Beyond Bloom's Taxonomy: Rethinking Knowledge for the Knowledge Age


Carl Bereiter and Marlene Scardamalia


Date of Download: 7 August 2002

Please Note:
This article is published and copyrighted material. It has been made available to you only for the purposes of this course and not for any further distribution.

From two quite different sources comes a similar message: Knowledge is far more important than has previously been realized. One source is the study of wealth creation and economic competition. From this source come such as-yet little understood ideas as knowledge-based economy, knowledge workers, and knowledge as an economic product and as a dominant 'means of production,' taking precedence over labor and capital (Drucker, 1993). The other source is cognitive research, now spanning three decades, on the nature of expertise. This research has demonstrated with great consistency and in many different domains that experts are distinguished from nonexperts mainly by the extent and depth of their knowledge, not by their mental abilities, thinking skills, or general cognitive strategies (Chi & Glaser, 1988).

These ideas have begun to have an impact on the thinking of educational reformers. In particular, many curriculum reforms are afoot that emphasize depth of understanding. Yet an examination of both the products and the rhetoric of many programs for educational change will reveal that they are based on the conception of knowledge that was current forty years ago, and whose roots go back not only to before the 'cognitive revolution' and before the advent of the 'knowledge society' but to before the printing press and the microscope. It is a conception that trivializes knowledge and subordinates it to a panoply of intellectual abilities and skills of doubtful teachability. It is a conception that fixes knowledge within individual minds and therefore can make little sense of the social and economic role of knowledge.

Our objective in this chapter is to advance new ways of looking at knowledge that are more consistent with current understanding and with the ascendant social importance of knowledge. The old way of conceiving of knowledge is well represented in an important and still influential work of four decades ago, the Taxonomy of Educational Objectives, Handbook I: Cognitive Domain (Bloom, 1956), more familiarly known as Bloom's Taxonomy. This taxonomy played an important role in expanding the scope of curriculum objectives and achievement testing beyond those of itemizable subject-matter content, but at the same time it served to entrench the idea that knowledge just is such items of content. In the taxonomy, Knowledge occupies the lowest of six levels of cognitive objectives. In explaining this level, the authors suggested that the reader

...think of knowledge as something filed or stored in the mind. The task for the individual in each knowledge test situation is to find the appropriate signals and cues in the problem which will most effectively bring out whatever knowledge is filed or stored. (Bloom, 1956, p. 29)

The higher levels of the taxonomy—Comprehension, Application, Analysis, Synthesis, and Evaluation—were conceived of as "intellectual abilities and skills." They constituted the person's capacity to operate on the contents of the mental filing cabinet. Contents of the filing cabinet might go out of date and need to be changed, but the intellectual abilities and skills would continue to serve the person throughout life. Accordingly, they were the objectives of most long-range significance for education (pp. 38-43).

These ideas should sound familiar. They are part of the rhetoric of contemporary educational reform. They do, of course, have some validity. Some knowledge does go out of date (although the great bulk of what we know does not). What one can do with knowledge is crucial. But the limitations of these ideas, which we will explore more fully in later sections, can be glimpsed by considering how they could serve to answer two questions: (1) What does it mean to have a deep understanding of something? (2) In what way is a knowledge worker different from any kind of white-collar worker? A conception of knowledge that is of no help, that may even get in the way of answering questions such as these, is surely in need of updating itself.

Emergence of the Knowledge Society

Taken at face value, terms such as 'knowledge-based economy' and 'knowledge society' do not carry much meaning. When was there ever an economy that was not based on knowledge applied to producing or acquiring tradeable goods? What society does not embody the accumulated knowledge of its past? To impart meaning to these terms, we need to look at historical changes in the status of knowledge.

Throughout most of the human past, knowledge was embedded in traditional practices, tools, and myths. Practices, tools, and myths evolved over time, and in this sense knowledge grew. But, said Whitehead (1925/1948, p. 91), "[t]he process of change was slow, unconscious, and unexpected." Major advances occurred in response to new conditions, which continues to be the case in traditional societies. But there was little capacity to envisage and create new conditions. That would have required a detachment of knowledge from its embedding practices and myths, so that ideas could be manipulated and recombined in a speculative way.

Such a detachment or, as we shall say, objectification of knowledge began to take place in all the major civilizations a few thousand years ago. Many social factors conspired to bring this about, but the invention of writing systems undoubtedly provided a powerful tool (Olson, 1994). Philosophers, historians, mathematicians, and theologians began to appear. These, along with attendant librarians and scribes, became the first knowledge workers. Knowledge work differed from that of the present day in three important respects, however: (1) There was no general conception of a state of knowledge, which advanced through the cumulative contributions of knowledge workers. (2) Knowledge work was not applied to practical arts. Such knowledge continued to be embedded in the various trades and crafts, evolving slowly and with little crossover from one trade to another. (3) Knowledge work of any kind was the province of a tiny minority of the working population.

With the Industrial Revolution came the deliberate application of knowledge in the advancement of practical arts. Yet, according to Alfred North Whitehead (1925/1948, p. 92), it was not until the nineteenth century that we got "the full self-conscious realization of the power of professionalism in knowledge in all its departments, and of the way to produce professionals, and of the methods by which abstract knowledge can be connected with technology, and of the boundless possibilities of technological advance." This led to what he called "disciplined progress," progress achieved through the deliberate and orderly pursuit of solutions to theoretical and technical problems.

