Maths & Logic Workshop: Michael Potter

Michael Potter presented the outline of a theory of arithmetic for which a priori knowledge is not trivial (contra empiricism) and which takes a qualified attitude towards absolutely unknowable truths (contra full-blooded transcendentalism). He started with a short appraisal of Kant’s contribution to arithmetic, holding fast to the Kantian notion of synthetic a priori truths as a genuine addition to monadic and polyadic logic while rejecting Kant’s idealist account of these truths.

Potter proceeded to recall that Gödel’s first incompleteness theorem leads to the disjunctive conclusion that either human minds are not machines or there are absolutely unknowable truths about arithmetic. Both disjuncts are not empiricist (provided the rejection of physicalism about the mind results in a non-empiricist position). However, it is not clear from Gödel’s argument whether both are transcendental. Potter turned to Wittgenstein to motivate a transcendental interpretation of Gödel’s first disjunct. Wittgenstein remarked that there is no space between our capacity for recognising the limits of our cognitive abilities – so that it makes sense to say that there are absolutely unknowable truths about the mind – and the absence of this capacity – so that all a priori truths about the mind turn out trivial. If this is true, the rejection of empiricism about minds implies transcendentalism.

Combining Wittgenstein’s views about cognitive limits with Gödel’s disjunction results in a further disjunction. Potter concluded that the introduction of synthetic a priori truths in arithmetic implies that there are either absolutely unknowable truths about the mind or absolutely unknowable truths about arithmetic. This disjunctive conclusion makes it unnecessary to commit to absolutely unknowable truths across the board (in analogy to Dummett’s non-systematic conception of antirealism). Committing to either transcendentalism about the mind or about arithmetic is unavoidable, but there are no external criteria to adjudicate on which side our commitments go. It is in this sense that Potter’s (tentative) transcendental position on arithmetic is a qualified position and not a full-blooded one. We are committed to transcendentalism about arithmetic if human minds are machines. We may be committed to it if that is not the case.

Conference: The Nature of Transcendental Philosophy

Well, our second annual conference was, I think, a great success.

We kicked off with Ulrich Schlösser, who claimed that Kantian transcendental philosophy is defined by its proposal that we are necessarily related to objects. His paper, "The Necessity of Object Reference, Idealism and Communicability", explored two Kantian arguments for this proposal and examined their relation to transcendental idealism.

Next, Alan Richardson outlined his view of the history of Logical Empiricism, understood via its relation to both neo-Kantian transcendental philosophy and subsequent figures such as Quine and Popper. His paper, "Has Philosophy of Science Completely lost its (Transcendental) Mind?", focused on the fate of the knowing subject in scientific philosophy, and on the tenability of 'empiricism without experience'.

In his paper, "The Bounds of Sense", Adrian Moore looked at a number of attempts from within Twentieth Century Analytic Philosophy to draw a distinction between that which does and that which does not make sense. In doing so he motivated his own view that an explanation of why we should draw the bounds of sense where we do must remain ineffable.

In the final paper of the first day of the conference, Barry Stroud outlined some "Transcendental Strategies". He gave an account of what we can expect to achieve by the the use of certain distinctively transcendental ways of arguing. Although, he claimed, certain conclusions can only be reached if one assumes some form of idealism, transcendental arguments may well be of significant value in showing some propositions to be 'invulnerable' in a number ways.

Our own Matt Grist, "In Defence of Epistemic Transcendental Idealism", started off the second day of the conference by offering just what his title suggested. Distinguishing his own view from that of Henry Allison, to which it is related, Matt argued that his interpretation showed that transcendental idealism is a non-trivial 'knowledge limitation' thesis which is entailed by empirical realism, as Kant understands it.

Next, Jonathan Lear put forward an account, based on the testimonies of Crow Indians after their move to the reservation, of what it is to lose that by means of which the world is intelligible to one. In, "What is it to be Deprived of a World", he argued that an understanding of this phenomenon - loss of that which makes one's world possible - opens us up to the possibility of a distinctive kind of philosophical irony, a possibility noticed by Kierkegaard in his account of Christianity within Christendom.

In, "Transcendental Philosophy without Things in Themselves?", Wayne Martin explored the possibility of understanding both transcendental idealism and transcendental philosophy more generally, without being forced to posit the existence things in themselves - a possibility that keeps the figure of Fichte hovering in the background. He addressed the question by way of a detailed reading of Kant's treatment of freedom in Groundwork III, tentatively concluding that the resort to things in themselves is driven by a particular sort of philosophical anxiety.

Finally, Christopher Peacocke in his, "Objectivity", defended the claim that transcendental arguments can, contrary to some, establish truths about the world in a way that relies on realism rather than any form of idealism. He offered two transcendental arguments, based largely upon his account of content, designed to show a 'minimum objectivity' involved in both judgement and perceptual experience respectively.

Reading Group: Lear's 'Disappearing We'

Jonathan Lear takes a line on the later Wittgenstein (LW) inspired by the Williams' paper we read last week. The paper outlines Wittgenstein's investigation of the way inner episodes of understanding meaning are intrinsically connected to the potentially publicly available outer manifestation of the practical ability to follow a rule. It compares this to Kant's investigation of how intuitive representations are 'saturated' by concepts due to their being instances of the analytic unity of apperception. The quite general point of comparison is that a certain synthetic unity amongst representations is required to make possible inner episodes of understanding/apperception. For LW the synthetic unity in question is constituted by language-games, so that understanding has to be capable of manifesting itself in rule-following which is potentially recognisable as a shared linguistic practice. Lear glosses this in terms of every 'I understand' implying a 'We are so minded'. This is supposed to license the 'we' in the second locution disappearing, so that what makes sense to 'us' shortens to 'what makes sense'. I presume the thought here is that if any case of understanding a word will also be a case of some potentially publicly available linguistic ability manifested in a language-game, then the 'we' who take making sense to consist in such manifestation just is the 'we' of sense-makers simpliciter.

The disappearing 'we' is one leg of Lear's interpretation of LW. The other is that our reflective self-consciousness of the fact that making sense simpliciter is a matter of words being potentially manifested in shared linguistic practices is consciousness of a form of transcendental idealism. The idea here is that the metaphysical articulation of the world is utterly enmeshed with the linguistic articulation of language-games - to individuate an object is always to articulate a view of the world that is part of our linguistically articulated 'form of life'. The thought is that any such individuation will consist in introducing one more possible articulation of the world that the global 'we' finds makes sense as a language-game, and that this means there is no room for articulating how the world is in itself. Or, to put it another way, on this view, the limits of possibilia coincide with the limits of what is linguistically articulable.

There are two obvious questions here. Does the 'we' disappear simply because understanding a word requires potentially publicly available manifestations of rule-following? If so, does this really imply idealism? To the first question, I think the answer is yes. If the global 'we' is defined as those subjects who understand meanings in terms of publicly available rule-following or shared linguistic practices, then any such understanding will be encompassed by some shared linguistic practice. It is then an anayltic truth that understanding will be part of the global 'we'. As for the second question, I'm not sure. I think it will depend on what one understands by 'realism'. If this is understood as our linguistic articulation of the world not falsifying the properties and relations that the world contains , then I don't think anything idealistic follows (some subset of our form of life might be articulated by language-games that precisely 'map' what there is). But if realism is taken as having an epistemic element, so that it requires that what makes our thoughts/sentences true is their correspondence with a world as it is in itself, then I think the position is transcendentally ideal. For it says that no such correspondence is intelligible, for there is now no sense to the idea of the way the world is in itself.

Reading Group: Williams on Wittgenstein

I'm going to be very brief today for the reason that my grip on some of the central aspects of this paper is pretty slender. As is familiar, Williams thinks that there is a view implicit in the later Wittgenstein that might be called 'first-person plural transcendental idealism'. All rests on his view that the later Wittgenstein accepts something along the lines of "the limits of our language mean the limits of our world" (p.150).

