Untitled

Age-Related Differences in Skilled Performance and Skill Acquisition

Elizabeth A. Bosman

University of Toronto

Neil Charness

Florida State University

(Preprint of Bosman, E. A. & Charness, N. (1996). Age differences in skilled performance and skill acquisition. In T. Hess & F. Blanchard-Fields (Eds.) Perspectives on cognitive change in adulthood and aging (pp. 428-453). New York: McGraw-Hill.)

Skilled performance is an integral part of everyday life. Most of our time is spent in activities that are highly practiced, such as our professional occupation, and activities of daily living, like driving and shopping. Much of the training and education children and young adults receive is intended to help them develop the skills they will require to function as adults. The ability to continuously acquire new skills throughout life is also important. The rapid rate of technological advance means that many jobs and everyday activities change frequently, and new ways of doing familiar tasks must be learned. Rapid change also means that new activities are being continuously created. For example, 30 years ago no one had heard of word processing, programming a VCR, or using a bank machine, yet these activities are now commonplace.

To acquire skill within a specific domain typically requires many years of study, and considerable experience, or practice. For example, doctors must spend many years at university, and then must complete several years of internship before they can be licensed to practice on their own. Skilled tasks thus differ fundamentally from the tasks described in previous chapters. The tasks typically employed by experimental psychologists are unskilled tasks, that is, novel tasks with which the study participants have had no prior experience. This difference is critical because tremendous changes occur in performance as a result of practice. In fact, it is often suggested that the amount of experience with a task is a better predictor of performance than age per se (e.g., Charness & Campbell, 1988). This does not however, imply that all age-related differences could be eliminated through the provision of extensive practice. Nor does this imply that the results of experimental studies employing novel tasks should be disregarded. The goal of many studies is to examine age-related differences in perceptual, motor, and cognitive abilities independently of the effects of experience. Clearly, in order to achieve this goal novel tasks must be used. However, such studies obviously do not provide information regarding the nature of age-related differences for highly practiced tasks, or age-related differences in the effect of practice.

The goal of this chapter is to review empirical studies examining age-related differences in skilled performance and skill acquisition. The chapter is divided into three sections. The first describes the characteristics of skilled performance and outlines a theory of skill acquisition. The second and third sections discuss age-related differences in effects associated with acquired skill (i.e., skills acquired prior to participation in an experimental study), and skill acquisition, respectively.

The Nature of Skill and Skill Acquisition

The differences between the performance of a novice and an expert are quite striking. Consider for example, the differences in performance between someone just learning to skate and a champion figure skater, or, the wobbly motion of a novice cyclist compared to an expert who glides smoothly through narrow spaces. More generally, the performance of a novice is slow, error prone, and awkward, whereas the performance of the expert is fast, virtually errorless, and extremely smooth and fluid. The novice often cannot perform the simplest of tasks while the expert can perform the most difficult tasks with ease. These novice/expert differences exist because the expert has over the course of many years of study and practice developed extensive knowledge within his/her domain of expertise. During performance the expert draws upon this domain specific knowledge, and it is this knowledge, at least in part, that underlies their skilled, or expert, performance. In contrast, the novice has not yet developed domain specific knowledge, and thus their performance lacks the hallmarks of expertise.

Following from Fitts and Posner (1967), Anderson (1983) proposed that skill acquisition is characterized by three stages. The initial declarative stage is one in which the individual learns basic information and facts about the skill he/she is trying to master. This basic information is termed declarative knowledge. When attempting to perform the skill, declarative knowledge is retrieved from long term memory and held in working memory. The information held in working memory is then used to infer the means of accomplishing the desired goal. During the second stage, termed knowledge compilation, active reliance upon the interpretation of declarative knowledge retrieved from memory is reduced. Instead, the individual develops methods for performing the task, called procedures, that directly apply his/her knowledge of the task. Procedures thus represent knowledge about how to perform the task, and correspond to what is termed procedural knowledge. During the final procedural stage, the individual continues to refine the procedures developed during knowledge compilation. In general, during this stage performance becomes much faster. This speed-up in performance is well described by the power law of practice illustrated in Figure 1. This figure illustrates data drawn from Charness and Campbell (1988), and indicates the effect of practice upon the time required to mentally square a two digit number for young, middle aged, and elderly adults. As can be seen in the figure, speed of performance increased with practice for all three groups. However, the increase in speed was greatest during the initial stages of practice, and least during the later stages of practice. Specifically, the most dramatic increases in speed occurred during the first three sessions of practice. During the last two sessions of practice there was very little increase in speed. This finding of minimal improvements in speed during later practice sessions is quite typical of studies examining the relationship between speed and practice, and is taken to indicate that the later stages of skill acquisition have been reached. Further, once the final stages of skill acquisition have been reached, further improvements in performance are unlikely. Finally, in addition to increased speed of performance, during the procedural stage, performance becomes more finely tuned; fewer errors are made, and the individual becomes more skilled at making decisions about how to respond to different aspects of the task environment.

Data from Charness and Campbell (1988) indicating changes in time to mentally square a two digit number as a function of practice session and age group. (Data are for the Class 1 problems reported in the study.)

These stages of skill acquisition can be illustrated by considering the process of learning to touch type. During the initial declarative stage the individual learns information such as how to position their hands, where different characters are located on the keyboard, and which finger is associated with each key. When attempting to type the individual spends a lot of time trying to remember where different characters are located and which finger is supposed to strike each key. In general, performance is extremely slow and error prone, and is characterized by many long pauses while the individual searches the keyboard for a particular key. During knowledge compilation the individual spends less time actively trying to remember where keys are located and which fingers strike each key. Instead, procedures are developed that enable the individual to directly strike the appropriate key. Performance becomes faster and less error prone, and the keyboard is consulted less often. Finally, during the procedural stage, typing speed continues to increase while the number of errors decrease. As well, the individual becomes more skilled at typing characters that occur infrequently, such as numbers.

The amount of practice required to develop expertise depends upon the complexity of the task. Very simple tasks can be mastered in a matter of a few hours while more complex tasks may require hundreds or thousands of hours of practice, and years of study. For example, it has been estimated that approximately 600 hours of practice are required to achieve a typing speed of 60 net words/minute (Gentner, 1988). For more complex domains such as chess, music, and mathematics, it appears that at least 10 years of intensive study and practice is required to achieve the highest levels of performance (Ericsson & Charness, 1994; Ericsson, Krampe, Tesch-Römer 1993).

