PaperlayerView exact revision on arXiv

Source-linked AI summary

Spreading Dynamics on Higher-Order Networks: A Structure-Mechanism-Behavior Perspective with Applications

Ming Xie, Shibo He, Qihao Huang, Yuzhen Fan, Jiming Chen

arXiv:2610.39297v1cs.SIphysics.app-ph

TL;DR

Many real-world contagion processes involve group interactions, while research on higher-order spreading remains fragmented across models and representations. This review organizes higher-order structures, contagion mechanisms, and emergent behaviors in a shared perspective. It connects structure–mechanism combinations to phenomena including multistability, discontinuous or hybrid transitions, and propagation inertia.

  • Problem

    Many contagion processes involve group interactions, but higher-order spreading research remains fragmented and lacks a mechanism-driven organization across interaction structures and modeling frameworks.

    Source: §1 · pp. 1, 3

  • Method

    The review compares higher-order representations, classifies contagion mechanisms, and maps their structural carrying capacity to emergent behaviors.

    Source: §1 · p. 3

  • Results

    The review connects higher-order structure–mechanism combinations with behaviors including multistability, discontinuous or hybrid transitions, and propagation inertia.

    Source: §7 · p. 17

  • Takeaways & Limitations

    The structure–mechanism–behavior mapping offers practical guidance for designing maximization and intervention strategies.

    Source: §7 · p. 17

    A systematic account of representation minimality and equivalence, mesoscopic structure, and reconstruction from incomplete empirical data remains needed.

    Source: §6.1 · p. 16

Abstract

from arXiv · show

Spreading in real-world networks, including modern communication and networked systems, often unfolds through group interactions rather than isolated pairwise contacts. Higher-order network representations such as simplicial complexes, hypergraphs, and their temporal, multilayer, and directed extensions provide a principled way to encode group interactions. Unfortunately, existing studies on spreading over these substrates remain fragmented and largely model-centric, offering limited insight into how higher-order representations and contagion mechanisms jointly shape emergent dynamics. This review addresses this gap by introducing a structure-mechanism-behavior three-level perspective that provides a systematic mapping from higher-order network structures and spreading mechanisms to collective behaviors. Within this framework, we compare how different higher-order network representations encode group interactions, introduce a taxonomy of contagion mechanisms, and analyze their structural carrying capacity, defined as the ability of a given representation to support specific contagion mechanisms and the associated dynamical behaviors. We further map specific structure-mechanism combinations to three hallmark emergent behaviors, thereby offering a unified cross-level perspective beyond existing model-centric surveys. These insights are particularly relevant for communication systems involving group interactions, such as social information spreading, cooperative protocols, and distributed networked processes. Lastly, leveraging this three-level perspective, we offer a roadmap for guiding and controlling dynamics on higher-order networks, while discussing future directions in representation choice, mechanism design, and data-driven inference.

1 INTRODUCTION

The introduction motivates higher-order network models by noting that many real-world contagion processes unfold through group interactions, which pairwise models cannot represent directly. It frames the review as a response to fragmented, model-centric work, organizing structures, mechanisms, and emergent behaviors in a shared perspective.

1 INTRODUCTION

  • Group interactions, rather than isolated pairwise contacts, are the effective unit of contagion in many social, information-sharing, biochemical, and animal systems.Examples include group chats, family or workplace decisions, joint molecular reactions, and multi-body coordination.Source: §1 · p. 1
  • Pairwise graph models have limited ability to capture group-driven spreading, motivating representations such as simplicial complexes and hypergraphs.These treat interactions among three or more nodes as fundamental building blocks rather than reconstructing them from edges.Source: §1 · p. 1
  • Higher-order spreading studies report multistability, discontinuous or hybrid phase transitions, and propagation inertia, behaviors absent or more constrained in purely pairwise networks.These behaviors highlight sensitivity to initial conditions, abrupt collective responses, and temporal or memory effects.Source: §1 · p. 2
  • Existing reviews and studies remain fragmented and model-centric, without systematically disentangling structures, contagion mechanisms, and their effects on emergent behavior.The introduction identifies mechanism-driven organization across interaction structures and modeling frameworks as an open question.Source: §1 · pp. 2–3
  • A systematic structure–mechanism–behavior mapping is presented as relevant to data-driven inference, system-risk anticipation, and intervention design.The introduction names infection kernels and thresholds as inference targets.Source: §1 · p. 3
  • The review maps higher-order representations and contagion mechanisms to emergent behaviors, then uses that perspective to guide applications and control.Its scope includes simplicial complexes, hypergraphs, temporal and multiplex networks; mechanisms include reinforcement, inhibition, memory-aware, and threshold-activated processes.Source: Figure 1 · p. 2, §1 · p. 3

2 DESCRIPTIONS OF STRUCTURES

The section connects higher-order representations to the group interactions they encode and to the contagion mechanisms they can express. It also establishes structural, modeling, and dynamical concepts used to relate network structure to spreading behavior.

