Experts attain their admirable status through experience of a special kind. In the domains in which they are acquiring their abilities, developing experts learn more from experience than do the rest of us. It is highly motivated learning in which they are engaged, whether it is the acquisition of baseball knowledge, chess moves, or computer programming. Their learning probably is also reflected upon more than is the learning in which others engage. What is learned in the domain in which they have their talent is better coded and remembered by the expert than that which is learned in other domains or that which is learned by others who are not as motivated to learn in a particular domain as the expert. What is learned by the expert appears to be linked better to other knowledge that the expert possesses. It also appears that such knowledge is more easily retrievable in appropriate situations and more transferable to new situations than is most other kinds of knowledge that they acquire or that is possessed by other individuals without the commitment that the expert has to excel in a particular domain. In the domain in which they have acquired their unique skills, experts usually perform appropriately and effortlessly, hallmarks of an exemplary performance, whether it be in chess, radiology, physics problem solving, taxi driving, or teaching.

Although expertise is universally admired, it was not systematically studied by psychologists and others until recently. Certain areas in psychology were related to contemporary scholarship on expertise, however, such as research on idiot savants; training studies on the acquisition of complex skills, such as diagnosis in medicine or in electronic systems repair; and research on the acquisition of psychomotor skills. The training of telegraphers at the turn of the century by William Bryan and his student, Noble Harter (1899), was a remarkable example of the latter, revealing the importance of practice, organization, and automaticity in the development of expertise:

Psychology had been preparing to change when deGroot's book was published in English. The domination of the field by behaviorists was quickly coming to an end, in part because they had no methods to study intrinsically interesting cognitive phenomena. DeGroot presented an interesting inquiry as he examined the differences between chess masters who win tournaments and those who are less able. He showed, for example, that the chess masters had spent thousands of hours staring at chess boards studying chess, that they had stored in memory the moves from thousands of games, and that they had remarkably fast and accurate pattern recognition abilities. These domain-specific abilities were not possessed by those who were less able at the game of chess. With further studies of chess experts by Newell and Simon (1972) and Chase and Simon (1973), we gain a perspective on the commitment it takes to become an expert in a particular field:

[A] student who spends 40 hours a week for 33 weeks spent 320 hours studying. Imagine spending more than ten years in college studying one subject, chess, and you get some appreciation of the time commitment of master level players. . . . It is reasonable to assume a chess master can recognize 50,000 different configurations of chess, not too far different from the number of different words an English reader may be able to recognize. (Posner, 1988, p. xxxi)

A second issue affecting the zeitgeist at this time was the recognition of the computer as a tool of singular importance. Computer scientists of that time period were beginning to develop expert systems and became aware of the deGroot studies, immediately recognizing their relevance for that work. An expert system for diagnosing infectious diseases or playing chess is nothing more or less than a computer program that can match the expert physicians or expert chess players (e.g., Duda & Shortliffe, 1983). To program a computer to perform as an expert requires studying human experts in a particular domain to learn how they solve problems, organize information in memory, perceive patterns, and so forth. Advances in artificial intelligence then, in part, are dependent upon some understanding of human intelligences. So the cognitive scientists from the fields of computer science and psychology both became enamored with expert-novice studies, and scores of such studies were carried out in just a few years (Chi, Glaser, & Farr, 1988). The deGroot studies, therefore, achieved instantaneous recognition and launched a wave of investigations about expert performance because psychology was changing paradigms and computer science was emerging as a powerful new field of inquiry.

What was learned from studies of expertise? Glaser (1987, 1990) has reviewed the literature on expertise and believes that about two dozen propositions about the development of expertise are defensible. Paraphrased and abbreviated, some of these propositions are:

1. Expertise is specific to a domain, developed over hundreds and thousands of hours, and it continues to develop.
2. Development of expertise is not linear. Non-monotonicities and plateaus occur, indicating shifts in understanding and stabilization of automaticity.
3. Expert knowledge is structured better for use in performances than is novice knowledge.
4. Experts represent problems in qualitatively different ways than do novices. Their representations are deeper and richer.
5. Experts recognize meaningful patterns faster than novices.
6. Experts are more flexible, are more opportunistic planners, and can change representations faster when it is appropriate to do so. Novices are more rigid in their conceptions.
7. Experts impose meaning on ambiguous stimuli. They are much more "top down processors." Novices are misled by ambiguity and are more likely to be "bottom up" processors.
8. Experts may start to solve a problem slower than a novice, but overall they are faster problem solvers.
9. Experts are usually more constrained by the task requirements and the social constraints of the situation than are novices.
10. Experts develop automaticity in their behavior to allow conscious processing of ongoing information.
11. Experts develop self-regulatory processes as they engage in their activities.

It is no surprise that in the initial rush to conduct studies of expertise none of the investigators thought to study teachers, even though many of those scientists have close ties to educational psychology. Pedagogical knowledge has always been devalued. It is erroneously believed by many legislators and parents that teaching is like child care, an occupation that seems to require no special abilities. Furthermore, teaching is seen as primarily women's work and, therefore, surely not as complex as, say, physics problem solving, a male domain. Finally, teaching is seen as an ill-structured domain of endeavor, without any inherent "proper" and "elegant" solutions such as occur in chess or physics. I would submit, however, that because teaching is ill-structured and takes place in a public setting that requires mastery of a complex social and political environment, it is inherently more difficult than many other professions. I believe it is a far tougher job in which to excel than is radiological analysis, computer programming, or tournament bridge play. It is for these reasons that my work focused on the development of pedagogical abilities. I wanted to study in more depth the wondrous performance of a small number of teachers that had enthralled me as I visited public school classrooms.

THE DEVELOPMENT OF EXEMPLARY PERFORMANCE

As experience is gained in teaching and other areas, such as computer programming, nursing, piloting an airplane, or playing chess, some individuals get better at what they do. Psychological theories of performance acquisition, corresponding to both common sense and empirical data, usually specify three levels or stages of development: (a) a novice stage, where errors are frequent; (b) an intermediate stage, where some consolidation of learning takes place and automaticity is developed; and (c) for some who work hard at acquisition of their skills, there is a stage where high levels of performance occur (Shuell, 1990, reviews some of these models). The developmental model of Dreyfus and Dreyfus (1986) is a bit more complex, but with adaptations, it fits the data we collected on the acquisition of pedagogical expertise. This heuristic model specifies five stages as an individual moves from novice to expert. It is described next, using examples from many fields, but primarily illustrating the model with examples appropriate to learning to teach.

Stage 1: Novice Level

We begin with the greenhorn, the raw recruit, the novice. At this stage the commonplaces of an environment must be discriminated, the elements of the tasks to be performed need to be labeled and learned, and the novice must be given a set of context-free rules. In learning to drive an automobile, for example, one is taught the meaning of yellow blinking lights, double-yellow lines, the "yield" sign, and other commonplaces of driving. When teaching a novice to drive a standard shift automobile, we might tell them to shift from first gear at 12 miles an hour, despite the fact that such a rule is terribly inadequate for driving on hills, slippery roads, or for certain engine/gear ratios. An expert driver shifts when it is appropriate to do so or when the sound of the engine reveals to that experienced individual that it is time to shift. Such "know-how" is virtually impossible to communicate to a novice. So to start them off, we often provide an inadequate, context-free rule, and it suffices until more experience is gained. There are similarities to these situations in learning to teach. The novice teacher is taught the meaning of terms like higher-order questions, reinforcement, and learning disabled. Novices are taught context-free rules such as "give praise for right answers," "wait three seconds after asking a higher-order question," and "never criticize a student." Understanding of the commonplaces and some context-free rules are what is needed to begin to teach. The behavior of the novice, whether automobile driver, chess player, or teacher is usually rational, relatively inflexible, and tends to conform to whatever rules and procedures they were told to follow. Only minimal skill at the tasks of driving, chess playing, or teaching should really be expected of a novice. This is a stage for learning the objective facts and features of situations. It is a stage for the gaining of experience and the one at which real-world experience appears to the learner to be far more important than verbal information, as attested to by generations of drivers, chess players, and student teachers. Student teachers and many first-year teachers may be considered novices.

