Philosophy Blog

Life at the thought-face.

Tuesday, July 29, 2003

That last post of mine clearly shows that I am spending too much time on admin, and even beginning to care about these things. Time to get back to some philosophy. I have been reading Richard Foley’s book Intellectual Trust in Oneself and Others, which is very stimulating, though so far rather lacking in detailed argument. Anyway, there is one thing I wanted to take issue with. Foley appears to assume that the following are equivalent (e.g. 28, 39):

1. S’s opinion that p is ‘in accord’ with S’s other reflective first and second order opinions.
2. S’s opinion that p will survive S’s most thorough and searching reflection.

Foley’s view is thus that if someone accepts p because q, but has a second-order opinion about what counts as good evidence, which opinion entails that q is not a good reason for p, then that person will be able to discover and resolve the tension purely by solitary reflection. Foley accepts that such reflection may sometimes be unwise, because it will take time and energy which could be put to better use, but it is always possible for the individual to discover and resolve tensions in her opinions.

What this view neglects (caveat: I am only on p.48, so Foley may discuss this later), is that reflective self-criticism is a skill, and a difficult one at that. Someone unskilled in spotting incompatibilities in their opinions may well reflect away as hard as Foley cares to require and never spot an inconsistency. And yet when it is pointed out to him by a suitably skilled teacher, she accepts it and revises her opinions accordingly. Some of the evidence on the Wason selection tasks suggests exactly this phenomenon: subjects fail the task and are told that they have failed, but cannot work out on their own what their mistake was. Foley talks a great deal about ‘self-criticism’, but this could mean either criticism achievable by oneself alone, or criticism of someone by her own standards. The two come apart when X’s ability to criticize herself (by her own standards) falls short of Y’s ability to criticize X by X’s standards.

If Foley is really overlooking this point, then it would seem that he is assuming some sort of transparency of the semantic relations between our beliefs/opinions. For to be a good (self-)critic one has to be able to spot how different opinions are related, to spot what is incompatible with what, and if everyone is potentially as good as their own best critic, then there can be no incompatibility between someone’s opinions which is hidden to her most careful reflection. Maybe Foley is happy with this assumption, but with the current state of the philosophy of mind, I think it needs a fairly substantial defence.

.: posted by Tom Stoneham 11:32 AM

Rather long rambling post, initially about examination rules but ending up commenting on the point of getting a degree …

I got into a surprisingly heated (on my part) argument about examinations a few weeks ago. The question was this: suppose a student turns up very late or even misses an examination completely, should she be given a chance to take the examination?

The view I was opposing says:
Yes if the following conditions are met:
1. The lateness/absence was a genuine mistake
2. The examiner believes there has been no collusion

My view is that we should answer affirmatively only if there are genuine medical or compassionate circumstances explaining the lateness/absence, and that these have been confirmed by the an appropriate independent source, for example, there is a medical certificate. The disagreement, then, was over cases where the student forgets, gets the time wrong, or sets her alarm wrong, or some similar organizational error. I wanted to argue that these are culpable mistakes and the student should suffer the consequences. My colleagues were of the mind that we should be more sympathetic and do what we can to enable the student to prove her intellectual merit.

Less heated reflection suggests that the real disagreement here is over the very nature of assessment by examination. If one wants to assess someone’s ability to do X, then one has to choose whether to have a form of assessment which minimizes the gap between competence and performance (to adapt a useful distinction of Chomsky’s), or to simply test performance. For example, the person who wins the gold medal in the 100m at the Olympics is not necessarily ‘the fastest man in the world’, rather it is the person who ran the fastest at a particular time in particular conditions. To win the gold medal, one has not only to be very fast, but also to to perform on the day. In contrast, holding the world record is completely different: you can keep on trying whenever it suits you. Of course, performance is a function of competence, so testing performance also tests competence, but it tests other things as well. In particular it tests whether someone can exercise her ability to run fast, or to argue philosophy, on demand and under pressure.

Now here at York we use two methods of assessment of our students: closed examinations and essays (aka term papers). Assessment by essay aims to reduce the performance-competence gap, because the students have months and months to write their essays, they can show drafts to their tutors, discuss their ideas with anyone who will listen, check and recheck what they have written etc. etc. Of course, they have to get in it by a deadline, but even deadlines can be extended for sympathetic reasons. In contrast, assessment by closed examination explicitly tests performance. The students must do their best in very specific and rather artificial circumstances. To prepare for a closed examination one must not only make sure one knows lots of philosophy, but also that one can recall and express it under pressure, and in a very short time. There are special skills here, because not only must material be recalled, but it has to be marshalled and pared down in order to answer a precise and unexpected question, all in a short period of time. A well-prepared student may have the competence to write a 10,000 word dissertation on, say, free will, but in an examination she is being asked to select the 800 or 1000 words which best answer the question which was set.

