DarwinSys

Eight Short Research Notes

An English guide to the questions behind eight research pieces.

English reading edition · November 9, 2026

Two global workspaces

One workspace acts; the other scrutinizes. The theory asks whether their continuing revisions to each other’s memory and predictions could sustain an ongoing stream of cognition.

Carbon–silicon symbiosis

The proposed substrate transitions preserve earlier levels while adding a new medium for processing information. The paper asks whether human adaptation limits the rate at which coupled human and silicon systems can grow.

Why two units?

Evolutionary, neural-network and game-theoretic models compare the performance, stability and cost of one, two and three units. Their results are model-dependent comparisons.

The cost of replicas

Differences in initial state, memory decay and communication can produce divergence under shared input. The replica-cost paper asks how many samples are needed to cover that divergence as the number of units increases.

A finite-time limit

If successive transition intervals shrink geometrically, infinitely many intervals have a finite sum. Whether technological transitions actually sustain such a ratio is an empirical question.

Routing experts

The routing note asks whether a low-rank representation can reduce the scheduling cost of a mixture of experts. Its proposed logarithmic dependence on the expert count rests on the structure assumed for the router.

Coordination across cells

The note uses bioelectric communication to explore how local actions can support a coordinated body plan. It draws an analogy with the tension between local and collective objectives in a modular AI system.

The vanished third eye

The parietal-eye study examines how opposing spectral signals can encode a timing cue. It asks how far comparison between opposing inputs can explain other sensory and cognitive circuits.

Research library