Future University Hakodate, JapanMeta-learning with EV3
MindstormsDr. Michael VallanceEdD., MSc., BSc.(Hons), PGCE
iVERG lab | mvallance.netFUN Japanterms usedmotivationresearch
outcomes
meta - learningMeta-learning / making sense of ones experience
of learning.what is learned (declarative knowledge) how something
is learned (procedural knowledge). advanced in meta-learning =
realize why something is learned (metacognitive knowledge). Generic
competency: willingness to learn, have new ideas, consolidate
previously studied information. What are your
capabilities?Declarative competency - the application of knowledge
in problem solving.How are you using your capabilities (what you
know)?Epistemic competency - displayed by evidence of increased
flexibility in the utilization of knowledge, making connections.Why
are you using your capabilities?Flow is a holistic sensation of
total involvement.
LEGO EV3 Mindstorms
motors, sensors, read & write svariables, loops,
switches
EV3 program solution
MotivationJapanese educationJapan is in top 3 OECD countries for
maths, science & reading. PISA (2012).Yet students lack
confidence, have high anxiety & little interest in STEM
subjects. PISA (2012).Good at test-taking.But lack creativity and
self-directed learning skills.
Educating students for the 21st centuryIn STEM and Arts &
Humanities courses, students need to be provided with meaningful
and relevant opportunities to:engage with problems that require the
retrieval of prior knowledge, offer multiple perspectives of
problems and solutions, and facilitate a challenging process which
results in achievable, diverse outcomes. Learning should not occur
in isolation but involve communication, cooperation and
collaboration with fellow learners and experts to foster long term
understanding and transference of learned concepts. # commonalities
among all literature on 21st century skills
Creativity Spiral creativity - the way people deal with
problems
Tsukutte (Create) Katatte (Share) Furikaeru (Reflect).T is
Making/ Creating.K is Talking/ Sharing.F is Reflecting/
Discussing.
Mitch Resnick, Director, MIT Media Lab, 2007
RESEARCH HYPOTHESISH1: Learners actively engaged in iteratively
challenging robot-mediated interactive tasks develop generic,
declarative and epistemic competencies.H2: Increased meta-learning
is associated with the development of generic, declarative and
epistemic competencies.H3: Students need to be in a state of flow
for meta-learning to occur.Research is not about acquisition of
particular knowledge (e.g. content) but attempts to learn more
about the processes of acquiring knowledge (e.g. competencies/
knowledge put into action).
participantsJAPAN: n = 6 / Yr2 & Yr3 at FUNUK: n = 6 /
Baccalaureate students (ages 16 ~ 18) studying Design &
TechnologyTASKS in AY2015 SEM 1 & SEM 2 = 20 Once a week at 5F
iVERG Lab and MACS LEGO lab (UK)
instrument: surVEY Data & Task complexity values
Post-task surveyTask quantified by its complexity valuesCTC = (d
+ m + s+ o)RTC = Mv1 + Sv2 + SW + Lv3
implementation
Tasks in oder of CTC complexityDATA
Results of Pearsons Chi Square Test of Association between CTC
& RTCp value: 0.09982Pearsons Chi Square statistic:
140.25Degrees of Freedom (df): 120p is small but NOT statistically
significant (less than 0.01 required for statistical significance)
Tasks
DATA
Tasks
DATA
Tasks
What is the correlation between REFLECT & COMPETENCIES as
circuit task complexity increased (that is, as problems became more
challenging?
REFLECT versus G.total subset by CTC (color as CTC value)
REFLECTResults of Pearsons Chi Square Test of Association
between REFLECT & Gtotalp value: 0.2751Pearsons Chi Square
statistic: 85.0Degrees of Freedom (df): 78p is NOT statistically
significantG total
REFLECT versus D.total subset by CTC (colour as CTC value)
REFLECTResults of Pearsons Chi Square Test of Association
between REFLECT & Dtotalp value: 0.1924Pearsons Chi Square
statistic: 69.303Degrees of Freedom (df): 60p is NOT statistically
significantD total
REFLECT versus E.total subset by CTC (colour as CTC value)
REFLECTResults of Pearsons Chi Square Test of Association
between REFLECT & Etotalp value: 0.1224Pearsons Chi Square
statistic: 86.133Degrees of Freedom (df): 72p is NOT statistically
significantE total
FLOW dataMost tasks appeared in FLOW segment
students comments: REFLECTIONBeing flexible.When I saw other
solutions, I felt that flexibility can be so helpful and important
to make smart solutions. There are various ways to make robot turn,
such as using time, sensors, number of rotations, and so on. But
they can be easily affected by some conditions, like friction,
structure of circuits, sensor's feeling.Thinking differently, being
creative, taking risks and being aware of taking risks.I tried to
make several parts that were quite different from what I usually
make.Seeing different idea in same task was very interesting and
impressive for me! This would absolutely make us more productive
because the more challenges we do, the more ideas we can see, get,
and make!Comparing ideas and other points of view.Through this
challenge, I talked and had communication with foreign students for
the first time. Then I could see different views and ideas from
them.When it comes to the solution for today's task, I got
interesting idea from my partner. He thought and said that moving a
robot in entire circuit to find balls can be done by moving it,
like drawing smaller circles or spirals. This was quite different
idea from mine. I could see another point of view.Perseverance:
start again and remake.I and Keitaro mainly organized the circuit.