Age-related Differences in Acquired Skills

It is often noted that although experimental studies consistently demonstrate age-related declines on the vast majority of laboratory tasks, older adults perform as well as, if not better than, younger adults on a wide range of skills and occupations. For example, no relationship has been found between job productivity and age (McEvoy & Cascio, 1989). One potential explanation for this paradox is that experimental studies typically examine unskilled performance, whereas everyday performance usually involves tasks that are highly familiar and well practiced (Salthouse, 1990). Given that relative to their performance on experimental tasks, older adults perform quite well on everyday tasks, it has been suggested that skill may enable older adults to perform at levels comparable to younger adults despite age-related declines in basic cognitive processes (Bosman, 1993; Charness, 1981a, 1981b; Charness & Bosman, 1990; Rybash, Hoyer, & Roodin, 1986; Salthouse, 1987, 1989, 1990). Consequently an important goal of research examining age-related differences in skilled performance is to identify the means by which older adults may be able to perform at levels comparable to younger adults. As part of this objective, four conceptual frameworks, accommodation, compensation, maintenance, and encapsulation, have been proposed as outlining the means by which older adults may be able to perform at levels comparable to younger adults (Salthouse, 1990).

According to the accommodation perspective older adults have acquired metacognitive knowledge related to their domain of expertise that enables them to continue to perform well. Specifically, within their domain of expertise, older adults have learned to identify the conditions under which their performance is adversely affected by age-related declines, and to selectively avoid these conditions (Salthouse, 1990). For example, an elderly individual may give up driving at night because under these conditions their driving performance becomes unsafe. However, this same individual may continue to drive during the day because under these conditions their driving performance is unaffected by age-related declines. This person has accommodated to an age-related decrease in their driving skill, specifically, unsafe driving at night, by restricting their driving to daytime conditions. Although it may appear to the casual observer that the individual's driving performance has been unaffected by age-related declines, this appearance is the result of the individual selectively avoiding conditions under which he/she can no longer continue to perform well. The essence of accommodation is an ability to maintain high levels of performance by selectively avoiding conditions under which it is impossible to perform well. Thus, to the extent that older adults can identify these conditions, and, to the extent that these conditions can be avoided, older adults will continue to exhibit high levels of skilled performance.

In contrast, the compensation perspective suggests that the negative impact of age-related changes is offset by the development of compensatory mechanisms that enable older adults to perform at levels comparable to younger adults. According to such a proposal, aging is characterized as consisting of two interacting processes; a decline in perceptual, motor, and cognitive processing capabilities, and an accumulation of specialized knowledge (i.e., procedural and declarative knowledge, Anderson, 1983) that offsets these declines in the domain of expertise (Charness & Bosman, 1990; Salthouse, 1987, 1989, 1990). More specifically, according to this perspective, older adults develop a compensatory mechanism, or strategy, which enables them to achieve comparable levels of performance in a qualitatively different manner relative to younger adults of equivalent skill. For example, an elderly person may find that their driving performance is adversely affected by age-related slowing. However, the individual may compensate for these age-related declines by increasing the skill with which he/she evaluates traffic and road conditions. Thus, although the individual may react more slowly the events on the road, an increased ability to evaluate traffic and road conditions may compensate for this age-related slowing. It should be noted that compensation is not restricted to strategies that are consciously employed by the older adult. Compensatory mechanisms may develop unconsciously over time as the individual attempts to maintain performance in the presence of age-related declines.

Unlike the compensation perspective which suggests that skill enables older adults to compensate for age-related declines, the maintenance, or selective sparing perspective argues that skill prevents age-related declines from occurring. More specifically, the maintenance perspective suggests that older adults are able to maintain skilled performance because extensive practice has prevented age-related declines from occurring in the component processes underlying skill (Charness & Bosman, 1990; Salthouse, 1987, 1989, 1990). For example, according to this perspective, decades of experience driving would prevent an older person from experiencing age-related declines in their driving skill.

The final perspective, encapsulation (also referred to as compilation), argues that once a skill has been acquired, it is not affected by age-related declines in the efficiency of the underlying component processes. An analogy that has been used to describe the encapsulation perspective is that of a computer program that has been compiled from several subroutines. Once the program is compiled, changes to the subroutines will not affect the compiled program (Salthouse, 1989, 1990). Similarly, during the initial stages of skill acquisition different component processes may contribute to skilled performance. However, once the skill is established, it may become independent of these component processes. Consequently, changes in the efficiency of these component processes should not influence the overall level of performance. Thus, once a skill is established, age-related declines in the efficiency of the underlying component processes should not affect performance because the skill has become independent of these component processes. The skill can thus be said to incorporate, or encapsulate, a more efficient form of functioning (Rybash et al., 1986; Salthouse, 1989; 1990). For example, according to this perspective, an elderly individual should not exhibit declines in their driving performance because of the encapsulation of their driving skill.

An interesting implication of the encapsulation perspective is that whether or not an individual exhibits age-related declines in the performance of a particular skill may depend upon the age at which the skill was learned. Specifically, if a skill was learned at a young age before age-related declines began to affect performance, then the encapsulated skill would presumably reflect the level of performance the individual was capable of when they were young. However, if the skill was learned during middle or old age when the influences of age-related declines upon performance were becoming apparent, then presumably the encapsulated skill would reflect a lower level of performance.

Currently, there is very little empirical evidence regarding the importance of accommodation, compensation, maintenance, and encapsulation in enabling older adults to perform at levels comparable to younger adults. No research has been found that has examined the importance of accommodation and encapsulation. However, some research has examined the roles of compensation and maintenance in the skilled performance of older adults. Following is a discussion of this research.

Evidence for Compensation

Evidence suggesting that older adults may compensate for age-related declines comes from quasi-experimental studies examining age and skill-related differences in chess (Charness, 1981a, 1981b, 1981c), bridge (Charness, 1979, 1983, 1987), and transcription typing (Bosman, 1993, Salthouse, 1984; Salthouse & Saults, 1987). Before providing a more detailed description of the results of these studies, it is useful to examine in some detail the methodological approach underlying most of the above studies. The basic feature of the approach is the selection of a sample that varies widely in terms of both age and skill level, but in which age and skill are uncorrelated (e.g., Bosman, 1993, Charness, 1981a, 1981b, 1983, 1987; Salthouse, 1984; Salthouse, & Saults, 1987). The rationale is that such a sample makes it possible to assess independently the effects of age and skill. Further, it is also possible to assess if there are qualitative differences in the means by which individuals of different ages achieve the same overall level of performance. Specifically, an examination of age effects on descriptors of skill other than overall performance may reveal differences that indicate how older adults are able to perform as well as younger adults.

A potential disadvantage of using a sample in which age and skill are uncorrelated is that the lack of correlation may be achieved by selecting the fittest old, and the least fit young, resulting in an unrepresentative sample. That is, the elderly adults in such a sample may be more elite than the younger adults. For example, when young, the elderly adults may have exhibited higher overall ability relative to the younger adults. Such a sample makes it difficult to determine how skill enables older adults to perform at levels comparable to younger adults because the pattern of results could be attributable to the biased sample, rather than the effect of skill (for a discussion see Charness & Bosman, 1990; Salthouse, 1989). Consequently, in addition to establishing a lack of correlation between skill and age, it is important to demonstrate that the sample is representative of normal aging. This issue can be addressed by demonstrating that within the sample there are the typical age-related patterns on domain-unrelated measures (Charness & Bosman, 1990). Examples of measures that can be employed for this purpose are measures of crystallized intelligence, such as vocabulary tests, which typically show little, if any decline, with age, and measures of processing speed, such as reaction time, which show significant slowing with age (Kausler, 1982; Salthouse, 1982). Finally, a limitation of this approach is that it does not address the issue of whether or not the overall level of performance declines with age. The requirement of obtaining a sample in which skill and age are uncorrelated precludes this possibility. For example, the sample may consist of elderly individuals who have consistently performed at a high level throughout their adult life, and of elderly individuals whose performance level has declined during their adult years, but who still perform at a higher level than novices.

