Agent-based systems are used to simulate the behaviour of people in various
spatial environments. Designers of such systems need to give attention to
ontological and epistemological concerns during the early stages of develop
ment to assure that the agent-based system is built upon a sound foundation
. Only then is it possible to model the agent's behaviour in a cognitively
plausible way. This paper describes an ecological approach to model ontolog
y and epistemology for agent-based wayfinding simulation. The ontology of t
he wayfinding environment is based on a subdivision into medium, substances
, and surfaces. The epistemological model uses the concept of affordances,
which we divide into physical, social-institutional, and mental affordances
. Ontology and epistemology are both grounded in people's descriptions of t
heir wayfinding experiences. A case study of wayfinding in airports is used
to demonstrate the applicability of the method for designing a simulation
with a cognizing agent.