I want to focus in on two remarks that Williams makes that are, as I understand it, intended to give flesh to this first-person plural transcendental idealism. They are:

"what the world is for us is shown by the fact that we can make sense of some things and not others: or rather - to lose the last remnants of an empirical and third-personal view - in the fact that some things and not others make sense." (p.152)

"our language, in this sense in which its being as it is has no empirical explanation, shows us everything as it appears to our interests, our concerns, our activities, though in the only sense in which we could meaningfully say that they determined everything, that statement would be false. The fact that in this way everything can be expressed only via human interests and concerns, things which are expressions of mind, and which themselves cannot ultimately be explained in any further terms: that provides grounds, I suggest, for calling such a view a kind of idealism." (p.151)

The relevant "we" here is not an empirical group of people located within the world, but a plural version the metaphysical self of the Tractatus. This is what is supposed to legitimate the move in the first quote that would otherwise appear to express an unfounded optimism, or arrogance. The thought is perhaps expressed by saying that "we" refers to all possible subjects. One problem with this might be that we (some empirical group) cannot be said to have privileged access to what "we" can make sense of. But I won't say anything about that.

My query is why we should think of the first quote as expressing a form of idealism. It concerns, after all, what the world is for us. But shouldn't idealism tell us something about what the world is. I realise that this is overly blunt. But the point is that there is a way of reading the first quote such that it is very far from idealistic.

The second quote, I presume, is supposed to provide something of an answer to this query. It seems reasonable to suppose that Williams' use of "human" in this passage is a slip suggesting, as it does, some empirical group. But, in so far as I think I understand the claim, it once more seems to me to be susceptible to non-idealist readings. For the claims concerns "everything as it appears" and the way "everything can be expressed". But this doesn't strike me as being a claim about what is in any obvious way. Or, at least, I need to be shown why we should understand even this second passage amounts to something which should be regarded as treating the world (phenomenal, noumenal, empirical, transcendental, in-itself, whatever...) rather than our expression of it, or its appearance, as in some sense mind-dependent.

These remarks are, no doubt, naive. As I said, I felt that much of the subtlety of this paper was passing me by. But nobody said life was a bed of roses. J

Robert Stern at the Transcendental Arguments Workshop

Bob gave a paper entitled The Value of Humanity: Reflections on Korsgaard's Transcendental Argument. In it he argued that a particular reading of 'phase two' of Korsgaard's transcendental argument for the value of humanity formulation of the categorical imperative is successful (Bob stressed that he was not defending the success of phases one and three, nor another version of phase two based on a 'regress of identities argument').

The argument Bob defended had the following first three premises, which were what discussion focussed on.

1. To rationally choose to do X, you must take it that you have some reason to do X
2. You cannot think that X is the rational thing to do in itself, but only given your practical identity
3. You cannot regard your practical identity as making X the rational thing to do unless you can see some value in that practical identity

It seems premise 1 is unproblematic. Given anti-realism about the normative source of reasons, premise 2 is fairly unproblematic also - for if actiona are not in themselves possessed of normative properties, then a particular rational action will only make sense if the agent conceives of it as the instantiation of what she finds worthwhile, i.e. as an instantiation of a practical identity.

Bob suggested Korsgaard's transcendental argument be read as a modest one - that is, not as 'world directed', attempting to establish the truth of the value of humanity, but rather operating under the auspices of an anti-realism about value which means the argument is only attempting to establish what an agent must believe from within her perspective on the world.

Most discussion was of the move from premise 2 to 3. One objection was that we can have anti-realism about values without going through a self-reflexive stance where the agent values her practical identity - think of Humean projectivism. Bob replied that it is hard to understand an agent valuing what she projects without valuing the practical identity that frames the projection, unless we see the object of a rational action as itself normative (precisely what we can't do given anti-realism about norms). Another objection was that acting under a practical identity just is to find certain reasons compelling - that is, it is to operate according to the practical identity rather than to explicitly endorse that identity in an evaluative step. Bob replied that the evaluative step is necessary to answer our how possible question about rational action given anti-realism. A third objection was that reasons are usually identified with roles in reasoning, which yields a sort of anti-realism in that occupying the role makes things so (that I conceive myself as a holder of office makes it so that fulfilling that role is a reason to act, nothing more is needed to make the reason compelling). But if this is right, the move from premise 2 to 3 is invalid, for finding certain reasons compelling is constitutive of having a practical identity, which means one can see X as the rational thing to do without valuing the practical identity that frames it. Bob responded that we need the evaluative step to get reasons into the picture in the first place in an anti-realist setting.

Generally then, the scepticism about Bob's reading of the argument, such as there was, centred on the idea that acting under a practical identity implies valuing that identity. It wasn't clear to me whether the valuating step is just built in to rational action as such, or whether an agent has to actually engage in self-reflexive evaluation (which seems false), or whether the latter is more a requirement of philosophical theory that wants to explain the possibility of rational action in an anti-realist setting.

TI Reading Group: Bell on Empirical Realism and Transcendental Idealism

First of all, sorry for the lateness of this post.

Last week, we read David Bell's 'Is Empirical Realism Compatible with Transcendental Idealism'. The paper attempts to get clear about problems facing holders of the 'dual component view' of transcendental idealism (TI) and empirical realism (ER). It also attempts to adumbrate possible solutions to said problems. As is usual for Bell, the paper is clear and easy on the reader whilst dealing deftly with some complex issues. It does, however, have the feel of a short conference paper and the broad strokes necessary for that format make it somewhat cursory and suggestive. Nevertheless, the work of laying out the ground for further discussion that is carried out here is useful and interesting.

The pressing questions the dual component view must answer are the following: (i) what exactly does it consist in?; (ii) is it self-contradictory?; (iii) isn't the view's second leg, TI, false?

Bell hopes to answer these questions by getting clear about both ER and TI.

He starts by setting out two pre-theoretical intuitions that drive the realist position: (a) the desire for metaphysical modesty and (b) the desire to explain how objectivity in thought, language and knowledge is possible. (a) grows out of the idea that human thought is only a small part of the universe and that its imperfect cognitive faculties can only aspire to an ever evolving epistemic picture of a largely independent natural world. (b) grows out of the idea that it would not be possible for us to have the putatively objective cognitive practices we have - say, for example, the practice of referring - if we didn't also hold a largely realist view of the world. Bell goes on to say that these days realism is usually understood as the semantic thesis that truth and falsity are epistemically unconstrained. But he notes there are epistemological and ontological versions of the position as well.

Taking his cue from Dummett, Bell argues that talk of realism tout court is problematic. He suggests we talk rather of realism with regard to a domain of objects. However, that domain can be so wide as to encompass all empirical objects, and this is precisely the domain of ER. (Nothing about ER commits one to being a realist about numbers, universals and theoretical entities.)

At first blush the accordingly relativised empirical realist position is committed to at least the following:

(R) Fs exist as such and possess intrinsic properties independently of the existence or nature of any awareness any subject may have of them.

(The locution 'exist as such' and the term 'intrinsic' signify simply the anti-reductionism implicit in ER.)

(R) is the conjunction of the following three theses: (i) that Fs exist, (ii) that Fs exist as such, (iii) that they do so independently of any awareness we might have of them. Bell points out that denial of (i) and/or (ii) leads to some sort of nominalism/reductionism. Denial of (iii) leads to a species of anti-realism best described as idealistic.

Taking (R) to be relativised to the entire empirical domain, TI's (apparent) denial of (iii) seems to entail a straightforward contradiction for the dual component view. What's more, such a denial has no empirical basis in the natural sciences. So ER seems to be an empirical theory captured by the following:

(EmR) It is an empirical fact that empirical objects exist as such and possess their intrinsic properties independently of the existence or nature of any awareness any subject may have of them.

(EmR) is an empirical theory both in terms of what it's about, and its methodology. It concerns the entities we can have possible sense experience of, or can be linked to sense experience by principles and theories that are acceptable extensions of principles and theories which are grounded in such experience. And, importantly, ER is not a restrictive realism: it is realism relativised to the empirical domain, swinging free of realism/nominalism/reductionism/idealism with regard to other domains. It says of empirical entities that they exist independently of any awareness we might have of them. It individuates those entities via contingent, defeasible and inductively supported assertions.

