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.

Reading Group: David Macarthur, 'Naturalism and Skepticism'

Kant called it a veritable scandal for philosophy that claims about the existence of the external world have to be taken on faith. Macarthur echoes this sentiment. He is alarmed by the fact that most philosophers today seem indifferent to this scandal although there is still no satisfactory proof for the existence of the external world. In his view, the culprit is (scientific) naturalism. Naturalism dismisses the epistemological preoccupation with refuting the skeptic as an exaggeration of the underlying problem. It takes external world skepticism (knowledge of the external world is not rationally justified) to be defective in certain respects so that its refutation is no precondition for satisfactory accounts of knowledge. Macarthur concedes that quietism with regard to the skeptical problem is an option. However, he thinks that naturalism lacks the resources for sustaining this option. If that is true, then naturalism simply begs the question against the skeptic and thus exacerbates external world skepticism.

Hume was unambiguous about the skeptical consequences of his naturalism. According to him, all mental processes are causal processes. If all mental processes are causal processes, then subjective experiences are intramental effects of some cause. Intramental effects can be produced by an infinitely large number of causes and for that reason do not allow us to tell what their actual effective causes were. So it may be the case that, contrary to our presumptions, none of these causes issues from the external world. This causal theory of experience results in a systematic mismatch between mind and world which leads to external world skepticism.

Naturalism shares Hume’s premiss but seeks to avoid the conclusion. All attempts to do this, Macarthur argues, are dogmatic in one way or other. (Before he turns to representative naturalistic responses to external world skepticism, he rejects inference to the best explanation as a suitable anti-skeptical instrument. Even if references to the external world explain best what makes our experiences coherent, these reference are pointless if our subjective experiences are not coherent, as seems to be the case on the causal theory of experience. This is simply to reinforce the point that if inference to the best explanation is one of the most powerful tools of naturalism, it cannot hope to refute the skeptic but must settle for the quietist option.)

Macarthur discusses three representative examples for naturalistic responses to external world skepticism. These are

(1) Quine’s position that skepticism cannot doubt all of knowledge at the same time. As far as skepticism merits an answer, there is common ground with science – those bits of empirical knowledge which are not currently in doubt. External world skepticism doubts all of knowledge and thus is unanswerable.

(2) The reliabilist view that replaces rational justification with a causal account of belief formation. If beliefs are produced by reliable processes, we can have true beliefs about the external world without having a subjective experience of its justification. So the skeptic’s view that there is no external world because we do not have rationally justified knowledge of it is badly motivated.

(3) Finally, Strawson’s interpretation of Hume according to which the belief in the external world is natural, not rational. Although what the skeptic says against the external world is intelligible, the lack of rational justifications for beliefs in the external world does not affect the natural commitment to this notion.

Macarthur finds all these responses unsatisfactory. Here are (what I take to be) his main reasons for this assessment:

(ad 1) It is wrong to assume that external world skepticism doubts all of knowledge. It just doubts knowledge about the external world. So external world skepticism is answerable by naturalist standards.

(ad 2) Reliabilism may have a point on possible knowledge. But actual knowledge involves some subjective experience of justification. So the skeptical point retains its force. Furthermore, reliabilism cannot properly account for the reliability of belief-forming processes. Reliability is either posited dogmatically as the property of some process, or accounting for it leads to infinite regress because reliability is accounted for in terms of other reliable processes.

(ad 3) Natural belief cannot be revised or confirmed empirically. This contradicts naturalistic principles. Then, the notion of natural belief obscures the connection between belief and reason. Contrary to assumptions or hypotheses, Macarthur claims, beliefs are always open to rational criticism. External world criticism, like any other skepticism, insists that beliefs be given up if they become untenable. Strawson’s Humean “natural belief” is untenable by these standards.

Macarthur concludes that these representative naturalistic responses fail to stop the Humean move from naturalism (his causal theory of experience) to skepticism. Naturalism has not shown that the skeptic need not be refuted and therefore begs the question against the skeptic. So quietism on naturalist terms is not a viable solution.

Just two quick observations:

(I) Part of Macarthur’s point against (2) is that accounts for reliability are infinitely regressive. But this may not be as damaging for reliabilism as he seems to think. The accuracy of inductive reasoning may be enhanced by further induction (Popper, not Hume). Similarly, subsequent reliable processes may have corrective effects on the original process. So reliability may not be the dogmatic notion as which Macarthur portrays it.

