Learning and Instruction 55 (2018) 32–40

Contents lists available at ScienceDirect

Learning and Instruction

journal homepage: www.elsevier.com/locate/learninstruc

Optimizing learning from animation: Examining the impact of biofeedback T

∗
Lijia Lin , Ming Li
Faculty of Education, East China Normal University, China

A R T I C L E I N F O A B S T R A C T

Keywords: The purpose of this study was to investigate whether EEG biofeedback would help engage learners in obtaining
Animation procedural knowledge from animated or static visualizations. Two specific forms of this feedback were in-
Feedback vestigated. The first was active feedback that required learners to interact with the feedback using hand gestures.
Engagement The second was constructive feedback that instructed learners to verbally reflect on their learning processes by
utilizing the feedback. A total of 116 college students were randomly assigned to one of six experimental con-
ditions formed by a 2 X 3 factorial design with the visualization type (animated vs. static) as one factor and the
feedback type as the other factor (constructive vs. active vs. no feedback). The results revealed that learners who
received no feedback had significantly higher perceived attentiveness and value than their peers who received
constructive feedback. Additionally, when constructive feedback was provided, learning procedural knowledge
from animated visualizations reduced perceived difficulty in learners when compared to static visualizations.
These findings were discussed in terms of the implications, limitations, and future research directions.

1. Introduction from 1970s to early 2000s, which resulted in 76 comparisons between

Dynamic visualizations, or animations, have been increasingly uti-
lized in educational and training settings over recent years. Not only
has decades of research demonstrated an overall positive effect of
animations, but has also shown a link between animations and specific
types of knowledge acquisition (Höffler & Leutner, 2007). Moreover,
the animation effect on learning also depends upon a variety of factors
(Berney & Betrancourt, 2016). In the current study, we investigated
using two specific types of biofeedback to engage learners in the pro-
cessing of animated visualizations.
1.1. Learning with animations
Animations are a type of visualization that “generates a series of
frames, so that each frame appears as an alternation of the previous
one, and where the sequence of frames is determined either by the
designer or the user” (Bétrancourt and Tversky, 2000, pp. 313). Based
on this definition, animations have an advantage over static images in
their ability to illustrate conceptual changes, procedures, and dynamic
processes. Animations provide external scaffolding for learners to build
correct mental models regarding learned knowledge and skills so that
learners do not need to make (incorrect) inferences from static graphics.
In the existing literature, two meta-analytic studies have shown the
potential benefits of using animations for learning and instruction.
Höffler and Leutner (2007) reviewed 26 relevant studies conducted
animations and static pictures. Their analysis revealed a medium-sized
effect favoring animations. Furthermore, three factors were identified
in their meta-analysis as the moderating factors for the effectiveness of
animations. The first is the role of animation. Specifically, animations
that are representational in nature were found to be more effective than
static pictures. The second moderator is the type of knowledge studied.
Using animations was found to be more effective for learning proce-
dural knowledge, as opposed to learning declarative knowledge and
solving problems. The third moderator identified by Höffler and
Leutner is the degree of realism. Animations with a high degree of
realism were found to be more effective than static pictures. Berney and
Betrancourt (2016) meta-analyzed 140 pairwise comparisons (ani-
mated vs. static) from 50 research reports published from 1976 to 2013.
In contrast to the work of Höffler and Leutner, they used Bloom's re-
vised taxonomy to code the outcomes with two dimensions: a cognitive
dimension (to remember, understand, apply, analyze, evaluate, and
create) and a knowledge dimension (factual, conceptual, procedural,
and meta-cognitive knowledge). Berney and Betrancourt found that the
positive effects of animation did not vary across different learning
outcomes, no matter whether they were factual, conceptual, or proce-
dural knowledge. However, they did find that system-paced animations
with audio narrations or without accompanying text were superior to
static visualizations. In summary, the results of these two meta-analytic
studies are inconsistent and somewhat contradictory.
Recent empirical studies reported in the literature also reveal mixed

∗
Corresponding author. Faculty of Education, East China Normal University, Shanghai, 200062, China.
E-mail addresses: ljlin@psy.ecnu.edu.cn, lijia.lin615@gmail.com (L. Lin).

