Investigating the Semantic Perceptual Space of Synthetic Textures on an

Ultrasonic based Haptic Tablet

Maxime Dariosecq

, Patricia Pl

enacoste

, Florent Berthaut

, Anis Kaci

and Fr

eric Giraud

CRIStAL, CNRS, University of Lille, France

L2EP, University of Lille, France

Keywords:

Haptic Feedback, Tactile Device, Ultrasonic Vibrations, Perceptual Space, Material Perception, Texture

Rendering.

Abstract:

This paper aims to investigate the semantic perceptual space of synthetic tactile textures rendered via an

ultrasonic based haptic tablet and the parameters inﬂuencing this space. Through a closed card sorting task,

30 participants had to explore 32 tactile-only textures and describe each texture using adjectives. A factorial

analysis of mixed data was conducted. Results suggest a 2 dimensional space with tactile textures belonging

to a continuum of rough to smooth adjectives. Inﬂuence of waveform and amplitude is shown to play an

important role in perceiving a texture as smooth or rough, and spatial period is a possible modulator of different

degrees of roughness or smoothness. Finally, we discuss how these ﬁndings can be used by designers on tactile

feedback devices.

1 INTRODUCTION

The vast majority of our daily touch interactions on

devices such as phones or tablets, consisting mainly

of visual and auditory information (Levesque et al.,

2011), are still lacking colourful tactile experience,

which has been shown to add to their emotional im-

pact (Wilson and Brewster, 2017). While real materi-

als vary in shape but also size or textures, the simula-

tion of these materials is often limited, due to the tech-

nological constraints of the devices. As such, various

devices can be found, ranging from deformation crust

devices, pin arrays, electrovibratory devices, midair

vibrations or friction modulation surfaces (Culbert-

son et al., 2018). Focusing speciﬁcally on tactile tex-

tures, ultrasonic based tablets have shown promising

enhancements to the user’s exploratory movements,

allowing applications such as haptic text messages,

image sharing and virtual touch (Mullenbach et al.,

2014) or haptic e-books (Cingel and Piper, 2017)

which can enhance children’s reading activity on a

tablet with tactile textures. These devices rely on a

phenomenon called squeeze ﬁlm effect (Biet et al.,

2007) which leverages ultrasonic vibrations to create

an air-gap between the user’s ﬁnger and the tablet, re-

sulting in friction reduction.

1.1 Related Work

Thus far, most of the studies on tactile perception con-

ducted on these devices were focused on perception

thresholds of friction, size or sharpness(Biet et al.,

2007; Samur et al., 2009; Kalantari et al., 2016;

Kalantari et al., 2018b; Kalantari et al., 2018a). In the

context of interaction design however, many oppor-

tunities will arise from the ability of rendering tac-

tile textures that provide users with speciﬁc percep-

tual sensations (i.e. rough/smooth, soft/hard). Lit-

tle research has addressed the perception of simu-

lated textures (Brewer et al., 2016; Messaoud et al.,

2016). In these studies, real and simulated surfaces

were explored and compared by the participants. Re-

sults show acceptable classiﬁcations between real and

re-created textures, therefore illustrating the inﬂuence

of parameters such as temporal frequency and spa-

tial period, amplitude, friction, with speciﬁc signal

shapes. Fabrics or material with unique extracted fea-

tures gave participants enough perceptual cues to suc-

cessfully distinguish between compliant and harder

surfaces, between low friction surfaces and high fric-

tion surfaces, but also between different recreated fab-

ric such as velvet and cotton.

In Messaoud (Messaoud et al., 2016) virtual tac-

tile textures were created using a multi level approach

with both signal shape (the average of user’s friction

Dariosecq, M., Plénacoste, P., Berthaut, F., Kaci, A. and Giraud, F.

Investigating the Semantic Perceptual Space of Synthetic Textures on an Ultrasonic based Haptic Tablet.

