A Priori Workshop: Finn Spicer

In his talk ("A priori knowledge - you either love it or hate it"), Finn Spicer claimed that there is no urgent reason for naturalism to resist the concept of a priori knowledge. To support this claim, he raised a number of semantic points. If meaning is reference, to begin with, we encounter difficulties discriminating between a priori and a posteriori truths. The sentence “Napoléon is Napoléon” is an uninformative tautology and a priori true. The sentence “Napoléon is Bonaparte” is informative and a posteriori true. Both sentences have the same meaning, however, because they have a common referent. So, reference doesn’t give us any clue why they differ with regard to aprioricity. If the picture is extended to include sense, Finn claimed, these difficulties do not arise. For instance, the sense of ‘Napoléon’ is the knowledge necessary for having a concept of Napoléon. Or it is the conception associated with ‘Napoléon’ (Napoléon was a man from Corsica, won the battle of Austerlitz, spent his last years in exile etc).

Finn now suggested (if I understood him correctly) that aprioricity comes back into this in a naturalism-friendly way if the conception of ‘Napoléon’ is such that it contains bits of knowledge from a reliable source, and is needed to have the concept expressed by the name ‘Napoléon’. This is to say that some Napoléon-knowledge (e.g. Napoléon was a man from Corsica) is necessary to have the Napoléon-concept, and is a priori in that sense, without being knowledge in virtue of its a priori status.

A recurrent criticism of Finn’s approach was that on his account epistemic and semantic matters easily come apart. For the conception of Napoléon to have a priori items (those necessary to have a concept expressed by ‘Napoléon’) is one thing. For these items to be knowledge from a reliable source is another. Even if a conception contains both, they may not (never) be congruent.

A Priori Workshop: Peter Railton

In his paper, "A Naturalistic Transcendental Argument? (or, Why should Kantians have all the fun?)" Peter Railton argued that a naturalistically acceptable quasi-transcendental argument can be mounted against the sceptic. If I understood it (and I'm far from sure I do), the argument relied on the conditions necessary for a system to be in a position to learn from experience. The condition is, roughly speaking, that the epistemic options must be investigated with an expectation that assigns differential probabilities to each. If each possibility is assigned equal probability, then a system will not be able to rely on the information it gains to adjust the probabilities of further options. Further, it doesn't matter so much exactly which expectations the system begins with as, if there are epistemic reasons for the system to latch onto, then there will be a tendency for systems to converge on those reasons.

What this means for the sceptic is that his demand that his demand that we put aside all prior assignments of probabilities to the sceptical versus the non-sceptical hypotheses, is inconsistent with the supposition (sure made by the sceptic) that we are creatures capable of learning. This doesn't show that the sceptical hypothesis is false (there may be no epistemic reasons), nor that it is meaningless. Rather, it suggests that, contrary to what the sceptic wants to insist, we are justified in making assumptions (embodied in our behaving in accordance with certain norms) concerning those things the sceptic doubts despite not having an apriori 'from-scratch' argument that shows that these assumptions are true or objectively likely to be true.

A priori workshop: John Skorupski

John gave a paper entitled 'Aprioricity and Normativity'. In it he talked about a certain form of a priori knowledge we have about the relations between reasons. A thinker's epistemic reasons are relative to what John called her 'epistemic field'. This is a set of facts at a specific time that are ascertainable by the thinker, but not necessarily ascertained. If there is outright sufficient reason to believe p then that belief has a priori support in the relevant epistemic field. John then talked of a narrowed epistemic field which he called an 'epistemic state'. A fact p is part of an epistemic state if the obtaining of p at a particular time gives a monotonically sufficient reason for the thinker in this state to believe that p. An example of such a fact might be that if I'm thinking and I have the thought that I'm thinking, I can know a priori that the latter thought is true (no further adduced evidence can falsify the thought).

John then went on to describe 'a priori warrant' as a phenomenological attitude towards a thinker's belief that she has sufficient reason to believe a fact p. If she bases that sufficient reason on her being in an epistemic state which gives outright sufficient reason to believe p, then she has a priori warrant. He then gave what he called the 'critical argument' where he argued that a thinker could not generate the relevant attitude to undewrite a priori warrant with regard to any a posteriori fact. Given his argument for a priori warrant, the conclusion was what John called the cognitive irrealism of the 'normative view' - that the apparently factual a priori propositions we know about reasons are in fact non-factual offshoots of synthetic a priori knowledge of reasons. What such knowledge has as its object is something like what Kant called the pure principles of the understanding - presumably principles of epistemic reasoning as such.

