Tuesday, December 3, 2013

Another round on the lottery paradox

I've had a few rounds with Thomas Kroedel over the lottery paradox. His latest response to my latest response is coming soon to Logos & Episteme.  

As he sees it, the lottery paradox (a version of it, at least) arises because these are plausible but inconsistent: 

(1-J) For each ticket, I’m justified in believing that it will lose. 
(2-J) If, for each ticket, I’m justified in believing that it will lose, then I’m justified in believing that all the tickets will lose. 
(3-J) I’m not justified in believing that all the tickets will lose. 

His solution is to reject (2-J), denying that epistemic permissions agglomerate. 

My objections are these: 

O1: Agglomeration isn't an issue if there aren't permissions to begin with. I'd reject (1-J).
O2: There's a good reason to think that we should reject (1-J), which is that we know that we cannot know the relevant propositions.  
O3: Once we get started, so to speak, adding beliefs to the belief set, there's nothing available to him that could explain why we should 'draw the breaks'.  Once we're clear on ex post and ex ante rationality, I don't think that you can both say that it is ex ante rational to believe each of the lottery propositions and then say that it is not ex post rational to believe all of the lottery propositions.  

Anyway, in my last round, I challenged Kroedel to explain why epistemic permissions do not agglomerate. I probably wasn't as clear with my challenge as I should have been, so I'll just note a worry about his latest response.  He offers as examples that show that permissions do not agglomerate cases in which we're permitted to believe conjuncts but not conjunctions.  Here's a general worry about the principle he assumes.

He wants a case with this structure: Jp, Jq, ~J(p&q).  Here's a worry about this. Suppose you were to argue for the conjunction, p&q, from two premises, p and q.  

* Since you believe (a) the premises and (b) believe the premises entail the conclusion, you should (in some sense) believe the argument is sound. (At the very least, you know the conclusion follows and are thought to be permitted to believe the premises, so I'd think you can permissibly believe that the argument is sound.)

* Since the argument is a counterexample (allegedly) to agglomeration of permission, you should not believe the conclusion is true.
* If you do what you should, does that mean you should be like this: I believe the argument is sound, but I don't believe the argument's conclusion?

To my mind, that's a crazy way to be.  I'm pretty sure that this is similar (if not identical) to Adler's remarks concerning the lottery and the conjunction rule from Belief's Own Ethics.  

Another problem that simply doesn't arise if we simply reject (1-J).  Since I don't think there's yet been an adequate response to O3, I still don't think that the worries I've raised can be put to rest.  

Of course, there's this persisting problem.  If I ask you whether the ticket won or lost, you might say, 'I don't know'. Kroedel and I agree that you'd speak the truth. If you then added 'Of course, it will lose', I think you're being irrational.  This is a point on which we disagree, but I haven't yet seen any plausible story about how that could be a rational state of mind to be in.

Saturday, August 17, 2013

E=K and Infallibilism

Jessica Brown (forthcoming) identifies a potential problem for Williamson’s (2000) approach to evidence and knowledge. Williamson identifies your evidence with the propositions that you know:
E=K: S’s evidence includes p iff S knows p.
He also accepts this account of evidential probability:
EP: The evidential probability of a proposition p for you is the conditional probability of p on your total evidence.  
Taken together, these two claims commit Williamson to a form of infallibilism:
Infallibilism: If S knows p, the evidential probability of p on S’s evidence is 1.

Why is Williamson forced to choose between inductive skepticism and the possibility of p being evidence for p? Consider a standard approach to evidential support, one that Williamson accepts:
EV: e is part of S’s evidence for h iff S’s evidence includes e and P(h/e) > P(h).
In the case of inductive inference, prior to believing p the evidential probability of p is less than 1. After adding p to your evidence, it's probability raises to 1. So, in the case of inductive inference that results in knowledge, p is evidence for p because (i) p is part of your evidence (by E=K and the anti-skeptical assumption) and (ii) P(p/p) > P(p).

How serious a problem is this for Williamson? Brown thinks it's quite serious. I disagree, but that's for another post.  

