The type of knowledge that subserves this proficiency

The first systematic investigation of skill learning involved a skill that is now obsolete. This was the study of the acquisition of the American Morse code by employees of railway and telegraph companies (Bryan and Harter 1897, 1899). On the basis of these studies, Bryan and Harter proposed that the formation of skill is a multi*phase process, which involves the mastery of specific elementary "habits that become associated in a hierarchy. The expert is then able to perform the hierarchy of habits automatically, at a speed that in certain individuals approaches the physical limit of the task (Figure 60). The command of automaticity should therefore be given full respect in education: 'It is quite useless to raise the question whether or not children should acquire specific automatic habits. There is no escape from such habits ... The wolf does not escape. Neither Shakespeare nor Caliban escape... Automatism is not genius, but it is the hands and feet of genius' (Bryan and Harter 1899).

1. The spectrum of skills. Sending Morse messages involves a motor, or perceptual-motor skill. Decoding such messages involves primarily a perceptual skill. Other types of skill are cognitive, for example, problem solving. 'Skill' is also used, mainly in colloquial language, to refer to proficiency in other behavioural domains, such as emotional or social, without necessarily distinguishing between an innate ("a priori) predisposition and an acquired ability. Emotional and social skill could be regarded as subtypes of cognitive skill. Cognitive skill is occasionally dubbed 'intellectual skill', but this applies only to humans and may connote

105 106 107 Number produced

Fig. 60 The acquisition of skill. (a) The average speed of production of cigars by a group of girls operating cigar-making machines. The curve for the first 2years fits a power function (see text). After producing about 3 million cigars, the performance of the workers approached the machine cycle time. (Adapted from Crossman 1959.) (b) The average reading time of a group of undergraduates for mirror-reading English text.Again, the curves fit a power function. (Adapted from Kolers 1968.)

Days

Fig. 60 The acquisition of skill. (a) The average speed of production of cigars by a group of girls operating cigar-making machines. The curve for the first 2years fits a power function (see text). After producing about 3 million cigars, the performance of the workers approached the machine cycle time. (Adapted from Crossman 1959.) (b) The average reading time of a group of undergraduates for mirror-reading English text.Again, the curves fit a power function. (Adapted from Kolers 1968.)

pretentious capabilities. The term 'mental skill' should be avoided as all skills are mental. Skills could be performed in the absence of "conscious awareness, hence, are classified as non"declarative memory (Squire and Zola 1996). Their acquisition, however, may either involve conscious awareness, i.e. be declarative or explicit, or be independent of conscious awareness, i.e. nondeclarative or implicit.

2. The acquisition of skills. The analysis of all types of skill shows that their acquisition involves multiple phases.1 The nature of these phases is the subject matter of theories of skill acquisition (e.g. Snoddy 1926; Crossman 1959; Fitts 1964; Anderson 1982; Logan

1988). These theories differ in the nature and number of phases, but basically share the view that the first phase involves selection of the proper "algorithm, and the later phase(s) involve(s) perfection and automatization of the suitable solution. Hence the first phase is called either 'adaptation' (Snoddy 1926), 'algorithmic' (Logan 1988), 'understanding' or formation of 'cognitive set' (Fitts 1964),2 or 'declarative' (Anderson 1982); the later phase is 'facilitation' (Snoddy 1926), or 'automatization' (Logan 1988), or 'procedural' (Anderson 1982). The existence of phases in skill acquisition is supported by behavioural data, and more recently, by "functional neuroimaging data that show the recruitment of different brain circuits with practice (see below).

A common observation is that the improvement of performance with practice follows a power law: log C = log B + nlog X, where C = a measure of performance, X = number of trials or time on task, B, n = constants (Snoddy 1926; DeJong 1957; Crossman 1959; Newell and Rosenbloom 1981) (Figure 60). This is called the 'power law of performance', or 'of practice', or 'of learning'. Power functions are commonly associated in the literature with perceptual-motor skill, and sometimes are even taken as evidence that the type of learning is indeed such skill. In fact they are common to all learning that involves repetitive training (Newell and Rosenbloom 1981), including learning to "forget (Rubin and Wenzel 1996).

