NOVEMBER 2010 | VOL. 53 | NO.

11 | COMMUNI CATI ONS OF THE ACM 27

L

L
A
B
Education
K12 Computational Learning
Enhancing student learning and understanding by combining theories
of learning with the computers unique attributes.
DOI:10.1145/1839676.1839686 Stephen Cooper, Lance C. Prez, and Daphne Rainey
I
N COMPUTATIONAL THINKING,
14

Jeannette Wing struck a chord
that has resonated strongly
(generating positive as well as
negative responses) with many
computer scientists and non-comput-
er scientists. Wing has subsequently
defined computational thinking as
the process of abstraction,
15
guided
by various engineering-type concerns
including efciency, correctness, and
several software engineering -ilities
(maintainability, usability, modifi-
ability, and so forth). Some have inter-
preted computational thinking as an
attempt to capture the set of computer
science skills essential to virtually ev-
ery person in a technological society,
while others view it as a new descrip-
tion of the fundamental discipline
that represents computer science and
its intersection with other elds. The
National Academies report
1
captures
both of these views, as well as present-
ing others.
While we can live with such deni-
tions/descriptions in the higher educa-
tion arena, we struggle with these no-
tions of computational thinking in the
K12 arena (note that we primarily con-
sider K12 education within the U.S.).
Several concerns spring to mind:
1. Computer science does not ap-
pear within the core topics covered in
high school. We would have a tough
time justifying a computer science
course, even the great ideas AP Princi-
ples course (being developed as part of
Denning
4
) replacing Algebra 2, Biology,
or American Government. K12 educa-
tion is a zero-sum game. If one wishes
to add a course, one must also propose
a course to be removed.
2. Even as an elective topic, comput-
er science tends to be disproportionate-
ly available to those wealthy suburban
schools. Margolis et al.
6
explore this
situation in depth within the urban Los
Angeles school district.
3. Too few K12 computing teachers
are available to implement a national-
scale computing requirement. CUNYs
ambitious 10,000 teacher project
2
will
not produce sufcient numbers of com-
puting teachers required to instruct all
schoolchildren in the U.S. It would not
even get one qualied teacher into each
of the nations 30,000+ high schools
(see http://nces.ed.gov/programs/di-
gest/d09/tables/dt09_086.asp).
4. It is not clear to us how teachers in
other K12 subjects would take advan-
tage of school children who had been
trained in computational thinking.
5. The most common denitions of
computational thinking are confusing
when explained to non-computer scien-
tists. And many K12 computing teach-
ers are not computer scientists.
We nd the above denitions of
computational thinking not espe-
cially useful when considered in the
context of K12 education, and, more
specically, K12 science, technol-
ogy, engineering, and mathematics
(STEM) education. We propose an al-
ternative to the common denition
of computational thinking we believe
is appropriate for operationalization
in K12 education and consider its
broader implications.
Berkeley public elementary school students on a eld trip learn how computers enable
science.
28 COMMUNI CATI ONS OF THE ACM | NOVEMBER 2010 | VOL. 53 | NO. 11
viewpoints
A K12 View of
Computational Thinking
We have struggled with how computa-
tional thinking might be different from
mathematical thinking, algorithmic
thinking, quantitative reasoning, de-
sign thinking, and several other models
of math, science, and even engineering
to critical thinking and problem solv-
ing. It was after struggling with the lat-
est type of thinking that we realized that
perhaps even the term computational
thinking was misleading (from a K12
perspective), and we were approaching
the denition incorrectly. Rather than
considering computational thinking as
a part of the process for problem solv-
ing, we instead developed a model of
computational learning that empha-
sizes the central role that a computer
(and possibly its abstraction) can play in
enhancing the learning process and im-
proving achievement of K12 students
in STEM and other courses. The gure
here depicts our current working model
of computational learning. It should be
noted that this model is explicit in its
use of a computer and specically ex-
cludes non-cognitive uses of technology
(Powerpoint, wikis, blogs, clickers, and
so forth).
Similar to Wings original vision of
computational thinking, we see com-
putational learning as an iterative and
interactive process between the hu-
man (the K12 student in our case) and
the computer (or, in a more theoretical
construct, a model of computation).
We also make explicit the two conse-
quences of the human cognitive pro-
cess, namely, the capacity for abstrac-
tion and for problem formulation, and
two strengths of the computer, namely,
their ability to present complex data
sets, often visually, and their capacity
for storing factual and relational knowl-
edge. These four elements frame and
establish the boundaries of the iterative
interaction between the human being
and the computer. Note that the accom-
panying gure does not explicitly in-
clude a teacher, not because we believe
teachers are unnecessary, but rather
because the role of the teacher in this
model is complex and requires further
investigation.
In developing this model, we ob-
served that it includes other extant
models in scientic learning and in-
quiry. For example, one can view com-
putational science as the interaction
between the human and the computer
that is contained within the box where
a human being formulates a problem
and provides a representation suitable
for a computer. The computer then acts
on this representation and returns the
results of these actions to the human
being through, for example, a visual
representation. Computational learn-
ing expands this interaction by allow-
ing the computer to add foundational
knowledge, not just data, unknown to
the human and by having the results of
the computers actions represented in
a form compatible with the humans
current capacity for abstraction. In the
more interesting instances of computa-
tional learning, both of these processes
are likely to be adaptive and personal-
ized to the individual.
We make several observations about
computational learning:

