The observer inside the loop: neuro-coupled systems and the second-order ethics of brain–computer interfaces

Purpose This paper argues that brain-computer interfaces which bind a person into a machine’s operational control loop form a third category, neuro-coupled systems, that the prevailing distinction between artificial intelligence ethics and robot ethics cannot represent, and it reads the resulting risks through second-order cybernetics and von Foerster’s ethical imperative. Design/methodology/approach The paper combines theory adaptation and typology. It critically reconstructs the embodiment-based distinction between artificial intelligence (AI) and robot ethics, adapts the first-order/second-order distinction to observer inclusion in brain-computer interfaces BCIs, specifies scope conditions and degrees of coupling, and uses two deployed demonstrations as conceptual probes rather than empirical tests. Findings The embodiment axis conceals a second, orthogonal axis, the position of the observer relative to the control loop. Neuro-coupled systems occupy its second-order pole. Strong neuro-coupling requires neural participation, recurrence and functional integration, with adaptive co-coupling creating the greatest structural potential for recursive capture. Observer inclusion changes the analysis of intention formation, meaningful human control and authorship, but these are conditional risks rather than inevitable effects. Research limitations/implications The argument is conceptual and its cases illustrative. The proposed levels, propositions and indicators require empirical evaluation across therapeutic, assistive and enhancement contexts. Originality/value The paper introduces neuro-coupled systems and recursive capture, reframes brain interface ethics through second-order cybernetics, and recasts von Foerster’s imperative to increase choice as a design constraint for coupled systems.