DOI: 10.5220/0008979800450052

In Proceedings of the 15th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2020) - Volume 2: HUCAPP, pages

45-52

ISBN: 978-989-758-402-2; ISSN: 2184-4321

contrast) and real textile speciﬁc spatial periods (num-

bers of grooves of the textile’s surface) gathered by

tribological extraction. Instead of relying on existing

materials, we focus here on synthetic textures, with

the goal of allowing interface designers to generate

semantically deﬁned (e.g. smooth, rough ...) textures

from a set of parameters. A similar approach was pro-

posed by Friesen et al (Friesen et al., 2018), who study

the users’ ability to recognise the spatial frequency of

synthetic tactile textures. We believe that observing

how individuals perceive and give perceptual descrip-

tors to randomly simulated textures will offer an in-

sight into which parameters tend to inﬂuence percep-

tion. In this paper, we aim to strengthen this method-

ology by giving a more general view about the inﬂu-

ence of these parameters on the user’s perception, al-

lowing designers to approach speciﬁc tactile percep-

tions without the need of tribological extractions of

real materials.

Human perceptual and affective experiences in

touching materials are expressed by a semantically

multilayered and multidimensional information space

(Okamoto et al., 2016). The bottom layer of this

perception system involves the perception of phys-

ical properties, allowed by tactile receptors dis-

tributed by clusters around the human skin. Activated

mechanoreceptors or thermoreceptors will then trans-

mit information through nerve signals to the brain

(Dargahi and Najarian, 2004). This psychophysical

layer, or layer of textures has been widely investi-

gated, with more than 20 papers offering a relatively

homogeneous description of the dimensions required

to identify textures. While results may vary, previ-

ous attempts to describe this texture perception space

used multidimensional scaling (MDS) to converge to

an optimal 3 to 5 dimensions space (Holliins et al.,

1993; Hollins et al., 2000; Picard et al., 2003; Tiest

and Kappers, 2006; Guest et al., 2011; Okamoto et al.,

2016). These main textural dimensions supposedly

include continuum such as ”Rough/Smoothness”,

”Hard/softness”, ”Sticky/Slipperiness”, ”Hot/cold”

and ”Dry/Wetness”. Going further, Guest et al.

(Guest et al., 2011) broadened the lexicon used to

describe the experience of touch using similar MDS

based experimental design to represent the semantic

perceptual spaces underlying the words answered in

touch perception tasks. If the importance of a clear

lexicon to describe the perceptual experiences of real

materials is stressed by Guest et al., a lexicon describ-

ing artiﬁcially rendered textures on haptic devices

would greatly help further developing these technolo-

gies.

The semantic perception space for real materials

has been thoroughly studied, but it remains mostly

unknown for synthetic tactile textures on an ultrasonic

tablet. Moreover, there are no clear guidelines on how

to tune the parameters used for texture rendering to

elicit a speciﬁc perception. The goal of this study is

to broadly investigate the user’s semantic perceptual

space of ultrasonic rendered textures. We also aim to

reveal the relationship between speciﬁc semantic pat-

terns and the synthesis parameters used to create the

textures. More precisely, we aim to observe if there is

a speciﬁc relation between the waveform, amplitude,

and frequency and the descriptors used to qualify the

perception of speciﬁc textures.

2 EXPERIMENT

A closed card sorting task was performed to inves-

tigate the participant’s semantic perceptual space af-

ter they explored a sample of synthetic tactile texture.

Participants had to explore simulated textures on an

ultrasonic tactile device and qualify each texture us-

ing descriptors of either consistency and/or texture.

It is worth noting that there was no visual feedback

for the participants, the screen remained blank and

the textures were only tactile. Textures were explored

only once, without time restriction and with the pre-

ferred hand. During this task noise-cancelling head-

phones were used to prevent interference with noises

that could be emitted by the device. Factor analysis of

mixed data was used to investigate both the semantic

perceptual space derived from the chosen adjectives

and its relation with the texture’s parameters.

