The Road To Excellence

the Acquisition of Expert Performance in the Arts and Sciences, Sports, and Games

K. Anders Ericsson

• Publisher: Psychology Press
• Year: 2014
• Print ISBN: 9780805822311
• eBook ISBN: 9781317779056
• Topic: Psychology
• Subtopic: Cognitive Psychology & Cognition
• Index: Psychology

Excellence and the highest levels of performance in the arts and sciences, sports, and games have always been an object of fascination to both scientists and lay people. Only during the last 20 years have scientists studied these levels of performance in the laboratory in order to identify their mediating mechanisms. Contrary to the common belief that innate talents are the critical factors for exceptional performance, investigators have found that acquired skills, knowledge, and physiological adaptations in response to intense practice are the primary mechanisms, mediating the highest levels of performance.

This is the first and only book to examine how elite performers effect their exceptional accomplishments. The world's leading researchers on expert performance and creative achievement review theories and recent findings from many different domains of expertise on how experts optimize improvement in their performance and eventually attain excellence. Elite performers are shown to have engaged in deliberate-practice activities specifically designed to improve their performance from an early age. By age 20 they have often accumulated over 10,000 hours of practice! The essential elements of deliberate practice, such as specific goals to improve performance, successive refinement through repetition, feedback and instruction, are explicated for different domains. Although the content of practice tasks will necessarily differ from domain to domain, investigators have found invariant characteristics for the optimal duration of practice sessions, maximal amounts of daily practice, the length of intense preparation (around 10 years), and ages of peak performance. Some of the book's chapters extend the review to the acquisition of everyday-life skills such as reading, to the performance of teams of experts, and to the development of creative achievement, geniuses, and artistic child prodigies. The book concludes with commentaries by several outstanding scientists in psychology, education, and history of science who discuss the generalizability of presented ideas and raise issues for future issues.

EXTRA COPY...It could be said that striving for excellence is what characterizes humanity, or perhaps what characterizes humanity at its best. Why do so few individuals ever reach the highest levels when so many start out on the Road to Excellence? In this book, the world's foremost researchers of expert performance in domains as diverse as sports, medicine, chess, and the arts explore the similarities and differences in the extended and strenuous Road to Excellence taken by the successful individuals in each domain. Their findings will intrigue and inspire readers who are themselves driven to achieve or who simply want to better understand the processes involved.

1

---

The Acquisition of Expert Performance: An Introduction to Some of the Issues

---

K. Anders Ericsson

The Florida State University

In every domain of expertise, many start on the “road to excellence” but few reach the highest levels of achievement and performance. Through the centuries, a large body of historical facts, interviews, and observations have been collected about these exceptional individuals in popular and academic biographies. In many cases, the most detailed information is given retrospectively by the exceptional individuals themselves years later when their outstanding achievements have been recognized by their peers and society. For example, the famous accounts of creative experience by Kekule, Poincare, and others were first written down years or even decades after the events they describe, and consequently suffer from many of the problematic biases of recollection (Gruber, 1981b). The scarcity of relevant evidence often forces biographers to lower their standards of evidence to the point where it is difficult to separate hard facts from biased recollections and myths. Much of the popular evidence for talent and inexplicable creativity is based on accounts that cannot be subjected to scientific analysis (Ericsson & Charness, 1994). For example, many of the famous descriptions of the young genius Gauss are based on unsupported anecdotes told by Gauss himself as an old man (Bühler, 1981). The first step toward a scientific account of exceptional achievements requires rejection of unverifiable evidence and the identification and systematic accumulation of all empirical evidence that meets standard scientific criteria.

Outstanding human achievements usually consist of actual products, such as works of art or scientific publications. In these cases, the date of completion and the individual responsible for the achievement are known. However, the psychological processes that led to the achievement are far more difficult to investigate and explain scientifically. Part of the problem concerns their original and creative nature. By definition, an innovative achievement goes beyond the application of available knowledge in the domain at the time of its completion. Hence, major innovations and discoveries emerge unpredictably and often considerable time passes before they are widely accepted as valid great accomplishments. We are thus primarily limited to the evaluation of post-hoc explanations of major specific discoveries based on careful analysis of the historical records such as notebooks and earlier drafts (Gruber, 1981a; Holmes, 1989). A complementary experimental approach is to recreate the historical conditions in which a discovery of a mathematical law was made and study how individuals who are unaware of that discovery but have all the necessary knowledge are able to rediscover the original law (Qin & Simon, 1990).

An alternative approach to creative achievements involves a search for general patterns of development across exceptional individuals. This approach is well represented in this book. Simonton (chap. 8, this volume) has focused on time course of creative productivity in different domains during the life span and Howe (chap. 9, this volume) describes general characteristics of the childhoods and early careers of geniuses. Furthermore, the chapters by Richman, Gobet, Staszewski, and Simon (chap. 6, this volume) and Simonton (chap. 8, this volume) discuss related mechanisms to account for the generation of innovative ideas and creative products.

