared of the charge of question-begging against Kant’s target. In accordance with my suggestion in section 4, we should now make an intuitive holistic assessment of whether the premises beg the question in combination. Intuitively, I claim, there is no holistic problem here. Unlike the case mentioned above (P or R, P or not-R, therefore P), this reconstruction does not involve a question-begging structural trick. The reconstruction therefore does not beg the question against the Cartesian sceptic.

Issues of interpretative plausibility and question-begging aside, the reconstruction has other desired features. It is logically valid, and its conclusion is not Berkeleyan. Though the conclusion is consistent with counting spatiality as a secondary quality, it does not construe the object as mere representations. Nor does it require identifying spatial properties with mere representations. This reconstruction therefore avoids the interpretative dilemma I described in the introduction.

So interpreted, the argument has two significant limitations. First: it does not establish that spatial objects exist in the way that Descartes thought they did. Descartes thought spatial qualities were primary qualities of objects, and made no appeal to criteria of experience. In Kant’s terms, he was a transcendental realist. Moreover, Descartes held (e.g. CSM, II. 55) that only an appeal to God’s existence can show that spatial objects exist. So Descartes might not have regarded the argument as a satisfactory answer to scepticism. Nonetheless, Kant aims to refute the Cartesian sceptic, not the actual Descartes. This sceptic, in the first two Meditations, takes no stand on whether spatial qualities are secondary. He does, in Meditation I (CSM, II. 14), suggest that simple, general geometrical qualities are ‘real’, and that geometry contains something ‘indubitable’. These statements are neutral about the primary vs. secondary metaphysical status of space, however (cf. B41, B70–1).

Second: the argument leaves open what the spatial object is in itself, that is, what the thing is that affects the mind and gives rise to spatial perceptions. The object could be something with primary spatial qualities, an all-powerful evil genius, God, or some unknown part of ourselves. This last possibility is especially disappointing if we were hoping Kant’s Refutation would rule out the frightening Twainian scenario. Unfortunately, Kant seems open to something like this possibility. He grants that we can affect ourselves (see e.g. B67–8), and insists on our ignorance about whether we differ in ourselves from outer things in themselves (see A379–80, B409, B427–8). This is the main reason I do not claim that my reconstruction will assuage all feelings of disappointment. Yet we should remember that all Kant explicitly promised in the Refutation was a proof of the existence of real things in space. Given his idealism, that conclusion leaves open whether, at some deeper level, we (or, even worse: you) are all alone.

That said, Kant may have further resources for lessening our disappointment. First of all, his confidence that thought must have a deeper ground shows that he would reject the Twainian possibility of the self being a lonely thought. That is a consequence of Affection, which requires some perception-independent entity. Second, Kant never says that I might be the ground of outer appearances, and he denies that this ground is a thinking being in itself (A379–80). Given our limited knowledge about things in themselves, why is Kant so confident about this?

The answer, I suspect, is that Kant has an account of the self that makes it impossible for the self to be identical to any thing in itself. Elsewhere, I have argued that for Kant the self is a thing or things insofar as that thing or things perform acts of judgement and synthesis that unify some course of experience (by analogy, a fist is a hand insofar as it is clenched).Footnote ⁴² Say, then, that there is a single thing that gives rise to our perceptions and to our unifying thoughts/syntheses. That thing is not the self, nor is the self that thing insofar as it gives rise to mere perceptions, even though all of these entities metaphysically overlap (in the way that my hand, fist and open hand metaphysically overlap). Given this account, then, the self cannot be completely alone. This is a broadly Spinozist view: the self is not a fundamental entity, but instead shares an ultimate ground with other things. To be sure, if we are looking for metaphysically non-overlapping company, we will still be disappointed with this conclusion.Footnote ⁴³ But Spinozism is very different from solipsism.

The above non-Cartesian account of the self would not beg the question against the Cartesian sceptic since the sceptic never considers this kind of insofar-as metaphysical construction. Descartes himself would have rejected this account, though, and we might reasonably ask whether Kant has arguments for his account that would not beg the question against Descartes. For now, though, the point is that Kant may have further grounds for denying that we are all alone.

9. Further Merits of the Argument

Kant’s proof, so reconstructed, has several philosophical virtues. It is valid, it does not beg the question against the Cartesian sceptic, and its conclusion is not Berkeleyan. Nevertheless, a bad argument could have all these virtues. In this section, I focus on two further evaluative topics: the complex structure of the argument and the plausibility of its crucial premises.