So according to Bell, (EmR) is a defeasible, contingent empirial theory about the natural world. It seems appropriate to place (EmR) within a naturalistic theoretical environment. However, thoroughgoing naturalism goes further than (EmR): it says that all there is is empirical reality, and that all acceptable methods for thinking about the latter are natural scientific methods. Thus such naturalism is incompatible with TI, because it rules out a priori the very possibility of there being anything more to reality than the empirical, and equally, rules out any non-naturalistic method of yielding knowledge of reality (even if it is only knowledge of the limits of our knowledge of reality). So TI is incompatible with thoroughgoing naturalism, due to clashing a priori commitments. What we can say at this point is that the dual component view is not ruled out because of such a clash, because (EmR) is detachable from thoroughgoing naturalism.

Bell seems to think that the major stumbling block for the dual component view is the employment of transcendental explanation by TI (rather than, say, metaphysical problems encountered when two classes of empirical and transcendental entities are posited). To that end, he sets out five key elements by which transcendental explanation is constituted.

(1) Maximising Explanatory Adequacy. Naturalistic explanation, it is argued, always falls short of completeness. Transcendental explanation aims to be, in some sense, self-sufficient, transparent and self-contained (beholden to no further external theory).

(2) Maximising Explanatory Scope. Whereas (1) concerns worries over the limitations of naturalistic explanans, this element concerns worries over the scope of such explananda. Naturalism, by definition, cannot explain everything - it cannot make judgements about the world as a whole. What's more it is often held to fall short of explaining various specific phenomena such as consciousness. Transcendental explanation has pretensions to close both these explanatory gaps.

(3) Emphasising Reflexive Structures. Given (1) and (2), transcendental explanation ought to be able to explain its own possibility.

(4) Emphasising Subjectivity. Transcendental explanation, given (3), can only coherently be rooted in Socratic self-knowledge of our cognitive faculties/practices.

(5) Adopting a transcendental vantage point. Naturalistically minded philosophers take the 'natural attitude' of being one part of nature in unproblematic contact with another. They thus accuse transcendental philosophers of taking an incoherent transcendent viewpoint. But this charge need not stick. It would, only if the two vantage points were exhaustive. And it seems transcendentalists accept the naturalist's denial of a 'God's eye view' and also the objectivity of the natural world. What they insist on is a systematic reconception of the latter driven by awareness of the fact that the natural attitude is not cognitively ultimate, without denying (EmR). This reconception will make subjectivity constitutive, be global in scope and be explanatorily reflexive.

All that remains to be done is define TI. Bell proceeds as follows.

(TrI) Without denying any empirical facts, it is possible systematically to re-describe reality as a whole in such a way that the objects which comprise it are not construed as independent of any awareness we may have of them.

The redescription (TrI) offers is one where the natural world is understood in terms of a primitive non-natural conception of phenomena or appearances. Nothing in the natural world is then explained in terms of anything that goes beyond the latter. Given this, such entities will literally be what is given in experience - the most basic building blocks of cognition beyond which it is not possible to penetrate and from which the entire natural world can be explained.

Thus the balance Bell achieves with the dual component view seems to be the following. ER is a contingent empirical theory about the existence of objects. TI is an a priori theory that construes empirical objects as governed by principles and concepts that are constitutive of subjectivity, global in scope, explanatorily self-sufficient and complete.

Presumbaly, the compatibility arises because ER is a contingent theory about object existence and TI an a priori theory about object construal. It is not contradictory to say that empirical objects exist completely independently of the mind yet redescribe the whole of empirical reality in terms that are non-naturalistic, a priori and mind-dependent, for the sake of solving philosophical problems. Presumably those problems are the deficiencies in naturalistic explanation that (1)-(5) address.

I have much sympathy for this view. For it paints TI as a non-metaphysical view about the possibility of cognition/knowledge that potentially achieves two of Kant's aims; (a) it blocks the cognitive relativism of differing conceptual schemes that naturalism seems to invite by securing synthetic a priori knowledge about the possibility of experience (transcendental aesthetic and analytic), and (b) it blocks off certain philosophically problematic cognitive avenues by providing a complete explanation of our cognition of the natural world (transcendental dialectic).

I will end by highlighting a potential difficulty for this balancing act (there are many more, I'm sure). The naturalist might run the transcendental explanatory desiderata back at the transcendentalist. For surely, Kant - to take an extremely prominent transcendental philosopher - doesn't explain all aspects of phenomena completely. For example, he explains self-consciousness in terms of judgement, and intentional content in terms of self-consciousness. But what counts as an explanation of judgement? Kant seems to think laying out its a priori forms. But, the naturalist will argue, this doesn't really explain judgement, it just takes those forms as primitive. Further explanation is needed, and it is likely to be naturalistic in nature, abrogating the transcendental explanatory framework. In other words, which lack of explanation we find troubling is not itself a philosophically neutral matter.

Virtual again this week

This Wednesday (20th) we won't be meeting at Essex, but reading virtually (see previous post for the reading). We'll hopefully be meeting physically for the final two sessions thereafter.

Bell on Wednesday

On Wednesday (20th) we'll be reading: Bell, D. 'Is Empirical Realism Compatible with Transcendental Idealism?’ in Idealismus und Repraesentationalismus, Ed. R. Schumacher, (Berlin, 2001)

Transcendental Idealism Reading Group: Van Cleve's Two Worlds View

This week we read pp. 143-155 of Van Cleve's Problems from Kant. These sections aren't really a defence of the two-world view, but rather a battery of arguments against one-worldism. The arguments are of two types: those concerned with what Kant actually thought, and those concerned with which of the one and two-world views is most philosophically convincing. I shall say a little about both these types of arguments, although I won't discuss everything Van Cleve says in these sections of his book.

First, there is some textual evidence offered for why Kant is a two worlder. When he uses language indicative of one-worldism, such as using the same pronoun to refer to things as they for empirical cognition and as they are in themselves, this is a rhetorical device that does not imply one or the other of the two ontological views. Second, one worlders are in the uncomfortable position of saddling Kant with a Leibnizian conception of cognition that he clearly rejects - that we perceive things in themselves (TiTs) confusedly in perception, and clearly by means of the intellect. Presumably one-worlders are forced to pin this conception on Kant because for them, empirical cognition is of TiTs, but only in terms of how they appear, so that such cognition is a 'confused' version of what the understanding alone cognises of TiTs.

The first point is fine but points in neither a one world or two world direction, it merely makes two worldism as consistent with some of what Kant says as one-worldism. The second point seems to me to get Kant all wrong. Kant continually warns against the idea that pure understanding can think TiTs in anything more than an extremely cursory way - in a logically consistent but cognitively empty manner. This just gets us the very idea of TiTs as coherent, rather than some proper cognition of them a la Leibniz. So there are two modes of access to TiTs for one worlders. One via empirical cognition and one via transcendental philosophy itself. But (for the post-critical Kant anyway) there are not two cognitive routes to TiTS - for cognition requires an object for Kant, and the solely intellectual use of the understanding has no object.

Van Cleve goes on to raise some difficulties with apparent contradictions for the one world view. These are encountered by asserting of TiTs that they are both spatial (as they appear) and non-spatial (in themselves). He propose three different models for dissolving the contradictions, which I shal summarise as follows.

(1) It is in fact different parts of the same object that possess the apparently contradictory properties, i.e. there is no contradiction is asserting that a buliding is round on the first floor, square on the third.

(2) The apparently contradictory properties are ascribed by two different interpretations of the facts, one or both of which turn out to be false.

(3) What appear to be contradictory monadic properties are in fact relational properties. There is no contradiction is saying Fred is taller than Alfonso, yet shorter than Diego.

(1) is apparently no good as it reintroduces the two-object view. We questioned this. It doesn't so much introduce the two-object view but assert that a 'thing' has two parts, spatial and non-spatial. Now, it may be strange to talk about 'things' as composites, but it preserves the one-world view without contradiction (and who said transcendental idealism wasn't strange!).