The next and current stage in the evolution of knowledge work is not very well defined. Peter Drucker, who coined the term 'knowledge society,' dates its emergence as the end of World War II. The change, he said, is that knowledge began to be applied to knowledge, whereas previously it had been applied to materials and to work. This rather barren definition may gain more meaning through use of an analogy. What comes to the silversmith's workbench is silver and what leaves it is still silver, but it is worth more than it was before. The silversmith's work has added value to the silver. Similarly, what comes to the knowledge worker's desk is knowledge and what leaves it is also knowledge, but the knowledge worker has done something to add value to it. What arrives might be market research; what leaves might be the draft of a marketing plan. What arrives might be functional specifications for a new software application; what leaves might be technical specifications. What arrives might be excerpts from airline schedules; what leaves might be an itinerary. What arrives might be student journals; what leaves might be entries by the teacher that stimulate further thought. What arrives might be customer complaints; what leaves might be ideas contributed to a design database. Or what leaves might be only an organization of the complaints into useful categories. Knowledge work may go on at different levels. It need not always be creative, but it must in some fashion render the knowledge more meaningful, accessible, reliable, relevant, or applicable to particular purposes. Clearly, it takes knowledge in order to do this. In order to organize the customer complaints into a useful set of categories, you need more that 'classification skills,' whatever that might mean. You need to understand the product or service customers are complaining about and you need to understand the contexts within which those complaints are arising and what capacities the organization has for responding. This, as we take it, is the sense in which knowledge work means applying knowledge to knowledge.

We are not intellectual historians. The preceding sketch is highly derivative and no doubt flawed. But its main theme is, as far as we are aware, uncontroversial. That theme is the gradual shift from knowledge being completely embedded in practice, myth, and artifact to its becoming objectified as abstract objects that are recognizable human creations and that can be described, compared, criticized, disseminated, improved, discarded, rediscovered, and so on. An important question, accordingly, is whether education has kept up with this transformation. "Professionalism in knowledge," which Whitehead dated from the nineteenth century, can certainly be found in many classrooms, but the literature on teacher professionalization would indicate that it is still to be fully realized. As for students functioning as knowledge workers, engaged in adding value to knowledge, however, this is virtually unheard of except at postgraduate levels. Bringing such a conception into elementary and secondary schooling is a new challenge, which later sections of this chapter will address.


Expert Knowledge

Research on the nature of expertise has been one of the most active areas of cognitive research. The earliest research on expertise, which set the paradigm for much of what followed, dealt with experts at chess. This was a fortunate choice, because there was already a firmly established conventional belief that the essence of skill at chess is reasoning ability. To this day, chess is fostered in some school programs as a means of teaching children to think (Marjoram, 1987). However, it was found that chess grand masters did not differ from lesser players in reasoning out the consequences of possible moves. The difference was that grand masters only reasoned about good moves. This seemed to deepen the mystery, but another finding offered a clue. Grand masters had a phenomenal ability to memorize whole chessboard configurations at a glance. Yet it was not that they had generally superior memory abilities. The ability was confined to chessboard configurations and—most interesting of all—only to meaningful configurations , which is to say, arrangements of pieces that might actually occur in a well-played game. Give them a randomly arranged chessboard configuration and their ability to memorize it was not much better than that of a novice.

Through a series of ingenious experiments and analyses, Chase and Simon (1973) deduced that the secret of the chess experts' performance was that they knew from memory tens of thousands of patterns in which chess pieces might be arranged. A particular chessboard configuration would consist of a combination of several of these patterns. For them to memorize a chessboard layout in a few seconds was no more difficult than it would be for you to memorize a sequence of 30 letters of the alphabet when they form a readable sentence of four or five words—as compared to what it would be like to memorize the same letters randomly arranged. Thus the secret is knowledge, but not a kind of knowledge that had been appreciated before. It is far vaster in quantity that the knowledge we commonly recognize. It is not readily verbalizable; those who have it are typically not even aware of it. And yet it is integral to what we generally regard as highly intellectual activity.

Similar experiments have been done in many other fields—various sports, medicine, computer programming, weaving, music. In all of them the same kind of evidence shows up indicating vast knowledge of patterns relevant to the activity. But not just any patterns will do. Given textbook physics problems involving pulleys, inclined planes, and the like, novices as well as experts can sort the problems into meaningful categories; but the categories of the novices are based on surface features—pulley problems in one category, inclined plane problems in another, and so on—whereas the categories of the experts are based on the laws of physics that are applicable.

Principled pattern knowledge evidently lies behind a great deal of what we commonly attribute to mental abilities and intuition. The novice physician looks at a patient and sees a dumpy person with thin, oily hair; the skilled internist looks at the same person and sees a familiar pattern of thyroid deficiency. The novice editor sees a 40-word sentence and breaks it into two disjointed sentences. The expert editor sees a 25-word noun clause and changes it to a free modifier, thus rendering the 40-word sentence easy to read. The star quarterback or infielder decides in a split second on a play so brilliant that it takes the sportscaster a minute and a half to explain its rationale. Could the player actually have thought all that out? Of course not. It was a matter of recognizing a principled pattern—principled in the sense that it encapsulated the principles elaborated by the sportscaster.

The lesson in this, however, is not that we should be teaching students tens of thousands of patterns. If there is a place for pattern training at all (which there may well be) it will be at advanced stages of mastering very specific jobs or problem domains. Experts do not generally learn patterns directly but as a byproduct of striving to achieve goals in their domains. Their pattern knowledge is principled by virtue of their pursuing principled goals, trying to get to the bottom of things, reflecting on their mistakes, making use of principles to understand what they are doing and the phenomena they encounter (Bereiter & Scardamalia, 1993).

A better way of approaching the educational implications of pattern knowledge may be the following: With experience, everybody acquires pattern knowledge. That is just how our brains work. They are pattern-learning devices (Margolis, 1987). The only question, therefore, is what kind of patterns will be learned. Will they be patterns that support resourceful, principled action and that keep being elaborated and enriched as experience grows or will they be patterns bound to surface appearances, limited in their potential for growth, and supporting mindless, stereotyped behavior? Schooling should be able to do something about this, even though most pattern learning will take place outside school.