Stage 2: Advanced Beginner Level

As experience is gained, the novice becomes an advanced beginner (see Bullough, 1989, for a case study of a teacher in the transition from a novice to an advanced beginner). Many second- and third-year teachers are likely to be in this developmental stage. This is when experience can become melded with verbal knowledge and where episodic and case knowledge is accumulated. Similarities across contexts are recognized. Without meaningful past episodes and cases to which to relate the experience of the present, individuals are unsure of themselves, they do not know what to do or what not to do. This is true of learning to drive a car, when the advanced beginner is suddenly confronted with fog, ice, or a traffic jam. In education we see advanced beginners having difficulty knowing what to do when a child challenges the teacher's authority, neurotically seeks the teacher's attention, or boasts of an "A" performance. Such incidents in driving or teaching are understood better after the second and third time they happen. Strategic knowledge-when to ignore or break rules and when to follow them is also developed in this stage as context begins to guide behavior. For example, in learning to teach, you team that praise doesn't always have the desired effect, as when a low ability child interprets it as communicating low expectations. You also learn that criticism after a bad performance by a usually good student can be quite motivating. Experience affects behavior but the advanced beginner may still have no sense of what is important. Benner (1986, pp. 23-24) makes this point in describing the difference between novices and advanced beginners on the one hand, and competent nurses on the other:

Stage 3: Competent Level

With further experience, and some motivation to succeed, most of the advanced beginners become competent performers of the skills needed in their domain of interest: nursing, piloting a plane, driving a car, or teaching. Not all advanced beginners, however, are likely to reach this level. Evidence shows that some teachers remain "fixed" at a less than competent level of performance (Borko, 1992; Eisenhart & Jones, 1992). This is consistent with other facets of life; we all have muttered at drivers who, though experienced, are not in our judgment competent. Nevertheless it is believed that many third- and fourth-year teachers, as well as more experienced teachers, reach a level of performance that we consider to be competent.

There are two distinguishing characteristics of the competent performer of a skill. First, they make conscious choices about what they are going to do. They set priorities and decide on plans. They have rational goals and choose sensible means for reaching the ends they have in mind. Secondly, while enacting their skill, they can determine what is and what is not important. From their experience they know what to attend to and what to ignore. This is the stage in which teachers learn not to make timing and targeting errors, because one has learned through experience what to attend to and what to ignore in the classroom. This is also when teachers learn to make curriculum and instruction decisions, such as when to stay with a topic and when to move on, based on a unique teaching context and a particular group of students.

Because they are more personally in control of the events around them, following their own plans and responding only to the information that they choose to, teachers at the competent stage tend to feel more responsibility for what happens. They are not detached. Thus they often feel emotional about success and failure in their area in a way that is different and more intense than it is for novices or advanced beginners, and they have more vivid memories of their successes and failures as well. Competent performers still are not very fast, fluid, or flexible in their behavior. These are characteristics of the last two stages in the development of expertise.

Stage 4: Proficient Level

Perhaps about the fifth year, a modest number of teachers may move into the proficient stage of development. This is the stage at which intuition or know-how becomes prominent. Nothing mysterious is meant by these terms. Consider the micro-adjustments made in riding a bicycle or learning a dance step. At some point in learning to ride a bicycle or performing the mambo, individuals no longer think about the kinds of adjustments needed. They no longer worry about balance and stop counting their steps to keep time to the music. In both cases they simply develop a more "intuitive" sense of the situation. Furthermore, out of the wealth of experience that the proficient individual has accumulated comes a holistic way of viewing the situations they encounter. They recognize similarities among events that the novice fails to see. That is the residue of experience. For example, the proficient teacher may notice, without conscious effort, that today's mathematics lesson is bogging down for the same reason that last week's spelling lesson failed. At some higher level of pattern categorization, the similarities between disparate events are understood. This holistic recognition of patterns as similar allows the proficient individual to predict events more precisely, because they see more things as alike and therefore as having been experienced before. Their rich case knowledge can be brought to bear on the problem. Chess masters, bridge masters, expert air traffic controllers, and expert radiologists rely heavily upon this ability. The proficient performer, however, while intuitive in pattern recognition and in ways of knowing, is still likely to be analytic and deliberative in deciding what to do. While the proficient stage is that of most tournament chess and bridge players, the grand masters are those few who move a stage higher, to the expert level.

Stage 5: Expert Level

If the novice is deliberate, the advanced beginner insightful, the competent performer rational, and the proficient performer intuitive, we might categorize the expert as often being arational. Experts have both an intuitive grasp of the situation and seem to sense in nonanalytic and nondeliberative ways the appropriate response to be made. They show fluid performance, as we all do when we no longer have to choose our words when speaking or think about where to place our feet when walking.

We simply talk and walk in an apparently effortless manner. The expert safety in football, the expert martial artist in combat, the expert chess master, and the expert teacher in classroom recitations all seem to know where to be or what to do at the right time. The great hockey player Wayne Gretsky, when asked by an interviewer for the secret of his remarkable success, answered modestly, "I don't know; I just go to where the puck is going to be." If the rest of us were that prescient, we would all be experts in our chosen fields.

As mentioned earlier, experts engage in performances in a qualitatively different way than do the novice or the competent performer. They are more like the race car driver or fighter pilot who talk of becoming one with their machine, or the science teacher who reports that her lesson just moved along so beautifully today that she never really had to teach. The experts are not consciously choosing what to attend to and what to do. They are acting effortlessly, fluidly, and in a sense, that is arational because it is not easily described as deductive or analytic behavior. Though beyond the usual meaning of rational, since neither calculation nor deliberative thought are involved, the behavior of the expert is certainly not irrational. Insight into the behavior of the expert can be obtained from the writings of Schon (1983), as he discusses knowledge in action, and from the work of Polya (1954), in his discussion of the role played by tacit knowledge in problem solving.

Experts do things that usually work, and thus, when things are proceeding without a hitch, experts are not solving problems or making decisions in the usual sense of those terms. They "go with the flow," as it is sometimes described. When anomalies occur, when things do not work out as planned or something atypical is noted, deliberate analytic processes are brought to bear upon the situation. Conversely, when things are going smoothly, experts rarely appear to be reflective about their performances.

This theory of development has heuristic value for thinking about the education and evaluation of teachers. It is also reasonably well supported by data my colleagues and I have collected over the last few years, although we did not know about the theory when we started our work. Bents and Bents (1990) also studied the transitions that occur between novice, advanced beginner, and expert teachers and verified the sensibility of these stages. They characterized the transition as moving from a teacher-centered, or self-centered, novice, to a more student-centered advanced beginner, to a stage in which the expert teacher was a much more integrated individual.