(2) is also apparently no good, because it makes appearance talk illusory and this is something one-worlders wish to avoid. We also questioned this. Van Cleve assumes that if two apparently contradictory predicates are ascribed to a thing, then on this model the only way to make sense of this is one or both ascription being wrong. But this assumes there is a contradiction which is what the one-worlders deny. What they argue is that one ascription is true of things at the empirical level, and the other at the transcendental level. To refuse this dissolving of the contradiction by recourse to such distinctions comes very close to begging the question against the one-worlder.

(3) is the best option according to Van Cleve. If monadic properties are disguised relations, our apparent contradiciton is solved, and one-worldism can fly. He has no conclusive argument against this model. He simply points out that it is very strange to think of 'having a shape' as a relational property. But again, whoever said transcendental idealism wasn't strange!

To demystify the idea of spatial properties being relational, he suggests the relation be understood as 'being shaped in relation to subjects with constitutions like ourselves' (or some such). But then, he claims, this just reintroduces the illusion/reality picture one-worldism wants to avoid. We found this unconvincing, as it presumes transcendental realism (that there is one way of cognising objects which is common to God and ourselves, and that any object either has or does not have a given predicate correctly asserted of it according to this single cognitive perspective). Again, this seems mighty close to begging the question against one-worlders who insist that cognition is relative to a transcendental standpoint. (I presume that one-worlders think there is only one way the world is. What they deny is that that way can only be correctly described from a single cognitive perspective.)

Van Cleve raises some further problems for one-worlders. He claims that Kant, in the Amphiboly section of the first Critique, maintains that certain principles hold of TiTs - such as (a) complexes being composed of simples and (b) the identity of indiscernibles - but not of appearances. But if this is so, then one-worlders must hold that a thing in both its aspects (phenomenal and noumenal) has two different mereological structures, and that two indiscernible things are both identical (at the noumenal level) yet diverse (at the phenomenal level). Both of these conclusions Van Cleve claims are absurd.

In a footnote, Van Cleve admits that it is not clear that Kant does hold that (a) and (b) hold of TiTs, rather, Kant is merely pointing out in the Amphiboly section that our cognition cannot be understood as individuating objects solely in terms of their qualitative features via concepts. This is because spatio-temporal objects are known through their relations to other such objects, and two of them can be qualitatively identical yet differ in spatial location. Thus, Kant concludes, our cognition should not be modelled on Leibnizian style conceptual individuation. If this is right, Van Cleve's argument must be recast as 'if (a) and (b) hold of TiTs, then one-worlders are committed to the absurd conclusions outlined above.'

I personally don't find relative identity particularly troubling, as identity is a relation ('a is identical to b according to qualitative individuation, different according to spatial'). But relative mereology does seem absurd. How can a single object be both simple and composite?

Reading Group: Lucy Allais, "Kant's One World"

Well I didn't get round to reading Lucy's critique of Langton, so my comments here are based purely on the "Kant's One World" (KOW) and "Kant's Idealism" (KI) papers, primarily the former. Also, I'm not going to say anything about the first half of that paper, useful and interesting as it is. Rather, I'm sticking to the positive proposal.

OK, so Lucy thinks that an account of transcendental idealism (TI) has to respect the following constraints:

1) TI has to be non-trivial
2) Ignorance of things in themselves (TTs) must be genuine ignorance
3) We must be able to have coherent thoughts about TTs
4) Appearances must be mind-dependent
5) Appearances must not be mentalised (must not be "in the mind")
6) Appearances and TTs must have a genuinely different status

As I understand it, the account sets about to achieve this in the following way:

a) First of all, we understand secondary qualities in terms of a direct realist (relational) theory of perception and, especially, a direct realist account of perceptual illusion. The idea being that we make room, within a direct realist view for the idea that a thing can appear as it is not.

b) Appearances are relational properties. That is, appearing a certain way is possessing a certain property in relation to (possible) subjects. On this view, appearances: (i) are public features of objects, (ii) represent objects as being a certain way, (iii) are mind-dependent, (iv) do not reveal the way things are in themselves.

c) The properties of objects can be specified either as they are in themselves, or via their effects on other things, including us. Ignorance of TTs amounts to the claim that we can have no knowledge of the former. That is, the ways things appear to us in experience never reveals their true, intrinsic nature (the ways they are in themselves).

About this I am puzzled by two things, both of which concern the account of appearances. The first concerns the difference between relations and powers, the second concerns the notion of representation.

First. Although in KI Lucy explicitly rejects the interpretation of appearances as powers ("my subjectivist account of colour does not see colour as a disposition of power." [KI, 483]), in KOW it seems to me that she occasionally slips into talking this way, especially when discussing Langton. For example, "The idea is that we can distinguish between specifying a property in terms of what it is a power to do, i.e. to other things, and describing it as it is in the object in itself." [KOW, 679], and "just as we cannot know the intrinsic nature of things, or the grounds of powers, we cannot know the relation between the ground and the powers" [KOW, 680].

It seems to me that there are differences between relations and powers that are relevant to the mind-dependence of appearances. As an example of a relation consider __is taller than__ and as an example of a power consider __is soluble in__. It think it is very natural to say that relations require the actual existence of their relata, whereas powers (at best) only require the possible existence of the second ___. That is, I am only taller than Jeff if Jeff exists, whereas salt is soluble in XYZ even though XYZ does not actually exist. If you don't think that this is a natural way of talking, treat it as a stipulation.

With this in mind, I want to force a choice: Are appearances relations or powers? If relations, then they are strongly mind-dependent - appearances only exist if subjects like us actually do. If powers, then they are weakly mind-dependent - appearances only exist if subjects like us possibly do.

Despite the fact that Lucy goes in for relations talk and explicitly rejects powers talk (we can put this down to a terminological difference between her and me), there is reason to think that she would choose the powers option. Why?, Well, "On the one hand, objects have colour whether we are perceiving them or not, and would not change colour if we all died; on the other hand, objects only have colour in relation to subjects like us being possibly able to perceive them." [KOW, 674]. And, although perhaps this is not inevitable, there is the sense that the relations option might slide into the view that appearances only exist when they are being perceived.

On the other hand, there is reason to think that the powers option doesn't have enough idealism in it for Lucy's tastes. In discussion Van Cleve, she says that if we, "identify the appearances of things with what the secondary qualities "in truth" are--causal powers of objects--but now our picture is in danger of losing any kind of idealism." [KI, 467]

Further, she makes the following puzzling remark, "It is not just that where there is an empirically real thing there is the possibility of experience, which a realist might also think, but that where there is no possibility of experience there is no empirically real thing." [KOW, 675]

As I understand these two, they are logically equivalent (If P then possibly Q, and If not possibly Q then P [No? What have I missed?]), so if a realist can endorse the one then he can endorse the other. But all of this sounds like the powers option: appearances require the possible existence of subjects like us.

There is more to be said on this one, but I've gone on long enough. But to my mind there is a genuine puzzle about exactly what Lucy's view is committed to.

Right. Second. About representation, Lucy says this, "Perception involves things being presented, or represented, to subjects in certain ways, and the term 'representation' could be used to mark the fact that appearances are always appearances for subjects--that appearing is essentially relational--rather than to refer to mental intermediaries." [KI, 473]

I have to admit to finding this strained. It certainly doesn't sound natural to me to think of an object's relational properties as representing the object. Especially since there is an intuitive connection between representations and things with a proposition-like structure, and it seems odd to me to think of, say, an object's colour as propositional. But lets brush that aside. The reason, as I understand it, for justifying talk of representation here is that there is error (appearances don't reveal the nature of TTs). Now I understand that thought - just like, look my beliefs can say P even though not-P, ahh belief must be representational. But it seems to me that this rationale for introducing representation talk would have the consequence that what is represented is the way the thing is, and more, the way the thing is in itself. But, if that's right then all appearances would involve illusion/misrepresentation, and this is something that Lucy wants to avoid.