https://doi.org/10.1016/j.learninstruc.2018.02.005
Received 16 August 2017; Received in revised form 16 February 2018; Accepted 21 February 2018
0959-4752/ © 2018 Published by Elsevier Ltd.
L. Lin, M. Li
findings in terms of the superiority of animations over static graphics.
For instance, Turkay (2016) compared the relative benefits of using
whiteboard animations (a type of animated video that demonstrates the
drawing processes on a whiteboard to explain concepts or processes) to
static images in promoting learners’ acquisition of physics concepts.
The results showed that the whiteboard animations had positive effects
on retention, engagement, and enjoyment. Such positive effects of
animation were also found in other recent empirical research, where
learners were involved in learning concepts, processes, and procedures
(e.g., Castro-Alonso, Ayres, & Paas, 2015; Stebner, Kühl, Hoffler, Wirth,
& Ayres, 2017). However, a number of empirical studies have revealed
different findings in a variety of learning tasks. For instance, Castro-
Alonso, Ayres, and Paas (2014) reported two experiments in which
learners were asked to memorize abstract symbols through either ani-
mations or static pictures. They found that static visuals were better
than their animated counterparts in terms of accuracy and efficiency for
a task with a high level of transience (i.e., visible for only 12 s). In an
experiment conducted by Wong, Leahy, Marcus, and Sweller (2012),
learners were instructed to complete a paper-folding task by viewing
either static graphics or animated videos. Their findings turned out to
be a little complicated: Animations were found to be superior to static
graphics only when instructional materials were presented in short
segments; but these two types of visualizations were equivalent in terms
of impacting learning when materials were presented in longer seg-
ments.
A variety of theoretical accounts have emerged to explain the cog-
nitive processes involved in learning with animations. Specifically,
some researchers argue that both perceptual and cognitive processing
are involved (Wagner & Schnotz, 2017). Learners initially process
animations on the perceptual level by parsing the dynamic flow into
separate elements. They later build their mental models by making
sense of how these elements operate together (Lowe & Boucheix, 2017).
Static and animated visualizations may be equally effective when the
learning goal is to understand the composition of a system because both
types of visualizations present the key elements or states (Wagner &
Schnotz, 2017). Nonetheless, static visualizations may be less effective
than animated visualizations when the learning goal is to understand
the mechanism of how a system works, since learners may experience
more difficulty making correct inferences from static visualizations
(Hegarty, 1992). Thus, by externally providing information, animations
can reduce a learner's cognitive burden. Other researchers have looked
at the animated-static comparison from an embodied perspective. They
argue that the mirror neuron system (a system that plays an important
role in understanding actions) in humans can be activated during the
process of learning procedural knowledge from dynamic visualizations
(van Gog, Paas, Marcus, Ayres, & Sweller, 2009). Such learning, trig-
gered by this activation, imposes substantially less mental effort on
learners due to its nature of being evolutionary primary knowledge
(Paas & Sweller, 2012). De Koning and Tabbers (2011, 2013) further
argued that engaging learners in gestures, object manipulation, and
other human movements can improve learning from animations. This is
because learners' motor experiences related to the content presented in
the animation help enhance the construction of their mental models.
1.2. The engagement framework
One critical factor, learner engagement, seems to be missing from
the discussion on the animated-static comparison. The mixed findings
reported in the existing literature could be attributed to learners' dif-
ferent levels of engagement. Theoretically, engagement is a multi-di-
mensional construct, which encompasses behavioral, emotional, and
cognitive aspects. Behavioral engagement refers to students' participa-
tion, attention, and effort. Emotional engagement refers to students’
interest, boredom, and other affective constructs. And cognitive en-
gagement refers to the investment of learning (Fredicks, Blumenfeld, &
33
Learning and Instruction 55 (2018) 32–40
Paris, 2004). This broad and ambiguous conceptual delineation not
only shows that intrinsic motivation (e.g., interest, value, and effort)
and engagement are reciprocally connected, but also extends engage-
ment to internal cognition (such as cognitive load). According to
Blumenfeld, Kempler, and Krajcik (2004), if individuals are focused and
interested in the materials or tasks presented in a learning environment,
they may be more engaged in the learning processes, which leads to
germane cognitive processing and enhanced learning outcome. They
also argue that the success of knowledge or skill acquisition may result
in their perceptions of competence and value, as well as reduced
pressure.
An emergent engagement framework provides a new perspective to
explain this inconclusive literature with regard to the animated-static
comparison. Based on learners' overt behaviors, Chi and Wylie (2014)
distinguish four modes of engagement: interactive, constructive, active,
and passive (ICAP). In the passive mode of engagement, learners are
seen as the information containers because they only receive informa-
tion. For example, they just passively listen to lectures or view videos.
In contrast, active engagement involves learners' physical manipulation
or action, such as pointing to and underlining the reading materials.
Furthermore, a learner's engagement can be considered constructive
when this individual generates something beyond the presented mate-
rial, such as self-explaining and summarizing. Finally, interactive en-
gagement refers to the situations when constructive learning occurs in a
group of individuals. Chi and Wylie hypothesize that learning outcomes
associated with interaction should be better than learning outcomes
associated with construction, which is better than activeness, which is
superior to passiveness. According to ICAP, merely presenting different
types of visualizations without other learning strategies is a passive
mode of engagement, as learners neither generate products nor ma-
nipulate materials during the learning process. As a result, without the
support of other learning strategies, learning outcomes associated with
animated and static visualizations should have no significant differ-
ences.
The advancement of digital technologies, such as smart phones and
tablets, offers great opportunities for educators to engage learners in
novel and multifaceted ways. This is especially the case as human
movement like gesturing can be applied with the help from these
technologies. A recent empirical study conducted by Agostinho et al.
(2015) revealed promising evidence in this respect. The researchers
asked some primary-school students to trace temperature graphs with
their fingers on an iPad while a remaining group of students simply
read the graphs without any finger tracing. The study's results indicated
that learners who physically traced elements of a graph outperformed
their peers who did not on transfer tests. Although using digital tech-
nologies has shown its potential of enhancing engagement, it must be
noted that providing feedback to learners can also be utilized to further
assist learners in correcting any faults in their mental models as well as
to monitor their learning processes more effectively.
1.3. Biofeedback in learning and instruction
Feedback is the information a teacher, a peer, a parent, a computer-
based environment or other agents provided to improve an individual's
learning and performance (Hattie & Timperley, 2007). Feedback has
been long recognized by scholars as one of the most powerful techni-
ques in learning and instruction, as learners can benefit from it by
confirming or modifying their current mental models (Cohen, 1985).
Results from several early meta-analytical studies provide supporting
evidence for the effectiveness of feedback (Hattie, 1999; Kluger &
DeNisi, 1996). For instance, Kluger and DeNisi meta-analyzed over 600
effect sizes obtained from 131 papers. These papers reported empirical
evidence regarding using feedback as a way of intervention. The re-
