Animations showing Lego manipulative tasks: Three potential moderators of effectiveness

Highlights

• Learning a Lego task was similar when tested in physical and virtual environments.

• In both environments, animations were slightly better than static pictures.

• Not showing the hands solving the task was better than showing them.

• Focused rather than unfocused complete visual information of the task was better.

Abstract

Evidence suggests that transient visual information, such as animations, may be more challenging to learn than static visualizations. However, when a procedural-manipulative task is involved, our evolved embodied cognition seems to reverse this transitory challenge. Hence, for object manipulative tasks, instructional animations may be more suitable than statics. We investigated this argument further by comparing animations with statics using a Lego task shown to university students, by examining three potential moderators of effectiveness: (a) the environment of manipulation (virtual or physical), (b) the quality of visual information (focused or unfocused), and (c) the presence of hands (no hands or with hands). In Experiment 1 we found an advantage of animation over statics, and no differences among the environments. In Experiment 2, we again observed an animation advantage, a small advantage of focused static information compared to unfocused static information, and a positive effect of not showing the hands.

Introduction

Research into the effectiveness of instructional dynamic visualizations (e.g., animation and video) has been extensive. A common strategy has been to compare animated presentations with static equivalents, with the desired aim of showing that dynamic visualizations are the superior format. Even though some domains most conducive to learning from animations and videos have been identified (see Höffler and Leutner, 2007, van Gog et al., 2009), research has frequently produced mixed outcomes and identified a number of moderating variables (see Tversky, Morrison, & Betrancourt, 2002). Consequently, there is still much to be done in order to understand and identify the multitude of factors that impact on the effectiveness of instructional animations.

With the current study we aimed to extend the research comparing animated with static presentations by using a manipulative task (Lego bricks), which required the memorization of its final position (object memorization task). We also examined some of the moderating variables that could impact on the effectiveness of both the visualization and the execution of the manipulative task. In particular, we investigated two visualization moderators: (a) the presence of an embodied element (hands), and (b) the quality of visual information shown. We also studied one moderator for the execution of the task: whether it was performed in a virtual or in a physical environment.

There has been much expectation that animated visualizations should provide a more effective learning environment than static presentations (see Chandler, 2009). The meta-analysis by Höffler and Leutner (2007) identified some moderating variables (such as knowledge depicted or task content) that promoted instructional animations over statics (see Section 1.1.2). However, if these moderators are not controlled, animations may be less or equally effective than static images for learning. An example where groups studying static illustrations outperformed groups studying animations was reported by Mayer, Hegarty, Mayer, and Campbell (2005) in four experiments where university students had to learn about mechanical systems. Similarly, statics have been observed to be better or more efficient educational resources than animations in learning about topics of probability (Scheiter, Gerjets, & Catrambone, 2006), biology (Koroghlanian & Klein, 2004), and symbol memorization (Castro-Alonso, Ayres, & Paas, 2014b). Also, there are studies showing no significant differences between dynamic and static images, in tasks such as learning the mechanisms of brake systems (Mayer, DeLeeuw, & Ayres, 2007), or a flushing toilet (Narayanan & Hegarty, 2002). Because of the uncertainty surrounding the learning performance due to instructional animations, researchers have started to examine the reasons why they might not be effective. One reason gaining considerable support is the transient information effect (see Ayres and Paas, 2007a, Ayres and Paas, 2007b).

The transient information effect is a relatively new research area (see Sweller, Ayres, & Kalyuga, 2011), and occurs when a permanently displayed educational material produces higher learning outcomes than an equivalent transient format. Arguably the most ubiquitous form of transient information in education is speech, which disappears as soon as it is spoken (unless it is recorded in some fashion). Studies (e.g., Leahy and Sweller, 2011, Singh et al., 2012) have shown that lengthy spoken explanatory texts (transient form) may lead to less learning than identical written texts (permanent form).

The transient information effect can also be observed with instructional animations, where many of these depictions can include fleeting images that do not stay visible on the screen for very long. When studying from a transient animation, learners may have to remember previous critical information that has disappeared, and integrate it with new information, in order to understand and learn about a new concept (see Ayres & Paas, 2007b). According to cognitive load theory (a theory that considers how instructional design impacts on working memory and learning; see Sweller et al., 2011) this type of mental processing is very demanding for working memory, and leads to learning deficiencies (e.g., Castro-Alonso et al., 2014b). In contrast, static visualizations, which are more permanent forms of information, can be re-examined as needed and impose less cognitive burdens. Under such conditions, instructional static pictures may produce better learning outcomes than comparable dynamic visualizations, as some empirical findings described above have shown.

