LUCY ALLAIS | Manifest Reality: Kant’s Idealism and his Realism | Oxford University Press 2015


 

By Colin Marshall 

Lucy Allais’s Manifest Reality is a terrific book, comparable in importance to landmark works like Henry Allison’s Kant’s Transcendental Idealism and Rae Langton’s Kantian Humility. In my view, the interpretation of Kant’s idealism Allais offers is significantly more plausible than either Allison’s or Langton’s. Not just that, but Allais’s presentation of the issues is thorough and exciting. Both students and seasoned Kant scholars have much to gain from Manifest Reality.

Since most of this review is devoted to challenging various parts of Allais’s account, I want to start by describing two of the features that, in my view, make this book so good. First, I think Allais offers a compelling middle ground between two competing approaches to Kant’s distinctive form of idealism. In some passages, Kant really sounds like a realist, while in others, he really sounds like an idealist. Allais provides a reading that lets us take all the relevant passages seriously, instead of antecedently prioritising one sub-set of passages and trying to deflate the other.

Second, Allais’s methodology shows how productive the intersection can be between contemporary and historical philosophy. Here again, we find a compelling middle ground between two extremes. On the one hand, some historians simply impose contemporary philosophical frameworks on historical texts. This approach runs the risk of anachronism, and ignores the possibility that historical philosophy might provide unexpected challenges to contemporary views. On the other hand, some historians refuse to draw any explicit connections between historical and contemporary philosophy. This latter approach also precludes the possibility of finding useful challenges to contemporary thought and, more often than not, the historian’s own philosophical sensibilities end up sneaking in implicitly anyway (particularly, I think, in how he or she reads ambiguous passages).

Allais avoids both extremes. She uses resources from contemporary philosophy of mind to expand our sense of interpretive options, but simultaneously lets textual considerations motivate interpretations that cut across familiar contemporary philosophical categories. Interpreters who disagree with Allais on textual grounds are faced with long, detailed lists of passages they must wrestle with. Not just that, but Allais maintains a constant focus on Kant’s larger motivations as a guide towards making sense of the most difficult passages. By contrast, interpreters who disagree with Allais on philosophical grounds are faced with the challenge of justifying their alternative philosophical approach in as much detail as Allais provides for hers. Either way, Allais’s book promises to significantly elevate the discussion of Kant’s idealism on all fronts.

Some readers will be bothered by how little attention Allais devotes to Kant’s development or his intellectual context. She barely mentions Kant’s pre-Critical work, makes no mention of Wolff, Crusius, or Baumgarten, and discusses Leibniz only via Kant’s own discussion in the Amphiboly chapter. My suspicion, however, is that attention to the historical context would reinforce Allais’s reading on a number of points. Even if that suspicion is wrong, however, I do not know of any extant work on Kant’s development or predecessors that would seriously threaten Allais’s central claims. If other historians of philosophy can use context to show that Allais is mistaken, therefore, the result would be a significant improvement in our current understanding of seventeenth century philosophy.

I now turn to more critical remarks. I find the majority Allais’s textual claims persuasive, so I focus more on the coherence of and philosophical plausibility of her interpretation. The aspects of Allais’s reading I call for scrutiny are as follows: the identity/sameness relation between appearances and things in themselves (Section 1), the nature of ‘essentially manifest qualities’ (Section 2), the relationship between sensation and intuition (Section 3), three neglected possibilities in her account of Kant’s argument in the Transcendental Aesthetic (Section 4), and her appeal to the generality of thought in the Transcendental Deduction (Section 5).

1. The Sameness or Identity of Appearances and Things in Themselves

Allais’s reading falls within the tradition of one-world/two-aspect interpretations of Kant’s idealism (though she now has reservations about those labels). The central claim in this tradition is that each appearing object of experience is, in some sense, the same thing as some thing in itself.[1] Some interpreters in this tradition construe this sameness claim in terms of numerical identity, saying that each appearance or appearing object is identical to some thing in itself. However, this identity claim generates at least two problems. First, Kant seems to ascribe incompatible properties to appearances and things in themselves, in particular, spatiotemporality. Second, we would expect Kant to deny that we could possibly know how things in themselves are individuated. But if each appearance is identical to some thing in itself, and we know how appearances are individuated, then knowledge about the individuation of things in themselves would seem to be within easy reach.

Allais is aware of these potential problems. For example, she writes that “to say that the things we experience have a way they are independently of their appearing to us is not to say that picking out objects at the level of experience tells us how they are individuated in themselves” (p. 73). Now, one way to avoid such problems would be to just deny the identity claim in question. Allais does not do this, however. Instead, the main argument she offers to block the individuation worry is as follows:

Normally, when we ask with respect to two putatively distinct things whether they are in fact numerically identical we have some way of individuating and referring to each […]. On Kant’s account of cognition, the only way we have of individuating or picking out things is at the level of spatio-temporal appearances […]. We cannot pose a question about the numerical identity or distinctness of things in themselves and appearances without having some way of identifying the objects that are things in themselves. (p. 71, see also p. 242)

It sounds like Allais wants to avoid taking a stand on the identity of appearances and things in themselves. I do not think this is a good way for Allais to deal with the potential problems. Instead, I believe, she should just deny that appearances are identical to things in themselves.

