Maximal Rationality Redux

This past Friday I gave a talk at UN-L on my paper “On Maximal Rationality.” In that paper I defend an analysis of maximal rationality, which attempts to delineate what it takes to be immune from rational criticism. Luckily for me, I received some excellent feedback on Friday. In lieu of said feedback, I have revised the view considerably. Below the fold is the current view.

First, we must define some terminology. Let JSPA stand in for judgment sensitive propositional attitude.  Call the following type of belief a (1)-type belief:

(1)    I believe that I have decisive reason on my evidence to [JSPA] that ϕ.

So, the (1)-type belief for belief is,

(1b)    I believe that I have decisive reason (on my evidence) to believe that ϕ

and the (1)-type belief for intention is,

(1i)    I believe that I have decisive reason (on my evidence) to intend that ϕ

and et cetera for all of the other JSPAs.  Now, let’s call the proposition that is the content of the (1)-type belief a (1)-type proposition. Hence, the (1)-type proposition for (1b) is,

(1b-p) I have decisive reason (on my evidence) to believe that ϕ.

Finally, call the JSPAs of the following form (2)-type attitudes:

(2) I [JSPA] that ϕ.

Therefore, the (2)-type attitude for belief is the following,

(2b) I believe that ϕ,

and et cetera for the other JSPAs.

Second, the analysis of maximal rationality is the following:

MR: one is maximally rational iff one is both procedurally rational and substantively rational.

Hence, maximal rationality is a function of procedural rationality and substantive rationality. Procedural rationality can be defined thusly:

PR: one is procedurally rational iff one’s attitudes are structurally valid.

One’s attitudes are structurally valid when they are formally coherent.  In order to meet the demands of PR, an agent has to fulfill certain rational requirements.  Some examples are,

Modus Ponens Principle (MPP): One must: (If one believes ‘α’ and believes ‘if α, then ω’, then one believes ‘ω’).

Non-Contradiction Principle (NCP): One must: (If one believes ‘α’, then not believe ‘¬α’).

Means to an End Principle (MEP): One must: (If one’s end is E, and one believes M is a means to E, one intends to M).

Take a Stand Principle (TSP): One must: (If one has a (2)-type attitude, then one has a (1)-type belief).

The examples above all have to do with the structure of one’s attitudes, not about which attitudes are rational to have.  Thus, in terms of PR, if one violates NCP, then one can right her wrong simply by ceasing to believe either ‘α’ or ‘¬α’.

You might think that PR is all that needs to be satisfied in order to be immune from criticism. I think that is false. In support of this, consider two cases from my paper:

Case one: Emmett grows up in rural Louisiana. His parents are fundamentalist Christians who believe that the earth is only 10,000 years old. Currently, he is 25 years old and has always lived in rural Louisiana. He lacks all access to the outside world. He has a fully consistent set of attitudes. He believes that the earth is only 10,000 years old. From this belief, he forms the intention to assert that the earth is only 10,000 years old. He asserts that the earth is only 10,000 years old.

Case two: Emmett grows up in rural Louisiana. His parents are fundamentalist Christians who believe that the earth is only 10,000 years old. Currently, he just began as an assistant professor of geology at a major American university. He has also just finished a PhD dissertation in which he assumed that the earth is much older than 10,000 years old (he even unfaithfully asserted that there is irrefutable evidence that the earth is older than 10,000 years old). Despite this, he still believes that the earth is only 10,000 years old. This belief is part of a consistent set of attitudes. Because of this belief, he forms the intention to assert that the earth is 10,000 years old. He asserts that the earth is only 10,000 years old.

By stipulation, Emmett’s attitudes are structurally consistent in both cases. In both cases he has the (1)-type belief that he has decisive reason to believe the earth is only 10,000 years old. Moreover, in each case he has the corresponding (2)-type belief (that the earth is only 10,000 years old). Moreover, it’s stipulated that Emmett satisfies all of the other principles relevant to PR. However, in case 2 he seems to be clearly open to a type of criticism he is not open to in case 1. Namely, in case 2 he is badly mistaken about the relevance of certain facts he is in a position to know. If this is right, then PR cannot be necessary and sufficient in determining the realm of criticizability. Thus, we need an additional concept.