The major problem confronting the scientific investigation of extraordinary achievements and their creative nature is their uniqueness. By focusing instead on the highly replicable skills of exceptional performers (e.g., professional musicians) one can identify high (expert) levels of performance (Ericsson, Krampe, & Tesch-Römer, 1993) that correspond to phenomena that are more tractable to analysis with scientific methods.

EXPERT PERFORMANCE AS AN EMPIRICAL PHENOMENON

Empirical phenomena can be studied with scientific methods when they meet several criteria. The first and most important criterion is that the phenomenon occurs reliably in clearly specified situations with distinctive observable characteristics. Current findings on the superior performance of experts shows that it occurs reliably in many domains of expertise. Second, the phenomena should be reproducible under controlled conditions in the laboratory to allow for experimental variation and systematic observation of the mediating processes. Following Ericsson and Smith (1991), I show that many forms of expert performance can be reproduced and studied under laboratory conditions. Finally, the observed phenomena for a specific situation should be predictable and describable by objective absolute measurements. In some domains of expertise such as individual sport events (e.g., running the 100-meter dash), the performance is measured by absolute units of time. In other domains, performance is evaluated in relative terms through comparison with other contemporary performers (e.g., gymnastics). I propose methods for measuring and describing even these types of expert performance by absolute standards that are independent of the social and historical context of the studied expert performance. In the next three sections I discuss the following characteristics of expert performance: (a) its reliability, (b) its reproducibility in the laboratory, and (c) its measurement in absolute terms.

Reliability of Superior Expert Performance

There are many domains of expertise in which individuals consistently exhibit outstanding and superior performance under standardized testing conditions. In perceptual motor activities, such as individualized sports and typing, the outcome of performance can be directly measured. For competitive domains, such as tennis and chess, results from tournaments can be analyzed to rank individuals, often on an interval scale (Elo, 1978). The individual differences in recorded performance between the best and the least accomplished performers in these domains are among the largest reproducible differences in performance observed for normal adults.

Similar large differences in performance would be expected in many professional domains of expertise, such as medicine, auditing, physics, and business, in which a long period of preparatory education followed by an apprenticeship is required. Only after many years of further experience are some individuals recognized as experts in the domain. Surprisingly, the prediction that expert professionals exhibit clearly superior performance on relevant activities in the domain has not been well supported when tests are performed under standardized conditions. For example, experts in decision making and judgment have often failed to display superior accuracy of performance, especially for tasks involving prediction of future outcomes (Camerer & Johnson, 1991). Not all experts in computer programming (Doane, Pellegrino, & Klatzky, 1990) and physics (Reif & Allen, 1992) perform consistently at a superior level on representative tasks. Professional mathematicians’ performance on algebra problems was not reliably superior to that of the best third of a sample of college students (Lewis, 1981). A particularly striking finding is reported by Wagner and Stanovich (chap. 7, this volume), who show that experts at speed reading can scan text rapidly but without any understanding of the content. The only validated aspect of the speed readers’ skill is the high speed of turning pages.

The lack of general superiority of experts’ performance has forced investigators to identify those activities that are central to experts’ responsibilities. It is precisely at those activities that experts usually excel. However, expertise may never develop for some tasks, such as forecasting events. For example, it has been impossible to identify expert investors on the stock market that consistently outperform the average market indices (McClosky, 1990). There is reason to believe that the stock market is continuously adapting to publicly available information, thus making superior prediction virtually impossible in the absence of illegal information obtained from insiders. Unless one can identify at least one expert who exhibits reliably superior performance we do not know that such performance is possible.

In many types of professional expertise, most of the time is spent on relatively routine cases and complex difficult cases are encountered more rarely. In medicine, the superior diagnostic performance of experts has been readily demonstrated with difficult cases (Lesgold et al., 1988; Norman, Trott, Brooks, & Smith, 1994; Patel & Groen, 1991) even compared to medical residents with completed training and some work experience. On the other hand, evidence for performance differences with more routine medical cases has been more difficult to demonstrate (Norman, Coblentz, Brooks, & Babcook, 1992; Schmidt, Norman, & Boshuizen, 1990).

An important general finding has been that the number of years of experience in the domain is only weakly related to the level of attained performance (Ericsson, Krampe, & Tesch-Römer, 1993). Even more refined measures, such as the number of chess competitions attended (Charness, Krampe, & Mayr, chap. 2, this volume) and the number of baseball games played in the major leagues (Schulz, Musa, Staszewski, & Siegler, 1994), do not accurately predict performance in samples of skilled performers. However, the amount of time that children spend reading is related to various aspects of reading performance (Wagner & Stanovich, chap. 7, this volume).

In summary, reliably superior performance by experts is exhibited under conditions that capture the essence of expert performance of the domain, such as the conditions of competition for athletes or difficult cases in medical diagnosis for medical experts. In a later section, I explore whether insights into the necessary conditions for attaining superior performance might also explain those striking instances when experts do not outperform novices.