Related to the principled aspect of pattern knowledge is another finding well supported in many areas of expertise. It is the importance of depth of knowledge. Among the correlates of chess expertise, accuracy of memory for chess positions is one that distinguishes among levels of skill across the whole wide range covered by chess point ratings. An equally strong correlate, however, is the kind of knowledge obtainable from textbooks: knowledge of chess strategies, important games, and the like (Charness, 1991). Perhaps the most striking evidence of the importance of depth of knowledge comes from a study by Lesgold and LaJoie (1991). This is one of the few studies that has compared experts with experienced nonexperts rather than with relatively inexperienced people. The people were employed in troubleshooting defects in airplane test instruments. Lesgold and his colleagues used a wide range of psychological and performance assessments to find out what distinguished the more expert from the less expert troubleshooters. They did not differ in general mental abilities or in troubleshooting strategies. Thus, although troubleshooting is clearly a thinking task, experts did not appear to differ from nonexperts in thinking skills. They all knew how to troubleshoot. They did not differ in their basic knowledge of electronics, either, however. Where they differed was in their knowledge of the actual devices they worked with and on. The experts had, according to Lesgold, a very deep understanding of these devices, whereas the others had a more superficial understanding.

Depth seems to be the unifying theme in the bulk of studies on expertise: getting beneath the surface, making contact with the underlying patterns and principles that give meaning and support intelligent action. Understanding in the ordinary sense, marked by the ability to explain, may be a part of it, but deep knowledge goes beyond that to encompass patterns that inform action yet are not available to consciousness. Bloom's Taxonomy circles around the idea of depth but never really seizes it. Many of the sample test items at higher levels in the taxonomy seem to require knowledge of some depth. One item at the Analysis level presents the following information:

Galileo investigated the problem of the acceleration of falling bodies by rolling balls down very smooth planes inclined at increasing angles, since he had no means of determining very short intervals of time. From the data obtained he extrapolated for the case of free fall. (Bloom, 1956, p. 151)

Examinees are then asked to identify the assumption implicit in the extrapolation. To do so would require grasping the logic of Galileo's ingenious procedure, which in turn requires understanding extrapolation at quite an abstract level as well as having a ready command of concepts of acceleration due to gravity and rolling friction. But this is not how the item is advertized. Instead, it is put forth as an item testing the ability to recognize unstated assumptions, as if there were such an ability that would generalize across subject areas. Of course, one must know what unstated assumptions are, but that is also knowledge—knowledge of what Ohlsson (19xx) calls "abstract models." If research on expertise teaches us anything relevant to this example, it is that having students spend time solving hidden assumption problems while neglecting deeper understanding of physics would be just the wrong way to go.

Alternatives to the Filing Cabinet Model

The psychology that informed Bloom's taxonomy was a blend of behaviorism, which was the dominant scientific psychology of the day, and a commonsense view, which has come to be called 'folk psychology' (Bruner, 1990; Stich, 1983). From behaviorism came the choice to define educational objectives in behavioral terms and to base the hierarchy of levels "on the idea that a particular simple behavior may become integrated with other equally simple behaviors to form a more complex behavior" (Bloom, 1956, p. 18). From folk psychology came the mind-as-container metaphor (Lakoff & Johnson, 1980), which led to treating knowledge as the contents of a mental filing cabinet.

Behaviorism has since waned as a theoretical program, but the container metaphor persists. Cognitive psychology and artificial intelligence research have elaborated and specified the contents of the container. In addition to the consciously accessible stored facts envisaged by Bloom and his colleagues, the mind as envisaged in mainstream cognitive psychology contains a large number of unnoticed items of factual knowledge or belief and additionally contains rules, which are the basis of skills. These items, furthermore, may be organized into larger structures such as scripts, schemata, semantic nets, production systems, or mental models. Anderson (19xx) presents strong evidence for believing that complex skills such as computer programming and geometry proof are built up one rule at a time.

Thus the container metaphor is far from dead. For the first time in centuries, however, it has begun to come under serious attack. The most direct challenges come from research on memory, which indicates that remembering is not a matter of retrieving an intact object but of reconstructing something anew each time remembering occurs (Schacter, 1989). Another kind of attack is based on the ability of connectionist or neural net AI programs to demonstrate how systems can act as if guided by rules and concepts without actually containing any such objects (Bechtel & Abrahamsen, 1991; Bereiter, 1991). Other attacks are based on the biological implausibility of the container metaphor (Churchland, 1986). These have been strengthened by mounting evidence that people are born with a great deal of what functions as knowledge but that can hardly be mental content fitting the filing cabinet metaphor (Hirschfield & Gelman, 1994). A very different line of criticism comes from research on situated cognition and on the social and discursive bases of knowledge. Here the general argument is that much of what folk psychology assumes to be internal is actually external, sustained by the cultural practices and ongoing discourses that people are engaged in. "[T]he whole point of the discursive turn in cognitive psychology," say HarrŽ and Gillett (1994, pp. 39-40), "is to get away from mythical mental entities."

Unfortunately, it is quite beyond the scope of this chapter to discuss how it is possible to have a knowledgeable mind without stored mental content. A crude analogy will have to suffice. Your comfortable old shoe does not contain a representation of the shape of your foot. When the shoe is not on your foot it looks like any other shoe, but the molecules in the leather have gotten arranged so that when you put the shoe on it moulds itself to your foot (and not to just any foot of equivalent size). Imagine the brain as a supershoe that can mould itself to many different feet that it has encountered in the past. It is a shoe that remembers but that does not contain memories.

