Given my focus on practical rather than theoretical reasoning, I may at this point consider weakening the principle of total evidence and examining a presumption instead. The presumption to be examined establishes, for purposes of rational action, a generic bias in favor of more knowledge rather than on less. To defend the adoption of the presumption in favor of being maximally informed amounts to defending the belief that following it will lead, in the long run, to better overall results, in terms of goal fulfillment, than the results of following its antithesis (i.e., a presumption establishing a generic bias in favor of acting on the basis of less knowledge rather than on more), or indeed better than the results of a case-by-case balancing (i.e., of following no rule or presumption at all).

An important feature of a presumption as distinct from a general principle or rule is that it is rebuttable in concrete instances. In any concrete instance in which we consider applying the presumption, there may be reasons, or counter-indication, that will caution us against applying it in that concrete instance. When this happens, we say that the presumption is being rebutted in that particular case. A presumption is by its nature rebuttable: it has an implicit unless-and-until clause attached to it. (Thus, the presumption of legitimacy in the Anglo-American legal system establishes a generic bias in favor of treating the mother’s husband as the father of her child, unless and until there are counter-indications in a specific case.) Irrespective of being rebutted, a presumption may be revised: the generic rule itself may come to be viewed as ill motivated, or as having outlived its usefulness, or as unfitting to changed circumstances, and so on, and therefore may come to require revision. Revising a presumption may mean changing it, reversing it, or discarding it altogether.