Cognitive Continuation After Interruption: A Causal-Role Grammar of Commitment, Resumption, and Protected Continuation State

What persists when a system stops processing a task but remains able, and in some sense committed, to return later? Research on interruption, prospective memory, unfinished intentions, and goal pursuit shows that recall, spontaneous resumption, affective pressure, and execution can dissociate. Yet a single notion of attention is still credited with current processing, content availability, goal maintenance, return, and control over action. This paper proposes a causal-role grammar of cognitive continuation. It separates active processing from preserved continuation state; task lifecycle from validity knowledge and validity in the world; retrieval from authorized resumption; alarm reduction from a valid closure ground; and the causal reach of a local control mode from its authorized transitions. Commitment is treated minimally as the continuing control-relevance of a valid open task, not as uninterrupted attention or affective pressure. The five distinctions largely synthesize established work. The residual proposal is a criterion for selecting protected continuation state: preserve inherited states whose loss would collapse outcomes requiring different grounds, such as lawfully closed versus lost or invalid versus unchecked. A three-level program tests a human off-target profile, declared violations in a guarded agent, and emergent surrogate closure in an unguarded agent. Within the guarded branch, a collapse-discriminability test measures whether removing candidate protected states destroys the later configuration's ability to distinguish ground-dependent outcomes. The result is a testable grammar for preserving distinguishable lawful routes through interruption without substituting one causal function for another. Preprint version 1.0. Submitted to New Ideas in Psychology on 13 August 2026.