Education for Thinking
by Deanna Kuhn - 1986
The educational system has not been particularly successful in teaching thinking skills. In order to do so, it is necessary to identify what thinking skills are and understand their development. Current research and design of future research are discussed, as is the problem of transfer. (Source: ERIC)
Preparation of this article was supported in part by a grant from the Spencer Foundation.
The idea of education for thinking has ancient origins, but never has it been held in greater esteem than in recent history. In the early 1960s, a report by the Educational Policies Commission of the National Education Association titled The Central Purpose of American Education claimed, The purpose which runs through and strengthens all other educational purposesthe common thread of educationis the development of the ability to think.1 In the 1970s, the Rockefeller Foundation sponsored a commission to study the humanities in American life. A major conclusion of that commissions report reflected a similar belief: The Department of Education should define critical thinking as one of the basic skills that provides the foundation for advanced skills of all kinds.2 Today, the call for thinking as the focus of education continues to be heard in a wide range of reports criticizing the U.S. educational system.3
These calls have not gone unheeded. Early in the 1980s, the title of a major news article proclaimed Classes in How to Think Spring Up around the Nation.4 The article described a diverse set of efforts aimed at a wide range of age groups. In 1980, with other higher education institutions soon to follow suit, the mammoth California State University and College system issued executive order 338, mandating instruction in critical thinking for all students. This instruction was to be designed
to achieve an understanding of the relationship of language to logic, which should lead to the ability to analyze, criticize, and advocate ideas, to reason inductively and deductively, and to reach factual or judgmental conclusions based on sound inferences drawn from unambiguous statements of knowledge or belief. The minimal competence to be expected at the successful conclusion of instruction in critical thinking should be the ability to distinguish fact from judgment, belief from knowledge, and skills in elementary inductive and deductive processes.5
Are such ambitions likely to be realized? The available data suggest that to date the U.S. educational system has not been particularly successful in teaching thinking skills. Educational assessments repeatedly reflect that, as important as they may be, the so-called skills of thinking, reasoning, and problem solving remain the most difficult to impart to students. The reports that have proliferated in recent years paint a uniformly gloomy picture. The 1981 National Assessment of Educational Progress in reading and literature indicated that a majority of nine- to seventeen-year-old students are competent in initial comprehension of a passage, summary of its major theme or content, and superficial statements of personal reactions to it. However, students show little ability to analyze or evaluate a passage, drawing on portions of the text as evidence to support their judgments. Similarly, in a 1981 policy statement, the National Council of Teachers of Mathematics called for new teaching methods that address problem solving. Students are often competent in basic computational skills, the statement claims, but are unable to apply those skills in situations that require quantitative reasoning and problem solving.
The evidence, then, suggests that education for thinking may be a goal on the order of goals like world peace. There is striking accord as to the desirability of the goal, especially for such a traditionally pluralistic field as education, yet only minimal progress has been made in realizing it. Why has there been such limited success in implementing such a highly and widely valued goal?
One possibility is that educators lack the necessary knowledge base with respect to thinking skills and the mechanisms that govern their development. The limited ways in which the new computer technology has been employed in education provide a hint that this is the case. The computer has been widely lauded as offering great promise for the enhancement of education, notably in the area of developing reasoning and problem-solving skills. Classroom use of computers, however, is overwhelmingly for drill in routine facts and skills. The small minority of software developers who have devoted their efforts to programs for teaching thinking skills have tended to make very general (and undocumented) claims that their programs teach problem-solving or reasoning skills. The generality of these claims gives the impression that no one has been able to provide these program developers with much of an idea of what such skills are in any more specific sense. They are thus confined to generalities, unable to make specific claims that a particular activity involved in their program does or does not foster a particular cognitive skill. The software industry, one gets the sense, is doing the best it can within the confines of the state of the art of educational knowledge. A similar judgment might be made in the case of many other, non-computer-based experimental programs for teaching thinking. In sum, then, what evidence there is from students academic performance suggests that the schools have not been effective in fostering thinking skills; at the same time, there is little sense that the education profession possesses very clear ideas as to how to remedy this state of affairs.
