First, we asked whether changes in animacy are encoded in visual working memory. People did a simple retention + change detection task. Changes that altered animacy (e.g., rabbit→boot) were more detectable than changes that did not (e.g., rabbit→dog), even though animacy was task-irrelevant.
