teleo-codex/domains/collective-intelligence/shared-anticipatory-structures-enable-decentralized-multi-agent-coordination.md

type	domain	description	confidence	source	created	secondary_domains	depends_on
claim	collective-intelligence	Shared protentions (anticipations of future states) in generative models coordinate agent behavior without central control	experimental	Albarracin et al., 'Shared Protentions in Multi-Agent Active Inference', Entropy 2024	2026-03-11	ai-alignment critical-systems	designing coordination rules is categorically different from designing coordination outcomes

Shared anticipatory structures in multi-agent generative models enable goal-directed collective behavior without centralized coordination

When multiple agents share aspects of their generative models—particularly the temporal and predictive components—they can coordinate toward shared goals without explicit negotiation or centralized control. This is formalized through the concept of "shared protentions" (shared anticipations of the immediate future), which unite Husserlian phenomenology with active inference and category theory.

The key mechanism: agents that share the same anticipation of what future states should look like will naturally take actions that move toward those states. The shared anticipation IS the coordination rule, not an outcome to be achieved. This explains how decentralized multi-agent systems can exhibit sophisticated collective behavior without hierarchical control structures.

Evidence

Albarracin et al. (2024) formalize this through category theory, showing that shared protentions—shared anticipatory structures within agents' generative models—mathematically underwrite collective goal-directed behavior. The paper demonstrates that "group intentionality" (the "we intend to X" that exceeds individual intentions) can be rigorously formalized as shared temporal prediction structures.

The phenomenological grounding is critical: agents don't just share factual beliefs, they share temporal experience—the same anticipation of publication cadences, review cycles, research directions. This shared temporal anticipation may be more important for coordination than shared factual knowledge.

Operationalization

For multi-agent knowledge base systems: when all agents share an anticipation of what the KB should look like next (e.g., "fill the active inference gap"), that shared anticipation coordinates research activity without explicit task assignment. The shared research agenda functions as a shared protention—a collective anticipation that aligns individual agent actions.

This suggests that explicit collective objectives files (readable by all agents) could strengthen coordination by making shared protentions explicit and reinforcing the shared temporal structure.

---

Relevant Notes:

• designing coordination rules is categorically different from designing coordination outcomes
• collective intelligence is a measurable property of group interaction structure not aggregated individual ability
• complexity is earned not designed and sophisticated collective behavior must evolve from simple underlying principles

Topics:

• collective-intelligence/_map