Reproducibility of Superior Expert Performance

In order to predict human behavior, one must take into account motivational factors. Fortunately, most expert performers face the problem of controlling motivation frequently when they have to perform at specific times during competitions or when medical doctors and other professionals have to respond to emergencies. Hence, expert performers learn to control all relevant (including motivational) factors necessary for attaining their superior performance at a time determined by external factors. This control over the elicitation of one’s performance suggests that it would be possible to reproduce expert performance outside of its everyday-life context under controlled conditions in the laboratory.

In many domains of expertise, such as individual sports, the conditions for competition are already standardized for all participants and expert performance can be reproduced under similar test conditions in the laboratory. Studying expert performance in such domains as medical diagnosis and computer programming is more difficult because individual experts hardly ever encounter the same problem or cases in their normal professional practice. The standard procedure to measure expert performance in these domains has involved the collection of a set of standardized tasks. These tasks are then administered to experts and other professionals in the domain under controlled conditions that allow the researchers to study the processes mediating superior performance.

Many representative situations encountered by experts cannot be captured by fixed descriptions because they involve ongoing interaction with other experts or active participants. Patel, Kaufman, and Magder (chap. 5, this volume) review the large body of research on performance of teams of experts in complex contexts. They also propose a methodology for studying and analyzing interactions between experts in a team. A different approach to deal with the complexity of human interaction is to identify important goal-directed activities of a particular participant in the interaction (Ericsson & Smith, 1991). The relevant context of this activity and its goal can then be converted to a standardized task that can be presented to any expert responsible for that type of activity. This approach for studying expert performance has been successfully applied to chess, in which two opponents face each other during a match. Although the starting position in a chess match is always the same, the sequence of chess moves for two specific opponents is virtually never reproduced exactly. Because the index of skill in interactive domains is derived from competitions with many different opponents, it is reasonable to assume that the superior performance should be reflected in how well the performer responded to any of the individual situations within the interchange. For example, an expert chess player would be expected to select as good or better moves for each of the chess positions of a chess game (de Groot, 1946/1978). Hence, the standard method for studying expert chess players is to present them with an unfamiliar chess position from an actual chess game and ask them to select the best move. The quality of selected moves under such conditions is highly correlated with chess skill determined from tournaments (Charness, 1991).

By breaking up the interactive interchange into a sequence of its component situations, well-defined tasks can be identified that can be presented to expert performers in the corresponding domain. This methodology has been extended to analyses of expert performance within a team context. Helsen and Pauwels (1993) analyzed soccer matches in the World Cup to identify situations when a world-class soccer player got the ball and had to rapidly pass the ball to another player. Helson then carefully reconstructed the context for the situations using other soccer players so a video recording could be made from the exact view of the critical player. They then recreated the perceptual situation by projecting the video on a large screen in a gymnasium, and at the appropriate time the subject was passed a real soccer ball and the speed and accuracy of the subject’s actions were recorded. Under those conditions Helsen and Pauwels (1993) were able to establish superior performance of soccer experts compared to less accomplished soccer players. How much context with prior interchanges between the players needs to be presented before the critical test situation and the degree of perceptual fidelity required raise empirical questions concerning the necessary conditions for capturing a particular type of expert performance?

It is important that each selected situation or task is associated with a correct or consensually established best action that can be objectively specified. This is not typically a problem because the investigators have access to more information and more time to analyze the situation than the participating experts. However, in some domains, such as auditing (Bedard & Chi, 1993), “experts” do not agree about the correct judgment of representative cases, which virtually precludes efforts to identify tasks measuring any associated performance.

In summary, expert performance can be reproduced in many domains for a collection of standardized tasks with consensually established best responses. Once the superior performance of experts can be reliably reproduced in the laboratory, its mediating mechanisms can be des...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Table of Contents
• Preface
• 1 The Acquisition of Expert Performance: An Introduction to Some of the Issues
• 2 The Role of Practice and Coaching in Entrepreneurial Skill Domains: An International Comparison of Life-Span Chess Skill Acquisition
• 3 Deliberate Practice in Sports: What Is It Anyway?
• 4 The Acquisition of Musical Performance Expertise: Deconstructing the “Talent” Account of Individual Differences in Musical Expressivity
• 5 The Acquisition of Medical Expertise in Complex Dynamic Environments
• 6 Perceptual and Memory Processes in the Acquisition of Expert Performance: The EPAM Model
• 7 Expertise in Reading
• 8 Creative Expertise: A Life-Span Developmental Perspective
• 9 The Childhoods and Early Lives of Geniuses: Combining Psychological and Biographical Evidence
• 10 The Rage to Master: The Decisive Role of Talent in the Visual Arts
• 11 Changing the Agency for Learning: Acquiring Expert Performance
• 12 Expert Performance and the History of Science
• 13 Capturing Expertise in Sports
• 14 Laboratory Experimentation on the Genesis of Expertise
• 15 Costs of Expertise
• Author Index
• Subject Index