Architecture of Intelligence

Stephen Feber FRSA · 2026 · Twelve interactive graphics for the web publication and RSA lecture. Intelligence is an evolved capacity for ecological coupling, not a scalar property.

Theoretical framework
1The Intelligence Spectrum
Five Thresholds as expanding ecological range. Fourteen-dimensional profiles of HI, AI, and AuHI. The non-scalar argument visualised.
2Frame of Intention: Reaching for Coffee
The four-phase coffee cup reach as multi-system active inference. Eleven body systems, precision-weighting, micro threshold crossings within a single intelligent act.
3The Frame as a Landscape
The free-energy landscape beneath the frame of intention. The apex prior deforms the surface and the act descends into a basin; deep priors are canalised valleys, shallow priors revisable, the moving frame a moving threshold.
4Two Time Bases: Tying a Shoelace
A single tied bow on two timescales: five-to-seven years of developmental assembly through organism–environment coupling, and a twenty-second act. Diachronic and synchronic on one view, with a coda setting active inference against object-relations reification.
The network problem
5Network States
3D interactive network. Three clusters with adjustable flow. Cross-cluster flow is the network condition.
6The Lombard Effect
Acoustic analogy for DVAP. The feedback loop of diminishing intelligibility. Why RSA events feel like networking but produce no network.
7Density–Value–Accessibility Paradox
As connection density increases, theoretical value rises but accessibility falls. The peak of usable value and the collapse beyond it.
The hybrid architecture
8Four Network Conditions
Two failure modes (connection without signal, semantic overload) and the architecture that resolves them. Mesh, profiles, heuristic enquiry, system learns.
9Fellow Dashboard
Five heuristic sliders shaping GNN recommendations in real time. The interface between human intention and machine intelligence.
10Cognitive Signature
Gardner's eight intelligences as a multimodal profile. Fellow A vs Fellow B with computed semantic overlap. Complementarity as important as similarity.
11Cluster Growth
Adding a Fellow does not add to the cluster. It reforms it. Watch the network flex and settle into a new equilibrium.
12Network Emergence
Force-directed graph with Fellow nodes, prune slider, BFS path finder, and topic labels. The Fellowship as a living, navigable network.