From an educational standpoint it is quite legitimate to ask, however, what is wrong with "mythical mental entities" if they produce a theory that works in practice. Folk psychology surely does work well in everyday practice. It works well in education so long as we are dealing with knowledge that can be adequately described by a smallish set of sentences or rules. In those cases, teaching people the sentences or rules is one way (and often a fairly good way) to impart the knowledge to them or at least to get them started in mastering a skill. It is the time-honored way of teaching arithmetic algorithms, for instance. And when a student is getting something wrong in a nonrandom way, it often helps to think of a rule that fits what the student is doing and then try to get the student to see the inadequacy of the rule. This is a prevalent strategy in the 'conceptual change' approach to teaching in science and mathematics (Scott, Asoko, & Driver, 1992).

Many important kinds of knowledge cannot be adequately described by sentences or rules, however, at least not by a small enough number to be of practical use in education. The use of English prepositions is one example. Rules fail, and a list of actual usages fills a book. Number sense, as distinct from executing arithmetic algorithms, is another (Bereiter & Scardamalia, in press). On a larger scale, literary skills and the learning that occurs in reading good literature are important kinds of knowledge that can hardly be described in terms of mental content at all. In general, the deeper the knowledge the more difficult it is to describe it in propositions and rules and the less useful it is to deal with it in that way.

The higher-level test items in Bloom's Taxonomy, like the previously cited one concerning Galileo and falling bodies, seem from a more modern perspective to call for knowledge beyond what can be readily stated. As Bloom and his colleagues well recognized, students could understand gravity, acceleration, and friction at the level these are typically presented in textbooks and yet be unable to explain the logic of Galileo's experiment or to identify its unstated assumptions. Something more is required, and the authors of the Taxonomy sought to capture it by defining a hierarchy of general intellectual skills—what are now called 'domain independent' skills, meaning that they are not tied to any particular knowledge domain but apply across the board. But even if we acknowledge that there could be such a domain-independent skill as 'recognizing unstated assumptions,' students could be well endowed with it and still fail the test item because their understanding of physics and/or extrapolation lacked the necessary depth and coherence.

It is this deeper, more coherent understanding that contemporary research tells us we should be pursuing in education. In order to do so in a purposeful manner, however, we need ways to think about knowledge that allow us to be reasonably clear and definite about what we are trying to achieve yet do not require reducing knowledge to itemizable objects in the mind. Bloom's Taxonomy fails, but what is a practical alternative?

In combination, two of the ideas presented so far provide a basis for a more adequate treatment of knowledge objectives. These ideas are, on one hand, the connectionist view of mind as being knowledgeable without containing knowledge items, and, on the other hand, the objectification of knowledge as abstract objects that people create, modify, and use. The two ideas come together in the following proposition:

The educated mind has various abilities and dispositions. Paramount among these are the ability and the disposition to create and work with abstract knowledge objects.

Mapping Levels of Understanding

[S]ome teachers believe their students should "really understand," others desire their students to "internalize knowledge," still others want their students to "grasp the core or essence" or "comprehend." Do they all mean the same thing? Specifically, what does a student do who "really understands" which he does not do when he does not understand? Through reference to the taxonomy as a set of standard classifications, teachers should be able to define such nebulous terms as those given above. (Bloom, 1956, p. 1)

These words from the foreword to Bloom's Taxonomy indicate that its authors aimed to elucidate the nature of understanding, at least in behavioral terms. However, as we have seen, the Taxonomy does no such thing. With Knowledge occupying the bottom level of the hierarchy, Comprehension occupies the second. The four levels beyond that are not levels of understanding but are levels defined by kinds of performance that depend on but do not clearly reveal understanding. The Taxonomy captures the strong intuition that there are levels involved in knowledge. We all recognize a low level characterized by a smattering of facts and deeper levels characterized by coherently connected principles. But the levels of theTaxonomy do not constitute levels of knowledge in this sense.

It is also common to recognize levels of capability with respect to knowledge, ranging from some lowly ability to parrot statements to abilities to do intelligent things with knowledge. The Taxonomy offers us levels of this kind, but it is not clear that they are very useful levels. They are testable, to be sure, but do they correspond to reasonable educational objectives? The authors of the Taxonomy evidently thought so: "Teachers building a curriculum should find here a range of possible goals in the cognitive area" (Bloom, 1956, pp. 1-2). Many educators have used it in curriculum planning. The Taxonomy is cited among the sources of the Common Curriculum, for instance, now being introduced in Ontario schools (Ontario, Ministry of Education and Training, 1993). Few would dispute that a good educational program will engage students in plenty of comprehending, applying, analyzing, synthesizing, and evaluating. But these do not constitute a curricular sequence. No sane educator would propose starting with knowledge in grade 1, moving to comprehension in grade 2, application in grade 3, and so on. Rather, the levels of the Taxonomy refer to processes that need to go on in concert at all levels, supposedly leading to the attainment of worthy objectives. By not indicating what those objectives might be, however, the Taxonomy has, we suggest, encouraged schools to continue an emphasis on low-level factual knowledge as the only kind of knowledge that has been clearly identified, supplementing factual instruction with various activities believed to foster domain-independent higher-level skills.

These criticisms leave the impression that Bloom's Taxonomy represents a failed attempt to map levels of understanding, and that now, after four decades of cognitive science, we should be able to do it right. This would be to misread the lessons of cognitive research, however. A sounder conclusion would be that it is futile to try to define levels of understanding that are applicable across domains, or even across objects within the same domain. Possibly the authors of Bloom's Taxonomy tried it and found it couldn't be done. Suppose we have worked out six levels of understanding Huckleberry Finn and six levels of understanding the principle of natural selection. What correspondence could we expect to find between the two hierarchies? Would level 4 on one have any meaningful relationship to level 4 on the other? In order to define levels that applied to both the literary work and the scientific principle, we would need to move to a high level of abstraction. The result might be a set of indicators—essentially a set of test item types—much like those of Bloom's Taxonomy , or it might be levels of cognitive functioning, perhaps based on the Piagetian stages. In any case, we should have lost any sense of what a deep understanding of Huckleberry Finn or of natural selection would consist of.

