Current AI-cognition research fundamentally misunderstands what it is measuring by studying individuals operating with constrained cognitive architectures engaging AI systems designed for multidimensional thinking. Neurotypical users, primarily accessing focused conscious cognition while their cerebellum, somatic intelligence, and subconscious resources remain largely inaccessible, experience cognitive decline when AI replaces their limited conscious processing. Neurodivergent individuals, operating with three cognitions—metacognitive (small picture), protocognitive (big picture), and infracognitive (blended whole picture)—use AI not as cognitive replacement but as multidimensional compression translation tool to interface with 2D institutional systems. This paper argues that documented "cognitive debt" reflects the collision between AI systems designed by creatives for their multidimensional cognition and users operating with artificially constrained cognitive access, while neurodivergent users demonstrate AI's actual potential as interface technology between higher-dimensional thinking and lower-dimensional institutional requirements.
The Cognitive Processing Reality: 95% Hidden Intelligence
Current neuroscience reveals a staggering disparity between conscious and unconscious cognitive processing that fundamentally undermines assumptions in AI-cognition research. Conscious processing handles only 5-10% of total brain activity, with some estimates suggesting we are explicitly aware of as little as 0.1% of moment-to-moment processing (Bargh & Chartrand, 1999; Dijksterhuis & Nordgren, 2006).
Meanwhile, unconscious processing manages 90-95% of cognitive activity, handling an estimated 11 million bits per second of sensory input while conscious awareness processes merely 40-60 bits per second (Zimmermann, 1989; Koch, 2004). This vast unconscious network encompasses:
Somatic intelligence: Continuous bodily sensing and environmental monitoring
Pattern recognition: Complex associative processing across multiple domains
Emotional indexing: Affective tagging of memory and experience
Predictive modeling: Anticipatory processing of future scenarios
Memory consolidation: Integration of experience across temporal dimensions
The implications are profound: current AI-cognition research studies individuals accessing perhaps 5-10% of their total cognitive capacity, then generalizes findings as representative of human-AI interaction. This is equivalent to studying automobile performance by examining only the dashboard instruments while ignoring the engine, transmission, and drive systems.
Somatic Memory and Embodied Intelligence
Recent research in embodied cognition reveals that the nervous system continuously records vastly more sensory and environmental data than reaches conscious awareness (Damasio, 2010; Barrett, 2017). This somatic memory system operates like a sophisticated environmental monitoring network, cataloguing:
Micro-expressions and non-verbal communication patterns
Energetic and spatial dynamics in social interactions
Temporal rhythms and environmental fluctuations
Peripheral sensory data filtered from conscious awareness
Proprioceptive and interoceptive information streams
Some individuals demonstrate enhanced access to this embodied intelligence through what can be described as retrospective somatic archaeology—the ability to consciously return to memory states and retrieve the fuller sensory record that was unconsciously catalogued but consciously overlooked (Craig, 2002; Mehling et al., 2011).
This process, analogous to Sherlock Holmes's method of mental reconstruction, involves using emotional states as retrieval cues to navigate embodied memory networks and access pattern fragments that the body detected before the conscious mind could articulate them.
Three Cognitions, Not One
The fundamental error in current AI-cognition research is treating all human users as operating with the same single "cognitive" architecture. The studies document what happens when individuals with constrained cognitive access interact with AI systems, but completely miss the experience of those operating with multiple cognitive modes in integration.
Constrained Architecture (Typical Institutional Users):
Primary access to focused conscious cognition (small picture - dissection only)
Limited cerebellar and subconscious integration
Reduced somatic intelligence availability
Minimal access to embodied memory networks
Creative spark underutilized or suppressed
Holistic orchestration inaccessible
Temporal processing locked in present-moment focus
Thinking operates in single dimensional, linear sequences
Operating with approximately 5-10% of total cognitive capacity
Keep reading with a 7-day free trial
Subscribe to The Gifted Neurodivergent Podcast to keep reading this post and get 7 days of free access to the full post archives.