searchers found that feedback had an overall positive effect on per-
formance with a moderate effect size. However, over one third of such
L. Lin, M. Li
interventions were found to hinder learner performance, indicating a
large amount of variability in the feedback effect. More specifically,
they found that feedback was most powerful when it was provided to
students to correct their responses and when it was based on previous
changes. Also, the impact of feedback was dependent upon task goals
and task complexity.
In order to ameliorate the ambiguity in the feedback literature, re-
searchers have resorted to investigating specific types of feedback. For
instance, some researchers compared different elaboration levels of
feedback in computer-based learning environments (Lin et al., 2013;
Moreno, 2004). Lin et al. (2013) compared two types of feedback
provided after learners viewed animations on thermodynamics and
responded to practice questions. The feedback was either simple feed-
back that provided verification to students' responses (e.g., “Yes. That's
correct.”) or elaborate feedback that not only included verification but
also detailed instructional explanations. The study revealed that the
effect of feedback depended upon the visual presence of an animated
virtual agent. When the virtual agent was visually present, elaborative
feedback was better at promoting learning when compared with simple
feedback. However, when the virtual agent was not visually present but
only with her voice present, the effect of elaborate feedback was
equivalent to simple feedback. Other aspects of feedback have also been
empirically investigated, such as the timing of feedback (immediate
feedback vs. delayed feedback, Schroth, 1992) and the modality of
feedback (audio feedback vs. visual text feedback, Fiorella, Vogel-
Walcutt, & Schatz, 2012). But most researchers have looked into the
feedback issue from a cognitive perspective, i.e., whether information
provided in the feedback can directly improve learners' knowledge and
correct their misconceptions. Rarely have researchers studied using
feedback to impact learners' meta-cognition and affect. One exception
in this regard is a study conducted by Terzis, Moridis, and Economides
(2012), in which a female virtual agent with empathetic encourage-
ment behaviors was used as the source of emotional feedback. The
study's findings demonstrated that this type of emotional feedback en-
hanced learners' behavioral intention to use a computer-based assess-
ment system.
Based on their theoretical model for feedback, Hattie and Timperley
(2007) argue that not only does effective feedback need to be task-
oriented, but also needs to facilitate learning processes and individuals'
monitoring of their own learning. Technological advances have offered
learners more opportunities to receive feedback via educational tech-
nology (Schaeffer, Margulieux, Chen, & Catrambone, 2016). Some re-
searchers have pointed out the potential benefits of using emerging
brain-computer-interface technology as a cutting-edge tool for giving
learners feedback (Chen & Huang, 2014; Sun & Yeh, 2017). The ratio-
nale behind this claim is that the combination of measurement, col-
lection, analysis, and reporting of the learners' rich data (e.g., brain-
wave data) can offer valuable insights on learning and affect states
(Baker & Yacef, 2009). A couple of empirical case studies in this vein
have been reported in the literature. Amongst them, Sun and Yeh
(2017) utilized mobile EEG headsets to monitor learners' attention and
provided learners audio feedback when they were not paying sufficient
attention to their learning materials. They found that this biofeedback
approach significantly enhanced these learners' attention, even though
the learners' self-efficacy and achievement were equivalent to their
peers who were not provided with such feedback. However, a closer
examination revealed that learners in these studies were only studying
factual knowledge (specifically, knowledge about anti-phishing in Sun
& Yeh, 2017 and foreign language learning in Chen & Huang, 2014) and
that the instructional materials were mostly textual. Consequently,
despite the fact that this type of biofeedback has a good potential for
engaging learners in terms of the enhanced attention, it is unknown
whether the biofeedback has positive effects on learning, cognitive
load, and intrinsic motivation when learners are acquiring procedural
knowledge from visualizations. In this study, we utilized EEG
34
Learning and Instruction 55 (2018) 32–40
biofeedback, which presented an individual's attention and relaxation,
to help learners obtain procedural knowledge.
1.4. Overview of the study
The purpose of this study was to investigate whether two types of
feedback would engage learners in obtaining procedural knowledge
from animated or static visualizations. Specifically, college students
learned to create simulated scenes related to physics by interacting with
a sandbox game. They followed step-by-step tutorials that were pre-
sented in either animated or static format. During that process, their
brain waves were measured and collected through a portable EEG
headset. This headset, along with its connected smartphone and soft-
ware application, transformed the EEG data into waves of attention and
relaxation (see Fig. 1). Learners were able to zoom in and out to view
their own attention and relaxation by tapping, pinching, and stretching
on the screen using their fingers. According to the ICAP framework, this
mode of engagement is active, which we considered as active feedback.
In the current study, we also designed a second feedback condition, in
which learners were prompted to verbally reflect on their learning
processes based on their attention and relaxation from the biofeedback.
The intention was to engage learners in processing their biofeedback on
a deeper level and help them better monitor their own learning. Ac-
cording to the ICAP framework, this mode of engagement is con-
structive and we considered it as constructive feedback. Therefore, in
addition to the type of visualization (animated vs. static), we also ex-
perimentally manipulated the type of feedback that the learners re-
ceived (constructive feedback vs. active feedback vs. no feedback). The
dependent variables involved in the study were intrinsic motivation,
cognitive load, learning time, and learning outcomes.
The research questions addressed in the current study are: (a) Are
the two types of feedback (constructive and active feedback) better than no
feedback to assist learning procedural knowledge? (b) Is constructive feed-
back better than active feedback to assist learning procedural knowledge? (c)
Are the positive effects of animated visualizations dependent upon the type of
feedback? Building on the ICAP engagement framework, we hypothe-
sized that constructive feedback would be better than active feedback,
which would in turn better than no feedback regarding the improved
intrinsic motivation, cognitive load, learning time, and learning out-
comes (Hypothesis 1). Taking into account the research arguing the
existence of moderating factors in the efficacy of animations, as well as
the literature regarding the use of biofeedback in learning, we hy-
pothesized that the use of biofeedback (constructive feedback and ac-
tive feedback) would assist learners in processing animations, leading
to the improved intrinsic motivation, cognitive load, learning time, and
learning outcomes (Hypothesis 2).
2. Method
2.1. Participants & design
The participants of this study consisted of 116 undergraduate and
graduate students from a public university in East China. They were
recruited from the general campus population via flyers and social
media outreach, as well as from the psychology participant pool. The
participating students were from a wide range of disciplines and majors,
including education, psychology, music, business, and history. There
were 30 (26%) males and 86 (74%) females, which represented the
typical student population of the university. The average age of the
participants was 20.31 years old (standard deviation = 2.30 years old).
There were 52 freshmen, 18 sophomore, 13 junior, 19 senior, 12
graduate students, and 2 unspecified. The participants were paid a
small stipend for their participation.
The study used a 2 X 3 between-subjects design: the first factor was
the visualization type (animated vs. static) and the second factor was
L. Lin, M. Li Learning and Instruction 55 (2018) 32–40

