Challenging Topics in Science 1 Helping students understand challenging topics in science through ontology training

Chi (2005) has proposed that students experience difficulty in learning about physics concepts such as light, heat, or electric current because they attribute these concepts with an inappropriate ontological status of material substances rather than the more veridical status of emergent processes. Conceptual change could thus be facilitated by training students in the appropriate ontology prior to physics instruction. We tested this prediction by developing a computer-based module where subjects learned about emergent processes. Control subjects completed a computer-based task that was uninformative with respect to ontology. Both groups then studied a physics text concerned with electricity, including explanations and a post-test. Verbal explanations and qualitative problem solutions reveal that experimental students gained a deeper understanding of electric current. Challenging Topics in Science 2 Introduction Students' understandings of concepts like force, light, heat, or electricity are well-established and quite distinct from the conventional scientific views offered by instructors. For decades, cognitive and science education research has examined the science knowledge of novices and experts in a widespread effort to identify and characterize preconceptions of various science concepts. Many of the earliest studies of naive science conceptions (e.g., King, 1961; Kuethe, 1963; Doran, 1972; Viennot, 1979; Minstrell, 1982; Shipstone, 1984) sought to document the existence of firmly held preconceptions or so-called "robust misconceptions" that are particularly resistant to instruction. Not surprisingly, a consensus has emerged that "young children do have firmly held views about many science topics prior to being taught science at school" (Osborne & Wittrock, 1983, p. 489). This simple but important statement is reflected by nearly 6000 published studies of students’ misconceptions and instructional attempts at their removal (Pfundt & Duit, 1988; Duit, 2004). Research on student misconceptions has focused primarily on those concepts for which students exhibit robust misconceptions. In reviewing a wide range of such studies, Reiner, Slotta, Chi and Resnick (2000) found that students often attribute difficult concepts with materialistic properties. They reviewed numerous studies that were concerned with physics novices’ conceptualizations of force, light, heat, and electricity (all of which are notoriously difficult concepts), and found a pattern of robust misconceptions across all these topics. Specifically, Reiner et al. (2000) observed that physics novices tend to think of these concepts as if they are material substances, or have certain properties of material substances. This conclusion was based directly upon arguments offered within the research articles they reviewed, as well as on inferences from particular attributions in the misconceptions that were reported. For example, if a novice reasoned that a moving object slows down because it has "used up all its force," this reasoning was taken as evidence of a substance-based conception of force – in contrast to the more conventional view of force as a process of interaction between two or more objects. Similarly, Reiner et al. (2000) observed that novices’ conceptualizations of heat were drawn from reasoning that involved heat (or Challenging Topics in Science 3 cold) being "blocked" or "trapped," suggesting a substance-like view. The Reiner et al. (2000) review established a pattern across concepts, suggesting a common origination of naive conceptions. Any theory of conceptual change is therefore challenged to account for this pattern of misconceptions, and ideally to respond with an effective method of instruction that responds to the robust substance-like nature of conceptualizations in these various topics. Ontological Attributions Chi (1992; 1997) has hypothesized that some misconceptions are robust because they involve changing one's commitment about the ontological nature of the concept. This view assumes that people associate concepts with distinct ontologies (c.f., Keil, 1981), such as processes, ideas, and material substances to name a few. (Throughout the paper, ontological categories will be italicized) When encountering a novel concept, the learner forms an ontological commitment that guides his or her understanding of fundamental aspects of that concept and leads to attributions of features or properties. Thus, in learning about a new concept such as osmosis, a person may commit to a process ontology,1 which implies the attribute "occurs over time," since this is a common characteristic of all processes. Misconceptions result from commitments to an inappropriate ontology. In learning about the concept of "heat,” for example, many children assume a material substance ontology, perhaps because of language conventions such as "close the door, you're letting all the heat out." However, in the scientifically normative view, the concept of heat is associated with a process ontology, as it involves the transfer of kinetic energy between molecules (Slotta, Chi & Joram, 1995). Unfortunately, once an ontological commitment is made with respect to a concept, it is difficult through any stages of mental transformation to change our fundamental conception from a substance to a process (Chi & Roscoe, 2002). Thus, ontologically misattributed concepts would require an extraordinary process of conceptual change. Further contributing to the robust quality of such alternative conceptions is the fact that students may lack any notion of the appropriate ontology with which certain concepts should be attributed. Chi (2005) has proposed that many concepts of this nature are not only processes (as opposed to substances), but moreover, they are a specific kind of processes that she calls emergent Challenging Topics in Science 4 (as opposed to direct), which are particularly for students to understand in a scientifically normative way. These processes typically involve emergent properties of a system, such as equilibrium states or net statistical changes of certain system properties (e.g., inside and outside temperature; voltages; air pressures; etc.). Such emergent relationships are often difficult for students to understand, in part because they can involve misleading perceptual correlates. For example, the diffusion of a beaker