Psy 416: Reasoning and Problem Solving
Expertise and its development
 Erwin Segal

What is expertise?

Currently the most prevalent is that an expert is a person who has some skill or knowledge in some domain that is matched by only a few other people. These people are thus extraordinary in some way. Anders Ericsson, probably the best known of the researchers on expertise defines expertise as Relatively stable outstanding performance.Experts are often labeled as such. People called exceptional, superior, gifted, talented, specialist, expert, etc. tend to belong to the set of experts.

There is no doubt that there are large differences in the quality of performance of different people on different tasks or in different domains. We can think of this difference as a scale of expertise. Novices are those who do not perform very well, and we can move through different levels of expertise until we find some individuals that we might say are skilled or knowledgeable beyond that of almost everyone else in the world, or world class.  What is the nature of this dimension? What are the categories within which this level of expertise motif applies?

Becoming an expert in any domain requires experience and effort. Don Norman introduced the notion that an someone requires 10,000 hours of experience and practice for reasonably complex domains to have the possibility of being an expert. Most people seem to agree with that assessment. In order for someone to become an expert in physics, music, chess, psychology, mathematics, baseball, etc. takes many hours, even years, of hard work and practice.

***Keith Ericsson in viewing the development of expertise argued that the most important factor, perhaps even necessary and sufficient for developing expertise is deliberative practice. Deliberative practice has four properties: (1) it is at an appropriate level of difficulty, (2) the participant receives informative feedback, (3) the participant has many opportunities for repetition, and (4) the participant has th opportunity to correct for errors (from Ericsson (1996; found in Sternberg & Ben-Zeev (2001).

If we focus on the process of becoming an expert rather than the claim that only a few become expert, we may come to a position I first heard from Micki Chi. Children are universal novices. They have not developed very many of the component skills needed for any domain. Decalage is the order of the day; many of the skills needed are relatively domain specific. The topics in the topics of cognitive development are a selection of the topics that could be studied under the rubric of expertise. Following Vygotsky, people learn to perceive and think in the domains that they come into direct interaction with. Some skills are fairly general in that they underlie many other skills. Those are the ones that children tend to learn first.

Domains that show movement from novice to expert include: Walking, talking, reading English text, writing, riding a bicycle, driving a car, getting around campus, talking to friends, studying particular courses, taking notes, seeing mathematical relationships, understanding formal arguments, taking multiple choice tests, taking essay tests. As we develop skills in these areas the structure of our  performance changes.

Hypothesis: The development of expertise parallels the development of cognition in children.

As people develop expertise their skills and performance becomes structurally different than it was prior to that development. The performance of experts is qualitatively different from those of novices. When one learns a new task she is automatically at an early stage of development. How early depends to a great extent on how many of the component skills have already been developed.

Some examples--(What is a domain?) playing chess, looking at blood slides, listening to music, looking at an x-ray, looking at 3-D pictures, riding a bicycle, skiing. dancing, gymnastics, solving physics problems, walking, reading, riding a bicycle, driving a car, dancing, typing, computer programming, cab driving, radiology, medical diagnosis, playing a piano, violin, basketball, baseball, football, gymnastics, golf, writing papers, taking tests.
 

 Propositional knowledge--Philosophically--knowing what. When novices learn in a new domain, the information tends to be represented as independent single propositions, with the development of expertise the separate pieces become integrated into larger units. Mayer's categories of propositional knowledge and comments Kinds of propositional knowledge--(a) factual e.g. equations, (b) syntactic--recognize appropriate forms, (c) semantic e.g. physical world, (d) schematic e.g. structural relations among parts and types, (e) strategic e.g. approach to solving problems

I am not real happy about the five categories of knowledge that Mayer uses. Its not a bad first step, but I have some problems with it. More now that I am more of an expert than when I first encountered it. Particularly, I have trouble understanding Mayer's difference between Factual and Semantic knowledge. Both of these refer to having some knowledge of the Domain under study.

The richness of understanding grows with the gaining of expertise. The schemata that one uses to account for the empirical observations get larger and more interconnected.
 

