Monday, March 14, 2011

Cognitive science arguments against religious beliefs

This is the first in a series of posts about whether psychology can debunk religious beliefs. I will begin by examining debunking arguments highlighted by Gregory Peterson in his recent article "Are Evolutionary/Cognitive Theories of Religion Relevant for Philosophy of Religion?" (Zygon,Volume 45, Issue 3, pages 545–557, September 2010). Peterson considers psychological debunking arguments from three different fields: cognitive science, neuroscience, and evolutionary biology. In this post, I will focus on debunking arguments from cognitive science.

Cognitive science debunking arguments
The first cognitive science argument goes like this:

(1) According to cognitive science, the formation of all religious beliefs is completely explained in terms of certain psychological mechanisms.
(2) According to religious believers, the formation of all religious beliefs is not completely explained in terms of certain psychological mechanisms.
(3) Religious believers are mistaken about the explanation of why they hold their religious beliefs: the real explanation is given by cognitive science.
(4) Therefore, all religious beliefs are undermined by cognitive science.

Peterson points out, rightly enough, that this argument seems to commit the genetic fallacy: explaining why a belief is held does not show it to be false or rationally unjustified. 

We should notice some further problems with the argument as well:
A. The first premise is very strong, claiming that the cognitive science explanation of religious beliefs is complete. This would seem to beg the question against certain kinds of interventionist theism. If we weaken the premise, however, then the argument will not be as strong.
B. The second premise is not something religious believers need to be committed to. All they need is that the formation of their beliefs be justified, not that there be some special divine influence on their belief formation.
C. Finally, the argument relies on pointing out a mismatch between the explanations for religious beliefs given by cognitive scientists and by individual religious believers. But it’s unclear why this is relevant. Suppose a religious believer reads a lot of cognitive science and becomes convinced that he holds his religious beliefs because of the relevant psychological mechanisms, so that there is no mismatch between the explanations of the believer and those of the cognitive scientist. In this case, the argument simply doesn’t go through, which is puzzling. If anything, we’d expect the threat to religious belief to be greater the more the religious believer knows about the supposedly debunking psychological explanations. 

In light of these issues, as well as Peterson's objection, we can consider a more powerful version of the argument, the one that Peterson focuses on:
(1) Cognitive science completely explains how all religious beliefs are formed in terms of certain psychological mechanisms that have no truth-conducive relation to supernatural entities.
(2) If a belief about supernatural entities can be completely explained in terms of something that has no truth-conducive relation to X, then that belief is thereby undermined.
(3) Religious beliefs are about supernatural entities, and (by (1)) are explained completely by psychological mechanisms, which have no truth-conducive relation to supernatural entities.
(4) Therefore, all religious beliefs are undermined by cognitive science.

This argument has nothing to do with what explanation a believer gives for their beliefs, so it avoids the third problem I pointed out for Peterson’s version. It also avoids the second problem, since it doesn’t require theists to be committed to a special causal history for the formation of religious beliefs. The worry about the first premise remains, however.

Peterson fleshes out the argument by focusing on Pascal Boyer's account of religious belief formation, which goes like this:
-Our modular cognitive architecture, which evolved based on selection pressures, includes both (i) an agency detection module and (ii) an intuitive folk concept of agency.
-The agency module is hyperactive, sometimes detecting agency where there is none, including (so the argument goes) supernatural agency. Supernatural agency violates the intuitive folk concept of agency. So, supernatural agency is biologically counterintuitive, and belief in such agency is therefore more likely to be remembered, passed on, etc.
-In summary: The origin of religious beliefs is accounted for by the agency detection module, and the perpetuation of religious beliefs is accounted for by the fact that supernatural agency is counterintuitive.

As an account of the origin of religion, this argument does not undermine religious beliefs, according to Peterson. He thinks that it would only apply to the first religions, and would require some additional premises to be undermining at all.

As an account of the perpetuation of religion, the argument has some force as a hermeneutic of suspicion. How much force it has depends on the particular religious tradition in question.

I agree with Peterson that the argument is generally unsuccessful, partly because of the problem I mentioned earlier with the first premise. A second, and even deeper, concern is with appealing to Boyer's account of religious beliefs. In order for the argument to go through, we need to know that the agency detection module is not just generally hyperactive (i.e. giving lots of false positives), but also that in the case of religious beliefs in particular it is giving false positives. But we obviously cannot know that, unless we knew whether the religious beliefs in question were true or false, or justified or unjustified. To know that a belief, considered as an output of a psychological process, is a false positive, we have to know whether that is the result that process should have produced. Obviously we make very many valid inferences about agency, as well as plenty of invalid ones. The valid ones are justified on the basis of the subject's available evidence, the invalid ones are not. The question of when the evidence justifies the inference is a substantive question in epistemology, and one that is closely connected with the very issue at stake: the rationality of religious beliefs.   To give the cognitive science argument on the basis of Boyer's theory is, therefore, to beg the question against religious belief. 


  1. You never explain why Boyer needs to show that the agency detection module is especially hyperactive with respect to religious beliefs. I don't see why we need to identify particular instances of false positives in the case of religious beliefs to effectively use the hyperactivity of our agency detection module to undermine confidence in religious beliefs.

    General hyperactivity might be sufficient, for example, if false positives turn out to be correlated with certain features of contexts in which the relevant ascriptions were made and these features also hold true of contexts in which supernatural agency is ascribed. One such feature may be that the fact of agency would be especially comforting for the ascriber.


  2. Emad, thanks for your comment.

    To begin with, it's compatible with agency detection being hyperactive in general that it is fairly or even perfectly reliable with respect to beliefs about supernatural agents in particular. To get a debunking argument, then, we need something more.

    You give a suggestion: that "if false positives turn out to be correlated with certain features of contexts in which the relevant ascriptions were made and these features also hold true of contexts in which supernatural agency is ascribed," we might have enough to rationally undermine religious belief.

    This is interesting, but I don’t think it helps the debunker. Certain features of circumstances might make some beliefs less reliable, without having that effect on others. Darkness makes my beliefs about the colors of objects less reliable, but does not affect the reliability of my beliefs about objects’ shapes. Likewise, in the case of religion— the positive impact on subjective well-being due to inferring agency (supposing that it’s generally positive, though this is questionable- a malign demon is less comforting then a fluke natural disaster, so there might be a net neutral impact on subjective well-being from inferring agency) might make some of my beliefs about agency less reliable, without doing so for others. It’s that possibility that the debunker needs to address.

    Perhaps the debunker can make progress by appealing to religious disagreement. People hold an array of incompatible religious beliefs, one might say, so obviously our mechanism of forming them is unreliable. But general disagreement tells us nothing about the reliability of particular believers. Supposing there is a true religion, those who hold the tenets of that religion for rationally appropriate reasons will have reliably formed true beliefs, even if many other adherents of that religion, and the adherents of other religions, did not form their beliefs reliably.