Thermal Perception and Thermal Devices Used on Body Parts Other Than Hand Or Face

Thermal perception and thermal devices used on body parts other than hand or face

Citation for published version (APA): Kappers, A., & Plaisier, M. (2019). Thermal perception and thermal devices used on body parts other than hand or face. IEEE Transactions on Haptics, 12(4), 386-399. [8747382]. https://doi.org/10.1109/TOH.2019.2925339

DOI: 10.1109/TOH.2019.2925339

Document status and date: Published: 01/10/2019

Document Version: Accepted manuscript including changes made at the peer-review stage

Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website. • The final author version and the galley proof are versions of the publication after peer review. • The final published version features the final layout of the paper including the volume, issue and page numbers. Link to publication

General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement: www.tue.nl/taverne

Take down policy If you believe that this document breaches copyright please contact us at: [email protected] providing details and we will investigate your claim.

Download date: 30. Sep. 2021 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TOH.2019.2925339, IEEE Transactions on Haptics

1 Thermal Perception and Thermal Devices used on Body Parts other than Hand or Face

Astrid M.L. Kappers and Myrthe A. Plaisier

Abstract—Most fundamental research on thermal perception focuses on the ﬁngers or the hand. Also most existing and proposed thermal devices are meant to be applied to hand or ﬁngers. However, if the hands are needed for other tasks, application of thermal stimulation to other body regions should be considered. This paper surveys the literature on thermal perception and thermal devices relevant to such other body regions. It starts with a short description of the experimental methods used in the various studies, such as the methods of limits, the two-alternative forced choice method, and magnitude estimation. This is followed by thermal psychophysical studies on detection, adaptation, spatial summation and resolution. Next some striking thermal illusions are presented, such as a thermal grill and a seemingly continuously warming or cooling stimulus. Finally, the few studies on thermal communication and applications are summarized. These latter studies mainly focus on communicating emotions or playing computer games. The overall conclusion of this survey is that thermal devices should not focus on conveying complex messages, but especially in the areas of gaming or communication there seem to be interesting possibilities for further developments.

Index Terms—Overview, thermal stimulation, psychophysics, devices, body !

1 INTRODUCTION

HE majority of studies on thermal perception and ther- sensitivity variation over the human body surface. Finally, T mal devices have made use of stimulation on the hand, there are many different experimental methods to assess (e.g., [1]). This needs not be surprising as hands are easily psychophysical performance of individuals, but it is very accessible and flexible in reaching a stimulus. Moreover important to realize that different methods will lead to dif- hands and fingers are more sensitive to thermal stimulation ferent outcomes; a threshold measured via a two-alternative than most other body regions, although lips, cheek and forced choice method will not necessarily be the same as a forehead are even more sensitive [2]. However, if a thermal threshold obtained via the method of limits (see below for device is meant to be used as, for example, a feedback, an explanation of these methods). alerting or communication system, it might be desirable The studies discussed in this overview are all published to apply the stimulation on other body parts in order to in peer-reviewed scientific journals or conference proceed- keep the hands free for other tasks. In the special case of ings, so this excludes devices that might be on the market devices designed for use in daily life by individuals with without being published. All references of the included impaired vision, it is also undesirable to apply stimula- papers are checked for further relevant studies. In addition, tion to the face as this would involve some clearly visible papers that cite the included studies have also been checked apparatus connected to the head. The goal of the present for relevance. In several of the studies described below also paper is to give an extensive overview of studies on thermal thermal stimulation on hands or face has been studied in perception on other body parts than the hands or the face. addition to other body regions, but these results will be The studies included in this overview range from rigorous omitted from this overview. psychophysics to rather informal pilot studies of thermal As it is important to have some basic understanding devices. about the psychophysical methodology used in the various When comparing the results of various studies, many studies, this overview will start with a brief introduction issues have to be taken into account. Age and gender of of relevant terms and methods. Next, studies on thermal different populations of participants may directly influence detection, adaptation, summation and resolution will be the results [2]. In the present case of thermal stimulation, presented. As thermal perception is also susceptible to illu- the physical parameters of the stimulus such as contact sions, a separate section is devoted to illusions. Finally, more size, rate of change of temperature, and baseline temper- applied research on thermal communication and other ap- ature are all important and may have huge influences on plications will be presented. In the conclusion, the important the results. Also the body region stimulated is of major factors to take into account when designing applications importance, as Stevens and Choo [2] showed a 100-fold with thermal feedback are discussed.

• A.M.L. Kappers and M.A. Plaisier are both with group Dynamics 2 METHODSANDTERMSUSEDINTHESTUDIES and Control of Eindhoven University of Technology, The Netherlands. A.M.L. Kappers is also with the groups Control Systems Technology and Perceptual performance of humans is often expressed in Human Technology Interaction of the same university. terms of thresholds. A threshold indicates the amount or E-mail: [email protected] intensity of a stimulus needed for the stimulus to be con- Manuscript received ??; revised ??. sciously perceived. Often the threshold is more precisely

1939-1412 (c) 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information. This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TOH.2019.2925339, IEEE Transactions on Haptics

