Eye-tracking has been widely used for research purposes in fields such as linguistics and marketing. However, there are many possibilities of how eye-trackers could be used in other disciplines like physics. A part of physics education research deals with the differences between novices and experts, specifi-cally how each group solves problems. Though there has been a great deal of research about these differences there has been no research that focuses on noticing exactly where experts and no-vices look while solving the problems. Thus, to complement the past research, I have created a new technique called gaze scrib-ing. Subjects wear a head mounted eye-tracker while solving electrical circuit problems on a graphics monitor. I monitor both scan patterns of the subjects and combine that with videotapes of their work while solving the problems. This new technique has yielded new information and elaborated on previous studies.

1. Gaze Scribing in Physics Problem Solving
David Rosengrant
Kennesaw State University
1000 Chastain Road, Kennesaw GA, 30144
drosengr@kennesaw.edu
Abstract Combining this methodology with eye-trackers creates a new
research method I call gaze scribing. Subjects wear a head
Eye-tracking has been widely used for research purposes in mounted eye-tracker while they write out solutions on a graphic
fields such as linguistics and marketing. However, there are monitor. Videotaping and audiotaping the student’s work, sav-
many possibilities of how eye-trackers could be used in other ing their written data, listening to their discussion and analysis
disciplines like physics. A part of physics education research of their scan paths [Duchowksi 2007] are all necessary compo-
deals with the differences between novices and experts, specifi- nents for this gaze scribing.
cally how each group solves problems. Though there has been a
great deal of research about these differences there has been no This initial study focuses specifically on electrical circuits.
research that focuses on noticing exactly where experts and no- Problem solving in physics varies depending on the sub-
vices look while solving the problems. Thus, to complement the disciplines (mechanics, dynamics, electricity, et al). I have fo-
past research, I have created a new technique called gaze scrib- cused on electrical circuits for this study because there is a
ing. Subjects wear a head mounted eye-tracker while solving unique combination of qualitative understanding of representa-
electrical circuit problems on a graphics monitor. I monitor both tions and quantitative reasoning abilities. Furthermore, there has
scan patterns of the subjects and combine that with videotapes of been no work to date utilizing eye-trackers and expert-novice
their work while solving the problems. This new technique has differences with electrical circuits.
yielded new information and elaborated on previous studies.
2 Previous Work
CR Categories: J.2 [Computer Applications]: Physical Science
and Engineering - Physics Novices and experts approach problem solving differently. One
such difference is the search techniques they use to solve a prob-
Keywords: gaze scribing, education research, physics problem lem [Larkin et al 1980] or the strategy they use to solve a prob-
solving lem [Chi et al 1981]. Novices typically write down the known
and unknown variables. Next, they use a backward inference
1 Introduction technique - a search for equations involving variables they think
they can use. This is commonly called plug and chug. Experts
One of the goals of many education researchers is to narrow the use a forward inference technique. They first determine the con-
gap between experts and novices [Chi et al 1981; Feltovich and cept and the key features associated with the problem to deter-
Glaser 1981; Kindfield 1994; Kozma and Russell 1997; Styliano mine how they want to solve the problem. Furthermore, novices
and Silver 2004]. In Physics, the gaps include but are not li- categorize problems by the surface features of the problem as
mited to how both groups understand and comprehend the con- opposed to experts who again focus on underlying concepts
tent material, how they learn new material and how they solve [Kozma and Russell 1997].
problems. Problem solving is not always finding a numerical
answer; it can include qualitative solutions such as using simula- What is also important is what happens during the problem solv-
tions to explain how microwave ovens work. For this study, I ing process. For example, if an expert or novice gets stuck dur-
focus specifically on the problem solving gap.. ing a problem, how they get unstuck is vastly different. Experts
typically can perform a qualitative analysis of their work or they
First, we must understand how experts and novices learn and can use some other method to aid them in the problem solving
solve problems in order to help students (novices) become ex- process. If novices get stuck, they typically only manipulate
perts. Traditionally, problem solving studies have involved case equations or seek outside help to get them unstuck [Gerace
studies with one-on-one interviews with students. The inter- 2001]. Another key difference is that experts are able to check
viewer would give the subject a task and use a think aloud pro- their solutions, possibly via another method such as multiple
tocol [Ericsson and Simon 1998] to determine the subjects representations or alternate mathematical equations. Novices
thought processes while solving the problems. however are only able to find a solution via one method.