2.1 Textures and Descriptors

Thirty two synthetic tactile textures were generated

on the tablet’s surface using different waveforms, spa-

tial periods, amplitudes and ratios with a constant

offset. Two waveforms were used, namely sine and

square waves which is depicted in Figure 1. Spa-

tial periods ranged from 1000 to 10000 µm including

1000, 2500, 5000 and 10000 µm. Ratios, which cor-

respond to the proportion of high voltage in a period

for the square waveform, consisted of 10, 50, and 90

percents. Finally, two amplitudes ranges of 10% and

40% were used with an offset of 50%, 100% being the

maximal voltage actuated by the piezoelectric cells.

Participants were asked to describe textures by se-

lecting adjectives from a set of 20 descriptors. Instead

of relying on descriptors designed for visual textures

(Rao and Lohse, 1996), we chose to build a tactile

speciﬁc set of descriptors. They were therefore cho-

sen by narrowing down answers from a pre-testing

open card sorting task with 22 participants in which

HUCAPP 2020 - 4th International Conference on Human Computer Interaction Theory and Applications

Figure 1: Example textures from a selection of waveforms (sine and rect/square waves), spatial periods, amplitude and ratios

(only for square waves) used in the study. 50 corresponds to the offset (50% of voltage).

they were asked to propose as many adjectives as they

wanted on the same set of synthetic textures. From

these answers, the ten more frequent adjectives were

kept. This pre-testing task also revealed a number

of terms describing consistency rather than texture,

which we chose to keep as a separate category and

complete with additional adjectives.

Table 1: Summary of the descriptors.

Texture Consistency

Delicate Hard

Spiky Soft

Sticky Mellow

Rough Pasty

Smooth Slimy

Granular Elastic

Wet Spongy

Oily Gelatinous

Sandy Viscous

Slippery Friable

2.2 Apparatus

Friction reduction comes with a number of limita-

tions, such as the absence of feedback when the ﬁnger

is not moving. However it allows for ﬁner control of

synthetic texture parameters compared to other tech-

nologies such as vibrotactile feedback, which would

not permit the range of spatial periods that we inves-

tigate in this study.

A tactile feedback display based on ultrasonic vi-

brations for haptic rendering was used, which is de-

picted in Figure 2. The enhanced visual tactile ac-

tuator (E-ViTa) is a Banana Pi based single-board

computer (Shenzhen LeMaker Technology Co. Ltd,

China),running a 1 GHz ARM Cortex-A7, a dual-core

CPU and 1 GB RAM. A STM32f4 microcontroller

(STMicroelectronics, France) is allowed to work in

parallel with the computer via Serial Peripheral In-

terface (SPI) bus at 10 kHz. To be able to detect

positions of the ﬁngers, this setup is connected to a

12.5cm capacitive touchscreen (Banana-LCD 5”-TS,

MAREL, China), with a sampling frequency of 62

Hz.

Ten 14x6x0.5 mm piezoelectric cells actuate a

154x81x1.6 mm ﬁxed glass plate, resonating at 60750

Investigating the Semantic Perceptual Space of Synthetic Textures on an Ultrasonic based Haptic Tablet

Hz with a half wavelength of 8 mm. A power

electronic circuit converts a 12V DC voltage source

into an AC voltage, controlled in amplitude and fre-

quency and supplied to the piezoelectric cells. The

microcontroller synthesizes a pulse-width modulation

(PWM) signal to drive a voltage inverter that actuates

the piezoceramics. Details on the friction reduction

which can be achieved with this technology can be

found in (Vezzoli et al., 2017).

Figure 2: Structure of the E-Vita ultrasonic based haptic

display.

2.3 Participants

A sample of 30 participants (24 males, 6 females,

M=29, SD=8.8) were recruited for this experiment.

They were all naive regarding the haptic technology

used in the experiment to avoid any interference dur-

ing the response phase of the protocol. They where

all right-handed, without motor impairment, and par-

ticipated freely in the experiment.

