Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

H.

Chad Lane
Kalina Yacef
Jack Mostow
Philip Pavlik (Eds.)

Artificial Intelligence
LNAI 7926

in Education
16th International Conference, AIED 2013
Memphis, TN, USA, July 2013
Proceedings

123
A Computational Thinking Approach
to Learning Middle School Science

Satabdi Basu and Gautam Biswas

Institute of Software Integrated Systems, Vanderbilt University, Nashville, TN, U.S.A.


{satabdi.basu,gautam.biswas}@vanderbilt.edu

Abstract. Computational Thinking (CT) defines a domain-general, analytic ap-


proach to problem solving, combining computer science concepts with practices
central to modeling and reasoning in STEM (Science, Technology, Engineering
and Mathematics) domains. In our research, we exploit this synergy to develop
CTSiM (Computational Thinking in Simulation and Modeling) - a cross-domain,
visual programming and agent based, scaffolded environment for learning CT and
science concepts simultaneously. CTSiM allows students to conceptualize and
build computational models of scientific phenomena, execute the models as simu-
lations, conduct experiments to verify the simulation behaviors against ‘expert
behavior’, and use the models to solve real world problems.

Keywords: Computational Thinking, Science education, Visual Programming,


Agent-based modeling and simulation, Learning by design, Scaffolding.

1 Introduction

Computational Thinking (CT) provides a domain-general approach to modeling phe-


nomena, solving problems, and designing systems by drawing on core computer
science concepts [9]. It supports practices like abstraction, decomposition, recursion,
simulation, and verification, several of which are also central to the development
of STEM expertise [5]. For example, formally representing scientific laws and phe-
nomena resembles the object-oriented programming concepts of encapsulation, ab-
straction, and generalization. Conversely, the biological concepts of taxonomy and
inheritance inspire the class inheritance concepts in CT. Research has also shown that
novice misconceptions have similar patterns in science, math, and programming do-
mains: they have both domain-specific and domain general roots (e.g., challenging
concepts, and difficulties pertaining to conducting inquiry and problem solving) [6].
Exploiting the assumption that CT concepts are learnt best when anchored in real
world problem contexts and that CT concepts parallel important aspects of science
learning, several researchers have focused on leveraging the synergies between CT
and scientific expertise [2,3,6]. Though research suggests that programming and com-
putational modeling can serve as effective vehicles for learning challenging STEM
concepts [2,3], we still know of no CT-based environments for science education that
have been integrated into classrooms in any significant way.

K. Yacef et al. (Eds.): AIED 2013, LNAI 7926, pp. 920–923, 2013.
© Springer-Verlag Berlin Heidelberg 2013
A Computational Thinking Approach to Learning Middle School Science 921

This motivates our research, which aims to develop a computer-based learning


environment for synergistic teaching of CT and science concepts in middle school
classrooms supported by an adaptive scaffolding framework. Developing such an
environment involves several challenges including (1) selecting a pedagogical ap-
proach supporting science learning as well as CT practices, (2) designing an activity
sequence progressing from conceptualization of a phenomena to problem-solving
using the acquired knowledge, (3) designing an interface where the functionality of
the environment is decomposed into manageable modules, (4) choosing a program-
ming paradigm that makes the CT principles explicit without the challenges of
learning a programming language syntax, (5) making explicit the computational
commonalities across different science domains, and (6) diagnosing problems faced
by students at varying levels of understanding and developing supporting tools and
scaffolds to help them learn CT and science concepts simultaneously. The following
sections describe our approach to addressing these challenges by developing the
CTSiM (Computational Thinking in Simulation and Modeling) environment.

2 Research Methodology

In keeping with the core epistemic and representational scientific practice of ‘model-
ing’ [4], and a core CT practice of developing models and simulations of problems
[9], our research adopts a learning-by-design pedagogy, as described below.
CTSiM is decomposed into multiple worlds [1,8] to make the learning process more
manageable. In the Conceptual Modeling (CM) World, students develop initial ab-
stractions of the phenomena being studied by identifying the types of agents involved,
their properties and behaviors, and specifying how the properties and behaviors are
related. We choose an agent-based paradigm since it is believed to leverage students’
pre-instructional intuitions, and help learn emergent phenomena in science domains.
In the Construction (C) World, students build executable computational models using
a visual programming language, thus reducing students’ challenges in learning pro-
gramming language syntax. Some visual primitives are domain-specific, while others
related to CT principles (conditionals, loops, operators) can be reused across domains.
Each visual primitive is defined in terms of one or more domain-independent compu-
tational primitives, which is translated to NetLogo code to produce multi-agent simu-
lations, which students visualize in the Enactment (E) world. Then, students design
experiments in the Envisionment (V) World to compare their model behaviors with
that of an ‘expert’ model, and demonstrate their understanding in the Problem-solving
(PS) World by using their models to predict, explain and solve real-world problems.
For supporting students’ tasks in CTSiM, various tools have been and will continue
to be developed. These include (a) a set of searchable hypermedia resources with
text, diagrams, videos, and simulations acting as a domain knowledge source for the
phenomena being studied, (b) a model-tracing tool in the E-World that enables trac-
ing the code command-by-command with the simulation to help students correlate
their models with the resultant simulations, (c) a code commenting-out tool that
enables students to test their code in parts, and (d) a guided dynamic workflow in the
922 S. Basu and G. Biswas

