Cognitive bias and the poetics of surprise

Abstract

The 'curse of knowledge' is a pervasive cognitive bias that makes it very difficult for us accurately to imagine, once we know something, what it is like not to know it. This article analyzes examples drawn from both novels and films to demonstrate that this bias plays a substantial and previously unexamined role in narrative structure. I argue that narratives often take advantage of the curse of knowledge to solve an ongoing storytelling dilemma: how to engineer satisfying twists that genuinely surprise audiences but also avoid coming off as non-sequiturs or cheats. The curse of knowledge provides a useful mechanism to encourage readers to over-generalize propositions in predictable and reproducible ways, while making it likely that they will also agree, in retrospect, that these generalizations were mistaken. The same bias serves to enhance the impression, in hindsight, that the narrative's outcome was indeed possible to predict. Finally, building on Mental Spaces theory (Fauconnier, 1985, 1997) and simulation-based theories of language processing (e.g. Barsalou, 1999; MacWhinney, 2005), I argue that the curse of knowledge is an artifact of a more general cognitive shortcut that is implicated in features of 'correct' sentence interpretation such as presupposition projection as well as in phenomena traditionally described as curse-of-knowledge errors. This account unifies the discussed examples and helps to explain why certain devices, particularly unreliable narration, emerge so frequently as aids to narrative surprise.