Vous êtes sur la page 1sur 8

Ergonomics

ISSN: 0014-0139 (Print) 1366-5847 (Online) Journal homepage: http://www.tandfonline.com/loi/terg20

Evaluating the ergonomics of BCI devices for


research and experimentation

Joshua I. Ekandem , Timothy A. Davis , Ignacio Alvarez , Melva T. James &


Juan E. Gilbert

To cite this article: Joshua I. Ekandem , Timothy A. Davis , Ignacio Alvarez , Melva T.
James & Juan E. Gilbert (2012) Evaluating the ergonomics of BCI devices for research and
experimentation, Ergonomics, 55:5, 592-598, DOI: 10.1080/00140139.2012.662527

To link to this article: http://dx.doi.org/10.1080/00140139.2012.662527

Published online: 16 Apr 2012.

Submit your article to this journal

Article views: 1074

View related articles

Citing articles: 16 View citing articles

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=terg20

Download by: [University of Valencia] Date: 16 January 2017, At: 04:48


Ergonomics
Vol. 55, No. 5, May 2012, 592598

Evaluating the ergonomics of BCI devices for research and experimentation


Joshua I. Ekandema, Timothy A. Davisb*, Ignacio Alvareza, Melva T. Jamesa and Juan E. Gilberta
a
Human Centered Computing Division, School of Computing, Clemson University, Clemson, SC 29634, USA; bVisual Computing
Division, School of Computing, Clemson University, Clemson, SC 29634, USA
(Received 7 June 2011; nal version received 26 January 2012)

The use of brain computer interface (BCI) devices in research and applications has exploded in recent years.
Applications such as lie detectors that use functional magnetic resonance imaging (fMRI) to video games controlled
using electroencephalography (EEG) are currently in use. These developments, coupled with the emergence of
inexpensive commercial BCI headsets, such as the Emotiv EPOC (http://emotiv.com/index.php) and the Neurosky
MindWave, have also highlighted the need of performing basic ergonomics research since such devices have usability
issues, such as comfort during prolonged use, and reduced performance for individuals with common physical
attributes, such as long or coarse hair. This paper examines the feasibility of using consumer BCIs in scientic
research. In particular, we compare user comfort, experiment preparation time, signal reliability and ease of use in
light of individual dierences among subjects for two commercially available hardware devices, the Emotiv EPOC
and the Neurosky MindWave. Based on these results, we suggest some basic considerations for selecting a
commercial BCI for research and experimentation.

Statement of Relevance: Despite increased usage, few studies have examined the usability of commercial BCI
hardware. This study assesses usability and experimentation factors of two commercial BCI models, for the purpose
of creating basic guidelines for increased usability. Finding that more sensors can be less comfortable and accurate
than devices with fewer sensors.
Keywords: brain-computer interface (BCI); comfort; usability

1. Introduction
Brain computer interfaces (BCIs) are physical devices that allow one-way communication from the human brain to
an external machine by interpreting electroencephalography (EEG) brain wave data. While such devices have been
studied since the 1970s (Vidal 1973, 1977), research using BCIs has recently experienced a surge in popularity across
a diversity of disciplines, including mathematics, computer science, psychology, physical rehabilitation, design and
policy studies (Bimber 2008, Harris 2010, Peck et al. 2010). One primary reason for this new wave of interest is the
advent of BCI device accessibility. Although EEG monitoring was once an expensive procedure performed in a
controlled environment, BCI devices today are low cost and portable.
Much of the BCI research to date has focused on several topics: data acquisition and signal processing (Stern et al.
2001, Andreassi 2007, Cichocki et al. 2008, McFarland and Wolpaw 2011), theoretical foundations of BCI including
vocabulary and methodology, and conceptual applications or novel prototypes (Palaniappan and Mandic 2007, Lin
et al. 2008, Campbell et al. 2010, Peck et al. 2010). While these topics are necessary to ensure the reliability of BCI
technology, basic research that focuses on the ergonomics of the devices is scarce. Specically, studies are needed to
explore how issues such as comfort, limits of use and physical characteristics of the user aect the suitability of BCIs for
formal experimentation. Such studies may be essential for future research to unlock the full potential of BCI devices for
a wide variety of applications.
BCI headsets were designed to be more user friendly than research-grade BCI devices and have gained
popularity among researchers and developers. For example, the Neurophone is a novel application that uses a
commercial-grade wireless EEG headset, the Emotiv EPOC, for dialing phone numbers on a smartphone (Campbell
et al. 2010). Similarly, the Neurosky MindSet, another commercially available wireless EEG headset, was used as a
physiological research tool to measure attention and meditation levels (Crowley et al. 2010). Finally, research has
been conducted to test and evaluate the benets of prototype wireless EEG headbands (Lin et al. 2008).
These commercial-grade BCI devices are non-invasive EEG headsets that promise ease of use, low cost, and
untethered mobility, which are acknowledged by researchers and developers alike as highly desirable (Campbell

