Dual-Process Computation
The Intention Space framework draws a structural parallel between Kahneman's dual-process theory of cognition and two fundamental modes of computation.
System 1 and System 2 in Computing
Kahneman distinguishes between fast, automatic System 1 thinking and slow, deliberate System 2 reasoning. We propose an analogous distinction in computational systems:
| Cognitive Mode | Computational Analogue | Intention Space Role |
|---|---|---|
| System 1 (Fast) | Pattern matching, caching, reactive pipelines | Perceptual Computation — pulse monitoring and signal propagation |
| System 2 (Slow) | Algorithmic reasoning, complex transformations | Generative Computation — Design Node processing |
The Key Insight
Traditional software conflates these two modes. A single function might both perceive a state change and generate a response, making it impossible to reason about which mode is active. Intention Space separates them structurally.
Implications for AI Interpretability
This separation has direct implications for understanding transformer attention mechanisms. When we can distinguish between perceptual and generative attention heads, we gain a principled framework for mechanistic interpretability.