Few arguments against intentional states in animals have stood the test of time. But one objection by Stich and Davidson has never been rebutted. In my reconstruction it runs: Ascribing beliefs to animals is vacuous, unless something counts as an animal believing one specific “content” rather than another; Nothing counts as an animal believing one specific content rather than another, because of their lack of language; Ergo: Ascribing beliefs to animals is vacuous. Several attempts to block the argument challenge the first premise, notably the appeals to “naked” belief ascriptions and alternative representational formats. This essay defends the first premise and instead challenges the second premise. There are non-linguistic “modes of presentation”; these can be determined by attributing to animals specific needs and capacities—a “ hermeneutic ethology” based on lessons from the debate about radical translation/interpretation in the human case. On that basis we can narrow down content by exclusion. What remains is an “imponderability of the mental” which does not rule out attributions of intentional states to animals.