A priori workshop: John Callanan

John gave a paper entitled 'Kant's a priori and seeing how things must be'. In it he distinguished between two types of necessity that get talked about, and not always kept apart, in the first Critique. The first is the familiar necessity of synthetic a priori truths about the transcendental structure of experience - the necesary conditions of possibilty of experience that hold with 'strict universality' for any thinker/perceiver as such. The second is the necessity which attaches to certain experiences, so that we literally, in the course of experience, 'see' that, say, a certain temporal ordering of events is necessary rather than contingent. In uncovering this second source of necessary truths, Kant is very roughly doing a kind of phenomenology influenced still very much by the early modern philosophy that preceded him. For instance, very much influenced by Hume's reflections on how sense impressions are given to the mind and ordered by it in certain ways. What Kant changed was the Humean demarcation between the given and constituting elements of experience (for example, the representation from which the concept space is derived is constituted not given for Kant). Also, concepts derived from what is constituted are legitimate for Kant, whereas they are not for Hume. This is because for Kant they are a necessay condition of the very experience in question (for example, the constituting element of the causal principle is a necessary condition for Kant of the possibility of us experiencing certain sorts of temporal order at all).

But John also argued that for a lot of Kant's arguments in the Critique, he is concerned not with so called transcendental arguments, but rather with a phenomenological analysis of the understanding at work in experience. Kant works with quite a rich conception of experience and John's focus was his analysis of how we 'see' that something is the case, like when we see that in turning a cube a certain coloured side is bound to reappear at a certain point. We might say then, that the focus is on an a 'seeing that' which is suffuse with the conceptual activity of the understanding.

This general approach has ramifications for how we view Kant's philosophical project. John argued that we tend to see Kant through Frege's anti-psychologistic spectacles, that we see him as concerned with the conditions of possibility of thinking per se. But on John's view, Kant is far closer to Hume. He is involved in a phenomenology of experience that uncovers constituting elements that are merely human. Of course, this interpretation will effect how we think of the Kantian a priori. That is, on John's account it quantifies over a far more limited domain of individuals than it is often taken to.

I for one think John's approach is fruitful and probably right, certainly as a reading of the Aesthetic and the Analogies. But there seems to be a problem over the Transcendental Deduction, which appears to be concerned with the conditions of possibility of thinking as such. But there is a way out: perhaps Kant is concerned in those sections, as McDowell has argued, with another problem: that of intentionality. The TD could then be seen as arguing for certain a priori concepts that constitute thoughts and perceptions alike, so that perceptions have a certain minimal conceptual content. Kant's universalism about logic then becomes detatchable from the TD and it can fit with John's overrall interpretative approach of seeing Kant as concerned with analysing phenomenologically the understanding at work in experience.

Maths and Logic workshop: Emily Carson

Emily Carson gave a paper entitled ‘Arithmetic and the Possibility of Experience’. In it she sought to work out Kant’s position on how arithmetic is grounded in the transcendental structures of experience. This questions asks, very roughly, how we can explain that every ‘discursive’ thinker - as Kant puts it - possesses some arithmetical knowledge, and how arithmetic applies absolutely generally to every empirical object (every such object is quantifiable). In the case of geometry Kant’s position is fairly clear: geometry is the pure science of space and our pure intuition of space is the source of our a priori geometrical knowledge. But there is no comparable science of number for Kant, the source of which would be our pure intuition of time.

Emily argued that the source of arithmetical knowledge are the categories of quantity, categories that are involved in every synthesis of any manifold of intuition. She explained this by arguing that the threefold synthesis cited in the A-edition Deduction, a synthesis apparently necessary for any determinate thought content whatsoever, is a synthesis that minimally involves the categories of quantity. Thus the formal element of possible experience that arithmetic is grounded in is this synthesis, and this explains Kant’s cryptic statement that the formal element in experience that grounds arithmetical knowledge is‘the universal in the synthesis of one and the same thing in time and space, and the magnitude of an intuition in general (number) that arises from that.’

I for one found this a very convincing account of Kant’s position. But one question that arose was that if all that grounds arithmetic is the categorical synthesis of quantity, doesn’t that give us an unKantian answer to how arithmetic and the possibility of experience relate? That is, arithmetical capacities now start to look wholly conceptual. It came up in discussion that it seems necessary for there to be some intuitive structure to index the categories to, so that arithmetical abilities didn’t depend on contingencies like the amount of objects in the world (if there were only four objects, the sentence 7+5+12 couldn’t be formulated unless there were an infinite iterative structure like time that has enough places to index the constituent numbers of the sentence to). What was then asked was, why should this indexing structure be specifically time or space? To my mind, against Kant, this question is wrongheaded: he is not trying to show how any discursive thinker would be capable of arithmetic, but how we are.