I have written more about this in a paper that will probably never see the light of day (unless someone can think of a good journal for it):


Monday, August 12, 2013

Fetch for one

There's a video of a dog playing fetch with himself on Gawker (here). That's old hat. Agnes invented that game ages ago:

Wednesday, July 3, 2013

Converting to knowledge-first epistemology

I've been critical of the knowledge-first approach to issues in epistemology, but around the beginning of this year I started to reconsider. There were three reasons for this. The first is that the intuitions that I thought caused the most trouble for this approach had to do with a specific kind of Gettier case. Now my intuitions about cases of environmental luck have started to shift. I'll post some about this soon when I get a decent draft of a paper on robust virtue epistemology. The second is related to this and that's that I've started to think that the link between knowledge and ability is a bit tighter than I had initially thought. Third, I think that the truth-first approach I favored really cannot explained why epistemic assessment has the inward looking focus that it does. That's the topic of this paper, which should be out in The Proceedings of the Aristotelian Society in the not so distant future. In the meantime: The Russellian Retreat

Tuesday, July 2, 2013

Justification, Access, and Moore's Paradox

Good news! Received word last night that Episteme has accepted my paper, "A Note Concerning Justification and Access". In this paper, I evaluate Declan Smithies' Moorean arguments for access internalism and argue that we shouldn't appeal to the access principles he defends to explain why various combinations of attitudes are not rationally co-tenable. You can find a draft of the paper here. If interested, you can find a new page with links to papers here.

Friday, May 31, 2013

Reasons vs. Causes (again)

Thinking about this last night before bed, it seems to me that there's a simple response to those who would argue for statism and against factualism on the grounds that reasons are causes.   

P1. Causes have to have to be fully determinate, fully specific entities. 
P2. Reasons cannot be fully determinate, fully specific entities.
C. Reasons cannot be causes.

The motivation for P1 might initially seem a bit obscure since some authors (e.g., Mellor) have argued that facts can be causes.  Remember, however, that we're looking at an argument that's an argument against those who have already decided that facts cannot be causes, presumably on the grounds that they are not concrete.  

As for the motivation for P2, the idea is that the rational role of reasons requires an entity that's graspable.  Facts, propositions, and states by virtue of the fact that their natures seem to be determined entirely and exhaustively by the way they are canonically picked out linguistically.  Whereas causes have to satisfy Steward's 'secret life requirement', reasons cannot have secret lives upon pain of failing to be a graspable.

This is all back of the envelope stuff, but I think it's a start.  

Knowledge, Reasons, and Causalism (Draft)

I've finished another draft of a paper on knowledge, reasons, and causes. (I've changed the title because I've discovered that Harman has a paper with the title I thought I'd use.)

Knowledge, Reasons, and Causalism

In this paper, I explain why I don't think the considerations that support the causalist view of rationalizing explanations lends any support to a statist view of epistemic reasons. Almost all of the extant views of the basing relation treat the basis for one's beliefs as states of mind. I don't argue that the view is without any motivation, but I do argue that the only motivation I know of for the view is no motivation at all.  I then point to some surprising consequences of the factualist view of reasons I prefer. (I was surprised by some of these consequences!  Not all the surprises were happy surprises.)

Sunday, May 26, 2013

Swinburne on designators and dualism

Another stab at Swinburne's argument for dualism. I don't think the introduction of the notion of an informative rigid designator is at all helpful. Here's why.

Swinburne has a new argument for property dualism.  If successful, the argument would seem to show that many familiar forms of physicalism are false.  I don’t think the argument is successful.

The Anti-Physicalist Argument
To understand his argument, we need to understand some of Swinburne’s machinery. Swinburne introduces the notion of an informative rigid designator (IRD) to help introduce identity criteria for properties.  On his view of property identity, properties are individuated by the IRDs that pick them out (2013: 24). Whereas a rigid designator will designate the same thing in every possible world (where it designates anything at all), an IRD is such that one who grasps its meaning will know what something has to be to be designated by that designator:
For a rigid designator … to be an informative rigid designator it must be the case that anyone who knows what the word means … knows a certain set of conditions necessary and sufficient (in any possible world) for a thing to be that thing (whether or not he can state those conditions in words) (2013: 12).
To use a well worn example, think about ‘lightning’ and ‘electrical discharge’.  If it is true that lightning is just electrical discharge, we might expect that it is necessarily the case that lightning is electrical discharge. The expressions seem to rigidly designate the same thing and so there is no possible world in which it’s false that lightning is electrical discharge.  While someone might grasp fully the meaning of lightning without being in a position to say anything at all about electrical discharge (e.g., speakers who were ignorant of the relevant scientific discoveries and had no concept of electricity), one might think that someone who knows what ‘electrical discharge’ means knows what it would take for something to be an electrical discharge and see that lightning fits the bill.
Let me note two interesting (apparent) consequences of this approach.  The first is that an identity sentence involving a pair of IRDs is not simply necessarily true if true but logically necessary if true.  The denial of the sentence would entail a contradiction (2013: 19). The second is that any identity sentence involving a pair of IRDs will be knowable apriori if true (2013: 24).  Thus, if it is not apriori that a pair of IRDs designate the same thing, it is supposed to be necessarily false that they designate the same thing. The sentence that asserts that they designate the same thing would entail a contradiction.
With this much in place, there is a quick argument against the identification of any mental property with any physical property.  To show that mental properties are distinct from any and all physical properties, Swinburne argues that it is not apriori that the IRDs that designate the physical properties or the mental properties designate the mental properties or the physical properties:
Since the informative designators of any physical properties are not logically equivalent to those of any mental properties (since there are different criteria for applying the designators), no mental property is identical to a physical property. The criteria for being in pain are not the same as the criteria for having some brain property … The criteria for being in pain are how the subject feels, and the criteria for brain and behavioural events are what anyone could perceive (2013: 70).  
The argument is quick, but is it effective?  I fear that there are two notions of IRD at play in his discussion. Once we’re clear on what they are, we shall see that his argument is no more persuasive than familiar unpersuasive conceivability arguments against physicalism that don’t make use of his notion of an IRD.  