*Corresponding author. Email: tadavis@cs.clemson.edu

ISSN 0014-0139 print/ISSN 1366-5847 online


2012 Taylor & Francis
http://dx.doi.org/10.1080/00140139.2012.662527
http://www.tandfonline.com
Ergonomics 593

et al. 2010). Benets gained, however, are often oset by low signal reliability (Campbell et al. 2010, McFarland and
Wolpaw 2011), fatigue associated with extended wear, and in some cases, diculty of use. These issues bring into
question the suitability of such devices for both research and real-world applications.
This exploratory study examines the feasibility of using commercially available consumer-grade BCI devices
for formal research and experimentation from both the experimenter and participants perspectives. For the
experimenter, feasibility concerns include costs associated with the device, ease of use, signal reliability and
length of preparation time. While a participant may also desire signal reliability and, in turn, eectiveness of the
device in interpreting brain waves to perform tasks on the computer, another important consideration is
comfort while wearing the headset. In this study, we examine these factors as they apply to a heterogeneous
population of subjects with varying physical characteristics using two commercially available hardware models:
the Emotiv EPOC and the Neurosky MindWave. The data collected during this preliminary qualitative
study are used to determine the suitability of such devices in research studies where prolonged use may be
required.

2. Materials and methods


2.1. Apparatus
Four commercially available BCI devices have been released to date: the OCZ neural impulse actuator (NIA; http://
www.ocztechnology.com/nia-game-controller.html), the Emotiv EPOC, the NeuroSky MindSet and the NeuroSky
MindWave. The OCZ NIA is a wired BCI device designed primarily as a gaming controller with limited capabilities.
It has been discontinued by the manufacturer and is not included in this study.
The Emotiv EPOC is a wireless BCI device that uses 14 active sensors placed across the participants head to
monitor changes in electrical signals from the brain. Each sensor contains a felt pad that must be dampened with
saline solution before positioning on the head. Further, each of these sensors must be placed in direct contact with
the scalp. The Emotiv Control Panel, the software application bundled with the hardware, provides user training as
well as visual feedback for proper placement of the sensors.
The NeuroSky MindSet was the rst multi-purpose commercial BCI device created for both research and
applications in areas such as education, gaming and assistive tools for the disabled. The structure of the MindSet
device is similar to that of a high-quality stereo headset with a primary sensor positioned on the boom microphone
for placement on the forehead, and secondary sensors located on one ear piece to contact the area close to the left
temple. Similar to the EPOC, the MindSet is a wireless device that is bundled with software for training and gaming,
in addition to an optional developers kit for creating new applications.
More recently, NeuroSky has released a second-generation BCI headset model, the MindWave, that is also
promoted as a research-grade apparatus. The new model appears more rened in its physical design and operation;
however, the MindWave diers from the MindSet in several fundamental ways that aect the ergonomics of the device.
First, unlike its predecessor, which contained size adjusters on the headset, the MindWave is a one-size-ts-all device
from child to adult. Second, the primary sensor on the MindSet is rounded on the contact point, while the MindWave
uses a at rectangular sensor, which must be pressed directly against the forehead. Finally, the secondary contact points
on the MindSet are located on a cushioned earphone, while the MindWave employs secondary sensors in an ear lobe
clip. All of these features undoubtedly aect the ergonomics of the MindWave model.
In this study, ergonomics tests were performed on the Emotive EPOC and the Neurosky MindWave. These two
devices were chosen since they are the current state-of-the-art BCI products specically designed for research
purposes. Both headsets are wireless, non-invasive and allow for multiple hours of continued use. Further, simple
games are included with both that allow the user to move items on the screen simply by focusing thought on certain
tasks. For our tests, each device interfaced with a Dell Optiplex 780 running Windows 7 through a Bluetooth
adaptor.