a difference threshold, which is the minimum change in in- and 100, are fixed by the experimenter. There exist many tensity needed to perceive the stimulus as different from variations of this method, all with their own advantages a reference stimulus. For thermal perception the threshold and disadvantages. indicates the minimal difference in temperature (either heat- Finally, some of the studies make use of scoring on a ing or cooling) from a baseline temperature that is noticed Likert scale. Participants have to score a property in more by a participant. The threshold is also termed just-noticeable psychological terms like dislike–like, not comfortable–very difference or JND. comfortable, etc. Such scales are typically used in question- Thresholds can be measured in different ways and the naires after an experiment. outcome will depend on the method. A method often used There are many books and papers that explain the above- for thermal thresholds is the method of limits (MoL). Starting mentioned terms and methods in much more detail. A from a baseline temperature (usually close to skin tempera- selection can be found here: [4], [5], [6], [7], [8]. Thermal ture), the temperature of the stimulus increases or decreases physiology and thermal receptors lie outside the scope of in small steps until the participant indicates that s/he no- this survey, but readers interested in these topics can find a ticed a change in temperature. Usually, this procedure is selection of papers here: [9], [10], [11]. repeated several times and the average end temperature is defined as the threshold. In this way, separate warm thresh- 3 PSYCHOPHYSICS olds (WT) and cold thresholds (CT) can be determined. In some studies, the stimulation reverses sign when a threshold is 3.1 Thermal Detection - Basic Psychophysics reached and stimulation continues to the opposite threshold. There are quite a number of studies that investigated how Thus, one goes back and forth between the warm threshold well participants can detect a change in temperature (see and the cold threshold. The average value of this difference Table 1). Thermal detection thresholds are the smallest dif- is the so-called difference limen (DL). ference in temperature that can be detected when the skin Another method to determine thresholds is the two- warms or cools starting from a certain baseline temperature. alternative forced choice method (2AFC). In each trial, a partic- The detection thresholds will somewhat depend on the ipant is presented with two stimuli and s/he has to choose method used to determine these thresholds. The studies which of the pair contained the actual stimulation. This described in this subsection focus on fundamental aspects procedure is a more sensitive way to measure thresholds, of thermal detection. but it is also much more time-consuming as many pairs Kenshalo et al. [12] first adapted the skin of the dorsal have to be presented to the participants. A variation of this forearm to a temperature of 31.5◦C. Using the method of method is the three-alternative forced choice (3AFC) method, limits (MoL) they determined both warm and cold thresh- where the participant has to choose which of three intervals olds (WT and CT) for different rates of temperature change contained the stimulus. Such forced choice methods can (RoC). The warm thresholds were about 0.4◦C if the rate be used either in combination with the method of constant of change was 0.1◦C/s or higher. Lower rates of change led stimuli, where all stimulus pairs or triples are presented an to substantial increases of the threshold. The cold thresholds equal number of times, or with a staircase procedure. With were about 0.2◦C if the rate of change was 0.1◦C/s or higher this latter procedure, the difference between the stimuli of and thus somewhat smaller than the warm thresholds. The a pair is made smaller after one or more correct answers same group also showed that warm thresholds measured and made larger after one or more incorrect answers. Af- with either radiant stimuli or with contact stimuli are sim- ter how many correct or incorrect answers the difficulty ilar [25]. Also Gray et al. [13] measured warm and cold reverses determines the percentage correct to which this thresholds, for both adults and children. They used a 2AFC procedure converges. For example, a one-down/one-up staircase method in which pairs consisting of one stimulus staircase converges to a 77.85% correct threshold, whereas with a stable temperature and one stimulus with a warming a four-down/one-up staircase converges to 85.84% [3]. or cooling stimulus were presented to participants, who had The physical characteristics of the stimulus will influ- to decide which of the stimuli of a pair contained the chang- ence the threshold. An important parameter for thermal ing temperature stimulus. Although thresholds could differ experiments is the rate of change or RoC. A change in the about an order of magnitude between participants, within temperature of stimulation cannot be instantaneous and the participants the thresholds turned out to be quite stable. rate of change indicates whether this change occurs slow Age and gender did not influence the thresholds. Like in or fast. If the rate of change is very small, the change the study by Kenshalo et al. [12], they found somewhat in temperature might go unnoticed, leading to increased smaller thresholds for cooling. Using a more sophisticated thresholds (see next sections). set-up to measure thresholds, Jamal et al. [14] report smaller Another common psychophysical method which is also thresholds than the earlier studies. Warm and cold thresh- used in some of the studies discussed in this paper is magni- olds measured on wrist, forearm or thigh hardly differed tude estimation. In this method, the task of the participant is (about 0.23 and 0.15◦C, respectively), but warm thresholds to rate the “magnitude” of the stimulation. This magnitude measured on the ankle were substantially larger (1.35◦C). can be any property such as intensity, size, or duration. In The influence of much higher rates of change (1.4, 2.4, most cases, the scale participants have to use for their rating and 3.9◦C/s) on the warm and cold thresholds on forearm, is arbitrary, as long as they use the scale consistently over leg, cheek and hand was investigated by Pertovaara and all stimuli: if a stimulus is perceived as having twice the Kojo [15]. For all tested skin regions, they found an increase magnitude of another stimulus, they should double their of the warm threshold with increasing rate of change, while rating. Sometimes the extremes of the scale, for example 1 the cold thresholds remained the same. Although this seems

TABLE 1 Thermal detection - basic psychophysics.

Reference Location Participants Stimuli Task Outcome Kenshalo et al. arm dorsal 19–27 (3) 31.5◦C 3AFC - WT, CT constant above 0.1◦C/s [12] site near warming, cooling: MoL - rapid increase below 0.1◦C/s elbow 0.3, 0.1, 0.05, 0.02, 0.01◦C/s

Gray et al. [13] volar 19–55 (13) 3.6 cm2 2AFC detection - median threshold: forearm 7–9 (11) staircase warming +1.04◦C, cooling -0.15◦C - no inﬂuence of age or sex - substantial differences between participants - high test-retest correlation Jamal et al. [14] wrist, 6–73 (106) 34–35◦C 2AFC thresholds WT/CT: forearm, warming, cooling: 1◦C/s detection - wrist: 0.23◦C / 0.15◦C thigh, ankle - forearm: 0.24◦C / 0.15◦C - thigh: 0.23◦C / 0.15◦C - ankle: 1.35◦C / 0.17◦C Pertovaara & distal 20–40 (2f, 6m) warming, cooling: detect: - inﬂuence of RoC the same for all body sites Kojo [15] forearm, 1.4, 2.4, 3.9◦C/s - warm, heat, - warm, heat, pain thresholds increase with RoC leg pain, cool - CT independent of RoC

Claus et al. [16] - wrist, 19-78 (77) 35◦C detect - most sensitive: legs with 2.5-2.8◦C/s ankle 16–65 (32) - warming: 0.9–4.2◦C/s - CT lower than WT - leg - cooling: 1.1–2.8◦C/s - 2.5 cm×5 cm Hilz et al. [17] forearm 18–56 (20) 30◦C, 35◦C MoL: - WT: 1.6–2.3◦C 4 cm from - warming, cooling - WT - CT: 1.6–2.2◦C wrist - CT - DL: 3.9–4.6◦C volar site - DL Hilz et al. [18] foot, calf, 3–7 (74) 32◦C MoL: - no difference between age groups forearm - warming, cooling: 1◦C/s - WT, - foot: WT 2.9◦C, CT 3.5◦C - 1.5 cm×2.5 cm or - CT - calf: WT 3.1◦C, CT 3.7◦C 2.5 cm×5 cm - forearm: WT 2.5◦C, CT 2.6◦C - larger probe → lower threshold Hilz et al. [19] foot, calf, 7–18 (225) 32◦C MoL: - larger probe → lower threshold forearm - warming, cooling: 1◦C/s - WT - MoL suitable for quick evaluation - 1.5 cm×2.5 cm or - CT of thermal sensitivity 2.5 cm×5 cm