Evaluating the Argument
In the passage quoted earlier, Swinburne suggested that a speaker who knows the meaning of an IRD will associate a set of necessary and sufficient conditions that something must satisfy to be designated by the IRD in this or in any possible world:
IRD1: If a speaker knows the meaning of ‘a’ where ‘a’ is an IRD that designates a, there is a condition C that is necessary and sufficient for being a in any possible world that the speaker associates with being a.
We can introduce another notion of an IRD, one that’s a bit more demanding:
IRD2: If a speaker knows the meaning of ‘a’ where ‘a’ is an IRD that designates a, for any condition C, a knows whether C is necessary or sufficient for being a in any possible world.
To see that these notions of an IRD are distinct, suppose there are two conditions, C1 and C2, that a meets.  Suppose that satisfying C1 is necessary and sufficient for being a in every possible world. Suppose that satisfying C2 is also necessary and sufficient for being a in every possible world.  A speaker might satisfy the conditions set out by IRD1 for knowing the meaning of ‘a’ by associating C1 with ‘a’ without associating C2 with ‘a’. (Similarly, the speaker might associate C2 with ‘a’ without associating C1 with ‘a’ and still know the meaning of ‘a’.)  While it would be apriori that a satisfies C1, it might not be apriori that a satisfies C2.  According to IRD1, the speaker would know the meaning of ‘a’. According to IRD2, the speaker would not.
It’s quite possible that the notion of knowledge at meaning is itself one that’s subject to various ambiguities. I would like to avoid tricky disagreements as to whether IRD1 or IRD2 captures the true meaning of ‘knowledge of meaning’.  There might be a weak sense in which knowing the meaning of ‘a’ requires satisfying the conditions set out in IRD1. There might be a strong sense in which knowing the meaning requires satisfying the more stringent conditions set out in IRD2.  To fix meanings, let’s refer to these two kinds of knowledge as knowing the meaning in the weak and the strong sense respectively.
A concrete example might help to illustrate the distinction and help to explain why there’s a potential worry here with Swinburne’s argument.  On Swinburne’s view of persons, each person has a thisness that ‘makes’ a person the person he or she is (2013: 151).  Swinburne holds that we know the meaning of various IRDs that we can use to designate ourselves:
Now what sort of designator is ‘I,’ or ‘Richard Swinburne,’ as used by me? These seem to be informative designators. If I know how to use these words, I cannot be mistaken about whether or not they apply to a certain person--given that I am favorably positioned (e.g., his body is my body) with faculties in working order, and not subject to illusion. And when I am considering applying these words to a person in virtue of his having some conscious event, these conditions will be maximally satisfied and no mistake is possible. I am, in Shoemaker’s … phrase, ‘immune to error through misidentification’ (2012: 119).
Suppose that that is so.  This is consistent with the further idea, defended by Kripke (1980), that certain facts about our origins are necessary for us.  If Kripke’s arguments are to be believed (and their conclusions are surely consistent with everything Swinburne has argued for in arguing for dualism), we could not have had our origins in any combination of egg and sperm other than the combination actually was involved in our generation.  We can exploit this to introduce further IRDs that designate us that identify conditions we do not associate with ourselves in using ‘I’ to refer to ourselves.  Indeed, if successful use of ‘I’ turned on knowing our origins, we would lose this immunity to error through misidentification that Swinburne associates with our use of ‘I’.  Thus, there is a perfectly good sense in which George Bush would have grasped the meaning of ‘George Bush’ even if he was in no position to say whether George Bush could have been born into the Kennedy family.  As the example hopefully illustrates, knowing the meaning in this weak sense is both consistent with knowledge of meaning in a stronger sense but is not for that knowledge of meaning in a totally uninteresting sense.
Do we know the meaning of, say, ‘pain’ and ‘firing of C-fibers’?  Swinburne is surely right that there is a perfectly good sense in which we know the meaning of ‘pain’ if we have had painful experiences and can use this term to describe them in the normal way.  We’ve also all read enough philosophy of mind to know something about the firing of C-fibers.  The crucial question is what we know in knowing the meaning.  Do we know the meaning of ‘pain’ in the weak sense or the strong sense?  Suppose that we know the meaning of ‘pain’ and ‘firing of C-fibers’ only in the weak sense. If so, it is consistent with what we know that the criteria of application of ‘pain’ and ‘firing of C-fibers’ will coincide in this and in every possible world.  Thus, knowing that ‘pain’ properly applies to something will not put us in any position to know that ‘firing of C-fibers’ does not also apply.  Thus, while it might not be apriori that anything that is a pain is a firing of C-fibers, it could nevertheless be necessarily true that it is.
The success of Swinburne’s argument, then, depends upon something quite strong.  For any of us to know apriori that a pain is not just the firing of C-fibers, we would have to know the meaning of ‘pain’ and ‘firing of C-fibers’ in the strong sense, not merely the weak sense.  I doubt that any of us have this knowledge. I certainly doubt that having this knowledge is what we have when we know what pain is by learning the criteria for its correct application (i.e., how certain things feel).  In knowing that I feel a certain way, I can know that it is appropriate to say that I’m in pain but I’m not thereby well positioned to assert the stronger claim that what I’m feeling does not require the firing of C-fibers.  What a physicalist should say is that we do not know the meaning of ‘pain’ or ‘firing of C-fibers’ in the strong sense. Once we have distinguished between different senses of knowledge of meaning, there is no reason to be ashamed to say that we don’t have this knowledge now and may never have this knowledge.
The introduction of the distinction between two senses of IRD and two kinds of knowledge of meaning can be independently motivated. Interestingly, it seems to save Swinburne from trouble.  Consider the debates between the Cartesian interactionists and the occasionalists.  Both parties to this debate presumably knew what ‘pain’ meant in what I’ve called the weak sense.  They certainly grasped the criteria for the correct application of the term.  What they disagreed about was whether pain was the sort of thing that could be caused or cause a physical thing without divine intervention.  The Cartesians would impress nobody with this argument:
Our debate is about ‘pain’. The criteria for the correct application of this term has to do with how the subject feels. Knowing the meaning of ‘pain’ involves grasping facts about how things feel to the subject. We could introduce an IRD, ‘o-pain’, that designates a state that must meet two conditions: having the feel that is characteristic of ‘pain’ and being such that it cannot causally interact with anything physical without divine assistance. Since it is not apriori that pain is o-pain and both ‘pain’ and ‘o-pain’ are IRDs, it is not true that pain is o-pain. Thus, your occassionalist view of pain is mistaken.
The occassionalists could run a parallel argument for ‘c-pain’, an IRD that designates a state that has the felt quality of pain and is such that it can causally interact with something physical without divine assistance.  Neither the Cartesian nor the occassionalist would be impressed by the claim that their opponent knew what ‘pain’ meant in the weak and the strong sense.  What both parties should say that what they know about the meaning of ‘pain’ is sufficient for them to apply it correctly and concede that their knowledge of the meaning of ‘pain’ leaves unsettled questions about whether pain is c-pain or o-pain.
Suppose Swinburne were to reject my suggestion that there are two notions of IRD at play and two notions of knowing the meaning of an IRD.  Suppose he insisted that in knowing the criteria that would enable one to correctly apply an IRD, one is thereby in a position to know apriori the conditions it meets.  If he insisted on this, the debate between the Cartesians and occasionalists should worry him.  