2 DESCRIPTIONS OF STRUCTURES

  • The section’s role is to provide structural and conceptual vocabulary for tracing how higher-order substrates and contagion mechanisms relate to phase diagrams, multistability, and propagation delays.The reviewed structures include simplicial complexes, hypergraphs, and temporal and multiplex higher-order networks.Source: §2 · p. 3, §2.5 · p. 6

2.1 Simplicial complexes as geometric carriers of group interactions

  • Simplicial complexes encode a group as a simplex and require every non-empty subgroup of that simplex to be included.This closure property represents groups embedded in fully connected local neighborhoods, which can support reinforcement and threshold effects.Source: §2.1 · pp. 3–4

2.2 Hypergraphs as flexible models of group interactions

  • Hypergraphs represent groups as hyperedges without requiring their members’ subgroups to interact, allowing group exposure without an underlying dense pairwise network.Uniform hypergraphs isolate a single interaction size, while hyperedge overlap, higher-order components, and cores describe mesoscopic organization linked to spreading outcomes.Source: §2.2 · pp. 4–5

2.3 Adding extra features to higher-order structures

  • Temporal, multilayer, and directional features enrich higher-order structures to represent time-ordered groups, coupled interaction contexts, and asymmetric or weighted influence.Temporal order constrains contagion to time-respecting paths; multilayer structures couple processes across contexts, while signed interactions can distinguish facilitating from inhibiting influence.Source: §2.3 · p. 5

2.4 Structural abstractions in higher-order contagion models

  • Model construction depends on whether states belong to nodes or groups and whether interactions remain explicit or are projected onto pairwise links.Explicit representations preserve group sizes and configurations, whereas projections are simpler but may suppress or distort group-dependent discontinuous and multistable behavior; the review matches mechanisms to suitable structures.Source: §2.4 · p. 6

2.5 Dynamical observables and basic notions

  • The section establishes prevalence, incidence, outbreak size, transmission parameters, equilibria, and phase transitions as concepts for describing spreading dynamics.Prevalence measures the infected or active fraction at time t, while discontinuous transitions involve a finite order-parameter jump and often bistability and hysteresis.Source: §2.5 · p. 6

3 MECHANISMS OF SPREADING DYNAMICS

This section organizes higher-order spreading mechanisms by how they use structural features and additional dynamical dimensions. It presents four mechanism families and compares which network representations support them and where modeling evidence remains limited.

3 MECHANISMS OF SPREADING DYNAMICS

  • The reviewed mechanisms use group structures directly or add temporal correlations, multi-state transitions, and history-dependent rules.The section emphasizes how different higher-order structures vary in their capacity to support and express these mechanisms.Source: §3 · pp. 6–7
  • Figure 5 maps five higher-order representations against contagion mechanisms, distinguishing naturally supported mechanisms from those requiring additional assumptions and combinations with limited evidence.The five representations are simplicial complexes, hypergraphs, directed, temporal, and multilayer higher-order networks.Source: Figure 5 · p. 7
  • The review groups higher-order contagion into synergistic reinforcement, competitive inhibition, memory-aware, and threshold-activated mechanisms.For each family, it outlines principles, surveys representative models, and considers how representations constrain or enhance mechanism expression.Source: §3 · p. 7

3.1 Synergistic-reinforcement contagion

Synergistic reinforcement makes infection depend nonlinearly on group activity. The section distinguishes simplex-based, order-dependent rates from nonlinear hyperedge kernels, then surveys the structures that can carry these mechanisms.

3.1 Synergistic-reinforcement contagion

  • The review treats synergistic reinforcement as nonlinear amplification of transmission as infected group members accumulate, an effect pairwise models cannot represent.Simplicial complexes and hypergraphs are identified as natural carriers of these cooperative effects.Source: §3.1 · p. 7, §3.1.2 · pp. 7–8

3.1.1 Classification of Synergistic-Reinforcement Contagion

  • Simplicial models combine pairwise transmission with higher-order reinforcement, with collective or individual activation rules determining when a simplex can transmit.The cited model assigns distinct rates to interactions of different orders.Source: §3.1.1 · p. 7, §3.1.2 · p. 8
  • Two complementary implementations encode synergy: simplices use interaction-order-dependent transmission rates, while hyperedges use nonlinear infection kernels.Both make infection rates depend on the number and configuration of active group members.Source: §3.1.1 · p. 7