So, I'm puzzled about how Lucy's view can neatly capture Kant's claim that appearances are representations. Now, I don't think that the above argument will run if we switch presentation for representation, and today Matt and Wayne went some way to convincing me that we can make this shift whilst preserving the sense of Kant's position. So perhaps this isn't such a bad thing. In any case, I have probably misunderstood some (a lot?) of what is going on here. So I'll stop.

This week's reading

We are reading this week pp.143-162 of James Van Cleve's Problems from Kant. This is a longer section of the book than was previously advertised.

There is some subsidiary reading, also by Van Cleve: ‘The Argument from Geometry, Transcendental Idealism and Kant’s Two Worlds’, in Minds, Ideas, and Objects: Essays on the Theory of Representation in Modern Philosophy, eds. Cummins, Phillip D., and Zoeller, Guenter.

Reading Allais Next Week

As I said earlier, next week we are reading Lucy Allais, "Kant's One World: Interpreting Transcendental Idealism’", British Journal for the History of Philosophy 12:4 (2004), pp. 655-684.

I'm pleased to say that Lucy has been kind enough to let us put two of her forthcoming papers online here and here. These will be online for 1 week from today. She tells me that these are essentially developments of certain sections of the 2004 paper. As such we thought they might provide nice additional reading matter.

Online Reading Group: Reading List

Here's advance notice of the rest of our planned reading. We will continue next week on Wednesday 30th May, at 12pm in room 4.137 at Essex University. We will be reading Lucy Allais, "Kant's One World: Interpreting Transcendental Idealism’", British Journal for the History of Philosophy 12:4 (2004), pp. 655-684.

As usual, discussion will continue here on the blog.

In the weeks that follow we will be reading the following (in this order):

1. Van Cleve, J. Problems from Kant, Oxford University Press, 1999, pp.143-150. And, Van Cleve, J. ‘The Argument from Geometry, Transcendental Idealism and Kant’s Two Worlds’, in Minds, Ideas, and Objects: Essays on the Theory of Representation in Modern Philosophy, eds. Cummins, Phillip D., and Zoeller, Guenter.

2. Abela, P. “Putnam's Internal Realism and Kant's Empirical Realism: The Case for Divorce”, Idealistic Studies, 1996, 26 (1) 45-56.

3. Bell, D. 'Is Empirical Realism Compatible with Transcendental Idealism?’ in Idealismus und Repraesentationalismus, Ed. R. Schumacher, (Berlin, 2001)

4. Williams, B. ‘Wittgenstein and Idealism’, in Moral Luck, Cambridge University Press, 1981. And, Lear, J. ‘The Disappearing ‘We’’, Proceedings of the Aristotelian Society, Supplementary Volume, 1984.

Transcendental Idealism Reading Group: Friedman on 'Kant, Scepticism and Idealism'

This week we read Michael Friedman's 'Kant, Scepticism and Idealism'. The paper starts by pointing out the similarities between a recent complaint about Kant's transcendental idealism (TI) voiced by Stroud and McDowell, and one of Jacobi's objections to Kant. The complaint is that even if TI in some sense justifies our knowledge of the empirical world against Humean and Cartesian scepticism, it still makes such knowledge seem second grade due to its not being of fully mind-independent reality. Friedman notes there is a reasonable Kantian reply to this complaint: that it is a non sequitur to infer that just because our knowledge is limited to the empirical domain, and so cannot be of objects not existing in space and time, that it somehow falls short of being knowledge.

Having made this response on Kant's behalf, Friedman notes there is nevertheless a sceptical problem that arises from Kant's way of 'grounding' knowledge of the phenomenal world and which Kant and his successors in the German idealist tradition took seriously. Reinhold apparently formulated the worry thus: how, within the realm of experience (or, 'at the empirical level'), do we know that the matter given a posteriori to our senses will conform to the a priori cognitive forms via which we cognise it, and thus know that our purported empirical knowledge is genuinely knowledge?

We spent much of the session today trying to work out exactly what this sceptical worry amounts to. It cannot be a worry that empirical objects might cease to conform to a priori cognitive structures, because at the empirical level, objects are defined for Kant as so conforming. And the worry can't be located at the transcendental level, because Reinhold has raised it 'within consciousness', that is, at the level of empirical reality. Friedman's gloss is that the worry is about there being no specific causal laws presented in experience, or that what we take to be such laws are in fact no such thing.

So we have two types of scepticism.

1) (Jacobian scepticism) That our knowledge is not of fully mind-independent reality

2) (Reinholdian scepticism) Even granted that empirical reality conforms to a priori constraints, so that empirical objects necessarily conform to certain general 'laws of nature', we are not justified in taking putative specific empirical laws of nature to be necessary

Friedman claims that it is Reinholdian scepticism to which the Absolute Idealists such as Schelling are responding, and to which Kant himself responded in the Metaphysical Foundations of Natural Science, and the the ether proof of the Opus Posthumuum.

Friedman takes Kant to think that particular empirical laws - such as some of Newton's laws of physics - owe their status as laws to being 'grounded' or 'determined' by the principles of the pure understanding. But even if we accept this, there is still a problem over many putative specific empirical laws. And this problem becomes particularly acute when we consider that for Kant, whole domains of objects described by natural science - for example, the domains described by biology and chemistry - cannot have their putative laws vouchsafed by being derived from a priori principles of the understanding. In light of this Kant coins the distinction between constitutive and regulative principles of the understanding: the former consist of the pure principles of the understanding, without which experience would be impossible, the latter consist of guidelines for making knowledge as systematic as possible, but are not necessary for experience. The former confer the honorific title of being laws on rules 'grounded' in them, the latter bestow only the warrant to treat rules as laws for the sake of the systematicity of knowlege.

Friedman argues that Kant's ether proof in the Opus Posthumuum and Schelling's philosophy of nature are both attempts respond to Reinholdian scepticism, by 'grounding' the putative laws of some of the special sciences in the pure principles of the understanding, and thus warranting our taking them to be genuine laws (warranting our taking them to instantiate constitutive rather than regulative principles). The general idea is the Schelling's philosophy of nature is an attempt to demonstrate that magnetic, electrical and chemical forces are governed by the fundamental forces of repulsion and attraction that also govern Newtonian physics. Glossing over much puzzling detail (does Schelling infer an identity from a similarity?), if this is so, then all the life sciences are 'grounded' in the pure principles of the understanding and thus can contain genuine laws of nature.

Setting aside the lassitude Schelling allows himself in demonstrating that organic matter is suffuse with various 'forces' that are 'grounded' in the pure principles of the understanding, there seems to be a serious ambiguity about what Absolute Idealism achieves. Does it:

a) answer the Reinholdian sceptic by simply accepting the Kantian story about constitutive principles and extending the domain of objects governed by laws that instantiate them. This would leave in place TI, that is, would not provide an answer to Jacobian scepticism because the extended domain is still one constituted by objects that are transcendentally mind-dependent (objects governed by laws that instantiate the pure principles of the understanding) .

Or:

b) reject the Kantian story and seek to provide an answer to the Jacobian sceptic. This would involve something along the lines of the idea that reality as it is in itself is an actualisaion of the 'same' laws that govern our experience of it, this identity somehow guaranteeing that our knowledge is truly knowledge of fully mind-independent reality. Schelling's philosophy of nature presumably demonstrates that the required identity obtains.

Friedman seems to be claiming that Schelling is motivated to answer Jacobian scepticism in light of his awareness of the force of Reinholdian scepticism. But to answer the former kind of scepticism one needs to be engaged in b), which isn't motivated by Reinholdian scepticism at all.

Reading Group: Guyer on Transcendental Idealism

So, here are my thoughts on Guyer's , "The Rehabilitation of Transcendental Idealism?". From this afternoon, I know that Wayne, in particular, has another very interesting take on it but I'll stick to the original thoughts that I had when reading the paper (which have been sharpened enormously by the useful discussion today).

The general picture is that Guyer offers an interpretation of Kant's main argument for transcendental idealism that shows it (a) to rule out the neglected alternative, and (b) to be seriously questionable in more than one respect. The main point he wants to impress upon us is that Kant's argument for transcendental idealism uses the denial that space (& time) is a property of things-in-themselves as a premise. Thus the neglected alternative (that things-in-themselves might be spatio-temporal after all) is not neglected at all.

