Level of Consciousness Signatures Across Biological and Artificial Minds: A Unified Framework for Measuring Cognition in Human EEG and Large Language Models

Jamaludheen K N · AIME Research · 2026

Abstract

We present the first direct comparison of cognitive coherence between human brains and large language models using the AIME LOC framework. Using the same 13-function cognitive model applied to both EEG frequency power and transformer layer activations, we conducted four validation studies across 5 model architectures (12B–70B parameters) and 21 EEG subjects.

Study 1 demonstrated that 12 of 13 cognitive functions are detectable at the individual token level (p < 0.001). Study 2 achieved sentence-level classification across all architectures. Study 3 detected cognitive phase transitions at sentence boundaries. Study 4 provided zero-confound within-sentence validation. Cross-substrate analysis revealed that 7 of 13 functions show same-direction effects in both biological and artificial networks, with causal validation achieving 6.5× chance detection via layer isolation.

Key Findings

12/13 cognitive functions significant at token level in Qwen-35B (p < 0.001)
10–12/13 functions significant per model across all 5 architectures tested
7/13 functions show convergent effects between EEG and LLM substrates
Causal proof: layer isolation achieves 6.5× chance detection
Model size scales with True Coherence: Llama-70B (15.4%) vs Gemma-12B (7.4%)
Knowledge distillation reduces TC by 1.17 percentage points (−11%)

Figures

Effect size heatmap across 13 cognitive functions — Figure 1. Cross-network effect sizes for 13 cognitive functions across EEG subjects and LLM models.

Summary of all 4 validation experiments — Figure 4. Summary of all four validation experiments.

Cross-substrate scatter plot — Figure 5. Human vs AI cognitive profile comparison.

Citation

@article{jamaludheen2026loc,
  title={Level of Consciousness Signatures Across
         Biological and Artificial Minds},
  author={Jamaludheen, K N},
  journal={AIME Research},
  year={2026},
  url={https://aimindengine.com/research/paper-1}
}