Wednesday, March 26, 2008

"Discursive Dilemma"

Today at the LOGOS Colloquium, Stephan Hartmann discussed the so-called “discursive dilemma.” I was convinced by Genoveva Martí that it is not clear how to get a real dilemma from the examples. Suppose a hiring committee agrees to appoint a candidate if but only if s/he is strong both at research and at teaching. One third of them think s/he is, one other third think s/he is strong only at research, and the final third that s/he is strong only at teaching. It seems to me that a collective decision-making mechanism that allows the candidate to be hired in this situation is not the most reasonable one.

Pettit (2001) seems to suggest that, were the candidate not to be hired, the group would suffer from a certain sort of deficiency in “collective rationality”, as the majority think the candidate is strong at research, and the majority think that s/he is strong at teaching. That is true, but it certainly does not follow that the majority think that s/he’s strong both at research and at teaching—actually, the majority think s/he lacks one essential requirement to be appointable. Why should they hire the candidate??

19 comments:

Esa said...

Hi! Thanks for the post, it's really interesting.
As you say, in this case "the majority think s/he lacks one essential requirement to be appointable", that is, being both a good researcher and a good teacher. But I was just wondering what is the crucial question in this case: whether the candidate lacks the essential requirement, or whether the majority think the candidate lacks the essential requirement. If the former is the central question at issue here, then it might be argued that in the described case, we have good reasons to think that s/he satisfies the essential requirement: s/he is a good researcher (since the majority thinks so) and s/he is a good teacher (since the majority thinks so). So maybe s/he should be appointed after all?

Dan López de Sa said...

Well, on your assumption that the majority thinking something is so-and-so gives us good reasons to think it is, we are to conclude that s/he does not satisfy the requirement: s/he is not strong both at research and at teaching (since the majority think so), right? So it still seems to me that no, s/he shouldn't be hired, even under your assumption.

Esa said...

Thanks a lot for your comment! I think that this is how the dilemma could be generated: on the assumption that the majority thinking something is so-and-so gives us good reason to think it is, on the one hand we get the conclusion that she is NOT strong both at research and at teaching, as you argue, and on the other hand, we get the opposite conclusion that she is strong both at research and at teaching, since the majority thinks she is good at research and the majority thinks she is good at teaching. So on this assumption, we seem to get a contradiction.

Dan López de Sa said...

Which kind of goes against the assumption ;-)?

Esa said...

Perhaps so! ;-) But without such assumption, would we have reasons at all to think that the candidate should not be hired?

Dan López de Sa said...

We do: s/he should not be hired given that the committee agreed to hired the candidate on condition that s/he was strong both at research and at teaching, and most think s/he is not.

(True: most think s/he is one and most think s/he is the other, but this is not enough for most thinking s/he is both.)

The assumption which seems to lead to contradiction is not required for not hiring.

Manolo Martínez said...

Well, the committee is interested in doing the right thing: having the right candidate for the job (someone that excels at teaching and research).
If you defend the CBP it's, presumably, because you and the committee think that a majority of members believing that [she excels at research and teaching] is good evidence that she in fact so excels.
This is so, presumably again, because it is a particular case of the principle that whenever the majority of the committee believe that so-and-so, most probably so-and-so. So you still need the assumption for the votation to have a point, don't you?

Dan López de Sa said...

Not sure I follow you completely, sorry. We need to assume that if they agreed on doing A only if B holds and most think that it's not the case that B, then they shouldn't A. Is your thought here that this in turn requires or somehow presupposes the previous problematic assumption that if most think that B then (most probably) B?

Esa said...

Hi there! I am not sure either whether the previous problematic assumption is needed, but I think that at least something like the following is required: (1) "If a group of people agree on doing A only if B holds and most think that it's not the case that B, then they shouldn't do A."
This assumption sounds very reasonable. The problem, as I see it, is the following: there is another assumption that seems equally reasonable, but which, if applied to the same case, would lead to the opposite conclusion. And the assumption is: (2) "If a group of people agree on doing A only if (B & C) holds and most think that it's case that B, and most think that it is the case that C, then they should do A." To my lights, this does not commit the fallacy that Dan correctly pointed out above (namely, that if most think that A, and most think that B, then most think that (A & B)). I cannot think of any reason to favour one assumption over the other. I think that this is what gives rise to the so-called "discursive dilemma".

Dan López de Sa said...

(1) indeed sounds plausible to me. Even in the stronger biconditional version. And so does in particular (1*) "If a group of people agree on doing A if (B & C) holds and most think that it's case that (B & C), then they should do A." But I'm afraid I still fail to see how this should give any support whatsoever to (2).

Dan López de Sa said...

(Nor (2*): "If a group of people agree on doing A if (B & C) holds and most think that it's case that B, and most think that it is the case that C, then they should do A.")

Esa said...

Hi again! Yes, (2*) seems to be what is needed (for my argument), thanks for the correction. As I understand the problem (for instance in (Pettit (2001)), (2*) is supposed to have independent support (from that of (1) and (1*)).

Dan López de Sa said...