(II) Macarthur suggests that external world skepticism withstands all these responses. However, to withstand them, it has to be modified in each case. The skepticism that survives (1) is not the same as the skepticism that survives (2). As (1) and (2) can be held jointly, naturalism might try to corner external world skepticism by combining these responses. I hope I can make this a little clearer:

To manoeuvre external world skepticism past (1), the skeptic must accept that there is some rationally justified knowledge, just not that about the external world. If positioned against (2), this appears to amount to the concession that there is some actual reliabilist knowledge – as there is rationally justified knowledge about something, though not the external world, the reliability of the process by which this knowledge is reached is confirmed as well. It does not seem impossible, then, to expand this reliable process tentatively, to close the causal gap between mind and world and to include beliefs about the external world. These cannot be not true because the process is unreliable, and this seems to hold regardless of the availability of rational justifications for them.

Mind Workshop: Avramides

Anita Avramides, "Knowing vs. Acting with Others: Some Background", argued that an otherwise promising approach to knowledge of others, exemplified by early knowledge externalists such as Goldman and Dretske, fail to account adequately for the relations we bear to others. This is for the reason that they attempt to account for knowledge of others against a background that leaves open as a possibility the sceptical hypothesis that we lack knowledge of others - we can have knowledge of what another is thinking without knowledge that another is thinking. Anita argued that a recent view, proposed by Quassim Cassam, that we can see that others are in mental states, does not really progress very far from this basic externalist position. Instead, she argued, we should look to Wittgenstein who conceives of the background to our knowledge of others in terms a kind of social interaction that leaves no room for the sceptical possibility to take root. In this way, we can take a more satisfactory view of what is wrong with the sceptical arguments.

VRG: Maddy on Hume

The chapter on Hume (I.3) is entitled “Hume’s naturalism”. However, Maddy’s chief interest here is in Hume’s skepticism and in the connection between Hume’s skepticism and his naturalism. Her general line is that for skepticism to follow from naturalism, as is the case in (some interpretations of) Hume, this naturalism must be defective.

It is not difficult to see how this chapter connects to the previous ones. Against straightforward Cartesian skepticism, the SP maintains that an inquiry yielding reasonable beliefs can go ahead without the seal of certainty. Against extraordinary Cartesian skepticism, she maintains that an external justification of ordinary knowledge is not asked for. Humean skepticism is in line with the SP’s thinking so far. It raises cautionary rather than fundamental doubts, because it has roots in everyday empirical reasoning and does not adopt an “external” perspective on knowledge. However, Humean skepticism poses a different and more serious threat to the SP’s project. If skepticism is the “internal” consequence of an investigation based on ordinary perception and reasoning, she cannot simply undercut or ignore the skeptical challenge. In this case, the very stuff of her worldview, reasonable belief, generates doubts about rational warrants for such beliefs and thus undermines her investigation.

There are two basic ways of looking at the relation between naturalism and skepticism in Hume. Either his skepticism and his commonsense approach are distinct from each other, or his position is of one piece. The first set of accounts is less interesting. If seen as a separate issue, Hume’s skepticism can be treated more or less along the lines of the skepticism about external justification. Given the problem of an “internal” skepticism, Maddy concentrates on the single Hume for whom skepticism is a consequence of naturalism. She considers two versions of this interpretation. On the first of these (Broughton), Hume’s skepticism reflects clashes between norms of ordinary perception and reasoning. On the second of these (Stroud), Hume’s skepticism arises from his naturalistic rejection of the Cartesian ego. On the first view, there is a direct connection between Hume’s naturalism and skepticism. On the second view, this connection is indirect. It has the character of a critical appropriation, on the part of ordinary knowledge, of answers to extraordinary skepticism.

The direct connection between Hume’s naturalism and skepticism (after Broughton) can be pictured as follows. Ordinary perception and reasoning is subject to norms of clarity, consistency, carefulness, certain styles of causal argumentation and the existence of an external world. According to Hume, causal reasoning and the commitment to the existence of an external world can come apart in ordinary perception. I won’t rehearse Hume’s argument here, nor Maddy’s brief gloss on it. The familiar outcome is that ordinary reasoning can only rely on resemblance and contiguity, not cause and effect, when making connections between matters of fact. It can motivate us to attend to different ideas but falls short of reasonable beliefs. So if norms of ordinary perception and reasoning can come apart, we are left with certain habits of thought, not with stringent reasoning. This inspires cautionary skepticism.

