Spatial resolution of vibrotactile perception on the human forearm when exploiting funneling illusion

Abstract

Recent advances in man-machine interaction, telerobotics, telepresence and teleaction have shown that introducing the haptic modality to multimedia applications has the power to significantly widen their application range and to dramatically improve the user experience. Especially, the human body surface has been considered as an additional means of presenting information using vibrotactile display devices. In this context, spatial displacement of a vibrotactile stimulus can be deployed for information display. By exploiting a psychophysical illusion called ¿funneling illusion¿, we are able to increase the spatial resolution of vibrotactile displays. In this paper, we aim at investigating the spatial resolution of vibrotactile perception on the human forearm when applying multiple ¿funneling¿ stimuli. In our psychophysical experiments, we revealed the human spatial perception ability on the human forearm for stationary and moving vibrotactile stimuli.

Full text

Page 1

Spatial Resolution of Vibrotactile Perception on the
Human Forearm when exploiting Funneling Illusion

Ahmad Barghout, Jongeun Cha, Abdulmotaleb El
Saddik
Multimedia Communications Research Lab
University of Ottawa
Ottawa, Ontario, Canada
{abarghout, abed}@mcrlab.uottawa.ca,
jcha@discover.uottawa.ca
Julius Kammerl, Eckehard Steinbach

Institute for Media Technology (LMT)
Technische Universität München
D-80290 Munich, Germany
{kammerl,eckehard.steinbach}@tum.de


Abstract—
Recent advances in man-machine interaction, telerobotics,
telepresence and teleaction have shown that introducing the
haptic modality to multimedia applications has the power to
significantly widen their application range and to dramatically
improve the user experience. Especially, the human body surface
has been considered as an additional means of presenting
information using vibrotactile display devices. In this context,
spatial displacement of a vibrotactile stimulus can be deployed
for information display. By exploiting a psychophysical illusion
called “funneling illusion”, we are able to increase the spatial
resolution of vibrotactile displays. In this paper, we aim at
investigating the spatial resolution of vibrotactile perception on
the human forearm when applying multiple “funneling” stimuli.
In our psychophysical experiments, we revealed the human
spatial perception ability on the human forearm for stationary
and moving vibrotactile stimuli.

haptics; vibrotactile display; perceptual spatial resolution;
psychophysics; teleoperation; funneling illusion
I.
INTRODUCTION
We, humans, rely heavily on the haptic modality to interact
with our environment. However, the haptic modality is rarely
used in modern multimedia systems. Novel multimodal
human-computer interfaces exploit the human body surface as
an additional means of presenting information using
vibrotactile devices [1]. This allows for presenting extra
information such as a directional cue in a car [2], interaction
for touch screen mobile devices [3], tactile music [4], a
grabbing force in teleoperation [5], touch sensation in a remote
interpersonal communication [6], etc. A variety of tactile and
other haptic interfaces and applications are also introduced in
[21, 22].
Additional interest can be found in telepresence and
teleaction systems, which allow a human user to immerse into
a remote or inaccessible environment. To enable a realistic
immersion into the distant environment, a multimodal interface
device displays visual, auditory and haptic information to the
human operator which is sensed within and received from the
remote environment. To avoid overloading the visual modality,
the vibrotactile modality can be used for displaying critical
information, such as remotely measured distance-to-object
information. By controlling the displacement of vibrotactile
stimulus on the skin of the human operator, additional
information can be displayed to the human. In that context, the
human localization ability of vibrotactile feedback is
fundamental. In our previous work [7], we demonstrated the
use of a psychophysical phenomenon, called the funneling
illusion, for overcoming limitations in spatial resolution of
vibrotactile arrays. In this paper, we conduct psychophysical
experiments to investigate the performance of localizing
“funneled” vibrotactile stimuli.
The remainder of this paper is structured as follows. In
Section 2, we present our proposed methods for creating the
stationary and moving stimuli on the forearm. In addition, we
present the deployed vibrotactile actuators. Section 3 discusses
the experimental apparatus, design and procedure of our
psychophysical experiments as well as results. Section 4
concludes the paper.
II.
DISPLAYING VIBROTACTILE STIMULI
A. Funneling Illusion
The funneling illusion describes a phantom sensation
midway between multiple stimuli when they are presented
simultaneously at adjacent locations on the human skin [8, 9].
To create a stimulus between two vibrotactile actuators, they
are activated simultaneously. If the two actuators have the
same intensity, the illusionary sensation is created in the
middle between them. However, if they have different
intensities, the sensation is “funneled” and shifted towards the
actuator with higher intensity. This illusion allows us to create
an apparently continuously moving vibrotactile stimulus [7,
10]. An illustration of this approach is shown in Figure 1. By
varying the intensities of two adjacent actuators, we can
smoothly shift the in-between sensation from one actuator
location to the other.
In [7] it has been shown that a good distance between two
actuators for displaying the apparent movement sensation on
the human forearm is around 40-80mm. As the average adult's
forearm length is 252mm [11], four vibration motors spaced at

