Act-Dependent Direction of Answerability: How the Kind of Act Decides What Must Match What

This conceptual paper introduces Act-Dependent Direction of Answerability, a distinction concerning what must answer to what when words and events diverge. Descriptions and predictions are assessed against situations; later conduct is assessed against commitments; expressed plans direct action toward intended states without automatically creating promises. Classification depends on context and conventions, and acts may combine functions. A statement of intention can also be assessed for truth or sincerity. For commitments, the paper asks who is bound, to whom, and who may release the obligation. Fault and remedy require further assessment. The discussion draws on Anscombe’s and Searle’s treatments of direction of fit and examines Leonard v. Pepsico, Inc. (1999). The court held that a commercial depicting a fighter jet made no contractual offer; its decision also rested on a separate writing requirement. The case illustrates act classification in a specific legal setting. Part of the Standards and Answerability series within On Human Modeling, the paper offers questions for distinguishing inaccurate reports, unfulfilled commitments, and revised intentions. Its framework can inform human–AI research on how interfaces distinguish predictions, plans, and commitments, including the authority and conditions supporting a claimed commitment.