Maddy’s SP concedes that ordinary cognitive norms can clash. Where Hume sees irrevocable confusion, however, the SP still sees potential for improvement. In her view, Hume is mistaken to conclude from the failure of some norms of ordinary perception and reasoning that they fail all at once. From the SP’s perspective, the norms of ordinary perception and reasoning are open to reassessment and refinement, much like the claims of second philosophy in general. So if there is a clash between these norms, some norms should be amended or changed. This process leaves enough ground for the SP to stand on, even if there is an incompatibility between, say, principles of causal reasoning and commonsense realism. In short, Hume’s direct “internal” skepticism can be defused by appealing to a more thoroughgoing fallibilism. According to the SP, Hume’s declaration that these clashes are irrevocable and irremediable falls short of this “spirit of open inquiry” characteristic of naturalism. So his skepticism does not follow from naturalism properly conceived, and the SP prevails.

The indirect connection between naturalism and skepticism (after Stroud) does not involve specific ordinary cognitive norms. For Hume, reasoning as such is part of the natural causal order. Hence its opposition to the detached thinking substance of Cartesian philosophy. Hume’s basic argument against the Cartesian ego is that standards of detached reasoning fail, because they do not reflect how human beings are like from an empirical perspective. If measured by the standards of detached reasoning, empirical human beliefs could never be reasonable. Clearly, the SP agrees with this critique.

But the indirect view doesn’t leave it at that. It describes Hume’s skepticism as something like an after-image of his critical engagement with the Cartesian ego. Stroud argues (in oddly dialectical fashion) that this engagement prompts the transfer into the commonsense view the traditional philosophical attitude that something is lacking from ordinary practices, turning this attitude against the traditional philosophical view. What Hume has learned from the traditional philosophical perspective, according to Stroud, is that ordinary practices can be accounted for comprehensively by distancing the account from ordinary terms. Against the detached standpoint, however, Stroud’s Hume envisages a distancing from the ordinary view that is continuous with this view. The perspective anticipated by Hume—as I understand Stroud, and I’m not sure if I get his point—is that of a sustained naturalistic investigation of human life which is a fully philosophical naturalism, a philosophical vindication, rather than a mere restatement, of ordinary perception and reasoning in the medium of that reasoning. Hume’s skepticism is an after-image of this Cartesian impulse, because Hume is kept from making sense of a comprehensive account of human nature in the medium of ordinary reasoning by his associationism. He fails to carry forward his naturalistic project, so the indirect view, because he cannot revert to a realist picture of human nature.

To the SP, the most important virtue of the indirect view is to have exposed another insufficiency in Hume’s naturalism. As before, this confirms to her that Hume’s “internal” skepticism does not follow from naturalism properly conceived and thus doesn’t present a threat to the SP’s own project. (She does not make much of Stroud’s transfer from Descartes in this context. To her, the naturalistic project of a Hume cured from his doubts would be a plain triumph over traditional philosophy.) In the concluding lines, the SP suggests how to remedy the limitations of Hume’s naturalism. An obvious course of action (already suggested by Stroud) is to fully naturalise the mental side of Hume’s views on perception and reasoning. The SP’s therapy (briefly hinted at here and presented in the second part of Maddy’s book) will replace associationism with evolutionary psychology.

Questions: Can the SP expand her fallibilism about reasonable belief to the norms of ordinary perception and reasoning without further ado? Are these norms just reasonable beliefs? Is this second order fallibilism? Either of these would be strange, or require more argument, it seems to me. Aren’t reasonable beliefs about the world revised against the background of cognitive norms as well? If new evidence forces us to revise our theory, doesn’t this involve readjusting e.g. offset explanatory models against norms of consistency rather than revising these norms?
It may be a bit unfair to make much of Maddy’s concluding gestures about evolutionary psychology. Still, given that Hume’s skepticism culminates in the claim that our reasoning is truly just a habit of association, we might wonder whether the SP’s evolutionary psychology doesn’t simply discover further habits and enshrine them as reasoning. How would this defeat skepticism?

VRG: Maddy on Descartes & Stroud

Maddy begins the book with what she calls a 'character study' of someone referred to as 'the Second Philosopher' (SP). It's tempting to cut out the middle-woman and just call SP 'Maddy', but I shan't - on the off-chance that there is some distinction to be made between their views. In any case, the proposed plan of Part One of the book is to offer an account of what second philosophy (read 'austere naturalism') is by way of seeing how SP responds to the views of various philosophical figures. These are, primarily, Descartes, Stroud, Hume, Kant, Carnap, Quine and Putnam, each of whom is devoted a section to themselves. In this first post, I'll discuss sections 1.1 and 1.2, on Descartes and Stroud respectively.

