|
How Explanations Facilitate Scientific Reasoning
Computational models of reasoning demonstrate that effective reasoning depends on the proper coordination of modes of representation with procedures of inference. These models together with the psychological studies of human reasoning described above suggest that people's proficiency in reasoning about phenomena depends on their ability to construct representations of problem content that include those dependency relations that are necessary to make the appropriate inferences. To coordinate representations and inferences in this way, people must achieve an understanding of the potential effects of all variables that can influence target phenomena and the dependency relations that account for those influences. Scientific explanations provide this understanding by specifying all explanatorily relevant dependency relations among the phenomena they explain and those variables that can influence the occurrence and/or manifestations of those phenomena. Some kinds of phenomena, that are objects of study in the physical sciences, can be explained by subsuming them under general statements that that describe fundamental laws of physics. Because these generalizations (which were derived through model-based reasoning procedures) are assumed to be true, scientists can reason about the occurrence and manifestations of the phenomena they subsume in the following way: First, the covering laws and observed antecedent and background conditions are represented in propositional form. Then, inferences are derived by operating on these propositional representations in accordance with formal rules of logic. However, subsumption under fundamental laws of physics is not sufficient to explain phenomena that are produced by special kinds of systems, such as those that are the objects of study in sciences like biology and psychology. To explain these kinds of phenomena, scientists must specify additional dependency relations that account for the effects of the organizational structures of the target system on the phenomena they produce. Moreover, most law-like generalizations in the biological and social sciences are not explanatory because they do not specify all relevant dependency relations. To explain phenomena that are produced by these kinds of systems, scientists must describe the mechanisms by which relevant components of the target system act and interact to produce the phenomena to be explained. A mechanism is a set of entities that act and interact in systematic ways within a specified context to produce the explanandum phenomena. Complete descriptions of mechanisms specify all relevant dependency relations that hold among relevant component entities within a specific context. Scientific reasoning about the phenomena produced by these kinds of systems involves (1) constructing a representation of the relevant causal dependencies that hold between system components within the context in which the target phenomenon occurs and (2) making inferences by mentally simulating the activities of relevant system components in accordance with the represented causal dependencies. Mechanistic explanations facilitate this form of reasoning by providing the understanding necessary to represent relevant causal relations in a way that facilitates the simulation of relevant system components. The kind of representation that serves this purpose is one that captures causal dependencies between both the actions of relevant system components and the way those actions are related to one another.
Current research in scientific cognition (i.e. the psychological underpinnings of scientific thinking) indicates that contemporary scientists explain and reason about these kinds of phenomena by (1) constructing a functional analog of the kinds of systems that produce them and (2) mentally simulating the activities of model components in accordance with the causal relations between model components.
Current Research in Scientific Cognition The search for the psychological underpinnings of scientific thought and activity is an interdisciplinary effort that involves co-operation among psychologists, philosophers, and other researchers in the cognitive and social sciences. Research in this area that focuses on the use of models and modeling has been of particular interest to science educators. The value of models and modeling to scientific research is well documented and there is a considerable amount of evidence that much of the training of professional scientists involves learning the models of a scientific community and developing facility with model manipulation. Models and modeling activities play many roles in science, such as making predictions, summarizing data, and providing heuristics for designing experiments. Our theoretical framework draws on research in scientific cognition that focuses on the use of representational models: those that serve as abstract representations of real world systems. These models are constructed to capture those commonalities in the class of systems they represent that are relevant to a specific purpose, such as prediction, explanation, or control. Researchers in scientific cognition have identified three kinds of cognitive skill that are essential to model-based reasoning that uses representational models. These skills are those that involve abstraction through generic modeling, model-based simulative reasoning, and coordinating knowledge claims with empirical evidence.
To explain a phenomenon, a representational model must capture all causal dependencies that are relevant to all aspects of the explanandum phenomena. Relevant aspects include its range of manifestations and contextual conditions that can produce, prevent, or alter those manifestations. Explanatory modeling consists of a cycle of activities that involve creating and transforming representational models until a model is achieved that can account for all of the relevant empirical evidence that is currently available. The initial model is constructed to be analogous to the class of systems in represents with respect to causal dependencies among relevant components. Then, that model is continually evaluated and refined by (1) mentally simulating its implications for the effects of potential precipitating, inhibiting, and modulating conditions in novel situations, (2) comparing the results of those mental simulations to relevant empirical evidence, and (3) using feedback from those comparisons to confirm, reject, or revise the model. There is significant evidence that the explanatory modeling practices employed by scientists are productive methods for conceptual change in science. These practices can help scientists identify areas of needed conceptual change and restructure a body of scientific knowledge to resolve incoherencies, such as logical and probabilistic inconsistencies and explanatory anomalies. Current research in science education demonstrates that engage ment in model-based reasoning activities can promote similar conceptual change in students.
The Role of Explanatory Modeling in Teaching Psychology
When properly integrated with other kinds of instructional methods, explanatory modeling activities can help students to better understand and evaluate psychological explanations and arguments (the public expression of reasoning). Evaluation presupposes understanding because students cannot properly evaluate what they do not properly understand. Evaluation is particularly important in the social and behavioral sciences where alternative theories often compete for acceptance. To understand the subject matter of psychology, students must be able to establish relations between psychological explanations, the facts these explanations purport to account for, and relevant empirical evidence. GLN Consulting can recommend ways of helping students learn to establish these relations by bringing together information resources, explanatory modeling activities, and instructional strategies. Learning to establish these kinds of relations can help students develop a better understanding of the conventions and idiosyncrasies of psychological arguments and explanations. ___________________________________________________________________________
1. Consider, for example, an early and now classic study by Wason and Johnson-laird (1972). In a series of experiments, they found that only a small number(12%) of subjects were able to solve a simple logical problem when presented in abstract form, but a much larger number of subjects (60%) were capable of solving a formally equivalent version of the same problem when the content involved material they were familiar with.
| 