Due to these problems with some animations, instructional strategies such as segmenting (to show shortened versions rather than a whole animation; e.g., Wong, Leahy, Marcus, & Sweller, 2012) and learner control (to include buttons to slow down or pause the animation; e.g., Höffler & Schwartz, 2011) have been used effectively to manage the transitory information of animations (see Ayres and Paas, 2007a, Ayres and Paas, 2007b, Castro-Alonso et al., 2014a). However, research has also shown that some domains are particularly suited to learning from animations, regardless of transitory effects. One such collective domain is learning about human movement through procedural–manipulative tasks.

Höffler and Leutner (2007) identified several moderators for the effects of animated or static presentations on learning. One of these moderators was the type of knowledge depicted, whether procedural-motor, declarative, or problem-solving knowledge. The highest advantage of animated over static presentations (d = 1.06) was reported when procedural–motor knowledge was to be learnt. For example, greater outcomes from animations over static images have been reported in procedural–manipulative tasks as diverse as (a) disassembling a machine gun (Spangenberg, 1973), (b) bandaging a hand (Michas & Berry, 2000), (c) replicating origami models (Wong et al., 2009), (d) solving puzzle rings (Ayres, Marcus, Chan, & Qian, 2009), and (e) copying different knots (Ayres et al., 2009, Garland and Sánchez, 2013).

To explain this phenomenon of better learning with animations (in spite of their transient information) when the task involves procedural–manipulative knowledge, or human movement, researchers have proposed links with the work of Geary and evolutionary biology. Geary distinguishes between biologically primary skills—evolved and thus relatively effortless cognitive abilities—and biologically secondary skills—not evolved and thus effortful abilities that must be learnt through instruction (see Geary, 1995, Geary, 2002, Geary, 2007). As biologically primary skills are easier to learn, they consume less working memory capacity than secondary skills. In consequence, when learning from animations that show primary skills (e.g., a manipulative task), students can handle the transiency problem of these depictions in a much effective manner than when dealing with secondary skills. Paas and Sweller (2012) have termed this phenomenon as the human movement effect.

In conclusion, humans learn manipulative tasks easily because this is an evolved effortless primary skill. In other words, tasks that involve employing the hands to manipulate objects following a sequence or procedure are relatively easy to humans. This implies that there are evolved cognitive mechanisms that allow humans to manage the transiency of the manipulations. Arguably, the most important of these mechanisms is the mirror neuron system (see van Gog et al., 2009).

Mirror neurons are visuomotor cells that are activated not only when individuals perform an object manipulation, but also when they watch other individuals doing the same action (see Rizzolatti & Craighero, 2004). These neurons compose the mirror neuron system, which, in connection to other perception–action mechanisms, provide an extensive brain representation to aid understanding of human movement (e.g., Cross, Hamilton, & Grafton, 2006) and manipulative tasks. Noteworthy, the existence of these systems is an indicator that humans have evolved an embodied cognition, namely a cognitive architecture that links perception and bodily action to allow humans to thrive in their environment (see Barsalou, 2010, Wilson, 2002).

Because the simultaneous activation of perceptual and motoric streams is an evolved phenomenon, this activation is expected to be more pronounced with natural and evolved manipulative tasks rather than non-natural object manipulations. For example, embodied cognitive systems are activated to a greater extent when watching human-as compared to robotic-arm motions (e.g., Kilner et al., 2003, Press et al., 2005). Similarly, Shimada and Oki (2012) reported that an area of the mirror neuron system was triggered more when watching fluent and natural rather than jerky and paused arm movements. This result can explain the human movement effect and why animated visualizations that show natural motions are better than static images to model manipulative tasks. In conclusion, these findings suggest that the mirror neuron system and related embodied mechanisms are preferentially triggered in natural situations, aligned with their evolution. This could imply that, when designing an instructional visualization, natural situations should be preferred. In other words, the effectiveness of an instructional visualization to learn a manipulative task may be moderated by embodied mechanisms, as discussed next.

