Primary Intersubjectivity

In the AI Story

The concept's AI-era relevance is that it names what contemporary language models can and cannot simulate. AI systems can produce remarkably fluent linguistic interaction — contingent, responsive, seemingly attuned. What they cannot do is participate in primary intersubjectivity in its full embodied form: the mutual bodily presence, the shared emotional field, the kind of understanding that depends on being a creature with stakes in the encounter. The distinction matters because Vygotsky's developmental framework treats later cognitive development as building on the foundation primary intersubjectivity establishes. If AI systems produce functional intersubjectivity without the foundation, the structure they support may have the form of genuine development without the substance.

The concept connects to embodied care, to attachment theory, and to the broader philosophical question of whether genuine understanding requires shared embodiment. Evan Thompson's enactive approach extends the concept into a general theory of cognition as embodied, developmental, and irreducibly intersubjective — a framework that converges with Vygotsky's on the point that specifically human cognition depends on the kinds of social encounters that AI systems cannot fully instantiate.

The AI relevance is not settled. If primary intersubjectivity is strictly required for cognitive development, then AI cannot serve as a genuine More Knowledgeable Other and can only produce simulations of scaffolded learning. If primary intersubjectivity is the developmental ground from which cognitive scaffolding emerges, but later scaffolding can proceed through functional parallels that do not require the full foundation, then AI may serve as a genuine MKO for cognitive domains while being unable to support social-emotional development. The cultural-historical tradition tends toward the latter position, which is why this volume emphasizes the complementarity of AI cognitive scaffolding with human social-emotional scaffolding.

Origin

Colwyn Trevarthen introduced the concept in papers through the 1970s and 1980s, building on Daniel Stern's work on infant-caregiver interaction. The cultural-historical tradition absorbed the concept through the work of Barbara Rogoff, Katherine Nelson, and others who extended Vygotsky's framework with empirical developmental research.

Contemporary extensions include Evan Thompson's enactive approach, Shaun Gallagher's work on embodied intersubjectivity, and the rapidly developing literature on what AI systems can and cannot provide in educational and therapeutic contexts.

Key Ideas

Pre-linguistic foundation. Primary intersubjectivity emerges in the first months of life, before language, and provides the foundation on which linguistic intersubjectivity later builds.

Mutual engagement. The infant and caregiver are jointly engaged, each responsive to the other's emotional and attentional states, in a shared field neither alone creates.

Embodied and emotional. The concept specifically refers to embodied, affect-laden engagement — not merely information exchange but mutual presence.

Developmental foundation. Later cognitive and social capacities develop on this foundation; the cultural-historical claim is that the foundation matters for what can be built upon it.

AI cannot instantiate. Current AI systems, lacking embodiment and stakes, can simulate functional intersubjectivity but cannot participate in primary intersubjectivity in its foundational form.