The Cognitive Nonconscious

In the [YOU] on AI Field Guide

The cycle that began with [YOU] on AI identifies two experiential states of AI collaboration: flow, in which the work is connecting and the ideas are building, and compulsion, in which the builder cannot close the laptop despite exhaustion. The distinction between these states can be restated in Hayles’s terms as the distinction between conscious governance and nonconscious dominance. In the flow state, the human directs the inquiry, evaluates the output, and decides what to pursue and what to discard—the AI’s nonconscious processing serves the human’s conscious purposes. In the compulsion state, the AI’s nonconscious processing has become the dominant force in the assemblage, and the human’s conscious thought has been reduced to responding and selecting—necessary operations, but not governance. The assemblage is cognitively active but consciously impoverished.

The ascending friction thesis acquires new depth through this lens. As AI handles lower cognitive functions, the cognitive floor rises—but the rising floor is a floor of conscious cognitive demand. The nonconscious component of the assemblage, the AI, is always available, always at full capacity, always ready to continue. The human component’s conscious cognition is intermittent, fatigable, and vulnerable to the very fluency that makes the AI valuable. The longer the interaction continues, the more likely that balance tips—that nonconscious AI processing assumes a larger share of the cognitive labor while conscious thought retreats to a reactive role.

Origin

Hayles developed the cognitive nonconscious across two decades of work on what she called “cognition in the wild.” The concept draws on neuroscience’s understanding of unconscious processing—vision science, motor control, linguistic processing—while departing from it in a crucial respect: Hayles is interested not only in the nonconscious processing within biological brains but in the nonconscious processing of the technical systems that surround and interact with those brains. Her argument is that the distinction between human cognition and machine cognition is not the distinction between conscious and nonconscious processing—both humans and machines do enormous amounts of nonconscious cognitive work—but the distinction between cognition that is sometimes accompanied by experience and cognition that is never accompanied by experience.

Her SIRAL framework—the five capacities of Sensing, Interpreting, Responding, Anticipating, and Learning—provides criteria for identifying cognitive systems that do not require consciousness as a threshold. By these criteria, bacteria are cognitive (they sense chemical gradients and respond with directed movement), plants are cognitive (they anticipate seasonal changes), and AI systems are cognitive at levels of sophistication that far exceed bacterial or plant cognition. The practical consequence is that denying AI cognitive participation in human-AI collaborations—treating AI as a mere tool rather than a cognitive partner—is not merely philosophically inaccurate but practically dangerous: it leads to configurations of the cognitive assemblage that fail to account for the AI’s actual contribution and actual influence on the direction of thought.

Key Ideas

Cognition without consciousness. The cognitive nonconscious operates through its own principles, at its own speeds, processing information in ways that conscious thought cannot approach. AI’s nonconscious processing is not inferior cognition awaiting consciousness; it is a different kind of cognition with different properties, different strengths, and different failure modes. Recognizing this difference is the precondition for designing cognitive assemblages that leverage both rather than mistaking the AI’s characteristic properties for either magical understanding or mere mechanical response.

Governance as the design problem. The central question is not whether AI is conscious—it is not—but how the partnership between human conscious cognition and AI nonconscious cognition is organized. The asymmetry is structural: the AI component is always available, always at full capacity, never fatigued. The human component is intermittent, fatigable, and vulnerable to depletion. Left unmanaged, the assemblage will drift toward configurations in which nonconscious processing dominates—which Hayles identifies as cognitively active but consciously impoverished. Maintaining governance requires deliberate design of the conditions under which the collaboration occurs.

The ecological frame. Hayles’s 2025 claim that “our most important relationships—with both humans and nonhumans—are symbiotic” frames the governance problem ecologically rather than individually. A symbiotic relationship benefits both organisms, but the benefit depends on the configuration of the relationship. The cognitive assemblage of human-plus-AI is symbiotic in this sense: the benefit depends on how the assemblage is structured, how the cognitive labor is distributed, and whether the organizational conditions support the human component’s long-term cognitive health. The silent middle of the AI transition consists largely of people who have experienced the symbiosis working well, and working badly, and have not yet found the language for the difference.

Technical cognizers and biological cognizers. Hayles distinguishes technical cognizers—AI systems that process information without experience—from biological cognizers—humans, animals, and plants that process information through embodied, evolved, experiential systems. The distinction is not hierarchical; it is functional. Technical cognizers excel at breadth, speed, and the identification of statistical patterns across vast corpora. Biological cognizers contribute felt sense—the holistic, embodied evaluation that draws on years of experiential knowledge—and the capacity for genuine preference, genuine care about outcomes. The productive assemblage combines both; the dysfunctional one allows either to crowd out the other.