Presumably, the sort of ‘normal’ identity question Allais has in mind is something like: “Is the morning star identical to the evening star?” In this case, we do indeed have two independent ways of individuating and referring to the relevant object, namely, as a certain celestial object appearing at morning and as a certain celestial object appearing at evening. But this is a feature of only one type of informative identity question. We can also ask whether the morning star is identical to the morning star. Normally we don’t ask this sort of question, because the answer is obvious, but that questions with obvious answers are still meaningful. So it is not true that meaningful identity questions require anything like independent ways of individuating the relevant thing(s).

An analogous case can get us closer to the relevant issue, however. Consider the question: “Are Ted Cruz and Hillary Clinton each identical to one of God’s creations?” The answer to that question may or may not be obvious. But say that we know that every US presidential candidate is a creation of God. Then it will be true that Cruz is identical to one of God’s creations, and Clinton is as well. That all makes sense even if we have no independent way of individuating God’s creations. We might come to know the individuation fact that God has at least two creations, then, from knowing that Cruz and Clinton exist, that they are not identical, and that each is identical to one of God’s creations. Similarly, on the basis of knowing that Cruz and Clinton each have the property of being a presidential candidate, we could come to know that two of God’s creations have the property of being a presidential candidate.

The above example has the same structure as Allais’s claim about appearances and things in themselves. We have the general description ‘things in themselves’ that apparently can be used to refer to things (since, after all, Allais claims Kant commits himself to there being things in themselves). That, like the general description ‘God’s creations’ is enough to generate the worries about individuating things in themselves on the basis of appearances and about things in themselves and appearances having different properties.

Now, my hunch is that Allais herself does not really think that, in the strict sense of ‘numerical identity’, appearances might be numerically identical with things in themselves. For example, in the same section in which she makes the above claims about questions of identity, she writes:

[W]e still have the problem of contradictory properties being attributed to the same things. The response to this is that there is no contradiction in saying that things have properties in certain relations which they do not have independently of these relations. (p. 73)

Later in the book, moreover, Allais states that “Kant thinks […] that the essentially manifest qualities of things we are able to cognize exhaust the characteristics of spatio-temporal empirically real objects” (p. 135). It is hard to see how something whose characteristics are exhausted by essentially manifest qualities (which Allais takes to be mind-dependent; see below) could be numerically identical with objects with a mind-independent existence.

To see how Allais’s view might be understood without any problematic identity claim, consider another example of a relational property. I have the property of being an older brother. I have this property in virtue of having a younger sibling. Independently of my relation to any sibling, though, I do not have the property of being an older brother. But this does not mean that there is someone who is not an older brother who is identical to someone who is an older brother. Instead, we might reasonably say, there is just one person here, who is an older brother, but who also has some properties independently of his relations to any sibling. Carrying the analogy over: saying that each appearance is ‘the same thing as’ a thing in itself can be understood as simply saying that the things that have appearance-properties also have properties in themselves. Likewise, saying that things in themselves are not spatial would be simply saying that spatiality is not among the properties objects have independently of their relations to us. There would be no question of identifying distinct entities in this case.

Alternatively, if we are impressed with Kant’s tendency to talk of appearances and things in themselves as different entities, we could do some subtle metaphysics here. Sticking with the above example, we could say there are two non-identical but overlapping objects in the brother case, only one of which has the property of being an older brother. There is me-qua-brother and me-qua-independent-of-siblings. We could model this on the example of the relation between a hand and a fist—a fist is essentially clenched, while a hand is not, so no hand is identical to a fist (this is the solution I offer in Marshall 2013). On either of these approaches, there would be no need to get into issues about ways of individuating objects in identity questions, as Allais does.

I should note that there is a more radical option available as well. One could insist that each appearance is numerically identical to some thing in itself, and yet deny this relation supports inferences about individuation and properties (that is, deny the Indiscernibility of Identicals). To me, it seems like this would be changing the topic away from numerical identity, but perhaps I have not properly grasped how metaphysics breaks down when it comes to things in themselves. Even so, it would take some serious exegetical work to motivate attributing such a view to Kant. Nowhere does Kant suggest there are limits to the Indiscernibility of Identicals.

2. Two Challenges about Allaisian Essentially Manifest Qualities

At the core of Allais’s interpretation of Kant’s idealism is the notion of an essentially manifest quality. These qualities

(a) are properties of objects that are (at least partly) mind-independent, contra phenomenalism (see pp. 34, 53, 103);

(b) “manifest themselves to us in perception” (p. 120),

(c) are “feature[s] the objects have in relation to the possibility of their being experienced by minds of a certain sort” (p. 116, emphasis added).

