Mechanisms underlying mixed effects of interaction on collective performance

Determine the underlying mechanisms that drive the mixed effects of interaction on collective outcomes, including circumstances in which interaction improves or impairs group performance relative to independent aggregation.

Background

The introduction contrasts findings showing that interaction can enhance group performance with findings showing that interaction can suppress the benefits of independent exploration and reduce collective outcomes. Independent aggregation can also outperform interacting groups in some settings, but the mechanisms responsible for these divergent effects are not established. The paper presents dynamical modeling as one approach to investigating these mechanisms, although the broader question concerning mixed empirical observations remains open.

References

There are open questions about the underlying mechanisms driving these mixed observations.

— Collective Benefits and Efficiency in Coupled Echo State Networks  (2610.17506 - Bruna et al., 13 Oct 2026) in Section 1, Introduction

Performance appears to begin to asymptote at 16 ESNs, although it is unclear how many ESNs can be added while performance gains continue to accrue.

— Collective Benefits and Efficiency in Coupled Echo State Networks  (2610.17506 - Bruna et al., 13 Oct 2026) in Section 6.1, Experiment 2: Efficiency of Group Interaction, subsection “Dyadic Advantage Extends to Larger Ensembles”