Chapter 4: Determining Our Ends 1. Outline of the Chapter
2. The Problem of Self-Defeat
2.2. Self-Defeat and the Design Method
In the previous section, we did some work in attempting to isolate out and identity the phenomena of self-defeat. In this section, we will proceed in two stages. First, we will
go through a self-defeat argument from the literature to isolate issues that are likely to cause philosopher’s worries about the potential for self-defeat in philosophical positions.
Second, we will examine what I take to be the strongest version of a problem of self-defeat for the design view.
Michael Huemer has forcefully argued that anyone that denies the truth of phenomenal conservativism will ultimately hold a self-defeating belief (2001, 2007).
Phenomenal conservativism is the view that if it seems to me that P is true, and I don’t have relevant defeaters, then I have some degree of justification in believing P. His self-defeat argument goes as follows:
1. When we form beliefs, with few exceptions not relevant here, our beliefs are based on the way things seem to us.
2. If one’s belief that p is based on something that does not constitute a source of justification for believing that p, then one’s belief that p is unjustified.
3. Therefore, if seemings or appearances do not constitute a source of justification for us, then our beliefs are generally unjustified (including the belief that PC is false). (2007)
The core idea of this argument is that any theory of justification will be unjustified if it excluded the claim that phenomenal conservatism is true because the belief in that theory will ultimately have its basis in something that only the phenomenal conservative says can justify a belief. Our goal here is not to evaluate whether this argument works19, but only to extract something about the argument structure given. I think this argument as
19I think it ultimately fails due to a mistake surrounding how we should understand the notion of epistemic basis. Being the proximate cause of a belief is not the same thing as being its basis in the relevant sense, but that is a subject for another time.
well as ones given by philosopher’s like BonJour (1998) and Bealer (1996) in defense of intuition have a common structure.
Both Huemer and some defenders of a priori intuition begin with some method that they think is the basis of a large subset or even all of our beliefs. To defend that method from people who question whether it is a justifiable method they argue that any reason for questioning the justificatory status of the method ultimately relies on the use of that very method, thus the denier of their method is placed in a bind, either (i) their belief is unjustified because it has no proper source of justification, or (ii) it does have a
potentially good source, but is in fact false. This is an interesting kind of defense because it doesn’t establish that what the critic claims is false. It only establishes that if the critic of phenomenal conservativism or of a priori intuition is right, then ironically, they have no justification for saying so and thus their case should be unpersuasive.
Given this structure, we should expect that many self-defeat worries for the position I am advocating will occur if the method I am advocating brings into doubt the justificatory status of some highly general and ubiquitous belief forming method, and thus we will begin by turning our attention towards this prospect.
To begin let’s start by considering the set of beliefs that form the basis for the justification of epistemic norms on my picture. A huge number of beliefs go into shaping the contents of the norms on my view, but we can divide them into these rough
categories:
1) Beliefs about what are the best and/or most desirable epistemic ends.
2) Beliefs about the cognitive capacities of creatures like us.
3) Beliefs about the general functioning and maintenance of social norms.
4) Beliefs about the general environment we find ourselves in.
5) Beliefs about the likely result of the general adherence to a social epistemic norm given (2), (3), and (4).
These beliefs are formed in all sorts of ways by untold numbers of methods and thus there are lots of at least potential places where self-defeat might arise. But, let’s begin with a salient example. We might suppose that any initial beliefs about (1) are formed by something that appears to be intuition.20 It is imaginable that we might have the following issue. Suppose that we come to believe that allowing people to rely on intuitive beliefs tends to produce worse epistemic outcomes than forbidding it or limiting it more than we do now. If this were the case, then the Design Method would seem to sanction the belief in a self-undermining norm. To highlight the potential problem, let’s consider the following self-defeating argument for an anti-intuition norm that the Design Method might potentially sanction.
(P1) Believing truth and avoiding falsehood is the final epistemic end.
(P2) Social sanction of a norm endorsing beliefs formed via mere intuition tends to lead to less true beliefs and more false beliefs that alternative norm X.
(P3) We shouldn’t sanction norms that are worse at serving our final epistemic ends than some alternative.
(C) We shouldn’t sanction a norm endorsing beliefs formed via mere intuition.
20It's probably worth reminding the reader here that at least as far as I have characterized our way of forming beliefs like (1) in chapter 4, I have not rejected the use of intuition out right. Though I have mostly talked about knowledge of things like (1) as being grounded in knowledge or our own interests, which is likely a different phenomenon than linguistic intuitions or a priori intellectual seemings that some think ground knowledge of logic and metaphysics. I don't think anything in the design method is fundamentally hostile to these either, though I harbor some personal doubts about robust intellectual intuitions.
The self-defeat in this argument arises in so far as we have reason to think that the truth of the conclusion would undermine the source of the positive epistemic status of (P1) for us. In other words, if the design method told us to give up intuition, but intuition was how we came to be justified in believing that a particular purpose was the correct epistemic purpose in the first place, then it appears that we would have lost whatever initial justification we might have had to follow the method at all.
Of course, this case is only the most contextually salient possible case of self-defeat. The method advocated here has a high potential of producing significantly revised norms. It doesn’t just potentially threaten to bring into doubt our reliance on intuition, but potentially just about any other common sense means of forming beliefs. For example, it seems highly unlikely that the norms of reasoning or testimony that most people naively follow in their everyday lives are optimal for the service of their epistemic ends. This raise a potential problem. Anyone attempting to follow the design method will have to begin with some set of beliefs to initially begin the process of deciding what norms would best serve their ends, but all those beliefs were formed by following common sense norms. For example, I have some beliefs about how my cognitive system functions, but it is highly unlikely that in coming to form these beliefs I did so in accordance with whatever norms I would conclude were best suited for a creature like me to follow to achieve its ends. The design method sees norms as tools that we engineer through a process of optimization. There is essentially no chance that the tools we naively began with are identical to the optimized epistemic tools the method aims to work us towards.
This seemingly obvious point leads to the following potential argument against the design method:
(P1) A method is self-defeating if an agent correctly using the method would form a belief that entails that the method is an unjustified method of forming beliefs.
(P2) Using the design method is likely to produce a belief that our everyday methods of belief formation are unjustified.
(P3) The design method initially relies on beliefs produced by our everyday methods of belief formation.
(P4) If a method relies on another unjustified method, then it is an unjustifiable method of forming beliefs.
(C1) Using the design method is likely to produce a belief that entails that the conceptual engineering method is unjustified. (from P2, P3, and P4)
(C2) The design method is likely self-defeating. (from P1 and C1) It is to responses to this argument that we will turn in the next section.