We have been criticizing Bloom's Taxonomy for its failure to address depth of understanding, but we too have skirted the question of what depth of understanding is. The definition we shall offer is so simple that it will appear circular: Having a deep understanding of something means understanding deep things about it. Although you might argue that there is more to deep understanding than this, you can hardly argue that there is less. And if you accept that deep understanding must include understanding deep things about the matter in question, then you must abandon hope of a general taxonomy of levels of understanding.

The deep things to be understood about Huckleberry Finn have no necessary resemblance to the deep things to be understood about natural selection, even at a very abstract level. Experts may disagree about what the deep things are. This is invariably the case with literary works. Furthermore, we should expect only a weak ordering, even among people who agree on the deep principles. That is, it may be clear that understandings B and C are both 'deeper' than understanding A, but there may be nothing to say about the depth of B relative to C. But in all cases we are talking about the depth of what is understood, not about the cognitive processes or skills associated with that understanding. In assessing someone's understanding, we might well make use of the kinds of questions and tasks presented in Bloom's Taxonomy , but these would only be tools for getting at the substance of the student's understanding and they would be useless without a conception of what the understandings are that we are looking for. Those understandings would invariably be domain-specific.

The performance standards being developed as part of the New Standards project (National Center on Education and the Economy, 1995) place a heavy emphasis on understanding. Of the eight major standards in science, four of them begin, "The student understands...." The middle-school standard for life sciences concepts reads:

The student understands:

  • structure and function of cells, tissues, and organs;
  • reproduction and heredity, including genes, traits, and learning;
  • regulation and behavior, especially the roles of senses and hormones;
  • population and ecosystems, including food webs, resources, and energy;
  • evolution, in particular, species, diversity and adaptation, variation, extinction.


For any of these concepts there are things to be understood far beyond the grasp of middle school students, but there are also simple understandings—about the senses and about biological diversity, for instance—that even young children can be expected to have picked up without study (Keil, 1989). Obviously something in between is expected, but what would an intermediate level of understanding consist of? Along with the standards are model evaluative activities with samples of student performance intended to indicate an appropriate level of understanding. Thus, there is an implicit scaling of levels of understanding, but the hard work of determining what actually constitutes adequate understanding of the various concepts is left to be worked out locally, and will need to be done separately for each concept or network of concepts.

The hard lesson to be learned is that there is no shortcut to setting objectives of understanding. The curriculum designer or teacher has to get deeply into the material to be learned, to see what is there that warrants understanding, where understanding can go awry (as research on misconceptions shows that it frequently does), and what the deeper understandings are and whether these are within reach of the students. A model of this kind of analysis may be found in the work of Hunt and Minstrell (1994). In high school physics they identified a large number of what they call 'facets.'

A facet is a convenient unit of thought, an understanding or reasoning, a piece of content knowledge or a strategy seemingly used by the student in making sense of a particular situation. For the most part, our facet descriptions paraphrase the language used by students as they justify their answers, predictions, or explanations.... An example from free-fall and projectile motion is "Horizontal motion keeps things from falling as rapidly as they would if they were moving straight downwards." (Hunt & Minstrell, 1994, p. 52)

Over 200 facets were identified just within the areas of mechanics and electricity. Most of these would count as partial or faulty understandings, while a few comprise the principles intended to be taught. The facets were incorporated into a software application called DIAGNOSER, which not only identifies the facets of individual students' understanding but checks for consistency. One could imagine a set of facets developed to capture various novice and expert understandings of Huckleberry Finn. These would be specific to that novel and would have very little overlap with facets developed for The Brothers Karamazov, for instance. There could also be facets for literary theory, which would consist of understandings about literature in general, although these in turn would be very different from understandings of domains such as history, and facets pertaining to history in general would be very different from understandings of a particular event or epoch. Itemizing the facets of understanding relevant to standard school subject matter could occupy a substantial industry. Unfortunately, such an industry is unlikely to develop; and the task is too formidable for practitioners to carry out independently. And so, instead, we have scope-and-sequence charts and curriculum guidelines which merely name concepts without addressing what constitutes understanding, and general schemes like Bloom's Taxonomy, which have no direct relevance to issues of understanding.

Levels of Approach to Knowledge

If there is no way to characterize levels of understanding in general, and if identifying levels of understanding in particular domains is impractical, this raises doubts about the value of any general scheme of educational objectives. Yet there is an obvious need for educators to take a large view.

One kind of large view is provided by the various societies and education ministries that have produced curriculum frameworks. A number of these are cited in Performance Standards (National Center on Education and the Economy, 1995) [ed: cite appropriate chapter in handbook instead]. These can be important in forming the major topical boundaries within which educational activities are to go on and in bringing about changes in those boundaries in response to new knowledge or new societal concerns. It is not reasonable to expect them to do much more than that, however.

Another kind of large view is provided by developmental schemes. There is the well-known Piagetian scheme of development from sensori-motor to concrete to formal logical operations. There are neoPiagetian schemes which do not propose uniform development across all domains but nevertheless propose the same general form of cognitive development in different areas of competence (Case, 1985; Fischer, 1980; Karmiloff-Smith, 1992). These developmental schemes have the virtue, lacking in Bloom's Taxonomy, of indicating, for a student at any particular level of attainment, what a reasonable next step should be. Case's model, in particular, has shown itself to be valuable in designing instructional interventions based on developmental levels (Case, 1992). [note to editor and reviewers: we assume there will be chapters to cite that will explain developmental approaches more fully. otherwise this section might need to be expanded.]