Hilz et al. [20] foot, calf, 18–80 (225) 32◦C MoL: - larger probe → lower threshold forearm - warming, cooling: 1◦C/s - WT, - MoL suitable for quick evaluation - 1.5 cm×2.5 cm or - CT of thermal sensitivity 2.5 cm×5 cm

Madera et al. forearm 20–29 (2f, 5m) radiant stimuli MoL: - larger RoC gives higher detection rate [21] warming: - WT - larger RoC gives shorter detection time 0.033, 0.01, 0.2◦C/s Stevens & 13 body 18-28 (20) 33◦C staircase 2AFC - sensitivity varies 100-fold over body: Choo [2] regions 40-60 (20) 3 s stimulus leg/foot relatively poor ≥ 65 (20) - warming: 2.1◦C/s presence - sensitivity cold > warmth - cooling: 1.9◦C/s - sensitivity declines with age - sensitivity for warm and cold correlates Lee et al. [22] 12 body ∼21(20m) warming, cooling: 0.1◦C/s MoL - tendency for different thresholds for sites tropical and temperate natives Heldestad foot, leg, 16–72 (75) 32◦C MoL: - thresholds depend on body site Lillieskold¨ & thigh, warming, cooling: 1◦C/s - WT - thresholds higher in lower parts of body North [23] forearm, - CT - WT > CT upper arm

Filingeri et al. 54 locations ∼30 (8m), 31◦C magnitude - thermosensitivity to warm and cold varies [24] on foot ∼28 (8f) 5 s estimation by up to 5-fold across glabrous and hairy sites warming, cooling: 1◦C/s - hairy skin is more sensitive than glabrous skin - sole and dorsum are more sensitive than toes - hands are twice as thermosensitive as feet

For participants age or age range is given in years and between brackets the number of participants (f females, m males). For stimuli the following values are given (if available): skin temperature in ◦C, duration in s, RoC in ◦C/s, and stimulus size in cm×cm. 2(3)AFC stands for two (three) alternative forced choice. MoL is method of limits. WT is warm threshold, CT is cold threshold. DL is difference limen. RoC is rate of change.

to be in contradiction with the results presented by Kenshalo thresholds at all body regions, and the thresholds of the et al. [12], it should be noted that the rates of change used lower parts of the body were higher than those of the upper lie in quite a different range. Claus et al. [16] showed that parts. the legs are most sensitive to thermal stimulation with an In another recent study, Filingeri et al. [24] mapped the rate of change between 2.5 and 2.8◦C/s. thermosensitivity of 54 locations on the foot and 49 on the Hilz et al. [17] determined warm and cold thresholds hand. They found that different regions on the foot may using the method of limits. Their major aim was to compare differ by a factor 5 in thermosensitivity. Hairy skin was these to difference limen (DL) thresholds. The latter thresh- found to be more thermosensitive than glabrous skin, and olds are determined by alternating warm and cold stimula- sole and dorsum were more sensitive than toes. Especially tion; when a participant notices a temperature change, the this latter finding was new and surprising as the opposite is temperature change is reversed. Thus, one of the questions true for touch and pain sensitivity. In comparison, the hand was whether the difference limen threshold is equal to the was twice as thermosensitive as the foot. sum of the warm and cold thresholds. As the warm and cold From the studies mentioned in this subsection the fol- thresholds turned out to be much less variable than differ- lowing general findings emerge: Thermal thresholds vary ence limen thresholds, they advocate a strong preference widely over body location, with mostly lower thresholds for for separate measurements of warm and cold thresholds. higher body parts. Thresholds for warm stimuli are usually The same group measured warm and cold thresholds on somewhat higher than for cold stimuli. The method used to forearm, foot and calf for large groups of children, juveniles measure thermal thresholds has a direct influence, so one and adults in order to obtain normative data for comparison should be careful with comparing thresholds measured in with patients (in particular, patients with neuropathy) [18], different ways. Finally, lower rates of change will increase [19], [20]. They found no difference between the thresholds the detection threshold. of age groups 3–4-, 4–5-, 5–6- and 6–7-years, nor between boys and girls [18]. A larger area of stimulation yielded 3.2 Thermal Detection - Applied lower thresholds [18], [19], [20]. The main conclusions from this series of studies are that the method of limits is well The studies described in the previous subsection aimed at suitable for a quick determination of the thresholds, that gathering fundamental knowledge on how well humans can there is sufficient intertrial reproducibility, that a larger detect thermal stimulation and how this depends on method probe is preferred if possible (i.e. can be placed flat on the and body location. The studies mentioned in the current skin) and that it is not necessary to pre-warm the skin before subsection also use psychophysical methods to determine testing. A recent study by Madera et al. [21] showed that thermal detection with body parts other than the hand or also for radiant stimuli, a larger rate of change leads to a face, but these have clear applications in mind (see Table 2). higher detection rate and shorter detection times. Wilson et al. [26] studied thermal detection at body An extensive investigation of the influence of both age sites where potentially a mobile device could be carried, (between 18 and 88 years) and body region (13 regions, like the fingers, the thenar, the forearm and the upper among which lower and upper arms, belly, back, thigh and arm. They found that the intensity of the stimulus, that calf) on the warm and cold thresholds was presented by is the rise or fall of temperature with respect to base- Stevens and Choo [2]. For all ages, an enormous influence of line, influences detection: higher intensities result in better body region was found: the face was about 100 times more detection and faster response times. Cooling is detected sensitive than the foot. Cold thresholds were everywhere faster than warming. A higher rate of change improves lower than warm thresholds. There was a strong correlation detection, but reduces comfort. Thresholds for the different between the warm and cold thresholds of participants. body sites were as follows: thenar (1.9◦C), forearm (2.2◦C), Finally, the study also showed a decline of thermal sensi- upper arm (2.3◦C) and finger (2.9◦C). Similar tests in an tivity with age, with the most prominent declines in the indoor mobile setting (as opposed to a static situation) extremities. showed that performance dropped significantly, although Lee et al. [22] investigated whether natives from differ- most of the observed patterns and dependencies remained. ent climate zones (tropical versus temperate) had different The same group also investigated the influence of placing warm and cold thresholds. Using the method of limits, a textile between the skin and the thermal stimulus [27]. they tested 12 body regions of 10 individuals living in They placed the stimulus either directly on the skin or on a temperate climate zone and 10 individuals living in a a piece of cotton or nylon on the thenar, the thigh or the tropical climate zone. All experimental tests took place at the waist. Also in these experiments, higher intensity and faster same location, so conditions could be kept the same for all rate of change improved performance. Detection percentage participants. Consistent with previous studies, thresholds depended strongly on material, namely 65%, 47%, and 36% widely depended on body region, with the calf the less sen- for none, nylon and cotton, respectively. This indicates that sitive. They further reported a tendency that the individuals materials with lower thermal conductivity require a higher from the tropical zone have a somewhat lower sensitivity to intensity. However, users rated comfort higher when the detect warmth. stimulation was provided via a textile. In a recent study, the warm and cold thresholds of a Experiments with a quite different focus were done by large group of 75 participants ranging in age from 16 to 72 Bolton et al. [28] who tested a device worn around the years were measured using the method of limits at 8 body wrist capable of gracefully interupting attention of the user. regions [23]. Their results were consistent with the earlier They found, similar to the previously mentioned studies, studies: Warm thresholds were generally higher than cold that intensity and rate of change both influence detection