Fig. 1. Two sample screenshots of the biofeedback interface. The graph on the left shows an example of the visual without any hand manipulation of the touch screen. The graph on the
right shows an example of the visual after a learner taps the screen. Text was presented in Chinese during the experiment. English translations have been added for the purpose of
understanding.

the type of feedback (constructive feedback vs. active feedback vs. no

feedback). Participants were randomly assigned to one of the six ex-
perimental conditions:

(a) animated visualization with constructive feedback (n = 20);

(b) animated visualization with active feedback (n = 18);
(c) animated visualization with no feedback (n = 20);
(d) static visualization with constructive feedback (n = 20);
(e) static visualization with active feedback (n = 19);
(f) static visualization with no feedback (n = 19).

The participants across the six conditions were comparable in terms

of age (F(5, 110) = .18, p = .97) and gender (χ2(5) = 10.06, p = .08).
None of them had used the Algodoo software before. Moreover, no
participants were majoring in physics.

2.2. Learning environment

The learning environment included a Lenovo laptop computer, an
Apple iPad mini, an Apple iPhone, and a BrainLink portable EEG
headset (see Fig. 2). The laptop computer allowed learners to use a
software program called Algodoo (http://www.algodoo.com/), a
sandbox game, to create simulated physics scenes. Learners could use
their mouse to draw two-dimensional objects (e.g., circles and boxes) in
the Algodoo program and apply gravity, friction, fluid, and other
physics functions to the objects in the program. This allowed the lear-
ners to create playful, cartoony scenes (simulations) that demonstrate
physics principles and phenomena.
The iPad mini was used to present three step-by-step tutorials for
learners to create three physics scenes within Algodoo. An educational
technologist created a 4.5-min-long animation via screen recording plus
his narration of the procedures. The animation was then uploaded to
cloud storage so that it could be easily accessed on the iPad mini using
various applications (e.g., Baidu NetDrive). Twenty screenshots of the
animation were taken as key frames. These key frames were chosen by
the authors based on whether a new object was drawn or a default
Fig. 2. The experiment's set-up.
parameter was changed. They were presented to learners in a
PowerPoint format using a Microsoft PowerPoint application pre-
installed onto the iPad mini. Each static key frame was accompanied
with brief text-based explanations that were transcribed from the audio
narration from the animation tutorial (see Fig. 3).
The BrainLink headset is a commercially developed portable EEG
device with a 512 HZ sampling frequency. It has three electrodes that
are attached on the head to obtain brain waves in humans. Two elec-
trodes are placed on the forehead, approximately one inch above the
eyebrows, while the third electrode is mounted on the device's ear clip,
which needs to be attached to the wearer's left earlobe (see Fig. 2). This
portable headset must be connected to an iPhone via Bluetooth.
The BrainLink iOS application was used in the study so that the
headset would record each individual participant's brain waves, as well
as learning time. By using the eSense algorithm developed by

35
L. Lin, M. Li Learning and Instruction 55 (2018) 32–40

Table 1
Intrinsic motivation items.

