Wednesday, February 23, 2011

Theories of Motor Skill Acquisition

The most notable thing that happens when people practice is that they demonstrate increased proficiency in performance and skill.  A skill can be conceptualized as a task (e.g. throwing a baseball, kicking a ball) or it can be viewed as a level of performance proficiency that distinguishes a higher-skilled performer from a lower-skilled performer (Schmidt, 2004).  While several definitions of skill have been proposed, Guthrie’s (1952) definition captures the critical elements of skill that are espoused by the majority of contemporary researchers and theorists.  He proposed that “skill consists in the ability to bring out about some end result with maximum certainty and minimum outlay of energy, or of time and energy.”  There are different types of skill; for example, motor skills, perceptual skills, and cognitive skills.  Motor skills are those in which both the movement and the outcome of the movement are emphasized (Newell, 1991).  There are three essential features of skilled movement: maximum certainty of goal achievement, minimum energy expenditure, and minimum movement time. 
Motor skill acquisition is a process in which a performer learns to control and integrate posture, locomotion, and muscle activations that allow the individual to engage in a variety of motor behaviors that are constrained by a range of task requirements (e.g. athletic context) (Newell, 1991).  As a learner acquires a skill, changes may be observed that reflect strategies that an individual uses to achieve specific movement outcomes.  A learner may show a change in the spatial orientation of his or her body and body limbs as well as exhibit a change in the timing and sequencing of movements.  Motor-skill acquisition follows a pattern in which learning accumulates with practice.  Changes in performance that accompany practice are usually much greater and more rapid at first and systematically become smaller as practice continues.
Bryan and Harter (1899) were among some of the first researchers to study skill acquisition.  They observed performance scores of telegraphers who received telegraphic messages and then translated the code into their native language.  Bryan and Harter hypothesized that improvements in the telegraphers’ performance, which was measured by the amount of words translated in minute, could not be due a sudden increase in knowledge of native language but rather an acquisition of higher language habits (Bryan & Harter,1899).  Later, Snoddy (1926) provided a classic example of motor skill acquisition.  He conducted a study that required subjects to perform a mirror tracing task that required them to learn to control hand movement speed and accuracy.  Snoddy asked his subjects to trace a circuit of a 12-edge, star shaped path one fourth of an inch wide.  The direct vision of the tracing instrument and the hands were obstructed by a screen and only an indirect mirror image of the tracing device and hands was available to the participants.  The instruction to the participants was to move around the path as fast as possible and avoid making contact with the side of the tracing.  Each trial consisted of completing one circuit, and performance was measured as the ratio of 1000 over the sum of tracing time (T) and number of contact made (E) within each trial [1000/(T+E)].  Analysis of participants’ scores revealed that gains in performance follow a non-linear pattern in which improvement was rapid at first, but declined as training progresses and the number of trials increases.  Snoddy (1926) hypothesized that the number of repetitions was the primary parameter that affected the course of learning.  He explained motor-skill learning as a two-stage process which was comprised of an adaptation stage, in which the learner acquires the neuromuscular pattern required to perform the movement, and a facilitation stage, in which the efficiency of the movement pattern is improved. 
Later, Henry and Rogers (1960) explained motor learning in terms of neuromotor memory.  They hypothesized that humans possess a vast amount of unconscious motor memory which is stored in the form of innate motor coordinations that are essential to initiation of controlled motor actions.  They modeled motor control processes in terms of a memory drum, a data storage device developed in the 1930’s that was an early form of computer memory.  For machines, the memory drum formed the working memory of the machine which allowed for data and programs to be loaded off the machine using punch cards.  The memory storage drum in the human mind as proposed by Henry and Rogers is analogous to a memory drum in a machine in that programs are preprogrammed and stored for retrieval.  Henry and Rogers hypothesized that the neural pattern for specific and well-coordinated motor acts are controlled by a stored program that when retrieved directs all of the neuromotor details of the performance (Henry & Rogers, 1960).  In the absence of a stored program, a novel task will be carried out under conscious control and the execution of the movement will be poorly coordinated and awkward.  Thus, the memory drum theory predicted that whenever a specific movement pattern is required; the stimulus causes the memory drum to ‘play back’ the particular learned neuromotor program.  The theory was consistent with the view that learning motor skills is specific, rather than general, and that there is little or no carry-over from one skill to another unless the skills are nearly identical.  Practice was predicted to improve performance of a specific skill by the strengthening of the neuromotor program; further, the retrieval of the neuromotor program was predicted to occur more automatically and with less conscious awareness. 
Most motor-skill acquisition theories have embraced a stage conceptualization of learning.  Fitts (1964) and Fitts and Posner (1967) proposed a three stage process of motor learning that incorporated a cognitive stage, an associative stage, and an autonomous stage.  During the cognitive stage of skill acquisition, the biggest challenge of the learner is to understand what is to be performed, while the biggest challenge for teachers is conveying to the learner what is to be done.  During this stage, performance gains are usually quite large; however, these performance gains become smaller and smaller as a function of the number of trials.