TABLE 2 Thermal detection - applied.

Reference Location Participants Stimuli Task Outcome Wilson et al. - forearm, 21–57 (9m, 5f) duration: 10 s - detect - intensity influences detection: [26] upper arm 23–41 (10m, 4f) intensities: 1, 3, 6◦C - rate: 53%, 91% and 97% - palm, - warming, cooling: - intensity - cooling detection faster than warming forearm, 1, 3◦C/s - comfort - thresholds: upper arm - conditions: forearm (2.2◦C), upper arm (2.3◦C) - static - response time decreases with intensity - mobile - RoC influences detection - comfort decreased with RoC - static better than mobile Halvey et al. thigh, 22–39 (15) duration: 10 s - detect - intensity influences detection: [27] waist intensities: 1, 3, 6◦C - rate: 19% (1◦C), 58% (3◦C) and 71% (6◦C) - warming, cooling: - intensity - interface influences detection: 1, 3◦C/s - comfort 65% (none), 47% (nylon), 36% (cotton) interface: none, nylon, - RoC influences detection cotton - no detection difference between warming, cooling - detection faster with cooling - detection slower with thigh - comfort decreased with intensity Bolton et al. [28] wrist (15) duration: 10 s detection - intensity influences detection intensities: 0.5, 1, 1.5◦C - RoC influences detection - warming, cooling: - cooling better than warming slow, high, medium

Janjeng & wrist 24 intensities: 3, 6◦C detection - faster detection with RoC 3◦C/s Leelanupab - warming, cooling: rate: - preference for cold stimuli [29] 1, 3◦C/s - comfort - significant influence of ambient different ambient - intensity humidity temperatures - significant influence of ambient temperature - no influence of sudden change of ambient temperature Ketna & wrist 21–30 (12f, 12m) duration: 10 s detection in noisy - detection rate higher with thermal Leelanupab cooling: 3◦C/s or vibratory stimuli [30] environment For participants the age or age range is given in years and between brackets the number of participants (f indicates females, m males). RoC is rate of change.

and that cooling has a stronger effect than warming. Janjeng was significantly better with thermal stimulation than with and Leelanupab [29] focused on the usefulness of thermal the other three types. Performance further improved with stimulation in tropical circumstances. In the tropics, ambient multisensory stimulation. temperatures might change suddenly, for example, when From these studies with a more applied focus, sev- moving from inside an airconditioned building to outside. eral of the findings from the more fundamental studies They reported that both ambient temperature and ambient were confirmed. Among others, these studies also showed humidity have a significant influence on detection of a that different body parts have different thermal detection thermal stimulus, with higher detection rates for higher thresholds, that faster rates of temperature change improve temperatures, and lower detection rates for humidity above detection and that cooling is detected faster than warming. 60%. Although they also conclude that sudden change of The more applied findings from these studies are harder to ambient environment has no influence on detection, it is not generalize as these are only based on single studies. The clear on which data that conclusion is based. From subjec- most relevant findings are that thermal detection becomes tive comfort ratings it followed that participants preferred harder in mobile conditions and that easier detection due to the cooling stimulus. Ketna and Leelanupab [30] tested higher intensity of the stimulus or faster rate of change has whether thermal feedback would be suitable in a noisy or a a negative effect on experienced comfort. bumpy environment. The noisy environment was created by exposing the participants to 120 decibel noise from a 3.3 Thermal Adaptation concert; in the bumpy environment condition, participant It is of interest to know how thermal perception changes as had to stand on a pivotal vibration plate machine with a function of adaptation to a certain temperature. Only one 30 hertz vibration. They compared the detection rates of study investigated thermal adaptation at a body site other two types of auditory stimulation, vibratory stimulation than the hands (see Table 3) [31]. The focus of this study lies and thermal stimulation in noisy or bumpy environments. on introducing a better method to measure adaptation and The stimuli could be presented either in isolation or in to investigate the time course of adaptation in more detail. various combinations. In both environments, performance At the start of the experiment, participants first had to sit

TABLE 3 Thermal adaptation.

Reference Location Participants Stimuli Task Outcome Kenshalo & Scott forearm (2f, 2m) size: 14.4 cm2 adjust just detectable - complete adaptation after 25 minutes [31] RoC: 0.3◦C/s warm or cold - rapid adaptation for temperatures close pressure: 11.8 g/cm2 to skin temperature - adaptation above skin temperature more rapid than below skin temperature

The number of participants is given between brackets (f indicates females, m males). RoC is rate of change.