It does seem that it is apriori that pain is either c-pain or o-pain, but not apriori that it is one or the other.  Notice that if all necessary truths are apriori and I’m right about the lack of apriori entailments between ‘pain’, ‘o-pain’, and ‘c-pain’, we can generate a contradiction from Swinburne’s system. These three claims, after all, would constitute an inconsistent triad if all necessary truths are apriori:
(i) Necessarily, pain is either c-pain or o-pain.
(ii) Although I grasp the meaning of ‘pain’ and ‘c-pain’, it is not apriori that pain is c-pain.
(iii) Although I grasp the meaning of ‘pain’ and ‘o-pain’, it is not apriori that pain is o-pain.
To see that these three constitute an inconsistent triad in Swinburne’s system, suppose (ii). Since both ‘pain’ and ‘c-pain’ are IRDs, this entails that pain is not c-pain. This entails by (i) that it is o-pain. This, however, is incompatible with (iii) because (iii) entails that pain is distinct from o-pain.
Such contradictions are easily avoidable if tinker with Swinburne’s system. We can say that there are necessary truths about the relationships between things designated by IRDs that are not apriori. If we said this, we could say that (i)-(iii) are true, but then the fact that it is not apriori true that pain is C-fibers firing would be consistent with the fact that pain just is the firing of C-fibers.  Alternatively, we could hold onto the idea that the all necessary truths are apriori and assert that (ii) or (iii) must be mistaken if we concede that what we know about the meaning of ‘pain’, ‘o-pain’, and ‘c-pain’ is sufficient to understand what these terms mean without our understanding putting us in a position to determine which of these claims is true.  If we say this, we shouldn’t be moved by Swinburne’s anti-physicalist argument because we’d have to concede that we know what ‘pain’ is only in the weak sense that doesn’t put us in a position to know apriori whether or not it is the firing of C-fibers.  There are lots of ways of rearranging the deck chairs so as to avoid an apparent contradiction, but on none of these ways of doing so should we think that we’re in any good position to say what pain isn’t simply by virtue of having a painfully good grasp of what it is.   

