Trauma-focused psychotherapies seek to change how old learning shapes present responses. Imagery Rescripting (ImRs), hypnosis, and some interventions used during EMDR reprocessing can introduce protection, agency, information, support, or other new outcomes while a traumatic memory or representation is active. This article proposes a theoretical and clinical model centered on functionally relevant mismatch (FRM): a new outcome that is both unexpected relative to the activated learning and matched to the functional dimension organizing the traumatic response. The distinctive prediction is an interaction. At comparable levels of surprise, a functionally matched outcome should produce greater updating than an irrelevant surprise; at comparable levels of functional relevance, an unexpected outcome should produce greater updating than an expected correction. FRM is proposed as a testable construct, not as an established mechanism. Changes in beliefs, agency, distress, need satisfaction, or action tendencies are treated as candidate mediators, while new learning, inhibitory learning, retrieval processes, and other explanations remain competitors. Reconsolidation is examined as a candidate mechanistic hypothesis of particular interest, not as the empirically preferred explanation of clinical change. Three predictions test FRM, its relation to mediators, and indices that are compatible with - but do not identify - a contribution from reconsolidation. Hypnosis and EMDR are considered distinct clinical contexts in which reparative imagery can occur, without assuming that they share a single neurobiological mechanism.