An especial skill: Support for a learned parameters hypothesis

Abstract

We tested the ‘learned parameters’ hypothesis as an explanation of the ‘especial skill effect’. Outcome attainment and movement kinematics were recorded for 10 expert and 10 novice players performing basketball free-throw shots at five distances (11–19 ft) with a regular and heavy weight basketball. As predicted, experts performed better than expected relative to the regression equation at the 15 ft, free-throw line with the regular basketball, supporting the ‘especial skill effect’. This effect was not present for the experts when shooting with the heavy ball. Novices did not show an advantage at the free-throw line when performing with either ball. Although the outcome attainment scores support the ‘learned parameters’ hypotheses, kinematic analysis failed to identify differences in the movement pattern for the especial skill, suggesting that these skills (i.e., shooting at different distances) are not governed by separate motor programs.

Introduction

Most cognitive theories of motor learning are based on the view that large amounts of practice lead to the formation of a specific memory representation for that specific class of motor actions. The proposal that there is a single memory representation for a class of motor skills as opposed to a unique representation for each single movement within a class has been debated for decades (for a review, see Schmidt & Lee, 2005). The formation of a single memory representation for a class of actions has been proposed to provide two advantages; first, it alleviates the storage problem that may be encountered if each motor skill within a class had an associated memory representation, and second, it provides motor flexibility when a performer encounters unfamiliar tasks (Schmidt, 1975, Schmidt, 2003).

In recent years renewed interest in this topic has emerged, challenging the concept across all motor skills that a single generalized memory structure represents all actions within a class. In studying the performance of expert basketball players executing the set shot from the foul line (i.e., 15 ft from the basket), this well-practised shot was found to be more accurate than at closer distances and hence from the performance scores predicted based on a generalized motor program view of behaviour. Keetch, Schmidt, Lee, and Young (2005) referred to this type of skill as an ‘especial skill’, based on the assumption that its uniqueness is a result of massive amounts of practice at a specific distance. There has been some debate as to the characteristics responsible for this enhanced performance as well as the resulting specificity of this particular action within the general class of actions (i.e., set shots).

One hypothesis is that success is tied to the visual characteristics of the task, such that the visual information pertaining to the ball and the distance and height of the basket remain constant relative to the player. There has also been research to show that the angle of elevation of the basket relative to the shooter is the most important visual cue underlying shooting accuracy (de Oliveira, Oudejans, & Beek, 2009). A second hypothesis is based on the familiarization the player has developed to the specific movement parameters of the task (e.g., calibrating their force output to the specific 15 ft distance required). There have been relatively few attempts to distinguish between these viewpoints, beyond research conducted by the original authors (see also Keetch, Lee, & Schmidt, 2008) and one study in baseball by Simons, Wilson, Wilson, and Theall (2009). As we discuss, there is evidence to suggest that the specific location (not just the distance), and the shot type, are critical characteristics for an especial skill, and as such the current viewpoint is that the visual characteristics pertaining to the foul line and basket distance are the main features underlying the emergence of this skill. However, this conclusion was based on manipulations to shot location (where the parameter distance remained constant but the visual information changed). Changes to the weight characteristics of the ball impacting on the absolute force output characteristics of the movement would enable a more direct manipulation of a task parameter, while keeping visual information constant, to more directly test this hypothesis. Furthermore, as yet, no attempt has been made to explore whether this phenomenon is a unique property of expert performance, and whether movement kinematics change across the various distances supporting the idea that especial skills are controlled by different motor programs (see Keetch et al., 2005).

According to schema theory (Schmidt, 1975, Schmidt, 2003) a class of actions is represented by two memory representations, namely, a generalized motor program (GMP) and a recall schema. The GMP is conceptualized as a store of invariant information for a class of specific motor skills, with each action in the class having similar form characteristics. For example, overarm throwing can be classified as a class of motor skills, as all attempts at the task have similar form characteristics (i.e., the temporal spatial position of the upper and lower arm in relation to each other remain the same). Based on experience of actions within a class, the recall schema is generated relating to the movement time and force required for specific conditions. Schema theory is an example of a generality view of motor skill learning in memory. Attempts at performing a movement are not stored, but rather what is stored is a general memory representation that can be modified via the recall schema to meet task demands or a changing environment. According to this generality view, improvements in performance are achieved by practicing slight variations in the movement.

