An argument is premise-circular if its conclusion is contained among the premises, and rule-circular if it endorses the reliability of the inference rule by which it was attained.
A rule-circular argument might go as follows:
Induction, say, is the rule that if most observed A's are B's, then most A's are B's.
Is Induction reliable? Well, yes:
- Most inductive inferences have been successful so far, so
- most inductive inferences are successful.
Reliabilists argue that rule-cirularity (in contrast to premise-circularity) is alright because the only thing that matters for the justification of the conclusion is whether the inference rule is in fact reliable; others object that this is too permissible. I'm quoting Stathis Psillos (The Scope and Limits of the No Miracles Argument, 2011):
But couldn’t any mode of reasoning (no matter how crazy or invalid) be justified by rule-circular arguments? Take for instance what may be called (due to Igor Douven) Inference to the Worst Explanation:
Scientific theories are generally quite unsuccessful - These theories are arrived at by application of IWE - What is the worst explanation of this? - That IWE is a reliable rule of inference
Let’s call this, following Boghossian the bad company objection. How can it be avoided?
I am wondering about his response:
The reply here is that the employment of rule-circular arguments rests on or requires the absence of specifi c reasons to doubt the reliability of a rule of inference. We can call this, the Fair-Treatment Principle: a doxastic/ inferential practice is innocent until proven guilty. This puts the onus on those who want to show guilt. I take this to be a fundamental epistemic principle.
What about this "Fair-Treatment Principle", is there anything to say in favor of (or against) it?
Why should it be fundamental?
Is there any specific literature about it?