20.4: Observation

Technology is constantly at our fingertips. Therefore, it is not shocking to witness an athlete pulling out their phone in the middle of training to watch a video of themselves, or someone else,performing a skill. If asked why they did this, a typical response relates to using the video to help them get better. For example, a baseball player may be seen reviewing their recent up to bat during practice in order to gain information on how to perfect the technique of the swing. Or an aspiring dancer may watch a YouTube video of a professional dancer performing a pirouette in hopes that they will be just like the professional dancer when they grow up. This is known as observational learning or modeling.

20.4.1. What is Observation Learning/Modeling?

Observational learning is the process by which individuals view a demonstration of either them self or someone else performing an action and, from this, is able to gain the ability to produce that same action (Cumming et al., 2005). There are many forms in which one can observe (for more detailed definitions see Ste-Marie et al., 2012); one can view the self or someone else and either of these techniques can be live or on video. Viewing the self is known as a self-as-a-model technique. Here, one can view themselves at their current skill level which could potentially include errors (i.e., self-observation) or at their best with little to no errors (i.e., self-modeling). The self-modeling technique can be broken down into two categories. The positive self-review technique is an edited video of the learner performing a skill to show the individuals best ever performance, such as those featured in highlight videos of athletes. The feedforward self-modeling technique uses footage of prior attempts which are then edited and spliced together to create a video of a behavior not quite achieved (Dowrick, 1999). For example, a gymnast may be struggling with certain skills in their floor routine. Their coach may decide to create a feedforward self-modeling video in which clips of each of their skills performed independently, but correctly, are spliced together to create an optimal full routine. In this situation, the gymnasts would see themselves performing their entire routine at the best they have every performed each skill independently: A snapshot into their future perhaps!

When viewing someone else, one can choose from either a skilled, unskilled, learning, mastery, or coping model. The most common forms used in sport are either the skilled or unskilled models. Thus, we will elaborate on these two types but direct readers to Ste-Marie et al. (2012) for further information on other modeling types if interested. A skilled model would demonstrate the skill with proper technique whereas an unskilled model would demonstrate that same skill but there would be obvious errors viewed as the individual has not yet acquired the proper technique. For example, a beginner volleyball player could watch a skilled player serving the volleyball in order to gain the accurate information on how the timing and proper technique of the overhand serve are carried out in hopes to be able to transfer that information into their own attempts. No matter what modeling technique one chooses, the consensus is that observing a demonstration works (Ste-Marie et al., 2020).

20.4.2. Conceptualizing Observation/Modeling

The two most prominent perspectives regarding the effectiveness of observing a demonstration stems from Bandura’s social-cognitive theory and Gibson’s direct perception perspective (for more on social-cognitive theory, see Chapter 5; Delli Paoli, 2021). Albert Bandura’s (1986) view relates to observation and social learning. This view suggests that we learn through observation by symbolically coding the observed behavior and translating that information into a cognitive representation. The cognitive representation is stored in our memory which then serves as a guide to later perform that skill. There are four subprocesses necessary for the individual to learn through observation: attention, retention, behavior reproduction, and motivation. Therefore, an individual must attend to the observed behavior in order to pick up the relevant information presented to them by the demonstration. From here, the information picked up would be retained in the form of a cognitive representation that can later be used to guide their own behavior when attempting to reproduce that same action. Motivation plays a key role as well. As such, if an individual is motivated to reproduce the behavior, they will selectively attend to the information presented to them via the demonstration and as a result are more likely to produce an accurate cognitive representation that would guide subsequent attempts to a desired reproduction of that movement. Past research has aligned with this perspective and has supported the notion of a cognitive representation in fact influencing subsequent actions (e.g., Frank et al., 2018).