3.1.2 Structural Carrying Capacity of Synergistic Reinforcement Mechanisms

  • Hypergraph models make infection pressure a nonlinear function of infected members in incident hyperedges, including power-law and saturating forms.Uniform, layered, and heterogeneous hypergraphs provide different ways to assign these rates.Source: §3.1.1 · p. 7, §3.1.2 · p. 8
  • Higher-order components and nested hypergraphs carry reinforcement through infected-node-dependent probabilities and size- or load-dependent rates.These structures use shared nodes among hyperedges or containment of smaller hyperedges within larger ones.Source: §3.1.2 · p. 8
  • Directed, multilayer, and temporal hypergraphs can support synergy with additional constraints, whereas path-based structures lack multi-node group interactions.Directionality, inter-layer reinforcement, and time-resolved group activation provide distinct extensions.Source: §3.1.2 · p. 8

3.2 Competitive–inhibition Contagion

Competitive–inhibition contagion covers higher-order processes that suppress or compete with spreading. The section distinguishes group-induced suppression from interactions among multiple contagions and reviews how different structures encode them.

3.2 Competitive–inhibition Contagion

  • Competitive–inhibition contagion includes transmission that is delayed, reduced, or blocked by competing pathways or suppressive group effects.The section distinguishes group-induced suppression of one contagion from direct competition among multiple contagions.Source: §3.2 · pp. 8–9

3.2.1 Classification of competitive–inhibition contagion

  • Models are classified by whether higher-order influence limits contagion through local saturation, resource competition, or global normalization.The two broad mechanisms are cooperative inhibition within groups and cross-inhibitory competition between contagions.Source: §3.2.1 · p. 9

3.2.2 Structural carrying capacity of competitive–inhibition contagion

  • Simplicial complexes encode inhibition through reduced or nonpositive higher-order rates and competition through contagion-specific node states; multilayer versions couple spreading processes across layers.Hypergraphs can express cooperative inhibition with sublinear reinforcement, but cross-inhibitory competition is less explored and multilayer hypergraph models remain largely undeveloped.Source: §3.2.2 · p. 9

3.3 Memory-Aware Contagion

Memory-aware contagion makes spreading depend on temporal history rather than only the current system state. The section separates memory of node-state duration from accumulated contact exposure and describes complementary structural implementations.

3.3 Memory-Aware Contagion

  • Memory-aware dynamics depend on past interactions, exposure order, or repeated signals, producing non-Markovian spreading and potentially delayed activation or propagation inertia.Memory also breaks the equivalence between static and temporal representations because spreading follows temporal paths.Source: §3.3 · pp. 9–10

3.3.1 Classification of Memory-Aware Contagion

  • State-duration memory makes transitions depend on a node’s trajectory, while exposure-accumulation memory makes activation depend on repeated, ordered, or temporally clustered contacts.The first includes time spent infected or time since exposure; the second accumulates influence across contacts.Source: §3.3.1 · p. 10

3.3.2 Structural carrying capacity of memory-aware contagion

  • Simplicial complexes encode node-level temporal dependence within group interactions, while temporal hypergraphs can encode group-level memory through recent activity windows.Memory-aware contagion in broader higher-order, multilayer, or hybrid structures remains limited and is an open challenge.Source: §3.3.2 · p. 10

3.4 Threshold-activated contagion

Threshold-activated contagion requires influence to exceed a predefined level, formalizing group effects through explicit activation rules. The section classifies thresholds by where they operate and how they vary, then compares structural support and remaining gaps.

3.4 Threshold-activated contagion

  • Threshold mechanisms activate a node or group only when influence exceeds a predefined level, linking collective input to cascades, critical transitions, and tipping behavior.The review notes abrupt phase transitions, cascade criticality, and multistable regimes in higher-order threshold models.Source: §3.4 · p. 10

3.4.1 Classification of threshold-activated contagion

  • Thresholds may operate at the node or group level, while single fixed thresholds dominate and evolving or history-dependent multi-threshold rules remain largely unexplored.Rules can compare active-neighbor fractions, simplex-order-specific fractions, or active members in a hyperedge against thresholds.Source: §3.4.1 · p. 10

3.4.2 Structural carrying capacity of threshold-activated contagion

  • Simplicial and hypergraph structures support order- or hyperedge-specific thresholds, while directed hypergraphs can encode asymmetric activation rules.Temporal interactions complicate fixed thresholds, and systematic exploration in multiplex higher-order settings remains limited.Source: §3.4.2 · pp. 10–11

4 EMERGENT DYNAMICAL BEHAVIORS

Higher-order contagion produces abrupt and hybrid transitions, coexisting stable outcomes, and delayed spreading trajectories through particular combinations of network structure and contagion mechanism. The section synthesizes these mechanisms and structural amplifiers into a mapping from pairings—not either factor alone—to emergent behavior.