Why should the field of education lack the knowledge it needs to teach thinking skills? If there were ever an area in which research ought to be able to make a substantial contribution to education, the implementation of education for thinking would be a promising candidate. We are not dealing in one of those realms of values, where empirical research is unable to dictate choices and decisions. The goal appears to be well articulated, and it is in the development of methods of implementation that assistance is required. Yet it appears that research has not filled this role. I quote again from the NEA report The Central Purpose of American Education:
Thus, in the general area of the development of the ability to think, there is a field of new research of the greatest importance. It is essential that those who have responsibility for management and policy determination in education commit themselves to expansion of such research and to the application of the fruits of this research. This is the context in which the significant answers to such issues as educational technology, length of school year, and content of teacher education must be sought and given. A new emphasis on this field by educational research may be expected to yield great dividends to the individual citizen and to the nation as a whole.6
In the twenty-five years since the publication of this report, little progress has been made in developing the sorts of research programs the report advocates. In this article, I explore why greater progress has not been made in acquiring knowledge of thinking skills and their development that would serve the needs of educators seeking to realize the goal of education for thinking. Part of the reason, I will suggest, has to do with the fields of educational and psychological research and ways in which they may have been poorly equipped to undertake this task. Yet another factor, almost certainly, is the inherent difficulty of the task.
The extent of this difficulty can be appreciated by contemplating what I will refer to as the transfer dilemma. The objective of the educator endeavoring to teach thinking skills is to equip the student with general skills that can be applied across diverse content domains. As soon as a teacher undertakes this task, however, he or she confronts the fact that thinking cannot be observed or exercised or taught in a completely general, contentless form. To think, one must think about something. But once thinking centers on a particular content, ambiguity creeps in as to what general skills are being observed or exercised or taught. For example, if we wish to give students experience in a skill such as evaluating evidence, we must give them evidence about something or other to evaluate. As soon as we do, the activity runs the risk of involving nothing more than specific thoughts about that something or other. Still, it is most unlikely that thinking and reasoning strategies are entirely specific to particular content. A thinking skill exhibited in one content domain no doubt has at least some potential for transfer to other content domains. Thus, thinking skills are neither completely wedded to specific content or contexts of use, nor are they completely general. The challenge to the researcher (and it is clearly a formidable one) is to identify them on this continuum. This, I shall claim, is equivalent to studying the thorny issue of transfer.
THE IDENTIFICATION OF THINKING SKILLS
Scientific research might contribute to the realization of education for thinking in two distinct ways. First, it might serve to identify with some precision just what these thinking skills are and how they manifest themselves across a wide variety of contexts. Second, it might establish what the mechanisms are that govern the acquisition of these skills. Without knowledge of both types, efforts to teach thinking skills or to guide, promote, or evaluate such teaching are severely handicapped.
The place one might expect to find research evidence of this sort would be in the educational and psychological literature on critical thinking. This literature has been of three major types. The first consists essentially of logical/rational analysis of the elements or features of sound thinking. A chapter by Ennis is a recent example. Ennis states at its outset that the chapter can be viewed as a set of guides for someone who wants to think rationally.7 Earlier examples of efforts of this type (though not necessarily having such an explicit prescriptive intent) are works by Black, Ennis, Smith, and Symonds.8 In many senses, this work is continuous with earlier theoretical analyses of thinking processes, for example, the classic works by Dewey and Mill,9 and indeed with the earliest philosophical works devoted to the analysis of rational thought.
A second type of literature is psychometric in character, consisting of the development and validation of instruments designed to measure critical thinking. The two most widely known examples are the Watson-Glaser Critical Thinking Appraisal and the Cornell Critical Thinking Test, developed by Ennis.10 Items on such tests tend to have been generated from the kinds of logical/rational analyses of critical thinking indicated above, or, more recently, to have been borrowed from existing tests. When the Educational Testing Service, for example, decided in the 1970s to develop a third subtest of its Graduate Record Examination to assess aptitude in analytic thinking, the initial item pool was generated from existing types of logical reasoning test items that the test constructors thought might have the desirable psychometric properties.11
The third type of literature is educational in nature, devoted to fostering critical thinking. For the most part, it is devoted to descriptions of programs or curricula designed to teach critical thinking. Occasionally, an educational program has been closely linked to a particular measure of critical thinking.12 More often, formal curricula devoted to teaching critical thinking are loosely based on literature of the first type and focus on acquainting students with some general principles of sound or critical thinking. The bulk of such literature has been aimed at students at the college level,13 but some recent efforts have been addressed to younger students. An example is a curriculum designed by Lipman and his colleagues, one of the most ambitious of this sort.14 Lipman undertakes to teach children the elements of sound thinking, as defined by classical logic. The following passage from Hurry Stottlemeiers Discovery, a text for middle-grade students, provides an example:
And then Harry had an idea. A sentence cant be reversed. If you put the last part of a sentence first, itll no longer be true. . . . Now, its true that all planets revolve about the sun. But if you turn the sentence I around and say that all things that revolve about the sun are planets, then its no longer trueits false!"15
How do we know such principles will be useful to children or young adolescents in making their own thinking more effective? This is a hard question to answer, for conspicuously missing from all of the critical-thinking literature are empirical data on the nature of reasoning (either valid or faulty) that children and adults in fact engage in, in the course of their own thinking. Without such data, how can we know what sorts of instruction in thinking will be useful to people and how can we assess whether that instruction has been effective?