Our data and that of other researchers in this field, are more compatible than might be expected from such a small set of qualitative studies. Some of the findings in this area will be reported in the form of descriptive propositions.

DESCRIPTIVE PROPOSITIONS ABOUT PEDAGOGICAL EXPERTISE

Almost all studies of pedagogical expertise have been small-scale and nonexperimental. Remarkably, however, there is some consistency across these studies, and a number of propositions about expertise in pedagogy seem to be robust. Some of these overlap with the general characteristics of experts, described above. Some are specific to the domain of teaching.

PROPOSITION ONE: Experts excel mainly in their own domain and in particular contexts.

Chi, Glaser, and Farr (1988) and Glaser (1987) note that experts excel primarily in a single domain. The reason for this is because experts have a great deal more experience in some domains than in others. In studies by Lesgold and his colleagues (1988), expert radiologists were estimated to be looking at their 100,000th X-ray. Chess experts have spent 10,000 to 20,000 hours staring at chess positions (deGroot, 1965; Newell & Simon, 1972; Chase & Simon, 1973). Considering the lengthy time commitments that are necessary to become expert in complex areas of human functioning, it is no wonder that individuals generally excel in only a single domain.

Time and experience play a similar role in the development of pedagogical expertise. Similarly, anecdotal reports lead us to estimate that teachers will not hit their peak performance until at least five years of on-the-job experience have been obtained. The expert teacher, say with ten years' experience, will have spent a minimum of 10,000 hours in classrooms as a teacher, preceded by at least 15,000 hours in classrooms as a student, though it is unknown if the latter experience is of any value. Experience alone certainly will not make a teacher an expert, but it is likely that almost every expert pedagogue has had extensive classroom experience.

In further support of the single-domain quality of expertise, we have found that this domain-specific knowledge is acquired through lengthy experience that is quite contextualized. For example, in one of our research studies, experts, advanced beginners, and novices were asked to teach a 30-minute lesson on probability to a group of high-school students (Berliner, Stein, Sabers, Clarridge, Cushing, & Pinnegar, 1988). The teachers were given 30 minutes to plan the lesson. While they taught the lesson, they were videotaped. After the lesson, during stimulated recall, teachers were asked to tell us about and to justify their actions during teaching. Despite the fact that, as a group, the experts were judged to be better teachers on a number of dimensions (Clarridge, 1988), the task triggered a good deal of anger among this group of teachers. One of them quit the study, another broke down and cried in the middle of the study, and all were unhappy. They all stated their fears about performing well when we moved them from their own classroom contexts to the laboratory situation that we had created for them to teach in. We had allocated 30 minutes for planning, enough for the advanced beginners and the novices to feet comfortable. But the experts said they needed more time. One suggested three hours, another claimed to need three weeks to prepare that material. Our interviews revealed that experts rarely entered their classrooms without having taken all the time they needed to (a) thoroughly understand the content they would teach, and (b) plan one or more activities to teach that content. In addition, the experts noted that they did not know the students in this situation and that their expertise depends, in part, on knowing their students in three different ways:

1. Experts know the cognitive abilities of the students they teach regularly, giving them insight for determining the level at which to teach.
2. They know their regular students personally; in their classrooms they did not need to rely on bureaucratic and formal mechanisms of control while teaching, as they did in the experiment.
3. They did not have a history with the students in this study. In their own schools, students knew about them and had certain expectations about what their teaching would be like. These teachers always had students who expected to be well taught and to learn a great deal, even though they knew they would likely be pushed to their intellectual limits.

By taking these experts out of their classrooms, we had taken away the particular context in which these pedagogues had learned to excel. Thus, we conclude that we should regard expert knowledge as, for the most part, contextually bound. Their cognitions are usually situated; they are not usually adrift in the brain, unconnected to actions and situations (e.g., Brown, Collins, & Duguid, 1989). And this raises problems for transfer. Transfer across contexts and domains of knowledge appears to be very difficult and does not often appear spontaneously. It seems to require cognitive work, some form of mental effort (Perkins & Solomon, 1989). Thus we can anticipate that expert pedagogues, like experts in many other fields, will excel mainly in their own domain and in particular contexts within that domain. Their expertise will not automatically transfer across domains.

For teachers, such a conclusion has policy implications. One is that a K-12 certificate, a license to teach at any level of schooling, is probably inappropriate. Furthermore, exemplary performance by a teacher at the 10th grade does not automatically mean one will see exemplary performance at the 4th grade, if that teacher were to change grades. Pedagogical knowledge is contextualized, it is not easily generalized. These interpretations of the data suggest also that too many false negatives will occur in the simulations and assessment-center exercises that are being created for the certification and licensure of teachers by national and state boards. Such tests may identify too many teachers as less than exemplary because those teachers were assessed outside of the context in which they excel. It may only be possible to obtain valid judgments about the degree of expertise a teacher possesses from observing them in their own classrooms, an expensive form of assessment.

PROPOSITION TWO - Experts often develop automaticity for the repetitive operations that are needed to accomplish their goals.

Glaser (1987) notes the efficient decoding skill of the expert comprehender in reading as an example of the way automaticity frees working memory to allow other more complex characteristics of a situation to be realized. Examples of the automaticity or routinization of some teaching functions among expert teachers serve the same purpose. For example, Leinhardt and Greeno (1986), in studying elementary-school mathematics lessons, compared an expert's opening homework review with that of a novice. The expert teacher was found to be quite brief, taking about one-third less time than the novice. This expert was able to pick up information about attendance, about who did or did not do the homework, and about who was going to need help later in the lesson. She elicited mostly correct answers throughout the activity and managed also to get all the homework corrected. Moreover, she did so at a brisk pace and never lost control of the lesson. She also had routines for recording attendance, handling oral responding during the homework checks, and hand-raising to get attention. This expert also used clear signals to start and finish the lesson segments. In contrast, when the novice was enacting an opening homework review as part of a mathematics lesson, she was not able to get a fix on who did and did not do the homework, she had problems with taking attendance, and she asked ambiguous questions that led to her misunderstanding the difficulty of the homework. At one point the novice lost control of the pace. She never did learn which students were going to have more difficulty later in the lesson. Of importance is that the novice showed a lack of familiarity with well-practiced routines. She seemed not to have habitual ways to act. Students, therefore, were unsure of their roles in the class.

The routines of novice and well-regarded experienced teachers were studied in a study by Krabbe and Tullgren (1989). English and language arts lessons at the junior high school level were analyzed. The experts took an average of 14 minutes to introduce a literature lesson, while the novices took 2 minutes to do so. The experts needed that time to follow a routine for the set inducement or introductory phase of the daily lesson. First, they briefly stated the immediate objective of the activity (e.g., "We will discuss several ways that we can learn about the qualities of people"). Then, they gave clear and explicit directions about what they wanted students to be doing (e.g., "Put everything away. Here are three situations for you to think about answering"). Next, they created a positive environment for the phase of the lesson that would follow. The experts found ways to increase student involvement, often arousing curiosity through use of analogies that had something to do with the central concept and theme of the lesson. The goal of the lesson was apparent throughout its introduction. This three-step routine was also accompanied by a mood shift among the expert teachers, from humorous and playful at the beginning of the introduction to serious and businesslike as the presentation, discussion, or oral reading phases of the lesson drew closer (Krabbe, McAdams, & Tullgren, 1988). This regularly occurring pattern of teaching, this routine, was not evident when the videotapes of the novice teachers were analyzed. Krabbe and Tullgren (1989) also identified a routine in the way that the presentation phase of the lesson was conducted by the experts. The expert teachers introduced material gradually and in hierarchical order, illustrated their points by using student background and daily experiences, and provided practice opportunities as they went along. Novice teachers during this phase of the lesson tended to ask text-specific, factual questions until the lesson was over. No sense of a routine was noted in the way the novices taught either a literature or language arts lesson.

