LI was reading a paper by a philosopher, Alexander Bird, the other day. The paper defended the view that scientific progress is measured by the accumulation of knowledge – on the Baconian scheme – rather than measured by its generation of true statements, as the semantic philosophy of science would have it. It is coming out in Nous.


So far, so good. But then we came across this counterfactual:


“Imagine a scientific community that has formed its beliefs using some very weak or even irrational method M, such as astrology. But by fluke this sequence of beliefs is a sequence of true beliefs. These true beliefs are believed solely because they are generated by M and they do not have independent confirmation. Now imagine that at time t an Archimedes-like scientist in this society realises and comes to know that M is weak. This scientist persuades (using different, reliable 4 methods) her colleagues that M is unreliable. This may be that society’s first piece of scientific knowledge. The scientific community now rejects its earlier beliefs as unsound, realising that they were formed solely on the basis of a poor method.


“On the semantic view this community was making progress until time t (it was accumulating true beliefs) and then regressed (it gave up those beliefs). This, it seems, contradicts the verdict of our intuitions about this episode. The acquisition of beliefs by an unreliable method cannot be genuine scientific progress, even if the beliefs so acquired are, by accident, true. Far from being a regressive move, giving up those unreliably produced beliefs, because of a now well-founded belief that they were unreliably produced, is positive, progressive step. So the semantic view yields a description in terms of progress and regress that conflicts with what we are intuitively inclined to say.”

We don’t mean to pick on Dr. Bird, but this is a rather neat demonstration of what we call the fallacy of the epistemologically deviant condition. The counterfactual only gets off the ground once we suppose “a scientific community that has formed its beliefs using some very weak or even irrational method M, such as astrology. But by fluke this sequence of beliefs is a sequence of true beliefs.” The last sentence gives us, as a sort of axiom, the framing epistemological conditions that will allow us to judge the Bird’s counterfactual.

However, the last sentence is actually a historically contingent statement, even though it is be treated as an axiom – something that is true a priori. Since it is historically contingent, the fact that it is true entails a story about the discovery that makes it true. Such a story would necessarily overlap with the example it is supposedly framing. This means that the story of how, by some fluke, a community’s irrational beliefs, M, were also true beliefs would entail an investigation, if true, that would be formally equivalent to the investigation mounted by the Archimedes like scientist in the story.

The epistemologically deviant condition is a form of begging the question. It is, unfortunately, all too common in analytic philosophy. My friend Alan and I have been arguing, on his site, about Chalmers, the consciousness-man. Chalmers has a weakness amounting to addiction for epistemologically deviant conditions. His most famous argument, which revolves around postulating a zombie human double that can cogitate, speak, and behave like a human being, but doesn’t have conscious experience of being like a human being, violates the conventions of framing in exactly the same way Bird does, above. It is for this reason that these arguments never work.

However, I am less interested in their implausibility than in their motivation. These philosophic fictions share a frustration with the more artistic fictions of novelists and film-makers: how to pack all the information the author has into the story. The voice-over in a film is a perfect example of the kind of artistic compromises that emerge in the struggle between the creator and the material. The voice over doesn’t really have a logical place. Is it supposed to represent the Still-Sprache going on in the head? Is it supposed to be the filmic equivalent of the inaugural moment in first person stories – the fiction that some “I” has sat down to write a story? Oftentimes, the voiceover presents itself in the conventions of written fiction’s first person. Anybody who writes fiction knows the frustration of sticking with the person of the teller – including the frustration of third person telling, which is always about the writer’s calculated interference in the angle and unrolling of the story.

ps -- because I'm an incompetent logician, and because, frankly, nobody cares, I usually don't bother with the technical side of my arguments. But in this case, the technical side would go something like this:

Given a framing condition, S, containing a fact, s, that entails an argument, z.

And given a counterfactual, T, such that S frames T, containing a fact t;

If t entails z, then I'd call the counterfactual badly formed.

There's nothing here, really, except another self reference paradox. Usually, this is disguised by suppressing the epistemological source of s -- in other words, suppressing the answer to the question, how do we know s? It has been my experience that counterfactuals involve assumptions that usually render them either superficial or badly formed. Why? Because, on the one hand, if we can mount a straightforward argument for the framing facts, then we don't need the counterfactual; and if we can't, and have to fall back on the counterfactual, then it is illegitimately prior in the line of argument to itself -- in other words, we have the problem of the vicious circle.

For those few and hardy souls who've actually reached this point in the post, congratulations. Most of you have justly fled -- but I'm not going to do this kind of thing too often.