A scaffolding framework to support learning in multi-agent based simulation environments

Simulations provide a suitable environment for discovery learning, but are not pedagogically effective unless exploration tasks are suitably scaffolded. We are developing an architecture for a computer based learning environment that includes a multi-agent based simulation, a causal modeling tool, and a set of contextualized scaffolds provided by a virtual agent to help middle school students learn important science concepts. The particular domain of study is ecosystems and ecological processes, and our goal is to help students gain a deep understanding of ecological processes.