A true, but often-sublimated fact of research is that one's conclusion depends on one's theory of the phenomenon under investigation. We illustrate this point using the inverted-face effect: The perceptual disadvantage of inverting a face is generally greater than the disadvantage of inverting other visual stimuli. Recognition memory was measured for novel faces or houses initially shown at varying durations either inverted or upright. As expected from past data, the inversion effect was greater for faces than for houses, based on the traditional stimulus type x orientation interaction. However, we generated a theory which assumes that, for both stimulus types, duration and orientation combine at perception into a single measure ("strength") and that face and house recognition performance are then determined by separate monotonic functions of strength. This theory, which implies the inversion effect to be identical for faces and houses, makes a very strong prediction that was perfectly confirmed..