Loading...
Please wait, while we are loading the content...
Similar Documents
Running head : REASONING THROUGH INSTRUCTIONAL ANALOGIES 1 Reasoning through Instructional Analogies
| Content Provider | Semantic Scholar |
|---|---|
| Author | Kapon, Shulamit |
| Copyright Year | 2012 |
| Abstract | This paper aims to account for students’ assessments of the plausibility and applicability of analogical explanations, and individual differences in these assessments, by analyzing properties of students’ underlying knowledge systems. We developed a model of explanation and change in explanation focusing on knowledge elements that provide a sense of satisfaction to those judging the explanation. We call these elements “explanatory primitives.” In this model, explanations are accepted or rejected on the basis of (a) the individual’s convictions concerning particular explanatory primitives and (b) the fit of these primitives to current circumstances. Data are drawn from clinical interviews with 3 high school students who worked through a bridging analogies tutoring sequence on the existence of the normal force in mechanics. Methodologically, our work involves fine-grain analysis of process data and explicit principles of empirical accountability; we believe it marks a methodological advance over most previously reported empirical studies of analogical reasoning. Running head: REASONING THROUGH INSTRUCTIONAL ANALOGIES 3 Over the last three decades, analogies and analogical reasoning have attracted the interest of many researchers. A core focus of attention has been how analogies foster understanding in some new situation or domain (the target), by comparing it to a more familiar one (the source). Analogies have been studied using diverse research methods and from a wide range of perspectives. Arguably the most influential theory of analogical reasoning is Structure Mapping Theory (SMT) (Gentner, 1983, 1989). SMT suggests that analogies provide insight by mapping between a source and a target that share a common structure. That is, objects and relations between them in one domain (source) are mapped onto similar ones in the other domain (target), leading to inferences about other possible objects and relations in the target. Employing a slightly different perspective, the Multi Constraint Theory (Holyoak & Thagard, 1989) suggests that analogical inference is the result of interplay between three constraints—structural similarity, semantic similarity, and pragmatic importance—which need to be satisfied simultaneously. Both of these theories have been studied extensively through laboratory experiments with humans and also via computational modeling. Other computational models describe analogy-making in terms of parallel processing and stochastic architectures, which are claimed to be closer to the way people reason (e.g., Hofstadter, 1995). Amid the complexity and diversity of perspectives, most if not all attempt to respond to a key issue concerning the effectiveness of an analogy: how one evaluates the degree to which the candidate transferred knowledge is applicable to the target domain and whether the analogical inference seems plausible. For instance, in an entry on computational modeling in analogical reasoning in the Encyclopedia of Cognitive Science, Kokinov and French (2003) define this as a core issue in analogical reasoning, along with mapping, representation building, and retrieval. Evaluation of plausibility and applicability has been modeled as being mediated chiefly by structural similarity across domains (e.g. Falkenhainer, Forbus, & Gentner, 1986; Forbus & Gentner, 1989) and pragmatic goals (e.g. Holyoak & Thagard, 1989; Keane, 1996). The present study does not contradict these claims but theorizes that there is a much more direct influence of the individual’s knowledge about the target domain on the evaluation of the analogical inference, particularly in the case of instructional analogies, where students do not generate the analogy but are required to reason with it. While other cognitive models of analogical reasoning acknowledge that processes of evaluation based on prior knowledge are important (e.g. Falkenhainer et al., 1986; Gentner & Colhoun, 2010), none so far has attempted to model these processes in detail. The current study may be seen, then, as a complement to other models of analogical reasoning: Using a methodology that is significantly different from the studies cited above, the paper extends the literature by examining in detail the ways in which knowledge of the target domain per se plays a role in evaluation of the plausibility and applicability of instructional analogies. As we will elaborate later in this section, this study attends specifically to differences in prior knowledge among individuals reasoning through the same instructional analogies and the role these differences play in the evaluation of analogical inferences. We will argue throughout this paper that differences among individuals are highly informative in understanding the process of evaluation of instructional analogies that have been or are intended to be used in classrooms, rather than simplified ones for experimental purposes. Recent findings in cognitive