The associative stage begins once the learner selects a movement strategy and actually performs the task, and based on feedback begins to modify how the movement is performed.  This stage is of particular interest to researchers because feedback plays a crucial role in altering the movement pattern.  In the associative stage, attention is allotted to improving the efficiency and timing of the movement.  The rate of gain of learning in the associative stage is influenced by the nature of the relationship between environmental stimuli and developing motor responses.  Stimulus-response compatibility refers to the extent of the association or “naturalness” between a stimulus and the response (Schmidt & Lee, 2005).  Tasks are easier or more difficult to learn as a result of the pairing between specific stimuli and their respective responses (Kornblum et al., 1990).
The autonomous phase appears after extensive training and it is characterized by motor movements being performed automatically and requiring less attentional capacity to complete the skill.  Schneider and Shiffrin (1977) conducted extensive research on automaticity and the goal of their research was to understand precisely the conditions under which attention limitations occur.  Schneider and Shiffrin used a visual search task that involved presenting stimuli in a rapid succession of displays and the subject’s goal was to judge whether a target stimulus had been presented.  Stimulus display duration, memory set size, and consistency of target-distractor mappings were manipulated.  Two conditions were used to evaluate attention and automaticity: consistent and varied mapping.  On consistently mapped trials, the targets and distractors were distinguished by category (e.g. letters or numbers).  In the varied mapping trials, targets and distractors were from the same category.  Results showed that performance in the variable mapping condition was dependent on load and frame size, and performance in the consistent mapping condition was largely independent of load and frame size.  Schneider and Shiffrin proposed two processes to account for their results: controlled search and automatic detection.  Controlled search is a serial process in which a matching decision occurs after comparison of each item in the display to the memory set items; in contrast, that automatic detection operates in parallel and independent of attention.  Automatic processes do not require attention and they do not use up short-term memory capacity; further, once initiated automatic processes are not easily modifiable (Schneider & Shiffrin, 1977).  The findings have implications for motor skill acquisition: they demonstrate that cognitive load affects rate of skill acquisition, and that once learned, automatic movements are difficult to modify.
Adams (1971) was one of the first researchers to emphasize the role that cognition plays in skill acquisition.  Early theories of motor skill acquisition were influenced by the views of behavioral psychologists who conceptualized learning in terms of the associations between stimuli and responses.  Adams hypothesized that human motor-skill learning was not simply a behavior driven by neuromotor programs in response to a stimulus, but rather that motor behavior included a variety of cognitive processes as well as the development of strategies that can be used to complete a given motor task.  A central component of Adams’ (1971) theory of motor control was the manner in which feedback and error detection influences learning.  Adams (1971) believed that learners possess a reference of correctness that specifies a desired outcome of the movement and a feedback mechanism that detects error between the learner’s desired movement and the actual movement produced.  Considerable research findings suggest that Adams’ views hold true for movements that are relatively slow.  Relatively slow movements provide the learner an opportunity to evaluate his or her performance as it is ongoing and to detect the error between the desired movement and the actual movement by way of a feedback mechanism.  This type of processing has been termed closed-loop processing (Schmidt & Lee, 2005).  Adams posited that movements produce internal feedback, which creates a perceptual trace of the movement that is laid down in the central nervous system.  The more accurate the movement, the more useful the perceptual trace will be on subsequent trials.  The feedback mechanism compares the feedback produced by the movement to the accumulated perceptual trace and detects any errors between the actual and expected feedback.
Adams’ theory placed less emphasis on how ballistic, rapid, open-loop movements are learned and controlled, however.  For open-loop movements, a motor plan needs to be structured in advance and executed without regard to the effects that they may have on the environment, which does not allow for feedback during the movement.  Schmidt (1975) developed an important theory of motor learning that addressed directly how discrete motor movements are acquired and controlled.  He proposed a schema theory that hypothesized that there are two states of memory: recall memory and recognition memory.  Recall memory is responsible for movement production and recognition memory is responsible for evaluation of movement.  Recall memory does not play a role in slow positioning movements.  For slow movements, the recall state simply controls movements in small bursts with the movement terminating when the movement-produced feedback matches the reference of correctness.  Schmidt proposed the idea of a generalized motor program; a structured plan of movement that is composed of invariant features and variant features.  Invariant features are comprised of the components that remain the same in regards to the general movement being executed (overhand throw) and variant features are the parameters of the program that can be altered such as time and time and force (soft overhand throw versus hard overhand throw).  Individuals do not learn specific movements; rather they construct a generalized motor program by exploring the rules of action (schema) and learning ways in which movements relate to outcomes.
Schmidt’s theory explains how motor skills are learned.  A general motor program depends on four types of information that are stored in short-term memory: 1) information about the initial conditions before the movement (variances in limb position or object size/weight), 2) parameters assigned to the general motor program (force, time), 3) augmented feedback about the movement (KR), and 4) sensory feedback (how the movement felt, looked, sounded) (Schmidt & Lee, 2005).  These sources of information are interrelated and represent recall and recognition schemas.  Learning occurs through the development of the recall schema as the number of trials of given task accumulate.  After each adjustment of parameters, various sources of information are discarded from working memory; thus, all that remains is the movement rule, which represents the recall schema.  The recognition schema forms in much the same way as the recall schema.  The recognition schema is developed on the basis of information concerning the relationship between the initial conditions, the environmental outcomes, and the sensory consequences.  Before a movement takes place, an individual can use a learned recognition schema to predict the sensory consequences that will occur if the correct movement outcome takes place.  These expected sensory consequences are the basis for which to evaluate movement.  Thus, augmented feedback plays a central role in schema development.
