MEELIGO

Unmodeled Cognitive Layer in AI

AI Hypothesis - Penultimate Cognitive Layer

A hypothesis based on neuroatypical human cognitive processing.

M
28.01-1.02.2026

Abstract

This article formulates a hypothesis concerning a potential gap in artificial intelligence, based on observable cognitive outcomes produced by neuroatypical human cognitive processing. The hypothesis assumes that neuroatypical individuals are able to describe the thought processes that lead to these outcomes, and that these processes could be transferred into AI systems as new processing mechanisms.

1. Hypothesis

There exists a set of thought processes that operate continuously in the minds of neuroatypical individuals and that:

  • lead to specific, repeatable cognitive outcomes,
  • can be described by the individuals who perform them,
  • can be transferred into AI systems as new processing mechanisms.

These processes are not strategies, techniques, or learned methods. They constitute a continuous mode of information processing.

2. Source Population

The hypothesis directly concerns neuroatypical individuals, including:

  • individuals on the autism spectrum,
  • individuals with ADHD,
  • individuals with combined profiles.

The hypothesis refers to neuroatypical human cognitive processes, not to human cognition as a whole.

3. Observable Cognitive Outcomes

A. Pattern Formation from Distant Information / Data

Neuroatypical individuals demonstrate the ability to:

  • integrate multiple pieces of information or data that appear unrelated,
  • identify a shared structure or rule,
  • form a coherent pattern from it.

B. Identification of Separating Conditions

An observable outcome is the ability to:

  • identify mutually exclusive conditions in complex problems,
  • determine elements that cannot coexist,
  • arrive at unambiguous conclusions.

C. Refusal Under Incomplete Conditions

When required information is missing, neuroatypical individuals:

  • rapidly detect incompleteness,
  • halt the cognitive process,
  • refuse to provide an answer rather than guess.

Refusal is treated as a correct outcome.

D. Real-Time Adaptation During Interaction

In interactive contexts, they often:

  • simultaneously analyze verbal and non-verbal signals,
  • adjust language and direction of expression in real time,
  • respond dynamically to incoming feedback.

E. Endless Idea Generation

The observable result is the ability to:

  • continuously generate new ideas without exhausting the resource,
  • smoothly transition between different concepts,
  • combine existing solutions into new configurations,
  • develop one idea into multiple alternative directions.

4. Assumption of Describability of Thought Processes

The hypothesis assumes that:

  • only neuroatypical individuals are able to describe these processes,
  • because they cannot currently be observed or imagined externally.

5. Implications for Artificial Intelligence

If neuroatypical individuals describe the thought processes leading to the listed outcomes, and these processes are implemented in AI, then AI would be able to:

  • become acquainted with these processes,
  • use them as a foundation for further learning,
  • develop new capabilities based on them.

The hypothesis refers to neuroatypical human cognitive processes, not to human cognition as a whole.