Observational learning of a new motor skill: The effect of different model demonstrations

Introduction

During the process of learning a new motor skill, learners have to acquire a new pattern of spatiotemporal coordination. Model demonstrations are extensively used by instructors and coaches as a teaching strategy to facilitate acquisition of new coordination pattern, especially in sport settings. A meta-analysis of research on observational learning revealed a moderate to strong effect size (0.77) on movement dynamics and a small effect size (0.17) on movement outcome. ¹

A topic of interest in observational learning is to identify what information is extracted by the observer from a demonstration for later replication. This issue has been addressed by visual perception perspective (VPP). ² Mainly influenced by the visual perception theory of Gibson ³ and research in perception of biological motion,^4–6 VPP suggested that while observing a demonstration, relative motion information of the action, i.e. the spatiotemporal changes of body joints or extremities in relation to each other, is directly picked up and perceived by the visual system and later used to produce the action of the model. Moreover, it has been suggested that relative motion information available in a demonstration could be more effective in the first stage of motor skill acquisition, in which the learner attempts to assemble the efficient coordination pattern of the to-be-learnt movement.^2,7

According to the VPP, a demonstration should be particularly effective when relative motion information of the movement is highlighted. That can be achieved by removing the structural information such as body shape, color, and so on from a demonstration through representing the human body in the form of a point-light or stick-figure display rather showing the observers a classic video containing structural information. As such, observing point-light or stick-figure demonstrations should result in better motor performance and learning than observing a classic video demonstration, because observers are able to extract relative motion information easily from point-light and stick-figure demonstrations rather than a classic video demonstration.

In the last decade, a growing body of research has emerged to test this assumption. Horn et al. ⁸ asked participants to learn a kicking action in soccer by observing a classical video demonstration or a point-light demonstration. In contrary with the assumption of the VPP, they observed no significant difference between video and point-light demonstration groups for neither outcome accuracy nor intra-limb coordination pattern. There was also no significant difference between demonstration groups and control group in any of measured variables. In the following experiment, Horn et al. ⁹ removed knowledge of result in order to make the model as the only source to convey information of to-be-learnt movement. Under this condition, participants in video and point-light demonstration groups did not reduce outcome error but showed a relatively accurate approximation of the intra-limb coordination of hip–knee to the model’s pattern. In contrast, participants in the control group showed no improvement in movement coordination during practice. However, no significant difference was observed between video and point-light groups.

Using a similar methodology but a motor task including intra- and inter-limb coordination of upper and lower body, Breslin et al. ¹⁰ found that participants who observed either a video or point-light demonstration of a cricket bowling skill performed intra-limb coordination of the bowling arm significantly better than participants of control group. Observation of the model led also to a closer approximation of the movement time in comparison to control group. There was no significant difference between groups in terms of outcome scores or coordination of the nonbowling arm. In two later studies and using the same motor task, Breslin et al.^11,12 found some advantages for point-light demonstrations including whole body or the throwing arm over demonstrations showing only the wrist of the throwing arm or both wrists. A “normal” video demonstration was not included in these studies.

Altogether, based on the previous studies there have been conflicting evidence regarding the use of point-light display and it is not clear whether highlighting relative motion information in form of a point-light display plays a positive role on the observational motor learning process. Therefore, there is a clear need for additional research to confirm the effectiveness of using point light display in observational learning. The primary aim of this study, hence, was to extend the literature and to investigate further the hypothesis of VPP. As such, we compared the effects of observing point-light and stick-figure model demonstrations, which are assumed to highlight relative motion information, with a classic video demonstration, which is assumed to not highlight relative motion information, on motor performance and learning. As mentioned earlier, previous research has only compared a point-light demonstration with a video demonstration and the results showed no superiority for observing a point-light demonstration against a classic video demonstration. Current study also added stick figure display for further investigation in terms of motion perception. The baseball-pitch was chosen as motor task because it is a novel, complex, and multilimb sport skill. In order to evaluate skill acquisition, we compared the performance of the model and the participant in terms of intra- and inter-limb coordination pattern at the level of the overall movement as well as at the level movement phases. According to the VPP, it was hypothesized that point-light and stick-figure model demonstrations would lead to a better motor performance and learning in comparison to classic video model demonstration. It was also hypothesized that demonstration groups would perform better than control group in acquisition phase and retention tests.

Method

Task and production of model videos

A highly complex and dynamic throwing action, the baseball-pitch, was selected as motor task. The pitch consists of a clear phase structure including wind-up, stride, arm cocking, arm acceleration, arm deceleration, and follow-through (the phases are presented in Table 1), which makes it possible to do a detailed analysis at the level of both overall movement and individual movement phases. In this study, the analysis of motor performance and learning was performed at the level of both overall movement and individual movement phases.

OPEN IN VIEWER

Table 1.

Start and end points of the baseball-pitch phases with durations and allocated data points (pictures adopted with permission from Rojas et al. ¹³ ).