3. The *persistence of skills. Skills could be retained over years in the absence of further practice (Fleishman and Parker 1962), and sometimes even in the absence of recollection of having previously performed the task (Cohen 1984). What loss occurs is quickly regained by re-practicing (Fleishman and Parker 1962; see 'saving' in "experimental extinction).

4. The relationship between skill and other types of procedural memory. Automaticity is an attribute shared by other forms of procedural memory (Squire and Zola 1996; "taxonomy). The question arises what is the relationship between skill and these other forms of memory. Of special interest is the relevance of skill to repetition "priming. Whereas skill learning refers to a "generalized improvement on a task, priming refers to the improvement on a given item within the task.3 Is the latter a special case of the former? The views differ; priming and skill learning are considered manifestation of the same type of incremental learning process or mechanism by some authors (Logan 1990; Poldrack et al. 1999), but not by others (Schwartz and Hashtroudi 1991; Kirsner and Speelman 1996). A data-based argument brought up by the opponents of the single-process view is that priming is independent of accumulative practice. A conceptual argument in favour of a unified-process approach is that even skill learning is limited to the specific task-related procedures, and therefore is not qualitatively so different from priming. It could be proposed, so goes this argument, that skill and repetition priming represent each a change along a continuum of generalization; the degree of generalization reflects the "level of the processing stream in which the modification occurs. Depending on the item variability in training and the amount of practice, the use-dependent alteration may result in learned information that does not "transfer to other items (repetition priming), or, alternatively, generalizes over a class of items (skill; Ofen-Noy et al. 2002). This view echoes the proposal that different types of learning, including declarative ones, involve a procedural component, yet differ, among others, in the resulting specificity of transfer (Kolers and Roediger 1984).

5. Which brain areas encode and retain skills? The approach to this question relies on two main types of "methodologies. One is the analysis of the performance of patients with 'global' "amnesia, Parkinson's disease, and Huntington's disease; the other, functional neuroimaging of brain activity during the learning and performance of skill in healthy human "subjects. The analysis of performance of amnesics, Parkinsonian and Huntingtonian patients provides information about brain areas that are obligatory for skill ("criterion). This type of studies is complemented by the investigation of the effect of circumscribed brain lesions on tasks that are considered to "model human skill in laboratory animals, e.g. the "monkey. The neuroimaging studies provide information about brain areas whose activity is correlated with skill learning and performance. This type of studies is complemented by cellular physiology methods in laboratory animals (e.g. Recanzone et al. 1993).

Despite their dense declarative amnesia, human patients with extensive damage to the mediotemporal lobe perform remarkably well on perceptual-motor, perceptual, and cognitive skill tasks (Milner et al. 1968; Brooks and Baddeley 1976; Cohen and Squire 1980; Cohen 1984). For example, with experience, they improve normally on motor tracking in a tactual maze (perceptual-motor skill), on mirror-reading (perceptual skill, Figure 60), and on solving the Tower of Hanoi puzzle (cognitive skill, Figure 61). All this, without recollection of having previously performed the task. These findings show that skill is not subserved by the same brain circuits that subserve declarative memory. In contrast to the success of amnesics on skill tasks, Huntington's and Parkinson's patients are impaired on

4K

/■ 1 1 Start

Fig. 61 The performance of amnesics on a cognitive skill. The Tower of Hanoi puzzle consists of five wooden blocks and three pegs (a). At the outset, all five blocks are arranged on the leftmost peg in size order, the smallest block on top.The subject is then asked to move the blocks from the leftmost peg to the rightmost peg, while moving only one block at a time, and without ever placing a larger block on top of a smaller one. The optimal solution, which involves shuttling the blocks back and forth on all three pegs, requires 31 moves. (Compare with the Tower of London, ^planning.) In this experiment, training consisted of solving the puzzle four times per session on each of four consecutive days. The performance of amnesics patients on this task is indistinguishable from that of normal controls (b). A follow-up study showed that patient H.M. ('amnesia, 'classic) retained the skill after 1 year. (Adapted from Cohen 1984.)