Page 2

intervals of 80mm are enough for utilizing most of the skin of
the forearm.
Actuator A
Actuator B
t
t1
Vibration
Intensity
Actuator A
Actuator B
Perceived Stimulus Location
Apparent Movement of Stimulus
t2

Figure 1: Illustration of exploiting the funneling illusion to create a
continuously moving vibrotactile stimulus.

B. Tactile Device
An easy-to-wear tactile device is designed to provide the
stimuli on the forearm which provides a continuous, relatively
flat surface to study, permitting the required separation of the
vibrators [12]. Our tactile feedback device consists of four
pancake-type vibrating DC motors that are usually used in cell
phones. These motors vibrate tangentially to the skin following
the recommendation of [9], which says the vertical vibration
can propagate on the skin and give deteriorated sensation. They
are lightweight, inexpensive and easy to deploy and consume
little power. Their operating voltage is 3.6 volt and their
operating frequency range is up to 220Hz. They can be
attached to and detached from the arm band using Velcro. To
fix the vibrators, an arm band is wrapped around the forearm so
that it softly presses the actuators to the skin. Figure 2 shows
the placement of four actuators on the forearm.

Figure 2: Vibrotactile Device

In order to control the intensity of the actuators, a
microcontroller, ATMega 128 is used to generate a pulse-width
modulation (PWM) signal which provides 16 levels of applied
intensity. According to previous studies [7] of our actuators, we
limit the range of control levels from zero to 12 to assure a
linear tactile perception when driving the actuators.
III.
EXPERIMENTAL EVALUATION
Psychophysical experiments are conducted to evaluate the
performance of spatial vibrotactile perception when exploiting
the funneling illusion. Additionally, to investigate temporal
dependencies when displaying a vibrotactile stimulus, we
evaluate the localization ability for stationary and for moving
stimuli separately.
A. Participants
Twelve participants (average age of 26.5, age range from
23 to 35 years; 11 males and 1 female), who are all students at
University of Ottawa, took part in this experiment. All of the
participants self-reported a normal sense of touch. Eleven of
the participants were right-handed, one was left-handed. The
experiment took approximately 45 minutes on average.
B. Apparatus and Experimental Design
An overview of the experimental setup is shown in Figure
3. The experimental apparatus incorporated a PC to provide the
subjects with a Graphical User Interface (GUI), a vibrotactile
display array equipped with 4 actuators located at the forearm
and a ruler (made of paper) placed on the arm for aiding the
subject to locate the stimulus.
Serial Port
tactile feedback
Microcontroller
Ruler & Actuators attached to the arm
Experiment
GUI

Figure 3: Overview of the experimental setup

A screenshot of the GUI is shown in Figure 4. It allows for
selecting the vibrotactile display mode (stationary or moving
stimulus), a start button for starting the experiment and a group
of radio buttons for entering the perceived sensation position.

Figure 4: Screen shot of the GUI of the experiment
Subject’s Information
Start the Experiment
Select Static/Moving
Stimulus Mode
Experiment GUI
Stimulus Location (estimated and
selected by the subject)
Start Trial/Save Result

Page 3

The four actuators of the vibrotactile display array are
separated with a distance of 80mm. Hence, the actuator array
covers a total distance of 240mm from the wrist towards the
elbow. Along the actuator array, we define a set of 13
locations to be discriminated by the subjects. The four
actuators are attached at positions 0, 4, 8 and 12 as shown in
Figure 5. The attached ruler is used by the subject for pointing
at the perceived stimulus location.

Figure 5: Ruler used to aid the subjects in identifying the stimulus location.
Locations range from 0 at the elbow to 12 at the wrist.

