Skip to ContentSkip to Navigation

Self-organization in social systems

Hemelrijk group - Research

The study of self-organization investigates how complexity at a higher level emerges from the interactions among simple units at a lower level. For instance, schooling behaviour of fish and flocking of birds may emerge from simple interactions among nearby group members. Using computer models we apply this process-oriented approach to uncover emergent phenomena of animal behaviour and their social organization, and to generate insights into how natural selection operates at several levels of complexity (e.g., individual, group, and community). Our specific aims are to understand the formation of social structure of fish schools, bird flocks and primate societies, and how the specifics of locomotion, aero- and hydrodynamics, individual personalities, memories, and social cognition affect these social structures. By integrating behaviour, interactions with others, aero- and hydrodynamics, and spatial structure in these simulations, our approach delivers new hypotheses that can be tested in natural settings. We test these model-based predictions by closely cooperating with empirical scientists.

Our research themes focus on understanding of:

Predatory attacks

A fundamental evolutionary question is whether and how the guidance strategy of birds of prey during aerial hunting has been shaped by natural selection. To answer this, we need to integrate mechanism and function in an evolutionary study (Evo-mecho approach). For this, we use a computational model within the framework of complexity science. Our aim is to:

  1. Gain insight how evolution may have shaped the prey-targeting system depending on the prey’s biomechanics and behaviour,
  2. elucidate the interplay between pursuit-evasion tactics, behavioural timing, biomechanics, flight control, prey-targeting and catch rate,
  3. generate and verify predictions in real raptors in cooperation with the Oxford Animal Flight Group, UK.

Collaborators: Robin Mills, Hanno Hildenbrandt, Graham Taylor

Collective escape by swarms

From the perspective of the prey, we aim to gain understanding about collective patterns of escape. In a combined modeling and empirical study in cooperation with University of Toulouse, France, we investigate the collective response of a school of fish when being attacked by a predator. Our aim is to:

  1. Gain insight in the process of transmission of information about the predator,
  2. elucidate how the different kinds of escape response relate to the strategy of the predator and the collective pattern of escape of prey, and
  3. find out by what behavior the prey confuses the predator.

Collaborators: Valentin Lecheval, Guy Theraulaz, Hanno Hildenbrandt

In flocks of starlings we have investigated what causes the agitation wave, when a flock is attacked by a falcon. An agitation wave is a black band that is moving away from the raptor. Since the flock is too far away, we do not know what escape manouevre of the individuals causes this. Our computational model shows that we observe an agitation wave only if the prey escapes by manoeunvring sidewards and back, in a kind of zigzag, not if prey moves temporarily closer together (Hemelrijk, Zuidam and Hildenbrandt 2015, blog).

In addition, we investigate vigilance and information transmission in a computational model with group-living animals.

Collaborators: Fransje van Weerden, Rineke Verbrugge

Social behavior in primates and corvids

In primates and corvids we investigate how behavioural rules that are cognitively unusually simple result in complex patterns of social interaction. These complex patterns have usually been supposed to represent reconciliation, consolation and attribution of knowledge to others.

In primates we are investigating the self-organisation of inter-sexual dominance. We develop new measures of dominance style.


Social interaction matrices are often analysed with the help of matrix-statistics, such as the TauKr test (Hemelrijk, 1990a; Hemelrijk, 1990b). For questions regarding matrix statistics for social interaction matrices or to obtain the program MatrixTester, contact Charlotte Hemelrijk.

Last modified:08 September 2020 4.42 p.m.