Computational learning is iterative,
requiring interaction between the com-
puter and the human.

In computational learning, the
computer can compensate for a hu-
mans lack of factual and relational
knowledge and mathematical and sci-
entic sophistication.

The computers ability to quickly
compute multiple examples and pres-
ent them via a modality appropriate to
a humans current development stage
and level of meta-cognitive awareness
can leverage the humans inherent,
though perhaps not fully conscious, ca-
pacity for abstraction.
This model for computational learn-
ing differs signicantly from other pro-
posed notions of computational think-
ing. For example, algorithmic thinking
does not require a computer and math-
ematical thinking is almost solely de-
pendent on the humans formalization
capacity for abstraction.
2

To better understand how our pro-
posed computational learning model
can be operationalized, we present
two examples: one from middle school
computing and one from high school
biology.
Digitizing Data
Cutler and Hutton
3
modied a CS Un-
plugged activity on image represen-
tation (see http://csunplugged.org/
sites/default/les/activity_pdfs_full/
unplugged-02-image_representation.
pdf) to enable middle school students
to work interactively with a computer
program as they learn about how com-
puters digitize images. The purpose of
these activities is to help students to un-
derstand what it means for a computer
to digitally represent an image. More
importantly, the students learn to move
from concrete representations of im-
ages to more abstract representations
of those images (as a digital represen-
tation), and from representation of im-
ages in 2D to representing objects in 3D.
And this ability to abstract is important
across all STEM disciplines.
Students work interactively with the
computer program, receiving feedback
to their attempts at digitizing their data.
Over the series of lessons, they develop
an initial ability to abstract away from
the physical representation of an im-
age to its digital representation. They
are then further able to develop their
ability to abstract as they move from 2D
A view of computational learning.
Capacity for Abstraction
External Knowledge
Memory Working Memory Visualization
Computation Knowledge Problem
Formulation
Computer
Computational Science
Human
viewpoints
NOVEMBER 2010 | VOL. 53 | NO. 11 | COMMUNI CATI ONS OF THE ACM 29
Other (though the computer obviously
cannot think at a higher level than the
student),
11
or even ts within Newell
and Simons Information Processing
Theory framework.
8
Our hope is that by
considering our model of computation-
al learning, we can better educate and
prepare teachers to benet from com-
puting in and outside the classroom,
and that approaches and computing
tools can be identied and built to im-
prove K12 student STEM learning.
References
1. Committee for the Workshops on Computational
Thinking, National Research Council. Report of a
Workshop on the Scope and Nature of Computational
Thinking. National Academies Press, Washington, D.C.,
2010.
2. Cuny, J. Finding 10,000 teachers. CSTA Voice, 5, 6
(2010), 12; http://www.csta.acm.org/Communications/
sub/CSTAVoice_Files/csta_voice_01_2010.pdf
3. Cutler, R. and Hutton, M. Digitizing data: Computational
thinking for middle school students through computer
graphics. In Proceedings of the 31st Annual Conference
of the European Association for Computer Graphics EG
2010Education Papers. (Norrkping, Sweden, May
2010), 1724.
4. Denning, P. Beyond computational thinking. Commun.
ACM 52, 6 (June 2009), 2830.
5. Gilbert, J.K., Justi, R., and Aksela, M. The visualization
of models: A metacognitive competence in the learning
of chemistry. Paper presented at the 4th Annual
Meeting of the European Science Education Research
Association, Noordwijkerhout, The Netherlands, 2003.
6. Margolis, J. et al. Stuck in the Shallow End: Education,
Race, and Computing. The MIT Press, 2008.
7. National Science Foundation. Molecular visualization
in science education. Report from the molecular
visualization in science education workshop. NCSA
access center, National Science Foundation, Arlington,
VA, 2001.
8. Newell, A., and Simon, H.A. Human Problem Solving.
Prentice-Hall, Englewood Cliffs, NJ, 1972.
9. Rotbain, Y., Marbach-Ad, G., and Stavy, R. Using a
computer animation to teach high school molecular
biology. Journal of Science Education and Technology
17 (2008), 4958.
10. Sewell R., Stevens, R., and Lewis, D. Multimedia
computer technology as a tool for teaching and
assessment of biological science. Journal of Biological
Education 29 (1995), 2732.
11. Vygotsky, L.S. Mind and Society: The Development of
Higher Mental Processes. Harvard University Press,
Cambridge, MA, 1978.
12. Williamson V.M. and Abraham, M.R. The effects of
computer animation on the particulate mental models