In order to control the location of the “funneled” midway
sensation between the actuators, they have to be driven with
different intensities. As our actuators are limited to 12 discrete
intensities [7], theoretically a maximum of 13 individual
intensity combinations between two adjacent actuators are
possible, {(12,0), (11,1), (10,2), (9,3), (8,4), (7,5), (6,6), (5,7),
(4,8), (3,9), (2,10), (1,11), (12,0)}, where the two numbers of a
couple represent intensities for each actuator. However, in our
psychophysical experiments, we preselected a reduced test set
of 13 equidistantly arranged funneled stimulus locations along
the vibrotactile actuator array by using {(12,0,0,0), (9,3,0,0),
(6,6,0,0), (3,9,0,0), (0,12,0,0), (0,9,3,0), (0,6,6,0), (0,3,9,0),
(0,0,12,0), (0,0,9,3), (0,0,6,6), (0,0,3,9), (0,0,0,12)} for actuator
A, actuator B, actuator C and actuator D, respectively.
Our psychophysical evaluation consists of two runs: one for
evaluating the spatial location perception of a stationary
stimulus and one for a moving stimulus. During each run, each
of the predefined 13 test locations is evaluated 7 times. Thus,
each run contains 91 trials per subject. In order to eliminate
trends of task learning, the order of the two runs was
counterbalanced and the investigated locations are randomized
for each subject. The whole experiment took on average 35
minutes with a 10 minutes break between the two display
modes.
Stationary stimuli were presented for a period of 1 second
(well above the 0.25 s threshold set in [9]). When displaying a
moving vibrotactile stimulus, we have to define a spatial range
within which the temporal displacement can take place. Along
the way of moving stimuli, stimulation at each location
persisted for 200ms. Hence, to travel over the whole arm
(240mm), it takes 2.6 seconds. To assure the user is not able to
perceive the location of the stimulus from its traveling time, the
moving stimulus was set to take the same time for each
location by following the trajectory shown in Figure 6, starting
and ending at the same location. Subjects were asked to
identify the ending point that the moving stimulus reaches.
C. Procedure
The subjects are comfortably seated at a desk facing a
visual display for instructions. During the experiment, the
vibrotactile stimuli are displayed on the subject’s left forearm.
Initially, all participants are trained to be familiar with the
experimental apparatus. Once they felt comfortable, the
experiment is started.
After entering the subject's personal information and
selecting the stimulus mode by the experimenter by
counterbalanced order, the subject is instructed to press the
“Start” button by using a mouse with his/her right hand and the
experiment begins.
After each trial, the “Estimated Location” radio button
group is enabled to allow the subject to select the location at
which he/she perceived the stimulus by referring to the ruler
attached to the forearm.
1
3
2
Location 0
Starting/Ending point
E.g. Location 9
Location 12
1
3
2
Location 0
Starting/Ending point
E.g. Location 5
Location 12

Figure 6: Two examples of the path followed by the moving stimulus for a
certain location (9 and 5 in this example).

D. Results and Discussion
In Figure 7 and Figure 8, the averaged perceived stimulus
locations with standard deviations are shown for the stationary
and the moving vibrotactile stimuli. Our results show good
localization performance for both display modes.

Figure 7: The average value of the perceived locations of the stationary
stimulus for each location.

Page 4

Figure 8: The average value of the perceived locations of the moving stimulus
for each location.