2.4 Procedure

Participants were seated and briefed in front of the

setup (see Figure 3). As proposed by Rao et al. (Rao

and Lohse, 1996) for visual textures, the 32 synthetic

tactile textures were presented one at a time and only

once, in random order. The textures were only tactile

and presented on a blank screen, i.e. the textures did

not have a visual representation presented to the par-

ticipants. Participant were free to explore each tex-

ture for as long as they needed, using a continuous

motion with their index ﬁnger. The hardware used for

this experiment does not allow for multitouch tactile

feedback, so we restrained the exploration to a sin-

gle ﬁnger. They then had to choose between the 20

descriptors listed in Table 1 on a laptop placed next

to the tablet. Participants were instructed to choose

descriptors they deemed accurate to represent the ex-

plored texture. There was no limitation regarding the

number of descriptors used. Experiment duration av-

eraged 35 minutes.

Figure 3: Experimental setup.

2.5 Statistical Analysis

A 2d square matrix was computed using the partici-

pants’ answers to each texture. Each cell expresses

the probability of each adjective being chosen for a

given texture.

As pointed out by Okamoto et al. (Okamoto et al.,

2016), factor analysis is a commonly used strategy

when assessing the psychophysical dimensions of tac-

tile perception of textures. However introducing both

qualitative (the texture’s parameters) and quantitative

(the probability of an adjective occurring) variables

requires the use of a factor analysis of mixed data

(FAMD). For each texture, we have K=4 quantita-

tive variables (the waveform, spatial period, ampli-

tude, ratio parameters) on a lower two dimensional

space, and Q=20 quantitative variables (probability

for each adjective) on a related subspace while maxi-

mizing inertia. Thus, after testing for correlation of

both K and Q dimensions, independent factors are

constructed representing a maximum portion of vari-

ance from the original data. Each factor can be con-

sidered as a perceptual dimension (Okamoto et al.,

2016).

2.6 Excluded Textures and Descriptors

After running this model a ﬁrst time, a selection

of textures and descriptors were removed from the

model, because they showed either a very low prob-

ability of occurrence, no correlations, or absence of

clear patterns. The following results will not account

for the following descriptors : hard, mellow, pasty

and spongy and the textures presented in table 2.

HUCAPP 2020 - 4th International Conference on Human Computer Interaction Theory and Applications

A B

C D

Figure 4: FAMD plots. A) screeplot, B) correlation circle for quantitative variables, C) subspace projection for the tex-

tures, D) subspace projection for the quantitative variables (waveform : sin(1)/square(2), amplitude : 10%/40%, period :

1000/2500/5000/10000 (µm)).

Table 2: Removed textures.

Texture Waveform Period Amplitude Ratio

1 rect 1000 10 90

16 rect 10000 10 50

19 rect 5000 10 50

23 rect 2500 10 50

27 sin 2500 40 na

28 sin 1000 40 na

3 RESULTS AND DISCUSSION

3.1 Semantic Perceptual Space

Results for the factor analysis of mixed data demon-

strate an optimal two dimensional space as shown

by the scree plot in Figure 4.A with a total of 44.6

% variance retained. Factor 1 (35.1 %) can be in-

terpreted as a rough/smoothness continuum with se-

mantically opposed clusters of descriptors including

rough descriptors all correlating negatively with F1

and smooth descriptors positively correlating (Figure

4.B). Table 3 shows the loading values for all fac-

tors, all loadings appearing as their maximum val-

ues. Loadings values can be interpreted as the cor-

relation coefﬁcient between the loading and the fac-

tor. All descriptors within F1 show correlation coefﬁ-

cients superior to .71, indicating a strong relationship

to the factor. With approximately 3 times the vari-

ance retained, the rough/smoothness factor accounts

for most of the variations in this model, although F2’s

(9.4 %) loadings tends to explains both smooth and

rough inner clusters variations, with either rougher

or smoother surfaces. These results indicate that a

methodology designed to assess the psychophysical

dimensions (Okamoto et al., 2016) and the seman-

tic (Guest et al., 2011) of tactile perception on real

textures can be transferred to rendered ultrasonic tex-

tures. The texture sample used for this study therefore

seem to elicit perception of various roughness levels.