In one of our studies, we had experts, advanced beginners, and novices teach a short lesson on probability to about 15 high-school students. As previously noted, the experts were very unhappy about their participation in this task, in part, because the students they had to teach were not trained in routines to make the classroom run smoothly (Berliner, 1988). One expert, reflecting on what was wrong with the task, said:

In another of our studies (Carter, Sabers, Cushing, Pinnegar, & Berliner, 1987), experts, advanced beginners, and novices talked about what they would do first if they had to take over a class that had been running for five weeks. The experts all made mention of the need for routines. One put it this way:

The well-practiced routines of expert telegraphers, surgeons, ice-skaters, tennis players, and concert pianists (Bloom, 1986), no less than expert teachers, are what give the appearance of fluidity and effortlessness to the performance of experts. What looks to be so easy for the expert and seems so clumsy for the novice is the result of thousands of hours of reflective practice, experience from which learning derives.

Again, some policy issues are suggested by the findings that support Proposition Two. First, because some of the problems of the novice teacher occur due to their lack of automaticity and routinization, novice teachers might be better off if their training included practice in automating the opening homework review; taking attendance; assigning, collecting, and giving back homework; testing; introducing a new topic; closing a lesson; and so forth. Second, while a good deal of learning to teach is cognitive, some of it is skill-like in its nature. These skill-like parts of teaching should be mastered and routinized during preservice education. That might also increase both the confidence and the efficiency of the novice teacher.

PROPOSITION THREE. Experts are more sensitive to the task demands and social situation when solving problems.

Glaser (1987) noted that the mental models that experts develop to guide their behavior are constrained by the requirements of the situation in which they must work. Housner and Griffey (1985), in a study of experienced and novice physical education teachers, provide evidence of the sensitivity of experienced teachers to those issues. They found that the number of requests for information made by experienced and novice teachers during the time they were planning instruction was about the same. Each group made reasonable requests for information about the number of students they would be teaching, their gender, their age, and so forth. However, in two areas, the experienced teachers made many more requests than did the novice teachers. They needed to know about the ability, experience, and background of the students they were to teach as well as about the facility in which they would be teaching. In fact, five of the eight experienced teachers in this study of planning and instruction demanded to see the facility in which they would teach before they could develop their plan! Novices made no such requests. The experienced teachers were sensitive to the social and physical environment in which instruction was to take place. Furthermore, when actually performing in the teaching role, the experienced teachers implemented changes in their instruction more often than did novices, using social cues to guide their interactive instructional decision making. The experienced teachers used their judgment about student performance as a cue to change instruction 24% more frequently than did novices. Similarly, they used their judgments about student involvement as a cue 41% more often. They also used student enjoyment of the activities as a cue 79% more frequently and their interpretation of mood and student feelings 82% more often as a cue to change the way they were conducting instruction. On the other hand, the novices were using student verbal statements about the activity as their primary cue for instituting a change in their instructional activity. They responded to these cues 31 % more often than did the experienced teachers. Clearly, the novice teachers changed what they were doing, primarily, when asked to; they seemed unable to decode the social cues emitted by students about the ways in which instruction was proceeding. The experienced teachers, however, were far more sensitive to the social cues emitted in the situation, and they used these social cues for adjusting their instruction.

Other examples of this sensitivity to the task demands and social setting abound. In one study we conducted, we asked experts, advanced beginners, and novices to look at a series of slides depicting a class from start to finish and talk about any of the slides that caught their interest. The remarks of one of our expert science teachers about certain slides follows (Cushing, Sabers, & Berliner, 1989):

In another study of instruction by an expert and a novice physical education teacher (Hawkins & Sharpe, in press), the novice teacher seemed unable to maintain central proximity to the students in the classroom, while the expert was found to almost always move within the center of the gymnasium. This allowed the expert to talk more easily and frequently to the students at many different activity stations at which instruction was occurring. The centrality of the teacher's location allowed for more effective monitoring, resulting in less bureaucratic and more informal control of the class. The expert also showed a markedly higher rate of change in teaching behavior throughout the lesson, responding to perceived student needs as instruction occurred, needs he was in a position to interpret through his close monitoring of instructional progress.

The rich episodic and case knowledge of an expert and the lack of that same kind of knowledge by a novice is vividly contrasted in protocols from a study by Karen Nelson (1988). The expert's comments include sensitivity to the task demands and the social setting, which display her richer knowledge base. Both teachers are commenting on a slide of a physical education class in which a student on the slide is not in gym clothes.

PROPOSITION FOUR. Experts are more opportunistic and flexible in their teaching than are novices.

Glaser (1987) reports that experts are opportunistic in their planning and their actions. They take advantage of new information, quickly bringing new interpretations and representations of the problem to light. Novices are seen as less flexible. Borko and Livingston (1988) discuss these same behaviors among novice and expert teachers. The term these researchers use to characterize the opportunistic quality of the lessons of expert teachers is that of the "improvisational performer." They see expert teachers as having a well-thought-out general script to follow, but being very flexible in following it in order to be responsive to what students do. One expert, discussing his planning, made clear the improvisational aspect of teaching:

Professor Richard E. Snow of Stanford University once described teaching using the metaphor of "the teacher as a Baysian sheep dog." The image this brings to mind is that of the teacher or sheep dog letting the flock of students wander this way and that, but nipping at their heels if they get too far afield. The teacher can be flexible, opportunistic, catch the teachable moments, and ad lib chunks of the lesson, but based on remembrances of past experiences (the Baysian calculation of prior probabilities), and the teacher also keeps the students on a course that has them end up at an appropriate place. The novice has neither the confidence to let the flock wander to see where it goes, nor the experience to know how to pull it back on course. Thus, opportunistic teaching is possible for the expert, but much more difficult for the novice. For novices, the pedagogical schemata necessary for improvisation or opportunism seem to be less elaborate, less interconnected, and clearly less accessible than are those of the experts.

Concerns about some teacher education programs may be derived from these conclusions. The exhortations to beginning teachers to be creative and spontaneous, to abandon teacher manuals and create their own lessons, may be impossible to follow until later in their careers. Beginners may need more structured classrooms; they may need the directions of the manuals, even though many are notoriously bad guides for instruction. The manuals may provide the structure that enables a novice to start teaching, even if they are inappropriate to use later in their careers (see description of the novice, previously in this chapter, about the need for rules, even inadequate rules, from which to begin teaching). Similarly, while colleges of education may eschew frontal teaching methods, most novices may be unable to experiment with alternatives. Perhaps cooperative teaching techniques, peer tutoring systems, and project methods of instruction, all of which call for a different teaching role, are nearly impossible instructional techniques to ask a novice to use. We seem to forget the notion of developmental stages in teaching, though we all accept the homily that children must team to crawl before they can walk. Perhaps teachers must learn to be structured before they can be unstructured, perhaps they must control before they can improvise. Opportunistic teaching may be the mark of the experienced teacher and not possible for the novice.