So now we have a way to defend my hard-line answer to the original question. If a student is late for a closed examination because his alarm did not go off, or he got the date and time wrong, then he should not be given extra time or other opportunities to redeem himself. For to do so would be to miss the point that the examination is testing the ability to answer certain questions, at a specified time and place, much like the Olympics is testing the ability to run fast on a certain track at a specified time and place.

But this argument is too hasty, for the point of the closed examination is not to test the students ability to perform on, say, Tuesday 15th June at 9.30 a.m., but her ability to perform under pressure and on demand. And whether she does the exam on Tuesday at 9.30 or Wednesday at 12.30 makes no difference to this (so long as there is no cheating, collusion or special advantage to taking the exam later). The ability to perform ‘under pressure and on demand’ is equally satisfied by someone who does the exam at a slightly different time to everyone else, because he still has to produce the philosophy there and then under exam conditions.

In order to maintain my hard-line position, I need to ask why we have closed examinations at all. What is so good about being able not merely to write good philosophy but to do so on demand and under pressure? Or to be more precise: what is the lifelong benefit that a student gets from being able to do this? Now it should be fairly obvious why medics or lawyers should be assessed like this, because in their careers they will be asked to apply their knowledge at specific times and places. (If you are wondering about the point of the Olympics, think of how such tests of skill indicated someone’s abilities as a warrior.) But no such consideration applies to philosophy students, not least of all because it is, for the overwhelming majority of our students, not a vocational degree.

However, we do hope that spending three years at University studying philosophy benefits our students in ways which have relevance to their future lives and careers. We often talk up the critical and analytical skills we teach, the ability to write clearly about complex issues, to argue coherently and persuasively, to see both sides of an issue etc. etc. How much better if our students can not merely display these skills but also display them on demand and under pressure? If you are an employer looking for someone who has good analytical skills, would you not also like to know that these skills will be employed whenever and wherever you ask? And if we think that one important point of closed examinations is to show potential employers that our students have not merely competence but can also perform at a high level, we devalue the goods by being too flexible. If we give the same certification of ability to perform under pressure and on demand, to someone who oversleeps and misses the exam, as we give to someone who turns up on time, then the ability displayed is not one of particular interest to potential employers.

There is no academic reason to insist that a closed examination be taken on time or not at all, but there is a reason for all that. We award degrees, and the value of those qualifications matters to our survival as a University. It would be ridiculous to think that the only value of a degree from York is the effect it has on our graduates’ employability, but that is one undeniable benefit, and, depressingly, it is the one which most of our students value most highly. If the point of teaching philosophy was only to bring our students one step closer to eudaimonia, then there would be no point in having closed examinations at all, let alone insisting they be taken at a specific time and place. But so long as we accept that it is some part of what we are doing to make our students better at their chosen careers, be that civil servant, diplomat, lawyer, accountant, journalist or whatever, we should take the hard-line on closed examinations. To give special treatment to someone who is late or absent through his own fault, is to reduce the value of the degree as an indicator of suitability for such careers.

.: posted by Tom Stoneham 10:54 AM

Thursday, July 10, 2003

Ralph Wedgwood emailed to say that something in my last post was wrong because a necessary truth cannot explain a contingent truth. Independently of whether what I wrote was wrong, this seemed an interesting issue (even, or perhaps especially, when we set aside all that stuff about necessary beings creating the universe).

My strategy was to try to get an example where p because (q and r), with q necessary and r contingent, and then argue that in that case ‘p because q’ was also true. The basic thought being that not all true explanations are complete explanations. So my example went:

Definition: x is a misanthrope iff (y)( x hates y if y is human)
Assumption: being human is an essential property, so John is necessarily human.

Suppose Peter is a misanthrope and he lacks all other reason to dislike John. So Peter hates John because John is human. Now a fuller explanation, offering a sufficient condition, might be that Peter hates John because Peter is a misanthrope and John is human. But that does not make the elliptical explanation false, just incomplete. It is true because it picks out the feature of John which is responsible for Peter’s hate. Suppose Mary is no misanthrope but still hates John, would not an adequate (though not logically sufficient) explanation be that John is boorish?