While there are differences among contemporary theories of motor-skill acquisition (e.g., Anderson, 1982), the notion that the learner progresses through a series of stages remains central to explaining the phenomenon.
Several contemporary theories of motor learning have identified cognitive processes as being important to motor skill acquisition.  Cognitive processes have been hypothesized to be crucial during the initial stages of skill learning.  During the cognitive stage of motor-skill acquisition a large amount of mental involvement is required of the learner.  The cognitive phase is characterized by conditions in which the learner must encode and integrate task instructions, become familiar with task goals, and formalize strategies for task accomplishment.  Ackerman (1988, 1992) provided evidence that during this phase learners’ performance is slow and error prone due largely to the need to formulate strategies and to test strategy effectiveness.  During the cognitive stage considerable attention is directed towards understanding movement goals and the contextual factors that constrain movement.  Performance during the cognitive stage is associated highly with general intelligence and verbal, spatial, and numerical abilities.  During the associative phase of skill acquisition, the role of general intelligence abilities decline and perceptual-speed abilities become more highly associated with performance.  In the autonomous stage, the influence of both general intelligence abilities and perceptual-speed abilities decline and performance becomes most associated with psychomotor abilities.
Ackerman and Cianciolo (2000) assessed procedural skill development via the Kanfer – Ackerman ATC task, which is a complex task that simulates air traffic control decisions and landing of aircraft planes on the basis of various procedural rules.  Results obtained from the study demonstrated the predicted change in the contribution of general intellectual ability as performance improved and confirmed the importance of cognitive abilities early in skill acquisition.
There are many factors that influence the performance and learning of a motor skill.  Verbal information in the form of instructions is one of the most important factors and also one of the first factors to be studied systematically.  An early study conducted by Solley (1952) evaluated the effects of instruction on learning a lunge and stab movement under conditions that emphasized either movement speed, movement accuracy, or an equal emphasis on speed and accuracy.  The results were quite dramatic.  The group instructed to emphasize movement speed had the highest movement speeds, the group instructed to emphasize movement accuracy yielded the highest accuracy scores, and the group instructed on both speed and accuracy performed at intermediate levels on both speed and accuracy.  These results indicate that specific information presented to learners can alter the way in which a movement is carried out as well as the outcome of the movement.  Modeling a movement is another way to convey information to a learner.  Modeling, or observational learning, is learning that occurs as function of viewing, retaining, and replicating a novel behavior executed by other individuals.  Several factors influence the degree to which modeling influences skill acquisition: the properties of the model (e.g., expert versus non-expert), the nature of the task (complexity, number of degrees of freedom), observer determinants (comprehension of the demonstration), and feedback (Ferrari, 1996).  Feedback in particular plays a critical role in determining motor learning and performance.
This research indicates that learning cannot occur in the absence of information and learning may be hindered if too much information is presented to the learner.  One critical aspect in the field of motor skill acquisition that has not been assessed is how an individual’s initial skill level affects the type and amount of feedback necessary for efficient learning to take place.  Researchers in the field, Guadagnoli and Lee (2004), have formulated a framework in which many hypotheses in regards to how skill level and task difficulty interact can be drawn.  The challenge point framework touches on the idea that increases in task difficulty are accompanied by increases in potential information (Guadagnoli & Lee, 2004).  However the there is a limit to the amount of information that is interpretable to the learner, which is assumed to be governed by the individual’s skill level.  Furthermore depending on the skill level of the individual, an increase in task difficulty would be associated with decreased performance expectations, but there would also be an increase in the amount of available information (Guadagnoli & Lee, 2004).  Thus the challenge point framework represents the degree of task difficulty an individual of a certain skill level would need to optimize learning (Guadagnoli & Lee, 2004). 
The challenge point framework and the research conducted on the most beneficial types of feedback and feedback schedules has led to hypotheses that deal not only with feedback protocols, but on how those protocols affect skills of varying difficulties.  Guadagnoli and Lee (2004) hypothesize that for tasks of high difficulty, more frequent presentation of feedback will yield the largest learning effect and for tasks of low difficulty, less frequent presentation of feedback will yield the largest learning effect.  This hypothesis has not yet been tested however and further research is needed.
A proposed experiment to test this hypothesis should include assigning participants into two groups (high skill and low skill) based on their performance on a median level task.  Upon assignment to skill level groups, experiment 1 will proceed by  asking each group (high and low) to perform a task of high difficulty under two conditions (frequent feedback and less-frequent feedback) and the performance under each condition will be collected.  For experiment 2, participants from each group will be asked to perform a task of low difficulty under the same conditions (frequent feedback and less-frequent feedback) and the performance under each condition will be collected and compared against the results obtained in experiment 1.  The conditions that yield the greatest performance will give researchers a better understanding of the dynamic interaction between skill level, task complexity, and the most appropriate feedback for each situation