Figure 9 and Figure 10 illustrate the percentage of correctly
identified test items. Surprisingly, the perceived location for
the stationary display mode seems to be shifted towards the
elbow. Independently from the display mode, the results show
best performance at the beginning and the end of the actuator
array (closest to the elbow and wrist). Here, the localization of
the stationary stimulus at location 0 (elbow) reached an
accuracy of 75% whereas at location 12 (wrist) it reached
60%. For the moving stimulus, we obtained 65% and 55%,
respectively. However, this observation does not necessarily
show improved performance. As towards the ends of the
vibrotactile actuator array the number of possible candidates
decreases, we obtain a reduced set of alternatives, which
facilitates the trial.
Nevertheless, a careful look at the literature [12] shows
that this result at the array limits is not related to the number
of alternatives as explained above. A similar result was
obtained with a 7-tactors array that showed better performance
at the human joints. When the authors in [12] moved the array
of tactors to have the middle tactor (number 4) at the elbow
(having the array to start from the upper arm), they still got the
best localization results at the elbow. Figure 11 illustrates their
results. The solid line shows localization results when the 7
tactors are at the forearm and the 7th tactor is at the elbow.
The dashed line shows localization results when the 7-tactors
array is shifted upwards so that tactor 4 is at the elbow.
Comparing the two lines emphasizes the higher localization
ability at the elbow. On an anatomical and physiological level,
this foundation is supported by many resources. Acuity, as
defined in [16], “refers to the ability to locate the site of the
initiation of a stimulus. High acuity allows for fine distinction
and requires a greater density of neurons”. In addition,
receptors in muscles and joints may contribute to the tactile
sensations besides the receptors in the skin [17]. Although
different receptors seem to respond best to particular types of
stimuli, they also respond to some degree to all types of
tactual stimuli [18]. This is the case here as the body joints,
tendons, and muscles hold a large amount of tactile receptors
[18, 19]. More specifically, Pacinian corpuscles, tactile
receptors mostly sensitive to vibrations [16, 20], normally
found under the skin, are scattered within the body,
particularly around muscles and joints [20]. This physiological
distribution of tactile receptors is behind the superior
localization performance at the edges of the actuators array.
When analyzing the performance in localization of
midway sensations, our results show a significant difference
between the stationary and moving stimuli. Here, the
performance of correctly identified test items ranges from 11-
48% for the stationary stimulus mode and 18-42% for the
moving stimulus mode. Please note, that the results for all test
items are clearly above the 100/13 = 7.7% chance
performance level for 13 test items.

Figure 9: The percentage of correctly perceived locations of the static stimulus
for each location.


Figure 10: The percentage of correctly perceived locations of the moving
stimulus for each starting/ending location.


Figure 11: The percentage of correctly perceived locations of the static
stimulus for each location in Cholewiak’s work [7].

Page 5

Furthermore, our results reveal that the performance is
better in the vicinity of the actuators than in-between them
where the virtual locations are generated. By looking at Figure
9 and Figure 10, we can see that the minimum accuracy is
obtained in the middle of two adjacent actuators (locations 2, 6
and 9). However, the results for the moving stimulus mode
show slightly better performance in presenting a “funneled”
midway stimulation between adjacent actuators.
After all, this result is consistent with the findings of
Weinstein [15] which states that the two-point discrimination
threshold ranges from 38mm to 40mm. It was difficult for the
subjects to discriminate adjacent locations separated by a
distance of 20mm which is less than the mentioned threshold.
Comparing the performance of the two display modes
reveals, that the moving stimulus is characterized by improved
spatial localization of “funneled” midway sensations (2, 3, 5,
6, 7, 9, 10) at the expense of impaired localizing ability at the
actuator locations (0, 4, 8, 12) when stimulated. The average
accuracy for localizing stationary stimulus is 31.5% within
which midway stimuli have a detection rate of 19.84%.
Respectively, the average accuracy for the moving stimulus is
32.69% within which the accuracy for virtual locations is
25.13%.
IV. CONCLUSION
With the growing trend of deploying haptically enabled
multimodal human-computer interfaces, the human body
surface has been considered as an additional means for
information display. By mapping a numeric quantity to a
moving vibrotactile stimulus on the forearm we are able to
establish an additional information channel, which can be used
to avoid overloading existing modalities. In this context, the
spatial perception of a vibrotactile stimulus on the human skin
is critical for the performance of such a display method. In this
work, we conduct a psychophysical experiment in order to
investigate the spatial resolution of vibrotactile stimuli on the
human forearm. 13 test items of stationary and moving
vibrotactile stimuli are evaluated. Both display methods show
best localization accuracy in the vicinity of the joints (elbow
and wrist), followed by the locations of the actuators
themselves. When displaying a moving vibrotactile stimulus,
improved performance of “funneled” midway sensations is
achieved.
Our future work addresses an extension of the spanning
area of the device to cover the whole arm. This is to allow
more separation between adjacent locations and possibly
higher correct localization rates.
REFERENCES