Everyone knows that there are potential problems with a certain style of apriori argument against physicalism.  The apparent cases of aposteriori necessities should worry us that the apparent power to conceive of a real distinction between two things is really just due to our ignorance of the complex ways that these things are.  The introduction of the distinction between rigid designators and informative rigid designators was supposed to help us overcome this apparent difficulty, but it hasn’t done so at all.  IRDs understood as IRD1s will not underwrite the sort of anti-physicalist argument that Swinburne is after because  it is possible to know the meaning of ‘pain’ in the relevant sense without knowing that nothing can be a pain unless it is a firing of C-fibers.  IRDs understood as IRD2s will not underwrite the anti-physicalist argument that Swinburne is after because it is not at all clear that we know the meaning of ‘pain’ or ‘firing of C-fibers’ in the strong sense.  Our ability to correctly and knowingly apply predicates like ‘pain’, ‘belief’, or ‘desire’ does turn on having criteria for correct application (and so knowledge of their meanings in a weak sense), but not on having the kind of knowledge that’s required for knowledge of meaning in the strong sense.  I submit that the introduction of the notion of an IRD has not helped to advance the case against physicalism at all.
In terms of Swinburne’s overall case against physicalism, the failure of his argument for property dualism is significant.  Physicalists have long held that it is possible to maintain a perfectly respectable sort of physicalism without any commitment to type-identity theory (i.e., the theory that all mental properties are identical to some physical properties).  The physicalists thought that even if type-identity theory might be false, it still might be true that all substances and events were physical substances and physical events.  As Swinburne conceives of the terrain, however, the failure of type-physicalism would seem to immediately lead to the failure of physicalism.  Events are, on his view, nothing more than property instantiations or changes (2013: 1).  If events are conceived of in this way, all mental events must be distinct from physical events once it has been shown that mental and physical properties are distinct.  Substances are, on his view, mental if they have any mental properties essentially (2013: 141).    

Kripke, S. 1980. Naming and Necessity.  Blackwell.
Swinburne, R. 2012. How to Determine which is the True Theory of Personal Identity. In G. Gasser and M. Stefan (ed.), Personal Identity: Simple or Complex. Cambridge University Press.

____. 2013.  Mind, Brain, and Free Will. Oxford University Press.

New Blog: Philosophers King's

Charles Cote-Bouchard and Antonio Delussu have started a new blog for KCL philosophy graduate students: Philosopher's Kings.

They have just two posts up, but I think it should be good once they get going.  (Get going!)

Saturday, May 25, 2013

Some thoughts on epistemological disjunctivism

Having read Duncan's work on epistemological disjunctivism a few times, I've written something up that explains why I don't think his approach can overcome the basis problem.  I'm in the midst of writing something up on a similar problem that confronts McDowell's epistemological argument for metaphysical disjunctivism. In the meantime, a paper on epistemological disjunctivism and the basis problem.