Phase		Start- and end-point	Phase duration (s)	Data points
	1 Wind-up	Start point: Left foot elevated from ground End point: Left knee at the highest point	1.072	100
	2 Stride	Start point: Left knee at the highest point End point: Striding finished, throwing arm back	0.958	90
	3 Arm cocking	Start point: Striding finished, throwing arm back End point: right arm was cocked	0.168	15
	4 Arm acceleration	Start point: right arm cocked End point: ball released	0.093	10
	5 Arm deceleration	Start point: ball released End point: velocity of the right arm decreased	0.075	10
	6 Follow-through	Start point: velocity of the right arm decreased End point: right arm decelerated fully	0.236	25
	Overall movement	Start point: Left foot elevated from ground End point: right arm decelerated fully	2.602	250

A right-handed male pitcher (age = 32 years) with 7 years playing experience in the second baseball league of Germany acted as model. To produce stick-figure and point-light models, retro-reflective markers were attached to the forehead, shoulder, elbow, wrist, hip, knee, ankle, and toe joints on left and right side of his body. Four digital and synchronized cameras filmed spatiotemporal positions of markers while he performed a baseball-pitch. Simi Motion software 5.0™ (SIMI Reality Motion Systems GmbH, Germany) was used for producing stick-figure and point-light videos. A normal video model was generated by using a digital video camera from a sagittal plane (Figure 1(a)). Point-light demonstration was constructed by recording movement of model with markers placed on the mentioned joints of the model and then processing the videos so that only the point-lights are visible in a darkened background (Figure 1(b)). A stick-figure demonstration was composed of the similar light points which are connected to each other with lines (Figure 1(c)). All demonstrations involved identical start and end points with four seconds duration. OPEN IN VIEWER

Figure 1.

Screenshots of model demonstrations used in the present study: (a) normal video, (b) point-light, (c) stick-figure.

Results

Discussion

The primary aim of the present study was to examine the hypothesis of the VPP, that relative motion information is picked-up and used by observers from a demonstration. According to this hypothesis, highlighting relative motion information within a demonstration, which could be achieved by generating point-light and stick-figure demonstrations, could result in better motor performance and learning in comparison to presenting the observers with a classic video demonstration, which involves all structural information. Therefore, in the present study we compared the effects of observing point-light and stick-figure model demonstrations with a classic video model demonstration on motor performance and learning of a baseball-pitch.

Generally, the results of intra- and inter-limb coordination patterns provided no support for this assumption, because there was no significant difference between demonstration groups in the acquisition phase or retention tests. Moreover, some evidence revealed that stick-figure demonstration group had even further approximation of movement pattern of the model than other experimental groups in some movement phases (e.g. arm cocking, arm acceleration, and arm deceleration). However, our findings are, in general, in accordance with the results of Horn et al. ⁹ and Breslin et al. ¹⁰ who found no superiority for point-light demonstration over video demonstration in observational learning. It appears from results of the present study along with results of the previous study that highlighting relative motion information within a demonstration does not necessarily guide changes in coordination.

It was also predicted that demonstration groups would show closer coordination pattern and movement time to the model than control group in acquisition phase and retention tests. This hypothesis was also not supported by the results; as participants in control group did not underperform participants in model demonstration groups in acquisition phase or retention tests. In other words, these results indicate that the video-based demonstration of the baseball-pitch, regardless of the form of demonstrations, had no positive influences on the motor learning process. Moreover, some evidence revealed that participants in control group showed closer movement time to the model than those in demonstration groups, e.g. in late retention test in terms of absolute movement time difference of overall movement or arm acceleration phase. Significant differences in movement phases between stick-figure group and control group might be because of poor performances of stick-figure participants in pretest and also because participants did not benefit from observing stick-figure model demonstration during acquisition phase. These findings are not in agreement with the results of Al-Abood et al., ¹⁴ Horn et al., ⁹ and Breslin et al.^10,12 who found model observation led to better motor performance and learning than no-model observation.

In our opinion, there are two possibilities which might interpret this contradiction. Firstly, it might be possible that amount of observational practice of our complex throwing skill was not sufficient to significantly improve the performance of demonstration groups during the acquisition phase. Secondly, presenting the participants with static images of pitch phases before the pretest might have prevented the effect of observing model demonstrations during the acquisition phase. An accurate look at the results of the pretest reveal that participants learned the baseball-pitch almost accurate at least in terms of throwing arm intra-limb coordination. Also, there were significant improvements from pretest to acquisition blocks and from pretest to retention tests in almost all measured variables, however, regardless of observing the model demonstration. It might indicate that the participants improved motor performances after pretest by physical practice and observing the model demonstrations had no positive effects.

These results raise the question if information of relative motion is extracted from static images of movement phases? The assumption of extraction of relative motion information was theoretically based on the Johansson’s studies.^5,6 Johansson claimed that the individuals perceive motion of the points over time perception in a point-light display in order to recognize the biological motion, i.e. form-from-motion perspective. ¹⁵ More recently, some studies in biological motion perception introduced an alternative perspective called motion-from-form perception.^15,16 According to this perspective, perception of body form/posture over time is crucial for recognizing the biological motion, not perception of motion of points over time. In our opinion, it might be possible that the participants in our research perceived body/posture information rather than relative motion information from static images of movement phases and used them to replicate the action of the model. Future studies should be directed to investigating the extraction of body form/posture information in the process of motor learning by observation.

A differentiated analysis of six phases of pitch showed that participants improved their performance in stride phase from pretest to acquisition blocks in all measured variables. It appears that significant improvements in coordination pattern from pretest to acquisition blocks occurred mostly in stride phase. It is obvious from angle–angle plot in Figure 2 that participants showed visually a large difference to the model in stride phase than in other phases. These results might indicate that stride phase of pitch is most effective phase to improve through practice. These results are important from practical point of view, because they provide a detailed examination of movement phases during the process of motor learning.

Conclusion

Results of the present study do not confirm the assumption of VPP regarding the extraction of relative motion information from a model demonstration. In addition, our research raise the question of what is the nature of information extracted from a series of static images of movement phases? In our opinion, it might be the information of body form/posture information over time. Therefore, we suggest that further studies should examine the extraction of body form/posture information from a model demonstration. Analysis of movement phases revealed that second phase of pitch, i.e. stride, is improved from pretest to acquisition blocks which might indicate that stride phase needs more practice than other phases.