Scully and Newell’s (1985) view aligns with Gibson’s (1950) direct perception perspective and suggests that we are able to simply “pick up” the information from a demonstration in which our visual system automatically processes the information without the need to create a symbolic representation. That is, we are able to directly perceive in a way that constrains our motor control system to act in accordance to what we see and produce a movement. Specifically, the individual viewing the demonstration directly perceives the relative timing of the joints and uses these motions and coordination patterns to develop their own movement patterns. An example of early related research was conducted by Johanson (1973), where light reflector markers were placed on the joints of a model while walking and running and individuals viewing the lighted dots in motion were able to distinguish the difference between movement patterns (i.e., walking vs. running). Within this perspective, point light displays (PLD) are typically used when models have light reflectors on their joints and it is the movement of the light reflectors that the learner observes on video. From these movements, the observer is able to gain information regarding the timing of the skill to be learned. Research conducted from this perspective suggests that PLD provide the relative motion information needed to learn a motor skill and that these displays are beneficial for highly complex skills (i.e., skills with multiple limbs involved) as well as less complex skills (i.e., skills with single limb movements; e.g., Kordi & Ghamary, 2014).

Photo by Budgeron Bach from Pexels

20.4.3. Why is Modeling Important?

No matter which perspective one is taking, there is ample evidence that support the notion that observation provides the necessary information to the learner and as a result the learner is better able to acquire or perfect the motor skill, create strategies, and assist in mental states (see Ste-Marie et al.,2020). As examples, observation techniques could highlight how to properly execute a free throw in basketball, help develop game strategies and routines when understanding power play formations in hockey, and help an athlete learn how to reach an optimal state of arousal before competing in their figure skating long program. The next section examines experimental results regarding the effectiveness of observation within these three functions.

In regard to evidence relating skill acquisition, the majority of research on the effects of using a skilled model has been promising. For example, the use of skilled models has been shown to increase the motor execution in areas as a power lifting (Sakadijanet et al., 2014), golf putting (Kim et al., 2017), badminton serving (Kamanga et al., 2013), throwing (e.g., Ghaehroudhani et al., 2016), and basketball shooting (e.g., Kordi & Ghamary, 2014). Another common model type is viewing the self. Research has shown positive effects in both the skill acquisition (e.g., Ste-Marie et al., 2011) and competitive enhancement (e.g., Rymal & Ste-Marie 2017). In Ste-Marie et al.’s (2011) work with competitive gymnasts, beam performance was significantly higher at competitions in which the gymnasts viewed a feedforward self-modeling video compared to competitions where no video was viewed. Recently, research has ventured into the exploration of combining models. This research has shown positive effects when learning how to perform sport skills such as hurdles (Amara et al., 2015), volleyball passing (Barzouka et al., 2007), gymnastic sequences (e.g., Robertson et al., 2018), as well as balancing tasks(Karlinsky & Hodges, 2018). In Robertson and colleagues’ (2018) research, participants either viewed the self or a combination of the self and a skilled model when attempting to learn a gymnastics sequence. The results indicated that both groups increased in performance however the group that received the combo model intervention outperformed the self-model group only for both skill learning and error recognition.

Cumming et al. (2005) reported developing strategies as the second most frequent use of observation, following the skill function. That is, the reason for which individuals use observation can also be tied to developing strategies/game plans to help achieve the motor skill. The use of observation as an actual intervention to influence strategies is sparse, but the few studies that have investigated this have shown promising results. For example, gymnasts in Rymal & Ste-Marie’s (2018) research used their self-modeling video as a means to strategically plan motor execution, create adaptive inferences, and analyze tasks specific to their bar routine. Frank and colleagues (2018) were interested in the effects of observation on physical performance and mental representations of the mechanics of the golf swing. Participants were assigned to one of two groups; an observational group and a combined observational and physical practice group. Despite both groups increasing physical performance and acquiring the mental representation of the golf swing (i.e., identifying correct mechanics of the swing when presented with a video), only the observation plus physical practice group was able to transfer that strategy into motor output. Thus, physically practicing alongside of an observation intervention helped identify skill mechanics which then lead to the use of those mechanics in future attempts.

