Seeing the Invisible: Perceptual-Cognitive Aspects of Expertise. Novices see only what is there; experts see what is not there. pdf 
Before going on, we need to be a bit clearer about what we mean by expertise. The achievement of expertise requires a large amount of experience, but simple accumulation of practice is not sufficient. If you endlessly repeat the same exercises, you will not develop very far. In research on firefighters (Klein, Calderwood, & Clinton-Cirocco, 1986), we observed that 10 years with a rural volunteer fire department were not as valuable for skill development as 1 year in a decaying inner city. Although some minimum amount of time is necessary it must be accompanied by a chance to accumulate a varied set of experiences.
A different approach is to model skill development on the cognitive maturation concepts of Piaget, because these are designed to accept variability and fluctuations in performance. The research of Campbell, Brown and DiBello (1991) on the development of expertise in computer programmers relied heavily on structured interviews with experts and trainees. The subjects were asked "meta-level" questions such as: How do you plan projects? How do you recognize problems? Can you compare how you do things now with how you did things when you were a beginner? Can you tell from a program how expert the programmer was? How did your knowledge of languages help when you learned a new one? The researchers also conducted a longitudinal study, using audiotaped diaries of programmers who were learning Smalltalk, an object-oriented language. The research showed the limitations of trying to distinguish experts from novices without considering intermediate stages of skill development.
Campbell et al. identified a number of developmental milestones in the learning of different program languages. With regard to Smalltalk, the researchers were able to specify seven distinct developmental levels. At each level, performance was distinguishable in operational terms (e.g., specific tasks the person could or could not perform well). This sets the developmental approach of Campbell et al. apart from others (e.g. Dreyfus & Dreyfus, 1986), which distinguish stages solely at a conceptual level, unanchored to empirical markers. The Campbell et al. approach may be a useful strategy for identifying milestones in skill development, and for allowing more useful assessment procedures.
# Re: Expertise as the development of higher order strategies Experts do not necessarily use different strategies than novices. We cannot envision training people to become experts by showing them the importance of top-down processing and analysis of deep structure... both experts and novices rely to some extent on top-down and bottom-up reasoning. Both utilize a divide and conquer strategy, and a cycle of forming and testing mental models. Both experts and novices rely on analogies and metaphors. All the various general strategies appear to some extent in almost all forms of reasoning.
# Re: Expertise as a function of knowledge base As people develop richer knowledge bases they are able to represent problems in more powerful ways, and to take more advantage of stronger reasoning strategies... The experts' representations are conceptually richer and more organized than those of the novices.... experts encode problems using deeper structure, whereas novices use surface features.
We suggest that the knowledge base and accumulated experiences may change the way people view their worlds. # Expertise as perceptual-cognitive differences
There is no way for a novice to judge what is normal and what is an exception. Consider a study by Chi, Hutchinson, and Robin (1988), in which descriptions of novel dinosaurs were presented to children who were "dinosaur expert" or novices. These novel dinosaurs were designed to be either typical or atypical. The experts, of course, realized immediately that a dinosaur was typical of a call of dinosaurs and were then able to attribute all the relevant features from the family to which the novel "typical" dinosaur belonged. The experts were equally proficient at determining that a novel dinosaur that was not typical did not belong to any of the familiar families. Novices lacked this ability to judge typicality and to use it to infer other characteristics.
The use of typicality judgements is important for problem solving as well as decision making. Elstein et al. (1978) studied the way physicians made diagnoses and found that the physicians rarely used a purely inductive method of letting the data drive the inferences. Even though the physicians were trained to reserve judgement rather than contaminate the process, they still could not resist forming early impressions. Elstein et al. referred to this strategy as hypothetico-deductive method, because the early judgements were hypotheses that helped to direct the subsequent information gathering. Without early hypothesis, the information gathering would have been inefficient and interminable. Weitzenfeld, Klein, Riedl, Freeman, and Musa (1991) observed the same phenomenon in a study of expertise in software troubleshooting. Experts were able to formulate initial hypotheses, or stories of how the problem might have arisen, and could use these hypotheses to direct the search for more evidence. Where do hypotheses come from? Presumably, they involve the same mechanism of situation assessment as was discussed earlier – using experience to judge typicality.
Experts are particularly better than novices and journeymen in making fine perceptual discriminations. Consider televised broadcasts of Olympics events such as gymnastics and diving, where expert analysts notice aspects of performance that we novices can detect only when shown the slow-motion replay. The reader who is interested in additional discussions of perceptual learning and expertise should consult Chi, Glaser, and Farr (1988) and Hoffman, Burton, Shanteau, and Shadbolt (1991).
# Implications and Applications of a Perceptual-Cognitive View of Expertise
Theory and application both benefit when they are linked. Without an applied focus, it is easy for theory to lose focus, and to address easy problems rather than hard problems that demand greater complexity. Application without theory can also lose shape and become a disorganized collection of practices. If our perspective on expertise is to develop, it must have value in shaping practice.
If expertise is so dependent on learning to perceive the world differently, then we should look at ways to sharpen perceptual skills, rather than ways to simply add to the experience base. This would enable people to make more rapid and accurate judgments about the nature of the situations they are in.
# Conclusion
We feel it may be useful to examine a somewhat different perspective, one that centers around the way the experts perceive tasks and situations: 1. There are several ways that experts can see things that others cannot: Experts can use their knowledge base to recognize typicality, experts can use perceptual learning to make fine discriminations, and experts can use mental simulation to represent antecedents and consequents. 2. Experts can also use their knowledge base to apply higher level rules, such as top-down processing. Such rules are also in the repertoire of novices and journeymen. People at lower skill levels infrequently use higher level strategies because they lack the experience base to make such strategies work. 3. Therefore, trying to train general strategies for thinking like experts may not be worthwhile. Attempts to teach such strategies are not useful in developing expertise, and research does not demonstrate the effectiveness of such attempts. 4. Training can address methods for sharpening perceptual skills and changing the way situations are experienced. There are ways of providing personal experiences, manufactured experiences, and vicarious experiences to accelerate the growth of expertise. We have shown how models of expertise can form the basis of knowledge engineering programs for training specialists in areas as diverse as nursing, computer programming, and research management. 5. It may be useful to develop a discipline of knowledge engineering, to direct us in eliciting and applying expert knowledge.
Campbell, R. L., Brown, N. R., & DiBello, L. A., (1991). The programmer's burden: Developing expertise in programming. In R. R. Hoffman (Ed.), _The cognition of experts: Psychological research and emperical AI_ (pp. 269-294). New Your: Springer-Verlag.
Chi, M. T. H., Glaser, R., & Farr, M. J. (1988). _The nature of expertise._ Hillsdale, NJ: Lawrence Erlbaum Associates.
Hoffman, R. R., Burton, A. M., Shanteau, J., & Shadbold, N. R. (1991). _Eliciting knowledge from experts: A metholological analysis._ Unpublished manuscript. Garden City, NY: Department of Psychology, Adelphi University.
.
John Allspaw pointed me at this PDF in response to some discussion at SNAFUcatchers 2018 workshops. Here, I'm doing my homework to collect the passages that resonate for me.
Ward offers a method for interviewing experts: lineup Reflective Interviews
I have learned my own deep appreciation for perception and practice of perception via the Feldenkrais practice of Awareness Through Movement which also changed my experience of aikido. Feldenkrais was influenced by Fechner. See Risk of Serious Inquiry