Sunteți pe pagina 1din 10

International Journal of Software Engineering and Its Applications

Vol. 9, No. 10 (2015), pp. 169-178


http://dx.doi.org/10.14257/ijseia.2015.9.10.17

Cognitive Load Changes in Pre-Service Teachers with


Computational Thinking Education1

Se-Young Park, Ki-Sang Song* and Soon-Hwa Kim


Korea National University of Education
nubo30@gmail.com, (*corresponding author)kssong@knue.ac.kr, soona6570@
gmail.com

Abstract
Since the emphasis of utilizing ICT in education moves from ICT to Computing in many
countries, curriculums adapt the Computational Thinking (CT) teaching as a basic
courses not only for primary school but also to CS education. Even though the depth of
CT teaching may be different, we may expect similar impact of CT teaching results to
student problem solving skills. With this rationale, we opened designed CT based
education program for freshman introductory of computer application course. Developed
course consists of 10 class hours program and the effectiveness of designed programs
have been tested, firstly for improving CT capabilities of learners in problem solving
environment, and secondarily the cognitive load changes appeared in the problem solving
process have been measured with direct brain wave measurement of EEG while solving
problems. The test results showed that designed CT program affected to the improvement
of problem solving skill, and also show the tendency of decrease in cognitive overload in
the experimental group.

Keywords: Computational thinking, Problem solving, Cognitive load, EEG, Theta/Beta


Ratio

1. Introduction
With an emphasis towards transitioning from Information and Communication
Technology (ICT) in education to Computing, many countries have started to adapt
the Computational Thinking (CT) skill as a basic course for primary school as well
as higher education [1, 2].
Computing competencies have become a major skill needed to carry out logical and
procedural thinking necessary to solve complex problems in various academic and real
life situations. Similarly, the computational thinking concepts were introduced by Janet
Wing in her seminal article as CT represents a universally applicable attitude and skill set
everyone, not just computer scientist, would be eager to learn and use[3]. Wing noted
that CT can be acknowledged in basic literacy terms as the 3Rs (Read, wRite,
aRithmetic) [3].
Even though CT is popular in computer science education and introducing computers
to schools, there are few researches that address the computational practice aspect of CT
that is applicable to problem-solving environments [4]. Therefore, in this paper, we want
to show the impact of a CT based curriculum on students problem solving competencies
and monitor cognitive load changes in problem solving environments in higher education
settings.
For higher education settings, students were asked to take the Scratch
programming as the common learning subject; Scratch programming with CT based
1Some part of this paper has been presented as a paper entitled EEG Analysis for Computational Thinking
based Education Effect on the Learners Cognitive Load at the Applied Computer and Applied
Computational Science (ACACOS '15), in Kuala Lumpur, Malaysia on April 23-25, 2015.

ISSN: 1738-9984 IJSEIA


Copyright 2015 SERSC
International Journal of Software Engineering and Its Applications
Vol. 9, No. 10 (2015)

education. To measure the effectiveness of the teachings, both experimental and


control groups were administered a traditional paper-and pencil CT competencies
test. We also tried to directly measure cognitive load changes through observing
students acquiring computational concepts in a problem solving process [4] with
EEG measurement.

2. Theoretical Background
2.1. Computational Thinking (CT)
As Wing [3] mentioned, CT is defined as the thought processes involved in
formulating problems and finding the solutions to be carried out by an information-
processing agent. Therefore, many researches stress that introducing CT in education may
be beneficial to learners in terms of developing learners' ability to think and promote the
strategies to put their knowledge to practical use [5].
CT skills not only help learners choose and utilize appropriate tools and strategies for
problems solving [2], but also can be used to establish a suitable strategy for solving the
problem. Such strategies may involve the use of suitable automated algorithms to solve
the problem utilizing a computing system. Therefore, if one has CT skills, he/she is
expected to improve their efficiency of solving problems across a variety of disciplines in
real-life settings.
Although the definition of computational thinking remains ambiguous due to its non-
dominant discourse reigns, CT education is effective in systematically improving the
following capabilities as shown in Table 1.