for 20 minutes in an airconditioned room. After this period, on the forehead was smallest and for the calf highest, with the temperature of their dorsal forearm was measured. cheek, chest, abdomen, shoulder, back, forearm, upper arm Subsequently, a Peltier element with the same temperature and thigh in increasing order in between. The same group as the skin was placed on the dorsal side of the forearm. also showed strong spatial summation for cooling stimuli on The task of the participant was to adjust the temperature both the back and the forearm [35]. However, in contrast to of the element such that its temperature was just detected the convergence found for higher levels of stimulation with as warm or cold (in different trials). Every five minutes, the warm stimuli [32], [33], no such convergence was found for experimenter changed the temperature of the element back cooling stimuli. This means that for cold stimulation near to neutral. Over a period of 40 minutes, the temperature the pain threshold, the relative contributions of size and difference needed to detect the temperature of the element intensity remained the same. as warm or cold increased from just a part of a degree to Rosza´ and Kenshalo [36] investigated whether thermal about 4◦C, clearly indicating adaptation. This temperature spatial summation also occurs for non-adjacent areas of range was much smaller than reported by other studies. stimulation. By stimulating either one forearm or both fore- The authors attribute this difference to the better control arms simultaneously, they could show that indeed spatial of their stimulus and experimental conditions, and size and summation also takes places over the two arms. They tested location of the stimulated area (hand versus forearm). They this for 5 different adaptation temperatures and 3 differ- also showed that adaptation above skin temperature occurs ent intensities (near threshold and suprathreshold) and in more rapidly than below skin temperature. all conditions they found spatial summation. The authors conclude that their results indicate that spatial summation occurs in the central nervous system. 3.4 Thermal Spatial Summation The studies described in this subsection all show that the An interesting thermal perceptual phenomenon is that of perceived sensation of thermal stimulation depends on both thermal summation. A stronger thermal sensation can be the area and intensity of stimulation. Doubling the intensity obtained by either increasing the area of stimulation or has the same perceptual effect as doubling the area. This increasing the intensity of the stimulation [37] (see also was found for all body parts tested. Interestingly, this was Table 4). As the intensity of the stimulation is summated even found for areas of stimulation that were far apart, as in over the area (i.e. thermal summation), spatial resolution of the case of stimulation on two forearms. thermal stimulation is quite poor. Stevens and Marks [32], [33] studied spatial summation 3.5 Thermal identification and localization of warmth on the back. With a magnitude estimation task, they showed that for a given area of stimulation, the per- For the design of thermal devices, it is of eminent impor- ceived warmth increased as a power function of intensity. tance to know how much thermal stimuli need to differ in The exponent of the power function was smaller for larger order to be distinguishable (see Table 5). In this subsection areas. The power functions for the various areas converged studies are described that investigate how well participants near the level of painful stimulation. This research was can identify temperature differences or localize stimuli. extended to include also forearm, upper arm, shoulder, The study by Song et al. [38] reports a recognition ex- chest and calf as investigated body regions, with basically periment in which participants had to recognize whether a the same results: strong spatial summation of intensity over thermal stimulus was 3 or 6◦C below or above skin temper- area [34]. Kenshalo et al. [25] showed that different methods ature, or just skin temperature. Of the total of 180 trials (15 of stimulation, namely radiant stimuli or contact stimuli, trials for each of 12 participants) only 70 stimuli were identi- lead to similar results. They also showed that at threshold, fied correctly. However, there were huge differences among the product of area and intensity is constant, indicating that participants, with some participants apparently performing a larger area can compensate a lower intensity, and vice close to chance level. The main confusions were made versa. between -3 and -6◦C, and between 3 and 6◦C, although also An important difference between the various body re- confusions between hot and cold stimuli occurred, causing gions was that the same stimulus caused quite different 22% of the errors. magnitude estimates, indicating that the stimulus warmth In the study by Taus et al. [39] participants received was perceived differently depending on body region, es- radiant stimuli on their volar forearm, and they were asked pecially for relatively weak stimulation [34]. For an equal whether the stimulation was closer to their wrist or to magnitude estimate, the intensity of the stimulation needed their elbow. For reference before each trial, the midline of

TABLE 4 Thermal spatial summation.

Reference Location Participants Stimuli Task Outcome Marks [32], back (14m) 3 s magnitude - subjective warmth grows with area and Stevens & Marks 31 – 324 cm2, estimation intensity [33] 36 – 282 mW/cm2 - rate of growth depends on area - convergence near pain threshold Stevens et al. [34] forearm, calf, (18m/2m) - 5 sizes magnitude - subjective warmth grows with intensity back, shoulder, - 6 intensities estimation - rate of growth depends on area chest, abdomen, - forearm and calf less sensitive than upper arm forehead - perceived intensity depends on location Kenshalo et al. [25] forearm, back (1f, 1m) 3 s MoL - complete summation: radiant stimuli: A (area) x I (intensity) = constant 1, 2, 3, 4, 6, 8, 12 cm2 - back least sensitive contact stimuli: - radiant and contact stimuli give same 1.7, 7.1, 14.4 cm2 results Stevens & Marks forearm, back (15) 2.0 – 19.6 cm2 magnitude - strong spatial summation [35] cooling: 2 – 12◦C estimation - subjective cold depends on intensity and RoC > 1◦C/s size of area - cold and warm summation follow different patterns

Rosza´ & Kenshalo forearms 20–26 (3m) 18.4 cm2 report: spatial summation over the two arms: [36] 19 cm from wrist cooling 3 s -temperature - both near threshold and suprathreshold RoC: 1◦C/s change - at all adaptation temperatures AT: 24, 28, 32, 36, 40◦C (yes/no) - at all intensities 3 intensities - conﬁdence unilateral/bilateral For participants the age or age range is given in years and between brackets the number of participants (m indicates males, f females). For stimuli the following values are given (if available): duration in s, stimulus size in cm2, intensity in mW/cm2, temperature difference in ◦C, and RoC in ◦C/s. MoL is method of limits. WT is warm threshold, CT is cold threshold. RoC is rate of change. AT is adaptation temperature.

TABLE 5 Thermal identiﬁcation and localization.