1.1 Descartes's first philosophy

The focus of this short section is the nature and purpose of Descartes' method of doubt. Maddy makes two key claims about Descartes and gives two important clues as to how she pictures SP's response.

The first claim (p.12) is that Descartes does not think that the method of doubt shows that our ordinary beliefs are unjustified or unreasonable. Rather, what the method shows is that they are not certain (more accurately, not certain at this point in the dialectic of the Meditations). I forgo the question of whether this is an accurate portrayal of Descartes. The important point is that the SP accepts it. She agrees that she is not absolutely certain that, for example, she is not dreaming but continues to think it reasonable to believe that she isn't and also that she knows that she isn't (p.16). The first point to note is that, we are talking here about ordinary dreaming as we all know it. Second, SP is endorsing the view that a belief's being reasonable (which I take to be something like justification) does not require certainty. Indeed, a belief's counting as knowledge does not require certainty. How is it that SP thinks she knows that she isn't dreaming, and so has hands? Well, she notes phenomenological differences between waking and dreaming consciousness, and notes that her current experience is of the waking sort. She then cites various facts about human perceptual and cognitive systems that she thinks allow her to come to beliefs about middle-sized objects hands. End of story.

But Descartes wants to push SP further and get her to suspend all her reasonable beliefs, or rather all those that are merely reasonable, i.e. not certain. The way to do this is to suppose the she is being deceived by an evil demon. But why? Here is the second key claim about Descartes. The purpose of the method of doubt is to place scientific enquiry upon foundations that can be known with certainty. A subsidiary reason, highly specific to Descartes' own project, is that the method will wipe the slate clean in order that Aristotelian metaphysics can be replaced by his own mechanistic corpuscularism. This is the purpose of the evil demon hypothesis. And SP's response to this latter reason is predictable. SP already believes the most up-to-date scientific theories, so lacks this motivation to go in for the method of doubt. To the first reason, SP agrees that it would be a good thing if her (reasonable but not certain) scientific theories could be based upon absolutely certain principles, and so she sees what Descartes has to offer. Noting that his project fails to deliver the result it promises, she leaves the method of doubt to one side and continues as she was before.

What are the clues here concerning SP's methods? They are related. First, when asked how she knows that she has hands (or that water=H20) she does not say that science says so and that she trusts science. Rather, she cites various facts. Facts that we would call scientific. Second, when confronted by Descartes' method of doubt she does not reject it as 'unscientific'. Rather, she judges it according to both its own and her own criteria of success. It fails, and so she moves on. Both of these are important. For SP does not have a general account of what counts as a scientific approach, by means of which she sorts the acceptable from the unacceptable theories and methods. Rather, her pragmatic attitude is that she will be happy to accept what works. This is a line that reoccurs throughout Part One of the book, and it marks a distinction between SP and a number of other naturalistic philosophers. Naturalists sometimes distinguish between the scientific and the non-scientific (or the empirical and the non-empirical) in order to draw the boundaries of legitimate enquiry and methodology. SP doesn't do this, and in this way avoids some of the sorts of objects that such proposals tend to receive.

1.2 Neo-Cartesian Skepticism

The section on Stroud (dubbed a neo-Cartesian) contains a great deal more than that on Descartes. It also expands upon some more controversial claims implicit in 1.1.

Stroud interprets Descartes differently. Or, at least, his assessment of the result of the dreaming argument is different. The possibility that we are dreaming robs our beliefs, thinks Stroud, not only of certainty but also of reasonableness. Stroud argues that we cannot rule out that we are dreaming, for any test a positive answer to which would rule it out, could itself be dreamed to have a positive answer. Now, to block SP's reply that she can very well tell the difference between finding a test to be positive and dreaming a test to be positive (because of the phenomenological differences between dreaming and waking consciousness) Stroud must reply that what needs to be ruled out is that what SP takes to be waking consciousness might itself be a dream. We are thus introduced to the 'extraordinary dreaming hypothesis' and the necessity of ruling out of a dream that is like no other of which we are aware.