Table 1. Concepts of CT (Source: ISTE, CSTA, NSF [6])


Concept Definition
Data Collection The process of gathering appropriate information
Making sense of data, finding patterns, and drawing
Data Analysis
conclusions
Depicting and organizing data in appropriate graphs,
Data Representation
charts, words, or images
Problem
Breaking down tasks into smaller, manageable parts
Decomposition
Abstraction Reducing complexity to define main idea
Algorithms & Series of ordered steps taken to solve a problem or achieve
Procedures some end
Automation Having computers or machines do repetitive or tedious
tasks.
Simulation Representation or model of a process. Simulation also
involves running experiments using models.
Parallelization Organize resources to simultaneously carry out tasks to
reach a common goal.

From these 9 concepts of CT skills of Table 1, we can create connections between


three domains of computational thinking of Scratch based programming [7] and Barrs
CT concepts as following Table 2 [8]. From both Table 1 and 2, it is clear that using
Scratch programming may improve CT skills in many aspects of CT definitions from
ISTE, CSTA, NSF [8]. This assumption is justified due to the fact that the problem
solving is usually approached using Means-Ends Analysis [9], and novices are
encouraged to focus on an initial state rather than a desired state in the Scratch
programming. Since the Scratch has the characteristics of easiness to check results even

170 Copyright 2015 SERSC


International Journal of Software Engineering and Its Applications
Vol. 9, No. 10 (2015)

the novice programmer does not need to consider the desired state at the initial learning
stage.

Table 2. CT Concepts from ISTE, CSTA, NSF, and Scratch Program [8]
ISTE, CSTA, NSF Scratch program based CT Concepts
Concepts Descriptions
- Data Collection
Computational Concepts that programmers use
- Data Analysis
concepts (Ex. Variables)
- Data Representation
- Algorithms &
Procedures Problem-solving practices that occur in the
Computational
- Automation process of programming
practices
- Simulation (Ex. Loops, Being increment and iterative)
- Parallelization
Students understandings of themselves,
their relationships to others, and the
Computational
- - Abstraction technological world around them
perspectives
(Ex. Testing, debugging, abstracting,
questioning

2.2. Computational Thinking and Cognitive Load


According to cognitive load theory (CLT), cognitive overload occurs when one
exceeds the required cognitive resources necessary to solve problems. Cognitive overload
is considered a major cause of learning failure. Thus, there have been many attempts to
identify the factors that cause unnecessary cognitive load and the teaching strategies to
manage the learners cognitive load during study [10].
Sweller [11] researched how to promote learners problem solving process by reducing
the cognitive effort of learners and obtaining a cognitive structure that is needed to
resolve the problem. It was about how learners can acquire the problem solving principle
and store it to their long-term memory.
Compared with previous instructional design approaches we are interesting the impact
of CT based education to learners; especially the decrease of cognitive load that learners
show after certain hours of training. For this experiment, we considered novice computer
science students to acquire CT skills for problem solving environments so that the skills
may contribute to reduce the cognitive load to solve problems.

2.3. Cognitive Load and Brain Activation

2.3.1 Brain Activation Measurement: As modern sensing technologies advance, various


types of physiological measurement techniques have been developed for assessing the
cognitive workload. Direct measurement of cognitive load in multimedia learning [12]
using fMRI or other physiological methods have been proposed. In this paper, we also
apply the EEG (Electroencephalography) technique for direct measure of cognitive load.
Brain wave comes from the summation of brain electrical activity that occurs in the
nerve cell population of cerebral cortex in vitro, and it is amplified and recorded from the
scalp phase (intact scalp). The signal is recorded as the horizontal axis showing the time
and the vertical axis to show electric potential. EEG is an objective, non-invasive,
continuous test to assess cerebral function and it is widely used today for cerebral
function evaluation.
Previous research has demonstrated that EEG signals are sensitive to cognitive load
changes in various tasks [5]. Among them, we referred Gevins and Smith [13] report that
spectral features of the theta and alpha frequency bands correlate with task difficulty

Copyright 2015 SERSC 171


International Journal of Software Engineering and Its Applications
Vol. 9, No. 10 (2015)

levels. Also, Fitzgibbon and his colleagues [14] showed that the power of gamma band
could be augmented by various cognitive tasks. These experiments clearly indicate that
we may find correlation between cognitive load and its relevance to CT education with
EEG signals.
To observe the cognitive load changes after CT education we have applied the
theta/beta ratio band power as the indicator for viewing the cognitive load level of
participants as Fitzgibbon did during the problem solving tasks, where high value
indicates high load to experiment.