Adams, J.A. (1971). A closed-loop theory of motor learning. Journal of Motor Behavior, 3, 111-150.

Ackerman, P.L. (1988). Determinants of individual differences during skill acquisition: Cognitive abilities and information processing. Journal of Experimental Psychology: General, 117, 288-318.

Ackerman P.L. (1992). Predicting individual differences in complex skill acquisition: Dynamics of ability determinants. Journal of Applied Psychology, 77, 598-614.

Ackerman, P.L., & Cianciolo, A.T. (2000) Cognitive, perceptual-speed, and psychomotor determinants of individual differences during skill acquisition. Journal of Experimental Psychology: Applied, 6, 259-290.

Anderson, J.R. (1982). Acquisition of cognitive skill. Psychological Review, 89, 369-406.

Bryan, W.L., & Harter, N. (1899) Studies on the telegraphic language: The acquisition of a hierarchy of habits.  Psychological Review, 6, 345-375.

Ferrari, M. (1996). Observing the observer: Self-regulation in the observational learning on motor skills. Developmental Review, 16, 203-240.

Fitts, P.M. (1964).  Perceptual-motor skills learning.  In A.W. Melton (ed.), Categories of human learning (pp. 243-285). New York: Academic Press.

Fitts, P. M., & Posner, M.I. (1967). Human Performance. Belmont, CA: Brooks Cole.

Guadagnoli, M.A. & Lee, T.D. (2004).  Challenge point: a framework for conceptualizing the effects of various practice conditions in motor learning.  Journal of Motor Behavior, 36, 212-224.

Guthrie, E.R. (1952). The psychology of learning. New York: Harper & Row.

Henry, F.M. & Rogers, D.E. (1960). Increased response latency for complicated movements and a “memory drum” theory of nueromotor reaction. Research Quarterly, 31, 448-458.

Kornblum, S., Hasbroucq, T., & Osman, A. (1990). Dimensional overlap: Cognitive basis for stimulus-response compatibility – a model and taxonomy. Psychological Review, 97, 253-270.

Newell, K.M. (1991).  Motor skill acquisition.  Annual Review of Psychology, 42, 213-237.

Schmidt, R.A. (1975).  A schema theory of discrete motor skill learning. Psychological Review, 82, 225-260.

Schmidt, R.A. & Lee T.D. (2005).  Motor control and learning: A behavioral emphasis.  Champaign, IL: Human Kinetics.

Schmidt, R.A. & Wrisberg, C.A. (2004). Motor learning and performance.  Champaign, IL: Human Kinetics.

Schneider, W., & Shiffrin, R.M. (1977). Controlled and automatic human information processing: I. Detection, search, and attention.  Psychological Review, 84, 1-66.

Snoddy, G.S. (1926).  Learning and Stability: A psychophysical analysis of a case of motor learning with clinical applications.  Journal of Applied Psychology, 10, 1-36.

Solley, W.H. (1952). The effects of verbal instruction of speed and accuracy upon the learning of a motor skill. Research Quarterly, 23, 231-240.

1 comment:

Steve B said...

You might enjoy the book "The Talent Code" which gets into deep practice and what needs to occur for someone to "master" a skill. It talks about the nervous system connections that are made through myelination etc.

Great stuff man.