Empirical studies intended to secure data of such a nature are not totally lacking, though one must search outside the traditional educational and psychological literature on critical thinking in order to find them. For the most part, neither are they to be found in the research literature in cognitive or developmental psychology. Despite the fact that a central concern of both of these fields is that of thinking and its development, in both fields there exists a long tradition of studying reasoning strategies that are often of great potential generality and significance (for example, goal recursion, studied by Simon in cognitive psychology, or isolation of variables, studied by Piaget in developmental psychology), but studying them in the context of artificial tasks that bear an uncertain relation to the cognitive activities that arise in the subjects own experience. Thus, despite the potential of such work in offering insight into basic cognitive processes, it remains unclear how much it reveals about peoples thinking in everyday contexts.
Surprisingly, perhaps, the research data that would appear most pertinent come from the field of social psychology, in particular recent work that reflects that fields evolution toward a cognitively based social psychology. A book by Nisbett and Ross, titled Human Inference: Strategies and Shortcomings of Social Judgment,16 brings together recent research on judgment (notably the work of Tversky and Kahneman17), attribution, and other social-inference processes to illustrate shortcomings in the performance of average individuals in inference tasks involving familiar social content. For example, individuals typically fail to take into account prior information about base rates in making predictions: If told that John is good at math, subjects judgments of the likelihood that he is an accountant rather than a lawyer typically are not affected by knowledge regarding overall frequencies of accountants and lawyers in the population. Other studies illustrate how prior knowledge or beliefs, when they are taken into account, distort the interpretation of evidence. In one study subjects were presented a set of peoples scores on a test of ability to delay gratification and a test of ability to resist temptation to cheat, and asked to use these scores to estimate the relationship between these two variables.18 The majority of subjects greatly overestimated the strength of the relationship, presumably due to their belief that these two variables ought to be related. In contrast, when asked to estimate correlations between neutral variables, for example, pairs of numbers, subjects tended to greatly underestimate the true correlation.
While such findings may reveal certain shortcomings in human inference processes, they reflect as well some serious shortcomings of the studies themselves. Though the content may be familiar, the data on which subjects are asked to base their inferences are necessarily presented in hypothetical (i.e., artificial), neatly packaged, and highly impoverished form (e.g., the John is good at math example cited above). Only rarely does a task involve nonartificial content that might be of genuine concern to subjects (for example, capital punishment or welfare in several studies by Nisbett or Ross and their colleagues). Subjects responses to such stimuli are of necessity similarly impoverished. For example, of two alternatives posed by the interviewer, the subject attributes an outcome to cause A rather than cause B. The choice between A and B is the sole source of evidence the researcher has regarding the subjects reasoning processes.
As a result of the first shortcoming, the relation that subjects performance in such inference tasks bears to the inference processes they use in thinking about real phenomena of genuine concern to them remains uncertain. In what contexts might it be important for someone to predict whether John is an accountant or a lawyer and would they ever do so having such limited knowledge of John? Indeed, this is a major criticism of the Nisbett and Ross book:19 Such shortcomings in laypersons inference processes may be confined to the context of artificial problems posed to them by psychologists. Nisbett and Ross acknowledge this uncertainty in their concluding chapter in stating that the errors they have described may not be significant and may in fact be cost-effective (i.e., not errors at all) in real life. They nevertheless derive educational implications from their review of studies, advocating programs designed to illustrate to people the sorts of inferential errors they are prone to, presumably in an attempt to inoculate them against such errors by heightening their awareness. Whether this method of education for thinking is likely to be the soundest one is a question we shall return to later.
