| Abstract |
Humans and animals react in recognizable ways to surprising events. However, there is a lack of models that generate surprise intensity and its effects on behaviour in a realistic way, leading to impoverished and non-humanlike behaviour of agents in situations where humans would react surprised. To fill in this gap in agent-based modelling, a compu-ational model is developed based on psychological empirical findings and theories from literature with which agents can display surprised behaviour. We tested this model in a simulated historical case from the domain of air combat and evaluated three behavioural properties against these simulated runs. The conclusion is that the model captures aspects of surprised behaviour and thus can help make agents behave more realistic in surprising situations.
|