Syntactic knowledge could be the learning of equations, vocabulary, abbreviations, symbols, particular forms of acceptable sentences, etc.
Semantic knowledge in the linguistic sense is knowing how to map the words, sentences, and equations onto the domain. This is not easy. After identifying the basic laws of motion Newton spent years trying to apply them to different domains and other physicists continued that process for two hundred more years.
Schematic knowledge looks at the domain and identifies similarities and differences among parts. What is deemed similar often changes as one gains expertise. As stated in the text, novices are much more likely to be captivated by surface similarities and experts abstract or implicit ones.
Strategic knowledge is more similar to procedural knowledge--knowing how. How one is likely to attempt to solve a problem is likely to vary as a function of the experience one has with the domain. As one becomes more familiar she is much more likely for 'obvious' problems solve them by forward chaining or working forward. However, the data for real problems, i.e., problems of a type that the expert has not previously seen, are not likely to be solved by forward chaining, but by some variant of means-ends analysis. Procedural knowledge--production systems, performance, skills, philosophically knowing how; Proceduralization--learning how to do what needs to be done; and also with practice being able to integrate what was originally separate implementations of procedures into larger performance units. If a problem has intermediate states, or if there are more than one or two operations, selecting a correct procedure can be a daunting task. Intermediate feedback from the system or from an external source may be necessary for improved performance.
If the specific procedure is unavailable, the problem solver may find himself in a state of unknowing or frustration. Segal calls this an inchoate state. Experts know the vocabulary of a particular domain. Some of it is quite specialized. This "jargon" allows the specialist to combine several logically independent concepts into a single unified whole. The novice often can not learn the jargon even when she hears it, because the components do not make a meaningful unit to her. So even the vocabulary implies an understanding of the structure of the domain. 'Renege', in Bridge; 'Castle' in chess; 'decalage' in Piaget's theory; 'Power' in statistics; 'Church-Turing thesis' in computation to name a few. The greater the expertise, the larger the vocabulary, and the better the understanding of the underlying schema that the terms abstract from.
Syntactic knowledge is knowing the forms of the domain. This might be the vocabulary, the equations, the sentences.

Two kinds of expertise, routine and adaptive, have been identified by some Japanese scholars (Hatano and Inagaki (1986)). (Referenced in Holyoak's chapter, p 310, in Ericsson & Smith (1991)).
Routine experts solve familiar problems quickly and accurately. If the problems are unfamiliar they may have trouble with them.
Adaptive experts can invent new procedures which vary from their basic methods. They might be the 'creative' experts. How one learns what to do is at least one important variable affecting this difference. Trying to understand why the methods work help lead to adaptive expertise. This is reminiscent of Wertheimer and Gestalt Psychology.
Experts differ on how broadly they can transfer their expertise. This is partially based on how varied the learning situation is.

World class performance on recognizable skills with large individual differences require many hours of dedicated practice. Having behavior scaffolded by an expert often leads to much better performance: Karoli and gymnasts, Tennessee State track stars, Writers from U of Iowa workshops; Miss America candidates from Texas; Prodigies of all sorts.

Winton Marsalis view on becoming an expert: commitment, listening, training, practice, confidence, independence.

Component skills and knowledges. Must borrow many of them, learn to apply them in the right places, and integrate them to the new task. Some knowledge and skill must be learned from scratch. Many skills need to be developed more highly. It is possible that all of the component skills can be decomposed into simple enough parts that they are known a priori; however, expertise still requires integrating and restructuring them into usable schemata.

What is the state of novice performance? Inchoate states, random trial and error, frustration, backward chaining, small units, surface form, separate nonintegrated components, bottom-up

Expert performance--focused, much forward chaining, top-down, coherent and integrated, abstract organization, large units, proceduralization, integrated sequences, skillful, selective.

Ericsson, K. A. (1996). The road to excellence. Mahwah, NJ: Erlbaum
Sternberg, R. J. & Ben-Zeev, T. (2001) Complex Cognition: The psychology of human thought. New York: Oxford.
 to syllabus