Given (c), a more apt label for these qualities might be “essentially manifestable qualities”, which Allais occasionally uses (pp. 13, 207). Together these features are meant to provide a middle ground between extreme realist and idealist understandings of Kant, and to do so without invoking something mysterious (like Henry Allison’s ‘epistemic conditions’). Feature (a) distances Allais’s interpretation from phenomenalist interpretations, while feature (c) distances her interpretation from non-idealist interpretations.

Setting interpretive issues aside, I want to raise two challenges to this notion. The first challenge asks: given the dependence on our minds, how are these qualities of objects in the strong sense Allais wants? The second challenge asks: given that these qualities are manifest to us in perception, how can we make sense of them depending on the possibility of their being experienced by minds? I use the word ‘challenge’ here deliberately. I suspect both challenges can be answered, but that doing so will require Allais to further develop the philosophical side of her view.

2.1. Challenge 1: Why think essential manifest qualities are qualities of mind-independent objects?

Allais claims that the relevant qualities are genuine properties of objects with some sort of mind-independent existence (though see below). This is meant to hold despite the fact that these qualities are mind-dependent in a certain sense. Though the possibility of mind-independent things having mind-dependent properties seems to have been overlooked by some commentators (see e.g. Van Cleve 1999:5), it is coherent. If there is a tension between mind-dependence and mind-independence here, it is slight. Consider a fairly straightforward example: a Lockean could agree that the mind-dependent property of being green is a property of mind-independent spruce tree. What makes this easy to conceive is that we can at least partly understand how this mind-dependent property could arise from some combination of features of minds and mind-independent features of the spruce, even though some details of that relation elude us. That is, the spatial features (which are mind-independent, for a Lockean) have some role in generating or constituting the property of being green. We cannot use straightforward examples like this to make sense of Allais’s reading, however, since the picture she offers is one on which we have no grasp of what the mind-independent features of things are.

Let me describe this challenge from a slightly different angle. Consider a traditional early modern view of the relationship between things and their properties or modes. Descartes, for instance, writes:

I find in myself faculties for certain special modes of thinking, namely imagination and sensory perception. Now I can clearly and distinctly understand myself as a whole without these faculties; but I cannot, conversely, understand these faculties without me, that is, without an intellectual substance to inhere in. This is because there is an intellectual act included in their essential definition; and hence I perceive that the distinction between them and myself corresponds to the distinction between the modes of a thing and the thing itself. (AT 7:78 [CSM II:54])

Descartes here seems to be saying that at least part of what makes his faculties of imagination and sensory perception modes of his mind are: (i) that their existence depends on his intellect/mind, but not vice-versa and (ii) that their “essential definition” involves reference to something mental (an “intellectual act”). Along similar lines, Spinoza defines a mode as “that which is in another through which it is also conceived” (Ethics 1d5). We would expect something similar to be true in the Lockean story about the relation between spatial properties and colour. Colours depend on spatial properties for a Lockean, but not vice-versa, and (somewhat more tendentiously) the essential definition of colour would have to involve reference to spatial properties. This manifests in some obvious ways in our experiences of colours: we cannot conceive a colour that is not on the surface of an object. Generalising, the idea is that metaphysically superficial properties of objects (modes) must asymmetrically depend on metaphysically deeper features of those same objects. This asymmetric dependence is not just existential, such as the claim that modes would not exist if there were no deeper features. Rather, it is the idea that modes arise from deeper features, and so are shaped by them in certain ways (hence the claims about essential definitions and conception). Descartes and Spinoza probably held that such asymmetric arising is both a necessary and sufficient condition for a mode to inhere in an object, but I am only concerned with it as a necessary condition. Let’s call the claim that asymmetric arising-from is a necessary condition for something to be a mode of an object the Inherence-Dependence Principle.

To see why this principle is appealing, consider one way of explaining the difference between hallucination and misperception. When I misperceive something as green, there must be an object in front of me that has the property of appearing green. Not so when I hallucinate something green—in that case, there might be no real object that has the property of appearing green (see Allais, p. 41). What makes the difference? Presumably, my state counts as a misperception only if the green-appearing arises from real properties of the real object, such as its location, shape, and proximity (e.g. there really is a cat there, he’s just not really green). On the other hand, if my apparent perception of something green does not arise from real properties of any real objects in my perceptual field (e.g. the only things in that location are air particles), then I am merely hallucinating, and no real object has the property of appearing green. This is part of what the Inherence-Dependence Principle is meant to capture.