As embodied mechanisms get activated more during natural rather than unnatural manipulations, it is logical to suppose that object manipulations showing the active role played by the hands would have greater effects than manipulations that do not show the hands solving the tasks. Evidence for this conclusion was found in a review by Brockmole, Davoli, Abrams, and Witt (2013) reporting diverse positive effects that the observation of hands had on perception, attention, and memory. This suggests that the visualization of hands may be an important factor in learning object manipulations. However, it has been shown that manipulative tasks can be sufficient to activate embodied cognition systems, without the need to depict the hands (e.g., Longcamp et al., 2006, Patuzzo et al., 2003, Wong et al., 2009; but see Fadiga, Fogassi, Pavesi, & Rizzolatti, 1995). In conclusion, although the inclusion of hands in the visualization may have beneficial effects on learning a manipulative task, it may not always be needed, for example when the objects alone are considered by the learners as sufficiently manipulative. Besides this embodied factor, another variable that may moderate the effectiveness of an instructional visualization is the quality of information depicted.

As pointed out by Tversky et al. (2002) instructional animations may convey more information than static pictures. In such situations, we could expect animations to be superior to statics. Thus, adopting the same argument for static presentations only, we could predict that a presentation of statics with all the steps of a learning task should contain more information (complete) and thus be a better learning tool than a presentation with less static pictures (incomplete). However, in contrast, Mayer et al. (2005) hypothesized that one of the advantages of static frames over animations was that the statics contained less and more focused information, thus showing only the most important learning steps. Hence, we could expect that a presentation with all the static steps (unfocused) would be an inferior learning tool than a presentation with less and key static frames (focused), because the latter would allow learners to invest less working memory resources by studying only the most important information. Combining these two arguments together, when presenting instructional information, its quality (unfocused versus focused) is more important than its quantity (complete versus incomplete), as having extra information not relevant to the task wastes the limited cognitive resources. To finish this theoretical background, in addition to the two moderators discussed, which concern the visualization of the manipulative task, there is an important variable that may affect the execution of the task itself, as described next.

Overall, students can attempt to perform a manipulative task in two ways: (a) physically, moving with their hands real elements; or (b) virtually, manipulating with virtual devices (mouse, joystick, etc.) digital replicas of the elements, in a virtual environment (cf. Zacharia & Olympiou, 2011). Evidence from various studies of different topics and students suggests that virtual and physical environments are largely equivalent for learning (e.g., Baki et al., 2011, de Jong et al., 2013, Klahr et al., 2007, Triona and Klahr, 2003, Zacharia and Olympiou, 2011). Also, there is a recent study showing the benefits of virtual over physical manipulations in learning about chemistry (Barrett, Stull, Hsu, & Hegarty, 2015). Acknowledging these findings, we planned to design our manipulative Lego task in the most convenient environment (especially for data collection), which is the virtual world (cf. Klahr et al., 2007, Triona and Klahr, 2003). However, the visualization of the task showed real (physical) Lego objects, and thus it followed both the identical elements theory (cf. Thorndike & Woodworth, 1901) and the congruence principle for effective graphics (Tversky et al., 2002). Consequently, there was a possibility that performing the task with the same physical objects watched in the learning presentations would be more effective than replicating the task with virtual replicas of the objects. Hence, we first investigated in Experiment 1 whether completing the tasks physically was better than virtually.

The study consisted of two experiments where the manipulative task was to construct a three-dimensional shape with Lego Duplo™ bricks from memory, having observed its construction through a presentation without textual explanation (observational learning). The present study had four main aims. The first aim was to discover if animated presentations would be superior to static presentations when learning how to construct Lego shapes. The second aim was to investigate potential differences between the learners when constructing the shape physically or virtually. The third aim was to investigate whether the number of frames shown during the static presentations made a difference. The final aim was to explore the potential effect of observing the hands moving the Lego bricks versus not observing them when viewing the different presentations.

Four general hypotheses were tested in the study to match these aims:

• The animation presentation would lead to higher learning outcomes than the statics presentation (Hypothesis 1).

• The physical environment would lead to higher learning outcomes than the virtual environment (Hypothesis 2).

• For static presentations, showing a fewer (focused) number of static frames would lead to higher learning outcomes than showing more (unfocused) static frames (Hypothesis 3).

• The with-hands condition would lead to higher learning outcomes than the no-hands condition (Hypothesis 4).

Hypothesis 1 was consistent with previous findings of animations outperforming statics when depicting procedural–manipulative tasks (e.g., Garland and Sánchez, 2013, Höffler and Leutner, 2007, Wong et al., 2009). Hypothesis 2 was supported using the identical elements theory (cf. Thorndike & Woodworth, 1901) and the congruence principle for effective graphics (Tversky et al., 2002). Hypothesis 3 was aligned with the notion that giving focused visual information about a task should be better than giving more but unfocused information (Mayer et al., 2005). Lastly, Hypothesis 4 was consistent with findings about the positive effects of observing hands on cognition (Brockmole et al., 2013).