Chess and Bridge

Several studies specifically designed to examine age-related differences in chess skill, and in which the correlation between age and skill was not significant, were conducted by Charness (1981a, 1981b). The measure used to determine chess skill was competitive chess rating, which is an index that varies as a function of the individual's performance in chess competitions. Age and skill-related differences were assessed for two tasks. In the first, the select-a-move task, players were asked to select the best move for their side for each of four chess positions. In another task, players were shown 20 end-game positions (i.e., a configuration of chess pieces that occurs near the end of the game), and were asked to categorize each position as win, lose, or draw. Both of these activities, move selection, and evaluating a board position, are integral parts of a chess game. Thus, not surprisingly, the accuracy of performance in both tasks increased with skill. More importantly, no age effects were observed for the value of the chosen moves, as well as for the number of correct evaluations.

Although there were no age-related differences in chess rating and in performance on these skill-related tasks, older chess players exhibited the expected memory deficits on a skill-related memory task. Memory performance was assessed by unexpectedly asking players to recall the 4 chess positions that were employed in the select-a-move task. There was the expected effect of skill indicating that more skilled players exhibited better recall. More importantly, there was an effect of age indicating that older players recalled less than comparably skilled younger players. A similar age-related decline in memory performance was documented in a study not specifically designed to examine the effects of age upon chess skill (Charness, 1981c). When asked to recall briefly presented (1, 2, 5 seconds) chess positions, older players recalled significantly less than younger players of equivalent skill.

Generally, studies examining age-related differences in bridge skill have produced a similar pattern of results indicating no age-related differences in skilled performance accompanied by age-related deficits on skill-related memory or speeded performance measures (Charness, 1979, 1983, 1987). One obstacle to research examining bridge skill is the absence of a valid measure of skill that is not confounded with age. Specifically, the commonly used index of bridge skill, master points, is cumulative across the life span. Master points can only be gained, not lost, and thus this measure may not accurately reflect a player's current skill level independently of age. Nonetheless, despite this problem, the logarithm of master point total proved to be sensitive to variations in performance on experimental tasks. Skill and age-related differences were assessed for a task in which players were shown a bridge hand and required to produce an opening bid as quickly as possible, and for a task that examined performance on a component of the bidding task, namely, summing the number of honour card points. For both tasks there was an effect of skill indicating superior performance by more skilled players. There was also no effect of age for either task indicating that older adults performed as well as comparably skilled younger players. However, the results did indicate that older players bid more slowly than younger ones. The latter finding is consistent with the usual findings of age-related slowing on speeded tasks. In tournament bridge, however, you are not permitted to bid quickly! All oral bids must be made in an even tempo when it is your turn to announce our bid.

Despite the absence of age-related effects for this aspect of bridge skill, there were age-related declines on a skill-related memory measure. When asked to recall previously presented bridge hands there was the expected effect of skill indicating better recall performance by more skilled players. More importantly, there was an effect of age indicating that older players recalled less than comparably skilled younger players (Charness, 1979).

Thus, a consistent pattern of empirical results emerges from studies examining age-related differences in chess and bridge skill. The results of these studies document a situation in which there were no age-related differences in overall skill, or on skill-related tasks, despite the presence of age-related memory deficits for skill-related memory tasks. If these memory tasks truly reflect a component of chess and bridge skill, then, in an indirect way, these studies suggest that the older players were able to compensate for memory deficits, although the means by which they were able to do so were not identified.

There is also some intriguing, though weak, evidence for selective preservation of practiced cognitive activities in bridge. Clarkson-Smith and Hartley (1990) showed that older bridge players who were comparable in age and activity levels to a control group of non-bridge playing older adults had greater working memory and reasoning capacity, but were not superior on other tasks such as reaction time or vocabulary. Clarkson-Smith and Hartley noted that this might be due either to bridge-playing demanding working memory engagement and reasoning hence preserving those practiced abilities, or to selective recruitment to bridge. That is, in the latter case, only those with better than average working-memory and reasoning ability take up the game in the first place, and even if there are declines with age, bridge players on average would retain their superiority. (The work on spatial ability in architects to be discussed later is consonant with the latter interpretation.) Longitudinal research would be required to distinguish between these two alternative interpretations.

Transcription Typing

Among the skills that have been examined by investigators interested in age-related differences in skilled performance, transcription typing is unique because it is the only skill for which there is more direct evidence suggesting how older adults may be compensating for age-related declines. Typing skill is typically indexed by net words/minute, a combined speed-accuracy measure commonly employed in educational and professional settings. Previous research has indicated that typing skill as indexed by net words/minute does not decline with age. That is, older typists perform just as well as younger typists. However, despite the absence of age-related declines in overall typing skill, older typists exhibit deficits on the motoric processes underlying typing (Bosman, 1993; Salthouse, 1984). Thus, similar to the results of studies examining age-related differences in chess and bridge skill, older typists exhibited no declines in overall skill coupled with declines on processes underlying skill. The implication is that somehow the older typists were able to compensate for their motoric deficits.

The motoric aspects of transcription typing are thought to involve at least two classes of processes, translation, or motor programming, and execution. Translation processes function to convert characters in the text into motor programs. Execution refers to those processes involved in implementing motor programs so that the appropriate keys are struck (see Salthouse, 1986 for discussion). Evidence for age-related deficits in the motoric processes of typing were obtained from a digraph typing task (Bosman, 1993; 1994). During this task, two letters (i.e., a digraph such as di or si), were displayed on the computer screen, and the typist had to type the two letters as quickly as possible. Performance on this task yielded two measures, initial latency, which reflected time to hit the first keystroke, and inter-keystroke latency which reflected the latency between the two keystrokes. The advantage of using these two measures is that they are differentially sensitive to the translation and execution components of motor performance. Specifically, initial latency primarily reflects the time associated with translation processes, whereas inter-keystroke latency primarily reflects the time associated with execution processes. The results of the digraph typing task were consistent with the interpretation that low skill older typists exhibit a deficit for both the translation and execution processes underlying transcription typing, whereas high skill older typists exhibit a deficit for translation processes only. Specifically, the results for initial latency indicated age-related slowing at all levels of typing skill. Thus, this result is consistent with an age-related deficit in translation, regardless of skill level. The results for inter-keystroke latency revealed an age by skill interaction indicating that at low skill levels the usual age-related slowing was observed, whereas at high skill levels no age-related slowing was apparent. The implication of this pattern of findings is that low skill older typists exhibit an execution deficit, whereas high skill older typists do not. Why the high skill older typists did not exhibit an execution deficit is unclear, but it may represent an instance in which expertise enables older adults to perform at levels comparable to younger adults despite age-related declines. However, despite this, while the nature of age-related deficits varied as a function of skill, older typists exhibited motoric deficits.

