Agent-Based Modelling

Agent-based modelling is a form of individual-based population model where individual simulants maintain internal data states and have senses that allow them to detect the environment they find themselves in. For a primate simulation they need a standard behavioural repertoire, a set of capabilitieis, and their goal is to maximise personal fitness. This should be sufficient to generate observed behaviours and we certainly plan to try and match the outcomes of our modelling with field-based observational data.

This figure illustrates the SimAgent system showing individual agents, their field of view and hearing ranges. Decision trees are used to allow the agent to decide on what to do in the next step depending on its internal state and sensory data. The internal state includes measures of energy balance, social rank and fatigue. Sensory data reveals the likely presense of food, conspecifics, and predators.

The environment can match existing ecological data in terms of productivity and patchiness. Foraging by agents can deplete resources in a realistic manner, and predation can either be a risk, or the predator itself can be another agent with a rather different decision tree. We are currently working on a pilot project using field data on Chacma Baboons (Papio cynocephalus ursinus) obtained at De Hoop Nature Reserve in South Africa. This includes detailed ecological and behavioural data that the agent based model can use and emulate to test theories about the effects of group size and seasonality on foraging patterns. The movie on the left represents 10 days from 6 months of data but it is really too small to see clearly. If you have a fast data connection you may wish to download a much larger but clearer movie.

Current visualisation techniques are usually rather basic although there are systems that allow 3D display and the technology will ultimately be linked with the locomotor simulations to provide a realistic depiction of individual and community activity. In particular in the Virtual Plio-Pleistocene Project we propose to produce a realistic reconstruction of the conditions in place during this key period of human evolution so that we can provide a workbench for testing the various theories that explain the divergence of the chimp and human lineages.