Why Dissect a Frog When You Can Simulate a Lion?
暂无分享,去创建一个
We are concerned with creating computer-based learning environments which provide students with opportunities to develop causal explanations of complex phenomena through experimentation and observation. We combine video and simulation to facilitate such exploration in high school biology classrooms. Specifically, we focus on issues in behavioral ecology and the predation behaviors of the Serengeti lion. The question explored by students concerns the effectiveness of the lion’s hunting strategies. Although popularized as a skilled predator, only 1530% of the hunts attempted by lions result in the successful capture of prey (Schaller 1972). Explaining the causal factors underlying this statistic requires an understanding of cooperative behavior, optimality, resource competition, and variation. The conjecture is that active explanation will result in a greater understanding of these concepts than passive lecture/textbook teaching approaches. Students first watch a collection of video clips showing lions employing various hunting strategies. Their task is to create narratives by explaining and comparing features of the videos. Video is useful for detecting dynamic features such as speed or use of cover, but factors that are “implicit” (e.g., time of day), invisible (e.g., wind direction), or ob scured (e.g., spatial positioning of predator/prey) in the film can be difficult to isolate and understand. To highlight such salient events, we are building a simulation of the lion hunt called the Animal Landlord. Animals are behavioral agents possessing perceptual, motor, and action selection routines. Students observe an aerial view of the creatures and can manipulate parameters influencing their behaviors (e.g., number of predators, vegetation density). Thus, the simulation provides a setting in which further data can be collected to investigate questions raised during the video exercise. Students require assistance in open-ended experimentation tasks, so unlike similar systems (e.g., Resnick 1994), we provide domain-specific guidance. We monitor the progress of agents to structure a debriefing episode (Johnson 1994) upon completion of a hunt. The rule system and logic-based truth maintenance system described in (Everett & Forbus 1996) records logical dependencies between aspects of the environment, creatures and their actions, and justifications for these actions. Agents can be queried about all actions taken during the hunt through a multimedia interface. The dependencies are
[1] Kenneth D. Forbus,et al. Scaling up Logic-Based Truth Maintenance Systems via Fact Garbage Collection , 1996, AAAI/IAAI, Vol. 1.
[2] W. Lewis Johnson,et al. Agents that Learn to Explain Themselves , 1994, AAAI.
[3] G. Schaller. The Serengeti Lion: A Study of Predator-Prey Relations , 1972 .
[4] Mitchel Resnick,et al. Turtles, termites, and traffic jams - explorations in massively parallel microworlds , 1994 .