Allais seems to hold that essentially manifest qualities are (relatively) metaphysically superficial properties of objects that arise or “flow from” (p. 226) their metaphysically deeper features. Against Rae Langton’s interpretation, for example, she says that “Kant does not just say that things have powers and that, in addition, they have intrinsic natures; rather, he says that the way things are in themselves is responsible for their relational appearances” (p. 245, see also p. 252). According to the Inherence-Dependence Principle, that means that essentially manifest qualities must depend on metaphysically deeper properties of those objects (such as their intrinsic natures). The problem, however, is that we are supposed to have no specific knowledge of these metaphysically deeper properties—this is how Allais understands Kant’s prohibition against our having cognition of things in themselves. Kant, she writes, thinks that “not only can we not know things’ intrinsic natures, we cannot understand how they ground appearances, or even what kind of relation this grounding might be” (p. 232). The problem I am describing for Allais is not a metaphysical problem, since there might be unknowable or uncognisable dependence relations between properties. Instead, it is an epistemological problem: without knowing what the metaphysically deep properties of mind-independent objects are, on what grounds can we say that essentially-manifest qualities are properties of those objects? That is, how can we know that the Inherence-Dependence Principle is satisfied for any of the properties we encounter in experience?

This challenge is distinct from the challenges about the justification of our beliefs that there are mind-independent objects, that those mind-independent objects have mind-independent properties, and that those mind-independent properties ground the properties we perceive (Allais addresses these latter challenges in Chapter 10). Returning to the hallucination/misperception analogy: you could know that there are mind-independent objects around you and know that those objects have mind-independent features and yet still wonder whether you are perceiving them, misperceiving them, or just hallucinating properties that do not belong to any real object. Nor would it be enough for you to rule out the possibility of hallucination if you knew that your states were caused by mind-independent objects. An object that causes dizziness does not thereby have the property of being dizzy or appearing dizzy (this is Allais’s example on p. 119). Rather, the only thing with the property of being dizzy might be you.

The question for Allais, then, is what grounds she thinks there could be, on Kant’s view, for taking the essentially manifest properties to be properties of mind-independent objects (as opposed to, say, merely being properties of us). Is this another one of Kant’s philosophical starting points, alongside the assumption of mind-independent objects with mind-independent features? Is there some explanation available to us? Or does Allais think Kant would reject the Inherence-Dependence Principle?

2.2. Challenge 2: How can essentially manifest qualities require merely the possibility of perception?

On a simple phenomenalist view, objects have properties only when those properties are perceived. That sort of radical mind-dependence is counterintuitive, so most phenomenalist readers of Kant have instead told a more complex story, according to which merely possible perception is sufficient for an object to have a property. Similarly, Allais holds that essentially manifest properties can exist without actually being perceived. They depend only on the possibility of being perceived. The sort of possibility in question involves a link to some actual perception. Hence, Allais writes that “Kant links empirical reality to possible experience” (p. 141) and that “anything which counts as part of possible experience must be causally linked to something we do actually perceive” (p. 144).

Allais view is ambiguous on this final point. Say (bracketing the idealism about time) that divine vengeance for idolatry wipes out all perceivers. Would that mean the end of all essentially manifest qualities? After all, there were some actual perceptions at some point, and there could still be some causal connections between future objects and past perceivers. Since Allais leans towards common-sense views, I shall assume that she thinks some connection to actual perceptions at some point or other are sufficient for empirical reality. In that case, there could be objects with essentially manifest qualities even if all perceivers ceased to be.

The challenge I want to raise here is subtler than the one in the previous sub-section. It concerns the notion of manifestation or revelation that Allais uses. Now, I believe these notions are very philosophically important, and Allais’s use of them is one of my favourite features of her interpretation. However, I think a problem arises here. Revelation (which I take to be equivalent to manifestation) is supposed to be metaphysically informative. Presumably, that is why appeals to revelation in the philosophy of colour have been used to resist certain reductive views, as Allais herself notes in Chapter 5. The challenge to Allais comes from the fact that nothing in the phenomenology of colour or spatiotemporal qualities suggests that they have the metaphysical profile Allais ascribes to them. If anything, the phenomenology seems to suggest a quite different metaphysical profile. This is why, in a defence of naïve realism about perception, Heather Logue writes that “the phenomenal character of veridical experience gives its subject insight into what things in one’s environment are like independently of one’s experiences of them” (Logue 2012:227; for a helpful general discussion about colour and revelation, see Maund 2012).[2]

As Allais describes them, essentially manifest qualities are at least something like dispositional properties. Though their nature makes reference to perception, objects can have these qualities even when not actually perceived. This makes them something like properties such as being nauseating. Presumably, something is nauseating only if it would cause nausea in someone under certain circumstances, but it can be nauseating even when it is not causing nausea in anyone. In the latter sort of case, its existence seems to consist in mere possibility, and this possibility ‘bursts into action’ in certain circumstances. Allaisian essentially manifest qualities seem to work roughly the same way. When not perceived, their existence consists in mere possibility, but they burst into action when perceived. Yet the phenomenology of colour and spatiotemporal properties in no way suggests this metaphysical profile. Rather, colours and spatiotemporal properties seem like stable properties of objects that always exist as more than mere possibilities. If that phenomenology is supposed to arise from the revelation of the properties, this is not what we would expect if Allais were right.