PROPOSITION FIVE. Experts represent problems in qualitatively different ways than do novices.

Chi, Glaser, and Farr (1988) note that experts seem to understand problems at a deeper level than do novices. Experts apply concepts and principles that are more relevant to the problem to be solved. The understanding of novices seems to be at a more superficial level, with fewer instances shown of principled reasoning. We have found support for that general statement from studies of expertise in the pedagogical domain. In one small study of ours, Hanninen (1985) created realistic scenarios about educational problems associated with gifted children. One scenario, for example, described Mark, an 8-year-old Asian boy with severe hearing deficits who likes mathematics and science and who has a strong interest in computers. Scenarios describing educational problems of this type were administered to 15 subjects. Five of the subjects were experts, experienced teachers of the gifted; another five were equally experienced teachers but without any background in gifted education; and five more were novice teachers of the gifted, still working on their certification. Just the opening sentence of some of the protocols revealed much about the thinking of experts and novices. The opening sentence from one novice reads: "Mark seems like a very talented individual with many diverse interests." Another novice comments: "Mark should be encouraged by his teacher to continue his science experiments and work on the computer." An experienced teacher who was a novice in the area of gifted education says: "He should be able to pursue his interests in greater depth." In contrast to these banal, unsophisticated beginnings to essays that attempted to address Mark's needs, in which only superficial characteristics of the problem were noted, one expert began right off with: "Mark's needs can be broken into three broad areas: academic enrichment, emotional adjustment, and training to cope with his handicap." The essay followed this form and was a more organized and sophisticated representation of the problems than was obtained from the novices. The experts also concentrated more on the affective characteristics of Mark's life than did the other teachers. This is a common occurrence among experts and will be noted again, below.

If one views pedagogical knowledge as a complex multidimensional domain of knowledge requiring sophisticated thinking, then it could be argued that the classification of a problem as solvable by using Newton's second law, or by considering it a conservation of energy problem, is no different than classifying, Mark's educational needs as falling into three categories and describing action relevant to each of those categories. 

This was an appropriate representation of a problem in the pedagogical domain. Experts in every domain must form a cognitive representation of the problems they face in the world. Whether it is to troubleshoot a faulty generator (Johnson, 1988) or develop a program for a special education child, the task of the problem solver is the same, to construct a mental representation of the problem and determine the parameters of the problem space. Experts, compared to novices, usually have better, more functional problem representations. Without these rich models of the situations, they cannot fix generators, debug computer programs, or develop educational programs that are likely to be successful.

The scenario methodology employed previously to learn about teachers' thinking was used in a better designed study by Nelson (1988), in which expert and novice physical education teachers were the subjects. She concluded that experts "displayed a greater variety of application of sound principles of teaching" than did novices. The experts were also more creative and thorough in descriptions of ways to address teaching problems, and provided more solutions to each problem that they addressed. In a different study, Peterson and Cameaux (1987) used videotapes to elicit comments from experienced and novice teachers. They found that the comments of experienced teachers "reflected an underlying knowledge structure in which they relied heavily on procedural knowledge of classroom events as well as on higher-order principles of effective classroom teaching." Similarities in these two studies are quite apparent. An expert in Nelson's study said of a problem concerning an exercise program designed for an overweight child:

Another study (Stein, Clarridge, & Berliner, under review) examined the ways that expert teachers predict how students would respond to mathematics and science items used in the National Assessment of Educational Progress. From the protocols obtained while the subjects "thought aloud," we learned that experts named or labeled items in a much more detailed and specific way than did novices or advanced beginners. The experts also engaged in a task analysis of the problem in a way that was quite sophisticated. They analyzed the demands of the task represented in the items, apparently to look deeper into the nature of the problems that students might experience. Task analysis was coded in their think-alouds when the subjects verbalized something about the reasons for an item's difficulty or when they traced out the various steps or competencies that a student would need to answer an item correctly. Eighty percent of the experts analyzed the task demands of the items, and they did this for between one and four of the five items for which they had to think aloud. This may be compared to the novices and the advanced beginners, where only 50% of each group engaged in task analysis of the items, and when they did so, it was for only one of the five items they analyzed. The task analyses of the experts were also more embellished or more clearly formed than those of others.

The experts also differed in their inferences about the student cognitions used in answering an item. Experts seemed to have a fund of knowledge about the way students thought and how those thoughts interacted with the content of the specific mathematics or science items. In addition, the experts seemed able to think through the misalgorithms that students might apply. The experts had more experience dealing with student errors and therefore knew what types of errors students might make. Novices rarely discussed the issue of misalgorithms that students might apply to solve a problem. It was concluded from this study that experts in teaching mathematics and science, in comparison to novices and advanced beginners, were more likely to represent the test items that students would address in a more sophisticated way, through their better labeling of problem types. Furthermore, they gained insight into the nature of a particular problem type by more frequently doing a task analysis of it from the students' perspective. From the labeling and task analysis, the experts could more often predict the kinds of errors that students would make when attempting to answer those test items. Because their predictions were more accurate, they appeared to be better explainers than novices or advanced beginners and had a better understanding of their students' ways of thinking.

The implications that follow from acceptance of this proposition about differences between experts and novices are similar to ones already discussed. Novices will be unable to provide the principled kind of reasoning we might expect of experts. They will be unable to do a high quality cognitive task analysis of new curricula and will have trouble sequencing instruction and estimating what will be difficult and what will be easy for their students to learn. Experience will teach those who choose to learn from it. Those individuals will gain the "wisdom of practice" that is so characteristic of expert teachers; but there may be few short-cuts. Thus, our expectations about what a novice can do and think deeply about may need to be quite modest.

PROPOSITION SIX: Experts have fast and accurate pattern recognition capabilities. Novices cannot always make sense of what they experience.

The accurate interpretation of cues and the recognition of patterns reduce the cognitive processing load for a person. Sense is instantaneously made of a field, such as a chess board. Quick pattern recognition allows an expert chess player to spot areas of the board where difficulties might occur. Novices are not as good at recognizing such patterns, and when they do note them, they are less likely to make proper inferences about the situation.

One task in the study used slides as the stimulus materials and called for an updating of information by a subject, as a slide was viewed briefly three different times. One expert in science, after the second viewing of a slide, said:

We may regard the reading of a classroom, like the reading of a chessboard, to be, in part, a pattern recognition phenomena based on hundreds and thousands of hours of experience. The ability of novices or other relatively inexperienced teachers to interpret classroom information in some reliable way is limited precisely because of their lack of experience. The information related to pedagogical events may be so rich and complex that novices and advanced beginners simply cannot agree on what is seen. In the study where three television screens had to be monitored simultaneously, novices and advanced beginners seemed to experience difficulty in making sense of their classroom observations and in providing plausible explanations about what was occurring within the classroom. For example, these two comments were made by advanced beginners who were asked to describe the learning environment in the classroom they were observing. According to one:

Another study also showed this difference in interpretive competency between experts and novices (Carter, Cushing, Sabers, Stein, & Berliner, 1988). In this study subjects went through a series of slides depicting science or mathematics instruction over a class period in a high school. The subjects held a remote control and were told to move through the 50 or so slides at their own pace, stopping to comment on any slides that they found interesting. Novices and advanced beginners seemed to show no particular pattern in what they stopped and commented upon, and they showed the same kinds of contradictions in their interpretations that were found in the study using videotapes. That is, one novice might say "everything looks fine; they're all paying attention," and another novice might say "it looks like they're starting to go off task; they're starting to drift." A pattern was noted among the experts that was quite different. The experts, more often than the subjects in the other groups, found the same slides worth commenting on and had the same kinds of comments to make. For example, on slides of a science class, three different experts made these comments about one slide (Cushing, Sabers, & Berliner, 1989):

It is interesting to note that the hiring of new teachers, immediately out of school, is seen by some principals, parents, and personnel directors as a good thing to do. They are said to possess both energy and commitment. Our data, however, inform us that they also cannot make a lot of sense out of what they experience. They have not accumulated the experience to do that. It is part of the continuing devaluation of pedagogical knowledge that leads some individuals to think that new teachers are more likely to be better at their jobs than experienced teachers. Would those individuals choose a surgeon in that way? I have never heard of criteria such as "enthusiastic," "bouncy," and "fresh" used to choose a surgeon, but such terms are used frequently to describe new teachers, indicating a deep underestimation of the complexity of teaching and the role of experience in the development of expertise.