3. Method
3.1. Characteristics of Courses

3.1.1 Computational Thinking Course (CT-group): This course was designed for
teaching CT concepts to pre-service teachers. After 10 weeks program, the students were
asked to develop Scratch program utilizing CT concepts of Table 1 and 2. The course
follows the steps of creative design, and participants were asked to develop program for 5
weeks as shown in the Table 3.

Table 3. Week Schedule of Computational Thinking Learning

Week Contents
Week 1 Orientation
Week 2 CT-Pre test
Introduction of Computational Thinking (CT) and CTs
Week 3
9 important concepts
Week 4 CT and Problem solving CT-
Week 5 Case Study : CT applicable Class situation (General) Application
Discussion : CT applicable Class situation (Subject Stage
Week 6
Specific) (3hrs,
Week 7 Mid-term 5weeks )
Project introduction : Computational Thinking and
Week 8
Scratch
Week 9 Scratch : Basic Class
Scratch : Intermediate Class
Week 10
Project : Planning Scratch project for my future class
Scratch
Scratch : Advanced Class
Application
Week 11 Project : Developing Scratch project for my future
Stage
class(1)
(3hrs,
Project : Developing Scratch program for my future
Week 12 5weeks)
class(2)
Project : Completing Scratch program for my future
Week 13
class
Project : Presentation of Scratch program for my future
Week 14
class
Week 15 CT- Post-test and EEG measurement

3.1.2 Basic Computer Skill Course (NCT-group): This course was designed for
introducing computers for pre-service teachers to expect for applying the skills in their
teaching after graduation. Similarly, students took 10 weeks courses, and the instructor
led students to experience the Scratch programming for 5 weeks as an introduction of
computer application for their teaching job.

172 Copyright 2015 SERSC


International Journal of Software Engineering and Its Applications
Vol. 9, No. 10 (2015)

3.2. Questionnaire
Items to measure the problem solving competency need to be designed so that it may
minimize the effects of a specific subjects prior knowledge, and enable the measurement
of actual competency during the problem solving process. For this, we selected the CT-
based problem solving test designed by [15] which is derived from the OECD / PISA
problem solving questions and modified for undergraduate students. The questionnaire
consists of 3 main categories and 8 sub-elements of CT competency. A total of 16
questions were designed and some examples are shown in the appendix.

Table 4. Questionnaire Designed for Assessing CT Competency [15]

Sub- element of Number of


Category
CT capacity Questions.
Problem Logical Thinking 2
Discovery Analytical Thinking 1
Problem Analytical Thinking 2
Analysis and
Abstract Thinking 1
Representation
Simultaneous Thinking 1
Problem Precedence Thinking 1
Solution Strategic Thinking 4
Strategy Procedural Thinking 2
Recursive Thinking 2
Total 16

3.3. Participants

3.3.1. Characteristics of Total Participants: A total of 34 students (16 male, 18 female),


ages 20 to 24 years old participated in the experiments. Among them 16 students were
selected from the group of participants to take part from the CT education course with
Scratch programing (CT group), and 18 students participated from the basic computer
skills course students with Scratch programing (NCT group).

3.3.2. Participants for EEG Measurement: 12 right handed (6 male, 6 female), ages 20
to 24 years old participated in the experiment. 6 students from the group of participants
completed a CT education course with Scratch programming (CT-group), and 6 students
from the group participated in a basic computer skill course with Scratch programming
(NCT-group).

3.4. Experiment Procedure


One of the purpose of this study is to investigate the effect of learners cognitive load
during the problem-solving process, after taking a CT skills course. In the experiment,
after the 10 learning sessions (30 hours in total), the problem solving competency test
(paper-based) and EEG measurement of the frontal lobe were conducted between a CT-
group (who participated CT skill education, and a NCT-group (who did not participate CT
course).
CT competency tests for college students [15] were applied to measure general
problem solving ability and problem solving questions of PISA was used for the EEG
measurement.