A beast with two backs

I've had this problem lately finding a place to sit.  In the old flat, I had more seats than I needed but no desk. Now I have a wonderful desk and can't seem to find a place to sit.  I've fixed this. In so doing, I've conclusively refuted Peter van Inwagen.

One chair with a broken back. One chair in need of a seat.


Where there had been two objects, we now have a chair with a back and a seat.

This is the most comfortable chair I've ever had.  I'm pleased with myself because I've just done what Peter van Inwagen takes to be impossible--make a comfortable chair out of chairs!

I think Martino Gamper would be proud.

New Blog! Fsopho

Hi all,

There's a new blog that I wanted to draw your attention to:


Fsopho has lots of good posts on issues near and dear to me.

Friday, May 24, 2013

The K took my baby away

On Monday I gave a talk at the Aristotelian society that had to do with the norms of belief (listen here). My view has long been that knowledge isn't the norm of belief (or assertion, practical reason, etc.), but I've started to think that it might be.  The view I defended in Justification and the Truth-Connection was a kind of truth-first view. Starting from the idea that the fundamental epistemic norm is a truth norm, we could say what needed to be said about epistemic normativity by identifying various derivative norms that derive their authority from the truth norm. At the end, we'd say what needs to be said about epistemic normativity without ever having to say anything at all about knowledge.

I've started to change my mind about this. I think there were two reasons for this and I don't think they're unrelated.  First, I had thought that having something as part of your evidence didn't require knowing that the thing was true.  I'm no longer convinced that this is so.  I'm not entirely certain that your evidence can't include p unless you know p, but I'm just not totally persuaded by my old Gettier-based arguments against E=K.  Since I still think that there are norms that govern belief that have to do with whether your beliefs can provide reasons, I've started to shift towards the idea that there is indeed a knowledge norm that governs belief.  

The second reason for the change is the subject of the Aristotelian society talk.  As I was thinking about what to talk about, I was struck by something I suppose I hadn't really thought about before.  Practical assessment seems to be largely outward looking. What I mean is that it seems that the deontic standing of your actions turns largely (if not exclusively) upon whether those actions were fitting in the circumstances.  (There are tricky cases that Steve Sverdlik describes in his excellent book Motive and Rightness, but we can set these aside for now because the point that I want to make doesn't turn on whether his arguments are successful or not.)  Thus, I think Judith Thomson is right when she says that it's an odd idea that we need to look into the agent's head to determine whether some prospective course of action would be permissible. (I'm not convinced that she's drawn the right lesson from this, but that's for another time.)  At any rate, it's striking that everyone seems to think that epistemic assessment is largely inward looking.  We care about whether the agent's reasons are good reasons and how the agent reasoned. The deontic standing of a belief seems to turn on this, not (just) whether it fits the facts.

So, there's an interesting asymmetry between practical and theoretical assessment that I've struggled to understand. Why does one seem to be so excessively concerned with the relationship between guiding and explanatory reasons when the other doesn't seem to be terribly concerned with this at all? 

It's because of this asymmetry that I started to think that the truth-first approach couldn't ultimately explain why the deontic status of one's beliefs turns on the reasons for which one believes and their relation to good reasons to believe.  It seems that any such explanation would have to start from the idea that the content of the fundamental epistemic norm has to do with truth and then try to derive additional norms by appeal to some ideas about what guiding reasons require or about how we're supposed to follow norms. Those claims look pretty good until you realize that these are supposed to be claims about how guiding reasons as such are supposed to work. Those claims threaten to undermine the asymmetry between practical and epistemic assessment.

A better approach, I thought, was a knowledge-first approach.  To the extent that knowing requires believing for good reasons, the knowledge norm could explain why epistemic assessment is inward looking without threatening to undermine the asymmetry by introducing some ancillary assumptions about how norms work or what guiding reasons require. So, I've shifted. You can listen above or read the paper here. I suppose the good news is that most of what I argue for in the book can stay pretty much the same. What changes is that the distinction between justification and knowledge that I thought could be maintained now seems to be undermined.  Oh well. 

Thursday, May 23, 2013

What are we supposed to learn from deviant causal chains?

I don't have the patience to work through most papers on causal deviance. This isn't an attempt to solve any of the problems having to do with deviant causal chains, only an attempt to say something about the significance of deviant causal chains.

The problem that deviant causal chains pose for a Davidsonian approach to action is one of the motivations for Steward's agency incompatibilism.  As she sees it, actions just aren't the sorts of things that have causal antecedents. It's a mistake to think of actions as events that have been brought about in the right way by the agent's states of mind or mental events.