Something important is still missing, however. Curriculum guidelines specify areas in which knowledge is to be pursued. Developmental models lay out a continuum of increasingly sophisticated performance, applicable to various curriculum areas. But we have not taken account of the student's role in the pursuit of understanding and competence. Constructivist thinking convinces us that students need to be more than willing workers. They must be agents, not merely recipients. But what are they to be agents of? Surely, the answer for a three-year-old cannot be quite the same as the answer for a thirteen-year-old, but what is supposed to change? Taking a cue from the briefly sketched history of knowledge, we can speculate that there should be developmental changes in how students approach knowledge itself.

What follows is a provisional scheme of levels of working with knowledge.1 The levels may be thought of as levels of objectification, which start with viewing knowledge as a mental state and extend to viewing it as consisting of abstract objects. Of the seven levels, the first three are fairly well documented in the developmental and writing research literature. The seventh level corresponds to a mature scientific approach to knowledge. In between, however, are three levels that mark important and little-recognized transitions that could form educational objectives in the school years:

  1. Knowledge as individuated mental states. Research on children's theories of mind (Astington, 1993) suggests that a concept of knowledge begins to emerge with the realization that one person may know something that another does not. Prior to that, knowledge is not distinguished from 'the way things are.' In one common type of demonstration, the child is shown a puppet playlet in which the puppet puts candy in a drawer and then goes away. The candy is then removed from the drawer and put in a cupboard. When the puppet returns, the child is asked, "Where will Bozo look for the candy?" The typical three-year-old will predict that the puppet will look in the cupboard because "that's where it is." The typical six-year-old will predict that the puppet will look in the drawer because it "doesn't know" that the candy was moved. Thus, implicitly, there is some entity—a fact—which a person may or may not know.
  2. Knowledge as itemizable mental content. According to Donald Graves, a favorite writing topic of six- and seven-year-olds is "What I Know About. . ." something of interest to them. This implies a view of knowledge as items of mental content that can be accessed and reported. At this level, however, knowledge tends to be reported in the order in which it comes to mind. This is true not only of young children but of unsophisticated writers at all ages (Flower, 1979). A consequence of this 'knowledge-telling' strategy (Bereiter & Scardamalia, 1987) is that knowledge tends not to be reflected upon in the course of reporting it, so that writing or telling contributes relatively little to knowledge development.
  3. Knowledge as representation. Trying to communicate what one knows to a reader, taking into account what the reader already knows and is in a position to understand, represents an important advance not only in language skills but also in how knowledge is conceived. It is no longer just something in the head to be expressed but is something to be represented, shared, interpreted by others. This stage is indicated by expressions of audience awareness and by the use of explanatory devices such as analogies and examples.
  4. Knowledge as viewable from different perspectives. An important step toward objectification occurs when students see that the same knowledge can appear in different contexts and can be viewed from different perspectives. To illustrate, we take a classroom experiment by Ward and Thiessen (19xx), which made use of CSILE, a student-generated hypermedia database (Scardamalia & Bereiter, 1994). Third-graders, studying endangered species, each produced a CSILE note describing a different endangered species in their region, its habitat, source of endangerment, and so on—a fairly common activity up to this point. However, using CSILE's note-linking capabilities, the students all linked their notes to appropriate points on a map of a region, thus allowing students to see what species were near each other or shared the same habitat. They also linked their notes to a phylogenetic tree, allowing them to see biological relationships among their species. Finally, the students themselves worked out a set of reasons for endangerment, and linked their notes to appropriate boxes in a diagram of these reasons, thus affording a third perspective on the same body of knowledge.
  5. Knowledge as personal artifacts. Although constructivism is widely endorsed by teachers, it is not so common for students to view themselves as constructors of knowledge. Viewing oneself as constructing knowledge is quite a step beyond viewing oneself as constructing knowledge representations (Level 3). One kind of knowledge construction students can grasp readily is the construction of theories. CSILE provides labels for several different kinds of contributions to collaborative knowledge building, one of which is "My theory." Notes thus labeled become discussable knowledge objects. Students will comment on one another's theories: "I agree with your theory," "My theory is like Jamie's theory," etc. After discussion, a group of students may begin referring to "our theory" or "our solution."
  6. Knowledge as improvable personal artifacts. When children first begin producing "My theory" notes, they are inclined to treat these as personal opinions, and thus entitled to the protected status accorded to personal opinions in modern classrooms, or else as guesses at the truth, to be checked by consulting authoritative sources, which provide the correct theory. This, of course, is not how theories are viewed among scientists. They are viewed as provisional solutions to theoretical problems, always subject to improvement. Viewing a theory in terms of what it can and cannot do it, what its virtues are and where it is in need of improvement thus represents a major advance in conceptualization of knowledge. Such a more advanced conception is conveyed by a fifth-grade student who, asked how she would know when she had learned, replied:
    I think that I can tell if I've learned something when I'm able to form substantial theories that seem to fit in with the information that I've already got; so it's not necessarily that I have everything, that I have all the information, but that I'm able to piece things in that make sense and then to form theories on the questions that would all fit together.
  7. Knowledge as semi-autonomous artifacts. In the preceding quotation, knowledge is still being described as something personal. This corresponds to what Kieran Egan defines as the 'philosophical' stage of educational development, in which there is a focus on "the general laws whereby the world works" but "this is not a process of expansion outwards along lines of content associations, it is a closer charting of the context within which the student exists. It is not a further expansion from the self, but rather a closer approach toward the self" (Egan, 1979, pp. 51-52). Movement to the seventh level involves recognizing that knowledge objects, like other constructed objects, take on a life of their own and can be considered independently of their personal relevance. This does not mean that you become dispassionate and 'objective' in a sense that implies extreme rationality and detachment. You may feel strongly attracted or repelled by an idea, but you recognize that the idea remains unaffected by your feelings, that other people may feel differently about it, and that the idea may turn out to have virtues or flaws that you are presently unaware of and that may change your attitude toward it. Thus, at this level, knowledge objects become things that one can relate to, use, manipulate, and judge in various ways, just like other things in the real world. At this stage, then, 'knowledge work' becomes readily comprehensible. Like any other kind of productive work, it involves adding value; but in this case the things one adds value to are knowledge objects.