Ok, so there is a dilemma only for those who find not only (1) but also the independent (2*) compelling, right?

Dan López de Sa said...

Or rather: there is a dilemma only for those who find not only (1) but also the independent (2**) compelling.

(2**) "If they agree on doing A if (B & C) holds and most think that it's case that B, and most think that it is the case that C, even if most think (B & C) does not hold, then they should all the same do A."

Manolo Martínez said...

I've put all the principles together for ease of reference:

(PA) If most think that B then (most probably) B

(1) "If a group of people agree on doing A only if B holds and most think that it's not the case that B, then they shouldn't do A."

(1*) "If a group of people agree on doing A if (B & C) holds and most think that it's the case that (B & C), then they should do A."

(2) "If a group of people agree on doing A only if (B & C) holds and most think that it's case that B, and most think that it is the case that C, then they should do A.

(2*): "If a group of people agree on doing A if (B & C) holds and most think that it's case that B, and most think that it is the case that C, then they should do A.

(2**) "If they agree on doing A if (B & C) holds and most think that it's case that B, and most think that it is the case that C, even if most think (B & C) does not hold, then they should all the same do A."

(PA) is the Problematic Assumption that Dan questions. First of all, I think I must say that I think the cognate principles proposed are somewhat cheating: they substitute a premise that relates probabilistically the truth of two propositions (that would be PA) with a premise for practical reasoning, that relates a proposition with a course of action. It is cheating because, in fact, the course of action in question [in this case, granting tenure] is made reasonable by some some proposition closely related to the consequent of PA [i.e., that the candidate excels in both teaching and research].
We can see this more easily when we make the action even more closely related to the consequent of PA:

In (1*),

Let doing A be: Issuing the following veredict: “Elsa is an excellent teacher”

Let (B&C) be: Elsa is an excellent teacher.

We readily have the following argument:

1.[Substitution of (1*)] If a group of people agree on issuing the following veredict: “Elsa is an excellent teacher” if Elsa is an excellent teacher and most think that it is the case that she is, then they should issue the veredict.
2.[Premise] The Committee agrees on issuing the following veredict: “Elsa is an excellent teacher” if Elsa is an excellent teacher.
3.[Premise] Most in the Committee think that Elsa is an excellent teacher
4.[MP 1,2,3] They should issue the veredict: “Elsa is an excellent teacher”

Mutatis mutandis for issuing the veredict "Elsa is an excellent researcher" and for "It is not the case that Elsa is both an excellent teacher and researcher".

The committee is forced to issue all three veredicts, and that is problematic enough for me Is this clear? These little "comment" windows are impossible to manage!

Dan López de Sa said...

Honestly, that the committee is, given the envisaged situation, forced to issue the three verdicts looks to me no more problematic than the underlying fact: that most of them might think A, most of them think B, and still not be the case that most think both A and B. I'm really sorry if I'm just being the most stubborn ever here!

Manolo Martínez said...

Let's see, I took your original point to be that there was no real problem about the decision of the Committee: they should just use the CBP and leave the candidate out. The perplexity about their granting that both premises are true is misplaced.

What I'm saying is that, even if this is true, there are other ways to bring the inconsistencies to the fore. Ïf I was the Dean and where to receive the following three memoes from the Committee:

"Elsa is an excellent teacher"
"Elsa is an excellent researcher"
"It is not the case that Elsa is both an excellent teacher and an excellent researcher"

I would have to think it twice before trusting their decision. I trust only those agents that observe the most basic logical constraints on discourse, and the Committee (as a whole) is not such an agent. I don't know, maybe I am the stubborn...

Dan López de Sa said...

Maybe we are reaching an agreement; let me check if you also think so.

On the one hand, if one is to credit collectives with “beliefs” and these are to satisfy constraints such as that if a collective “believes” that A and that B that it “believes” that A&B, then it cannot be that a collective “believes” that A if most of its members believe that A. On the other hand, there is no “discursive dilemma” and in particular no reason why the committee should appoint the candidate in our previous example. Do we agree?

Manolo Martínez said...

This is how I would put things now. Maybe we do agree, but I'm not quite sure :)

1. Maybe the point is too stringently put if put in terms of beliefs. Probably the constraints for having beliefs are very strong, and it is unlikely that Committees in general will qualify.

2. What is true is that Committees should observe some general rationality constraints if they are to be credited with decisions at all. In particular, their observance of Conjuntion Introduction and Detachment rules is, that would be the claim, needed for us to credit Committees with the ability to decide.

3. So, we can agree in that putting this rationality-failure in terms of a "discursive dilemma" may not be the best way to bring the problems with Committee-decisions into focus and, indeed, there may be a way around the Dilemma along the lines that you suggest. But the rationality problem, under other descriptions, remains.
For instance, I'm not sure that the Committee *has decided* that she is not to be appointed (with her not being good both at teaching and research) given that it has also decided that she is indeed good at teaching and good at research. Most likely it has not decided anything whatsoever.
What do you think? Why should we take the conclusion-based verdict at face value given that Committees are such a cognitive mess?