[ CogSci Summaries home | UP | email ]

Ericsson, K. A., & Staszewski, J. J. (1989). Skilled memory and expertise: Mechanisms of exceptional performance. In D. Klahr & K. Kotovsky (Eds.), Complex information processing: The impact of Herbert A. Simon (pp. 235-267). Hillsdale, NJ: Lawrence Erlbaum.

Author of the summary: David Zach Hambrick, 1998, gt8781a@prism.gatech.edu

The early work on expertise showed that novice-expert differences in domain-specific performance can be accounted for by differences in amount of knowledge: experts are more knowledgeable than novices. For example, Chase and Simon estimated that expert chess players have a vocabulary of up to 50,000 patterns representing familiar configurations of chess pieces. While appealing to differences in amount of knowledge to explain expert-novice differences makes good sense, it presents what Ericsson and Staszewski call a "thorny" problem: How do experts process an enormous amount of information given that they are subject to the same basic information processing demands as novices? More specifically, "How is it that experts bring more knowledge to bear on problem solving and skilled performance than novices and at the same time perform more quickly and accurately" (p. 237)?


The Chase-Simon Account


According to the memory skill perspective, through experience, experts not only acquire content knowledge, but also they acquire memory skills that allow for the efficient application of this knowledge. Chase and Simon proposed that domain-specific knowledge consists of organized patterns of information (or "chunks) stored in LTM, and claimed that experts used this knowledge to encode and retrieve newly presented domain-specific information. Labels corresponding to chunks were held in STM until recall. Experts presumably had larger, not more, chunks. Nevertheless, later research showed that the number of chunks for experts exceeded STM limitations and that STM interference had minimal effect on memory performance.


Skilled Memory Theory


Chase and Ericsson argued that experts acquire LTM encoding and retrieval skills, and that these skills expand the functional capacity of working memory. Organization of knowledge allows for encoding and retrieval rates typically associated with short-term memory. The result is computational efficiency, and as Chase and Ericsson explain, "From this theoretical perspective, the skilled memory effect, that is, expertsí superior recall for familiar materials, is viewed as a natural by-product of expertsí use of LTM to maintain information in an easily accessible state for processing in complex tasks" (p. 238). Encoding and retrieval skills result from organization of knowledge. Information is easier to encode and retrieve when it is well-organized, and limitations of STM are thereby circumvented. By contrast, novicesí knowledge is poorly organized; LTM encoding and retrieval processes are thus slow and require a great deal of effort. They are thus constrained by the limitations of STM.


Principles of Skilled Memory Theory


Chase and Ericssonís theory consists of three principles. The meaningful encoding principle states that experts use prior knowledge in the encoding of new domain-specific information. For example, SF encoded sequentially presented digits as running times (e.g., 3492 as 3 minutes, 49.2 seconds). Meaningful encoding creates "multiple potential cues and avenues for retrieval" (p. 239). The retrieval structure principle states that through practice individuals acquire memory mechanisms that facilitate the retrieval of information. Ericsson and Staszewski define retrieval structures as "organized systems for using retrieval cues" (p. 239). Retrieval structures are used during encoding as well. Put another way, retrieval structures are hierarchical organizational schemes used during encoding to assign retrieval cues. The Method of Loci is an example. Retrieval cues are assigned to locations within a familiar physical environment. Retrieval occurs through mentally "walking through" the environment. Finally, the speed up principle is an observation that LTM encoding and retrieval processes become faster with practice, equaling rates of short-term memory processes.


The three processes described above serve to increase the functional capacity of STM. That is, "By learning to quickly store and retrieve information in LTM, experts increase the amount of information they can easily access for processing beyond the limited amount that can be maintained in STM" (p. 240). Two predictions that are implied by this perspective are that 1) skilled memory is domain specific and involves LTM and 2) skilled memory operates within the constraints of STM. An additional point is that memory skills, which allow for the rapid encoding and retrieval of information, are used in many domains, and not just by mnemonists. That is, the rapid encoding and retrieval of information is required to be successful in most endeavors. The skilled memory effect is merely a by-product of this necessity.


Evidence for Principles


Evidence for the meaningful encoding principle is provided by studies showing that individuals with exceptional memory exploit extensive domain-specific knowledge, for example, knowledge of mathematics. The best evidence for the use of retrieval structures in exceptional memory comes from the study of DD, the runner who used an elaborate retrieval structure to recall digits. Staszewski found that the temporal patterns of encoding and retrieval were similar, suggesting that DD used the same mechanism during encoding and retrieval. Finally, research shows that LTM encoding and retrieval speed up with practice. For example, initially, NB (also a runner) required 1 minute to memorize 21 digits. By the end of training, however, NB could memorize 21 digits at a rate of one per second. In summary, "with practice and an appropriate knowledge base, normal subjects can considerably increase the amount of information that they can store in memory and then access rapidly, reliably, and flexibly" (p. 245).


Memory for Dinner Orders


Ericsson and Polson investigated exceptional memory for dinner orders in JC, a waiter who could memorize up to 20 dinner orders at a time. Verbal reports showed that JC used mnemonic strategies to encode items from different categories (e.g., meat temperature, salad dressing, etc.). For each order, he encoded an item from a category with previous items from the same category. For example, if an order for an individual included blue cheese dressing, JC would encode the initial letter of the salad dressing with the initial letters from previous orders (e.g., BOOT = blue cheese, oil-vinegar, oil-vinegar, thousand island). He recalled all items of a category clockwise. Thus, the retrieval structure was reflected at encoding and retrieval. When the presentation of information was designed to interfere with JCís encoding strategy, memory performance was greatly reduced.


Expert Mental Calculation


Research on mental calculators suggest that they use retrieval structures to encode and retrieve operands and intermediate results of computations. Chronometric analyses show that, when asked to recall problems, longer pauses occur between hypothesized chunk boundaries. New problems are encoded in terms of knowledge of numbers. A database of knowledge about numbers allows for encoding in terms of an organized semantic network, or for direct retrieval of a product. Speedup is also demonstrated.


Summary, Comments, and Questions


The crux of the skilled memory theory is this: "skilled memory enables experts to rapidly encode, store, and retrieve information within the domain of their expertise and thereby to circumvent the capacity limitations that typically constrain novice performance" (p. 263). The retrieval structure is the mechanism that accounts for this circumvention. A retrieval structure is defined as an abstract, hierarchical knowledge structure used to organize cues used in the encoding and retrieval of information.


The views articulated in this chapter are some of the central tenets of the expertise perspective. Central among these views, expertise reflects, or involves, the circumvention of basic processing capacity limitations. Freedom from such limitations is limited to familiar tasks. Expertise can be described within the confines of the limitations imposed by the human information processing architecture, and is an acquisition. The theory "provides and account of how [experts] employ their most distinguishing asset, knowledge, to overcome basic information-processing constraints that characterize the human mind and memory and achieve extraordinary levels of performance" (p. 264). Ericsson and Staszewski therefore propose that acquired memory skill is the foundation of expertise because it allows for the rapid processing of information.


Back to the Cognitive Science Summaries homepage
Cognitive Science Summaries Webmaster:
JimDavies (jim@jimdavies.org)
Last modified: Tue Mar 7 14:36:59 EST 2000