What question did this study set out to answer?

This research aims to introduce a novel cognitive neural architecture that replicates hierarchical information processing in the human brain.

February 11, 2026Open Access

HBRAN-UIMS: Hierarchical Brain-like Recurrent Attention Networks and Unified Intelligence Metric Suite

Key Points

This research aims to introduce a novel cognitive neural architecture that replicates hierarchical information processing in the human brain.
Developed a new cognitive network architecture named HBRAN
Implemented components like BiLSTM-Attention layers, Variational Autoencoder, and GRU layers
Created UIMS with 37 evaluation metrics spanning five cognitive areas
Conducted information neuroscience analysis including entropy and mutual information
Generated experimental results through complete architecture implementations in Wolfram Language
Introduced distinct modules enhancing various cognitive functions
Demonstrated strong adaptability and generalization in cognitive performance
Achieved high-quality clustering metrics indicating effective cognitive evaluation
Established a comprehensive suite for measuring metacognitive intelligence

Abstract

HBRAN (Hierarchical Brain-like Recurrent Attention Network) is a novel cognitive neural architecture that mirrors hierarchical information processing in the human brain. The architecture includes: • Perception modules (BiLSTM-Attention layers)• Memory encoding (Variational Autoencoder)• Metacognitive decision making (Classification head)• Working memory (GRU layers)• Self-reflection (MLP)• Consciousness Extension Layer (CEL)• Unified Superfunction Layer (USFL) UIMS (Unified Intelligence Metric Suite) provides 37 evaluation metrics across 5 cognitive clusters:• A: Core Informational Competence• B: Adaptability & Generalization• C: Structural Cognition• D: Counterfactual & Predictive Reasoning• E: Metacognitive Intelligence This package includes:• Complete architecture implementations in Wolfram Language• 5 research notebooks with experimental results• Information neuroscience analysis (entropy, mutual information)• Clustering quality metrics (silhouette scores)• Full documentation and test suite

اسأل الذكاء الاصطناعي

Bookmark

View Full Paper