Reference Location Participants Stimuli Task Outcome Song et al. [38] wrist 24–30 (8f, 4m) -6, -3, 0, +3, +6◦C with identify - RT not dependent on stimulus respect to skin temperature - many errors (110 of 180) - confusions: -3◦C and -6◦C and, +3◦C and +6◦C Taus et al. [39] forearm (3m) 4 cm×4 cm localize: - 59% correct for lowest intensity, shortest 6 sites 68, 138, 224, 398 mW/cm2 distal or proximal distance 3 s - 94% correct for highest intensity, greatest distance

Cain [40] torso exp 1: (2m) exp 1: exp 1 (dorsal side): exp 1: - dorsal exp 2: (6m) - 3.75 cm×2.5 cm - separated or not - modest spatial acuity at all irradiance - ventral - 93, 156, 256, 433 mW/cm2 (with feedback) levels - separation: 0, 1, 3, 5, 7, 9 cm - rate conﬁdence - stimuli never felt separated exp 2: exp 2: exp 2: - 348 mW/cm2 - localize (dorsal - 13.5% errors - diameter 7.8 cm or ventral?) - ventral and dorsal similar confusion rate

Tewell et al. forearm (2f, 10m) 7 stimuli identify on Likert - single site: 35% correct [41] 3 sites - 29, 30, 31, 32, 33, 34, 35◦C scale - 3 sites: 49% correct - 1, 2, 3 sites - mixed sites: 44% correct For participants the age or age range is given in years and between brackets the number of participants (f indicates females, m males). For stimuli the following values are given (if available): temperature difference with respect to skin temperature or stimulus temperature in ◦C, stimulus size in cm×cm or diameter in cm, intensity in mW/cm2, duration in s, separation between two stimuli in cm, stimulus temperature in ◦C, and number of simultaneously stimulated sites. exp stands for experiment. RT is reaction time.

their arm, halfway between wrist and elbow, was touched stimulus intensity. For the lowest intensity and shortest by the experimenter. Distance from the midline was either distance, performance was only just above chance level, but 2.5, 4, or 5.5 cm. Performance increased with distance and for the highest intensity and largest distance a performance

level of 94% was achieved. For one participant, the authors stimulation is intensified. compared this performance with a similar experiment in which light touch was used as stimulation. In this condition, 3.6 Illusory Thermal Patterns almost 100% correct was reached, indicating far superior performance with tactile as compared to thermal stimula- All sensory modalities are susceptible to perceptual illu- tion. sions. Many of the visual and auditory illusions are well- Cain [40] studied thermal resolution on the torso. In known, but there exist also haptic and tactile illusions [42]. his first experiment, participants were stimulated on their Based on the perceptual thermal properties described above, back on the right side of their body midline. The areas several authors have been able to create illusory thermal of two irradiant stimuli either touched or were separated stimuli and that is the topic of this subsection. by some distance, that varied systematically over sessions. As presented in Table 1, thermal thresholds depend Participants had to decide whether the areas were connected on the rate of change of the stimulation: higher rates of or not and what the confidence in their answer was. After change result in faster detection and thus lower detection every trial they received feedback about the correctness thresholds. In three studies [43], [44], [45], the authors make of their answer. Spatial acuity was relatively modest at use of this phenomenon in the creation of a stimulus that all levels of irradiance. Interestingly, even when the two is perceived as continously cooling, whereas actually the stimuli were separated by a distance of 9 cm, participants average temperature remains constant (see Table 6). They reported that they never really felt separated stimuli. An term this asymmetric cooling. The forearm is stimulated informal additional experiment showed that this was even with a small number of quickly cooling Peltier elements and the case with a separation of 23 cm. In a second experiment, a larger number of slowly heating elements. As the heating participants had to decide whether an irradiant stimulus of the elements is so slow that it is below threshold, the was presented to the front or to the back of their torso. Even heating is not perceived, resulting in the illusion of constant with these widely separated stimuli they made an error cooling. Manasrah et al. [43] investigate in detail the optimal in 13.4% of the trials. These experiments clearly show that arm locations, and cooling and heating cycles to induce thermal resolution is quite poor. this illusion. Of the three locations tested, stimulation of In one experimental condition in a study by Tewell et the dominant posterior forearm gave the strongest thermal al. [41], participants had to recognize the relative temper- sensation, as compared to the dominant and non-dominant ature of stimuli ranging from 29 to 35◦C in steps of 1◦C anterior forearms. Time cycles (in s) of heating/cooling of presented to a location in the center of their forearm. Perfor- 30/10 and 21/7 gave a stronger effect than 45/15. The same mance was only 35% correct (chance level 14%). In another authors [44] also investigate the opposite illusion (asym- condition, the same stimuli were presented simultaneously metric heating), a stimulus that is perceived as constantly to three locations on the forearm. The authors termed this heating. Here, the cooling of the elements is so slow that is the amplification condition, as now three locations were it below the perception threshold and thus only the heating always stimulated simultaneously and hence the total ther- stimulation is perceived. Both illusions were further investi- mal intensity was increased. In this condition, performance gated by Hojatmadani and Reed [45]. Among their findings increased somewhat to 49%. In a third condition, termed was a counterintuitive effect that asymmetric heating was quantification condition, 0, 1, 2 or 3 locations were presented perceived as cooling. simultaneously with either the lowest or the highest temper- A well-known thermal illusion is the thermal grill il- ature (so again 7 levels of intensity). The stimulus with the lusion, first demonstrated by Thurnberg in 1896 and later lowest temperature consisted of stimulation of 3 locations investigated in more detail by Craig and Bushnell [52]. with the minimum temperature (29◦C), and the warmest Placing one’s hand on a thermal “grill” consisting of inter- stimulus consisted of stimulation of 3 locations with the laced warm and cool bars of temperatures that are all well maximum temperature (35◦C). Stimuli in between consisted within harmless levels of warm and cold illicits a burning of 1 or 2 stimulated areas with the warmest or coldest and painful sensation. In a rather preliminary study, Oron- temperature, and one in which no stimulation was pre- Gilad et al. [46] describe their intention to use this illusion sented. Also in this condition, a similarly low performance as the basis for a tactile language. They developed a Ther- (44%) was achieved. The latter two methods achieved sig- moelectric Tactile Display (TTD) consisting of three thermal nificantly better performance than the single stimulus con- elements. Interestingly, they report that already with only dition, which is probably caused by the larger differences two thermal elements, one cold and one warm, participants in intensity between the 7 stimuli. The authors conclude experienced the thermal grill illusion. Unfortunately, they that the amplification method is the most promising for did not continue their research on this topic. indicating thermal states, but because the difference with Young [47] presented participants with an array of 4×4 the quantification method was only slight, both methods Peltier elements on their forearm. In the first phase of a merit further investigations (see Section 3.7). trial, all elements had the same temperature within a range The studies in this subsection show that thermal reso- of 22.5–37.5◦C. In the second phase, half of the elements lution is at most moderate. Error rates in identification and increased in temperature, the other half decreased, keeping localization tasks are quite high. Separated thermal stimuli the average temperature, and thus also skin temperature can be perceived as one connected stimulus, and even if constant. Cooling and warming was done in a checkerboard participants have to decide whether a thermal stimulus pattern. In the third phase, the temperatures of the elements is presented to their back or to their belly, they are far remained at the level they had reached during the sec- from perfect. Performance increases somewhat when the ond phase. Participants were constantly required to report

TABLE 6 Illusory Thermal Patterns.