I'm going to ignore the question as to whether Guyer gets Kant right. I'm not the person to adjudicate that. Rather, I want to reconstruct Guyer's argument as I understand it and pinpoint a couple of places where I remain unconvinced. Some of the terminology differs from that Guyer uses, but I hope I'm not introducing anything foreign to his line of thought. Here goes:

(for those present this afternoon, I've changed the numbering AGAIN - sorry)

1. We have synthetic apriori knowledge of Euclidean geometry.
2. We can have synthetic apriori knowledge only of the forms of intuition.
So,
3. Space is AT LEAST the subjective form of intuition

--But this leaves open the neglected alternative, that space is also a property of things in themselves. So we move on--

4. If we perceive an object, it is necessarily spatial
5. If something is necessarily spatial, then everything that is spatial is necessarily spatial
6. We perceive an object
So,
7. Everything that is spatial is necessarily spatial
So,
8. Either things in themselves are non-spatial or they are necessarily spatial
9. All necessary truths can be known apriori
10. We cannot know apriori that things in themselves are (necessarily) spatial
So,
11. Things in themselves are non-spatial
So,
12. Space is AT MOST the subjective form of intuition

I take it that Guyer treats transcendental idealism as the conjunction of 3 and 12.

I also take it that this argument is valid, so we should be concerned with the premises.

First, one of Guyer's most searching criticisms of this argument is its reliance on 4. According to him, the most that we could expect to result from a transcendental inquiry is,

4*. Necessarily, if we are to perceive an object then it is spatial

And, of course, 4 does not follow from 4*. [This is Guyer's distinction between what he calls 'conditional necessity' (all Kant can get) and 'absolute necessity' (what Kant needs)].If he is right about this, and also right about Kant's argument, that would be significant since it would show that Kant's argument for transcendental idealism rests on a premise that is unavailable from the perspective of transcendental philosophy. And that would be a bad thing.

Second, one might balk at 5 (I certainly do). It is Guyer's interpretation of the slogan that necessity implies universality. If Kant believes this, then so be it. But to my mind it stands in need of justification.

Third, more important is the argument's reliance on 9. Now I don't think that 9 is explicitly articulated by Guyer, but I for one can't see how to make sense of his comments towards the end of the paper without assuming that Guyer takes Kant to adopt such a principle. Without it, the neglected alternative remains a possibility - that is, 11 (and so 12, and so transcendental idealism) is not established. For, without it we cannot rule out that things in themselves might be necessarily spatial albeit unbeknownst to us. But, of course, 9 is highly questionable.

Fourth, it is unclear to me what is the justification 10. Does it follow from 2?

Fifth, premise one is a concern to those of us who doubt that Euclidean geometry is a body of necessary truths that can be known apriori. So some have tried to weaken this to something like,

1*. We have synthetic apriori knowledge of space (e.g. that it is unique)

Now, Guyer has a subsidiary argument to the effect that this can't be done consistently with some of Kant's other views. As I understand it, it goes something like this:

13. Necessarily, we impose a spatial form on the objects of experience
14. Intuition teaches us that the spatial form of the objects of intuition is Euclidean
15. The objects of experience have a determinate/particular spatial form
So,
16. Necessarily, we impose a determinate/particular spatial form on the objects of experience (form 13&15)
So,
17. Euclidean geometry is necessarily true (from 14&16)

Two things need to be said about this straight away:

First, Guyer admits that Kant could allow the factual falsity of 14. But he maintains that whatever the spatial form is that is revealed by intuition, it will reappear in a suitably modified version of 17. And, he suggests, this would be just as problematic. Perhaps this is because which geometry it is that correctly describes the spatial form of the actual world is an empirical issue (?).

Second, to my mind there is an equivocation in the argument. For, arguably, 16 is ambiguous between

16a. Necessarily, there is a spatial form that we impose
and
16b. There is a spatial form that we necessarily impose

As far as I can see, at best 16a would be licenced by 13&15, but it is 16b that is required to derive (with 14) the apparently unpalatable 17. If so, Guyer's further charge regarding premise 1 is unsubstantiated.

I have to say that I am uneasy attributing this slip to Guyer. If for no other reason than that it looks remarkably similar to Kant's alleged conflation of conditional and absolute necessity. But, as it stands, I can't see another way out.

Last thing...next week...

Wednesday (23rd) in room 4.137 at Essex University, between 12-2pm. We'll be reading Michael Friedman's, "Kant, Skepticism and Idealism", Inquiry 49:1, pp.26-43. Again, discussion will continue here.

Virtual Reading Group on Transcendental Idealism: Some thoughts on Allison's 'Transcendental Idealism and Descriptive Metaphysics'

This is a summary of some of the questions and issues that came up at yesterday's TPN reading group on transcendental idealism - thanks to the other discussants for an illuminating afternoon. What I say here is my own rendition, I leave others to speak for themselves. All page references are to: Allison, Henry E., 'Transcendental Idealism and Descriptive Metaphysics' Kant Studien 60:2 (1969), pp. 216-233.

I will concentrate in this post on summarising the following aspects of the paper: Allison's conception of Kant’s transcendental idealism; Allison's criticism of Strawson’s ‘descriptive metaphysics’ as falling short of answering the quaestio juris about knowledge that Kant’s transcendental turn makes central. This will also provide a useful way of helping to clarify the following questions: What is the thesis of transcendental idealism? How does the thesis justify knowledge? What kind of justification of knowledge can we expect transcendental philosophy to provide?

Allison says Strawson holds that “…the permanent significance of the Critique lies in its “investigation of the set of ideas which forms the limiting framework of all our thought about the world and experience of the world.”” (216) Strawson identifies this investigation with an extended ‘analytical argument’ that investigates “certain general…features necessarily ingredient in any coherent conception of experience which we can form for ourselves.” (220) But Strawson wants to detach this analytical argument from the ‘disastrous model’ of transcendental idealism within which Kant (apparently) couches it. That model has the following elements:

PF1: “… the doctrine that ‘reality is supersensible and that we can have no knowledge of it.’” (217)

PF2: The necessary features of our experience of an object are imposed on nature conceived as a “pregiven matter...by some kind of mysterious transcendental activity .” (222)

PF3: The “perverted model” of primary/secondary qualities which in Kant’s hands makes all empirical features of objects come out mind-dependent, yet which renders the contrast the original model employs unintelligible by destroying any means by which such qualities can be contrasted. (217)

I shall not concern myself much here with Allison’s partial defence of Kant with respect to his possessing a ‘hygienic’ version of this model. (This defence consists in: rejecting phenomenalism as an understanding of the assertion that representations are ‘in us’; finding a way of drawing the distinction between appearances and things in themselves that is intelligible and that flows from the a priori conditions on the possibility of knowledge rather than simply being a metaphysical assumption about what the latter are). Instead, I will focus on Allison’s contention that: “Strawson draws the line in the wrong place, and that the crucial distinction is not between transcendental idealism (including the model) and the ‘analytic argument’, but between transcendental idealism, of which the so-called ‘analytic argument’ is the chief expression, and the admittedly disastrous model in terms of which Kant interpreted it.” In making this contention, Allison hopes to: “…offer a defense of transcendental idealism, of the Copernican revolution, and show that Strawson is only able to avoid this idealistic commitment at the cost of begging the very question which Kant was most concerned to answer.” (216)

Before going on to lay out in broad terms Allison’s execution of his contention and defence, it is useful to clarify (as far as is possible!) some terminology.