Item Subscale

I really try my best in the learning tasks. Effort

For other similar tasks, I could work much better than I do now. Effort
(R)
I think that I rarely do my best in the learning tasks. (R) Effort
I tried very hard in learning the content. Effort
I think I am pretty competent at the learning tasks. Competence
Other people are better at learning than me. (R) Competence
I am usually slower at digesting the contents than other people. Competence
(R)
I fear that my performance on the tasks is bad. (R) Competence
I think I am good at learning. Competence
I can keep up well with the pace of the learning. Competence
I find it difficult to keep my mind on learning during the whole Attentiveness
task. (R)
I am often thinking about things that have nothing to do with the Attentiveness
learning task. (R)
I am easily distracted in class. (R) Attentiveness
Fig. 3. A sample screenshot of the tutorial. Text was presented in Chinese during the I can easily keep my attention on the work during the whole Attentiveness
experiment. English translations have been added for the purpose of understanding. learning task.
I enjoyed the learning tasks. Interest
I was really interested in the learning task. Interest
NeuroSky, the data were transformed and visually presented to learners I thought this was a boring activity. (R) Interest
on the iPhone screen, represented as waves of attention and relaxation I enjoyed doing this activity very much. Interest
I think I learn useful things in the task. Value
(Sun & Yeh, 2017; Sun, 2014). After the participant finished a simulated
To me, many things I have to learn in the learning task are Value
scene, the experimenters stopped the recording on the application and unimportant. (R)
showed the biofeedback information to the participant (see Fig. 1). I think that I have to learn things in the task that I won't ever need Value
in the future. (R)
I was very relaxed in doing these. (R) Pressure
2.3. Measures & instruments I felt very tense while doing this activity. Pressure
I felt pressured while doing these. Pressure
I did not feel nervous at all while doing this. (R) Pressure
Learners' intrinsic motivation was measured using a 7-point Likert
scale, ranging from 1 (“not at all true”) to 7 (“very true”). There were Note. R indicates reversely coded.
25 statements, assessing intrinsic motivation with six subscales: effort,
competence, attentiveness, interest, value, and pressure (see Table 1). paper and an individual who was not aware of the research design or
They were adapted from previous research (Van Damme, Fraine, purpose, independently coded 50% of the learning products and
Landeghem, Opdenakker, & Onghena, 2002; McAuley, Duncan, & reached a high level of agreement (98% agreement). Thereafter, the
Tammen, 1989; Ryan, 1982). Negatively worded items were reverse- first author coded the remaining products.
coded so that higher scores reflect more positive motivation. The The demographic information for each individual participant was
Cronbach's alphas for effort, competence, attentiveness, interest, value, also collected via a survey. Specifically, participants were asked for
and pressure were .83, .83, .76, .82, .78, and .80, respectively. their gender, age, academic status, major, and whether they had used
Learners’ perceived cognitive load was measured by the difficulty the Algodoo software before.
item (“It is easy for me to learn the materials and complete the tasks.”),
which was adapted from Paas, Tuovinen, Tabbers, and Van Gerven
(2003) and successfully implemented in the previous research (Lin 2.4. Procedure
et al., 2016, 2017). Like the intrinsic motivation items, this cognitive
load item was implemented on a 7-point Likert scale ranging from 1 The study was conducted in a controlled laboratory setting. First,
(“not at all true”) to 7 (“very true”). each individual participant signed a consent form and was then seated
The time taken for each individual to complete three physics scenes in front of a laptop computer. Next, an experimenter informed the
(i.e., the time taken to complete the first, second, and third scene, re- participant regarding the general goal and procedures of the experi-
spectively) was recorded by the BrainLink iOS application, and was ment. However, participants were left unaware of the experimental
visually presented to each learner thereafter (see Fig. 1). As mentioned conditions and the research questions. Thereafter, the participant
above, in order to show the EEG biofeedback to learners after they would complete the paper-based demographic survey, offered with no
completed each scene, the experimenters were required to stop re- time limit, and was randomly assigned to a condition using an experi-
cording in the BrainLink iOS application. Consequently, the time for ment ID number. Before starting to learn the tutorials and creating
learners to interact with the biofeedback (i.e., viewing their own at- scenes within Algodoo, the individual would be required to put on the
tention and relaxation by hand gestures) and to verbally reflect on their BrainLink headset, with the assistance of the experimenters. The ex-
learning processes was not recorded. The experimenters transformed perimenters would notify the participant to proceed only after the
these lengths of time into seconds and manually recorded these data hardware (the headset) and the software application was successfully
into an Excel spreadsheet for all participants. paired.
Learning outcomes were assessed by coding the products of For the next step, participants were provided with an iPad mini to
learning, which were those physics scenes created by participants using watch the tutorials and follow the instructions in order to create three
the Algodoo software. The experimenters developed a coding scheme scenes within the Algodoo program on the provided laptop computer.
where one point was awarded for each correctly drawn object or each The tutorials were presented to the participants in either animated or
correctly changed parameter in identical accordance to the criteria for static format, according to the experiment ID randomly assigned to each
creating the key static frames. Therefore, a maximum of 20 points could participant. There were no time limits enforced for learning from the
be attained for each participant. Two coders, the first author of the tutorials and creating the scenes. In addition, both animated and static