Interestingly, the much earlier social psychology literature on attitude and opinion represents one of the few bodies of research that deal with peoples cognitions about real (nonartificial) phenomena (effects of smoking on health, for example.)20 Unfortunately, as a consequence of both its theoretical framework and its methodological orientation, this work is even more severely handicapped, relative to more recent work in social psychology, in elucidating the cognitive processes that underlie a subjects judgments. Attitude (e.g., toward smoking) is treated as a unidimensional construct measured by a point on a quantitative scale ranging from highly unfavorable to highly favorable. The processes affecting such attitudes traditionally have been conceptualized within an associationist framework: Investigations have focused on the effects of such variables as frequency of exposure (to a persuasive message) or the amount of persuasive material contained in the message (number of arguments or pieces of evidence). Such studies provide little illumination of the nature or form of reasoning that underlies, for example, a subjects attitude toward smoking.
The more recent work in cognitive social psychology is similarly handicapped, though not as severely. Attribution of an outcome to cause A rather than cause B provides indirect, at best suggestive evidence regarding the nature or form of causal reasoning underlying that choice. Furthermore, in virtually all social psychology research, findings are limited by the fact that analysis is exclusively at a group level, the research strategy in general being one of the identifying variables that influence judgments of subjects treated as a group. Adequately understanding cognitive processes such as reasoning or inference, one can argue, requires investigation of these processes as they occur within the individual.
In sum, there exists little research evidence regarding the cognitive processes that are involved in peoples thinking about real topics contemplated in everyday life. Would it be correct to conclude that there is little general interest in what people think about such topics? Clearly not. Somewhat paradoxically, perhaps, billions of dollars and vast technical expertise are devoted to ascertaining with considerable accuracy what people think about all sorts of social and political topics. Why they think what they dothat is, the cognitive processes that yield the judgments and inferences they makein contrast, is something we know very little about.
As interesting and significant as such data would be, are they in fact critical to the educators objective of fostering thinking skills? One might accept the argument that efforts to improve thinking must be based on sound knowledge of the nature of the thinking that people in fact engage in. But perhaps the appropriate place to examine that thinking is in the context of the school activities that are the central concern of educators, that is, in students efforts to master academic subject matter. Indeed, cognitive psychologists recently have begun to turn their attention to analyses of the cognitive processes involved in school tasks, suggesting that this might be the most appropriate context in which to focus investigation of thinking skills. As fruitful and germane as much of this work may turn out to-be, an exclusive focus on the thinking involved in school learning implies that school learning is important for its own sake. It neglects the fact that underlying a societys formal education of its young is the faith that this education will make a difference in the quality of life outside of and following school. The experience of going to school cannot be justified on intrinsic grounds; hence, it is out-of-school and out-of-laboratory functioning that must remain the ultimate criterion in evaluating the success of educational endeavors. It is thus the thinking that occurs beyond the confines of the formal classroom that educators ultimately must care about and seek to improve; it is this thinking that we must understand.
I have suggested that a foundation of basic knowledge about the way people think is lacking, and as a result little progress has been possible in identifying with any specificity the thinking skills that might be the appropriate objectives of educators concerned with the improvement of thinking. How do things compare with respect to the second way in which it was proposed that research ought to contribute to education for thinkingIdentifying the mechanisms that govern the acquisition of thinking skills?