PROPOSITION SEVEN: Experts perceive meaningful patterns in the domain in which they are experienced.

The superior perceptual skills of experts is readily apparent, but there is no reason to believe it is due to any innate, superior perceptual abilities (Chi, Glaser, & Farr, 1988). Rather, their superiority is due to the way that experience affects perception. After 100,000 X-rays or the 10,000th hour of observing students, what is attended to and how that information is interpreted are likely to have changed for anyone motivated to learn from their experience. In one study (Carter, Cushing, Sabers, Stein, & Berliner, 1988), we asked expert, novice, and advanced beginners to view some slides of classrooms and tell us what they saw. The slides were flashed on the screen for only a very brief time. The responses of the novices and advanced beginners to the slides were clearly descriptive and usually quite accurate. A novice saw this:

In any field, the information that experts extract from the phenomenon with which they are confronted stems, in part, from the concepts and principles that they use to impose meaning on phenomena in their domain of expertise. That is, experts in all domains appear to be top-down processors. They impose meaning on the stimuli in their domain of expertise. In studying the interpretation of slides, a focus of the experts was on the notion of work: "students working at the blackboard," "students working independently," "teacher looking over a person working in lab," and so forth. This work orientation, of course, is part of what promotes high rates of achievement among the students of experts. But some other characteristics of the experts that are less clearly tied to effectiveness (though contributing to it) have also been found. In two studies Clarridge, 1988; Rottenberg, under review), it was found that expert teachers take student responsibility into account in a lesson, expecting students to somehow be involved in the creation of their own knowledge, perhaps through discussion, cooperative learning, questions, or projects. Somehow the experts seem to communicate this sense of responsibility and are sensitive to this when discussing their views of classes. In Clarridge's study (1988), videotapes of experts, novices, and advanced beginners were rated by a specialist in nonverbal communication. The specialist found the expert teachers to be high in incorporative behavior, behavior that invites the students to work jointly with them. The specialist also found that novice and advanced beginners set up barriers to keep authority in their own hands. The tapes, of course, were rated without knowledge of the experience level of the teachers that were viewed. Furthermore, the expert teachers seem to be unusually sensitive to the affective concerns of the students they teach and to individual differences among their students (Nelson, 1988; Bents & Bents, 1990; Rottenberg & Berliner, under review).

To place these data in context, a physicist may bring to bear Newtonian laws to make meaning of a problem in physics. A biologist may bring to bear concepts of homeostasis or ecological niche to make meaning of a problem in biology. A chemist, auto mechanic, or engineer will also bring to bear on the problems they face the most salient and useful concepts that they possess. Among the salient and useful pedagogical concepts with which teachers make meaning from phenomena that they encounter in their work are "attention," "work," "responsibility," "affect," and "individual differences." These concepts appear to have saliency among the experts and are not yet as important for novice teachers.

The implications of these ideas are relevant to the recent trend in many states to allow for alternative certification. Individuals who enter teaching in this way lack both pedagogical concepts and pedagogical experience. This leads us to predict that novice teachers from alternative certification programs will be ignorant about a good deal of what occurs in classrooms. They cannot help but to misinterpret events or interpret what goes on around them at a shallow level, and given the propositions stated above, we can expect such individuals to develop automaticity later, be slower to develop sensitivity to the task and social demands of the situation, be less opportunistic in their teaching, have more impoverished representations of pedagogical problems, and have slower pattern recognition. This does not bode well for the students of first-year teachers from alternative certification programs.

Our data do not suggest that the advanced beginner is markedly superior to the novice. On the contrary, we found it more difficult to separate these two groups than we had hypothesized. Although there were occasional hints in our studies that the advanced beginners were superior to the raw novices, both were usually clearly and markedly inferior in their performance to the experts. We do think, however, that there is a modicum of evidence to suggest that those who enter teaching from a teacher preparation program are likely to acquire the skills of a competent teacher sooner than those who do not.

PROPOSITION EIGHT. Experts may begin to solve problems slower, but they bring richer and more personal sources of information to bear on the problem that they are trying to solve.

In the study by Hanninen (1985) of expert and less expert teachers of the gifted, the mean time for a novice to read through a scenario and begin writing about ways to help solve the problems of a particular child was 2.6 minutes. The experienced teachers who had no background in gifted education took 3.0 minutes. The mean time for experienced/expert teachers of the gifted was 9.8 minutes. That is, from the start of reading their problem through the start of presenting their solutions, it took the experts three or four times as long as the other two groups. When scenarios were used in the same way with physical education teachers in Nelson's (1988) study, the experts also took more time to begin responding. Although the absolute differences between the groups were less dramatic than in our study, the experts still took 40% longer to begin responding to the scenarios. Thus, when one knows a good deal more about teaching, or some other domain, more time seems to be needed to represent the problem and to access the relevant knowledge needed to address the problem. If you do not have much information in storage, you need not look too long for information relevant to the task at hand.

Experts, however, unlike novices, can bring to bear rich personal stores of knowledge when they begin to address the problems they face. Peterson and Comeaux (1987) found that their experienced teachers "often gave more elaborated answers to the questions and gave more justifications for their decisions or comments." Nelson (1988) found that experts:

The implications of these data support previous comments made about new teachers and teachers that enter the profession through alternative certification. They are likely to be less adequate than are those who have some reflective experience under their belt.

Additional Propositions

As the data base grows, additional propositions about the differences between expert and less expert pedagogues are being revealed. For example, we are beginning to have evidence to suggest a ninth proposition, namely, that experts make substantially more inferences from and assumptions about the information presented to them than do novices. Their inferences, assumptions, and predictions allow them, like hockey pro Gretsky, to "go where the puck is going to be." Their experience lets them take a less literal and more inferential view of teaching. They can intuit what shall be coming and can therefore ignore or influence the flow of events. Novices cannot engage in such thought and action.

If evidence continues to mount, I will support a tenth proposition as well: Experts are more evaluative than are novices. When viewing the videos of other teachers or reading scenarios, the experts are likely to say things like "he shouldn't be wasting time taking attendance that way; there are better ways to do that," or "I think that's a poor way to find out if they know the concept because..." Novices have neither the confidence nor the successful past experience to be so evaluative.

For proposition eleven we might propose that experts, compared to novices, attend more to the atypical or unique events than to the typical or ordinary events in the domain in which they have expertise. Experts seem to ignore some things that go on. They seem to pick out anomalies to focus on, perhaps because they understand that not every bit of information is important to process.