I argue that that concept is substantive rationality. I define substantive rationality thusly,

SR: one is substantively rational iff  (i) for every (2)-type attitude that one has one knows the corresponding (1)-type proposition and (ii) one believes all of the (1)-type propositions that are salient for her.

The first conjunct says that for every (2)-type attitude one has, one must know the corresponding (1)-type proposition.  For example, if one holds (2*)

(2*) I believe that ϕ,

then one must know (1-p*)

(1-p*) I have decisive reason on my evidence to believe that  ϕ.

Since one must know (1-p*), it follows from the factivity of knowledge that (1-p*) is true.  If (1-p*) is true, then one is justified in holding (2*), and hence is not open to criticism for having an unjustified belief (mutatis mutandis for the other JSPAs).

The second conjunct requires one to believe all of the (1)-type propositions that are salient for her. We can define saliency in the following way:

Saliency Principle (SP): (1-p) ψ ((1)-type proposition ψ) is salient for A iff (1) A is in a position to know ψ and (2) A needs to take a stand about ψ in order to be procedurally rational.

SP1 is designed to eliminate problems with epistemic access that plague some theories of reasons with implausibility (or so it has been argued).   For example, in case 1 above Emmett’s only evidence about how old the earth is comes from theological sources.  In other words, the only information he has every heard about how old the earth is has come from sources that claim the earth is only 10,000 years old. Thus, it seems intuitive that he is rational in believing that the earth is only 10,000 years old, even in terms of substantive rationality. Emmett isn’t criticizable, that is, for failing to believe that the earth is older than 10,000 years old because no evidence that he is in a position to know supports this. Thus, in order for one to be criticizable for not having a (1)-type belief, she must be in a position to know the corresponding (1)-type proposition.

SP2 is designed to sufficiently narrow the set of reasons that the agent is in a position to know in response to cases like the following.  As I walk down the hall to my office, I am barraged by sensory information.  For example, I am in a position to know that there is a poster advertising a clothing store downtown, or a geology talk, etc.  In other words, I am in a position to know the following (1)-type proposition:

(1-pC) I have decisive reason to believe there is an advertisement for a clothing store on the wall of the department.

Now, if I have no prior (2)-type attitudes about the content of the walls, am I criticizable for failing to believe (1-pC)?  I take it that the overwhelming intuition is that I am not criticizable.  If I were, then I would be required to have way too many (1)-type beliefs.  Hence, we need some way of deciding which (1)-type propositions that I am in a position to know are the relevant ones for criticizability.  SP2 tells us which ones are relevant.  Namely, the (1)-type propositions that are relevant are those that we must take a stand on in order to be procedurally rational.  Normally, the Take a Stand Principle (TSP) is going to be the most relevant PR principle to meet this condition.  TSP requires an agent to have a corresponding (1)-type belief for every (2)-type attitude.   In other words, if the agent believes p, she is required by TSP to believe that she has decisive reason to believe p.  If she fails to have the latter belief, but she has the former, then she is accepting the former on blind faith.  She doesn’t think she has decisive reason to hold the belief, but she does anyway.  Thus, she is open to criticism.  Bringing this back to SP2, we get the following: since TSP is (ex hypothesi) a requirement of PR, the salient (1)-type propositions that agent A is in a position to know are those propositions that correspond to (2)-type attitudes the agent holds.  Hence, if I believed that there were no advertisements currently (as I walk down the hall) on the walls, then (1-pC) is salient for me currently.  Thus, since I don’t know (1-pC) (because it’s false), I am open to criticism for having an unjustified (2)-type attitude.
Let’s sum up the requirements of SR, starting with the second conjunct and ending with the first.  The second conjunct requires you to believe all of the (1)-type propositions that you are in a position to know and that correspond to (2)-type attitudes that you hold.   The first conjunct, in turn, requires you to know all of those (1)-type propositions.  Thus, if you don’t know them, you are open to criticism.  Hence, if you believe a false (1)-type proposition, you are criticizable.  But, if you know all of the (1)-type propositions that are salient for you, and you are procedurally rational (and hence have the corresponding (2)-type attitudes),  then you are immune from all criticism.

That is a short (and let me say that some of it is directly lifted from a draft of the paper) summary of the view as it stands now. Comments, as always, are greatly appreciated.

~ by Errol Lord on October 29, 2007.