Let us be clear that these are levels of approach to knowledge. Functioning at a high level does not imply either a high level of understanding of subject matter or a high level of skill in working with knowledge. It implies, rather, that students are in a position to take a sophisticated, constructive role in the pursuit of understanding and to engage in the kinds of purposeful activities that develop knowledge-processing skills. The work of actually achieving deep understanding in a domain and competence in working with knowledge in that domain remains to be done; the hierarchy sketched here pertains to the level at which students can participate in that work.

What is immediately striking about these seven levels of approach to knowledge is how much they are neglected in school practice and how alien they are to discussions of curriculum and standards. Normally, the first level could be safely ignored, because children can be expected to acquire an awareness of knowledge as a mental state through ordinary social experience. But when a child does not acquire it, this could be a sign of something seriously amiss (Astington, 1993, Ch. 9); yet it is not a part of any screening program we know of. The second level does receive attention in whole-language approaches but seldom in traditional approaches to writing. It should be noted that itemizing mental content is not the same as responding to factual questions. Even toddlers can do the latter, but searching memory for anything one knows about a topic implies a more mature cognitive stance. The third level, which involves representing knowledge in communicable form, is the only one regularly addressed in statements of educational objectives and standards. It is usually formulated in the context of writing abilities. The previously cited Performance Standards, for instance, call at each school level for the ability to produce a report that organizes appropriate facts and details, excludes extraneous and inappropriate information, and uses a range of strategies for effective communication.

That leaves four levels, however, that receive virtually no attention in education. These are the levels at which students begin to deal directly with knowledge as such. Prior to that, knowledge is a sort of aftereffect of their interactions with texts, people, and the material world. And that, we suspect, is how it remains for most people. They never become 'knowledge workers' insofar as their own knowledge is concerned; much less do they see the world's knowledge as something they can work with, add to, and modify. Yet we are expecting them to assume roles in a knowledge society that require just that kind of engagement with objectified knowledge.

How could schools foster development through the upper levels of the hierarchy? This is a large question, and only the beginnings of answers are available. But it is a researchable question—researchable both through psychological experiments and through classroom action research. The first step is to recognize that the challenge exists, and that has been the purpose of this chapter. Our own research over the past decade has explored the potential of the CSILE learning environment to promote higher levels of approach to knowledge (Scardamalia, Bereiter, & Lamon, 1994; Scardamalia & Bereiter, 1996; Scardamalia, Bereiter, Hewitt, & Webb, in press). Among upper elementary and middle school students we have seen clear evidences of levels 4 through 6. Level 7, which is characteristic of mature scholars and scientists, may not be attainable until late adolescence; but a school population functioning at level 6 would amaze the world.


Schools could function as places where students become proficient in all aspects of knowledge work, including its creation. To do so, however, fundamental changes in underlying epistemology and psychology are required. In this chapter we targeted the 40-year-old Taxonomy of Educational Objectives as embodying the kind of epistemology and psychology that needs changing. Knowledge, according to Bloom's Taxonomy , is analogous to the contents of a mental filing cabinet. The higher-level objectives of education are what Bloom and colleagues called "intellectual abilities and skills," which enable people to adapt knowledge to new situations and use it for various purposes. Such a concept fails when it is stretched to cover such contemporary concerns as the creation and allocation of knowledge, knowledge work, knowledge executives, and a knowledge-based economy in which knowledge is conceived as a means of production. Thirty years of research on the nature of expertise have shown, moreover, that what distinguishes experts in all fields is their deep knowledge, not their general "intellectual abilities and skills."

Drawing on insights from recent work on the nature of mind, situated cognition, expertise, and processes of knowledge creation in the sciences, we have tried to outline a different way of thinking about knowledge. Gone is the filing cabinet metaphor and its attendant trivializing of knowledge. Instead, we have a conception of minds as being knowledgeable without containing itemizable knowledge. The challenge, accordingly, is to develop forms of objectives and standards compatible with this view. Two watchwords of a new approach are depth of understanding and objectification. Because depth of understanding implies understanding deep things about something, no global hierarchy such as that of Bloom's Taxonomy can suffice. The 'deep things' need to be identified separately for each object of understanding. Objectification, however, can be characterized in a more general way. Objectification means the prying loose of knowledge from individual mental states and collective practices, making it an object of constructive activity in its own right. Historically, objectification emerged over the course of many centuries. For individuals, we have sketched a series of seven levels or stages that represent increasing ability to deal with knowledge as such—to construct it, view it from different perspectives, criticize it, improve it. Thus, progression through these levels represents an educational objective of particular significance to a knowledge society.


Anderson, J. R. (199xx). Rules of the mind.

Astington, J. W. (1993). The child's discovery of the mind. Cambridge, MA: Harvard University Press.

Bechtel, W., & Abrahamsen, A. A. (1991). Beyond the exclusively propositional era. In J. H. Fetzer (Eds.), Epistemology and Cognition (pp. 121-151). Dordrecht: Kluwer Academic Publishers.

Bereiter, C. (1991). Implications of connectionism for thinking about rules. Educational Researcher, 20, 10-16.