Expertise, apparently, lets us process less, rather than more, of the information available from the environment, thus allowing more efficient use of the very limited working memory system that all of us possess. When Kounin (1970) noted that one of the characteristics of effective classroom managers was their ability to handle simultaneous events, he was probably studying experienced teachers. Novices cannot do that as well because their information processing system is apparently cluttered with information that is unimportant, though they do not yet know that.

A candidate for proposition twelve is that experts appear to be more confident about their abilities to succeed at instructional tasks than are novices. This is not surprising, because the experts have succeeded regularly over the years. However, this proposition suggests that only expert, proficient, or experienced competent teachers have the affective characteristics as well as the wisdom to take on the piloting of a new management structure. Trying to deal effectively with an ungraded class or a multidimensional teaching environment, both of which require the simultaneous occurrence of many different educational activities, would tax a novice enormously. New programs such as these should be tried first by those with the confidence and skill to implement them well. Such individuals are likely to be experienced teachers.

Additional Policy Considerations

Policy implications have been noted as each proposition was stated, but some additional considerations should be noted. For example, we noted how a K-1 2 certificate or other broad-span certification and licensure is challenged by the data suggesting that knowledge is contextualized, and that expert knowledge is probably highly contextualized. Thus, we should also note that grade levels and subject matters should not be seen as interchangeable for teachers. Principals and superintendents often move teachers around without thinking about the importance of domain-specific knowledge and the lack of immediate transfer to new contexts. They often act as if a good first-grade teacher can teach at middle school and a good earth science teacher in the suburbs can teach computers in the barrio. Our evidence questions their actions. Furthermore, because knowledge is so contextualized, tests and simulations to judge teachers will usually yield many false negatives-candidates that are expert in their classes, but not in the testing situation. Many of the experts in our studies were not very articulate. They could not describe what they did very well. Nevertheless, they did those things in their classrooms quite well. On the other hand, test situations often favor articulate individuals, so there may also be too many false positives, candidates who are quite articulate about teaching, but who are not exemplary in their classroom performance. It appears that all assessment of expert teachers outside of their classrooms is subject to validity problems that may not be possible to overcome.

Routinization of some procedures was also seen as a missing component of some teacher preparation programs. We need to note also that routinization is not just needed for management procedures, such as handing in homework, to make the class more efficient. Routinization can occur, as well for cognitive aspects of teaching,  the learning of the rule-example - rule principle of instruction, the use of explaining links in chains of complex assertions that are to be learned, or the procedures that tie material to be learned to some knowledge that students already know, and so forth (see Gage & Berliner, 1992, for additional cognitive instructional behaviors that can be routinized). In most teacher education programs, novices never are provided practice in such routines, and that is not sensible because routinization "is the hands and feet of genius."

We now recognize that many pedagogical skills are gained slowly, through experience only, and so we have come to understand the inability of teacher education programs to develop competent teachers. Such programs produce only beginners, albeit beginners who have been primed to learn from experience. For this reason it is appropriate to ask why a new teacher is given the same responsibility as that of an experienced teacher? For example, given what we now know about the limited abilities of novices, why do they get the same number of children to teach or the same number of courses to prepare for as someone with ten years of experience? The first year might result in more learning about teaching if the demands on the novice were not so great. They should have fewer children to teach and fewer classes for which to prepare. Of course, if the craft knowledge of teaching is devalued, then it doesn't matter. Novices can be expected to do as well as ten-year veterans. All that is needed to succeed as a teacher is enthusiasm and motivation!

That same misunderstanding of a novice's ability results in schools of education throughout the nation teaching things that are inappropriate for the level of development of the novice teacher. They often teach about techniques that require complex managerial structures, such as those required when running a whole-language reading program, a cooperative learning program, or a peer tutoring program. Teacher educators often do not provide any practice opportunities in these techniques. Rather, they tell novice teachers about them and then expect the novice teacher to immediately implement such activities. This is not sensible, particularly when we know that novices are terrified about losing control of their classes.

To compound the issue, many of the novices and advanced beginners are assigned to schools that have the most difficult children to teach, and within that setting, they are often assigned the most difficult classes with the most difficult students to teach. This policy occurs frequently throughout the nation, is patently unfair to children and novice teachers, and is one reason for the high attrition rate among new teachers during their first few years of teaching.

CONCLUSION

The research on expertise reveals the wonders of exemplary performance in many areas of human endeavor, including teaching. At least two-and-one-half million teachers toil in the public schools at jobs that are inherently difficult, emotionally draining, and require public performances. Those teachers must also accept lower social status and lower remuneration than others with the same abilities and investment in education. Society is indeed fortunate that many talented individuals stay in the profession long enough to acquire sufficient craft knowledge to be competent at their jobs, and the nation is blessed that some small number of those teachers also become expert at what they do. Their display of virtuosity in the classroom is no different than that shown by the concert violinist in a concerto or the chess master in a tournament. When compared to the performance of the novice or even the competent pedagogue, it is exemplary performance with which the expert provides us. Society just seems to value it so much less than exemplary performances in other areas.

Capturing the uniqueness of classroom performance in teaching is difficult to do with our research. All the studies to date have been small, generally qualitative, and highly interpretive. They have tended to be descriptive rather than experimental. In addition, unlike chess, we have no "winners" of tournaments by which to evaluate exemplary performance. The equivalent of a Pillsbury teach-off would be helpful, if that were possible in education; but it is not. So we have continuing problems in adequately identifying expert, proficient, and competent teachers, and this leads to problems with theory development in this area.

On the other hand, we do not have an insurmountable problem for those in search of a heuristic rather than a scientifically adequate theory about the development of expertise. Theory, it should be remembered, is not to be judged on its truthfulness, but on its usefulness. The five-stage theory presented earlier fits data well enough to allow us to think sensibly about the development of knowledge and skill in teaching.

Generalizing from studies in this domain should be difficult, but remarkably, that has not been the case. The propositions offered by Glaser from studies of expertise across disparate fields were provided at the start of this chapter. As has been made clear in this review, those propositions are highly compatible with the ones we derived from studies of teaching. We, therefore, have an interesting case in which many imperfect studies, across many different fields, have yielded a coherent body of knowledge and a heuristic theory with which to think about the acquisition of expertise in teaching and in other domains.

The study of expertise is inherently interesting. Individuals that excel at a task whether it be physics, medical diagnosis, chess, or teaching fascinate us and also provide psychology with subjects that have unique perceptual, conceptual, and psychomotor skills. As we learn from them about how information can be organized and used efficiently, they remind us, as well, that motivated learning over extensive periods of time can lead to extraordinary levels of performance. The lesson for us all is that when people have reasons to learn some domain and the opportunity to pursue that learning, they can reach levels of competency, proficiency, and expertise not ordinarily achieved by others. Perhaps there is a lesson here for those who teach poor children in America. Without a reason for learning the subjects we teach in schools, the level of competency in those subjects that the nation can expect of such children will remain low. The causes of their generally poor performances may be in the economic systems of the nation and not in the schools themselves. Motivation, therefore, would seem to be a factor that needs greater attention. The expenditure of time to learn something well depends on it.

REFERENCES

Benner, P. (1984). From novice to expert. Reading, MA: Addison-Wesley.