2.2. Participants
The experiment consisted of 13 volunteer participants, nine male and four female, that ranged in age from 20 to 33
years old (average and median age both 25). Participants possessed various hair lengths and textures: ve (38.4%) had
short-ne hair, one (7.6%) had short coarse hair, one (7.6%) had medium ne hair, two (15.3%) had long coarse hair,
two (15.3%) had long ne hair and two (15.3%) had no hair at all. For this study, short hair is dened as roughly three
inches or less, medium hair is roughly three to six inches, and long hair is roughly six inches or greater in length. Hair
texture is not easily dened quantitatively and is used only as a subjective descriptor.
594 J.I. Ekandem et al.

2.3. Procedure
The 13 participants consented to test the two BCI devices in a two-day study. Participants were arbitrarily divided
into two groups, with 11 of the 13 completing both trials (the two participants who did not complete both trials were
dropped from the results). The rst group tested the Emotiv EPOC on the rst day, while the second group began
with the Neurosky MindWave. On the second day of the experiment, the rst group tested the MindWave, while the
second group worked with the EPOC. This approach allowed us to minimise bias, task fatigue, and residual eects
of discomfort across the two devices.Each trial began with the experimenter placing the headset on the participant.
This approach was taken for several reasons. First, the participants were largely unfamiliar with the BCI devices and
their proper placement. Second, proper headset placement is important to attain at least a minimal threshold
contact for accurate brain wave monitoring. Finally, this approach may be generally desired by researchers for
controlled experiments where accuracy of results are desired. In any case, after initial placement, the participant was
free to work with the experimenter to adjust the headset further for comfort while achieving at least 75% (e.g., 11 of
the 14 active sensors) active contact between the BCI sensors and the participant. Active contact in this context is
dened as the percentage of sensors registering full contact with the scalp on the devices monitor software (see
Figure 1).
Each experimenter received uniform training and followed the same protocol for placing, adjusting and
removing the headset with feedback from the participant. Total time from initial placement to nal adjustment was
recorded as part of the preparation time for each device. At this point, the participant completed a training phase
before engaging in several simple games included with each system (e.g., Pong, Tetris, and SpadeA). Figure 2 shows
two subjects engaged in the experiment. After wearing the BCI device for 15 minutes, the participant completed a
post-experiment comfort survey.

Figure 1. Emotiv sensor indicator.

Figure 2. Participants using BCI devices: Emotiv EPOC (left) and NeuroSky MindWave (right).
Ergonomics 595