Extraordinary dreaming is just like the evil demon, and SP admits that it can't be ruled out. It has been designed so that it can't be. Her reaction, however, is to deny that her being unable to rule it out undercuts the reasonableness of her beliefs. Indeed, that it undercuts her claim to knowledge. Here again we see SP's fallibilism. She maintains that we can be said to know things, despite our being unable to rule out certain alternatives. But failing that, she maintains that we can be said to have reasonable beliefs, despite our being so unable. After all, in ordinary life we don't require that people rule out the evil demon hypothesis before they can legitimately claim to know something.

Here Stroud distinguishes between practical and theoretical contexts, suggesting that although the concept of knowledge requires the ruling out of all alternatives, we do not make this requirement for practical purposes. Suppose that we give Stroud this. The question becomes what we should make of 'reasonable belief'. Should we think of reasonable belief as requiring certainty or not? Or might reasonable belief exist at a level below that of absolute certainty? On which side is common sense? Maddy suggests the latter, and I am inclined to agree.

But the question can be put acutely: is it legitimate for SP to employ her empirical methods of belief justification in order to justify those beliefs that she, in turn, uses to justify the adequacy of her empirical methods? Stroud's negative answer to this question seems unreasonable to SP, since it effectively ask us to justify our methods of belief formation without relying on any methods (or, at least, those methods that we take ourselves to have discovered are the best). The question that SP asks of Stroud is why we should suppose that in order for a belief to qualify as reasonable, it must have been generated by a method that can itself be justified without relying on the best available methods.

Questions: How much work is fallibilism (the thought that we can know that P even though we have not ruled out not-P) doing here? And is fallibilism really consonant with our everyday notion of knowledge? If not, then it is Maddy, not Stroud, who is departing from common sense in her defence of it.

But, in any case, can/should we make do with reasonable belief? Fallibilism about reasonable belief might seem more palatable than fallibilism about knowledge. Stroud appears to deny this also but, once more, where does common sense side here?

Finally, is it legitimate to demand that we justify our empirical belief forming methods 'from the outside', i.e. without employing any of them? Supposing that this can't be done, as the sceptic insists, what would it show? That we lack knowledge? That we lack reasonable belief? Or that it is a wrongheaded demand?

Transcendental Arguments Workshop: Vahid

In his paper, "Skepticism and Varieties of Transcendental Argument", Hamid Vahid took on a number of targets. First, he argued that transcendental arguments should be viewed as a species of inference to the best explanation (IBE). He then argued that recent 'modest' interpretations of transcendental arguments, for examples those proposed by Bob Stern, Barry Stroud and Mark Sacks, are epistemically ineffective.

First, non-modest transcendental arguments. Seeing transcendental arguments as a species of IBE means seeing them as exploiting subject matter relative assumptions which would ground the norms of simplicity, explanatory power, etc. But, if this is so, such arguments would beg the question if used to argue against, say, external world scepticism.

Second, modest transcendental arguments and the example of Bob Stern. Vahid argued that Stern's view is that transcendental arguments can be seen as justification conferring since their conclusions can be shown to enhance the coherence of the subject's belief-set. Vahid then poses a dilemma: Either coherence is truth-conducive or it is not. If it is not, then it is hard to see why the sceptic should be concerned. If it is, then the transcendental arguments become redundant, since we simply need to ask whether the proposition that appears as the consequence of the transcendental argument serves to enhance coherence - the transcendental argument itself can be bypassed.

One source of discontent with Vahid's position concerns the alignment of transcendental arguments with IBE. Take, for instance, Strawson's argument that self-ascription requires other-ascription. It is far from clear that the capacity for other ascription explains the capacity for self-ascription at all, let alone being the best explanation for it. So not all transcendental arguments can be plausibly understood in this way.

Second, with regards to the argument against Stern, can't Stern reply that the role of the transcendental argument is to identify which beliefs have a fundamental coherence enhancing role. Whilst coherence is the justificatory norm, transcendental arguments are not thereby made redundant.

Paul Boghossian at the Normativity Workshop

Paul gave a paper entitled 'Rules and Norms'. He began by reminding us of how central rule-following is to our understanding of what it is to be in the business of rational belief. He then went on to detail some differences between what he called rules in the guise of imperatival contents ('If C, do A') and normative propositions such as those that grant permission to do something on occasion of some state of affairs obtaining. He argued we have no clear understanding of which of these logical forms governs the rule-following constitutive of rational belief.

