Proceedings of the Annual Meeting of the Cognitive Science Society

Interpretation of perception reports with verbs like see varies depending on what the verb combines with. When see combines with a full tensed clause, the perception report is thought to be epistemically non-neutral: the perceiver must form an event-consistent belief. But it is `perceptually' non-committal --- the perceiver need not have directly witnessed the event. When the verb combines with something smaller than a full tensed clause, the pattern is claimed to flip. The perceiver must have directly witnessed the event described by the complement. But, it does not commit them to a belief about what is perceived. We experimentally test speakers' interpretations of these two types of perception reports by holding constant direct perception and manipulating epistemic neutrality. We found that irrespective of complement size, participants did not treat perception reports as epistemically-neutral. We discuss two possible avenues to account for these unexpectedly strong readings, one semantic and one pragmatic.