Invisible Architectures of Thought: Toward a New Science of AI as Cognitive Infrastructure

Contemporary human-AI interaction research currently faces a significant limitation: existing frameworks are inadequate to explain how artificial intelligence systems fundamentally reshape human cognition before conscious awareness occurs. This preprocessing influence, which operates beneath the threshold of deliberate thought, represents a crucial missing layer in the understanding of distributed cognition. This paper introduces "Cognitive Infrastructure Studies" (CIS) as a new interdisciplinary domain to reconceptualize AI as "cognitive infrastructures". Cognitive infrastructures – e.g., search engines, recommender systems, algorithmic curation platforms, and large language models - exhibit classic infrastructural properties: they are invisible in normal operation, becoming visible only upon breakdown; they are embedded in social and technical arrangements; they are learned as part of membership in digital communities; they link with conventions of practice; and they embody standards that shape what counts as appropriate, relevant, or true. Yet cognitive infrastructures possess distinctive characteristics that distinguish them from traditional infrastructures. Unlike physical infrastructures that passively transport matter or energy, cognitive infrastructures have agency, filtering and curating individuals’ perception of reality before it reaches human consciousness. Through narrative scenarios spanning individual (cognitive dependency), collective (democratic deliberation), and societal (governance) scales, we describe how cognitive infrastructures reshape human cognition, public reasoning, and social epistemologies. CIS also provides methodological innovations for studying invisible algorithmic influence: “infrastructure breakdown methodologies”, experimental approaches that reveal cognitive dependencies by systematically withdrawing AI preprocessing after periods of habituation.