The fundamental attribution error (Heider, 1958; Ross, 1977) has been extensively researched and explanations sought within a social cognitive framework. This work is reviewed, and it is noted that there is no unifying theory to account for the extensive catalogue of experimental work. Social cognitive explanations have proposed distinctions between perceptual, inferential and motivational functions within the person to account for the phenomenon. A phenomenological critique of this approach is then advanced. Drawing on the thought of Merleau-Ponty (1962) it is argued that our understanding of the phenomenon is enhanced by focusing not 'inside' people, but on interactions between them. In many ways, this brings us back to the project of Gestalt psychology, Heider's original framework for studying attribution.