modeling of analogical reasoning (Holyoak, Lee, & Lu, 2010) hint at the necessity and potential of our methodological approach. These researchers presented subjects with analogies where source and Running head: REASONING THROUGH INSTRUCTIONAL ANALOGIES 4 target were embedded in stories (experiment 3), instead of using synthetic short analogs (experiments 1 & 2), and examined their subjects’ judgments regarding analogical causal attributions. We are not concerned here with the specifics of the findings, but rather with a phenomenon that was reported to be present only in the story condition, which is much closer to the classroom-oriented instructional analogies that we study. In summarizing the results from experiment 3, the researchers write that “closer inspection of the data suggested that the overall pattern of means might be masking important individual differences in task performance” (p. 715). In this study, we focus on those important differences between individuals’ thinking in situ. Analogical reasoning in professional science has been studied from many perspectives. Historical accounts of scientific discoveries (Gentner et al., 1997; Nersessian, 1992), studies of authentic scientific work in research laboratories (Dunbar, 1997), and studies of expert scientists’ problem-solving (Clement, 1988) all suggest that the generation of analogies and the reasoning stemming from these analogies play a central role in scientific practice, thought, and creativity. Analogy is also a common explanatory device in mathematics and science classrooms and textbooks (Dagher & Cossman, 1992; Glynn & Takahashi, 1998; Richland, Holyoak, & Stigler, 2004; Sarantopoulos, Tsaparlis, & Strong, 2004). Educational researchers who have explored the outcomes of learning with analogies in science have documented its positive influence on students' learning (Chiu & Lin, 2005; Clement, 1993; Dagher, 1995; Duit, 1991; Duit, Roth, Komorek, & Wilbers, 2001; Gilbert, 1989; Glynn & Takahashi, 1998; Treagust, Duit, Joslin, & Lindauer, 1992). However, researchers have also found that learning with analogies can entail the generation of scientific misconceptions (Dagher, 1995; Duit et al., 2001; Harrison & Treagust, 2006; May, Hammer, & Roy, 2006; Spiro, Feltovich, Coulson, & Anderson, 1989; Wong, 1993; Yerrick, Doster, Nugent, Parke, & Crawley, 2003). In our view, the generation of misconceptions in the target domain suggests that the knowledge students bring with them to the learning event is likely to affect the ongoing reasoning induced by these instructional analogies, and, in particular, prior knowledge is likely to affect the assessment of the plausibility and applicability of the analogical inference. With the exception of one study, none of the studies cited above considered the assessment of the plausibility and applicability of the analogical inference. The exception is John Clement’s work on bridging analogies. Clement (1988, 1993, 1998) argued that direct mapping from the source (which he refers to as the anchor) to the target fails when students feel that the source and target are too remote. Clement suggested that in such cases an intermediate analogy called a bridging analogy must be presented. Bridging analogies have been successfully tested as a teaching model in classroom experiments using pre-post assessments. Brown and Clement (Brown & Clement, 1989; Clement & Brown, 2008) also examined some process data in a few case studies of tutoring with bridging pedagogy where they compared analogies. These case studies led them to argue that a good instructional analogy enriches the learner's representation of the target beyond projecting abstract mapping relations onto preexisting features of the target, and the analogy thereby helps the student to accept the scientific view as reasonable. They hypothesized that some analogies are better at facilitating this process than others. A good analogy according to this view (Brown, 1994; Brown & Clement, 1989; Clement & Brown, 2008) is one where the elements of the source can be seen as candidates for reality (Harre, 1972). Candidates for reality mean that elements of the anchor/source can be seen to operate in the target. In fact, these features of the target are simply “not there” for the student before the analogy is presented. Clement and Brown suggest that candidates for reality provide an explanatory model of the phenomenon at hand, arguing that Running head: REASONING THROUGH INSTRUCTIONAL ANALOGIES 5 students see the features of the anchor/source as operating in the target, but also attribute explanatory power to them by accounting for the way the target system operates (Brown, 1993, 1994; Brown & Clement, 1989; Cheng & Brown, 2010). Clement, Brown, and Zeitsman (Brown & Clement, 1989; Clement, 1998; Clement & Brown, 2008; Clement, Brown, & Zietsman, 1989) use the term “brittle” to describe anchors that do not create a candidate for reality. A brittle anchor is an anchor that has a particular feature that cannot exist in the tar |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | https://cloudfront.escholarship.org/dist/prd/content/qt4sh5363n/qt4sh5363n.pdf?t=p4tsfk |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