Perception elements of viscoelasticy (F3, 8.74%) and

wet/stickyness (F4, 8.13%) can also be extracted.

Investigating the Semantic Perceptual Space of Synthetic Textures on an Ultrasonic based Haptic Tablet

Table 3: FAMD factors extracted from the probability of adjective occurrence given a texture.

Factor F1: rough /

smoothness

F2: inner domains variations F3 : viscoelasticity F4 : wet/stickyness

%variance 35% 9.4% 8.74% 8.13%

Relationship

squared

Waveform .60 period .63

(qualitative) amplitude .42

Loadings granular -.82 viscous .52 sticky -.60

(quantitative) spiky -.72 elastic .45 wet .50

rough -.72

sandy -.71

friable -.71

smooth .88

delicate .86

slippery .70

3.2 Relation between Textures

Parameters and Semantic

Perceptual Space

In order to elicit speciﬁc perceptions, the underlying

inﬂuence of the parameters used to render these tex-

tures must be addressed. By maximizing the corre-

lations between qualitative and quantitative elements

and the factors, factor analysis of mixed data offers a

tool to assess these relations. Figure 4.C represents

the texture’s sub-space within the factors. We chose

not to provide the details for each of the 32 individ-

ual textures, to focus on the overall texture parame-

ters. Figure 4.D shows each modality of the texture’s

parameters, appearing as the centroid of the textures

possessing these properties in the 2D factor space.

Results show relationship squared coefﬁcients of .60

for waveform and .42 for amplitude with F1, and .63

between period and F2. Relationship squared metric

helps appreciate the strength of the relation between a

qualitative variable and the factor.

Figure 4.D and Table 3 show a strong inﬂuence of

waveform and amplitude on F1, meaning that these

variables inﬂuence heavily the participant’s percep-

tion, leading them to categorize a texture on a rough

or smooth continuum. Indeed, textures being identi-

ﬁed as rough-like (Figure 4.C), tend to possess spe-

ciﬁc properties such as being square waveforms with

higher amplitude, while smoother textures tend to be

sinusoidal waveforms with lower amplitudes. Varia-

tions within either continuum could here be explained

by the period, indeed variations located on the verti-

cal axis (Figure 4.B-dim2, Figure 4.D-dim2) of dif-

ferent roughness or softness tend to vary with differ-

ent magnitudes of periods. Also, Figure 4.C shows a

small cluster of square waveforms being identiﬁed as

smooth-like textures, due to lower amplitudes and a

probable effect of ratio, participants might not be able

to clearly perceive elements of macro roughness with

such reliefs.

3.3 Limitations

Regarding limitations, this experiment does not pro-

vide a speciﬁc list of textures eliciting very precise

and consistent perceptions. It rather provides general

guidelines allowing one to play with the dimension of

rough/smooth textures. As pointed out by Okamoto

et al. (Okamoto et al., 2013), adjective labels used in

these experiments are suffering from limitations, such

as limited vocabulary related to physical properties of

material, semantic variability or the lack of consensus

regarding the meaning of a descriptor. More appropri-

ately, these descriptors are more often used for the de-

tection of perceptual dimensionality, it is no surprises

that the descriptors presenting the best correlations,

probability of occurrences and representation in this

model tend to belong to the description of perceptual

dimensions.

Table 4: Tactile perception adjectives and possible corre-

sponding synthetic texture parameters.

Adjective Waveform Period Amplitude Ratio

granulous rect 5000 40 50

rough rect 2500 40 50

sandy rect 1000 40 50

smooth sin 10000 10 na

delicate sin 1000 10 na

HUCAPP 2020 - 4th International Conference on Human Computer Interaction Theory and Applications

Figure 5: Waveform sample resulting from the parameters

which can be used to elicit perceptions represented by the

adjectives in Table 4.