Reference Location Participants Stimuli Task Outcome Manasrah et al. forearm 18–55 (21) 12 Peltier elements describe thermal - a feeling of decreasing temperature was [43] (dom ant, asymmetric cooling and sensation elicited (without actual change in skin non-dom heating temperature) ant, 30/10, 21/7, 45/15 - strongest thermal sensation at dominant dom pos) heating/cooling time posterior forearm cycles (s) - 30/10 and 21/7 heating/cooling time cycles give stronger sensation than 45/15. Manasrah et al. forearm 18–55 (10) 12 Peltier elements - cooling threshold - a feeling of increasing temperature was [44] asymmetric cooling and - describe thermal elicited (without actual change in skin heating sensation temperature)

Hojatmadani & forearm 18–34 12 Peltier elements describe thermal - asymmetric heating created a cold Reed [45] (7f, 11m) asymmetric cooling and sensation perception heating - objects at skin temperature are perceived warmer

Oron-Gilad et al. forearm (3) warm/cold - thresholds - very informal pilot experiments [46] 3 Peltier elements - thermal grill - thermal grill illusion can be generated 1.5 cm×1.5 cm with only two “bars” Young [47] forearm 15–50 (5f) 16 Peltier elements report thermal - checkerboard pattern perceived warmer cooling and heating in sensation (colder) than homogeneous pattern for checkerboard pattern average skin temperatures above (below) 30◦C Watanabe et al. [48] forearm 21–25 (6m) warm/cold/neutral select perceived - thermal stimulation at one site inﬂuences 2 Peltier elements sensation (7 choices) perception of other site (thermal referral) 4 cm×4 cm - strong asymmetry between wrist and elbow site

Arai et al. [49] forearm 20–39 (10m) hot (44◦C)/cold (11◦C)/ report thermal - thermal referral: neutral site inﬂuenced neutral (32◦C) sensation of each by neighbor 3 Peltier elements spot (hot/null/cold) - confusion when one spot hot, other spot 4 cm×4 cm cold - confusion with hot/cold/hot or cold/hot/cold pattern - some participants perceived pain at center

Singhal & Jones forearm 24–29 4 pulses localization - spatio-temporal illusions [50] (1f, 9m) -8◦C - second pulse perceived towards third 2 s pulse if delay short Singhal & Jones forearm 24–36 (10m) 4 pulses localization - spatio-temporal illusions [51] 6◦C - second pulse perceived towards third 2 s pulse if delay short For participants the age range is given in years and between brackets the number of participants (f indicates females, m males). For stimuli the following values are given (if available): stimulus (Peltier) size in cm×cm, stimulus/pulse duration in s, stimulus temperature in ◦C, or difference in temperature with respect to skin temperature in ◦C. dom stands for dominant, ant for anterior, and pos for posterior.

their thermal sensation by adjusting a dial ranging from cold stimulus, the temperature felt by the middle finger was “painful cold” via “neutral” to “painful hot”. It was found the same, even though the actual temperature of the stim- that the checkerboard type of stimulation in phase 2 was ulus was neutral. The author termed this illusion “thermal perceived as warmer than the homogeneous stimulation referral”. Watanabe et al. [48] investigated how two thermal in phase 1 for skin temperatures above 30◦C, whereas for stimuli on the forearm, one near the wrist and one near the skin temperatures below 30◦C the checkerboard pattern elbow, and either of a warm, cold or neutral temperature, was perceived as cooler than that of the homogeneous would be perceived. As was expected on the basis of the temperature, even though the actual skin temperatures were thermal referral phenomenon, the perceived temperature at the same. Although the author did not discuss his findings one location was influenced by the temperature of the other in terms of the “thermal grill” illusion, it might be that his location, although this influence was asymmetrical: The findings could be thought of as a ”thermal grid” illusion. In temperature perceived near the elbow was influenced more both cases, the perceived temperature of the stimulation is by the temperature of the wrist stimulation than vice versa. changed towards more extreme temperatures. Some participants even experienced a burning sensation at the elbow such as in the thermal grill illusion when the wrist Green [53] showed the phenomenon that when the index was stimulated with a cold stimulus and the elbow with and ring fingers were stimulated with either a warm or a