Here's a worry.  It seems that you'll run into problems having to do with deviant causal chains in trying to give a causal account of believing for a reason, but it doesn't seem like beliefs are the sorts of things that shouldn't have causal antecedents. Agency incompatibilism has its merits, but I don't think that determinism is a threat to belief.

On Marcus' account, it looks like acting for a reason and believing for a reason are given the same treatment--to believe for a reason or act for a reason is to represent the belief or action as to be believed or done in light of something else. On one way of reading this, deviant chains cannot arise because that requires a causal chain between two distinct existences. A virtue of Marcus' approach is that it does away with the distinct existences. Believing for a reason doesn't involve a causal relation between a reason or some reasons and a belief. It involves representing the fact/proposition believed as to be believed in light of something else and that representation isn't a relation between two distinct existences.

Here are two worries. First, Marcus' account cannot be applied to the case of non-inferential belief.  Thus, the account seems insufficiently general. Whatever story we tell about believing for a reason in the non-inferential case, we'll need to rule out deviancy in this case without using his trick.  You might think that whatever we do to understand the relationship between the reason for which you believe and your belief in, say, the case of perceptual knowledge, we'll be able to tell a similar story in the case of inferential knowledge.  Second, there are deviant causal chains that don't have to do with belief, action, or reasons at all.  There are cases of that show that the connection between, say, the fragile glass' shattering and the striking of the glass isn't right for the glass to have shattered because of its fragility.  (I think I owe this point to Hyman).

Wednesday, May 22, 2013

Swinburne's designators

I'm reading Swinburne's Mind, Brain, and Free Will (OUP 2013) and trying to get a grip on his latest arguments against physicalism.  In the discussion, the notion of an informative rigid designator seems to play an important role. He says that rigid designators are supposed to designate the same thing in every possible world. (Let's bracket worries about what rigid designators designate in worlds where there's nothing that exists that could be designated by the expression.) Some of these (e.g., 'Richard Swinburne', 'This', 'Here') can be used by speakers who know the meaning of these expressions without knowing a set of necessary and sufficient conditions that the thing designated must meet to be the thing designated by the designator.  With informative rigid designators, however, 'it must be the case that anyone who knows what the word means ... knows a certain set of conditions necessary in sufficient (in any possible world) for a thing to be that thing' (12).  

The notion of an informative rigid designator does a lot of work in Swinburne's book. He says that properties are individuated by the informative designators that pick them out. As a consequence of this, it is a purely apriori matter whether one property is the same as another.  (As you might expect, the fact that there are no true apriori claims about the relationship between mental and physical properties is taken to be a reason to reject type identity claims. (And since events are defined in terms of properties and their instantiations, we get a quick argument against certain familiar forms of token identity theories. Editorial note: the quickness of that argument is a reason to worry about his approach to these issues, not a reason to worry about the status of token physicalism. The fact that you can't take seriously the idea that token physicalism might be true even if type physicalism is false just means that you aren't trying very hard to get the notion of an event right.)) 

So, what are these bloody informative rigid designators?  I fear that Swinburne's guidance isn't quite as helpful as it should be.  Here are two ways of thinking about IRDs that won't suit Swinburne's purposes but does seem to satisfy his initial gloss on the notion: 

IRD1: If I know the meaning of 'this' when I say 'This is F', I know that for any x, x = the semantic value of 'this' iff 'this is x' is true.  Since I'd know a condition that's necessary and sufficient for determining whether any x meets this (trivially because I know that they'd have to be identical to the semantic value of 'this'), all RDs are IRDs.
IRD2: If I introduce a descriptive name such as 'Julius' by means of a description, such as 'The person who actually invented the zip', I'd know a condition that's necessary and sufficient for any x to be Julius. For any x to be Julius, x has to be identical to the actual inventor of the zip.  This allows us to maintain some distinction between RDs and IRDs if we say that some such descriptive knowledge is necessary.  

It's obvious that the first proposal isn't at all in the spirit of Swinburne's suggestion and pretty clear that neither way of introducing IRDs would suit his purposes.  That's because it's supposed to be apriori whether two IRDs pick out the same thing and the case of descriptive names shows that it's quite possible to introduce a kind of RD where it's true that one thing is picked out in two ways without it being apriori that this is so.

It seems what Swinburne needs, then, is something stronger: 
IRD3: If I know the meaning of an IRD, I know _all_ the conditions necessary and sufficient for its proper application, not just a condition that's necessary and a condition that's sufficient.