THE DEVELOPMENT OF THINKING SKILLS
One finds a paradox in the educational literature on fostering effective thinking. The distinction underlying it is one of teaching people about thinking versus engaging them in it. Much of the educational literature on critical thinking is addressed to teaching principles of sound thinking. Yet there exists a long and distinguished theoretical literature in the field of education, one dating back to the beginning of the century, reflecting the view that the only effective way to teach people to think is to engage them in thinking. Deweys classic How We Think certainly reflects this view,21 as does the writing of other respected educational theorists of his time. Percival Symonds, for example, in Education and the Psychology of Thinking, wrote:
In order to learn to think one must practice thinking in the situation in which it is to be used and on material on which it is to be exercised. . . . In short, practice in thinking itself is necessary for the improvement of thinking. There is no substitute for the actual wrestling with real problems in the development of thinking.22
This view remained prevalent several decades later. To quote once more from the NEA report, The Central Purpose of American Education:
The rational powers of any person are developed gradually and continuously as and when he uses them successfully. There is no evidence that they can be developed in any other way. They do not emerge quickly or without effort. . . . Thus, the learner must be encouraged in his early efforts to grapple with problems that engage his rational abilities at their current level of development, and he must experience success in these efforts.23
Given this unusual consistency in theoretical position over seven or eight decades, it is quite striking to discover that there exists virtually no empirical research literature that pertains to it. Instead, experimental educational programs have focused on teaching students about good thinking, from the earliest efforts by Glaser24 and Osborn,25 whose programs were devoted to topics such as definition, evidence, inference, scientific method, and propaganda, to current programs such as the one by Lipman described earlier. Efforts such as Lipmans to teach general principles of sound thinking are likely to suffer the same fate as many of the intervention efforts in education and psychology: The competencies that are the object of instruction may be displayed in the narrow contexts in which they are taught but largely fail to generalize to a wider range of contexts in which they are equally appropriate. These all-too-characteristic failures of generalization limit not just the practical utility of such efforts but also their scientific interpretability. One remains unsure of what individuals have grasped in the original teaching situation and to what extent the process that transpired resembles the process by which such attainments occur under more natural conditions.26 What is at issue, in other words, is the fundamental problem of transfer that we considered earlier and shall return to again as the heart of the matter in understanding and teaching thinking skills.
Little formal evidence is available regarding the effects of the newer programs on the thinking abilities of their participants. The results of the early experimental programs, however, substantiate the concerns just raised. Glaser did not assess the effectiveness of his program in any broader context than that provided by an earlier version of the Watson-Glaser test, for which he reports a significant increase in scores following the program.27 Osborn, however, whose instructional program focused on analysis of propaganda, included in addition to mastery tests of the instructional material a subsequent test of pupils resistance to propaganda (in the form of a persuasive communication on capital punishment). Both immediate and delayed attitude shifts following the communication were no less for experimental subjects than they were for controls (not exposed to the educational program), leading Osborn to a conclusion in fact very similar to those quoted above:
The failure of the type of approach used in this study does not mean that it is impossible to teach critical thinking, Rather, it suggests that new attacks be made on the problem. . . . Possibly critical thinking can be developed best when pupils are taught in such a manner, throughout their school experiences, that they must constantly use information in problem-solving situations and in such a manner that they are constantly forced to make tentative conclusions as a result. In other words, it is just possible that the way to teach critical thinking is to give pupils long-term practice in it.28
The new attacks advocated by Osborn some forty years ago have not materialized. There exists virtually no experimental investigation or controlled observation of students engaged in continuing efforts to grapple with problems that engage [their] rational abilities at their current level of development,29 despite the repeated statements by educational theorists for the past several decades that it is through such exercise that thinking ability develops.
The philosophy of exercise as the mechanism by which thinking skills are developed is well captured in a newspaper cartoon in which the housewife on her front doorstep looks blankly at the pollster questioning her. If youd argue with me it would help me decide, she comments. A point of view is indeed often lacking to the extent that someone has never had occasion to exercise it. When individuals are provided with an opportunity to support and refute their own and others assertions (especially by means of appeal to evidence and verifiability), their thinking strategies obtain exercise, and, with that exercise, the potential for development. Implicit in this philosophy is the assumption that critical thinking is not something to be taught and mastered as a specialized form of thought, distinct from noncritical thinking. Rather, it is through exercise that thinking becomes critical thinking.
Efforts to engage students in thinking, rather than to teach them about it, are made every day by inspired and energetic teachers in classrooms across the nation and world. Yet the research on fostering thinking skills that might inform their efforts has been limited almost exclusively to experiments instructing students in general rules or principles of good thinking. This is so despite the counterindications that this is the way to approach the teaching of thinking and despite the limited interpretability of such experiments.30
The research that would bear on an exercise-based view of the development of thinking skills would be essentially observational in character, focusing on the manner in which thinking strategies evolve in the course of their exercise, and is of course dependent on the previous research task I discussed, identification of the specific effective or ineffective strategies employed by individuals in the course of their thinking. Researchers in the relevant fields, I suggested, have been poorly equipped to address the first task, due to a long tradition of studying thinking disembedded from the contexts in which it occurs naturally. Similarly, researchers in these fields may have been ill equipped and ill disposed to engage in research of the second sort, having to do with mechanisms of development. Influenced in part by the domination of stimulus-response theoretical models, treatment-outcome models have to a very strong extent governed the design of research in psychology and education. Observational studies of behavior change initiated by subjects themselves over time are not readily assimilable to such models.