Copyright 2015 SERSC 173


International Journal of Software Engineering and Its Applications
Vol. 9, No. 10 (2015)

t1 r1 t2 r2 t3

Test Resting Test Resting Test


3 min 5 min 3 min 5 min 3 min
Figure 1. EEG Test Schedule

Brain activity was measured with 2 channels (Fp1 and Fp2) of 10/20 EEG
measurement system. To accurately measure brain activeness, measurement an electrode
arrangement system was used as in Figure 1 [16]. EMOTIVs wireless device, EPOC
Neuro Headset was used to measure the cognitive load of the participants, and the EEG
data was processed with EMOTIVs Test Bench. The measured analog signal data is
transformed to the digital signal with the Complexity 2.0 of Laxtha and EEG was
analyzed with a Fast Fourier Transform (FFT).
EEG measurement results may be affected by location, the condition of participants,
illumination and noise. Thus, to prevent the possibility of outside environments factors
that may interfere the validity of the experiment, the measurement was conducted in a
shielded room.

Source : TCT [17]


Figure 2. 10/20 System Position of EEG Measurement

4. Results
4.1. Analysis of CT based Problem Solving Test
A pre-CT based problem solving test was conducted for both CT and NCT groups.
According to the t-test results, both groups showed similar values in most areas, thus it is
possible to assume that participants of both groups have identical knowledge of CT
competencies.

174 Copyright 2015 SERSC


International Journal of Software Engineering and Its Applications
Vol. 9, No. 10 (2015)

Table 4. t-test Result of pre-CT based Problem Solving Test (p<.05)


Classification Group Mean S.D. t Sig.
Analytical CT .8750 .22361
Analysis and -2.376 .024*
Thinking NCT 1.0000 .00000
Represen-
Abstract CT .8125 .40311
tation .243 .810
Thinking NCT .7778 .42779
Strategic CT .7344 .28090
Solution .140 .890
Thinking NCT .7222 .22506
Strategy
Simultaneous CT .8125 .40311
-1.181 .246.
Thinking NCT .9444 .23570
Logical CT .6250 .28868
1.547 .132
Discovery Thinking NCT .4444 .32970
Analysis CT .7500 .44721
1.173 .249
Thinking NCT .5556 .51131
Recursive CT .7500 .36515
1.176 .248
Thinking NCT .6111 .32338
Solution
Procedural CT .5000 .36515
Strategy .434 .667
Thinking NCT .4444 .37920
Precedence CT .6250 .50000
.636 .529
Thinking NCT .3889 .50163
(Source: S. Y. Park, et al. [21])

If the p-value associated with the t-test is less than the significance level (<.05), then
the null hypothesis is rejected, and both groups have similar CT capabilities. On the
contrary, if p-value is greater than the significance level (>.05), it fails to reject the null
hypothesis and both groups CT capabilities are different [18].
From observing Table 2, all the p-value of each subdomain of CT skills are greater
than the significance level (>.05) except analytical thinking domain (<0.5), thus we can
conclude that both groups have no CT capability difference except analytical thinking.
To check the effectiveness of designed CT education course, CT based problem
solving test with problem isomorphs was given to the participants of both groups after 10
sessions of CT education.

Table 5. t-test Results: post-CT based Problem Solving Test (p<.05)


Group Mean S.D. t Sig.
Analytical CT 1.0000 .00000
Analysis and Thinking NCT 1.0000 .00000
Representation CT 1.0000 .00000
Abstract Thinking
NCT 1.0000 .00000
Strategic CT .8438 .22127
2.329 .026*
Thinking NCT .6250 .31213
Solution Strategy
Simultaneous CT 1.0000 .00000
2.074 .046*
Thinking NCT .7778 .42779
CT 1.0000 .00000
Logical Thinking 1.618 .115
NCT .8889 .27416
Discovery
Analysis CT 1.0000 .00000
1.372 .180.
Thinking NCT .8889 .32338
Recursive CT .6875 .35940
2.768 .009**
Thinking NCT .3333 .38348
Solution Procedural CT .9375 .17078
1.430 .162
Strategy Thinking NCT .8333 .24254
Precedence CT .6250 .50000
.716 .479
Thinking NCT .5000 .51450
(Source: S. Y. Park, et al. [21])