of blue liquid through a valve into a beaker of clear liquid suggests a simple direct causal mechanism where the blue liquid continues flowing into the clear liquid until an equal amount exists in both beakers, when the process comes to a halt. However, while this appears to be a direct process where blue liquid flows from one side of the beaker to the other, it s actually a much more complicated emergent process, resulting from the continuing action of dyed water molecules that move independently of one another, but statistically even out in the two sides of the beaker over time. The movement of these molecules, and indeed the process itself, continues indefinitely, even after the equilibrium state has been achieved. Chi (2005) has identified such processes as emergent (as opposed to direct) ones, because their observable, “macro level” patterns are seen to emerge from the lower, or “micro level” in a characteristic way. Moreover, the need to focus on two or more levels in emergent processes may also contribute to students’ difficulties (Wilensky & Resnick (1999). Chi (2005) explains that emergent processes are particularly troublesome for students to understand because they misattribute the concept’s ontological nature, either as a material substance or as a kind of direct causal process. Other examples of emergent processes that have been studied by researchers include natural selection (Ferrari & Chi, 1998; Hallden, 1988; Jensen and Findley, 1996), light (Slotta, Chi & Joram, 1995), traffic jams (Resnick, 1996), heat and temperature (Wiser & Carey, 1985; Wiser, 1996); and electric current (Joshua & Dupin, 1987; McDermott and Shaffer, 1992; Slotta, Chi and Joram, 1995). Assessing Ontological Commitments Chi’s account of misattributed ontology suggests that novices may need to revise their ontological commitment in order to understand the scientifically normative view of certain concepts, Challenging Topics in Science 5 and escape the reasoning errors that follow from their initial misclassification. In order to investigate such a claim, we require an assessment of a student’s ontological commitments, which will allow us to measure what those commitments are and whether they have changed as a result of an instructional intervention. Slotta, Chi and Joram (1995) developed such a methodology, based on a content analysis of students’ verbal explanations, which provided a source of inferences concerning ontological commitments. For example, if a student explains that light or heat can be blocked by a wall, we infer that the student’s underlying conceptualization must have some properties of material substances, which possess an ontological aspect that can “move,” “be blocked,” “be contained,” etc. Slotta et al. (1995) sought to differentiate novices’ and experts’ conceptions of light, heat and electric current. When physics novices were asked to solve conceptual problems involving these topics, they demonstrated a clear bias towards substance-like mental models (e.g., reasoning about electric current in a wire as if it were a fluid flowing inside a hose). This result was determined by presenting subjects with isomorphic pairs of problems, one of which was concerned with an emergent process concept (e.g., light, heat, or electric current), and the other a material substance isomorph of that problem (e.g., water), assuming the relevant physics concept was viewed as a material substance. For example, a problem involving an electric circuit with several bulbs in series was accompanied by a corresponding isomorphic problem involving water flowing through a hose with several sprinklers in series -assuming the physics novice thought of electric current as something like a flowing fluid. Such isomorphic pairs of problems were constructed with several choices of answers, so that similar answers to the problems would reflect similar conceptual reasoning (e.g., "the bulbs closer to the battery come on before the bulbs farther away" is similar to "the sprinklers closer to the faucet will come on before the sprinklers farther away"). That is, choosing the answer “the bulbs closer to the battery come on before the bulbs farther away” is incorrect, but choosing it implies that subjects are analogizing it to the “sprinkler” problem. In reasoning about such problems, novices preferred the substance-like models, leading to incorrect Challenging Topics in Science 6 choices of the solution that were consistent with the correct choices to the corresponding material substance isomorph problem. Slotta et al. (1995) looked more deeply into students’ ontological commitments by examining patterns of verbal predication in the language used by physics novices and experts as they explained their solutions to these problems. This analysis drew inferences about a subject's ontological commitments based on the presence of particular verbal predicates in his or her explanation. For example, if a subject said, "The current comes down the wire and gets used up by the first bulb, so very little of it makes its way to the second bulb,” then these four (underlined) predicates were taken as evidence that subjects conceptualized current as a substance-like entity with attributes of (1) “moving,”(2) “can be consumed,”(3) “can be quantified,” and (4) “moves.” respectively. By measuring the degree to which subjects used these attributes in explaining their answers to a variety of conceptual problems, it was possible to quantitatively address the question of ontological association. Figure 1 displays the average level of process and substance predication used by experts and novices in the Slotta, et al. (1995) study. Whereas novices relied almost exclusively on substance attributes regardless of problem types, with very little use of process attributes (see the two parallel solid lines), experts used the same high levels of substance attributes for the substance concept problems but chose more process attributes for the physics concept problems. Furthermore, the pattern of substance predication (e.g., across all the substance attributes assessed: Moves, Consumed, Quantified, Blocked, etc.) was quite similar for novices between the physics concept problems and their material substance isomorphs. Thus, not only were the novices’ multiple choice responses similar between the two members of an isomorphic pair of problems, so was the pattern of verbal predication within their explanations. This analysis provided evidence that novices attribute