36
L. Lin, M. Li
tutorials were learner-controlled. That is, participants could pause, re-
wind, or restart the visuals according to their own leaning needs.
For the Active Feedback conditions, the experimenters presented
each participant's attention and relaxation during learning on the
iPhone screen after the participant completed each scene (see Fig. 1).
The participant was permitted 3 min to view and interact with this
biofeedback. Participants could use their fingers to zoom in and out to
better view and comprehend the biofeedback. They could additionally
tap the screen to read the explanatory texts (e.g., “This is the trend
graph regarding your relaxation and attention, generated by BrainLink
based on your raw EEG waves.”).
For the Constructive Feedback conditions, not only were the parti-
cipants engaged in interacting with the biofeedback, but were ad-
ditionally prompted to verbally reflect on their learning processes by
utilizing the biofeedback information. Two sample prompts were:
“Could you reflect on your level of attention, according to the in-
formation presented on the iPhone?” and “By looking at the informa-
tion presented on the iPhone, does anything comes to your mind?” The
prompts used in this study were adapted from previous research (Lin
et al., 2016; Chi et al., 2001) and were verbally communicated to
participants by the experimenters. The constructive feedback process
was also limited to 3 min. A small-scale pilot study was conducted with
eight participants from the same population to determine this period of
time as suitable.
For the No Feedback condition, participants received no feedback
after completing a scene, and the experimenters simply let the parti-
cipants move on to the next tutorial thereafter. However, in order to
make this condition equivalent to the other two feedback conditions,
the participants similarly wore a BrainLink headset during the entire
learning process.
For all six experimental conditions, the experimenters recorded the
time that each participant spent completing a given scene. Note that the
3-min feedback time was not included in this. After completing all three
scenes, the experimenter reminded the participants to save their work
on the laptop computer and directed them to a weblink thereafter,
where they were asked to respond to a 26-item survey that assessed
their intrinsic motivation and cognitive load. Finally, the experimenters
removed the headset from the participants, thanked them for their
participation, and paid them 15 RMB in cash each. It would take ap-
proximately 40–60 min for an individual participant to complete the
entire experiment.
3. Results
Family-wise type I error rate was set at the .05 level. We used partial
η2 or Cohen's d as the effect size index. Accordingly, .01, .06, and .14
are considered as the η2 values for small, medium, and large effect sizes,
respectively. And .20, .50, and .80 are considered as the d values for
small, medium, and large effect sizes (Cohen, 1988). Mean (M) and
standard deviation (SD) are presented in Table 2.
3.1. Intrinsic motivation
Six two-way between-subjects analysis of variance (ANOVAs) were
conducted to evaluate the potential effects of visualization type (ani-
mated vs. static) and feedback type (constructive vs. active vs. no
feedback) on learners’ perceived effort, competence, attentiveness, in-
terest, value, and pressure.
For effort, the results indicated that neither the two main effects,
nor the interaction between the two factors, was significant, all
Fs < 1.00. We found the same patterns for competence (all
Fs < 1.00), interest (all Fs < 1.00), and pressure (all Fs < 1.11).
The results of ANOVA revealed a significant main effect of feedback
on attentiveness, F(2, 110) = 3.19, MSE = 1.19, p = .05, partial
η2 = .06 (medium effect). Three pairwise comparisons were conducted
37
Learning and Instruction 55 (2018) 32–40
as post hoc analysis. The Holm's sequential Bonferroni procedure was
used to control for type I error (Abdi, 2010; Holm, 1979). At the alpha
level of .017 (.05/3), we found that learners who received no feedback
(M = 5.94, SD = .89) had significantly higher ratings on attentiveness
than their peers who received constructive feedback (M = 5.32,
SD = 1.11), p = .01, Cohen's d = .62 (medium-to-large effect). The
results of the other two comparisons, i.e., constructive feedback vs.
active feedback and active feedback vs. no feedback, were found to be
non-significant, both ps > .20. In addition, neither the main effect of
visualization type nor the interaction was significant, both Fs < 1.00.
For learners' perceived value, the results of ANOVA also revealed a
significant main effect of feedback, F(2, 110) = 3.47, MSE = 1.07,
p = .03, partial η2 = .06 (medium effect). Three pairwise comparisons
were conducted as post hoc analysis. The Holm's sequential Bonferroni
procedure was used to control for type I error. We found, at the alpha
level of .017 (.05/3), that the difference on the ratings of value between
the learners who received no feedback (M = 5.49, SD = 1.14) and
those who received constructive feedback (M = 4.94, SD = 1.05) was
marginally significant, p = .02, Cohen's d = .50 (medium effect). At the
alpha level of .025 (.05/2), we also found a marginal significant dif-
ference between learners who received constructive feedback and those
who received active feedback (M = 5.47, SD = .88), p = .03, Cohen's
d = .55 (medium effect). Nevertheless, there was no significant differ-
ence between the no feedback condition and the active feedback con-
dition, p = .92. In summary, learners who received no feedback and
active feedback had perceived their learning to be more valuable than
those who received constructive feedback.
In order to reveal whether feedback impacted intrinsic motivation,
we also conducted complex mean comparisons (constructive feed-
back + active feedback vs. no feedback) on learners' perceived effort,
competence, attentiveness, interest, value, and pressure. For individuals
who learned from the static visuals, the results demonstrated that those
who received no feedback had significantly higher ratings on atten-
tiveness than their peers who received feedback, mean differ-
ence = 1.35, t(55) = 2.26, p = .03, Cohen's d = .63 (medium-to-large
effect). Differences on effort, competence, interest, value, and pressure
were non-significant, all ps > .35. For individuals who learned from
the animated visuals, all results of such complex mean comparisons
were non-significant, all ps > .13.
3.2. Cognitive load
A two-way between-subjects ANOVA was conducted to assess
whether visualization type (animated vs. static) and feedback type
(constructive vs. active vs. no feedback) had any effects on the per-
ceived difficulty. Neither the visualization type main effect nor the
feedback type main effect was significant: for the visualization type
main effect, F(1, 110) = 3.19, MSE = 1.27, p = .18, partial η2 = .02;
for the feedback type main effect, F < 1.00. But there was a significant
interaction, F(2, 110) = 3.19, p = .05, partial η2 = .05 (medium effect).
To further clarify this significant interaction, simple main effect ana-
lysis was conducted. The Holm's sequential Bonferroni procedure was
used to control for type I error. In the two constructive feedback con-
ditions, learners who learned from animated visualizations (M = 6.10,
SD = 1.07) were found to perceive their learning tasks easier than their
peers who learned from static visualizations (M = 5.10, SD = 1.33),
p = .006, Cohen's d = .83 (large effect). The remaining two pairwise
comparisons were non-significant, both ps > .63. For the simple main
effects of feedback type, all pairwise comparisons were non-significant,
all ps > .08. We also conducted complex mean comparisons (con-
structive feedback + active feedback vs. no feedback) for learners who
learned from animated visualizations and static visualizations.
However, the results of these comparisons were non-significant, all
ps > .37.
L. Lin, M. Li Learning and Instruction 55 (2018) 32–40

Table 2
Descriptive statistics of intrinsic motivation, cognitive load, learning time, and learning outcomes.