[1] A. Gallace, H. Z. Tan, and C. Spence, “The Body Surface as a
Communication System: The State of the Art after 50 Years,” Presence:
Teleoperators & Virtual Environments, Vol. 16, No. 6, pp. 655-676,
2007
[2] H. Z. Tan, R. Gray, J. J. Young, and R. Traylor, “A Haptic Back Display
for Attentional and Directional Cueing,” Haptics-e, Vol. 3, No. 1 2003
[3] S. Brewster, F. Chohan, and L. Brown, “Tactile Feedback for Mobile
Interactions,” Proc. CHI 2007, pp. 159-162, 2007
[4] E. Gunther, G. Davenport, and S. O’Modhrain, “Cutaneous Grooves:
Composing for the Sense of Touch,” Proc. Int. Conf. New Interfaces for
Musical Expression, pp. 1-6, 2002
[5] A. M. Murray, R. L. Klatzky, and P. K. Khosla, “Psychophysical
Characterization and Testbed Validation of a Wearable Vibrotactile
Glove for Telemanipulation,” Presence: Teleoperators & Virtual
Environments, Vol. 12, No. 2, pp. 156-182, 2003
[6] J. Cha, M. Eid, L. Rahal, and A. El Saddik, “HugMe: An Interpersonal
Haptic Communication System,” Proc. IEEE Int. Workshop on Haptic
Audio Visual Environments and their Applications (HAVE2008), pp.
99-102, 2008
[7] J. Cha, L. Rahal and A. El Saddik, “A Pilot Study on Simulating
Continuous Sensation with Two Vibrating Motors,” Proc. IEEE Int.
Workshop on Haptic Audio Visual Environments and Games, pp. 143-
147, Oct 2008
[8] G. V. Bekesy, “Funneling in the Nervous System and its Role in
Loudness and Sensation Intensity on the Skin,” The Journal of the
Acoustical Society of America, Vol. 30, No. 5, pp. 399-412, 1958.
[9] D. S. Alles, “Information Transmission by Phantom Sensations,” IEEE
Trans. Man-Machine Systems, Vol. MMS-11, No. 1, pp. 8-91, 1970.
[10] L. Rahal, J. Kammerl, J. Cha, E. Steinbach, and A. El Saddik,
“Investigating the Influence of Temporal Intensity Changes on the
Apparent Movement Phenomenon,” Proc. Int. Conf. Virtual
Environments, Human-Computer Interfaces and Measurement Systems,
pp. 310-313, 2009.
[11] N. Alves, P. L. Cândido, and R. Frazao, “Innervation of the Pronator
Quadratus Muscle,” Int. Journal of Morphology, Vol. 22, No. 4, pp. 253-
256, 2004.
[12] R. W. Cholewiak and A. A. Collins, “Vibrotactile Localization on the
Arm: Effects of Place, Space, and Age,” Perception & Psychophysics,
Vol. 65, No. 7, pp. 1058-1077, 2003.
[13] E. H. Weber, “The Sense of Touch,” New York: Academic Press
(Original work published 1826), 1987.
[14] H. Hamburger, “Locognosia: the Ability to Localize Tactile Stimuli on
the Body Surface,” Ph.D. Thesis, University of Amsterdam, Nederlands,
1980.
[15] S. Weinstein, “Intensive and Extensive Aspects of Tactile Sensitivity as
a Function of Body Part, Sex, and Laterality,” The Skin Senses, Charles
C. Thomas, Springfield, Illinois, 1968.
[16] C. A. Braun and C. Anderson, “Pathophysiology: Functional Alterations
in Human Health,” Lippincott Williams & Wilkins, pp 256, 2007
[17] L. Stankov, T. Seizova-Cajic, Roberts, and D. Richard, “Tactile and
Kinesthetic Perceptual Processes within the Taxonomy of Human
Cognitive Abilities,” Intelligence, Vol. 29, No. 1, pp. 1-29, 2001
[18] K. Moss, “Some Things to Learn from Learning Through Touch,” Texas
Deafblind Outreach, 2005
[19] R. Books, “The Encyclopedic Atlas of the Human Body” Gordon
Cheers, pp. 140-141, 2004
[20] "Pacinian corpuscle." Encyclopædia Britannica. 2009. Encyclopædia
Britannica Online. 12 Sep. 2009
<http://www.britannica.com/EBchecked/topic/437813/Pacinian-
corpuscle>
[21] A. El Saddik “The Potential of Haptics Technologies” IEEE
Instrumentation & Measurement, Vol. 10, No. 1, pp. 10-17, 2007
[22] M. Eid, M. Orozco and A. El Saddik, “A Guided Tour in Haptic Audio
Visual Environment and Applications”, Int. J. of Advanced Media and
Communication, Vol. 1, No. 3, pp: 265 – 297, 2007