agent definitions — complex systems
on this page
overview
complex-systems scholarship treats agents as rule-based components whose local interactions yield emergent macro patterns. agent-based modeling (abm), adaptive systems theory, and thermodynamic approaches all appear in this cluster.
signature traits
- local rules, global effects: agents follow simple rules; collective behavior emerges from interaction networks (holland 1995, epstein & axtell 1996).
- adaptation: agents evolve strategies via learning, reproduction, or selection.
- environment coupling: agents are situated in explicit environments with resource flows or state variables.
illustrative definitions
- 1995 — john holland, hidden order: complex adaptive systems comprise many interacting agents that learn and co-evolve.
- 1996 — epstein & axtell, growing artificial societies: defines abm agents as autonomous heterogeneous individuals driving emergent social dynamics.
- 2000 — stuart kauffman, investigations: describes autonomous agents capable of reproduction and thermodynamic work cycles.
relation to other dimensions
- autonomy spectrum: varies from rule-following automata (moderate autonomy) to self-maintaining systems (high autonomy).
- entity frames: straddle hybrid and machine frames; agents may represent people, firms, or synthetic organisms.
- goal dynamics: often focuses on adaptation rather than explicit negotiation; objectives emerge through evolution or feedback.
- persistence & embodiment: many models assume persistent agents; some, like kauffman, insist on physical embodiment and metabolism.
open questions
- how can insights from abm about emergent coordination inform multi-agent llm orchestration?
- do thermodynamic definitions belong in the same comparative space as purely digital abm agents?
- what metrics best capture when rule-based agents transition from delegate behavior to self-maintaining systems?