V-World to help students design structured experiments, explicitly specify their goals,
and use the comparison results effectively to verify and refine their models. The Re-
sources and the Workflow can be used with any modeling task, while the other tools
are specific to the agent-based computational modeling paradigm we employ.
In addition to providing these tools, open-ended systems like CTSiM need to pro-
vide scaffolding to help learners who may not be proficient in using the systems’ tools
or regulating their own learning. Adaptive scaffolding refers to actions taken by the
learning environment, based on its interactions with the learner, with the intention to
support the learner in completing a task [7]. However, providing adaptive scaffolds is
challenging. It requires systematically diagnosing learners’ needs by interpreting how
learners at varying levels of understanding approach their tasks, and adapting dynam-
ically to the learners’ states. While several modeling, simulation, and problem-solving
environments have been developed for science domains, few provide adaptive scaf-
folding derived from systematic interpretations of the learners’ approach to the learn-
ing task. In CTSiM, we systematically analyzed the challenges faced by different
students and categorized them broadly into modeling, programming, domain know-
ledge and agent-based thinking challenges [1]. Accordingly, adaptive scaffolding in
CTSiM will focus on online detection of these challenges along with providing sup-
porting strategies for the broad categories of challenges identified.

3 Expected Contributions of this Research

Given the dearth of learning environments that exploit the synergy between CT and
science education, our research will significantly contribute to the AIED community.
In particular, its contributions will include: (1) the development of a learning envi-
ronment that fosters the development of model building and scientific reasoning on
the one hand, and algorithm design and verification on the other; (2) the development
of a multi-level Conceptual Modeling interface that makes explicit students’ concep-
tions about the model structure; (3) the seamless integration of a visual programming
and animation tool into a multi-agent modeling and simulation environment that im-
proves the understanding of science topics in middle school classrooms; (4) the de-
velopment of a limited set of domain-independent computational primitives such that
all visual primitives can be defined in terms of one or more of them; (5) the develop-
ment and testing of a guided dynamic workflow to help students experiment systemat-
ically, set goals for themselves, record observations, draw conclusions based on the
observations, and monitor their own progress; (6) a systematic analysis and categori-
zation of challenges students face while working with a CT based learning environ-
ment for science; and (7) the development and testing of an adaptive scaffolding
framework based on detecting and overcoming the identified challenges.

4 Next Steps and Expected Results

In our first iteration of designing and implementing CTSiM, only the C, E, and V
worlds were developed along with two curricular units for kinematics and ecology
A Computational Thinking Approach to Learning Middle School Science 923

(these units can be tried out at http://www.teachableagents.org/downloadctsim.php).


A preliminary research study conducted in 6th grade classrooms showed significant
pre-post test learning gains in science, while demonstrating the need for scaffolds and
supporting tools [1,8]. The study also showed that the verbal scaffolds provided were
useful and helped reduce the average number of challenges over time and increase the
learning gains. The next steps of our research will include: (1) incorporating more
support through the CM and PS worlds, the 4 tools described in Section 2, and an
adaptive scaffolding framework using a mixed-initiative dialogue between the stu-
dents and a pedagogical agent; (2) developing more curricular units and making the
computational commonalities across domains more explicit; (3) conducting experi-
ments with and without the different tools and scaffolds to demonstrate their individ-
ual effects; and (4) conducting experiments to show learning gains for both science
and CT concepts, transfer of CT skills across domains, and increased abilities to solve
real world problems, construct their abstractions and model them algorithmically.

Acknowledgements. This work was supported by the NSF (NSF Cyber-learning


grant #1237350).

References
1. Basu, S., Dickes, A., Kinnebrew, J.S., Sengupta, P., Biswas, G.: CTSiM: A Computational
Thinking Environment for Learning Science through Simulation and Modeling.
In: Proceedings of the 5th International Conference on Computer Supported Education,
Aachen, Germany, pp. 369–378 (2013)
2. Guzdial, M.: Software-realized scaffolding to facilitate programming for science learning.
Interactive Learning Environments 4(1), 1–44 (1995)
3. Hambrusch, S., Hoffmann, C., Korb, J.T., Haugan, M., Hosking, A.L.: A multidisciplinary
approach towards computational thinking for science majors. In: Proceedings of the 40th
ACM Technical Symposium on Computer Science Education (SIGCSE 2009), pp. 183–
187. ACM, New York (2009)
4. Lehrer, R., Schauble, L.: Cultivating model-based reasoning in science education. In:
Sawyer, R.K. (ed.) The Cambridge Handbook of the Learning Sciences, pp. 371–388.
Cambridge University Press, New York (2006)
5. National Research Council, A framework for K-12 Science Education: Practices, Crosscut-
ting Concepts, and Core Ideas. The National Academies Press, Washington, DC (2011)
6. Perkins, D.N., Simmons, R.: Patterns of misunderstanding: An integrative model for
science, math, and programming. Review of Educational Research 58(3), 303–326 (1988)
7. Puntambekar, S., Hübscher, R.: Tools for scaffolding students in a complex learning envi-
ronment: What have we gained and what have we missed? Educational Psychologist 40,
1–12 (2005)
8. Sengupta, P., Kinnebrew, J.S., Basu, S., Biswas, G., Clark, D.: Integrating Computational
Thinking with K-12 Science Education Using Agent-based Computation: A Theoretical
Framework. Education and Information Technologies 18(2), 351–380 (2013)
9. Wing, J.M.: Computational Thinking: What and Why? Link Magazine (2010)

You might also like