On the positive side, while it is true that researchers characteristically have eschewed real-world content in favor of artificial tasks and observational study of change over time in favor of treatment-outcome experiments, these choices may be due not so much to the intractability of such content or the infeasability of such research methods as they are to aspects of the historical evolution of modern psychology.31 There is thus reason to be optimistic that fulfillment of the research mandate contained in the 1961 NEA report is not beyond our reach and that a body of knowledge might in fact be produced that would provide a foundation for the realization of effective education for thinking. Let us look more specifically, then, at some promising signs.
THINKING IN CONTEXT
Research data, I have suggested, have not made the contribution they ought to make to the educational goal of teaching thinking skills. One reason to think this situation might change is the increasingly widespread concern about the effectiveness of education and belief that some ways to improve it need to be found. In addition, however, trends within the relevant research fields give cause for optimism.
Psychologists engaged in the study of cognitive development have been influenced most strongly by Piagets model of cognitive development. In one sense, Piagets model might seem an ideal one on which to base educational efforts to teach thinking skills. It describes a number of very broad, general, and powerful thinking strategies, for example, isolation of variables, associated with major stages of intellectual development. Orthodox Piagetian theory describes these strategies as very general attainments, not tied to any particular content or experiential context, and regards their emergence as minimally contributed to by specific experience.
The available research evidence now makes it appear increasingly unlikely that thinking strategies develop in this completely general manner. Accordingly, neo-Piagetian models that portray cognitive strategies as acquired in, and therefore initially wedded to, particular contexts have been increasingly influential.32 These strategies gradually become more independent of their original contexts and hence more general. Closely related to this conception has been increased recognition of the external, societal contribution to this process.33 These new models of acquisition in context promise to generate a quite different kind of research from that of their predecessors.
In a similar vein, psychologists engaged in the study of intelligence have made progress recently in defining intelligence in ways that make reference to real-life functioning.34 No longer is intelligence likely to be regarded solely as a capacity for processing and manipulation of abstract symbols, unrelated to an individuals effectiveness in real-life settings. A growing number of investigations of intelligence in natural settings, such as the workplace,35 relate to this theoretical development.
In a similar vein, psychologists engaged in the study of cognitive processes have shown an increased interest in modeling thinking processes in real-life content-rich domains, rather than in hypothetical laboratory problems like the well-known Tower of Hanoi. The virtue of the latter kinds of problem domains had been assumed to be their almost total impoverishment with respect to content, thus allowing the researcher to study thinking strategies in a pure form, unencumbered by the subjects knowledge, or lack of it, within a particular content domain. Cognitive psychologists have recently come to the view, however, that the cognitive strategies a subject uses are strongly affected by the knowledge base the subject has available in a particular content domain, something that had long been ignored as irrelevant to the question of strategy. The development of a new strategy in fact may depend on the acquisition of particular domain-related knowledge. Thus, the development of strategies of thinking and reasoning cannot be studied apart from the development of the relevant knowledge base.36 As a consequence, it is necessary to observe thinking strategies in the context of specific, content-rich domains, rather than in isolation from meaningful content.37
The significance of the research trends I have just noted is that they all point to an increasing focus on thinking in context. The implication with respect to the concerns addressed in this article is that researchers who are now undertaking study of the kinds of thinking skills of potential interest to educators are less likely to attempt to conceive of and study such skills in completely general, contentless form. Instead, they are likely to examine the kinds of thinking a subject brings to bear within a particular, content-rich domain. Such approaches, I believe, hold the promise of increased utility to educators concerned with developing thinking skills, enhancing the likelihood that their efforts might be built on a usable research base of knowledge about thinking skills and how they develop. I should hasten to add, therefore, that research of the sort I have been referring to is at this point still little more than a promise. Cognitive psychologists are quite likely right in their recognition that the acquisition of a certain knowledge base plays a critical role in the development of theories or strategies that bear on it, but this should not be taken to imply that the development of thinking consists of nothing but the accumulation of particular domain-specific bodies of knowledge. Only a few psychologists today would endorse this view.38 Once the interaction of thinking strategies with the particular content being thought about is recognized, the major tasks still remain of identifying these content-linked thinking strategies within the contexts in which they occur and understanding the ways in which they develop as the individuals knowledge base increases.