We could do that well, but making proper one wasn't easy to do. We
tried to make it according to our first plan. But we needed add
some changes to move our robot smoothly. Our robot is a little bit
big, so it requires, kind of, big space to go around. We measured
distances to make enough spaces, made sure walls are tall enough,
tried to choose proper stuffs for circuits, such as blocks, boxes,
and boards. It was difficult to make a robot which attached any
sensors. I may be going to remake that robot.
students comments: REFLECTIONPersistent challengesSensorsUsing
color sensor to balls that could be moved when detecting was very
difficult for me. We needed to adjust the values of motors and
sensors, accurately and properly, but we couldnt.AccuracyIn
operation of the irregular motor, the difference between normal
operation was hard to understand.TimingFinishing things in time can
be quite difficult. It's like my objective.Sensors, Accuracy &
TimingI could imagine the solution and the program I had to make.
But completing wasn't easy. I guess we were so close to success.
However, we couldn't finish and had some difficulties in adjusting
values, making sure the work of sensors, and adding proper program
blocksCommunication: person to personI had to explain about our
solution, what we tried to do, and functions of our robot to
foreign students, not only our members. That was quite difficult
for me because situations seemed to be so different. I wasn't sure
whether I could tell them well. Actually I was kind of worried.
observationsH1: Learners actively engaged in iteratively
challenging robot-mediated interactive tasks develop generic,
declarative and epistemic competencies.Quantitative dataRTC mainly
increases with increased CTC; though not statistically
significant.Generic, declarative and epistemic competencies are
mostly patterned per task.Qualitative dataStudents comments were
categorized as being flexible (E), thinking differently (G), being
creative (G), taking risks and being aware of taking risks (E),
comparing ideas and others points of view (D), perseverance (E).The
hypothesis can be considered partially true.
observationsH2: Increased meta-learning is associated with the
development of generic, declarative and epistemic
competencies.Whatever the CTC value, as GENERIC competency
increases so do REFLECT values; though not statistically
significant.DECLARATIVE competency & REFLECT values are not
statistically significant. However, with high CTC values,
DECLARATIVE competency and REFLECT values are high.EPISTEMIC
competency & REFLECT values are not statistically significant.
high - medium - low CTC values have no impact on EPISTEMIC
competency. Students did REFLECT but little evidence of UNDERSTAND
and APPLY their reflections.The hypothesis can be considered
partially true for GENERIC and DECLARATIVE competencies but not
true for EPISTEMIC competency.
observationsH3: Students need to be in a state of flow for
meta-learning to occur.Students data recorded 14 of 20 tasks in
FLOW segment. However, no pattern appears linking GENERIC,
DECLARATIVE or EPISTEMIC competency with Tasks in FLOW segment.T4
T6 T9 T15 T16 T20 outside FLOW segment.REFLECT values are variable
(high - medium - low)No pattern appears to suggest FLOW is required
for REFLECT; there is no statistical significance.Results of
Pearsons Chi Square Test of Association between REFLECT & FLOWp
value: 0.08836Pearsons Chi Square statistic: 102.0Degrees of
Freedom (df): 84p is small but NOT statistically significant (less
than 0.01 required for statistical significance)
The hypothesis is therefore not true.
The metrics for meta-learningmeasure - over time - students
self-reporting of GENERIC COMPETENCIES thinking differently (G),
being creative (G)DECLARATIVE COMPTENCIEScomparing ideas and others
points of view (D),EPISTEMIC COMPTENCIESbeing flexible (E), taking
risks and being aware of taking risks (E), perseverance (E).
students will need help in raising such awareness
SO WHAT? The research has shown that undertaking tasks that
promote particular competencies support the development of a
self-directed learner;
engaging students in creative designs, engineering application,
use of programming logic, and reflection-on-learning, support the
development of a self-directed learner.
This can be implemented at High School and early university
undergraduate education by tasks of quantifiable complexity using
LEGO Mindstorms EV3 robots.
SO WHAT? It is concluded that A learner who becomes aware of
his/her knowledge, gaps in knowledge, and what to do in order to
enhance knowledge (i.e meta-learning) is more prepared for dealing
with uncertainty.
However, students need more guidance transferring that learned
to other scenarios (subjects, situations, unique challenges).
Future University Hakodate, JapanMeta-learning with EV3
Mindstorms
iVERG labwww.mvallance.netEmail: [email protected]
