People make graded judgments about the inconceivable

Some impossible things are more impossible than others. Magically levitating a feather is easier than levitating a rock, even though both are impossible in the real world. But within the things that are inconceivable-e.g., "the concept of red writing a play" or "a girl being a prime number"-are some things more inconceivable than others? We first established that people have graded, systematic judgments about the likelihood of inconceivable events, using two stimulus sets that manipulate inconceivability in different ways (Experiments 1 & 5). We then examined three hypotheses as to how people make such judgments: the likelihood of the event description string (Experiments 2 & 6); the ease of a metaphorical interpretation (Experiments 3 & 7); and the difficulty of transforming a nonsense statement into a sensible one, as measured by distance in a type hierarchy (Experiment 5). We found that graded judgments of inconceivability are not captured by string likelihood, nor by metaphorizability, but do correspond to a measure of distance in a type hierarchy. Our results suggest that inconceivability is graded, and the perceived likelihood of an inconceivable event may be a product of one's ontology of the world.