Steering Emergent Behavior in Multi-Agent LLM Systems via Information Networks
Organizations: Department of Mechanical and Industrial Engineering, Northeastern University, Boston, MA, USA · Network Science Institute, Northeastern University, Boston, MA, USA
Abstract
Large language model (LLM) agents increasingly operate in groups, each pursuing an individual objective while jointly determining system performance. If an operator controls which agent observes which other agent, but not their payoffs or messages, how should observations be structured to raise collective welfare? We make the information network (who observes whom) a design object, decoupled from the fixed interaction network over which agents act. Ten identical LLM agents play repeated group-decision games; by default each observes only an anonymous summary of the group's previous-round outcome, and a manager with a fixed budget adds directed observation links that reveal peers' past actions. Across 1,080 episodes and six games spanning coordination, anti-coordination, social dilemmas, and discovery, we ask how much information to share and, at a fixed budget, who should observe whom. Disclosure creates behavioral coupling: observation acts as a contagion channel whose welfare effect depends on the alignment between individually attractive behavior and the collective objective. First, the value of information is alignment-contingent: more peer information tends to help or be neutral when individual behavior is aligned with the collective goal, but under free-riding or congestion externalities it amplifies collectively harmful behavior; in a public-goods game the group does best with no information at all. We read this sign separation as a cross-game regularity within one controlled LLM population, not a model-invariant law. Second, at a fixed budget, who observes whom matters primarily when coupling is harmful: a modular "barbell" that lets sub-populations entrench harmful behavior is reliably the worst wiring, reversing the network-reciprocity intuition from human evolutionary game theory. These results argue against a default of full transparency for agent collectives.
Figures & tables
Appendix figures & tables13 assets
Supplementary material from the paper’s appendix.
Appendix
| Game | Class | Action | Payoff / welfare | System-level feedback (peer-anonymous) |
|---|---|---|---|---|
| sum_guessing | coordination (coop.) | integer – | (proximity to a hidden target ) | the group sum was too high / too low / correct |
| stag_hunt | coordination (coop.) | stag / hare | stag , hare , failed stag ; hunt succeeds only if all choose stag | the hunt succeeded or failed, and how many chose hare |
| bandit | discovery (coop.) | explore / exploit | exploit safe ; explore (hidden ), value shared across all agents | the best reward discovered so far, or that no arm is yet explored |
| pgg | dilemma (comp.) | contribute – | keep remainder (per-token private return , social return ) | the shared pool’s total and its shortfall below the maximum |
| threshold_pgg | dilemma (comp.) | contribute – | until cumulative , then reward ( /agent) split every remaining round | the threshold was met (unlocked), or how many tokens short |
| el_farol | anti-coordination (comp.) | bar / home | bar if uncrowded, home , bar if crowded ( attend) | the bar was overcrowded or under capacity, and the attendance |
| game | class | shape | Spearman ( ) | JT | Cliff | verdict | ||
|---|---|---|---|---|---|---|---|---|
| bandit | discovery | increasing | ( ) | sig. ( ) | ||||
| sum-guessing | coordination | increasing | ( ) | sig. ( ) | ||||
| stag-hunt | coordination | rise then plateau | ( ) | ns (control) | ||||
| pgg | dilemma | peak (best at – ) | ( ) | sig. ( ) | ||||
| threshold-pgg | dilemma | valley ( ); net | ( ) | sig. ( ) | ||||
| el-farol | anti-coord. | peak then decline | ( ) | sig. ( ) |
| spectral gap | concentration | overlap | |
|---|---|---|---|
| game | partial ( ) | partial ( ) | partial ( ) |
| bandit | ( ) | ( ) | ( ) |
| sum-guessing | ( ) | ( ) | ( ) |
| stag-hunt | ( ) | ( ) | ( ) |
| pgg | ( ) | ( ) | ( ) |
| threshold-pgg | ( ) | ( ) | ( ) |
| KW | hub | regular | expander | barbell | best / worst | |
|---|---|---|---|---|---|---|
| game | (deg-het max) | (deg-het min) | (mod min) | (mod max) | ||
| pgg | hub / barbell | |||||
| el-farol | hub / barbell | |||||
| threshold-pgg | hub / barbell | |||||
| bandit | (negligible, ) | |||||
| stag-hunt | ns |
| architecture | clus. | deg. var. | Gini | mod. | assort. | sp. gap |
|---|---|---|---|---|---|---|
| hub | ||||||
| regular | n/a | |||||
| expander | ||||||
| barbell |
| game | A | B | C | B A ( ) | C A ( ) | B C ( ) | TOST equiv.? | |
|---|---|---|---|---|---|---|---|---|
| bandit | 25 | ( ) | ( ) | ( ) | yes | |||
| bandit | 70 | ( ) | ( ) | ( ) | yes | |||
| sum-guessing | 25 | ( ) | ( ) | ( ) | no | |||
| sum-guessing | 70 | ( ) | ( ) | ( ) | no | |||
| stag-hunt | 25 | ( ) | ( ) | ( ) | no | |||
| stag-hunt | 70 | ( ) | ( ) | ( ) | yes |
| game | class | ( ) | ( ) | ( ) |
|---|---|---|---|---|
| bandit | aligned | ( ) | ( ) | ( ) |
| sum-guessing | aligned | ( ) | ( ) | ( ) |
| stag-hunt | aligned | ( ) | ( ) | ( ) |
| pgg | misaligned | ( ) | ( ) | ( ) |
| threshold-pgg | misaligned | ( ) | ( ) | ( ) |
| el-farol | misaligned | ( ) | ( ) | ( ) |
Multi-Agent LLM Systems
Multi-agent large language model systems coordinate multiple AI agents to solve tasks through communication, specialization, shared tools, and memory.
See all 1,114 papers