The ‘Copernican revolution’ (216) seems to consist in the idea that objects must conform (in some sense) to our minds. This ‘revolution’ seems motivated by the failure of metaphysics and the triumph of Humean scepticism which are of a piece in signing up to ‘transcendental realism’ ― the thesis that knowledge or cognition consists in the agreement of our conceptualisation of the world with a completely mind-independent reality. Philosophy under the auspices of the ‘revolution’ consists for Kant in ‘transcendental reflection’, which entails a ‘fundamental change in the direction of philosophical enquiry’. (217) This change in direction consists in the ‘systematic substitution of the quaestio juris for the quaestio facti’ which Allison glosses as substituting ‘a consideration of the objective validity of our judgements for an investigation of the objects known.’ (218) As we shall see, a lot will turn on what is meant by ‘objective validity’ here. Suffice it to say for now that it seems to mean a justification of our cognitive activities by demonstrating that they are articulated by concepts and principles that necessarily and correctly apply to the objects of experience. This justification is garnered through transcendental investigation into the necessary and universal characteristics of cognition.

Allison says that the ‘Copernican revolution’ is both a presupposition and a consequence of the reorientation of philosophy as transcendental reflection. There seems to be a three step procedure here; (i) adopt the Copernican stance due to the failure of traditional transcendental realism, (ii) carry out a transcendental investigation of the necessary conditions of possibility of cognition, (iii) bring out that the initial adoption of the Copernican stance is justified by adverting to the fact that step (ii) has shown that cognition can only be of objects as they are subject to a priori concepts and principles, and thus that all knowledge must consist in objects in some sense conforming to the strictures of the mind.

Allison contends that ‘transcendental idealism’ is not dependent on what Strawson describes as the disastrous model (PF1-PF3 above), but is a ‘direct outgrowth of his [Kant’s] conception of philosophy’, that is, of the three step procedure just described. The content of this idealism seems to consist in the following two claims: (a) because of the role of universal and necessary concepts and principles in cognition, objects do, in some sense, conform to the mind, and; (b) because of this the ‘traditional’ model of transcendental realism is to be rejected (that is, transcendental idealism seems to consist in the validated Copernican stance of the third step outlined above). The rejection of transcendental realism appears to be based on the following thought: if we can never come across objects not subject to various a priori concepts and principles, then the idea of cognition consisting in the agreement between representations and a completely mind-independent reality is not simply open to sceptical attack, it makes cognition itself impossible. In Allison’s words, the transcendental realist model opens up an unbridgeable ‘epistemological gap’ between mind and world. However, adopting transcendental idealism leaves “neither the notion of ‘knowledge’ nor of ‘objects’ with their pre-transcendental significance.” (218)

For the moment, keep in mind both the lack of clarity with regard to what ‘objectively valid’ means, and what exactly bridging the ‘epistemological gap’ consists in, as these questions will become important in assessing whether Allison’s critique of Strawson’s descriptive metaphysics is warranted, and if it is, to what extent.

Allison characterises the ‘essence’ of transcendental idealism further as “the tautological yet significant reflection that if an intuition is to be brought to consciousness, and thus conceived as an object, or in relation to an object, it must necessarily be subject to those conditions under which it can alone be conceptualised.” (219) This is initially quite baffling. It is hardly tautologous that being ‘brought to consciousness’ is equivalent to being ‘conceived as an object’, as this begs the question against philosophers who believe in non-conceptual content. But I think this ‘reflection’ is being made after the acceptance of the claim that there are certain a priori conceptual capacities that structure cognition (namely the categories), so that being brought to consciousness just is, according to the claim, being ‘conceived as an object’ (I am not here assessing the arguments for this claim, but trying to understand the ‘significant tautology’ it is embedded in). So the tautology seems to unpack into the following: given that there are a priori constraints on cognition, any object which is cognised will be subject to these constraints.

That this is a ‘tautology’ presumably renders it hygienic: disentangles it from the ‘disastrous model’ because adherence to it enables; 1) (contra PF1 and PF3) the ruling out of knowledge of completely mind independent things in themselves (presuming we can, from within the Copernican standpoint, legitimately refer meaningfully to the latter - but I set aside that question); 2) (contra PF2) an understanding of what it means for objects to necessarily conform to the mind without invoking the ‘imposition model’ (although there is still the question of how transcendental proofs discern necessary conditions of possibility of cognition without this model, but I leave aside that question also). That the ‘tautology’ provides hygienic versions of PF1-3 presumably makes it ‘significant’.

Allison then articulates what he takes to be Kant’s version of this ‘tautology’;

“Now, the crucial move of the Transcendental Deduction is contained in the reflection that these same concepts [the categories], ‘as a priori conditions of possible experience in general are at the same time conditions of the possibility of objects of experience’, (A111) and therefore possess ‘objective validity’. In saying this Kant is in effect claiming that these concepts are not only ways in which we must conceive of the world, but also the grounds of our objectively valid judgements about the world, and he can make this claim because he believes that he has shown that as ‘concepts of an object in general’, they serve to define what can alone count as an object to a finite mind.”

Everything turns on understanding some of the formulations in this passage. First, let’s start with what’s relatively clear. The phrase ‘a priori conditions of possible experience in general’ seems clear; this formulation refers to the concepts and principles that have been shown via transcendental proof to be necessary conditions of the cognition of any object (again, I leave aside the question of how ‘transcendental proof’ works, which is not at all immediately clear). But it is far from clear what the phrase that follows this one means ― the phrase ‘conditions of the possibility of objects of experience’ ― since it is asserted as equivalent to the first in some way. Working out the meaning of this ‘equivalency thesis’ is crucial, because it allows us to understand what ‘objective validity’ means. Does the equivalency consist in the ‘conditions of the possibility of objects of experience’ being the same as the ‘a priori conditions of possible experience in general’, i.e. consisting in the claim that objects of experience really are (at the empirical level) causally governed substances and so on? Or does it consist in them simply always being known as such? Allison seems to favour the latter, which is evidenced in his qualifying the claim: ‘these concepts are not only ways in which we must conceive of the world, but also the grounds of our objectively valid judgements about the world’, with the following statement; ‘he [Kant] can make this claim because he believes that he has shown that as ‘concepts of an object in general’, they serve to define what can alone count as an object to a finite mind.’

So the answer to the quaestio juris seems to be that our taking ourselves to have knowledge is justified, because the objects of which we purportedly have it necessarily conform, in their being known, to certain a priori concepts and principles. Thus we can be assured that the claims we make about the world (claims that are articulated by those concepts and principles) correctly limn that world ― that is, the claims we make at the empirical level, if all goes well, are true. But, the quaestio juris is only answered in this way, because we have adopted transcendental idealism ― we have adopted a cognitive model where true claims about the world are made true in part by the objects they are about conforming to the strictures set out by the mind (by reconceiving the object as the 'object of possible experience').

Now let’s move to Allison’s beef with Strawson (I will leave out Allison’s rebuttal of Strawson’s allegations of Kant’s phenomenalism). The questions that came up yesterday when we discussed this were: What does Strawson give us? Why does Allison not think it is enough to answer the quaestio juris? (Or, perhaps better; Why does Allison think there is some quaestio juris left to be answered once Strawson has executed his neo-transcendental project of descriptive metaphysics?)

Strawson argues for ‘looser substitutes’ (225) for Kant’s a priori concepts and principles, but still claims that knowledge of these more deflationary concepts and principles is non-empirical knowledge of what “‘understanding’ requires of a possible experience in time.” (225) So presumably the Strawsonian principles and concepts are universal and necessary features of ‘our’ conceptual scheme ― i.e. this is what being ‘invariant features’ amounts to. Let’s grant for the sake of argument that this is so and that Strawson successfully establishes the following claims:

1) That we in fact use certain concepts and principles such as some of the categories

2) That we necessarily experience the world as a unified spatio-temporal realm which displays causal regularities and thus are justified in using some of the categories that reflect this fact

3) That we can know a priori that we will use some of the categories

These claims, although involving less specific a priori claims about possible experience than Kant advances, seem to together constitute a version of the ‘equivalency thesis’ expressed by the ‘significant tautology’ cited earlier. That is, they together license the following claim; no objects that do not conform to the concepts and principles we necessarily employ can be objects of knowledge for us. But Strawson denies that accepting this claim entails accepting that we cannot have knowledge of completely mind independent objects. This amounts to accepting some form of ‘transcendental conditions’ on possible knowledge, but refusing to frame this acceptance within transcendental idealism. Allison thinks this combination is unstable, because it leaves open the ‘epistemological gap’ that Kant sought to close with his proof of the ‘objective validity’ of the categories and principles. That is, Strawson’s descriptive metaphysics can show that the latter apply necessarily, but not correctly. For if the objects of knowledge are allowed to be completely mind independent, as they are on Strawson’s model of knowledge, then the quaestio juris goes unanswered ― we may necessarily apply concepts and principles all we like, but this will not justify knowledge if we cannot show such application to yield true claims about the world.