Animated Visualizations Static Visualizations

Constructive Feedback Active Feedback No Feedback Constructive Feedback Active Feedback No Feedback
(n = 20) (n = 18) (n = 20) (n = 20) (n = 19) (n = 19)

M (SD) M (SD) M (SD) M (SD) M (SD) M (SD)

Effort 4.98 (.91) 4.94 (.88) 5.01 (.56) 5.00 (.82) 4.84 (.88) 5.13 (.69)
Competence 4.87 (.86) 4.70 (1.09) 4.85 (.82) 4.98 (.87) 4.95 (.89) 4.88 (.95)
Attentiveness 5.41 (1.09) 5.71 (1.25) 5.83 (1.00) 5.23 (1.15) 5.53 (1.24) 6.05 (.76)
Interest 5.86 (.93) 6.06 (.98) 6.03 (.90) 6.00 (.79) 6.22 (.66) 6.00 (.67)
Value 4.88 (1.27) 5.24 (.84) 5.53 (1.14) 5.00 (.77) 5.68 (.89) 5.44 (1.18)
Pressure 3.34 (1.40) 2.94 (1.23) 3.05 (.98) 3.19 (1.17) 2.78 (1.06) 3.17 (1.24)
Difficulty 6.10 (1.07) 5.56 (1.15) 5.55 (1.15) 5.10 (1.33) 5.74 (1.10) 5.53 (.91)
Time 1a 296.90 (114.87) 311.89 (117.78) 295.75 (145.44) 277.00 (120.42) 306.05 (169.55) 347.68 (142.72)
Time 2b 169.35 (58.47) 175.89 (50.93) 183.00 (89.63) 171.30 (51.07) 196.00 (74.03) 203.00 (48.82)
Time 3c 243.25 (74.57) 248.39 (89.23) 255.00 (66.11) 216.55 (48.44) 244.58 (66.35) 234.16 (68.70)
Learning Outcome 19.90 (.31) 19.94 (.24) 20.00 (.00) 20.00 (.00) 19.89 (.46) 20.00 (.00)

Note. M = Mean. SD = Standard Deviation. The unit of time is second.

a
Time taken to complete the first scene.
b
Time taken to complete the second scene.
c
Time taken to complete the third scene.

3.3. Learning time
A 2 (visualization type: animated vs. static) X 3 (feedback type:
constructive vs. active vs. no feedback) X 3 (learning task: Scene 1 vs.
Scene 2 vs. Scene 3) mixed ANOVA was conducted with the first two
variables being the between-subjects factors and the third variable
being the within-subjects factor. There was a significant main effect of
learning task: Wilks' Λ = .46, F (2, 109) = 63.72, p < .001, partial
η2 = .54. Using the Holm's sequential Bonferroni procedure to control
for type I error, post hoc analysis revealed that (a) learners spent sig-
nificantly longer time in completing the first scene (M = 305.88 s,
SD = 135.20) than the second scene (M = 183.09 s, SD = 64.00),
p < .001, Cohen's d = 1.16, (b) learners spent significantly longer time
in completing the first scene than the third scene (M = 240.32 s,
SD = 69.21), p < .001, Cohen's d = .61, and (c) learners spent sig-
nificantly longer time in completing the third scene than the second
scene, p < .001, Cohen's d = .86. The visualization type main effect
was non-significant, F (1, 110) = 1.45, p = .24, partial η2 = .03. The
feedback type main effect was also non-significant, F < 1.00. All two-
way interactions, as well as the three-way interaction, were non-sig-
nificant, all Fs < 1.00.
3.4. Learning outcomes
The descriptive statistics indicated that almost all participants suc-
cessfully created three scenes using the Algodoo software program.
Subsequently, a two-way between-subjects ANOVA was conducted to
assess whether visualization type (animated vs. static) and feedback
type (constructive vs. active vs. no feedback) had effects on learning
outcomes. The results revealed that neither the visualization type main
effect nor the feedback type main effect was significant: for the visua-
lization type main effect, F < 1.00; for the feedback type main effect, F
(2, 110) = 1.06, MSE = .06, p = .35, partial η2 = .02. Also, the inter-
action was non-significant, F < 1.00. We also conducted complex
mean comparisons (constructive feedback + active feedback vs. no
feedback). However, the results of these comparisons were non-sig-
nificant, all ps > .17.
4. Discussion & conclusion
In the current study, a portable EEG headset and its accompanying
software were used to collect the brain wave data of learners and vi-
sually present the data as waves of attention and relaxation, so that
learners could use this biofeedback to monitor their own learning
processes. Learners viewed and comprehended the biofeedback with
hand gestures (active feedback), or verbally reflected on their learning
processes by utilizing the biofeedback (constructive feedback). The
purpose of this study was to investigate whether biofeedback would
help learners engage in acquiring procedural knowledge from animated
or static visualizations. Specifically, the study addressed three research
questions: (a) Are the two types of feedback (constructive and active
feedback) better than no feedback to assist learning procedural knowledge?
(b) Is constructive feedback better than active feedback to assist learning
procedural knowledge? (c) Are the positive effects of animated visualizations
dependent upon the type of feedback?
According to the ICAP framework (Chi & Wylie, 2014) and the lit-
erature related to learning from animations (Berney & Betrancourt,
2016; Castro-Alonso, Ayres, & Paas, 2016), we hypothesized that, in
terms of intrinsic motivation, cognitive load, learning time, and
learning outcomes, constructive feedback would be better than active
feedback, which would in turn better than no feedback (Hypothesis 1)
and the use of biofeedback would assist learners in processing anima-
tions, leading to the improved intrinsic motivation, cognitive load,
learning time, and learning outcomes (Hypothesis 2). The results,
which addressed the research questions, did not support Hypothesis 1
but partially supported Hypothesis 2.
First, regarding the different types of biofeedback, we found that
learners who received no feedback had significantly higher levels of
perceived attentiveness than their peers who received constructive
feedback. Moreover, learners who studied static visuals and received
feedback had significantly lower levels of attentiveness than their peers
who studied the same type of visuals but did not receive feedback. Both
of these results indicate that learners' attention can be more focused if
they receive no feedback regarding their attention and relaxation after
completing a scene within the software program. These findings diverge
not only from Hypothesis 1, but also from the few case studies showing
supporting evidence of using the biofeedback to enhance attention
(e.g., Sun & Yeh, 2017). One possible explanation for this divergence is
that learners’ attention may be divided after receiving the biofeedback
that showed their distributions of attention and relaxation. For the
constructive feedback conditions in particular, the verbal reflection
activity may have positive effects to engage learners in making sense of
their biofeedback, which should be helpful for them to monitor their
learning processes. However, this same activity may also be disruptive
for learners in reaching their original learning goal. As a result, learners
may allocate some of their limited cognitive resources on monitoring