3. Results
Although the experimenters had considerable experience using both BCI devices, the average preparation time of
the Emotiv EPOC was 99 seconds longer than the Neurosky MindWave. Specically, the MindWave required a
mean preparation time of 17 seconds [standard deviation (SD) 7.7986] while the EPOC had a mean preparation time
of 116 seconds (SD 100.1). A major contributing factor to the dierence in times was the number of sensors needed
to acquire a 75% signal. Specically, preparing and adjusting the 14 active electrodes and two reference sensors on
the EPOC took considerably longer than the one active sensor and ear lobe clip on the MindWave.
An SD of 101 seconds for the preparation time of the Emotiv EPOC suggests that factors other than the number
of electrodes contribute to the preparation time of the device. Further investigation shows that, for some
participants, the time needed to acquire a minimal signal was much longer than for the others. This discrepancy
correlated directly to the length and density of the participants hair. As expected, participants with longer and
coarser hair required more preparation time. Alternatively, the shortest preparation times were recorded for
participants with no hair. These results highlight the sensitivity of sensor placement to hair texture and length.
As mentioned previously, signal strength of 75% was maintained throughout the experiment. In the majority of
the cases, the signal strength met or exceeded this threshold; however, for 17% of the cases (four of the 24 trials), the
signal level was below the threshold. In all four cases, the participants were females using the Emotiv EPOC. The
strict requirement of sensor contact with the scalp simply prohibits certain users (i.e., those with thick or long hair)
from using the device eectively. Indeed, for some of our participants, we were not able to achieve signal strength
above zero. Since we were not testing participant performance using BCI devices, we allowed these participants to
complete the experiment simply by wearing the BCI headset for the required time and reporting their subjective
experiences.
After the experiment, the participants were asked to complete a survey that rated the overall comfort of the device
(very uncomfortable, uncomfortable, indierent, comfortable and very comfortable), the length of time the participant
felt they could comfortably wear the device (05 minutes, 520 minutes, 2060 minutes, 60120 minutes and more than
120 minutes) and the type of discomfort perceived (sharp, dull, itchy, heavy, throbbing, awkward, burning or other).
The overall average comfort score for both devices fell well within the comfortable range; however, the majority of
participants (nine out of 11, or 81%) indicated that the Emotiv EPOC was comfortable, with six of the participants
(54%) explicitly indicating that the Neurosky MindWave presented some degree of discomfort.
Although more than one third (three of 11) of the participants felt some degree of discomfort using either BCI
device during the experiments, the majority of the complaints (75%) were aimed at the NeuroSky MindWave
(Figure 3(A)). Many of the participants complained that the MindWave felt unusually awkward to wear. These
participants indicated that these feelings of discomfort primarily related to the pinching sensation of the ear clip and
the pressure of the front head sensor when blinking. One participant reported a sharp pain, another detailed a
throbbing feeling at the points in contact with the device and yet another stated that the device felt too heavy. The
Emotiv EPOC, on the other hand, appeared to feel more comfortable, according to participants comments. These
results may be due in part to the exible design of the EPOC, as opposed to the MindWaves one-size-ts-all design,
which was reported to be too tight.
Participants that were indierent, or felt that the device provided some degree of comfort, stated they would be
able to wear the device for at least 20 minutes. As Figure 4 shows, half of the participants believed that they could
wear the device for at least an hour.

Figure 3. (a) Comfort of BCI device. (b) Percentage of average comfort.


596 J.I. Ekandem et al.

Figure 4. Percentage distribution of the estimated time participants reported that they could wear a BCI device.

4. Discussion
Both of the BCI devices tested had advantages and disadvantages in terms of suitability for research and
experimentation. These considerations make neither device a clear winner over the other. Depending on the
requirements of a specic research study, however, a single advantage or disadvantage may ultimately favour one
device over the other.
In accordance with our results, the NeuroSky MindWave is well suited for multi-participant studies since the
time needed to prepare the hardware for a trial and adjust the headset on the participant is relatively short. After
each trial, the headset may require cleaning with a disinfecting wipe; however, such a process consumes only a small
amount of cost and time. Conversely, the Emotiv EPOC requires the application of saline solution to each of its 14
sensor pads before placing the headset on the participant, which takes additional time. Also, the issue of the pads
themselves must be considered. Although the pads appear to be simple cotton discs sized to the sensor area of the
device, they can be quite expensive to purchase from Emotiv, the sole supplier of this product. Further, when
experimenting with a large number of participants, this cost may be signicant since re-using pads, especially wet
pads in contact with the scalp, across participants is not practical due to hygiene issues. To alleviate such expenses,
low-cost alternatives could be explored, such as fabricating suitable replacements using readily available
pharmaceutical products.
Another notable factor for the experimenter is the signal quality of each device. The MindWave allows for easier
signal acquisition, as its sensors are not typically impeded by obstacles such as hair on the forehead and ear lobe. As
demonstrated in this study, the EPOC clearly has issues with contact on participants with obstructing hair. Once the
EPOC is connected and calibrated, however, we found that the signal was maintained and even improved during the
session, while the MindWave experienced more signal uctuations. The EPOC thus may be better suited for longer
experiments where preparation time between subjects is not a signicant issue.
Another contributing factor to obtaining a reliable signal on wireless devices is battery power. The MindWave
works with disposable batteries, while the EPOC uses a rechargeable battery. In our study, we found that signal
levels were higher when using disposable batteries since rechargeable batteries tend to lose cumulative power over
time and thus become less reliable. One recommendation for BCI manufacturers, therefore, is to design hardware
using either disposable batteries or rechargeable batteries that are interchangeable. This feature would allow
researchers to easily replace the battery when low signal issues occur.
From the participants perspective, comfort was a major factor in using the BCI headsets. As shown in our
results, the EPOC, though overall larger with more sensors and wet pads, was evaluated as more comfortable than
the MindWave. As noted previously, the main comfort issues with the MindWave include the tightness of the one-
size-ts-all style, the pressure of the rectangular forehead sensor, and the pinching of the ear lobe clip, all of which
are modications from the rst-generation design. Interestingly, comfort results for the MindSet from an earlier
study (Rebolledo-Mendez et al. 2009) are much more favourable than those from this study for the MindWave,
indicating that Neurosky may have taken a step backward in this regard.