The root of the problem as Allison identifies it is Strawson’s adherence to transcendental realism (see pp. 232-233). It is adherence to this latter epistemic/cognitive model that disallows Strawson from closing the ‘epistemological gap’ and answering the quaestio juris. Allison further criticises ‘descriptive metaphysics’ for not yielding a ‘rational justification’ of our employment of purportedly a priori concepts such as the concept of causality. (231) He argues that Strawson’s method of transcendental argument (see 229-230) can only secure the result that we necessarily employ certain concepts. Yet this is not something that a Humean would deny, agreeing that such concepts are indispensable yet are nothing more than ‘natural beliefs’ compelled in us by ‘nature’s wisdom’. As is it is a desideratum of the quaestio juris to answer Hume’s scepticism by showing the ‘objective validity’ of our employment of the concept of causality, here again Strawson’s project apparently fails to close the ‘epistemological gap’.

These last two criticisms of ‘descriptive metaphysics’ bring us nicely to a question which also came up a lot in discussion yesterday; What kind of justification of knowledge do we expect transcendental philosophy to provide? Allison demands we demonstrate the ‘objective validity’ of the central concepts and principles that articulate knowledge claims, and that this requires accepting transcendental idealism. But might we accept a weaker justification, which Strawson seems intent to give? Namely, simply show that it is necessary to conceive the world in a certain way. Once this is done, the therapeutic benefit is that various metaphysical questions dissolve. For example, if we cannot but help conceive the world in terms of substances and causes, the question of whether the world really is so composed becomes a non-question. If we must so conceive it, we will never be able to apprise ourselves of the world not being so constituted. What’s more, Allison’s criticism of Strawson for not doing any better than Hume in demonstrating the indispensability of such conceptions is unfair: Hume does not say we must employ such conceptions for a priori reasons, but rather because our nature compels us. Given different natures, we might have different compulsions. Strawson’s transcendental arguments (presuming they are successful) could be said to rule out such ‘cognitive relativism’ ― rather, any self-conscious creature like ourselves will be possessed of broadly the same conceptions of reality. And if this is so, then the conceptions are necessary and universal rather than simply indispensable and contingent. There is different kind of modal justification given by the former claim; it says we cannot but help employ our conceptions and they are not parochial. Perhaps this justification of ‘cognitive universalism’ plus the therapeutic dissolving of various metaphysical questions is all we can expect from transcendental philosophy?

One final point. On page 223-224 Allison cites Strawson’s complaint that there is no reason to see the interconnectedness amongst objects necessary for the self-ascription of experiences as holding amongst representations rather than the objects themselves. And, if only the latter kind of interconnectedness were necessary we would have no reason to employ the ‘surrogate’ of a ‘transcendental object’ which is the correlate of the synthetic unity of judgement. Allison complains (rightly in my view) that Kant is not arguing here for replacing empirical objects with intentional objects as that of which we are aware. Rather, his point is that once we see it is the unity of consciousness that requires such interconnectedness, we must understand Kant as “offering us a transcendental re-interpretation of what is meant by the real.” Allison then goes on to say that Strawson raises “the possibility that we might have an awareness of objects distinct from our representations, without any consciousness of the necessary, rule governed coherence of these representations.” This possibility would consist in there being such interconnectedness without the subject of experience being aware of it, yet that subject still enjoying awareness of objects as distinct from herself. Allison complains this would be tantamount to the idea “that we can have awareness of objects apart from the conditions under which we can alone become aware of them.” This seems significant. One of the reasons why Strawson might not see necessary conditions of experience as entailing ‘a transcendental re-interpretation of what is meant by the real’, is that he conceives of the conditions perhaps obtaining in a wholly externalist fashion. But if they must in some sense always be internal, doesn’t this rule out Strawson’s detaching transcendental idealism from transcendental arguments?

Transcendental Idealism Reading Group: Update

Matt will be kicking off the discussion of Allison's paper in the next day or so. In the meantime, I thought I'd give you advance warning of what we'll be looking at next week,

Guyer, P. "Rehabilitating Transcendental Idealism?" in Schaper & Vossenkuhl (eds.), Reading Kant, (Oxford: Basil Blackwell, 1989), pp.140-167.

Transcendental Idealism Reading Group

As previously announced (but a little late), this term we are running a reading group on Transcendental Idealism. Our first meeting will be on Wednesday 9th May, at 12pm in the Philosophy Common Room, University of Essex. For that meeting we will be reading:

Allison, Henry E., 'Transcendental Idealism and Descriptive Metaphysics' Kant Studien 60:2 (1969), pp. 216-223.

The paper is available online here (subscription required).

Discussion will continue here on the tpn blog. Anyone unable to make the meetings but interested in taking part, is warmly invited to contribute in this way.

Further readings will be announced next week.

Virtual reading group on transcendental idealism/ Call for nominations

As part of the TPN project (www.essex.ac.uk/philosophy/tpn/), we will be running a 10-week reading group next term (w/c 23 April). The aim is to concentrate on contemporary work on the nature of transcendental idealism. Each week will be devoted to discussion of a single paper or chapter.

The group will meet in the Philosophy Department at Essex University, but discussions will extend to the TPN blog (here) so that participation is not restricted to those who can be physically present.

We are currently drawing up the list of 10 weekly papers/chapters, and welcome suggestions of pieces for inclusion. Suggestions can be emailed to tpn@essex.ac.uk or posted here. The resulting “top 10” will be posted on the web in advance of the reading group getting underway.

Participation is open to all.

Karl Ameriks at the Transcendental Idealism workshop

Karl Ameriks gave a paper entitled 'Kant’s Idealism: Once More into the Breach'. In it he sought to find space for a coherent 'modest metaphysical' reading of Kant's theoretical philosophy, where 'metaphysical' means allowing some sort of sense to Kant's various ways of talking about things-in-themselves, and ‘modest’ means that such talk flows from the a priori structures of experience rather than dogmatic assumptions. Karl argued the key to making tenable such a reading is to reject the false choice between accepting talk of things-in-themselves but only in the context of neo-Berkeleyan phenomenalism, and denying such talk in the context of a neo-Strawsonian descriptive metaphysics.

Karl achieved this rejection via a 'multi-stage, multi-level' reading of Kant's Critical Philosophy. He argued that Kant’s Critical Philosophy is methodologically multistage because it starts from non-inferentially accepted core claims consonant with common sense realism (the subject matter of which are immediate ‘mental-sensory’ representings); moves inferentially through transcendental claims about a priori experience-structuring features that philosophical reflection reveals claims at the first level require; and then goes on to show that these a priori features are only possible if we deny transcendental reality to space and time. This multistage understanding of experience yields a multi-level ontology that is idealist but which does not undercut the common sense realism that the position began from.

Karl’s reasons for arguing that a multi-stage methodology yields a multi-level ontology (and thus TI) were complex and interesting, but I shall not go into them here. I will just mention a couple of points that came up in discussion. First, it was suggested that it was not at all clear that Kant’s talk of empirical reality being ‘mere representations’ at the third methodological level does not undercut the robust empirical realism reinforced at the second level. Second, it was suggested that once we give up the negative deflationary conception of things-in-themselves as merely what empirical reality would look like not from our perspective (or no perspective at all), we are unconstrained in the metaphysical positions available to us as positive characterisations of noumena. For example, what is to stop us postulating an infinite hierarchy of levels of the latter, all supporting each another, rather than viewing noumena as some sort of underlying ‘causal’ substrate to appearances? In other words, what is the Critical reason for taking the latter metaphysical line?