Unconditionals are syntactic conditionals whose affirmation affirms their consequent, unconditionally. Prominent instances were addressed by J.L. Austin ('There are biscuits if you want some') and Nelson Goodman (even-if 'semifactuals'). Their detailed features are explained in a Decision-Theoretic Semantics (DTS) which extends, by certainty and relevance conditions, the "CCCP" conditional probability construal of conditionals due to Ernest Adams and others. The construal of assertions of conditionals as conditional acts, defended by Keith DeRose and Richard Grandy in 1999 against objections arising from (...) Austin's unconditionals, is shown to be incompatible with any known version of CCCP. However, an ad-hoc decision-theoretic construal of their proposal is seen to work for Austin's unconditionals, albeit for these only. The DTS account of unconditionals hinges on stochastic independence or irrelevance conditions. For Austin's variant, it also involves changes in the expectations of non-indicator random variables, i.e. properly valuational relevance relations. The widely assumed speaker's knowledge presumption for assertion in general is reconstrued non-transcendently as a proposal to update to ostensible common certainty. This imperative replaces the factive 'T-axiom', KA-> A, of modal logical explications of knowledge to yield an doxastically imperatival, but evidence-sensitive theory of assertoric truth claims. On such a basis, Max Black's and G.E. Moore's ostensibilist account of assertors representing themselves as knowing is defended against Tim Williamson's 1996 and Igor Douven's 2006 respective deontic proposals for a knowledge or rational credibility requirement. Moreover, ostensibly unconditional, unhedged assertion is reconciled with a reality of dimly perceived potential defeaters on recognizing it to be a tacit semifactual, if need be. (shrink)

Bodies of collective knowledge evolve through individual action, like all products that have a use. They also can be evaluated from the engineer's optimizing design perspective. But can individual participants in their making recognize local optimality? Can they work to realize it? Are they unable to act seriosly in a way that would ensure acquisition of a certain suboptimal design feature? One might hope for a simple answer: appeal to innate constraints on the form of categorization. But such constraints cannot (...) wholly pre-empt the need for individual checks and thus for agency. The use of categorial schemes engages facts about the world that do not leap to the eye. They are unlike syntactic or phonological constraints of language, or edge-and orientation-detection algorithms of visual perception. What we should thus be looking for will be modest: constraints, innate or otherwise, that rule out the choice of categorial schemes which are clearly suboptimal and that are, in principle, open to agents' instant observation. We look to a limited if important domain of systematic knowledge: medical diagnostic knowledge. (shrink)