A forum for VE lucubration

Friday, May 04, 2007

Troubleshoot Your Swamping Solution: Some Test Cases

It is widely assumed that knowledge is valuable, and additionally, that it is more valuable than any proper subset of its parts. It’s not clear that this fact can be explained in terms of the instrumental value of knowledge, for the reason that knowing p has many (if not all) of the practical benefits of truly believing p. And so, to the extent that we want to preserve our pre-theoretical intuitions about the value of knowledge, we must find some way to explain why knowledge is more valuable than its subparts, and in a way that appeals to the non-instrumental value of knowledge.

Process reliabilism has undergone a barrage of criticism for the reason that it does not appear to have the theoretical apparatus to explain this fact.
According to process reliabilism: S knows p iff (i) S truly believes p, and (ii) S’s belief that p was produced by a reliable belief forming process.
We think that ‘being produced by a reliable belief forming process’ is a valuable property for a belief because it means that the belief is more likely to be true than if it were produced by an unreliable process. So far, so good. The twist is this: If ‘being produced by a reliable process’ is valuable for the reason that ‘being likely to be true’ is valuable, then how can we explain why the property of ‘being likely to be true’ adds any valuable to a true belief?
As Jon Kvanvig (2003) puts it, the property of ‘being likely to be true’ is swamped by the property of ‘being true.’ And, thus, the indictment of reliabilism is that it defines knowledge in such a way that knowledge is equally as valuable as mere true belief—and this contradicts the platitude that knowledge is more valuable than true belief.

The reliabilist appears to have two escape routes:

First escape route: Admit that being produced by a reliable process is valuable for a belief to have for the reason that ‘being likely to be true’ is a valuable property for a belief to have, and then deny that the property of ‘being likely to be true’ is swamped by the value of a true belief. (I don’t know of any who have taken this route).

Second escape route: Deny that being produced by a reliable process is a property for a belief such that is value is exhausted by whatever value ‘being likely to be true’ accrues to a belief. Of course, this second escape route will require that the reliabilist locate what it is about ‘being produced by a reliable process’ that gives it value over and above the value it would confer to a belief by making it likely to be true. Goldman and Olsson (forthcoming) and Kristofer Ahlstrom (forthcoming) have taken this route.

Both Goldman and Olsson and Ahlstrom, in taking the second escape route, try this approach: they say that being produced by a reliable belief forming process makes it more likely that you will have future true beliefs. And so, ‘being produced by a reliable process’ is valuable not only because it makes the belief likely to be true, but also because it raises the conditional probability that future beliefs will be true.

I will not comment here on these projects, but rather, want to mention a different type of reliabilist response, which I think offers a more appealing response.
Consider the difference between:

S’s true belief is a result of a reliable belief forming process.
S’s came to believe p truly because of her reliable belief forming process.

The second of these two articulations has promise where the first does not. This is because the second articulation requires that there be an important connection between the cognitive success and the cognitive process that brought it about, a connection with we shall see can be valuable in a way that a cognitive success merely produced by such a process is not.

To bring what’s valuable in this connection into focus, we begin by amending the locution of ‘belief forming process’ in a such a way that makes possible normative evaluation. We, thus, revise the proposal in (something like) the following way:
…believes p truly because of S’s cognitive ability.

Success because of ability amounts to an achievement. Reaching the truth because of exhibiting cognitive ability is something we take to be more valuable than reaching the truth as (for example) True Temp does. We are inclined to say that success through ability deserves credit, and is as such valuable.
This is the idea being advanced in recent projects by Ernest Sosa (forthcoming), John Greco (forthcoming) and Duncan Pritchard.

But in order to escape the initial dilemma, it must be that the success through ability is such that it is valuable in a way that is not ‘swamped’ by the value of truth. That is, our new condition must confer on a true belief value that is something other and above the value that ‘being likely to be true’ would confer to a belief.

On the ‘success through ability’ proposal, the idea is captured in two theses:

1. Achievements are valuable for their own sake.
2. The supervenience base of achievements lies in their relational (non-intrinsic) properties.

The first thesis is the first step out of the ‘swamp.’ That is, by principle, anything valuable for its own sake is going to be such that it is not going to be ‘swamped’ by the value of something else. Achievements, then, are not merely instrumentally valuable, as ‘being likely to be true’ is instrumentally valuable.
The second thesis is the second step out of the swamp. It is a vindication of (1). I’ll explain how this vindication works by example. Suppose we have two guitars. One was used by Neil Young to record his album Harvest (suppose further that everyone thinks—(quite reasonably)--that this is a great album!). The second guitar has all the same intrinsic properties as the first, but was not used by Neil but rather stored in a warehouse in New Jersey. Suppose that only you know the first was used by Neil and that no one would believe you if you told them. (This is meant to prevent us from thinking that the first is of more instrumental value than the second). We would still find the first more valuable than the second. The explanation for this cannot be that the first is more valuable because of its intrinsic properties, given that both guitars have the same intrinsic properties. We should say, then, that the guitar is valuable for its own sake (not instrumentally), and that is value lies in its relational properties.