To be sure, I don’t imagine anyone would say that revelation must tell us everything about a property. Otherwise, nobody could think that colours are revealed in perception while thinking that we need subtle arguments to figure out the metaphysical status of colour. Nonetheless, the more an experience doesn’t tell us about a property, the weaker its claim to count as revelation. And for Allais, the revelation of essentially manifest properties seems to tell us very little metaphysically. It tells us nothing about the underlying mind-independent properties of objects (unlike, say, the relation between colour and shape on a Lockean view, where the colour appears on the surface of the relevant objects), nor does it tell us that these properties are dispositional/essentially manifestable. So it is hard to see in what sense we could plausibly say that those properties are manifest or revealed to us.

Of course, we might well be wrong about the metaphysics of many qualities. Perhaps, moreover, I am taking Allais’s talk of ‘revelation’ too strongly. There is, however, a reason for thinking that, for Allais, the nature of essentially manifest qualities is revealed to us in perception in some strong sense: without this, it becomes harder to make sense of the claim that we are directly acquainted with properties of objects in perception. If the nature of these properties is largely hidden from us in perception, then it does not seem we are really in touch with those properties via perception.

I shall finish this section with a related question. Say that we grant that the revelation of essentially manifest qualities does not tell us that they are essentially manifestable in the particular way Allais describes. In that case, what exactly would Kant’s grounds be for taking those properties to be essentially manifestable in one way rather than another? To make the question concrete, consider the property of being spherical. Here are four metaphysical options for how sphericality might depend on minds:

  • An object is spherical only if it is actually perceived.
  • An object is spherical only if it is actually perceived, was actually perceived, or will be actually perceived.
  • An object is spherical only if it is connected to an actual perception, regardless of whether it is actually perceived. (Allais’s official view.)
  • An object is spherical only if it is connected to a possible perception, regardless of whether it is connected to an actual perception. (suggested by some of Allais’s formulation; see e.g. p. 211.)[3]

If our revelatory experience of sphericality does not decide between these, for Kant, what would decide? Allais does not directly address this question, but her interpretation might seem philosophically ad hoc without an answer to it. Here are two factors we should keep in mind. First, if we have reason to make Kant’s views about empirical reality close to common sense, then the fourth option should win out. For that option would best let Kant respect the common-sense view that spatial things could exist even without any perceivers. Second, as Allais discusses (pp. 93–4, 215–16), Kant’s idealism is meant to allow that the magnitude of the spatiotemporal world is indeterminate. Presumably, that would have to mean that there are some indeterminate facts about spatial properties. However, it seems like the facts about what is connected to possible perceptions via causal laws (‘actual possibility’, in Allais’s sense) would be determinate. And it seems like there would be a fact about whether the set of things connected to perceptions was infinite or finite—nothing about the notion of a perception-dependent quality (in this sense) suggests otherwise. Given indeterminacy about temporal facts, the second option above looks more promising to me on this front. Regardless, it would be helpful to hear more about Allais’s views on how Kant decided on the particular form of idealism about properties he accepted, given the range of possible views.

3. The Role of Sensation in Empirical Intuition

A central part of Allais’ interpretation is her understanding of intuition as acquaintance with particular objects. Overall, I find her account of intuition attractive. Several aspects of that account raise questions, though. I shall leave most of these questions to the other commentator here. There is one issue, however, that I would like to discuss: the relationship between intuitions and sensations. Here, I think Allais opts for a position that creates unnecessary obscurity.

Allais claims that, in the Critique, “Kant’s dominant use of ‘sensations’ indicates that they are non-intentional or non-referential” (p. 161). Sensations, she claims, are a “mere causal, sensory input from objects” (p. 160). Allais goes to some lengths to emphasise the difference between sensations and intuitions, since intuitions, unlike sensations, are supposed to be intentional. The view she most wants to resist is that intuitions arise from the operation of concepts on sensations (‘conceptualism’). According to Allais, intuitions exist and present their objects independently of concepts. How, then, does she think sensations are involved in intuitions? Allais states that, for Kant, “empirical intuitions are the outcome of sensation being ordered in the a priori forms of intuition [i.e., space and time], and it is intuitions, not sensations, that give us objects” (p. 160).

When combined with Allais’s other interpretive claims, this view about intuitions and sensations is puzzling, though. Allais claims that intuitions directly present their objects to the subject, and that these objects are things that exist outside the mind (contra phenomenalism). How, though, could ordering something non-intentional (sensations) in a mind-dependent form produce a direct acquaintance with a mind-independent object? The forms of space and time alone do not present us with particular objects, so how could adding something non-intentional to them make the difference? If anything, it sounds like the nonintentional nature of sensations would block our direct access to mind-independent things, creating what Jonathan Bennett famously called a “veil of perception” (Bennett 1971:69).