Given these results, an important issue is how the older typists were able to maintain overall typing performance despite their motoric deficits. The available evidence suggests that older typists may be compensating for motoric slowing by beginning keystroke preparation sooner than younger typists. Evidence for this compensatory mechanism comes from research examining age-related differences in the size of the preview span. The preview span is defined as follows. A typist always looks a few characters ahead of the character currently being executed. For example, when typing the sentence "The black dog chewed a bone.", while typing the b in black the typist will be looking ahead in the text at other letters, such as the d in dog. This gap between the character being fixated by the eye, and the character being typed by the hand is referred to as the preview span (e.g., Bosman, 1993; Salthouse, 1984, 1985, 1986; Salthouse & Saults, 1987). The size of the preview span varies with skill, with more skilled typists having a larger preview span than less skilled typists. As well, if preview is restricted so that the amount the typist can read ahead in the text is less than their preview span, there is a dramatic decrease in typing speed. For example, if a typist has a preview span of 6 characters, but text preview is restricted so that they can only read 3 characters ahead in the text, their typing speed is slowed considerably. This relationship between typing speed and preview span has been interpreted as indicating that the preview span represents the point at which keystroke preparation begins (Bosman, 1993; Salthouse, 1984, 1985, 1986). Research examining age-related differences in the size of the preview span have found that preview span size increases with age (Bosman, 1993; Salthouse, 1984). Figure 2 illustrates the age-related increase in preview span. As can be seen in the figure, preview span size increases with skill. However, more importantly, preview span size increases with age. This is indicated by the regression lines indicating the predicted preview span size for individuals aged 30, 45, and 60 years.

Scatterplot of preview span size by skill for data from Bosman (1993). The number representing each subject corresponds to the first number of the subject's age. The regression lines represent the predicted preview span for individuals aged 30, 45, and 60 years.

Given that the preview span is thought to indicate the point at which keystroke preparation begins, the implication of the age-related increase in preview span is that older typists may compensate for age-related slowing of motor performance by beginning keystroke preparation sooner (Bosman, 1993; Salthouse, 1984). Support for this interpretation has been obtained by manipulating the amount of text preview during typing, and examining the impact of this manipulation upon the relationship between typing speed and age. Given that there is no age-related decrease in typing speed under normal typing conditions when text preview is unrestricted, it would be expected that when the amount of text preview is equal to or greater than the average preview span (i.e., 7-10 characters), there should be no correlation between typing speed and age. However, when text preview is restricted so that it is smaller than the average preview span (i.e., 1-3 characters), if older adults do rely upon greater advance preparation to compensate for age-related motoric deficits, then it should be the case that there is a correlation between age and typing speed indicating that the older typists are slower. In general, it has been found that when text preview is large, and therefore advance preparation of keystrokes is possible, there is no correlation between age and typing speed. However, when text preview is small, there is a significant correlation between age and typing speed indicating that the older adults are slower (Bosman, 1993; Salthouse, 1984). This pattern of results is illustrated in Figure 3. Thus, the age-related increase in preview span size coupled with greater age-related slowing of typing speed under conditions of restricted preview suggests that a reliance on greater advance preparation may be a compensatory mechanism used by older typists.

However, a limitation of the above research examining age-related differences in preview span size is that the research is cross-sectional in nature. Consequently, it is possible that the age-related increase in preview span reflects cohort differences. That is, preview span size does not increase with age, rather, relative to the younger adults, the older adults simply had larger preview spans to begin with. Longitudinal research tracking the size of the preview span across the adult life span would be required to document that preview span size does increase with age, and that such an increase functions as a compensatory mechanism.

Correlation between age and inter-keystroke interval (IKI) by preview level for data from Bosman (1993).

Evidence Regarding Maintenance

Only one study has been found that examined whether or not increased experience is associated with the maintenance of cognitive abilities with age. The relevant study examined the effects of age and experience on spatial visualization ability for architects (Salthouse, Babcock, Mitchell, Skovronek, and Palmon, 1990). Before considering the results of this study, however, it is useful to examine certain relevant methodological issues. First, because the goal of such a study is to determine if extensive experience is associated with the maintenance of general cognitive abilities, specific measures of overall performance in the relevant domain are typically not obtained. Consequently, studies of this type do not address the issue of whether or not performance on skilled tasks declines with age. Another critical issue is that the underlying cognitive ability assessed must be shown to be an important determinant of performance in the skill domain of interest. Further, the measure of the underlying ability must have construct validity, that is, it must measure the ability of interest, and not something else. Finally, age-related differences for the ability measure should be compared for groups of individuals who are both skilled and unskilled in the domain of interest. To demonstrate that maintenance has occurred it must be shown that age-related declines do not occur for individuals with the relevant experience, and do occur for those without the relevant experience.

In the study of interest, Salthouse et al. (1990) defined spatial visualization as the ability to interpret two dimensional drawings of three dimensional objects. Spatial visualization ability has been found to be predictive of success in domains such as drafting and geometry, and more importantly, was rated by the architects participating in the study as highly relevant to their occupational activities. Equally important, the measures used to assess spatial visualization were rated by the architects as requiring the same type of cognitive processes as did their professional activities. The relationship between age and spatial visualization abilities was compared for two groups of individuals, those who were architects, and those who were not. The results indicated that there were age-related declines in spatial visualization for both groups, and that the rate of decline was roughly comparable. The implication is that extensive experience with spatial visualization is not associated with the maintenance of this cognitive ability. However, the results also indicated that older architects performed better on measures of spatial visualization than did older adults who were not architects. These results were interpreted as indicating preserved differentiation of cognitive abilities with age. That is, relative to the non-architects, the architects probably had superior spatial visualization abilities as young adults, and this initial difference was preserved with increased age.