38
L. Lin, M. Li
their attention and relaxation, thereby becoming less focused on the
learning tasks per se.
Second, our findings with respect to learners' perceived value re-
vealed similar patterns. Learners who received no feedback perceived
learning more valuable than those who received constructive feedback.
Moreover, those who received active feedback also considered learning
more valuable than their peers who received constructive feedback.
These results again suggest that prompting learners to monitor their
own learning processes by utilizing the biofeedback that shows their
attention and relaxation may prevent them from reaching their original
learning goals. This in turn leads to their lower perceived value in
learning. Taking the results on perceived attentiveness and perceived
value together, there may be another possible explanation, which is the
visual modality of the biofeedback used in the current study. Learners
in the constructive feedback condition may experience a high level of
cognitive load during the process of verbal reflection, as not only do
they need to reflect on their learning experiences by mentally holding
information obtained from the visualizations, but must also process
external visual biofeedback information. Unfortunately, we did not
measure learners’ cognitive load immediately after they received
feedback, which was one of the limitations of our study. Researchers in
the future may consider further improving cognitive load measurement.
In addition, future research could focus on the relative efficacy of this
type of feedback, which makes use of the brain-computer-interface
technology, compared to other types of feedback, such as feedback
provided by an animated virtual agent (Lin et al., 2013), or other in-
structional strategies.
The findings of the current study partially support Hypothesis 2,
that the use of biofeedback would assist learners to process animations,
leading to improvement in intrinsic motivation, cognitive load, learning
time, and learning outcomes. Specifically, if feedback takes the form of
prompting learners to construct their own understanding about their
learning processes (i.e., meta-cognition), instructional materials pre-
sented in the animated format can reduce learners’ perceived difficulty
than materials presented in the static format. From the engagement
perspective, animated visualizations have great potential relative to
static visualizations if learners are deeply engaged, such as using the
EEG biofeedback accompanied with verbal reflection that was im-
plemented in the current study. The possible underlying cognitive
mechanisms are that deeply engaged learners may notice more subtle
details that are only available in animated visualizations but not in
static visualizations. Consequently, the learners perceive learning tasks
easier than their peers, thereby possibly utilizing fewer cognitive re-
sources to yield equivalent learning outcomes. From that perspective,
the biofeedback designed to engage learners may potentially positively
impact learning efficiency (Hoffman & Schraw, 2010). However, a
ceiling effect of the learning outcome measures is also possible. Future
research could consider replicating the findings using other learning
tasks.
The findings of this study have both theoretical and practical im-
plications. Theoretically, they open many doors for researchers to apply
EEG biofeedback to engage learners in processing visualizations.
Although biofeedback, such as eye movement, has been empirically
investigated by a few researchers (e.g., van Gog, Paas, van Merrienboer,
& Witte, 2005), biofeedback based on modern brain-computer-interface
technology has not been applied in research on dynamic visualizations
in learning. Moreover, our findings imply that feedback, one of the most
powerful instructional techniques in learning and instruction, is a po-
tential moderator for the effects of animations, which past research on
learning and instruction has not considered. In terms of practical im-
plications, our findings indicate that feedback is not always beneficial
for learning, especially when it is offered to help learners monitor their
own learning and cognition. Instructional designers should carefully
design the biofeedback so that it not only helps learners monitor their
own learning, but also minimizes the possible distractions it may pre-
sent to learners. When the learning goal is to obtain procedural
39
Learning and Instruction 55 (2018) 32–40
knowledge, one approach for enhancing the efficacy of biofeedback and
learner engagement is to utilize animated visualizations.
In summary, the findings of our study indicate that the effects of
animations on acquiring procedural knowledge partially depend upon
the biofeedback that engages learners in verbal reflection on their
learning processes. However, the biofeedback with the intention to
engage learners negatively impacted some aspects of intrinsic motiva-
tion. Therefore, this type of feedback, which utilizes advanced brain-
computer-interface technology, needs to be carefully designed in order
to minimize possible interruptions to learning.
Acknowledgments
This research was supported by Shanghai Planning Office of
Philosophy and Social Science (2014JJY001) awarded to Lijia Lin.
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://dx.
doi.org/10.1016/j.learninstruc.2018.02.005.
References
Abdi, H. (2010). Holm's sequential Bonferroni procedure. In N. Salkind (Ed.). Encyclopedia
of research design (pp. 574–578). Thousand Oaks, VA: Sage Publications.
Agostinho, S., Tindall-Ford, S., Ginns, P., Howard, S. J., Leahy, W., & Paas, F. (2015).
Giving learning a helping hand: Finger tracing of temperature graphs on an iPad.
Educational Psychology Review, 27, 427–443.
Baker, R. S. J., & Yacef, K. (2009). The state of educational data mining in 2009: A review
and future visions. Journal of Educational Data Mining, 1(1), 3–16.
Berney, S., & Betrancourt, M. (2016). Does animation enhance learning? A meta-analysis.
Computers & Education, 101, 150–167.
Bétrancourt, M., & Tversky, B. (2000). Effect of computer animation on users' perfor-
mance: A review. Travail Humain, Le, 63(4), 311–329.
Blumenfeld, P. C., Kempler, T. M., & Krajcik, J. S. (2004). Motivation and cognitive en-
gagement in learning environments. In R. K. Sawyer (Ed.). The cambridge handbook of
the learning sciences (pp. 475–488). New York, NY: Cambridge University Press.
Castro-Alonso, J. C., Ayres, P., & Paas, F. (2014). Learning from observing hands in static
and animated versions of non-manipulative tasks. Learning and Instruction, 34, 11–21.
Castro-Alonso, J. C., Ayres, P., & Paas, F. (2015). Animations showing Lego manipulative
tasks: Three potential moderators of effectiveness. Computers & Education, 85, 1–13.
Castro-Alonso, J. C., Ayres, P., & Paas, F. (2016). Comparing apples and oranges? A
critical look at research on learning from statics versus animations. Computers &
Education, 102, 234–243.
Chen, C.-M., & Huang, S.-H. (2014). Web-based reading annotation system with an at-
tention-based self-regulated learning mechanism for promoting reading performance.
British Journal of Educational Technology, 45(5), 959–980.
Chi, M. T. H., & Wylie, R. (2014). The ICAP framework: Linking cognitive engagement to
active learning outcomes. Educational Psychologist, 49(4), 219–243.
Chi, M. T. H., Siler, S. A., Jeong, H., Yamauchi, T., & Hausmann, R. G. (2001). Learning
from human tutoring. Cognit. Sci. 25(4), 471–533.
Cohen, V. B. (1985). A reexamination of feedback in computer-based instruction:
Implications for instructional design. Educational Technology, 25(1), 33–37.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale:
N.J.: L. Erlbaum Associates.
Fiorella, L., Vogel-Walcutt, J. J., & Schatz, S. (2012). Applying the modality principle to
real-time feedback and the acquisition of higher-order cognitive skills. Educational
Technology Research & Development, 60, 223–238.
Fredicks, J. A., Blumenfeld, P. C., & Paris, A. H. (2004). School engagement: Potential of
the concept, state of the evidence. Review of Educational Research, 74, 59–109.
Hattie, J. A. (1999). Influences on student learning (Inaugural professorial address, University
of Auckland, New Zealand). Retrieved from https://www.researchgate.net/
publication/237248564_Influences_on_Student_Learning [June 3, 2017] .
Hattie, J., & Timperley, H. (2007). The power of feedback. Review of Educational Research,
77, 81–112.
Hegarty, M. (1992). Mental Animation: Inferring motion from static displays of me-
chanical systems. Journal of Experimental Psychology: Learning, Memory, and Cognition,
18(5), 1084–1102.
Höffler, T. N., & Leutner, D. (2007). Instructional animation versus static pictures: A
meta-analysis. Learning and Instruction, 17(6), 722–738.
Hoffman, B., & Schraw, G. (2010). Conceptions of efficiency: Applications in learning and
problem solving. Educational Psychologist, 45(1), 1–14.
Holm, S. (1979). A simple sequentially rejective multiple test procedure. Scandinavian
Journal of Statistics, 6(2), 65–70.
Kluger, A. N., & DeNisi, A. (1996). Effects of feedback intervention on performance: A
historical review, a meta-analysis, and a preliminary feedback intervention theory.
Psychological Bulletin, 119, 254–284.
De Koning, B. B., & Tabbers, H. K. (2011). Facilitating understanding of movements in
dynamic visualizations: An embodied perspective. Educational Psychology Review, 23,
L. Lin, M. Li
501–521.
De Koning, B. B., & Tabbers, H. K. (2013). Gestures in instructional animations: A helping
hand to understanding non-human movements? Applied Cognitive Psychology, 27,
683–689.
Lin, L., Atkinson, R. K., Christopherson, R. M., Joseph, S. S., & Harrison, C. J. (2013).
Animated agents and learning: does the type of verbal feedback they provide matter?
Comput. Educ. 67, 239–249.
Lin, L., Atkinson, R. K., Savenye, W. C., & Nelson, B. C. (2016). Effects of visual cues and
self-explanation prompts: empirical evidence in a multimedia environment. Interact.
Learn. Environ. 24, 799–813.
Lin, L., Lee, C. H., Kalyuga, S., Wang, Y., Guan, S., & Wu, H. (2017). The effect of learner-
generated drawing and imagination in comprehending a science text. J. Exp. Educ.
85(1), 142–154.
Lowe, R. K., & Boucheix, J.-M. (2017). A composition approach to design of educational
animations. In R. Lowe, & R. Ploetzner (Eds.). Learning from dynamic Visualizations:
Innovations in research and application (pp. 5–30). Berlin: Springer.
McAuley, E., Duncan, T., & Tammen, V. V. (1989). Psychometric properties of the in-
trinsic motivation inventory in a competitive sport setting: A confirmatory factor
analysis. Research Quarterly for Exercise & Sport, 60, 48–58.
Moreno, R. (2004). Decreasing cognitive load for novice students: Effects of explanatory
versus corrective feedback in discovery-based multimedia. Instructional Science,
32(1), 99–113.
Paas, F., & Sweller, J. (2012). An evolutionary upgrade of cognitive load theory: Using the
human motor system and collaboration to support the learning of complex cognitive
tasks. Educational Psychology Review, 24, 27–45.
Paas, F., Tuovinen, J., Tabbers, H., & Van Gerven, P. (2003). Cognitive load measurement
as a means to advance cognitive load theory. Educational Psychologist, 38, 63–71.
Ryan, R. M. (1982). Control and information in the intrapersonal sphere: An extension of
cognitive evaluation theory. Journal of Personality and Social Psychology, 43, 450–461.
Schaeffer, L. M., Margulieux, L. M., Chen, D.-W., & Catrambone, R. (2016). Feedback via
educational technology. In L. Lin, & R. K. Atkinson (Eds.). Educational Technologies:
Challenges, applications and learning outcomes (pp. 59–72). New York, NY: Nova
Science.
40
Learning and Instruction 55 (2018) 32–40
Schroth, M. L. (1992). The effects of delay of feedback on a delayed concept formation
transfer task. Contemporary Educational Psychology, 17, 78–82.
Stebner, F., Kühl, T., Hoffler, T., Wirth, J., & Ayres, P. (2017). The role of process in-
formation in narrations while learning with animations and static pictures. Computers
& Education, 104, 34–48.
Sun, J. C. Y. (2014). Influence of polling technologies on student engagement: An analysis
of student motivation, academic performance, and brainwave data. Computers &
Education, 72, 80–89.
Sun, J. C.-Y., & Yeh, K. P. C. (2017). The effects of attention monitoring with EEG bio-
feedback on university students' attention and self-efficacy: The case of anti-phishing
instructional materials. Computers & Education, 106, 73–82.
Terzis, V., Moridis, C. N., & Economides, A. A. (2012). The effect of emotional feedback
on behavioral intention to use computer based assessment. Computers & Education,
59, 710–721.
Turkay, S. (2016). The effects of whiteboard animations on retention and subjective ex-
periences when learning advanced physics topics. Computers & Education, 98,
102–114.
Van Damme, J., de Fraine, B., van Landeghem, G., Opdenakker, M.-C., & Onghena, P.
(2002). A new study on educational effectiveness in secondary schools in Flanders:
An introduction. School Effectiveness and School Improvement, 13(4), 383–397.
Van Gog, T., Paas, F., Marcus, N., Ayres, P., & Sweller, J. (2009). The mirror neuron
system and observational learning: Implications for the effectiveness of dynamic vi-
sualizations. Educational Psychology Review, 21, 21–30.
Van Gog, T., Paas, F., van Merrienboer, J. J. G., & Witte, P. (2005). Uncovering the
problem-solving process: Cued retrospective reporting versus concurrent and retro-
spective reporting. Journal of Experimental Psychology: Applied, 11(4), 237–244.
Wagner, I., & Schnotz, W. (2017). Learning from static and dynamic visualizations: What
kind of questions should we ask? In R. Lowe, & R. Ploetzner (Eds.). Learning from
dynamic Visualizations: Innovations in research and application (pp. 69–91). Berlin:
Springer.
Wong, A., Leahy, W., Marcus, N., & Sweller, J. (2012). Cognitive load theory, the tran-
sient information effect and e-learning. Learning and Instruction, 22, 449–457.