Network Hawkes Process Models for Exploring Latent Hierarchy in Social Animal Interactions

TL;DR

This paper introduces a novel cohort Markov-Modulated Hawkes process model that effectively captures dynamic social hierarchies in animal interactions by leveraging interaction timestamps and addressing key behavioral phenomena.

Contribution

The paper develops and evaluates a new network point process model, C-MMHP, that incorporates important behavioral features to better infer latent social hierarchies from interaction data.

Findings

C-MMHP outperforms existing models in capturing interaction patterns.

The model accurately infers latent ranks in simulated and real data.

It provides meaningful predictions of social hierarchy dynamics.

Abstract

Group-based social dominance hierarchies are of essential interest in animal behavior research. Studies often record aggressive interactions observed over time, and models that can capture such dynamic hierarchy are therefore crucial. Traditional ranking methods summarize interactions across time, using only aggregate counts. Instead, we take advantage of the interaction timestamps, proposing a series of network point process models with latent ranks. We carefully design these models to incorporate important characteristics of animal interaction data, including the winner effect, bursting and pair-flip phenomena. Through iteratively constructing and evaluating these models we arrive at the final cohort Markov-Modulated Hawkes process (C-MMHP), which best characterizes all aforementioned patterns observed in interaction data. We compare all models using simulated and real data. Using statistically developed diagnostic perspectives, we demonstrate that the C-MMHP model outperforms other methods, capturing relevant latent ranking structures that lead to meaningful predictions for real data.