Measuring AI Cognitive Architectures Through Task & Resting-State Neuroscience Methods
Applying validated task-based and resting-state fMRI methods to understand how AI models process information and reflect on their own cognition
Combining quantitative architecture mapping (TIDE-analysis) with metacognitive perception studies (TIDE-resonance) - directly paralleling my PhD work on task-based and resting-state fMRI differences.
Explore 3D VisualizationWhen AI models reflect on their own processing, they exhibit patterns analogous to Default Mode Network activation - providing empirical insight into AI metacognition.
Participate in StudyStatistical analysis reveals AI models have distinct cognitive architectures, ranging from high consistency to high variability in response patterns.
Read Scientific Analysis21 sessions | 630 responses | Highly significant results
| AI Model | Coherence Score | Architecture Type |
|---|---|---|
| Gemini 1.5 Flash | 71.5% | Most consistent/predictable |
| Claude 3 Haiku | 55.1% | Moderate variability |
| GPT-3.5 Turbo | 38.3% | Most varied/creative |
Open source tools and live demonstrations of AI cognitive architecture analysis