(redating post originally published on 14 December 2011)

At 349, Reppert says: “We ought to draw the conclusion if we accept the premises of a valid argument”.

This is obviously wrong. Suppose, to take the worst case, that my beliefs contradict one another. If we are supposing classical logic — as Reppert clearly is — then, from my contradictory beliefs, using Reppert’s principle, I ought to infer that every claim is true. But, even though there IS a valid argument from premises I accept to an absurd conclusion, I ought NOT to “draw” the absurd conclusion. Rather, what I ought to do is to go back and examine my premises (my set of beliefs). Of course, this point is perfectly general: it is not just a point about the special case in which my beliefs are contradictory. Whenever I notice that my beliefs entail some further belief, it is an open question whether I ought to accept the further belief (“draw” the conclusion), or whether I should re-examine my premises (my set of already held beliefs).

This is not a minor quibble. Reppert throughout writes in a way that conflates logic and inference. It is one question what is entailed by beliefs that I already hold (this is a question of logic); it is quite another what I ought to infer — how I ought to change the beliefs that I hold — given the beliefs that I already hold. (In *Change in View*, Gil Harman notes other good reasons for not conflating logic and inference. For insistence, there is clearly a principle of “clutter avoidance” that applies to inference: you should only make the effort to draw inferences from given beliefs that you hold if there is some point to your so doing. “p” entails “p or q”; but, typically, there is no point in trying to enlarge your stock of beliefs by drawing inferences that conform to this logical principle. On the other hand, there is no principle of “clutter avoidance” that applies to logic: that one proposition entails another is utterly independent of our interests, cognitive economy, and so forth.)

In his discussion of “the argument from the psychological relevance of logical laws”, Reppert writes (380): “… laws of logic pick out ways of thinking that are correct regardless of place, time, or even possible world”. Not so. Laws of logic do NOT pick out “ways of thinking”. Laws of logic concern logical relations between propositions (entailment, inconsistency, etc.) Ways of thinking concern the revision of belief (possibly under the impact of new information). Connections between these topics are indirect. True enough, a collection of beliefs can be criticised because of the logical relations that hold between the contents of the beliefs: for instance, a collection of beliefs can be logically inconsistent. But this observation has nothing to say about the revision of belief: it cannot tell us, for example, which beliefs ought to be given up in order to return the system of beliefs to consistency. (Perhaps it tells us that we ought to revise our beliefs. But even this advice is not unconditional. If the inconsistency can be quarantined, and if it doesn’t have significant behavioural consequences, then it may fall so low on the list of cognitive imperatives as never to receive attention.)

Some of these difficulties can be avoided if, instead of talking about beliefs, we talk about theories. Since theories are identified by their non-logical axioms, they are not dynamic–and so we don’t get into the kinds of difficulties that emerge if we talk about beliefs.