5. Conclusion
BCI technology has an incredible potential to revolutionise the way humans interact with machines. Many everyday
applications in education, gaming and assistive tools can greatly benet from this technology; however, all of these
Ergonomics 597

possibilities begin with basic research (McFarland and Wolpaw 2011). Although much of the literature related to
BCI devices examines technical issues or conceptual applications of BCIs, empirical research concerning ergonomic
considerations has been limited, especially with regard to consumer-based BCIs used in research studies. This study
is aimed at identifying and understanding the possible usability problems that arise from the interaction of these
devices.
As our results suggest, both the Emotiv EPOC and the Neurosky MindWave are suitable devices for use in BCI
research; however, tradeos exist in terms of participant comfort, device preparation time and signal reliability, as
presented in the Discussion section. Experimenters must make informed decisions among these factors to determine
which device would be most suitable for a specic study. Our hope is that the results presented may aid in that
process.
Though originally designed for gaming, a wide variety of applications developed using commercial BCI
devices demonstrate their applicability to researchers and those with severe neuromuscular disorders. However,
for practical applications and continued research, there are some drawbacks of the two most popular commercial
BCI devices. The most signicant drawback for both devices is the reliability of the signal quality, which is a
signicant impediment when these devices are used for active control. The problems with current BCI devices
and their reliance on direct scalp contact are highly unpractical for many users, especially those with long and/or
coarse texture hair. However, it is our hope that as new systems are developed, some of these issues will certainly
change or be resolved completely. One promising technology is the non-contact sensor, which allows EEG
monitoring through hair, clothing and other obstacles (Grozea et al. 2010). This will denitely make BCI devices
more inclusive, and such breakthroughs will undoubtedly push BCI research to the next level, where new horizons
can be explored.