Little research has yet been conducted that addresses either of these tasks. As was noted earlier, cognitive psychologists in the last several years have shown a surge of interest in the analysis of school tasks, giving rise to a whole new field known as instructional psychology. Little of this effort has yet been addressed to what might be identified as thinking skills. The bulk of the work in the new field of instructional psychology has been devoted to reading comprehension and rudimentary mathematical computation, though there are a few notable exceptions. The work that comes closest to thinking is that on writing skills. Tellingly, this work has tended to focus on the procedural aspects of writing, for example, the retrieval of ideas from memory and their linguistic expression, rather than the more fundamental task of conceiving those ideas. The fact that the field of instructional psychology has devoted its initial efforts to the study of the more mechanical and traditional school tasks such as reading and computation, rather than to investigation of the thinking skills currently of such great concern to educators, is still further evidence of the challenges of addressing the latter, for researcher and practitioner alike.
There is thus reason to claim that the research work still lies ahead but that we are beginning to have a much clearer and potentially more fruitful idea of what that work is. Such work is compatible with a view of thinking skills as emerging in specific contexts, to which they are initially wedded, and gradually, by means of exercise, increasing in strength and in generality. What might a program of research associated with this view look like?
Initially, the researcher would be likely to focus on identifying particular thinking strategies or sets of interrelated strategies as they emerged in the course of a subjects efforts to solve problems or reason within some specific content domain. This work would begin with identification of the strategies a subject was utilizing in the initial efforts observed and would then become longitudinal in character, as the researcher traced changes in those strategies as the strategies were exercised by the subject in continuing efforts to grapple with the topic or problem. In so doing, the researcher would learn something about the nature of the strategies and the nature of the changes they undergo as they develop. Once progress had been made in both of these respects within a particular content domain, the researcher might begin to explore the issue of generality by examining related content domains for evidence of similar strategies. Ultimately, this work would also become longitudinal, as the researcher tackled the most difficult question of transfer across domains, by examining whether and how change over time in one domain manifested itself in another domain.
It is interesting to note that an educator who subscribed to the same broad model of thinking skills and their development that has been proposed might be likely to devise an educational program to foster thinking skills that had a number of the same characteristics as the research program just described. Within a particular subject area, issues and problems would be posed designed to promote students exercise of their existing thinking strategies. As this exercise led to the development and consolidation of new, more powerful strategies (and the gradual discarding of inferior strategies), the educator would endeavor to provide opportunities for students to apply these or similar new strategies in related contexts that allowed such application. I shall have something further to say below about what might be taking place as the student exercises, develops, and extends thinking skills in the course of such activities.
One further point about both research and educational programs is to argue the importance of their having a life-span focus. Until recently, developmentalists have focused their attention on the period from early childhood to mid-adolescence, a period during which physical maturation and formal schooling both are being completed, making it difficult to disentangle the various sources of influence on a childs developing cognitive competencies. As I have argued, it is the thinking outside of and beyond childhood schooling that must be the ultimate criterion in evaluating the success of educational endeavors, including (indeed especially) education for thinking. It is this thinking we most need to study and understand, from its origins in childhood through its development and use during adult years. Similarly, experimentation will be necessary to devise the most effective methods for promoting critical thinking at various periods across the life span. Only with a great deal of further study might we establish the portions of the life cycle during which education for thinking might be most effectively concentrated.
CONFRONTING THE TRANSFER DILEMMA
Research efforts of the sort I have described should provide insight into some general characteristics of the process by means of which thinking skills develop, even though, somewhat paradoxically, the skills themselves may not be totally general entities in the sense that both educators and researchers have tended to regard them. In particular, such work may shed light on the thorny problem of transfer.