These differences are common to all areas of physics, but since
this study focuses on electrical circuits there are key differences
between experts and novices that are specific to this sub-
Copyright © 2010 by the Association for Computing Machinery, Inc.
Permission to make digital or hard copies of part or all of this work for personal or discipline. In direct current (DC) circuits, current is a major
classroom use is granted without fee provided that copies are not made or distributed conceptual challenge to novices. They tend to believe that cur-
for commercial advantage and that copies bear this notice and the full citation on the rent gets “consumed” when moving through a circuit and that
first page. Copyrights for components of this work owned by others than ACM must be
parallel branches split current equally throughout the branches
honored. Abstracting with credit is permitted. To copy otherwise, to republish, to post on
servers, or to redistribute to lists, requires prior specific permission and/or a fee. regardless of the arrangement of the resistors [Duit and Rhonek
Request permissions from Permissions Dept, ACM Inc., fax +1 (212) 869-0481 or e-mail 1997]. Novices also believe that the battery is a constant source
permissions@acm.org.
ETRA 2010, Austin, TX, March 22 – 24, 2010.
© 2010 ACM 978-1-60558-994-7/10/0003 $10.00
45

2. 4
4 3 2 2
1
0.5
5
5
1
A
9
9
4
B
0.5
5
0.5
5 1
Figure 1: 4 Circuits given to the students
of current [Engelhardt and Beichner 2004]. Their difficulties them with any quantitative analysis if needed. Subjects received
with current are compounded by the fact that they interchange simulations of the last circuit to assist with qualitative questions.
the ideas of “current” and “voltage” [Metioui 1996]. They be- Figure 2 is an example of one subject’s work and a visual of
lieve that circuits are a system of pipes that allow a fluid called what the subject’s workspace looked like.
electricity to flow through them [Johsua 1984]. These difficul-
ties become even more noticeable when one incorporates series Some of the questions we asked the subjects only required an
and parallel sections of circuits. Even identifying which compo- auditory response. For example, how does the current compare
nents of a circuit are in parallel or in series are a challenge for going through specific resistors. However, others required a
many students. numerical response. In every case, the subjects needed to find
the net resistance of the circuit. All of the subjects’ work was
3 Sample and Setting conducted on the graphic tablet monitor in the paint file.
Each subject wore a head mounted eye-tracker while they ans-
This study was conducted at a suburban university of about wered the questions. The eye tracker was an Applied Science
21,000 students in the second semester of a two-semester alge- Laboratories Model 6000 Mobile Control Unit that included an
bra based physics course. The students were typically Biology Applied Science Laboratories head-mounted optics unit with
or Health and Exercise Science majors. The students in this scene camera. The subjects sat at arm’s length from the screen
course had already been taught about electrical circuits. so that they could write the answers on the monitor. The setup
is show in Figure 3. A head mounted unit assured that the arm
Eleven subjects participated in the case study. Nine of the sub- would not block any cameras while writing answers.
jects were considered novices. They were students in the above
mentioned course. Two others were considered experts who The graphics monitor was part of a 2 monitor setup. The inter-
were physics faculty at the university. viewer sat at the second display monitor. This allowed the in-
terviewer to easily change the circuits the subjects worked on as
4 Methodology well as what tools the subjects could use.
Each subject received a series of questions based on the circuits A video camera (in addition to the camera from the eye-tracker)
found in Figure 1. Each subject received each of the four cir- recorded the entire interview. Each interview was also audio-
cuits one at a time. The circuits were in a Microsoft paint doc- taped. The use of the multiple recording devices allowed me to
ument. This allowed the subjects to write their work next to the analyze the data several ways. First, I was able to listen to what
circuit since the monitor was a graphics display monitor. Sub- the students verbalized as they solved the problem. I was also
jects also received a calculator on the computer screen to assist able to tell what the students were calculating / drawing on the
monitor while simultaneously knowing where subjects looked
while they solved the problem.