3.4 Implications for the Design of

Tactile Feedback

From the results, we can extract a set of texture pa-

rameters that elicit perceptions ranging from smooth

to rough. This set is shown in Table 4. Figure 5 de-

picts the corresponding waveform patterns used for

texture rendering. As shown by our results, elicit-

ing smoother perceptual experiences would imply us-

ing sinusoidal waveforms with lower amplitudes and

rough-like textures would be rendered using rectan-

gular waveforms with higher amplitudes. Spatial pe-

riod could imply ﬁner to coarser rough-like textures,

but this relationship remains uncertain for smooth-

like textures as there were fewer sinusoidal textures.

While this study was conducted on a tablet with ultra-

sonic friction reduction, we believe the same effects

could be obtained with other tactile rendering tech-

nologies which allow for at least creating synthetic

textures with two waveforms (square and sine) and

controllable spatial period.

4 CONCLUSION

Using models designed to investigate the perceptual

space of real textures, this study shows that it is

possible to investigate simulated textures using the

same strategies. We showed that a continuum of

smooth/roughness does exist in a given sample of ul-

trasonic rendered textures, with different degrees of

roughness and smoothness. Inﬂuence of waveform

and amplitude were found to play an important role

in perceiving a texture as smooth or rough, and spatial

period as a possible modulator of different degrees of

roughness or smoothness. For designers, these results

are offering insight on how to tune speciﬁc param-

eters such as wave grating, amplitude and period to

elicit perceptions of smooth/roughness, but also pos-

sible various degrees of roughness, without the need

to go through tribological extractions. In future work

we will explore other waveforms such as sawtooth or

triangle, with a broader range of amplitudes and peri-

ods. Finally the association of visual and tactile com-

ponents in the case of synthetic textures still needs to

be addressed.

REFERENCES

Biet, M., Giraud, F., and Lemaire-Semail, B. (2007).

Squeeze ﬁlm effect for the design of an ultrasonic tac-

tile plate. IEEE transactions on ultrasonics, Ferro-

electrics and Frequency control, 54(12):2678–2688.

Brewer, D. J., Meyer, D. J., Peshkin, M. A., and Col-

gate, J. E. (2016). Viscous textures: Velocity de-

pendence in ﬁngertip-surface scanning interaction. In

2016 IEEE Haptics Symposium (HAPTICS), pages

265–270. IEEE.

Cingel, D. and Piper, A. M. (2017). How parents engage

children in tablet-based reading experiences: An ex-

ploration of haptic feedback. In Proceedings of the

2017 ACM Conference on Computer Supported Coop-

erative Work and Social Computing, pages 505–510.

ACM.

Culbertson, H., Schorr, S. B., and Okamura, A. M. (2018).

Haptics: The present and future of artiﬁcial touch sen-

sation. Annual Review of Control, Robotics, and Au-

tonomous Systems, 1:385–409.

Dargahi, J. and Najarian, S. (2004). Human tactile percep-

tion as a standard for artiﬁcial tactile sensing—a re-

view. The International Journal of Medical Robotics

and Computer Assisted Surgery, 1(1):23–35.

Friesen, R. F., Klatzky, R. L., Peshkin, M. A., and Col-

gate, J. E. (2018). Single pitch perception of multi-

frequency textures. In 2018 IEEE Haptics Symposium

(HAPTICS), pages 290–295. IEEE.

Guest, S., Dessirier, J. M., Mehrabyan, A., McGlone, F., Es-

sick, G., Gescheider, G., Fontana, A., Xiong, R., Ack-

erley, R., and Blot, K. (2011). The development and

Investigating the Semantic Perceptual Space of Synthetic Textures on an Ultrasonic based Haptic Tablet

validation of sensory and emotional scales of touch

perception. Attention, Perception, & Psychophysics,

73(2):531–550.

Holliins, M., Faldowski, R., Rao, S., and Young, F. (1993).

Perceptual dimensions of tactile surface texture: A

multidimensional scaling analysis. Perception & psy-

chophysics, 54(6):697–705.

Hollins, M., Bensma

ıa, S., Karlof, K., and Young, F. (2000).

Individual differences in perceptual space for tactile

textures: Evidence from multidimensional scaling.