4.1 Discontinuous and hybrid phase transitions

  • Strong group reinforcement can produce discontinuous or hybrid transitions with bistability, while competition can cause abrupt shifts in which contagion dominates.In hypergraphs, greater overlap can instead yield continuous transitions; memory and threshold mechanisms also modulate transition behavior.Source: §4.1 · pp. 11–13

4.2 Multistability and Coexistence States

  • Multistability arises through reinforcement, competition, thresholds, and memory, allowing distinct long-term outcomes under identical settings and making outcomes sensitive to initial conditions.Threshold-driven multistability is less systematically characterized, and a fixed-point classification for general higher-order threshold models remains open.Source: §4.2 · pp. 13–14

4.3 Propagation Inertia and Delay Effects

  • Higher-order spreading can show slow growth, long plateaus, and delayed outbreaks when group reinforcement, competition, awareness, memory, or activation thresholds postpone spreading.Examples include insufficient early reinforcement, cross-layer awareness that suppresses infection, and group thresholds or higher-order cores that delay large-scale activation.Source: §4.3 · pp. 14–15

4.4 Synthesis of three-level relations

  • Similar behaviors can arise from different structure–mechanism pairings, with distinct structural amplifiers and observable signatures.The synthesis associates cohesive motifs and reinforcement or competition with multistability, sharp activation rules with abrupt transitions, and memory, competition, or reinforcement on temporal or multilayer architectures with inertia.Source: §4.4 · p. 15

5 APPLICATIONS

The applications section examines how higher-order structure and spreading mechanisms can be aligned to promote diffusion or regulate unwanted spread. The appropriate design or intervention depends on the mechanism and operating regime.

5.1 Maximizing diffusion via reinforcement mechanisms

  • Reinforcement-driven spreading can accelerate adoption, enlarge cascades, and shape diffusion timelines across social information, epidemiological, and IoT settings.Selecting seeds embedded in multiple influential groups and aligning higher-order connectivity with reinforcement can promote faster, broader, and more robust spreading.Source: §5.1 · p. 16

5.2 Shaping and controlling spreading processes

  • Higher-order groups can be targeted to suppress epidemics, misinformation, and cascading failures; intervention strategies should also match the spreading mechanism.Limiting repeated exposure can mitigate reinforcement, while memory-aware processes may require temporally adaptive interventions.Source: §5.2 · p. 16
  • Features that amplify diffusion in one setting may suppress it in another, so application-specific design should jointly consider structure, mechanism, and operating regime.The structure–mechanism–behavior perspective is presented as a practical guide for such design.Source: §5.2 · p. 16

6 OPEN ISSUES

The review frames open issues through a structure–mechanism–behavior lens, intended to support both intervention design and the choice of adequate models.

6.1 Structural representations

  • Structural research still needs to establish when higher-order patterns can be reduced without losing relevant behaviors, which mesoscopic features matter, and how to reconstruct structure from incomplete data.The open questions include representation minimality and equivalence, overlapping groups and higher-order motifs, and identifying which structural aspects are dynamically relevant.Source: §6.1 · p. 16

6.2 Dynamical regimes and theoretical challenges

  • Theory remains fragmented across models and substrates, without a common account of multistability, hybrid transitions, fluctuations, heterogeneity, or temporal variability.The co-occurrence of multistability, discontinuous thresholds, and inertia across parameter space, as well as control and optimization of these regimes, remains poorly charted.Source: §6.2 · pp. 16–17

6.3 Data, inference, and applications

  • Standard pipelines are lacking for inferring contagion mechanisms from time series and selecting among structure–mechanism combinations.The review calls for tools that distinguish synergistic, threshold-activated, and memory-aware dynamics and learn infection kernels or threshold functions from data.Source: §6.3 · p. 17
  • Turning insights into actionable design principles for epidemic control, social coordination, or infrastructure cascades remains an open challenge.Examples include identifying group structures to avoid discontinuous outbreaks and using awareness layers to induce benign coexistence.Source: §6.3 · p. 17

7 CONCLUSION

The conclusion presents a structure–mechanism–behavior perspective for reviewing higher-order spreading and connects it to modeling, analysis, and control.

7 CONCLUSION

  • The review maps how higher-order representations shape interaction patterns, how spreading mechanisms operate on them, and how their interplay yields emergent behaviors.Source: §7 · p. 17
  • The perspective moves beyond model-centric descriptions to connect modeling, analysis, and control of spreading processes.Source: §7 · p. 17
  • The mapping offers practical guidance for maximization and intervention strategies, from structure-aware modeling to mechanism-driven control.Source: §7 · p. 17
Loading 2610.39297v1…