Fortunately, I think Allais’s view can be easily modified on this point in either of two ways, each of which would eliminate (or at least reduce) this obscurity. First, instead of saying that intuitions are just organised sensations, she could say that sensations are the subjective side of intuitions. Consider the following analogy. Imagine that a blanket is covering two sleeping cats (call them ‘Lewis’ and ‘Mr. K’). The covering relation here is immediate, since the blanket does not cover Lewis and Mr. K by covering something else. In order to cover the cats, though, the blanket must have certain properties that do not essentially depend on its relation to the cats, such as being of a certain size and shape. The non-cat-relational properties are the blanket’s side of the covering relation, that is, the properties it has that allow it to cover the cats. Similarly, we might take Kant’s view to be that the immediate intuition relation between subjects and objects has a subjective side, and that sensations are that side. Just as the blanket’s non-relational properties in no way block its covering Lewis and Mr. K, sensations would in no way block our acquaintance with objects.

The second way that Allais’s view might be modified here would be to take the sensations to be constituents of the relevant essentially manifest qualities. If so, then the fact that sensations are included in intuition would be perfectly compatible with intuition directly presenting us with qualities of objects. This is, I take, one of the ideas in some phenomenalist readings of Kant. But I do not think the idea requires phenomenalism. For these properties need not be only constituted by sensations. They might also involve the (mind-independent) object in some essential way, such that the property would cease to be if the object disappeared and only sensations were left. It would take some work to spell out that account of the metaphysics of qualities, but if it can be done, then it would provide a less obscure way to locate sensations in Allais’s account of intuition.

There is another puzzle in this same area, however, and I am not sure whether either of the above modifications alone would be sufficient to address it. Allais claims that, according to Kant “we organise the sensory input into perceptual units using our forms of intuition (space and time), as well as biologically determined principles of association” (p. 275, cf. p. 279). Presumably, “perceptual units” are things like our perceptions as of two cats. These perceptual units are, according to Allais, divided up on the basis of internal facts about the mind, concerning our forms of intuition and principles of association. However, according to Allais’s direct view about perception, it sounds like intuitions are divided up on the basis of external facts, concerning the individuation of objects external to the mind. On this latter line of thought, it would seem that there would be (roughly) one intuition for each object that affects the mind. The internal/external contrast here opens up the possibility that perceptual units do not line up with intuitions. That is, it seems that there might be one intuition that is divided up into two (or twenty) perceptual units (or vice-versa). This would be somewhat puzzling in itself, but it becomes ever more puzzling if this is supposed to be compatible with something like a direct realist story. If perceptual units can diverge radically from the number of objects, then it is hard to see in what sense our perceptual experience really involves objects being directly presented to the mind.[4]

4. Three Neglected Alternatives

In Chapter 8, Allais argues for a reading of Kant’s argument in the Aesthetic about the ideality of space and time. Her central proposal is that Kant begins by arguing that we have a priori intuitions of space and time, and concludes from that that space and time are mind-dependent. This argument is valid, she claims, because intuitions involve the actual presence of their objects, and the only way an a priori intuition could present its object were if the object were mind-dependent in some way. I think that, at a minimum, Allais’s reconstruction of Kant’s thought respects the text at least as well as any of the other major proposals, and that it does significantly better than many other reconstructions. However, at three places in her discussion, she seems to neglect important alternatives to certain claims. This is unfortunate, since one of the issues she takes herself to be addressing is the famous neglected alternative objection to Kant, which holds that Kant overlooks the possibility that space and time might be both forms of intuition and features of things in themselves.

4.1. First neglected alternative: Transcendental arguments and using idealism/self-knowledge

Part of Allais’s argument for her reading is her attack on other interpretations of the Aesthetic. The interpretation Allais most opposes hinges on the slogan “we can have a priori knowledge of what our minds make” (p. 180), and then appeals to our synthetic a priori mathematical and metaphysical knowledge to establish idealism. On this interpretation, Kant would be using idealism to explain our knowledge (to close an ‘explanatory gap’), and would be resting heavily on our having special awareness of mental features along the way. Allais offers five quick objections to this interpretation. One of these objections calls for special attention.[5] According to her third objection,

if idealism were introduced to close an explanatory gap, the argument would be vulnerable to other possible explanations of how this gap could be closed […] Kant himself has an alternative account of the source of the justification of synthetic a priori metaphysical claims: his transcendental arguments. (p. 181)

Similarly, when referring back to the interpretation she sees as objectionable, Allais writes that our insight into claims about spatiotemporal objects “is not a matter of insight into how our minds make things, but insight into the conditions of cognition” (p. 302), as established by transcendental arguments. Here, Allais follows many other interpreters in thinking that transcendental arguments can provide a distinctive source of justification that are less problematic than appeals to insight into our own minds. I think this is a mistake, however. Transcendental arguments can just be formulations of other sources of justification, including insight into one’s own mind.