Fig. 61 The performance of amnesics on a cognitive skill. The Tower of Hanoi puzzle consists of five wooden blocks and three pegs (a). At the outset, all five blocks are arranged on the leftmost peg in size order, the smallest block on top.The subject is then asked to move the blocks from the leftmost peg to the rightmost peg, while moving only one block at a time, and without ever placing a larger block on top of a smaller one. The optimal solution, which involves shuttling the blocks back and forth on all three pegs, requires 31 moves. (Compare with the Tower of London, ^planning.) In this experiment, training consisted of solving the puzzle four times per session on each of four consecutive days. The performance of amnesics patients on this task is indistinguishable from that of normal controls (b). A follow-up study showed that patient H.M. ('amnesia, 'classic) retained the skill after 1 year. (Adapted from Cohen 1984.)

learning of perceptual-motor skill (Heindel et al. 1988), perceptual skill (Martone et al. 1984), and cognitive skill (Saint-Cyr et al. 1988). Huntington's disease and Parkinson's disease both involve striatal pathology, and therefore, the striatum has been proposed to subserve skill learning, similarly to its role in habit formation.

Functional neuroimaging can be used to identify areas that are activated during the acquisition and performance of skill in the normal brain (e.g. Seitz et al. 1990; Grafton et al. 1995; Karni et al. 1998; Petersen et al. 1998; Hund-Georgiadis and von Carmon 1999; Nadler et al. 2000; Poldrack and Gabrieli 2001).4 A few tentative generalizations could be extracted from these studies: (a) skill-related changes in brain activity are distributed over multiple brain areas, some of which are task-specific and some task-independent; (b) the task-specific areas include modality specific cortical areas, whereas the corticostriatal system, and possibly the "cerebellum, are involved in skills that span sensory modalities and tasks; and (c) often it is found that the involvement of brain areas in skill learning and performance changes over time in the course of practice. The picture that emerges, although yet preliminary, is that the acquisition of skill involves changes in two types of "internal representations, those of the specific behavioural acts of the particular skill, and those of the 'syntactic' hierarchical routines by which the particular acts are perfected, bound, and executed. Further, above a certain degree of practice, the relevant internal representations may shift to brain circuits other than those employed in the early phases of training. This can be shown when comparing the activation of brain areas over training in an individual, or the activation in the brain of novices vs. experts (e.g. Hund-Georgiadis and von Carmon 1999). Exceptional skill benefits from particular behavioural routines (Ericsson and Lehmann 1996), and the talented brain may process the talent-related material in a unique way (Schlaug et al. 1995; Pesenti et al. 2001). An interesting question is whether this results merely from the perfection of normal brain resources, or from an innate predisposition in the structure of the brain, or, most likely, from both.

Selected associations: Birdsong, Habit, Instrumental conditioning, Priming, Transfer

1'Phase' here means stages in the development of skill that are superimposed on the basic universal phases of memory, such as short-term memory, 'consolidation, or 'retrieval.

2On 'set' see 'learning set. The terms 'understanding', 'cognitive set', and 'declarative' fit the discussion of human skill but not necessarily that of skill in other species.

3That is, whereas skill refers to types of stimuli, priming refers to tokens; on types and tokens see 'system.

4It is noteworthy that a limited degree of gross localization of function can be obtained in certain skill paradigms by the use of behavioural methods in the absence of neuroimaging. Hence if a visual perceptual skill improves only in one eye without affecting the other, it could be inferred that the process occurs at low-level of the visual processing stream in the brain, before the site of binocular integration (Sagi and Tanne 1994).

0 0

Post a comment