An alternative to generality is a specificity view of motor skill learning. Accordingly, each skill is stored along with the contextual information (i.e., sensori-motor) that accompanied the practice of the skill (e.g., Proteau, 1992). In this way, performance on a specific skill is only enhanced to the extent that it has been practised in a manner similar to the way it is tested (see also Davies and Thomson, 1988, Smith and Vela, 2001). The massive amounts of practice on the basketball set shot at the foul line compared to other distances from the basket provides an interesting context to explore the content of generalized memory structures and determine the degree of generality or specificity in performance.

In a recent experiment, Keetch et al. (2005) required expert male basketball players to perform the set shot at seven distances from the basket. Performance was generally in line with force variability predictions based on general motor program ideas that as distance increased, outcome attainment decreased. However, when a regression model was applied to the data the authors found that success at the foul line was significantly better than what was predicted. This finding was specific to the set-shot (i.e., the typical shot performed at the foul line), and not observed when a jump-shot was required at the various distances. In a second experiment, conducted with females across five distances from the basket, findings from the first experiment were replicated. In addition, removal of visual cues from the surrounding environment (i.e., covering of markings on the floor) did not change this effect. The authors hypothesized that the enhanced performance of the set shot at the foul line was a result of either specificity in visual aspects of the movement (related to the perceived distance and location of the player from the net) or parameter aspects of the movement, such as force requirements. No empirical support was available to support one explanation over the other.

Subsequently, two further studies have been conducted; one by the original authors, again in basketball, and a second by a different group of researchers that assessed the especial skills’ phenomenon in baseball pitching. In the latter study, an attempt was made to merely replicate the ‘especial skills effect’ rather than to identify the underlying mechanisms. Simons et al. (2009) showed that throwing accuracy at the 60.5 ft distance (the typical pitching distance in baseball) was more accurate than two distances 1 ft in front or behind the distance where massive amounts of practice had been acquired. Keetch et al. (2008) contrasted the ‘learned parameters’ hypothesis with a ‘visual context’ hypothesis to determine whether the memory representation for the free throw had a specific visual context, given that the free-throw shot is always taken from the same visual angle and distance from the basket. Keetch and colleagues manipulated the horizontal visual angle at which the experts viewed the basket by requiring them to take a set shot at seven points around the basket at a distance of 15 ft (i.e., at the foul line and directly in front of the basket, three points to the left of the foul line at 45°, 60° and 75° angles and three locations to the right at 105°, 120° and 135° angles). Performance at the foul line was significantly better than shots taken from the left or right of the foul line. The detriment in performance at the various locations was attributed to a change in the visual angle, providing support for the ‘visual context’ hypothesis. As the movement parameter distance was held constant, Keetch et al. (2008) concluded that parameter scaling issues did not lead to the decrease in outcome attainment in this experiment.

No attempt has been made to test the ‘learned parameters’ hypothesis by directly manipulating the scaling aspects of the basketball free throw at the foul line. It is possible that the emergence of an especial skill is not only due to the visual context but also due to the movement-scaling characteristics during the shot at the foul line. Furthermore, analysis of movement kinematics, such as the relative timing of the limbs during the set shot, would allow an examination of a schema theory account of this phenomenon. If the set shot at the foul line is controlled by a different motor program than shots performed at nearby distances, we would expect to see a different pattern of relative timing (i.e., a change in the invariant characteristics) at the foul line compared to other distances. If the kinematics do not differ at each distance, including the foul line, the emergence of an especial skill would be attributed to improved movement scaling or parameterization of the GMP, via the recall schema.

To test these proposals we disrupted scaling of the action at the 15 ft foul line by manipulating the weight of a standard basketball while keeping the visual features the same. We predicted that if an especial skill is a result of movement parameter scaling, the introduction of an overweight basketball would produce a detriment in outcome performance at the foul line and hence disrupt the ‘especial effect’ (i.e., there would be no significant difference between the expected and actual outcome scores). In addition to recording movement kinematics, a novice group of basketball players was included in the design to more specifically test whether the especial skills phenomenon is a result of ‘massive amounts of practice’ at the 15 ft foul line, or rather a property of the sensori-motor conditions afforded at this specific location, regardless of experience. If the phenomenon is a result of practice, the novice group should not be advantaged at the 15 ft free-throw line compared to the other distances from the basket, nor will their outcome-location relationships depend on whether they are shooting with the regular or heavy weight basketball.