properties or behaviors of material substances to certain physics topics, while expert conceptualizations of the same topics show no sign of a substance-like ontology, but rather appear to be consistent with a process ontology. ------------Challenging Topics in Science 7 Insert Figure 1 about here ------------Testable Predications about Instruction for Conceptual Change Chi's framework suggests that students would be less likely to make faulty ontological commitments if (a) they were prepared with some knowledge of the appropriate ontology before instruction about those concepts, and (b) their initial experience with (or instruction about) the concepts did not provide any suggestions of the wrong ontology. But students of almost any age have had some exposure to concepts of electricity, and those initial exposures were most likely suggestive of substance ontology rather than that of emergent processes. Young children are almost certainly exposed to substance-based language and conceptualizations regarding topics in electricity (e.g., "the battery is out of juice"). Thus, students probably enter instruction already in possession of the substance-based misconceptions that we wish to avoid, as shown in the Reiner et al. (2000) review. We must therefore ask whether points (a), providing some knowledge of the appropriate ontology before instruction of a specific concepts, and (b), barring any association with the wrong ontology, could also apply to the removal of misconceptions and not just in their prevention. Of course, it remains an open question whether or not any conception is actually removed, or whether these early concepts are simply subordinated to the normative conceptualizations over the course of instruction. Indeed, there is some evidence (Clement, 1987; McDermott, 1979; Slotta, Chi & Joram, 1995) that physics experts do maintain substance-based conceptualizations in parallel with their more normative process-like views. In their everyday reasoning, physics experts often use substance-like models of heat, light, and electricity, although they are well aware of the limitations of such models, including when the models should be abandoned (Slotta, Chi & Joram, 1995). Thus, if the early substance-like conceptions are not actually removed or replaced, we can interpret conceptual change as a matter of developing new conceptualizations alongside existing ones, and understanding how and when to differentiate between alternatives. Nevertheless, the problem Challenging Topics in Science 8 remains the same: How do we prevent the instruction of a physics concept that is of the process-like ontology from being assimilated into a student’s pre-existing substance-like conceptualizations? This paper assumes that physics novices possess substance-based conceptualizations of electricity, based on prior research by Slotta, Chi and Joram (1995) and Reiner et al (2000). We hypothesize that we can help novices develop an understanding of the process-like nature of electric current by first providing them with some training about the target ontology (emergent processes) followed by direct instruction about electricity that avoids any use of terms or analogies that might promote the material substance ontology (e.g., the water flow analogy). Overview of the Study In order to test this hypothesis, a training study was designed in which one group of physics novices received direct training about the emergent process ontology, followed by an instructional text concerning electric current that omitted any suggestion of a material substance ontology. A control group received no training in the emergent process ontology, performing a control task instead, and then read the same instructional text about electric current as the experimental group. The question of interest is whether the experimental group demonstrated conceptual change, as measured by a shift in their ontological associations for the concept of electric current. Subjects’ conceptualizations of electric current were measured by a pre-post test consisting of eight qualitative physics problems concerned with electric current, for which subjects selected a response from multiple choices, then verbally explained their reasoning in an interview format. Subjects’ choice of response to these problems, as well as their verbal explanations, provided measures of their ontological commitments at preand post test, enabling comparisons between control and experiment groups, and assessment of the impact of ontology training. An essential feature of the design is that both the experimental and the control groups received exactly the same instruction about the target concept of electric current. The two groups differed only in that the experimental group received prior instruction about the emergent process ontology. This training included no mention of electricity, nor any foreshadowing of its application to electricity concepts. The role of the ontology training was to provide the students with some Challenging Topics in Science 9 knowledge of the emergent process ontology so that they might better succeed in making the correct ontological attribution when subsequently learning about the concept of electric current. Assessment of conceptual change was performed by analyzing verbal explanation data for the presence of two specific sets of ontological attributes based on the attributes used by Slotta et al., 1995), indicating subjects’ commitment to a either a material substance or an emergent process ontology, respectively. Thus, if a subject explained problems concerning electric current using verbal predicates relevant to a material substance, this was taken as a measure of an underlying ontological commitment a material substance2 view of the concept. Similarly, the use of verbal predicates reflecting ontological attributes of emergent process is taken to reflect the presence of an emergent process association. It was predicted that the experimental group would show a transition from the pre-test, where they explained problems in terms of a substance ontology, to the post-test, where they drew upon emergent process predicates in their explanations. In addition to this analysis of ontological commitment, the preand post-tests were also scored for accuracy of responses, as the tests consisted of qualitative problems physics problems that were designed to be sensitive to the existence (or removal) of substance-based misconceptions of electric current. Method