Summary and Conclusions

At the beginning of this section it was noted that older adults often perform comparably to younger adults in a variety of domains, and four conceptual frameworks that could account for this observation, accommodation, compensation, maintenance, and encapsulation, were outlined. Currently, much of the available evidence supports the compensation perspective. The results from studies examining age-related differences in chess, bridge and typing skill suggest either indirectly or directly that older adults perform at levels comparable to younger adults by compensating for age-related declines. Thus, this research provides some support for the argument advanced by the compensation perspective which suggests that aging can be characterized as consisting of two interacting processes; a decline in perceptual, motor, and cognitive processing capabilities, and an accumulation of specialized knowledge that offsets these declines in the domain of expertise. However, the results of these studies should not be regarded as conclusive. In the case of studies examining chess and bridge skill, only indirect evidence suggesting compensation was found. Similarly, findings suggesting that older typists compensate for age-related slowing through a greater reliance upon advance preparation of keystrokes are open to alternative interpretations such as cohort differences. No evidence was found to support the maintenance perspective, although the results of a single study cannot be definitive. Similarly, no evidence is currently available regarding the accommodation and encapsulation perspectives. A more complete understanding of age-related differences in the performance of acquired skills will require additional research to investigate the validity of the compensation, maintenance, accommodation, and encapsulation perspectives. It is unlikely that elderly adults employ compensatory mechanisms in every skill that they perform, there are undoubtedly many ways in which skilled performance changes during the adulthood.

Age-related Differences in Skill Acquisition

As indicated by the cliché "You can't teach an old dog new tricks", it is often assumed that older adults have difficulty acquiring new skills. However, despite this assumption there is considerable empirical evidence indicating that older adults are able to learn new skills. Numerous studies have indicated that the performance of older adults on a variety of tasks improves considerably with training (Willis, 1985, 1987). Accordingly, the focus of this section will not be on demonstrating that older adults can acquire new skills, but rather on examining age-related differences in skill acquisition. One important issue is whether or not age-related differences are reduced or eliminated during skill acquisition. Given the dramatic improvements in performance resulting from extensive practice, it might be expected that age-related differences will be minimal once young and elderly adults become practiced at a task. In contrast, it has also been argued that age-related differences will increase with extensive practice (Kliegl & Baltes, 1987; Kliegl, Smith & Baltes, 1989). The rationale for this prediction is as follows. Extensive practice will enable both young and elderly adults to optimize their performance, and thus after practice both groups will be performing close to their potential. However, despite this optimization, age-related declines in perceptual, motor, and cognitive abilities will still adversely affect the performance of older adults. Further, the optimization of performance means that the differences between the age groups will increase because both groups will be performing close to their maximum.

Another important issue is whether or not certain aspects of skill acquisition pose particular difficulty for older adults. For example, using Anderson's (1983) theory of skill acquisition as a framework, age-related deficits in working memory may cause the initial declarative stage of skill acquisition to be particularly difficult for older adults. This may further lead to an age-related difficulty in knowledge compilation and the proceduralization of skill. These issues will be considered in the following discussion of selected studies examining age-related differences in skill acquisition for mental arithmetic, memory and visual search tasks, and serial recall of word lists. However, before reviewing these studies certain methodological issues will be examined.

Methodological Issues

Given the importance of practice in skill acquisition, a critical methodological criterion for experimental studies of skill acquisition is to provide sufficient practice for skill to develop. Generally, this is established by demonstrating that additional practice is unlikely to result in further marked increases in performance. Typically, performance as a function of practice is examined to determine if the later stages of skill acquisition have been reached, and if further improvements in performance are unlikely. Specifically, referring back to Figure 1 which illustrates the power law of practice, it is important to obtain a functions such as the ones obtained in this figure in which improvements in performance during the later stages of practice are quite minimal. Such a finding would suggest that performance has reached asymptotic levels, and that additional practice is unlikely to result in additional marked improvements in performance. The need to provide extensive practice is further complicated by the fact that within the context of an experimental study it is not feasible to provide hundreds, let alone thousands of hours of practice. Thus, it is not possible to teach study participants how to type or play chess. Instead, the tasks employed in skill acquisition studies must be of modest complexity so that they can be mastered in a comparatively short period of time.

Mental Arithmetic

The first study to be considered is one in which subjects were taught to mentally square two digit numbers such as 64 or 85 using the algorithm illustrated in Figure 4 (Charness & Campbell, 1988). The logic behind the algorithm is that it provides a method for breaking the computation down into simpler components that can be easily computed, and then combined to produce the answer. Another feature of the algorithm is that it makes heavy demands upon working memory because the components must be maintained in memory while simultaneously implementing the squaring algorithm. Young, middle aged, and elderly adults practiced mentally squaring two digit numbers using the above algorithm for 6 one-hour sessions. In addition, performance calculating the different components of the algorithm in isolation was also measured.

To square a number, N, find a constant, C, to add to or subtract from N to bring it to the nearest multiple of ten, NMT. Then subtract or add C from N to produce the other number, OTN. Multiply OTN by NMT by first multiplying the decade of OTN by NMT to get the first product, P1. Then, multiply the right digit of OTN by NMT to get the second product, P2. Now add P1 and P2 to get the SUM. Square the constant, C, to get C2. Finally, add C2 and SUM to get the answer.

For example, the procedure for squaring 59 is as follows:

Step 1: 59 + 1(C) = 60, NMT
Step 2: 59 - 1 (C) = 58, OTN
Step 3: 50 (decade of OTN) x 60 (NMT) = 3000, P1
Step 4: 8 (right digit of OTN) x 60 (NMT) = 480, P2
Step 5: 3000 (P1) = 480 (P2) = 3480, SUM
Step 6: 1 (C) x 1 (C) = 1, C2
Step 7: 3480 (SUM) + 1 (C2) = 3481

Outline of the algorithm for mental squaring employed by Charness and Campbell (1988).

From the perspective of assessing age-related differences in skill acquisition, the mental squaring task provides an opportunity to examine age-related differences in speedup and knowledge compilation. More specifically, performance on the squaring task depends in part on the speed and accuracy with which the components can be computed. However, performance also depends upon how efficiently the squaring algorithm can be implemented. Thus, the total time required to square a number is determined by the amount of time required to compute the components, and the time required to implement the squaring algorithm. Reductions in the amount of time required to compute the components are indicative of the extent to which practice has lead to an increase in speed. Reductions in the amount of time required to implement the algorithm are also indicative of increases in speed, but more importantly, are indicative of the extent to which the squaring algorithm has been efficiently compiled into procedures.

An index of how much time is required to implement the squaring procedure can be obtained by determining the ratio of the time required to compute the components to the total time required to compute the answer. If there was no cost associated with implementing the algorithm, the total time required to compute the solution would be equal to the time required to compute the components. That is, the ratio of the summed components time to total algorithm time would equal one. To the extent that the ratio drops below one (and approaches zero), it indicates that relatively more time is spent implementing the algorithm, and less time is spent computing the components. Of course, it should be noted using the ratio of component time to total time to draw conclusions about the role of knowledge compilation is based on certain assumptions. First, it assumes that the different components are computed independently of each other when implementing the algorithm. If this is not the case it is not valid to estimate the time spent computing the components of the algorithm by summing the time required to compute each component. The ratio also assumes that computation of the components and implementation of the algorithm are independent. If both are performed in parallel the ratio is not a good indicator of the relative amount of time devoted to either. Currently it is not possible to assess whether or not these assumptions are correct. However, despite this, we still believe that it is instructive to examine the change in the ratio of component time to total time as a function of practice and age in order to assess age-related differences in knowledge compilation. Although the ratio may not precisely reflect the relative amount of time devoted to computing the components versus implementing the algorithm, it seems likely that it will at least provide a rough estimate.