But whatever we decide about that, we accept that rule-following involves essentially the following four factors:

1) (Acceptance) If S is following rule R, then S has somehow accepted R.
2) (Correctness) If S is following rule R by doing A, then S's acceptance of R determines whether S's doing A is correct relative to his acceptance of R.
3) (Explanation) If S is following rule R by doing A, then, S's acceptance of R explains S's doing A.
4) (Rationalisation) If S is following R by doing A, then, S's acceptance of R rationalises S's doing A.

A popluar and natural way to incorporate these factors into a response to the rule-following sceptic is to take the 'Intention View' - the view that when we accept a rule we form an intention to correctly apply it over a potentially infinite amount of cases and that such acceptance both explains (sets causal conditions for right responses) and rationalises our rule-following behaviour.

Paul raised problems for the Intention View and by extension any view that invokes intentional content in its response to rule-following scepticism. He pointed out that problems remain even if we reject Kripke's 'meaning assumption' (the assumption that the contents of thoughts are understood on account of following rules so that if we respond to the sceptic who doubts our grasp of the meaning of 'plus' in terms of words like 'count', she can just bring into doubt our grasp of such words, and so on to infinity). Rather, the central problem for the Intention View remains even if we allow that there can be meaningful mental content that is not understood in terms of rule-following. That problem is that grasping any rule at the personal level relies on grasping a general conditional content governed by a deductive inference rule such as Modus Ponens. But then, how are we to explain grasp of that rule? Thus an infinite regress is generated in the manner of Lewis Carroll's famous argument in 'What the Tortoise said to Achilles'. Except the regress applies not only to the justification of applying a rule, but to rationalising such application as well: in other words, the regress rears up even in cases of unjustified rule-following, such as where we apply a bad rule.

This regress leaves us with the unpalatable rejection of the idea that accepting a rule consists in an intentional state with a conditional content. Paul argued this in fact was Wittgenstein's view: that if rule-following is to be possible at all then some of it must be blind. This puts in danger our intuitive generalist understanding of Rational Belief as governed by epistemic rules.

Paul finished by arguing that a dispositionalist response will not work here because something cannot rationalise or explain itself, so that a disposition must be explained and rationalised by the following of a rule. This leaves us with the idea that not only must we take mental content to be primitive (as we did when we rejected the meaning assumption), but also 'a general conditional content that acts as a reason to believe something'. Paul confessed, although he didn't say why, to not understanding how there could be such a primitive contentful state.

Objections and questions were many during a very stimulating discussion period, but I shall mention the few that my meagre memory cells can adequately recall.

Several people wondered whether an imperatival account of rules and norms wasn't perhaps still a live option.

In response to the infinite regress Paul generated it was suggested that we take the classic line of response to the Lewis Carroll dilemma: we argue that a subject S is operating in accordance with a rule of inference such as Modus Ponens without explicitly accepting it via propositional understanding. To this Paul resonded that Modus Ponens is a general rule that is defined over infinite applications and so one cannot be in accord with it on account of some kind of particularistic grasp of its application.

It was suggested that perhaps the minimal case of following a rule (which involves for PB the four factors outlined above) was too rich. Perhaps some less thick notion of rule-following might suffice that obviated the appeal to a rule of inference that governs a general conditional content? But, it was responded, it is hard to see how something can be a rule without at least (1) and (2) (see above) being satisfied.

Finally, it was suggested that the infinite regress PB had highlighted was not Wittgenstein's problem. The latter, it was suggested, consists in finding an answer to the question 'How do I know what I mean?' rather than the problem of how a rule like Modus Ponens can be a premise in an explanation of how I understand some further rule.

I for one would like to know more about why it is mysterious that 'a general conditional content that acts as a reason to believe something' could be taken as generated by a primitive cognitive-semantic capacity. Perhaps the difficulty here is in thinking there can be such a primitive mental content without there also being a subject who thinks. If I am to be able to think the content Fa then presumably I can think the inferential relations between Fa and Ga, between say, 'being red' and 'being coloured' - that, after all, is what it is to 'think' (it is essential to the content Fa that it be inferentially related to the content Ga - without such relations, there is no such content). But this is just a case of operating with the logical form of the conditional that a grasp of Modus Ponens presupposes; 'if a is red, then a is coloured.' So if we are going to take mental content as primitive I don't see what is so strange about taking 'a general conditional content that acts as a reason to believe something' as primitive, once we accept that content is always for a thinking subject. This brings us close to something like endorsing a 'transcendental' subject without which rule-following would not be possible.