of college chemistry students. Journal of Research in
Science Teaching 32 (1995), 522534.
13. Windschitl, M.A. A practical guide for incorporating
computer-based simulations into science instruction.
The American Biology Teacher 60 (1998), 9297.
14. Wing, J.M. Computational thinking. Commun. ACM 49,
3 (Mar. 2006), 3335.
15. Wing, J.M. Computational thinking and thinking about
computing. Philosophical Transactions of the Royal
Society A, 366 (2008), 37173725.
Stephen Cooper ([email protected]) is an associate
professor (teaching) in the Computer Science Department
at Stanford University.
Lance C. Prez ([email protected]) is an associate professor
in the Department of Electrical Engineering at the
University of Nebraska-Lincoln where he also holds the
position of Associate Vice Chancellor for Academic Affairs.
Daphne Rainey ([email protected]) is a biologist
currently serving in a temporary assignment as a program
director at NSFs Division of Undergraduate Education
within its Education and Human Resources Directorate.
Copyright held by author.
images to 3D objects. Finally, students
develop a further ability to abstract. As
part of the creation of a single 3D object,
say a chair, the students are then chal-
lenged to place many chairs in a room.
They need to be able to recognize that
representing a 3D chair consists of two
parts: the relative coordinates of each of
the parts of the chair, and the absolute
location of one part (say the bottom-
right corner of the front-right leg).
Evolutionary Biology
The second example involves the teach-
ing of evolution using computational
learning. Our vision is of a 3D visualiza-
tion system that could simulate evolu-
tion. A student could specify an organ-
ism, with primitive appendages (arms,
legs, joints, and other attributes) to ac-
complish locomotion. Then, by provid-
ing an environment, the student could
run the simulation to watch how the
organisms ability to move evolves over
time as a function of its current loco-
motion capability coupled with the im-
pact of that organisms environment.
Students could change the appendag-
es and/or the environment to observe
how such changes lead to a difference
in the organisms evolution over time.
In computer science terms, this exam-
ple is similar to passing a program and
an initial state as input to a Universal
Turing Machine.
Such a simulation allows the stu-
dent to work interactively with the
computer program. The student learns
both from the impact of the changes
she makes to the initial conguration
of the organism and to the initial en-
vironment (which will lead to the or-
ganism evolving the ability to move
differently) as well as by the ability to
observe the simulation/visualization
as it is running. In science, researchers
have found that visualization is central
to increasing conceptual understand-
ing and prompting the formation of
dynamic mental models of particulate
matter and processes (see
5,7,9,12
). Vi-
sualization and computer interaction
through animation allow students to
engage more in the cognitive process,
and to select and organize more rele-
vant information for problem-solving.
7

Computer animations incorporated
into interactive simulations offer the
user a chance to manipulate variables
to observe the effect on the systems
behavior (see
9,10,13
).
While we know of no tool that pro-
vides the exact support/simulation we
are describing, there are several avail-
able visualization systems that can
simulate/model the world. Two of these
systems have helped to shape our vision
of the above-mentioned simulation:
The 3D visualization system, Fram-
sticks (www.framsticks.com) can be
used for modeling evolution, and the
2D simulation system, NetLogo (http://
ccl.northwestern.edu/netlogo/) has
many available pre-built simulations,
including those that model evolution
albeit in a different manner than what
we describe.
Conclusion
Most papers weve seen on compu-
tational thinking represent attempts
at repackaging computing science
concepts, especially in the form of al-
gorithmic thinking and introductory
programming, sometimes in other do-
mains. Though this may be useful in
some contexts, it is unlikely such a sim-
ple approach will have signicant im-
pact on student learningof computer
science or other disciplinesin the
K12 setting. The proposed model of
computational learning combines the-
ories of learning with the computers
superiority in dealing with complexity
and variability and its ability to present
results using modalities that appeal to
the learner in order to enhance student
learning and understanding. We believe
that computational learning can be
framed within various theories of learn-
ing, where the computer plays a similar
role as Vygotskys More Knowledgeable
This model for
computational
learning differs
signicantly from
other proposed
notions of
computational
thinking.