How does the observation that solutions judged as more similar psychophysically exhibit similar patterns of firing align with the predictions of distributed coding in the context of taste perception?