Developing Explanations of Emergent Phenomena in Physics Through Visualization and the Analysis of “Midlevel” Representations

• In Event: Developing Evidence of Understanding Complex Systems

Abstract

Emergence is the process by which complex collective behaviors arise out of individuals' properties and interactions, usually in non-obvious ways (Strogatz, 2003;Wilensky & Resnick, 1999). Levy & Wilensky (2008) found a pervasive strategy that novices employ when reasoning about everyday emergent phenomena - “mid-level” construction – which involves identifying aggregations of a few objects and their local interactions, which are treated as homogenous entities within the larger system. This paper proposes an instructional approach for learning emergence, based on the production, visualization and analysis of such mid-level representations. It examines whether and how novices (5th graders), using this approach, can come to understand emergent phenomena that have been shown to be particularly challenging to understand for even college students (Reiner et al., 2000) - electric current and heat in resistive circuits.

The design experiment (Cobb et al., 2002) reported here was conducted with 5th graders (n=20) in a metropolitan public school, lasting three class periods. Students, scaffolded by printed Activity Sheets, interacted with a multi-agent-based computer model of electrical conduction (Sengupta, 2010) designed in NetLogo (Wilensky, 1999). This model represents current and resistance as emergent from the collective movement of electrons through a series of obstacles (atoms) (see Figure 1), based on a well-accepted theory in physics (Drude, 1900). The mid-level representations used in this study represent local interactions between a few agents (e.g., collision between stationary atoms and moving electrons) (Figure 2A). Data sources include learner-generated artifacts (images and written analyses of images), semi-clinical interviews with 5 randomly selected students during the focal activity (described below), and written pre- and post-tests, where students explained a) how a light-bulb works, and b) whether (and why) electric current remains the same throughout a light-bulb circuit.

During the focal activity, each learner produced three or four different mid-level representations in the form of 2D images of local interactions between atoms and free-electrons in the wire, by selecting and modifying variables corresponding to agent-behaviors on the model interface (Figure 1, highlighted portion). Then, through qualitatively analyzing the relationships between different visual elements in the image, learners would estimate qualitatively and mechanistically how hot or cold each region (represented in each image) would be (Figure 2B).

Analysis of interviews and learner-generated artifacts reveals that a) learners first qualitatively explained the mechanisms of conversion of electric current to heat, and the local spreading of heat as mid-level phenomena, and b) then developed quantitative measures of the “hotness” each “mid-level” region would be through analyzing the structural relationships between salient spatial components in each mid-level image - by using their piecemeal intuitive physical knowledge (e.g., collisions produce heat; impetus dies with distance; c.f. diSessa, 1993). Post-test responses show a) all students correctly indicated that electric current would be equal on either side of the light-bulb (compared to 10% in the pre-test), and b) that all participants were able to provide a “mechanism” that produces light (compared to 0% in the pre test), based on their mid-level mechanistic explanations of conversion of electric current to heat.

Authors

• Pratim Sengupta, Vanderbilt University

• Gokul Krishnan, Vanderbilt University