The results indicated that for all three age groups, performance improved significantly with practice, and that by the end of the sixth practice session performance had reached asymptotic levels suggesting that subjects had reached the final stages of skill acquisition. Specifically, during practice the time required to square a number decreased dramatically. Part of this speedup was attributable to increased speed in computing the components. However, the greatest increase in speed was associated with greater efficiency in implementing the squaring algorithm. Specifically, at the beginning of practice, the ratio of component time to total time was significantly less than one suggesting that subjects spent most of their time implementing the algorithm. At the end of practice the ratio had increased substantially suggesting that subjects had become more efficient at implementing the algorithm. Thus, skill acquisition for this task was primarily attributable to knowledge compilation that resulted in the efficient implementation of the squaring algorithm.

An examination of age-related differences for time to compute the components and time to compute the answer indicated that generally the rate of speedup was greater for the older adults, but that age-related differences still remained at the end of practice. For time to compute the components, at the beginning of practice both the middle aged and elderly were significantly slower than the young adults. The middle aged showed the greatest improvement with practice, and by the end of the practice they were not significantly slower than the young adults. Although the elderly improved with practice, they were still slower than the young adults at the end of practice. The results for total time required to square a number produced a similar pattern of results. At the beginning of practice the elderly were slower than the middle aged who were in turn slower than the young. There were age-related differences in the rate of improvement with the performance of the middle aged adults increasing at a faster rate than for either the young or elderly adults. At the end of practice, time to square a number did not differ for young and middle age adults, but the elderly were still slower. Thus, the results for time to compute the components, and time to compute the final answer indicated that practice had the greatest impact upon age-related slowing for the middle aged, and had less impact upon age-related slowing for the elderly, although the latter did improve substantially.

Results examining the ratio of time to compute components to total time to compute the answer indicated that at the beginning of practice the ratio for the elderly was significantly smaller than for the young. The middle aged were intermediate between the young and elderly, their ratio did not differ significantly from either of the other groups. The implication is that at the start of practice the older adults spent significantly more time implementing the algorithm. With practice there was a significant increase in the ratio for all three groups suggesting the knowledge compilation was occurring for all three groups. However, age-related differences did not change as a function of practice. That is, the ratios for the elderly and middle aged adults were still significantly smaller than for the young adults. The implication is that the older adults were less efficient at implementing the squaring algorithm even after considerable practice. This suggests that increased age may be associated with increased difficulty in knowledge compilation and the proceduralization of skill.

Memory and Visual Search

Memory and visual search tasks have frequently been used to examine age-related differences in knowledge compilation (for discussion see Fisk and Rogers, 1991). Age-related differences in skill acquisition will be considered first for memory search tasks, and then for visual search tasks. In a typical memory search task, at the beginning of each trial the subject is shown the memory set (or targets), which usually consists of 1-4 symbols or letters. The subject studies the memory set for a several seconds, and then the memory set is replaced with a single probe item. The subject's task is to indicate whether or not the probe item is one of the target items, or if it is a distractor item. The task is termed memory search because the subject must presumably search the contents of their memory in order to determine whether or not the probe was part of the memory set (for discussion see Fisk & Rogers, 1991).

Another critical feature of memory search tasks is whether or not the relationship between target and distractor items is consistent or variable. When consistent mapping is employed some stimulus items are used only as targets (i.e., they are always part of the memory set and never distractors), while others are used only as distractors (i.e., they are never part of the memory set). For example, if the stimulus set consists of A E G L K M and P, under consistent mapping, these stimuli would be divided into two groups, targets and distractors. When variable mapping is employed the stimulus set is not divided into target and distractor groups; a stimulus can be used as both a target and a distractor. This difference between consistent and variable mapping is important because the development of skill at detecting targets as a consequence of extensive practice is greatest when consistent mapping is used. When variable mapping is used, subjects do not develop as much skill at detecting targets. The reasons for this are as follows. Under consistent mapping subjects eventually learn which stimuli are targets, and which are distractors. Subjects can then use this information during knowledge compilation to develop procedures for automatically detecting targets. In contrast, when variable mapping is employed, stimulus identity cannot be used to determine if a stimulus is a target or a distractor. Consequently, even with extensive practice, it is not possible for subjects to develop procedures for automatically detecting targets (for discussion see Fisk & Rogers, 1991). Given that only practice with consistent mapping has been shown to lead to skill acquisition, we will focus on studies that employed consistent mapping when examining age-related differences in skill acquisition for memory search tasks.

Performance on memory search tasks can be summarized by two measures, mean reaction time as a function of memory set size, and the slope of the function relating reaction time to set size. Figure 5 illustrates the changes in performance in this task as a result of practice. Before extensive practice it is typically found that reaction time increases with set size. The reason for this is that on each trial the subject must search their memory comparing each item in the memory set to the probe item. The slope of the function relating reaction time to set size is thought to indicate the amount of time required to evaluate one item in memory. The more items in the memory set, the more time required to decide if the probe is part of the memory set. However, after extensive practice, in addition to decreasing dramatically, reaction time no longer varies as a function of set size, and the slope of the function relating reaction time to set size approaches zero. This reduction in slope is thought to reflect the development of procedures during knowledge compilation for the automatic detection of targets. Such procedures eliminate the need to search memory in order to determine if the probe is part of the memory set, and consequently the slope of the function relating memory set size to reaction time is greatly reduced. From the perspective of assessing age-related differences in skill acquisition for this task there are two important issues. First, does reaction time decrease at the same rate for young and elderly adults, or is the rate of decrease faster or slower for the elderly. Second, does knowledge compilation, as indexed by decreases in slope, vary with age.

Hypothetical data indicating changes in performance on a memory search task as a function of practice.

Salthouse and Somberg (1982) conducted an extensive study in which young and elderly adults practiced a consistently mapped memory search task for 50 one-hour sessions. At the beginning of practice, relative to young adults, the elderly had significantly slower reaction times. The slope relating reaction time to set size was also larger for the elderly suggesting that they searched memory at a slower rate. During the initial stages of practice the results indicated that reaction time decreased at a faster rate for the elderly adults. However, after the initial stages of practice, the rate of decrease in reaction time was equivalent for young and elderly adults. After 50 hours of practice, the elderly adults were still significantly slower than the young. Further, an examination of reaction time as a function of practice indicated that performance had reached asymptotic levels suggesting that marked improvements in performance as a function of additional practice were unlikely. Thus, practice reduced, but did not eliminate, age-related differences in reaction time. A different pattern of results emerged for the slope measure. Although initially the slope for the elderly was larger, by the end of practice there were no age-related differences in slope, and the slope for both young and elderly approached zero. A similar pattern of results for both reaction time and slope was reported by Fisk and Rogers (1991), who also examined age-related differences in skill acquisition for a memory search task under consistent mapping. The implication is that for this type of task, extensive practice reduces, but does not eliminate age-related slowing of reaction time. However, in contrast to the results obtained for the mental squaring task, knowledge compilation was equally effective for the young and elderly.

