论文信息 - Second Life as a Simulation Environment : Rich , high-fidelity world , minus the hassles

Second Life as a Simulation Environment : Rich , high-fidelity world , minus the hassles

Second LifeTM is a 3D virtual world with unlimited potential as a tool for cognitive modeling. This paper discusses the many advantages of using Second Life versus other simulation environments, the aspects of cognitive modeling that this simulation environment may be appropriate for, the interface setup, and various technical issues. Two simulations are provided as examples of interfacing Second Life with cognitive models, including an example where the highfidelity complexity and constraints of Second Life may help to distinguish between models and/or parameter values that produce varying performance in different task environments.

V. D. Veksler

[1] C. Lebiere,et al. The Atomic Components of Thought , 1998 .

[2] N. Schmajuk,et al. Latent learning, shortcuts and detours: a computational model , 2002, Behavioural Processes.

[3] Aluizio F. R. Araújo,et al. A topological reinforcement learning agent for navigation , 2003, Neural Computing & Applications.

[4] Richard S. Sutton,et al. Reinforcement Learning: An Introduction , 1998, IEEE Trans. Neural Networks.

[5] Wayne D. Gray,et al. Goal-Proximity Decision Making: Who needs reward anyway? , 2009 .

[6] David J. H. Burden. Deploying embodied AI into virtual worlds , 2009, Knowl. Based Syst..

[7] Kathryn E. Merrick,et al. Motivated Learning from Interesting Events: Adaptive, Multitask Learning Agents for Complex Environments , 2009, Adapt. Behav..