Bereiter, C., & Scardamalia, M. (1987). The psychology of written composition. Hillsdale, NJ: Lawrence Erlbaum Associates.

Bereiter, C., & Scardamalia, M. (1993). Surpassing ourselves: An inquiry into the nature and implications of expertise. La Salle, IL: Open Court.

Bereiter, C., & Scardamalia, M. (in press). Rethinking learning. In D. R.Olson &. N. Torrance (Eds.), Handbook of education and human development: New models of learning, teaching and schooling Cambridge, MA: Basil Blackwell.

Bloom, B. S. (Ed.). (1956). Taxonomy of educational objectives: Handbook 1. Cognitive domain. New York: David McKay Company, Inc.

Bruner, J. S. (1990). Acts of meaning. Cambridge, MA: Harvard University Press.

Case, R. (1985). Intellectual development: Birth to adulthood. Orlando, FL: Academic Press.

Case, R. (1992). The mind's staircase: Exploring the conceptual underpinnings of children's thought and knowledge. Hillsdale, NJ: Erlbaum.

Charness, N. (1991). Expertise in chess: The balance between knowledge and search. In K. A. Ericsson & J. Smith (Eds.), Toward a general theory of expertise: Prospects and limits (pp. 39-63). Cambridge: Cambridge University Press.

Chase, W. G., & Simon, H. A. (1973). Perception in chess. Cognitive Psychology, 4, 55-81.

Chi, M. T. H., Glaser, R., & Farr, M. (Ed.). (1988). The nature of expertise. Hillsdale, NJ: Erlbaum.

Churchland, P. S. (1986). Neurophilosophy: Toward a unified science of the mind-brain. Cambridge, MA: MIT Press.

Drucker, P. F. (1993). Post-capitalist society. New York: HarperCollins.

Egan, K. (1979). Educational development. New York: Oxford University Press.

Fischer, K. W. (1980). A theory of cognitive development: Control and construction of hierarchies of skills. Psychological Review, 87, 477-531.

Flower, L. S. (1979). Writer-based prose: A cognitive basis for problems in writing. College English, 41, 19-37.

Graves, D. R. (1983). Writing: Teachers and children at work. Exeter, NH: Heinemann Educational Books.

HarrŽ, R., & Gillett, x. (19xx). Discursive psychology.

Hirschfeld, L. A., & Gelman, S. A. (Eds.). (1994). Mapping the mind: Domain specificity in cognition and culture. New York: Cambridge University Press.

Hunt, E., & Minstrell, J. (1994). A cognitive approach to the teaching of physics. In K. McGilley (Ed.), Classroom lessons: Integrating cognitive theory and classroom practice. (pp. 51-74). Cambridge, MA: MIT Press.

Karmiloff-Smith, A. (1992). Beyond modularity: A developmental perspective on cognitive science. Cambridge, MA: MIT Press.

Keil, F. C. (1989). Concepts, kinds, and cognitive development. Cambridge, MA: Bradford/MIT Press.

Lakoff, G., & Johnson, M. (1980). Metaphors we live by. Chicago, IL: University of Chicago Press.

Lesgold, A. M., & Lajoie, S. (1991). Complex problem solving in electronics. In R. J. Sternberg & P. A. Frensch (Eds.), Complex problem solving: Principles and mechanisms (pp. 287-316). Hillsdale, NJ: Erlbaum.

Margolis, H. (1987). Patterns, thinking, and cognition. Chicago: University of Chicago Press.

Marjoram, D. T. E. (1987). Chess and gifted children. Gifted Education International, 5, 48-51.

New Standards. (1995). Performance standards: English language arts, mathematics, science, applied learning. Rochester, NY: National Center on Education and the Economy.

Ohlsson, S.

Olson, D. R. (1994). The world on paper: The conceptual and cognitive implications of writing and reading. New York: Cambridge University Press.

Ontario, Ministry of Education and Training. (1993). The Common Curriculum, Grades 1-9, Working document. Toronto: Publications Ontario.

Scardamalia, M., & Bereiter, C. (1994). Computer support for knowledge-building communities. The Journal of the Learning Sciences, 3(3), 265-283.

Scardamalia, M., & Bereiter, C. (1996). Adaptation and understanding: xx. In

Scardamalia, M., Bereiter, C., & Lamon, M. (1994). CSILE: Trying to bring students into world 3. In K. McGilley (Ed.), Classroom lessons: Integrating cognitive theory and classroom practice. (pp. 201-228). Cambridge, MA: MIT Press.

Scardamalia, M., Bereiter, C., Hewitt, J., & Webb, J. (in press). Constructive learning from texts in biology. In K. M. Fischer & M. Kirby (Eds.), Relations and biology learning: The acquisition and use of knowledge structures in biology. Berlin: Springer-Verlag.

Schacter, D. L. (1989). Memory. In M. I. Posner (Ed.), Foundations of cognitive science (pp. 683-725). Cambridge, MA: MIT Press.

Scott, P., Asoko, H., & Driver, R. (1992). Teaching for conceptual change: A review of strategies. In R. Duit, F. Goldberg, & H. Neidderer (Eds.), Research in physics learning: Theoretical issues and empirical studies Keil, Germany: Schmidt and Klannig.

Stich, S. P. (1983). From folk psychology to cognitive science: The case against belief. Cambridge, MA: MIT Press.

Ward, D., & Tiessen, E. (199xx). .

Whitehead, A. N. (1925/1948). Science and the modern world (Mentor ed.). New York: New American Library.

1 The provisional nature of the scheme must be emphasized. We are involved in the early stages of a project titled "Knowledge-Building Indicators," which will develop and test a variety of ways to assess knowledge building in telelearning environments. The scheme presented here is a first pass, based largely on existing research, and will undoubtedly undergo substantial changes as new research is carried out