Research examining the effect of extensive practice upon age-related differences in visual search reveals a different pattern of results. In a typical visual search task, at the beginning of each trial the subject is shown a single target item, usually a symbol or letter. After the subject has studied the target, it is replaced with a display set that usually varies in size between 1-4 items. The display set may contain the target item, or it may consist only of distractor items. The subject's task is to indicate if the target item shown at the beginning of the trial appears in the display set. This task is termed visual search because the subject must visually scan the display in order to determine if the target is present. Visual search tasks are also presented using either consistent or varied mapping, and, as is the case with memory search, consistent mapping leads to the greatest development of skill at detecting targets (for discussion see Fisk & Rogers, 1991).

Performance on visual search tasks is also summarized by the two measures used to describe memory search performance, mean reaction time as a function of display set size, and the slope of the function relating reaction time to set size. The effect of practice upon these two measures is the same as that observed for memory search. Before extensive practice reaction time increases as a function of display set size, presumably because the subject must search the display comparing each display item to the target. However, after extensive practice, reaction time is significantly reduced, and the slope of the function relating reaction time to set size approaches zero. As was the case with memory search, this reduction in slope is thought to reflect the development of procedures during knowledge compilation for the automatic detection of targets. From the perspective of assessing age-related differences in skill acquisition for this task, again there are two important issues. Does reaction time decrease at the same rate for young and elderly adults, or does the rate of decrease vary with age. Secondly, does knowledge compilation, as indexed by decreases in slope, vary with age.

Fisk and Rogers (1991) had young and elderly subjects practice a visual search task under consistent mapping for either 6 (Experiment 1), or 12 (Experiment 2) hours. There results indicated that reaction time decreased significantly as a function of practice for both the young and elderly, but that the rate of decrease was greater for the elderly. However, at the end of practice the elderly were still significantly slower than the young. As well, the results indicated that by the end of practice reaction time performance had reached asymptotic levels suggesting that further improvements in performance were unlikely. With regard to the slope relating reaction time to set size, again the results indicated that slope decreased with practice for both the young and elderly adults suggesting that both groups were able to develop procedures for automatically detecting the targets. However, the decrease in slope was much less for the elderly, and at the end of practice the elderly had a significantly larger slope than the young. Thus, similar to the results obtained for the mental squaring task, for visual search tasks it appears that knowledge compilation is less effective for older adults (for discussion see Fisk & Rogers, 1991).

Memory Performance

The next study to be considered is one in which young and elderly adults were given extensive training using a mnemonic technique known as the Method of Loci (Kliegl et al. 1989). The Method of Loci involves the use of a highly familiar and ordered sequence of locations as a structure for encoding and retrieving new information. Each item to be remembered is associated with one of the locations in the sequence through the formation of a visual image. The first item to be remembered is associated with the first location, the second item with the second location, and so on. The rationale behind the Method of Loci is that having a well known sequence of locations with which to associate new information facilitates remembering because the sequence of locations provides a means of organizing new information. Successful use of the Method of Loci depends upon two things. First, the sequence of locations must be memorized. As well, the individual must become skilled at creating images associating the to be remembered item and the appropriate location. Previous research suggests that spatial visualization is an important component in determining successful use of the Method of Loci (see Lindenberger, Kliegl, and Baltes, 1992 for a discussion). Thus, acquiring skill with this mnemonic may require the development during knowledge compilation of procedures for generating and remembering visual images.

In the study of interest (Kliegl et al. 1989, Experiment 2), the sequence of locations consisted of 40 well known landmarks in West Berlin whose order the subjects memorized at the beginning of the experiment. After the Method of Loci technique was explained, subjects practiced using the mnemonic for 20 sessions. The information to be remembered consisted of lists of 30 words, and each word was presented for 20, 15, 10, 5, 3 or 1 second. At the beginning of practice all subjects began with a 20 second presentation rate. The next fastest presentation rate was introduced when the subject was able to recall 50% of two consecutive word lists at a given presentation rate. At the start of practice the young adults recalled more words that the elderly at all presentation rates except 1 second. After practice performance improved substantially for both young and elderly, however there was still a significant age-related difference indicating that young adults recalled more at all presentation rates except 1 second. If, as suggested previously, skill acquisition for this task is associated with the development of procedures for generating visual images associating the to-be-remembered item and the location, these results may indicate an age-related deficit in knowledge compilation. Further, the difference in performance between young and old was larger after practice. That is, practice had increased, not reduced, age-related differences. These results are in marked contrast to previously discussed studies which indicated that practice reduced, and in some instances eliminated, age-related differences in performance.

Benefits of Previously Acquired Expertise

In the study just discussed, the participants did not possess any skills or abilities that may have facilitated skill acquisition. An interesting question, and one that has generally been ignored, is whether or not existing expertise and abilities facilitate the acquisition of new, but related skills. This question was addressed in a study that also employed the Method of Loci, but whose subjects included young and elderly adults who were graphics designers in addition to typical young and elderly adults (Lindenberger et al. 1992). The rationale for selecting graphic designers is that their profession requires them to generate and work with visual images. Given that the Method of Loci also requires the generation of images, it might be expected that elderly graphics designers would achieve higher levels of performance using this technique relative to elderly adults lacking their specialized experience. More specifically, the elderly graphics designers could be advantaged in learning the Method of Loci for two reasons. Their professional experience could mean that they have acquired a body of factual and procedural knowledge for generating visual images. Another possibility is that they possess superior abilities on spatial visualization and other task relevant abilities.

Prior to training in the use of the Method of Loci, subjects completed several psychometric tests assessing visual creativity and spatial visualization. The rationale for including these measures was to determine if memory performance, and differences in performance between graphics designers and normal adults, would be predicted by these measures. Subjects then completed 7 training sessions that were approximately one hour in length. An examination of age-related differences at the end of training indicated that the performance of the younger normal and younger graphics designers did not differ, and that young adults performed better than both groups of older adults. However, although the older graphics designers performed more poorly than the young adults, their performance was superior to that of the normal older adults. The implication is that the older graphics designers' professional experience generating visual images enhanced their ability to acquire skill using the Method of Loci. An examination of memory performance as a function of performance on the criterion-related tests supported this conclusion. First, memory performance was significantly correlated with measures of spatial visualization even after the effects of age and expertise were controlled for. Secondly, the difference in performance between the normal elderly and the elderly graphics designers was not significant when group differences in visual creativity and spatial visualization were controlled for. The implication is that the recall advantage of the older graphics designers relative to normal older adults was attributable to the image generation required by the Method of Loci. Thus, it is possible that the elderly graphics designers' professional experience enabled them to outperform the normal elderly.