As Allais describes it, a transcendental argument “defends some doubted claim by showing it to be a necessary presupposition of something which is not doubted” (p. 182). This is more or less the canonical description of a transcendental argument. However, note that this description says nothing about what the claim to be defended is, or how one shows that something is a necessary presupposition of something else. In this case, the what could be idealism, and the how could be some sort of reflection on the nature of the mind. Of course, there are other transcendental arguments that involve different what‘s and how‘s, and Kant might well have some of those to offer. But calling something a transcendental argument by itself tells us almost nothing about that argument is supposed to work, and so what other sorts of arguments it might be an alternative to. For that reason, a proponent of the interpretation Allais rejects could simply say that the interpretation is about a transcendental argument. Transcendental arguments do not offer an alternative.

4.2. Second neglected alternative: Intuitions and quasi-intuitions

As Allais understands it, Kant’s argument in the Aesthetic rests heavily on his notion of intuition, which Allais understands as requiring the immediate presence, and so existence, of its object. So if we have an intuition of space, then space must be present to us. And if, as Allais takes Kant to argue, that intuition is a priori, then space cannot be something mind-independent.

I think the textual case for this interpretation is significant, and that it probably captures at least part of what Kant had in mind. However, it opens up a new neglected alternative objection to Kant. Consider the possibility that there is some mental state that is, from the subject’s point of view, just like an intuition, but that does not require the presence of its object. This need not be a radical sceptical possibility (like a Cartesian deception). Instead, we might just suggest there is some way we have of retaining the subjective, qualitative experience of mind-independent features that had been presented in experience. Call this a quasi-intuition. The threat of quasi-intuitions would seem to threaten Kant’s argument just as much as the original neglected alternative did. Even if it is part of the notion of intuition that it involves the real presence of its object, but that then just pushes the question back to: what grounds do we have for thinking we have a priori intuitions of space instead of quasi-intuitions?

I’ll note that a related question can be raised for Allais’s claim about Kant’s use of the conceptual truth that “where there is something that appears in a certain way to us there is something that has a way it is in itself” (p. 69). To use this conceptual truth to conclude that things have a way in themselves, one would have to have grounds for applying the relevant concept of appearance, instead of, say, the concept of a quasi-appearance (which is something subjectively indistinguishable from an appearance but which does not require that there be a way the quasi-appearing thing is in itself). A similar point holds for Allais’s claim that one “reason[] for thinking that there are intrinsic, non-relational natures […] [is that] they are part of our conception of what it is for a thing to exist” (p. 248). Even granting that, why take ourselves to be in a world of things, instead of quasi-things without intrinsic natures?

4.3. Third neglected alternative: Extended structure of space

On the traditional neglected alternative objection, one asks why space and time couldn’t be both features of things in themselves and objects of our a priori intuitions. While Allais regularly talks of the ‘objects’ of intuition, she notes that

space and time (what a priori intuition presents us with) are not objects in any obvious sense […] ‘object’ should be understood in an extremely general sense […]. Space and time are the structures or forms within which empirical intuitions are arranged. (p. 197)

Note the focus on structures here. When Allais then turns to the traditional objection, she writes that

our representations of space and time do not present us with mind-independent features of reality. It is compatible with this that there should be some structure in things as they are in themselves; the point is that this structure is not what it is that is present to us when we represent space and time. (p. 198)

If this avoids the traditional objection, then it faces a nearly identical one, however. For, unlike particular objects, structures can easily extend over multiple domains. For example, all traditional realists about time allow that our mental lives are temporally structured while also holding that mind-independent reality is temporally structured. These are not different structures that happen to resemble one another—rather, the very same structure is present in both mind-dependent and mind-independent reality. In being aware of the structure of one’s mental life, one could then be presented with temporal structure as such.

Looking at Allais’ reconstruction of Kant’s argument on p. 195, we can put this point as an objection to the third premise, which states that “[o]bjects which are independent of us can be present to consciousness only through affecting us”. There is an ambiguity in the phrase “independent of us”. This can be understood either as “are not intrinsically involved in our mental lives” or as “can exist/be instantiated without us”. Especially for structures, the premise is plausible only with the former understanding. If some structure is not intrinsically involved in our mental lives, then perhaps that structure must affect us to be present to consciousness (even that claim seems dubious to me, but set that aside). By contrast, it does not seem at all plausible to say that affection is required for us to be presented with structures that are intrinsically involved in our mental lives but which could exist or be instantiated without us. To block a neglected alternative objection, however, it is this latter claim that Allais’s reconstruction needs.

Even if this problem and the previous one are genuine, Allais’s interpretation of the Aesthetic might still be the best on offer, of course. And perhaps Allais thinks that the short arguments Kant gives in the Metaphysical Expositions in the Aesthetic somehow preempt these worries. If so, however, it would strengthen her interpretation if she could say how.

