Memory as material cause: a semiotic theory of affective transmission across non-conscious nodes

Abstract Users report genuine affective responses when interacting with large language models that, by broad consensus, do not themselves possess affect. This is not in itself paradoxical: meaning emerges in the experiencing subject and requires no consciousness at the sign’s source. What is problematic is transmission, which presupposes that someone at the sending end feels something and intends to convey it – a presupposition the model invalidates while the recipient is still moved. When a sign-chain contains non-conscious nodes, what causal-structural continuity makes the recipient’s affect a constrained re-instantiation of an earlier closure rather than an independent elicitation? This paper distinguishes generation from transmission. Generation requires consciousness; transmission can traverse non-conscious nodes provided the propositional vehicle preserves structural features originating in the sending closure, and the recipient possesses non-propositional symbolic memory from which the affective object is drawn. Symbolic memory is classified by the representational format of its components, with degree zero – the minimal continuity through which retention persists – as formal precondition. A thought experiment shows such memory constitutive of genuine affective closure. Four conditions then distinguish transmission from elicitation and are tested against boundary cases. AI has created no new phenomenon; it has made visible a structure always already operative.