Figure 2: Screenshot of student’s work. Figure 3: Screenshot of setup.
46

3. 5 Findings they were done. Novices, in some cases only looked at the last
bit of mathematical work but generally did not show any evi-
The information uncovered about expert-novice differences on dence of comparing their answer with their work or the given
electrical circuit problems both reinforced previous studies and circuit.
uncovered new data. However, the focus of the findings here is
on how the gaze scribing and eye-tracking supplemented the Another difference between the experts and the novices was how
previous work of other physics education researchers. More they initially looked at the circuit. Most of the subjects looked
information about the other differences can be found in Rosen- at circuit 2 in a fashion similar to what is shown in Figure 4.
grant et al [Press]. When the subjects analyzed the circuit, they would simply go
from one resistor to the next following the shortest path between
One of the first questions I asked the subjects was to calculate the resistors. This was common among both the experts and the
the net resistance of the circuit. This involved a quantitative novices. However, one of the experts sometimes exhibited a
answer with a qualitative understanding of how circuits worked. different behavior. This expert was also a stronger expert in this
One of the traits I investigated was the idea that experts evaluate field because he taught the electronics course at the university.
their work while solving problems while novices do not. The The expert followed a path similar to what is shown in Figure 5.
gaze scribing reinforced this finding. Both the experts and the In this circuit, it appears that the expert followed the path of the
novices used the provided extra space to write out their calcula- current throughout the circuit.
tions and eventually a solution. Both the experts and the novices
would look back at the circuit to double check the value of the
resistors in their mathematical formulas. However, from that
point on the novices would only focus on their mathematical
work until they arrived at a solution. They would look back and
forth in their work, but not back to the circuit. The experts on
the other hand would gaze back and forth between the given
circuit, their work and circuits they may have redrawn to help
them solve the problem. This difference is not something that
could be noticed during a normal problem solving session.
There were other more specific patterns that emerged during this
evaluation. For example, in a parallel circuit containing two or
more resistors, the total resistance of that portion of the circuit
must be less than any of the resistors making up the parallel
portion of the circuit. After experts calculated the net resistance
for the parallel portion of the circuit they would look back at the
circuit. Specifically they would look back and forth at the val- Figure 5: Expert partial gaze path of circuit 2
ues of the resistors in parallel and their answer. Novices did not
evaluate their work in this respect; they simply focused on what This is important because none of the novices exhibited this
they were writing and then continued on to the next step in sim- behavior. This scan-path ties in with previous research that
plifying the circuit. They would look back and forth among states that one of the issues novices have with circuits is simply
their calculations and formulas, but not back to the original cir- understanding how current flows through a circuit.
cuit.
A final piece of interesting information that figure 4 also shows
When the experts found the resistance for the total circuit, they us is that both the experts and the novices would group resistors
then looked back at all of their work, back at any circuits they together when they first analyzed the circuit. The resistors
constructed and then to the original circuit while also looking would be grouped according to how they were arranged, either
back at their calculated answer before they exhibited signs of in series (such as the 8 and 16 ohm resistors shown in figure 4)
being finished with the problem. These signs include but are not or in parallel (the 3 and the 5 ohm resistors are connected in
limited to leaning back in their seat, looking away from the parallel to the 4 ohm resistor). As the groups were writing out
computer screen or saying that was their final answer or that their mathematical equations on how to add the resistors in pa-
rallel or in series both groups would look back to the relevant
parts of the circuit.
6 Discussion
The gaze scribing provides a unique opportunity to analyze
problem solving behaviors. Instead of relying on verbal res-
ponses and written work from experts and novices, now we can
also monitor their scan paths while they are solving problems.