5. Generality and the Transcendental Deduction

Allais’s interpretation of Kant’s notorious Transcendental Deduction of the Categories is creative and intriguing. I find much of what she says about the ambitions of the Deduction plausible, even though it completely sets aside the role of apperception. Parts of her reading seem problematic, however. One such part concerns her view of the role of the generality of thought (partly following Evans 1982:101). Allais writes:

The generality of thought means that in order to think of something as having a property a subject must be in a position to think of it as a subject of other properties as well, and must be in a position to think of the property in question as potentially belonging to other objects. This, I submit, is what Kant thinks we cannot do in the chaotic situation in which there is no limit to what can count as a property, and correspondingly no limit to what can count as a subject of properties. Investigating an object’s properties requires a way of ruling out properties that do not belong to it. (p. 280)

According to Allais, the categories are what allow us to rule out problematic properties (such as Goodman’s grue), and this is how Kant establishes the objective validity of the categories.

Now, Kant certainly does think some sort of generality is essential to thought. Moreover, he does seem at least partly concerned with chaotic situations in the Deduction (see e.g. A100–1). Yet it seems to me that, on both textual and philosophical grounds, Allais tries to draw too quick a connection between these. The textual grounds are simple: I do not see any places where Kant claims that the generality of thought has the sort of implications Allais describes. The philosophical grounds are just that it does not follow from being able to think of an object as having other properties (and of a property being able to hold of other objects) that one be able to rule out or limit any subjects or properties. As Kant recognises, there are few limits on what one can think (see e.g. A96, B146–7). So someone could take herself to be in a very chaotic world, filled with gruesome properties and gerrymandered objects, and yet have no trouble in thinking about any possible combination of objects with properties. Generality does not itself suggest any substantive restrictions on representation.

Notice that, at the end of the quoted passage, Allais shifts to talk of investigation. Investigation requires much more than the generality of thought, so I think this is a much more promising point for Allais to focus on. On the other hand, while Kant obviously has the generality of concepts in mind in the Deduction, it is harder to show that he is concerned with investigation in any rich sense. So if Allais’s reading hinges on a thick notion of investigation, then we would want to hear more about why she thinks that notion of investigation is central to the Deduction.

Conclusion

Though I think Allais’s interpretation faces certain questions as it stands, I do not think those questions are harder than those facing any other textually-sensitive interpretation. Since most of my discussion has been negative, I want to emphasise again how valuable I think Allais’s book is. Not only does it offer an important and powerfully-motivated interpretation, it also provides deep and thorough discussions of the competing factors every interpreter must face. In my view, Manifest Reality should be a focal point for the discussion of Kant’s idealism for years to come.

Acknowledgement: In writing this review essay, I benefited from discussion with Dai Heide, Colin McLear and James Messina.

Invited: 23 June 2015; received: 23 November 2015.

Notes:

[1] For more on this and the other points in this paragraph, see Marshall (2013).

[2] Note that the point here is about the revelation of a property, not an object. Allais thinks that, for Kant, some properties of an object can be revealed without all its properties being revealed (see e.g. p. 159). But we would expect the revelation of a property to tell us about the nature of that property.

[3] This last view would give the weakest sort of mind-dependence to these properties. The properties would, perhaps in virtue of their essence, depend at most on the possibility of minds. That sort of dependence allows for the properties to exist without any minds actually existing.

[4] My own view, argued in Marshall (2013), is that some subtle metaphysics is needed to address this kind of worry, but Allais seems wary of such metaphysics.

[5] The other objections also strike me as problematic, for reasons I give in Marshall (2014).


References:

Bennett, J. (1971), Locke, Berkeley, Hume: Central Themes (Oxford: Oxford University Press).

Descartes, R. (1984–91), Philosophical Writings of Descartes, 3 vols., trans. and ed. John Cottingham et al. (Cambridge: Cambridge University Press). (abbreviated as CSM)

Evans, G. (1982), The Varieties of Reference, ed. J. McDowell (Oxford: Clarendon Press).

Logue, H. (2012), ‘Why Naive Realism?’, Proceedings of the Aristotelian Society 112:2: 211–37.

Marshall, C. (2013), ‘Kant’s Appearances and Things in Themselves as Qua-Objects’, Philosophical Quarterly 63:252: 520-45.

——— (2014), ‘Does Kant Demand Explanations for All Synthetic A Priori Claims?’, Journal of the History of Philosophy 52:3 (2014): 549–76.

Maund, B. (2012), ‘Color’, The Stanford Encyclopedia of Philosophy (Winter 2012 Edition), ed. E. Zalta, http://plato.stanford.edu/archives/win2012/entries/color/.

Spinoza, B. (1985), Collected Works of Spinoza, vol. 1, ed. E. Curley (Princeton: Princeton University Press).

Van Cleve, J. (1999), Problems from Kant (New York: Oxford University Press).

© Colin Marshall, 2016.


Colin Marshall is Assistant Professor of Philosophy at the University of Washington, Seattle. He specialises in Kant, early modern philosophy, Schopenhauer, and metaethics. His articles have appeared in, among others, Journal for the History of PhilosophyPhilosophy and Phenomenological Research, philosophical QuarterlyEuropean Journal of Philosophy, and Kant-Studien. He is currently completing a book about compassion.

Website

 

Advertisements