He says, "Since the informative designators of any physical properties are not logically equivalent to those of any mental properties (since there are different criteria for applying the designators), no mental property is identical to a physical property" (69).

He says, 'The criteria for being in pain are how the subject feels, and the criteria for brain and behavioral events are what anyone could perceive. And the same applies to any other pure or impure mental property' (70).  

Notice that his argument that's supposed to establish that mental properties are distinct from physical properties because the IRDs for mental and physical properties (and, by extension, events) is that the criteria for _applying_ IRDs for the physical properties make reference to things that are publicly observable and the IRDs for the mental properties make reference to how things feel.  Now, if the subject who knew the meanings for the relevant IRDs knew all the necessary and sufficient conditions for their applicability, it seems that they'd know _more_ than just the criteria of application. That just requires knowing a sufficient condition for the IRD's application.  What the argument needs to establish non-identity, however, is both knowledge that, say, pain requires just that something feels a certain way AND ALSO that there's no physical property that's necessarily instantiated by virtue of things feeling a certain way.  

So, there seems to be a slip. It seems that Swinburne's argument against physicalism starts with the plausible idea that we know the meaning of the IRDs we use to talk about the mental where the plausibility of that idea turns on thinking of IRDs on the IRD2 model. That won't support the argument he's offering, however, because that requires thinking of IRDs on the IRD3 model. If we're clear that that's the relevant notion we have in mind, then there's just no reason at all to think that ordinary speakers know the meanings of the IRDs they use to refer to the properties that they do.  

Maybe the quick way of putting the point is this. It's plausible that we know conditions _sufficient_ for the application of 'pain' or 'C-fiber firing', but only because this knowledge doesn't require knowledge of the total set of necessary and sufficient conditions that have to obtain for the term to be properly applied.  What we need to establish non-identity claims is knowledge that meeting the condition for successfully attributing 'pain' does not involve, inter alia, meeting the condition for successful of application of, say, 'C-fiber firing'.  Swinburne's argument goes nowhere precisely because it's supposed to provide us with this knowledge. Instead, it subtly assumes that we have this knowledge in an attempt to establish a non-identity claim.  Being slippery in how we talk about IRDs just obscures this fact.

Saturday, May 11, 2013

Narrow contents and justification: security vs. sufficiency

I've been thinking a bit about the varieties of mentalist views about propositional justification. There's a view that's attractive to those of us who have internalist instincts (not me!) according to which the only states of mind that contribute to justification are those that are phenomenally individuated.  This includes some intentional states, but not states with wide content.  (I'm following Smithies' discussion of the view here.)

Here's a rough worry about the view.  It doesn't deny that the belief that, say, this glass contains water can be justified on the basis of your experiences. It insists that its justification is provided by phenomenally individuated states which are supposed to be common to subjects on Earth and Twin Earth. Here's the objection to this proposal:

P1. To have sufficient justification to believe propositions about the external world, these propositions have to be more likely than their negations on the evidence one has for these propositions.
P2. If these propositions are the contents of beliefs with wide contents, phenomenal mentalism implies that these propositions will not be more likely than their negations on the evidence one has for these propositions.
C. Phenomenal mentalism implies that one cannot have sufficient justification to hold beliefs with wide contents.

The first premise says that you cannot justifiably believe p if you need evidence to believe p and the evidential probability of p is not greater than the evidential probability of ¬p.  I take this premise to be eminently plausible. To deny it, it would have to be possible to have beliefs about the external world that were not more likely to be true than not on one’s evidence.  It’s not at all plausible to deny that.

Let 'w' be the proposition that the glass contains water and 't' be the proposition that it contains t-water. Now, we might suppose that our subject only grasps the concepts to entertain w but the grounds she has for believing w are the same grounds as the grounds her counterpart has for believing t.

The second premise says that the evidence you have to believe this glass contains water (‘w’) is the same evidence you have to believe that this glass contains twater (‘t’). The evidential probability of w cannot be greater than .5 because it is equal to the evidential probability of t and w and t are incompatible. (Indeed, since the evidential probability of the disjunction of t and w isn't 1, it will be less than .5.) Even if these exhausted the possibilities (which they don’t), w wouldn’t be more likely than not given your evidence.

Might the phenomenal mentalist try to avoid this worry by arguing that probabilities are only defined relative to propositions that you grasp? I suppose that's a possibility, but it seems like an odd one. Without going this route, it looks like the security of the grounds for our beliefs about the external world undermines the thought that these grounds are sufficient.