"There exists vast and valuable bodies of research in philosophy, psychology, and cognitive science of how people define, generate, select, evaluate, and present explanations.... This paper argues that the field of explainable artificial intelligence should build on this existing research."