Anyway, whatever one’s views on that, Bryan Frances has come up with a better example. Explanation is an epistemic notion, and according to Kripke, necessity is not. So an a posteriori necessity might explain a contingent truth. Here is an example:

Suppose the author Ruth Rendell has a distinctive stylistic quirk. We then notice that the author Barbara Vine has the same quirk. We begin to wonder if they went to the same school etc. Then we discover that they are one and the same person, that she writes under two names. That explains it! However, it is not necessary that an author who writes under two names writes in the same style under each name, so a contingent fact has been explained by a necessary one.

This example, like mine, trades on the fact that a correct explanation might be incomplete, in that it does not cite a sufficient condition. For if we make the similarity between RR and BV such that RR=BV is sufficient for it, then what is being explained is not contingent.

.: posted by Tom Stoneham 11:36 AM

Monday, July 07, 2003

Deduction

Recent work by Paul Boghossian suggests that we should be very careful in distinguishing different questions which might fall under the general heading ‘the justification of deduction’. So I want to see how many I can separate.

1. What can we say to persuade someone who seriously doubts that all of our rules of inference are necessarily truth-preserving?
2. Can we justify, to ourselves, our use of deduction by showing each of the rules of inference to be necessarily truth-preserving?
3. Can we justify the practice of using truth-preserving rules of inference?
4. How can we show that a particular inference is valid?
5. How does inferring according to a valid rule from premises one is justified in believing justify one in believing the conclusion?

The answer to the first question is probably ‘Nothing’, because while one may be able to use non-deductive patterns of thought to persuade someone that deduction was useful, or generally reliable, it does not look like one could show that it was necessarily truth-preserving without using deduction itself. That is, of course, why the extreme foundationalist empiricism of Mill failed. The second question, however, looks to be the perfectly sensible one we are pursuing when we try to choose between various logics.

The third question is an interesting one. For many philosophers it is too ‘obvious’ to need addressing: valid inferences can never lead us astray. But there are two reasons for thinking this answer inadequate. (i) If we are not certain of our premises, which is the usual epistemic situation of humans, we may want a way of drawing conclusion from them which respects that uncertainty rather than conditionalizes it away. (ii) Deductive inference may have costs associated with it, such as time and computational complexity, which make rough and ready heuristics much more useful in real world situations. The extent to which you think that there is a disagreement here is a function of whether you think justification is merely a permissive matter. For one might think that deduction is always permitted, and that heuristics and probabilistic inferences are also permitted, though perhaps only in certain circumstances. But someone else might think that there is also a prescriptive notion in play around this area, a notion which tells us what inferences we ought to make, and if we have the prescriptive notion in play, it is not hard to generate a conflict between deduction and other rules.

Question 4 is one I find particularly interesting: do we recognize the validity of a particular inference directly, or by recognizing it as an instance of a valid form of reasoning? I think that the answer is ‘Both’. Sometimes we take an inference to be valid because we cannot imagine the premises being true and the conclusion false. Other times we do not need to engage in such imaginings because we recognize that the inference is an instance of a pattern we unhestitatingly accept (e.g. modus ponens). [Aside: Consider Vann McGee’s famous counter-examples to modus ponens. Only someone who has done some logic will recognize that as an instance of the same pattern as everyday modus ponens. Does learning logic teach us to overgeneralize the acceptable patterns? Or does it extend our recognitional capacities?]

Question 5 is the question Boghossian addresses in his most recent paper on the topic, ‘Blind Reasoning’. The difficulty here is seeing why, once we have answered all the others, this question remains to be answered. I guess just about everyone could agree on the following as jointly sufficient and severally necessary:
(i) Belief in the premises is justified.
(ii) The justification of the premises is independent of the thinker having justified belief in the conclusion.
(iii) The inference is valid.
(iv) The conclusion is believed because it follows validly from the premises.
Epistemological externalists and internalists will disagree on how to understand the ‘because’ in the conclusion, but should agree with the general form of the answer. So what is the problem? According to Boghossian, the problem is to flesh out (iv) in a way which is both true and non-circular. If the ‘because’ is merely causal, then it makes obviously unjustified conclusions justified. But if it is justificatory, then a vicious regress looms.

The difficulty lies in establishing these two claims.

.: posted by Tom Stoneham 9:01 AM