Abstract

On most theories, “probable” is equivalent to “more probable than not”, interpreted either as comparative epistemic possibility (e.g., Kratzer) or as numerical probability (> 50%). Psychological experiments have revealed that subjects’ willingness to judge an event “probable” depends on the distribution of alternatives: in particular, most subjects are willing to judge an event “probable” if it is more probable than all contextual alternatives, even if it is clearly less probable than its negation. Since this conflicts with the definition of “probable” assumed, these results have been taken as evidence that people’s judgments about probability are illogical or incoherent, and that probability judgments are made according to a “gut-level”, “associative” system. I give an alternative semantic analysis based on comparison classes, which is independently motivated by the fact that “probable” is focus-sensitive, and develop it within standard degree-based semantics for adjectives and Beaver & Clark’s semantics for focus. This relieves us of the need to treat the behavior of experimental subjects as systematically illogical, which I take to be a good thing on charity grounds: plausibly, the only fallacy here is in the experimenters’ semantic assumptions.