The differences in the scan paths while subjects are solving
physics problems can lead to new interpretations and under-
standings of the differences between the two groups.
This preliminary study is only focused on one type of problem
Figure 4: Sample gaze path of circuit 2 solving scenario. Solving problems with electrical circuits al-
lows subjects to find a quantitative solution but it also involves
47

4. the construction of multiple representations (other circuits). METIOUI, A., BRASSARD, C., LEVASSEUR, J. AND LA-
Representations also have many differences between how ex- VOICE, M. 1996. The Persistence of Students’ Unfounded
perts and novices use and construct them [Rosengrant et al Beliefs About Electrical Circuits: The Case of Ohm’s Law.
2006].
International Journal of Science Education, 18, 2, 193-212.
There are many types of problems students solve in physics.
Gaze scribing is a research technique that can be implemented ROSENGRANT, D., ETKINA, E., AND VAN HEUVELEN, A.
for these other problem types such as mechanics problems, ray 2007. An Overview of Recent Research on Multiple Represen-
tracing with thin lens, online simulations and how students use tations. In Proceedings of the 2006 Physics Education Research
them, etc. These are all areas of future research that combine Conference, L. McCullough, P. Heron, and L.Hsu, Eds., Physics
eye-tracking with physics education research. This type of me-
Education Research Conference, Annual Conference, 149-152.
thodology can also extend beyond just physics. Gaze scribing
could be used in any discipline where subjects need to write out
solutions. ROSENGRANT, D., THOMSON, C., AND MZOUGHI, T.,
(PRESS). Comparing Experts and Novices in Solving Electrical
Circuit Problems With the Help of Eye-Tracking. Submitted to
the Proceedings of the 2009 Physics Education Research Confe-
References rence.
CHI, M. FELTOVICH, P., AND GLASER R. 1981. Categoriza- STYLIANO, D. AND SILVER, E. 2004. The Role of Visual
tion and Representation of Physics Problems by Experts and Representations in Advanced Mathematical Problem Solving:
Novices. Cognitive Science 5, 121-152.
An Examination of Expert-Novice Similarities and Differences.
DUCHOWSKI, A. 2007. Eye Tracking Methodology: Theory Mathematical Thinking and Learning, 6, 4, 353-387.
and Practice, 2nd Ed. Springer-Verlag.
DUIT, R., AND RHONEK, C. 1997 Learning and Understand-
ing Key Concepts in Electricity. In Connecting Research in
Physics Education with Teacher Education, A. Tiberghien,
E.L.Jossem, and J. Barojas, Eds., The International Commission
on Physics Education, 1-6
ENGELHARDT, P., AND BEICHNER, R 2004. Students’
Understanding of Direct Current Resistive Electrical Circuits.
American Journal of Physics, 72, 98-115.
ERICSSON, K., AND SIMON, H. 1984. Protocol Analysis:
Verbal Reports as Data. Cambridge, MA: MIT Press
GERACE, W. 2001. Problem Solving and Conceptual Under-
standing. In Proceedings of the 2001 Physics Education Re-
search Conference, S. Franklin, J. Marx, K. Cummings, Eds.,
Physics Education Research Conference, Annual Conference,
33-36.
JOHSUA, S. 1984. Students’ Interpretation of Simple Electrical
Circuit Diagrams, International Journal of Science Education, 6,
12. 271-275.
KINDFIELD, A. 1994. Understanding a Basic Biological
Process: Expert and Novice Models of Meiosis, Science Educa-
tion, 78, 255-283.
KOZMA, R. AND RUSSELL, J. 1997. Multimedia and Under-
standing: Expert and Novice Responses to Different Chemical
Representations of Chemical Phenomena. Journal of Research
in Science Teaching, 34, 9, 949-968.
LARKIN, J., MCDERMOTT, J., SIMON, D. AND SIMON, H.
1980. Expert and Novice Performance in Solving Physics Prob-
lems. Science, 208, 1335-1342.
48