The argument has a structural vice: it could easily be shorter. It could have instead started with the claim that we have at least one perception as of a spatial thing. Together with Affection, Minimal Coherence and Perceived Space, that would yield the desired conclusion. Moreover, assuming at least one perception as of a spatial object would not beg the question against the Cartesian sceptic, since the sceptic accepts as much in the Meditations. If my reconstruction is right, then, why did Kant not give a shorter argument?

One thing to say is that Kant seems proud of the fact that his argument requires only one empirical assertion, namely, that I am. The assertion that we have at least one spatial perception would also be empirical, by his lights. So Kant seems to think that refuting the Cartesian sceptic requires not making empirical assumptions about spatial perceptions. This may be because the perception of spatial objects is not as central to Meditations I and II as awareness of temporally determined thinking. Moreover, the argument is more philosophically impressive if its only empirical premise is about something that is not even perceived as spatial. That may also be crucial to the second half of the argument which, as noted in section 3, I have set aside. So there are virtues to the longer argument, even if a shorter argument would have worked.

The next question is the plausibility of the argument’s premises. A complete defence of the Refutation would have to fully justify all its premises. My aim now is more modest: to show that the above premises have at least some plausibility. Premise 1 (that I am aware of my existence as determined in time) seems safe enough, provided the notion of determination is understood in some straightforward way. For reasons explained above, Minimal Coherence also seems fairly plausible, though that hinges on what criteria of experience are supposed to be. That leaves us with Premises 2 and 8, along with Affection, Perceived Time and Perceived Space.

Premise 2 states that any awareness of existence as determined in time requires a perception as of something persisting. Premise 8 states if there is at least one real, perceived persisting object, then there is at least one perception as of a spatial object. Both claims are hard to assess. At a minimum, though, they are consistent with ordinary experience. We typically do have perceptions as of something persisting and perceptions as of something spatial (e.g. the feeling of our own bodies). Moreover, it is hard to imagine experience without such perceptions. Consider Guyer’s case of images with ‘digital timers in their corners’, presented as a counterexample to Kant’s claim that time determination requires spatial objects. Such a digital timer in our representations would still be a representation as of a persisting, spatial thing.Footnote ⁴⁴ Getting a clear counterexample to these premises requires stretching our philosophical imagination further, towards a scenario where we are aware only of the temporal order of our thoughts (cf. Anscombe Reference Anscombe1994). Such far-flung thought experiments are hard to assess, and this shows that Premises 2 and 8 have at least some plausibility. That said, it is far from obvious how one would establish their truth.Footnote ⁴⁵ If my argument so far has been successful, then perhaps the most daunting challenge in providing a complete defence of the Refutation would be justifying these premises (though they would need to carry less weight on my reconstruction than on some others).

The next premise to consider is Affection. I have left open whether Kant has deeper reasons for accepting this claim. Kant’s own beliefs aside, it is hard to see how to prove it. Nonetheless, it has some intuitive plausibility. Consider how much easier it is to motivate scepticism with brain-in-vat scenarios than with a scenario in which perceptions come to be acausally. Even radical idealists like Berkeley accepted Affection. Of course, an argument using that premise would beg the question against the most radical sceptics – but those sceptics are not Kant’s target.

Above, I suggested that Perceived Time and Perceived Space can be motivated by a complex secondary-quality view of spatiotemporal qualities. Unlike Affection, this view is probably not part of common sense. However, the secondary-quality view can be understood as a view about some spatiotemporal qualities. We seem to perceive space as three-dimensional, and time as involving absolute simultaneity relations. Post-Newtonian physics gives us reasons to think that other spatiotemporal qualities are primary, such as 11-dimensional space or time as involving only relativistic simultaneity. Perceived Time and Perceived Space do not imply that all spatiotemporal properties are perceivable, despite what Kant thought (see e.g. A42/B59). If so, then the secondary quality view might be the closest we can come to saving the commonsense view of familiar spatiotemporal qualities (cf. Sellars Reference Sellars1968).

I therefore claim that all of the premises have at least some plausibility. Still, for all I have said, Kant’s argument might be unsound. In that case, he would not have refuted the Cartesian sceptic. Nonetheless, if the premises are sufficiently plausible, then we have reason to think that the Refutation is successful until shown otherwise.

There are further questions that could be raised about my reconstruction, including how it compares to other interpretations and whether it remains too idealist-friendly. I hope to have shown, though, that the argument has significant virtues. Disappointingly, it cannot defeat the most radical sceptics – nor even all sceptics who have been labelled ‘Cartesian’. However, it may refute the specific form of scepticism described in Descartes’s Meditations, and that would be a significant achievement on Kant’s part.Footnote ⁴⁶