Summary and Conclusions

At the beginning of this section it was suggested that two issues are particularly important when examining age-related differences in skill acquisition, the impact of practice upon the magnitude of age-related differences, and whether or not different stages of skill acquisition pose particular problems for older adults. With regard to the effect of practice, given the dramatic impact of practice upon performance, it might be expected that after extensive practice, age-related differences in performance would be minimized, if not eliminated. However, it has also been suggested that once performance has been optimized via extended practice, the impact of age-related declines upon performance would result in even larger age-related differences. In general, in the majority of studies, age-related differences were reduced, but not eliminated, by the end of practice. However, in the minority of studies age-related differences were either eliminated, or larger at the end of practice. Thus, the results are rather mixed, and it is possible that practice interacts with other factors, such as task difficulty, to influence the magnitude of age-related differences after practice. However, a limitation of all of these studies is that the amount of practice they provided is quite small compared to the amount of practice obtained in everyday activities. Many everyday activities have been practiced for thousands of hours. It may be the case that such extreme levels of practice are required to eliminate age-related differences (see Bosman, 1993, 1994, for a discussion). It is also unclear from the studies discussed if any particular aspect of skill acquisition is especially difficult for older adults. Some studies suggested that the knowledge compilation stage may be problematic for older adults, while others did not. Again, it may be the case that factors like task difficulty influence which stages of skill acquisition are particularly problematic for older adults.

Future Directions

As the previous discussion suggests, only a small body of research focusing upon a limited number of domains has investigated age-related differences in acquired skills, and skill acquisition. Clearly there are many unresolved issues regarding the nature of age-related differences in skilled performance. For example, the research investigating age-related differences in acquired skills suggests that older adults are to maintain their performance level by employing compensatory mechanisms. However, it seems unlikely that older adults always employ compensatory mechanisms, and other mechanisms for maintaining performance level should be investigated. The maintenance perspective suggests that sustained practice over the course of adulthood will prevent age-related declines from occurring. Currently, there are no systematic investigations of the effect of practice upon the preservation of skilled performance with age. Likewise the accommodation perspective suggests that older adults may possess metacognitive knowledge that enables them to avoid conditions under which they can no longer perform well. Research examining the extent to which older adults do and do not accommodate to age-related declines in this manner is required to address this issue.

Research investigating age-related differences in skill acquisition so far has not produced a consistent pattern of results that holds across all tasks. For example, in most of the research the magnitude of age-related differences is reduced, but not eliminated by practice, In some instances age-related differences are eliminated after extensive practice, although this is rare, and in other instances age-related differences are actually greater at the end of practice. It may be the case that one of the factors influencing the effect of practice is the age sensitivity of the underlying processes. Specifically, if a process is not particularly age sensitive, practice may be successful in reducing or eliminating age-related differences. However, if a process is particularly age sensitive practice may lead to an increase in the level of performance, but not eliminate the negative impact of aging upon performance. Similarly, research investigating which stage of skill acquisition is most difficult for older adults seems to suggest that knowledge compilation is particularly difficult for older adults, though again, this result is not always found. Again, whether or not a particular stage of skill acquisition is problematic for older adults may depend upon the age sensitivity of the underlying processes. Future research should attempt to address this issue.

A somewhat different picture of age-related differences in skilled performance emerges from research examining age-related differences in acquired skills on the one hand, and age-related differences in skill acquisition on the other. The research examining acquired skills suggested that older adults are able to compensate for age-related declines and thus achieve a level of performance comparable to that of younger adults. In contrast, the research examining skill acquisition suggested that older adults are typically not able to achieve the same level of performance as younger adults when learning new skills. A challenge for future research will be to account for these differing pattern of results. Currently, there are several factors that could account for the differences between the two types of studies. First, it should be recalled that studies examining age-related differences in acquired skills have generally not attempted to identify if overall performance declines with age. Thus, the results of these studies should not be interpreted as indicating that the level of skilled performance does not decline with age. It may be the case that skilled performance does decline with age, but that the development of compensatory mechanisms minimizes the magnitude of decline. In fact, chess skill, as indicated by competitive rating does decline with age, although the decline is modest (Charness & Bosman, 1990).

Another difference between studies examining acquired skill and studies examining skill acquisition is the amount of practice participants have had with the task of interest. Subjects in studies of acquired skills have typically had hundreds, if not thousands, of hours of practice at the relevant task prior to participation in an experimental study. Although by the standards of the experimental literature, the participants in studies of skill acquisition have received considerable practice, the amount is quite minimal when compared to the amount of practice everyday tasks and activities receive. Practice has a profound impact upon performance, and it may be the case that thousands of hours of practice are required to reduce or eliminate age-related differences. A final difference between studies examining acquired skill and studies of skill acquisition is the age at which the task is first learned. Participants in studies examining acquired skills typically learned the task of interest at a comparatively young age, while this is obviously not true of the older participants in skill acquisition studies. It may be the case that the impact of aging upon skilled performance varies as a function of age of acquisition. Specifically, skills acquired while young may be more robust with regard to the effects of aging, while skills acquired when older are not. Certainly, the encapsulation perspective would suggest that this may be the case.

Implications for Everyday Functioning

The results of the research reviewed in this chapter have very positive implications for the everyday functioning of older adults. The research examining age differences in acquired skills indicated that older adults are able to perform at levels comparable to younger adults, possibly by compensating for age-related declines. Further it is worth noting that the skills examined in this context, chess, bridge, and transcription typing, are representative of skills performed in everyday life. Certainly typing is a skill performed by numerous individuals, and is a skill that is likely to become an increasingly important as the usage of computers continues to spread. While chess and bridge represent recreational activities for the majority of their practitioners, the complexity of both games is comparable to that of many professional activities. The implication of this research is that older adults can probably be expected to continue to function very well in domains in which they have acquired expertise.

The research examining age-related differences in skill acquisition also indicated that older adults are capable of acquiring new skills, and that practice results in dramatic increases in performance. While it might be argued that the tasks examined in this research, mental squaring, memory and visual search, and memory for word lists, have little relevance for most daily activities, we suggest that these tasks correspond to the components of many common activities. We have probably all had the experience of standing in the grocery store attempting to calculate which brand of dish washing liquid represents the best buy. Although these calculations are much simpler than mentally squaring two digits, they still take some time to perform. Similarly, everyday most of us are confronted by a seemingly endless list of things that we have to do, and we frequently have the experience of remembering that we forgot to do something. Visual search is also a part of many tasks. For example, during driving we have to continuously scan the environment to detect relevant stimuli such as a traffic light turning red, or a pedestrian stepping out onto the street. Memory scanning is also part of many activities, such as deciding which of several different medications is the one we are currently supposed to take. Thus, the results of research examining age-related differences in skill acquisition have more relevance to everyday functioning than might be immediately apparent. Clearly this research suggests that older adults will be able to successfully acquire new skills.