AST-1: A predictive model representing and enabling control over the current state of attention

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• Attention is a generalization of convolution; all convolutions can be expressed as tensor products of fixed relative position attention patterns and weight matricesclaim0.717
  Mathematical equivalence showing the relationship between attention mechanisms and convolutional operations
• H1: Alignment training is attention training for models — Constitutional AI trains self-observation explicitly.hypothesis0.716
  Confirmatory hypothesis supported at p=0.006
• The Attention Schema Theory: A Foundation for Engineering Artificial Consciousness (Graziano, 2017)concept0.714
  Paper providing the biological framework analogy for ESR as a form of attentional control
• Scaled dot-product attention is the most faithful, structured, and theoretically grounded method for incorporating stimulus influence into response representations leading to an RN.claim0.712
  Justifies the methodological choice of attention over concatenation, mean pooling, residual connections, or joint embedding.
• A generalized model where Level 2 modulates all Level 1 precision parameters would capture mood, emotion, and broader phenomenology of mental states.hypothesis0.711
  Future work direction; extends current model beyond attentional precision to full space of emotional and metacognitive phenomena.
• Top-Down Attentional Controlconcept0.710
  Biological analogue to ESR where top-down mechanisms detect distracting inputs and redirect processing
• What predicts self-observation-like scores is training approach (alignment type), not model size or architecture.claim0.702
  Central interpretive claim from statistical analysis
• Perception-Action Model of Empathyframework0.700
  Neuroscientific model suggesting empathy is mediated by partial convergence of self and other representations