Copyright 2015 SERSC 175


International Journal of Software Engineering and Its Applications
Vol. 9, No. 10 (2015)

Even though both groups participants showed better scores than the pre-test, and the
score of CT-group participants are better than the NCT-group participants except analysis
and representation domains. However, the CT-group shows greater strength in questions
relating to solution strategy which consists of strategic and simultaneous thinking than
NCT-group. The CT-groups mean value of strategic thinking domain was 0.8438 while
the NCT-group got 0.6240 (<.05).
Also, in simultaneous thinking element, CT-groups mean value of 1.00 was higher
than the NCT-groups of 0.7778 (<.05). The t-test value was found to be highly significant
(<.01) between the CT-group and the NCT-group in terms of recursive thinking.
From these results, we can say that the designed course with Scratch programming was
effective to both groups participants, however course designed with considering CT
concepts was more effective than basic computer skill course with Scratch programming.

4.2. Cognitive Load Changes with EEG Measurements


Emotion researched through EEG devices revealed that there exists correlation
between brain waves and cognitive states. For example, the alpha wave asymmetry
measurement on the frontal lobe shows the presence of states showing low cognitive load.
Meanwhile, some researchers [19, 20] reported that the relative theta/beta ratio in Fp1 and
Fp2 decreases in ratio values during cognitive load conditions compared to relaxing states.
Based on these research, we also recorded brain wave from Fp1 and Fp2 of frontal lobe
and analyzed relative band power of theta/beta ratio are depicted in the Figure 6.

Figure 3. Theta/beta Relative Band Power Ratios Appeared in the Test with
EEG

As previous research shows, the theta/beta ratio decreases during the test period (t1, t2
and t3) more than rest time period (r1, r2) in the CT group. This pattern clearly happens
in the CT group, but the NCT groups result show less small differences between test
period and rest period.
Also, the theta/beta ratio of the CT group decreases more than the NCT group, and the
ratio changes between test time and rest time is greater than that of the NCT group. These
patterns imply that participants of the CT group may allocate cognitive power to solve the
given problems. Another implication of this results is that CT group participants more
effectively use their cognitive power than the NCT group. In other words, the CT group
spent cognitive power more skillfully, when they have to pay attention or take rest.

5. Conclusion
Enhancement of computational thinking capabilities for freshmen level pre-teachers is
achieved through 10 sessions of Scratch programming integrated with introductory CS
course. Among the improvements of participants, solution strategy which consists of

176 Copyright 2015 SERSC


International Journal of Software Engineering and Its Applications
Vol. 9, No. 10 (2015)

strategic and simultaneous thinking, and strategic thinking domain show bigger
differences.
If learners are exposed to the opportunity of courses providing CT skill training
combined with computer application subjects, it may help them to enhance students'
problem-solving skills especially those associated with the ability of strategic thinking.
On the other hand, if learners have the opportunity to learn problem solving
strategies repeatedly, they may be better prepared to solve complex problems effectively
allocate their cognitive power.
This assumption has been proved from the direct measurement of EEG signals and the
theta/beta ratio pattern of Fp1 and Fp2 in frontal lobe. Even though, the t-test results of
the EEG measurement was not supported by statistical significance, it is expected that if
larger samples and more EEG channels are utilized then the learners' cognitive load
change may be more clearly proved with statistical significance.

Appendix
Examples of questions used for the CT capacity test [15].

Q-A. The following is a description of instructions for generating a figure below. Fill the
blanks with the appropriate instruction.
Q-B. Write appropriates commands for generating the next figure below. (Use the
repeating structure, and a single variable (A)).