Perception & Psychophysics, 62(8):1534–1544.

Kalantari, F., Grisoni, L., Giraud, F., and Rekik, Y. (2016).

Finding the minimum perceivable size of a tactile el-

ement on an ultrasonic based haptic tablet. In Pro-

ceedings of the 2016 ACM on Interactive Surfaces and

Spaces, pages 379–384. ACM.

Kalantari, F., Gueorguiev, D., Lank, E., Bremard, N., and

Grisoni, L. (2018a). Exploring ﬁngers’ limitation of

texture density perception on ultrasonic haptic dis-

plays. In International Conference on Human Haptic

Sensing and Touch Enabled Computer Applications,

pages 354–365. Springer.

Kalantari, F., Lank, E., Rekik, Y., Grisoni, L., Giraud, F.,

Rateau, H., Rekik, Y., Lank, E., Grisoni, L., Vezzoli,

E., et al. (2018b). Determining the haptic feedback

position for optimizing the targeting performance on

ultrasonic tactile displays. In IEEE Haptics Sympo-

sium (HAPTICS 2018), San Fransisco, United States.

Levesque, V., Oram, L., MacLean, K., Cockburn, A.,

Marchuk, N. D., Johnson, D., Colgate, J. E., and

Peshkin, M. A. (2011). Enhancing physicality in touch

interaction with programmable friction. In Proceed-

ings of the SIGCHI Conference on Human Factors in

Computing Systems, pages 2481–2490. ACM.

Messaoud, W. B., Bueno, M.-A., and Lemaire-Semail,

B. (2016). Textile fabrics texture: from multi-level

feature extraction to tactile simulation. In Inter-

national Conference on Human Haptic Sensing and

Touch Enabled Computer Applications, pages 294–

303. Springer.

Mullenbach, J., Shultz, C., Colgate, J. E., and Piper, A. M.

(2014). Exploring affective communication through

variable-friction surface haptics. In Proceedings of the

SIGCHI Conference on Human Factors in Computing

Systems, pages 3963–3972. ACM.

Okamoto, S., Nagano, H., and Ho, H.-N. (2016). Psy-

chophysical dimensions of material perception and

methods to specify textural space. In Pervasive Hap-

tics, pages 3–20. Springer.

Okamoto, S., Nagano, H., and Yamada, Y. (2013). Psy-

chophysical dimensions of tactile perception of tex-

tures. IEEE Transactions on Haptics, 6(1):81–93.

Picard, D., Dacremont, C., Valentin, D., and Giboreau, A.

(2003). Perceptual dimensions of tactile textures. Acta

psychologica, 114(2):165–184.

Rao, A. R. and Lohse, G. L. (1996). Towards a texture nam-

ing system: Identifying relevant dimensions of tex-

ture. Vision Research, 36(11):1649–1669.

Samur, E., Colgate, J. E., and Peshkin, M. A. (2009). Psy-

chophysical evaluation of a variable friction tactile in-

terface. In Human vision and electronic imaging XIV,

volume 7240, page 72400J. International Society for

Optics and Photonics.

Tiest, W. M. B. and Kappers, A. M. (2006). Analysis of hap-

tic perception of materials by multidimensional scal-

ing and physical measurements of roughness and com-

pressibility. Acta psychologica, 121(1):1–20.

Vezzoli, E., Vidrih, Z., Giamundo, V., Lemaire-Semail,

B., Giraud, F., Rodic, T., Peric, D., and Adams, M.

(2017). Friction reduction through ultrasonic vibra-

tion part 1: Modelling intermittent contact. IEEE

transactions on haptics, 10(2):196–207.

Wilson, G. and Brewster, S. A. (2017). Multi-moji: Com-

bining thermal, vibrotactile & visual stimuli to expand

the affective range of feedback. In Proceedings of the

2017 CHI Conference on Human Factors in Comput-

ing Systems, pages 1743–1755. ACM.

HUCAPP 2020 - 4th International Conference on Human Computer Interaction Theory and Applications