This is the idea that the ‘success through ability’ theorist has in mind when explaining the value of knowledge. A true belief that is a result of a cognitive achievement is valuable for its own sake, and because of its relational properties, in a way that a true belief not a result of an achievement is not. This sort of value that achievements have is final value, and if knowledge requires achievement, then knowledge has final value which isn’t shared by mere true beliefs. Thus, the ‘success through ability’ theorist has a coherent and persuasive way to vindicate our assumption that knowledge is distinctively valuable.
Despite the promise this proposal has as a path out of the swamp, there are nonetheless challenges that face the account. I think that there are two central problems that rise to the forefront.

(1) Objection to ‘success through ability’ as a sufficient condition for knowledge: There seem to be cases in which an agent meets this condition but does not know. These cases are (usually) ones in which the achievement occurs in a case in which malignant luck undermines knowledge.

(2) Objection to ‘success through ability’ as a necessary condition for knowledge: There seem to be cases in which an agent knows, but is not such that her believing p truly constitutes an achievement.

Both of these objections, if not properly addressed, threaten the ‘success through ability’ approach as a viable solution to the value problem it is meant to assuage. If the first objection is not met, then what makes knowledge distinctively valuable, on the proposal, would also make states that fall short of knowledge distinctively valuable. Thus, the idea that knowledge is more valuable than that which falls short is threatened.

The second objection, if not met, would cast the stone from the other side: there would be cases of knowledge that, by virtue of not requiring an achievement, would be bereft of an explanation for why they are more valuable than that which falls short.

I’ll turn now to what I think might be some challenging cases for the ‘success through ability’ proposal. Some of these cases will challenge the idea that achievement is necessary for knowledge, others will challenge the idea that achievement is sufficient for knowledge.

Case 1: Flowers for Algernon
Flowers for Algernon is a novel by Daniel Keyes in which the protagonist, Charlie Gordon, is a mentally retarded janitor. He is chosen for a unique experiment, which is to undergo an experimental brain surgery that would raise his IQ. As it turns out, the experiment works (and ultimately, to his demise). Suppose that such a procedure were possible: I think that an initial response would be that an agent who comes to know things he would not have known had it not been for the experiment, is not properly creditworthy, as his cognitive success might be more appropriately attributable to the scientists, than to his agency. This seems to suggest that perhaps creditworthiness is not required for knowledge. But add this twist: Suppose that scientists are able to perform this procedure, but because it is so expensive, only one person will be the lucky recipient. To determine who gets to have the procedure done, the scientists have an intellectual contest to determine who is most worthy. Albert E., who is naturally very clever, prepares tenaciously for the contest and wins. Albert undergoes the experiment, and his IQ is raised to an even higher level. Suppose that Albert comes to learn X, Y and Z in such a way that he would not have had these true beliefs had it not been for the surgery. (Suppose that X, Y, and Z) are facts about theoretical physics too complex for even the greatest geniuses. Albert’s cognitive successes are ‘because of’ his surgery. Interestingly, that he had his surgery is ‘because of’ his cognitive abilities.
(1) Does Albert know X,Y and Z?
(2) Does Albert’s coming to believe X, Y and Z truly constitute a cognitive achievement?

Case 2: Math Exam
Suppose that Peter is a mathematical whiz, who has become bored with school and has elected not to attend class or lecture for weeks. Finally, on threat of expulsion, he shows up for the final exam, unprepared. The teacher passes out a handout that includes five versions of the quadratic formula, four of which are false and one correct. The exam consists of applying the formula to 10 problems. Peter is discouraged because he knows he is clever and that, if given the right formula, he would have no trouble executing its application to a tee. With a sigh, he finally picks one at random (the correct one) and solves all ten problems correctly. Suppose that, had he chosen any different formula, he would have solved them all incorrectly (albeit, in accordance with the false formula). As it turns out, Peter gets the highest grade in the class. Even those who memorized which formula was correct struggled to apply it correctly to the 10 problems—a task which was easy for Peter.
(1) Does Peter know the answers to the 10 problems?
(2) Do Peter’s correct answers to the 10 problems constitute cognitive achievements?

Case 3: Spellbound
Little Susie Speller is obsessed with winning first place in the Scripps-Howard National Spelling Bee. Her overbearing parents provide her with the latest OED dictionary and hire the finest spelling tutors to assist her in her quest. Susie immerses herself in Latin and Greek word roots, linguistic etiology and after a year and a half, has learned to spell almost every word in the dictionary. A week before the spelling bee, Susie Speller is hit by a drunk driver (or, to avoid a sentimentalist response, suppose she is so stressed that she gets drunk and causes a wreck). In any event, her memory is badly affected, and seems to come in spurts. During the week before the Bee, she finds herself going through intervals where she spells words (even the toughest ones) perfectly for about a half an hour, and then lapses into a subsequent thirty minutes in which she has trouble spelling any of the words. Nonetheless, Susie shows up to the Spelling Bee, and luckily for her, the rotations are such that everytime she is called to spell a word, she is operating in one of her lucid intervals. (And, of course, when she sits down between turns, she goes blank). Susie ends up winning the Spelling Bee. Clearly, she would not have won had it not been for her spelling ability. Had she not studied, then car wreck or not, she would have no chance. However, had she drawn a different rotation number, she would have been given a word during her blank periods and would have lost.

(1) Did Susie know the words she spelled?
(2) Was Susie’s success at the Bee an achievement?