About expert knowledge

Do experts simply know "more" than others, or is there something qualitatively different about an expert's knowledge compared to the knowledge of a non-expert?

While most of us are not aiming for an expert's knowledge in many of the subjects we study or learn about, it is worthwhile considering the ways in which expert knowledge is different, because it shows us how to learn, and teach, more effectively.

Experts are sensitive to patterns of meaningful information

A basic principle of perception is that it depends on the observer. What is green to you may be teal to me; a floppy disk to me may be a curious square of hard plastic to you. The observer always sees the world through her own existing knowledge.

An essential part of the difference between an expert and a novice can be seen in terms of this principle. A configuration of chess pieces on a board, seen briefly, will be bewildering and hard to remember for someone with no knowledge of chess, and even for someone with some experience of the game. But to a chess master, the configuration will be easily grasped, and easily remembered.

When chess pieces are placed randomly on a board, the chess master is no better than the novice at remembering briefly seen configurations. This is because the configuration is not meaningful. After tens of thousands of hours of playing chess, of studying the games of other masters, of memorizing patterns of moves, the master has hundreds of stored patterns in his memory. When he sees a configuration of pieces, he breaks it into meaningful elements that are related by an underlying strategy. Thus, while the novice would have to try and remember every single piece and its absolute or relative position on the board, the master only has to remember a few “chunks”.

The master can do this because he has a highly organized structure of knowledge relating to this domain. (It’s worth noting that expertise is highly specific to a domain of knowledge; a chess master will be no better than anyone at remembering, say, a shopping list.)

Expert knowledge is highly organized in deeply integrated schemas.

This sensitivity is thought to grow out of the deep conceptual schemas that experts develop in their area of expertise.

A schema is an organized body of knowledge that enables the user to understand a situation of set of facts. Schema theories include the idea of “scripts”, which help us deal with events. Thus, we are supposed to have a “restaurant script”, which we have developed from our various experiences with restaurants, and which tells us what to expect in a restaurant. Such a script would include the various activities that typically take place in a restaurant (being seated; ordering; eating; paying the bill, etc), and the various people we are likely to interact with (e.g., waiter/waitress; cashier).

Similarly, when we read or hear stories (and many aspects of our conversations with each other may be understood in terms of narratives, not simply those we read in books), we are assisted in our interpretation by “story schemas” or “story grammars”.

A number of studies have shown that memory is better for stories than other types of text; that we are inclined to remember events that didn’t happen if their happening is part of our mental script; that we find it hard to remember stories that we don’t understand, because they don’t fit into our scripts.

A schema then is a body of knowledge that provides a framework for understanding, for encoding new knowledge, for retrieving information. By having this framework, the expert can quickly understand and acquire new knowledge in her area of expertise, and can quickly find the relevant bits of knowledge when called on.

How is an expert schema different from a beginner’s one?

Building schemas is something we do naturally. How is an expert schema different from a beginner’s one?

An expert’s schema is based on deep principles; a beginner tends to organize her growing information around surface principles.

For example, in physics, when solving a problem, an expert usually looks first for the principle or law that is applicable to the problem (e.g., the first law of thermodynamics), then works out how one could apply this law to the problem. An experienced novice, on the other hand, tends to search for appropriate equations, then works out how to manipulate these equations (1). Similarly, when asked to sort problems according to the approach that could be used to solve them, experts group the problems in terms of the principles that can be used, while the novices sort them according to surface characteristics (such as “problems that contain inclined planes”) (2).

The different structure of expert knowledge is also revealed through the pattern of search times. Novices retrieve information at a rate that suggests a sequential search of information, as if they are methodically going down a list. Expert knowledge appears to be organized in a more conceptual manner, with information categorized in different chunks (mini-networks) which are organized around a central “deep” idea, and which have many connections to other chunks in the larger network.

These mini-networks, and the rich interconnections between them, help the expert look in the right place. One of the characteristics that differentiates experts from novices is the speed and ease with which experts retrieve the particular knowledge that is relevant to the problem in hand. Experts’ knowledge is said to be “conditionalized”, that is, knowledge about something includes knowledge as to the contexts in which that knowledge will be useful.

Expert knowledge contains information about when it will be useful.

Conditionalized knowledge is contrasted with “inert” knowledge. This concept is best illustrated by an example.

Gick and Holyoake (1980) presented college students with the following passage, which they were instructed to memorize:

After students had demonstrated their recall of this passage, they were asked to solve the following problem:

Although the students had recently memorized the military example, only 20% of them saw its relevance to the medical problem and successfully applied its lesson. Most of the students were unable to solve the problem until given the explicit hint that the passage they had learned contained information they could use. For them, the knowledge they had acquired was inert. However, when the analogy was pointed out to them, 90% of them were able to apply the principle successfully.

Much of the information “learned” in school is inert. A compelling demonstration of this comes from studies conducted by Perfetto, Bransford and Franks (1983), in which college students were given a number of “insight” problems, such as:

Some students were given clues to help them solve these problems:

These clues were given before the students were shown the problems. Some of the students given clues were also explicitly advised that the clues would help them solve the problems. They performed very well. Other students however, were not prompted to use the clues they had been given, and they performed as poorly as those students who weren’t given clues.

The poor performance of those students who were given clues but not prompted to use them surprised the authors of the study, because the clues were so obviously relevant to the problems, but it provides a compelling demonstration of inert knowledge.

The ability of students to apply relevant knowledge in new contexts tends to be grossly over-estimated by instructors. Most assume that it will happen “naturally”, but what this research tells us is that the conditionalization of knowledge is something that happens quite a long way down the track, and if students are to be able to use the information they have learned, they need help in understanding where, when and how to use new knowledge.