(a) (b)
Figure 4. Figures used in the Questionnaire (a) for Q-A, (b) for Q-B

References
[1] S. H. Kim, K.-S. Song and S. Y. Park, Exploring the Technological Factors Affecting Creativity in
Computational Thinking-Centered Learning Context, WSEAS Transactions on Advances in
Engineering Education, vol. 14, (2014), pp. 70-79.
[2] A. Yadav, N. Zhou and C. Mayfield, Introducing Computational Thinking in Education Courses,
Proceeding of SIGCSE11, Dallas, Texas, USA, (2011) March 9-12.
[3] J. M. Wing, Computational Thinking, Communication of the ACT, vol. 49, no. 3, (2006), pp. 33-35.
[4] S. Y. Lye and J. H. L. Koh, Review on teaching and learning of computational thinking through
programming: What is next K-12?, Computers in Human Behaviour, vol. 41, (2014), pp. 51-61.
[5] Y. H. Choi, Analysis of Electroencephalogram (EEG) Activities of Middle School Students in
Technological Problem Solving Thinking Process as Levels of Structured Problem, Journal of Korean
Practical Arts Education, vol. 17, no. 4, (2011), pp. 129-152.
[6] ISTE, CSTA, NSF, Computational Thinking Teacher Resources (2011), Retrieved from:
http://www.csta.acm.org/Curriculum/sub/CompThinking.html.
[7] K. Brennan and M. Resnick, New frameworks for studying and assessing the development of
computational thinking, AERA, (2012).
[8] V. Barr and C. Stephenson, Bringing computational thinking to K-12: what is involved and what is the
role of the computer science education community?, ACT Inroads, vol, 2., no. 1, (2011), pp. 48-54.
[9] D. Shaffer, Cohesion, coupling, and abstraction, H. Bigoli (Ed.) Encyclopedia of Information Systems,
New York: Academic Press, (2003), pp. 127-139.
[10] L. Corno and E. B. Mandinach, The role of cognitive engagement in classroom learning and
motivation, Educational Psychologist, vol. 18, no. 2, (1983), pp. 88-108.

Copyright 2015 SERSC 177


International Journal of Software Engineering and Its Applications
Vol. 9, No. 10 (2015)

[11] J. Sweller, Cognitive load during problem solving: Effects on learning, Cognitive Science, vol. 12,
(1988), pp. 257-285.
[12] R. Brnken, J. Plass and D. Leutner, Direct Measurement of Cognitive Load in Multimedia Learning,
Educational Psychologist, vol. 38, no. 1, (2003), pp. 53-61.
[13] A. Gevins and M. Smith, Neurophysiological measures of cognitive workload during human computer
interaction, Theoretical Issues in Ergonomic Science, vol. 4, (2003), pp. 113-131.
[14] S. Fitzgibbon, K. Pope, L. Mackenzie, C. Clark and J. Willoughby, Cognitive tasks augment gamma
EEG power, Clinical Neurophysiology, vol. 115, (2004), pp. 1802-1809.
[15] E. K. Lee, A Robot Programming Teaching and Learning Model to Enhance Computational Thinking
Ability, Ph.D Thesis of Korea National University of Education, (2009).
[16] G. H. Klem, O. H. Luders, H. H. Jasper and C. Elger, The ten-twenty electrode system of the
International Federation, Electroencephalography, Clinic Neurophysiology, vol. 52, suppl.: 3, (1999).
[17] TCT, 10/20 System Positioning manual, Hong Kong, Trans Cranial Technologies ltd., (2012).
[18] J. C. Song, S. Nam, K. S. Song and S. Y. Park, Development of a Personalized Learning System Using
Gaze Tracking System, WSEAS Transactions on Computers, vol. 14, (2015), pp. 264-271.
[19] F. H. Howells, V. L. Ives-Deliperi, N. R. Horn and D. J. Stein, Mindfulness based cognitive therapy
improves frontal control in bipolar disorder: a pilot EEG study, BMC Psychiatry, vol. 12, no, 15,
(2012).
[20] M. M. Lansbergen, M. Arns, M. V. Dongen-Boomsma, D. Spronk and J. K., Buitelaar, The increase in
theta/beta ratio on resting-state EEG in boys with attention-deficit/hyperactivity disorder is mediated by
slow alpha peak frequency, Neuropsychopharmacol Biol. Psychiatry, vol. 35, no. 1, (2011) January 15,
pp. 47-52.
[21] S. Y. Park, K. S. Song and S. H. Kim, EEG Analysis for Computational Thinking based Education
Effect on the Learners Cognitive Load, Proceedings of the Applied Computer and Applied
Computational Science (ACACOS '15), Kuala Lumpur, Malaysia, (2015) April 23-25.

178 Copyright 2015 SERSC

S-ar putea să vă placă și