Attributing mental states to animals is charged with anthropomorphism, and the charge is applied asymmetrically. Explaining a chimpanzee's behaviour by supposing that it wants something and believes something is treated as a lapse requiring justification, while explaining it by conditioning histories is treated as the null hypothesis, though the second is also a theory and is also underdetermined by the observations. The methodological rule that the simpler explanation should be preferred does not settle which is simpler, since a mentalistic account is often shorter and the behaviourist reconstruction correspondingly baroque. What began as a corrective against sentimental over-reading has hardened into a presumption that shapes what counts as evidence, which is a familiar way for a methodological caution to become a substantive commitment without ever being argued for. Defaults are rarely defended, because nobody notices the moment at which they are adopted.