Bents, M., & Bents, R. (1990, April). Perceptions of good teaching among novice, advanced beginner and expert teachers. Paper presented at meetings of the American Educational Research Association, Boston, MA.

Berliner, D. C. (1987). In pursuit of the expert pedagogue. Educational Researcher, 15, 5-13.

Berliner, D. C. (1988, April). Memory for teaching as a function of expertise. Paper presented at meetings of the American Educational Research Association, New Orleans, LA.

Berliner, D. C., Stein, P., Sabers, D., Clarridge, P. B., Cushing, K., & Pinnegar, S. (1988). Implications of research on pedagogical expertise and experience for mathematics teaching. In D. A. Grouws, & T. J. Cooney (Eds.), Perspectives on research on effective mathematics teaching. Reston, VI; National Council of Teachers of Mathematics.

Bloom, B. S. (Ed.). (1985). Developing talent in young people. New York: Ballentine.

Bloom, B. S. (1986, February). Automaticity. Educational Leadership, 70-77.

Borko, H., & Livingston, C. (1988, April). Expert and novice teachers' mathematics instruction: Planning, teaching and post-lesson reflections. Paper presented at meetings of the American Educational Research Association, New Orleans, LA.

Borko, H. (1992, April). Patterns across the profiles: A critical look at theories of learning to teach. Paper presented at meetings of the American Educational Research Association, San Francisco, CA.

Brooks, D. M., & Hawke, G. (1985, April). Effective and ineffective session opening teacher activity and task structures. Paper presented at the meetings of the American Educational Research Association, Chicago, IL.

Brown, J. S., Collins, A., & Duguid, P. (1989). Situated cognition and the culture of learning. Educational Researcher, 18, 32-42.

Bryan, W. L., & Harter, N. (1899). Studies of the telegraphic language: The acquisition of a hierarchy of habits. Psychological Review, 6, 345-375.

Bullough, (1989). First year teacher: A case study. New York: Teachers College Press.

Carter, K., Cushing, K., Sabers, D., Stein, P., & Berliner, D. C. (1988). Expert-novice differences in perceiving and processing visual information. Journal of Teacher Education, 39(3), 25-31.

Carter, K., Sabers, D., Cushing, K., Pinnegar, P., & Berliner, D. C. (1987). Processing and using information about students: A study of expert, novice and postulant teachers. Teaching and Teacher Education, 3, 147-157.

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

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

Clarridge, P. B. (1988). Alternative perspectives for analyzing expert, novice, and postulant teaching. Unpublished dissertation, University of Arizona, Tucson, AZ.

Cushing, K. S., Sabers, D., & Berliner, D, C. (1989, January). Expert-novice research studies: Implications for teacher empowerment. In J.A. Kerrins (Ed.), Empowering the teaching learning process, proceedings of the sixth annual conference on issues and trends in educational leadership, Colorado Springs, CO.

deGroot, A. D. (1965). Thought and choice in chess. The Hague: Mouton.

Dreyfus, H. L., & Dreyfus, S. E. (1986). Mind over machine. New York: Free Press.

Duda, R. 0., & Shortiffe, E. H. (1983). Expert systems research. Science, 220, 261-268.

Eisenhart, M., & Jones, D. (1992, April). Developing teacher expertise: Two theories and a study. Paper presented at meetings of the American Educational Research Association, San Francisco, CA.

Gage, N. L., & Berliner, D. C. (1992). Educational psychology (5th ed.). Boston, MA: Houghton Mifflin.

Glaser, R. (1987). Thoughts on expertise. In C. Schooler & W. Schaie (Eds.), Cognitive functioning and social structure over the life course. Norwood, NJ: Ablex.

Glaser, R. (1990). Expertise. In M. W. Eysenk, A. N. Ellis, E. Hunt, & P. Johnson-Laird (Eds.), The Blackwell dictionary of cognitive psychology. Oxford, England: Blackwell Reference.

Hanninen, G. (1985). Do experts exist in gifted education? Unpublished manuscript, University of Arizona, College of Education, Tucson, AZ.

Hawkins, A., & Sharpe, T. (in press). Field system analysis: In search of the expert pedagogue. The Journal of Teaching in Physical Education.

Housner, L. D., & Griffey, D. C. (1985). Teacher cognition: Differences in planning and interactive decision making between experienced and inexperienced teachers. Research Quarterly for Exercise and Sport, 56, 44-53.

Johnson, S. D. (1988). Cognitive analysis of expert and novice troubleshooting performance. Performance Improvement Quarterly, 1(3), 38-54.

Krabbe, M. A., McAdams, A. G., & Tullgren, R. (1988, April). Comparisons of experienced and novice verbal and nonverbal expressions during preview and directing instructional activity segments. Paper presented at meetings of the American Educational Research Association, New Orleans, LA.

Krabbe, M. A., & Tu I Igren, R. (1989, March). A comparison of experienced and novice teachers' routines and procedures during set and discussion instructional activity segments. Paper presented at meetings of the American Educational Research Association, San Francisco, CA.

Kounin, J. S. (1970). Discipline and group management in classrooms. New York: Holt, Rinehart & Winston.

Leinhardt, G., & Greeno, J. (1986). The cognitive skill of teaching. Journal of Educational Psychology, 78, 75-95.

Lesgold, A., Rubinson, H., Feltovitch, P., Glaser, R., Klopfer, D., & Wang, Y. (1988). Expertise in a complex skill: Diagnosing x-ray pictures. In M. T. H. Chi, R. Glaser, & M. Farr (Eds.), The nature of expertise. Hillsdale, NJ: Erlbaum.

Nelson, K. R. (1988). Thinking processes, management routines and student perceptions of expert and novice physical education teachers. Unpublished dissertation, Louisiana State University, Baton Rouge, LA.

Newell, A., & Simon, H. A.-(1972). Human problem solving. Englewood Cliffs, NJ: Prentice-Hall.

Perkins, D. N., & Salomon, G. (1989). Are cognitive skills contextually-bound? Educational Researcher, 18, 16-25.

Peterson, P. L., & Comeaux, M. A. (1987). Teachers' schemata for classroom events: The mental scaffolding of teachers' thinking during classroom instruction. Teaching and Teacher Education, 3, 319-331.

Pinheiro, V. (1989). Motor skill diagnosis: Diagnostic processes of expert and novice coaches. Unpublished doctoral dissertation, University of Pittsburgh.

Polya, G. (1954). How to solve it. Princeton, NJ: Princeton University Press.

Posner, M. 1. (1 988). Introduction: What is it to be an expert? In M. T. H. Chi, R. Glaser, & M. Farr (Eds.), The nature of expertise. Hillsdale, NJ: Erlbaum.

Rottenberg, C. V., & Berliner, D. C. (under review). Expert and novice conceptions of everyday classroom activities, unpublished manuscript.

Sabers, D., Cushing, K., & Berliner, D.C. (1991). Differences among teachers in a task characterized by simultaneity, multidimensionality and immediacy. American Educational Research Journal, 28,63-88.

Schon, D. (1983). The reflective practitioner. New York: Basic Books.

Shuell, T. (1990). Phases of meaningful learning. Review of Research in Education, 60, 531-548.

Stein, P., Clarridge, P. B., & Berliner, D. C. (under review). Teacher estimation of student knowledge: Accuracy, content and process.

Westerman, D. (1990, April). A study of expert and novice teacher decision making: An integrated approach. Paper presented at meetings of the American Educational Research Association, Boston, MA.