Coherence theory studies the semantic bonds between segments in a representation (e.g., the clauses in a discourse, or the individual shots in a film). These semantic bonds can be conceived of as implicit questions that interpose between segments (e.g., Mary is annoyed. [Why?] John ate soup.). In this paper, I consider the thesis that the source of these implicit questions is the attentional system (whose state can be approximated by a question meaning, or “attentional set” [Simons 2000]). Perhaps the question Why? corresponds to a natural shift of endogenous attention resultant on reading the initial segment Mary is annoyed.

This account is general enough to capture parallel coherence phenomena in film. We might achieve a similar interpretation by cutting from a close-up of Mary looking displeased to a shot of John eating soup. Indeed, the account was inspired by Hochberg and Brooks' (1978) theory of “visual momentum” in film, and in particular of the spatial coherence induced between shots by a “sight link.” On their theory, the shot of a character gazing off-screen at something raises the “visual” question, What is the character looking at?. This question captures the attentional state of the visual system at the cut-point—deictic gaze being a well established endogenous cue directing attention to its object (Smith 2005, sec. 3.2.2)—but it also captures the semantic bond between the two shots, since the second (or “object”) shot in a sight-link is interpreted as answering the visual question by revealing the object of the character's gaze.

Hochberg, J. and Brooks, V. (1978). Film cutting and visual momentum. In J. W. Senders, D. F. F. and Monty, R. A., editors, Eye Movements and the Higher Psychological Functions, pages 293–317. Lawrence Erlbaum, Hillsdale, NJ.
Simons, D. J. (2000). Attentional capture and inattentional blindness. Trends in Cognitive Sciences, 4(4):147–155.
Smith, T. J. (2005). An Attentional Theory of Continuity Editing. PhD thesis, University of Edinburgh.