References
Andreassi, J.L., 2007. Psychophysiology: human behavior and physiological response. New York: Psychology Press.
Bimber, O., 2008. Total recall. IEEE Computer, 41 (10), 3233.
Campbell, A.T., Choudhury, T., Hu, S., Lu, H., Mukerjee, M.K., Rabbi, M., and Raizada, R.D.S., 2010. NeuroPhone: brain-
mobile phone interface using a wireless EEG headset. In: Second ACM SIGCOMM workshop on networking, systems, and
applications on mobile handhelds. New Delhi: MobiHeld.
Cichocki, A., Washizawa, Y., Rutkowski, T., Bakardjian, H., Phan, A.-H., Choi, S., Lee, H., Zhao, Q., Zhang, L., and Li, Y.,
2008. Noninvasive BCIs: multiway signal-processing array decompositions. Computer, 3442.
Croal, N., 2007. Mind control. Newsweek International.
Crowley, K., Sliney, A., Pitt, I., and Murphy, D., 2010. Evaluating a brain-computer interface to categorise human emotional
response. In: 10th IEEE international conference on advanced learning technologies. New York: IEEE Computer Society, 276
278.
Dillow, C., 2010. The worlds rst commercial brain-computer interface. [online]. Available from: http://www.popsci.com/
science/article/2010-03/worlds-rst-commercial-brain computer-interface [Accessed 15 May 2011].
Dirjish, M., 2010. BCI technology set to break commercial ground. Electronic Design, 58 (15), 77.
Ford, J., 2011. Thought-wired allows disabled to control home appliances with mind alone. [online]. From Singularity Hub.
Available from: http://singularityhub.com/2011/05/01/thought-wired-allows-disabled-to-control-home-appliances-with-
mind-alone/ [Accessed 17 May 2011].
Grozea, C., Voinescu, C.D., and Fazli, S., 2010. Bristle-sensorslow-cost exible passive dry EEG electrodes for neurofeedback
and BCI applications. Journal of Neural Engineering, 8 (2), 18.
Guger, C., Bin, G., Gao, X., Guo, J., Hong, B., Liu, T., . . . and Schalk, G., 2011. State of the art in BCI research: BCI award
2010. In: R. Fazel-Rezai, ed. Recent advances in brain-computer interface systems. Rijeka, Croatia: InTech, 193222.
Harris, M., 2010. Liar: can brain scans show whether people are telling the truth. Spectrum, 4053.
Lin, C.-T., Chen, Y.-C., Huang, T.-Y., Chiu, T.-T., Ko, L.-W., Liang, S.-F., Hsieh, H.-Y., Hsu, S.-H., and Duann, J.-R., 2008.
Development of wireless brain computer interface with embedded multitask scheduling and its application on real-time
drivers drowsiness detection and warning. IEEE Transactions on Biomedical Engineering, 55 (5), 15821591.
Lin, C.-T., Ko, L.-W., Chang, C.-J., Wang, Y.-T., Chung, C.-H., Yang, F.-S., Duann, J.-R., Jung, T.-P., and Chiou, J.-C., 2009.
Wearable and wireless brain-computer interface and its applications. In: E. Nishimura, C. Russell, J. Stautzenberger, H. Ku,
and J. Downs, eds. Foundations of augmented cognition. Neuroergonomics and operational neuroscience, Berlin/Heidelberg:
Springer, 504513.
Mason, S.G. and Birch, G.E., 2003. A general framework for brain-computer interface design. IEEE Transactions on Neural
Systems and Rehabilitation Engineering, 11 (1), 7085.
McFarland, D.J. and Wolpaw, J.R., 2011. Braincomputer interfaces for communication and control. Communications of the
ACM, 54 (5), 6066.
Nam, C.S., Li, Y., and Johnson, S., 2010. Evaluation of P300-based brain computer interface in real-world contexts.
International Journal of Human-Computer Interaction, 26 (6), 621637.
Palaniappan, R. and Mandic, D.P., 2007. Biometrics from brain electrical activity: a machine learning approach. IEEE
Transactions on Pattern Analysis and Machine Intelligence, 29 (4), 738742.
598 J.I. Ekandem et al.

Peck, E., Chauncey, K., Girouard, A., Gulotta, R., Lalooses, F., Treacy Solovey, E., Weaver, D., and Jacob, R., 2010. From
brain to bytes. XRDS, 16 (4), 4247.
Rebolledo-Mendez, G., Dunwell, I., Martnez-Miron, E.A., Vargas-Cerdan, M.D., Freitas, S., Liarokapis, F., and Garca-
Gaona, A.R., 2009. Assessing neuroskys usability to detect attention levels in an assessment exercise. In: Proceedings of the
13th International Conference on Human-Computer Interaction. Part I: New Trends. Berlin: Springer, 149158.
Stern, R.M., Ray, W.J., and Quigley, K.S., 2001. Psychophysiological recording. New York: Oxford University Press.
Vaughan, T.M., McFarland, D.J., Schalk, G., Sarnacki, W.A., Krusienski, D.J., Sellers, E.W., and Wolpaw, J.R., 2006. The
Wadsworth BCI research and development program: at home with BCI. IEEE Transactions on Rehabilitation Engineering, 14
(2), 229.
Vidal, J., 1973. Toward direct brain computer communication. Annual Review of Biophysics and Bioengineering, 2, 157180.
Vidal, J., 1977. Real-time detection of brain events in EEG. Proceedings of the IEEE. 65, 633641.