Research investigating the performance function typically coincides with research investigating skill learning as the main outcome. Self-efficacy seems to be the most commonly measured psychological construct in such research. Increases in self-efficacy have been found in areas such as swimming (Clark & Ste-Marie, 2007), gymnastics (Robertson et al., 2018), as well as hockey (Feltz et al.,2008). Extending beyond self-efficacy measures, much of Rymal and colleagues’ work with divers and gymnasts (2010,2017,2019; Ste-Marie et al, 2011) investigated the relationship of self-modeling and self-regulatory processes during a competition. The results of both divers and gymnasts suggest that the observation technique promoted performance characteristics such as self-efficacy, motivation, satisfaction, and arousal control during a competitive event.

20.4.4. Application

As noted by Ste-Marie and colleagues (2012; 2020) there are many things to consider prior to determining an observation intervention; however, it is not our goal to review the entire Applied Model for the Use of Observation (i.e., AMUO). We will instead give some avenues and suggestions as to potential applications. Specifically, we will discuss some options with respect to the type of model to use (i.e., “who” should be observed), as well as when and how to implement observation with athletes (for a full review of the AMUO see Ste-Marie et al., 2012; 2020).

20.4.5. Who Should be Observed?

Should the learner observe someone else?Would it be more beneficial if it were a skilled model or an unskilled model? Or perhaps the self should be viewed, but should it be at their current skill level, at their best, or at a level that is slightly better than their current state? These are all worthwhile questions that have some evidence from the research literature to answer them! Overall, existing research demonstrates that no matter who one chooses, all models seem to be effective to some degree (Ste-Marie et al., 2012; 2020). If looking to increase skill, a skilled model would be appropriate as a skilled model provides the correct information regarding the timing of joint movements. A skilled model can also assist in the cognitive representation, providing information to guide the learner when attempting to perform skill (e.g., Frank et al., 2018). An unskilled model would also be appropriate, as an unskilled model is thought to increase the problem-solving process and thought to help with error detection and correction of mechanisms. A key feature to consider is model similarity. If an athlete perceives they are similar to a model then it is more likely they will pay attention, associate to similar timing of the joints, and thus have a greater effect on skill learning (McCullagh, et al., 2012).

However, no one is more similar to you than yourself. Previous research suggests that there is a greater advantage to viewing the self, as compared to viewing other models. Noteworthy is that Ste-Marie et al.’s (2020) re-examination of the AMUO suggest that it is the combination of different model types that are giving the most promising results for motor outcomes and psychological aspects (e.g., Robertson et al., 2018). Furthermore, Karlinsky & Hodges (2018) examined dyads practicing a balancing task and despite no differences between those who were paired to those whom were not, paired practice might help coaches that have limited time due to the fact that dyads practicing together can both concurrently learn a new skill by examining their teammate as a model. As such, we will not suggest one model type over the other when working with individuals acquiring a new skill, but we do suggest to use more than one model type (e.g., skilled and self-model) as a means to enhance skill acquisition and performance.

20.4.6. When and How to Implement Observation Techniques?

According to Ste-Marie et al., (2012; 2020) the majority of research is still unclear as to whether one should show a demonstration before, during, after, or combinations thereof. However, most research that implements the observation intervention before and during has shown positive results. To date, very few research studies have compared the scheduling in order to determine which is actually more beneficial. Outside of the sport environment, research has suggested that viewing either before or during physical practice seems to be advantageous to the motor task outcome (e.g., Herbert, 2018). Furthermore, researchers are exploring the use of a self-control protocol when using observation to enhance motor skills. Here, the learner chooses how much and when they would like to view the demonstration. For the most part, evidence is still unclear as to which is best for skill acquisition: self-controlled viewings or other-controlled viewings. There is, however, a trend moving towards self-controlled viewings over experimenter-controlled viewings (e.g., Marques & Corrêa, 2016). What is interesting however, is that much of the research in this area has shown consistent findings regarding how much a learner actually needs to view a demonstration in order produce a change in behavior.Specifically, if self-controlling the frequency of viewings, the learner still benefits from the observation technique but these benefits arise from fewer viewings than compared to experimenter-controlled viewings (e.g., St. Germain et al., 2019). From a practical standpoint, coaches and practitioners are not able to be with one learner at all times to ensure they do their viewings. However, by allowing athletes to be in control of their own viewings, they may get just as much of a performance improvement in a much shorter time.