To illustrate what I have in mind, I shall describe very briefly some of the work that my co-workers and I are currently involved in on the development of thinking strategies. The broad area of skill we are investigating is the ability to coordinate theory and evidence, for example, to identify evidence germane to a theory, to recognize the status of a piece of evidence with respect to different theories, to weigh evidence in order to evaluate competing theories, and to revise theories in the face of evidence. The skills involved are complex ones, as we find that subjects of all ages utilize existing theories as a means of interpreting evidence at the same time that they utilize evidence to evaluate and revise theories. One might, as a number of researchers have begun to do, regard cognitive development as a process of theory revision, that is, partially correct theories and mini-theories within particular subject domains are successively revised over the course of development until they eventually approximate the commonly accepted adult versions.39 Following from such a view, it has been suggested that the aim of education might be regarded as the facilitation of this process of theory revision.40 If so, it becomes very important to understand the kinds of skills I have described, having to do with the bringing of evidence to bear on theories. Almost certainly, such skills affect the process of theory revision and have some generality across related content domains.
Our work in this regard has focused on theories and evidence of causality, initially with artificial sets of evidence presented to subjects and more recently in asking subjects to think about real-world topics such as the causes of unemployment or school failure. In this work we have closely observed subjects over a period of months encountering essentially the same problem situation.41 The exercise in thinking strategies that they thereby receive has been sufficient to lead a majority of subjects to modify these strategies during the period of exercise. It is in observing these subjects that we have been able to learn something about the change process. The most striking findings are these: For the large majority of subjects, change appears to be a very gradual, uneven process. During the period of change, subjects use both more and less advanced strategies in conjunction with one another, within a single session as well as from one session to the next. The insight gained in one session often does not carry over to the next, and a particular challenge appears to be the abandonment of less adequate, invalid strategies. If subjects are already competent in the more advanced strategies, as evidenced by at least occasional display of these strategies early in the series of sessions, then why does the change process (to consolidated, exclusive use of the most advanced strategies) take so long? Our hypothesis is that during the period of observation subjects are not only acquiring exercise in the use of these strategies; they are also gaining in metacognitive understanding of themthat is, why they are the most useful strategies to apply to this task, what their range of application is, why their earlier strategies are inappropriate, and so forth, and this metacognitive development is a much more complex, laborious acquisition.
Such findings offer some suggestions with respect to the problem of transfer in fostering thinking skills. Accounts of transfer within traditional learning paradigms have focused on stimulus similarity as the critical feature governing the transfer of learned behaviors: To the extent that a new stimulus situation is similar to the one in which the behavior was originally learned, the behavior is likely to be elicited by the new situation. A different way to conceptualize transfer would be to focus on the behavior, rather than the setting or stimulus, associated with the display of that behavior. The research I described above suggests that exercise of a cognitive strategy serves not only to perfect its execution but also to promote metacognitive awareness of the strategy itself. Heightened awareness of the strategy itself as a tool in turn might increase the likelihood of the users recognizing the applicability of the strategy in other contexts. Similarly, increased metacognitive awareness of strategies that operate within limited contexts might promote recognition of the commonalities among these individual context-embedded strategies, thereby increasing their generality and thus their power. In both cases, increased metacognitive control of a strategy acquired through exercise promotes transfer and increases generality. Perhaps, paradoxically, exercise of strategies within very specific content-delimited contexts may promote their generalization, while didactic teaching of the strategy in general, contentless form may fail to achieve this same end.
If these ideas are at all on the right track, they suggest that the educators role is one of facilitator rather than teacher of thinking skills. This does not trivialize the educators role; quite to the contrary, it enhances it. The goal of getting students to exercise critical thinking may be quite clear, but again, as noted at the beginning of this article, the means are far from obvious and are likely to require considerable skill. Research has suggested that once problems become familiar, individuals tend to apply well-ingrained, more or less automatic routines to them, with little if any intervening thought. Such problems provide no opportunity for metacognitive deployment of strategic skills. The educator must therefore continually attempt to pose problems that are novel enough to require conscious, selective deployment of strategies but familiar enough to permit application of strategies within the students competence.
Despite the centrality of the development of critical thinking skills to the aims of education, until recently we have had only the most intuitive understanding of what a good teacher does in the course of meeting this challenge. Research by Collins and Stevens42 and some others is beginning to give us more specific knowledge of the behaviors exhibited by a skilled teacher in inquiry-based instruction. Such research efforts should become more specific as our knowledge of students thinking skills and the processes by means of which they develop increases.