a warm stimulus, or vice versa. Thermal referral on two and they were asked to explore the possible use of mes- locations on the forearm was confirmed by Arai et al. [49]. saging and to keep track of when and where they used In addition, these authors also tested simultaneous stimu- it. In all three studies it turned out that the devices were lation of three locations, yielding similar results. Moreover, used in many different ways, such as gaming, wishing if the center stimulus was different from the outer stimuli good night, secret messaging when others were present, (hot versus cold or vice versa), the perception was often warnings, and wake up calls. They conclude that there is confusing (“paradoxical”) and some participants perceived quite some potential for the use of thermal messaging in this as a painful sensation (like in the thermal grill illusion). interpersonal communication. An interesting tactile illusion is the so-termed “cuta- Suhonen et al. [59] gave their participants a hand-held neous rabbit” [54]. When the forearm of a participant is device they could squeeze in order to give a thermal stim- stimulated with a series of brief pulses at three locations, ulus to the forearm of another person, either their partner for example five pulses near the wrist, then five halfway the or a friend. The pairs were seated in different rooms. They forearm, followed by five near the elbow, the perception will were instructed to discuss various specific topics while em- be that of a series of more or less equally spaced pulses from phasizing their emotions by using the haptic device. Warm wrist to elbow (hence the name cutaneous rabbit). Singhal and cold stimuli were consistently used for positive and and Jones [50] investigated whether a similar illusion could negative emotions, respectively. The Thermal Array Display be created with thermal stimuli. Participants placed their (TAD) described by Tewell et al. [41] consists of three Peltier forearm on a device with three Peltier elements with con- elements placed on the volar forearm. In the experiment, all tacts near the wrist, halfway and near the elbow. They tested three elements always had the same temperature of either the perception of various patterns of four cooling pulses 1◦C, 2◦C, or 3◦C above or below a neutral temperature of (two of the pulses were generated by the same element). 32◦C. These thermal stimuli were combined with text mes- They indeed found patterns where the perception of the lo- sages and participants had to rate the valence of the message cation of the second pulse was substantially shifted towards and how much they were aroused by it. The valence of the location of the third pulse. The same authors showed the message was not influenced by the temperature of the that a similar shift can be found with patterns consisting of device, but especially for neutral messages, temperature warm pulses [51]. may influence the emotional arousal of the user. The studies in this subsection have shown that thermal Gooch and Watts [60] proposed a thermal harness con- perception is far from veridical. It is possible to create the sisting of a waist belt in which three Peltier elements were perception of ever increasing or decreasing temperatures placed. Warming up of these Peltier elements was intended without actually changing the average skin temperature. to simulate the warmth of the arm of a partner placed With temperatures well within the normally acceptable around the waist. Pairs of participants always consisted of range it is possible to create a very unpleasant burning one male and one female who had been close friends for effect, and thermal stimulation at one site influences the per- at least a year. One of the pair (the heatee) wore the belt, ceptual temperature at another site. Most of these illusions the other (the heater) had access to a “thermal hug button” can be understood or were even predicted on the basis of the that could activitate the belt. The study showed that thermal perceptual thresholds and effects described in the previous stimulation could increase the feeling of social presence subsections. between users. The studies in this subsection showed that there is some potential to use thermal stimulation for communication, 3.7 Thermal Communication although this should still be investigated during longer A number of studies investigated the possibilities of thermal periods of time and outside the lab. Possibly, it was just devices for communication over a distance (see Table 7). In fun to try it out for some time within an experimental set- these experiments mostly couples or families participated ting. Warm stimulation was mostly associated with positive and either the wrist [55], [56], the forearm [41], [57], [58], [59] emotions and cold stimulation with negative. or the waist [60] was the body location used for stimulation. The intention of the devices is often to convey feelings of affection (e.g. [41], [55]) or social presence (e.g. [60]) to 3.8 Thermal Applications the partner or child at a distance. Other studies let the There are a few studies that use thermal stimuli in an appli- participants free in the use of the device and investigated cation and test performance in controlled experiments (see how and in which circumstances they use the possibility of Table 8). Klamet et al. [61] describe and test a device they thermal messaging (e.g. [56], [57], [58]). termed “WeaRelaxAble”. As the name already suggests, A communciation tool termed “Lovelet” [55] gives users the device is intended to relax stress by means of thermal information about the temperature of the surroundings of stimulation (and not discussed here, also audio, light and their partner by displaying a certain color on a small LED. vibration). Body sites that received warm or cold stimula- If they noticed that their partner was in a cold environment, tion were shoulder, loin and groin. The cold stimulation was they could remotely stimulate a Peltier element on their either not felt at all or perceived as a strange sensation. On partner’s wrist to warm up. Two couples used this device the other hand, warm stimuli were rated as pleasant and during 20 days. It was found that the couples also used relaxing. A second experiment in which half of the partici- the device spontaneously during telephone calls or while pants were allowed to actively use the stimulation during a chatting. In three studies [56], [57], [58], pairs, couples or series of challenging tasks did not reveal a significant stress families, were provided with a prototype thermal device reduction.

TABLE 7 Thermal Communication.

Reference Location Participants Stimuli Task Outcome Fujita & Nishimoto wrist (2 pairs) warming the partner communication - informal experiment [55] - couples used device over 20 days Lee & Lim [56] wrist 6–39 warm or cool messages families and couples - exploration of interpersonal (11; 2 couples, had to choose their communication 2 families) own type of use - usage: ∼4 messages per person per day during 14 days

Lee & Lim [57] forearm (2 pairs) sending or receiving pairs had to - potentially of interest in interpersonal messages choose and report communication their own type of use

Lee & Schiphorst forearm 8–9, 32–41 warm stimulus from - interpersonal - usage: ∼3 messages per day during 14 [58] (7 families) parent to child communication days - report experience - used for different purposes and context - children interpret messages in different ways

Suhonen et al. [59] forearm 19–30 at least 4◦C above or discuss: - cold signaled negative or disagreement (10 pairs) below skin temperature - happy days - warmth signaled positive or agreement - sad event - stimulation enables emphasizing - restaurant questionnaire Tewell et al. [41] forearm (3f, 9m) 7 thermal stimuli: - rate valence - thermal stimuli cause arousal, but 29, 30, 31, 32, 33, 34, or - rate arousal provide no valence 35◦C combined with 5 - in neutral messages, temperature may text messages inﬂuence emotion Gooch & Watts [60] waist (10 pairs) thermal hug via harness questionnaires - difference in social presence between to the other person rate heaters and heatees - closeness - togetherness For participants the age range is given in years (if known) and between brackets the number of participants (f indicates females, m males).

TABLE 8 Thermal applications.

Reference Location Participants Stimuli Task Outcome Klamet et al. [61] shoulder, - 25–45 Peltier element rate - half of participants did not notice cold loin, (8f, 7m ) heat, cold - comfort stimuli groin ∼30 - relaxation - heat at loin more pleasant than at (14f, 12m) shoulder or groin - heat at loin or shoulder more relaxing than at groin Kotsev et al. [62] forearm 18–28 ±4◦C (easy) play computer game - RT (easy) > RT (hard) (4f, 9m) ±3◦C (hard) - correct inputs (easy) > (hard) RoC 1◦C/s - no effect on power bars, area of effect, cool down Tewell et al. [63] forearm ∼32 29◦C, 35◦C navigate 4 mazes - total time shorter with thermal feedback (3f, 9m) - fewer turns with thermal feedback

For participants the age or age range is given in years and between brackets the number of participants (f stands for females, m for males). For stimuli, the temperature difference with respect to skin temperature is given in ◦C or stimulus temperature in ◦C. RoC is rate of change, RT is response time.

Kotsev et al. [62] investigated whether thermal stim- used again) and Area of Effect (AoE; area in which an effect ulation on the forearm by means of a Peltier element is will take place). They conclude that especially QTE and suitable as feedback mechanism in a video game instead AoE are potentially of interest to use as thermal feedback of more commonly used vibratory or visual information. mechanism in video games, possibly in combination with They investigated four feedback mechanisms that are often other means of feedback. implemented in video games, namely Quick Time Events Another game application is the “Heat-Nav”, a thermal (QTE; requirement to quickly perform a certain action), device suitable for navigating through a twodimensional Power Bars (PB; stop a changing power value at a preferred maze [63]. Participants wore three Peltier elements on their level), Cooldowns (CD; waiting time till some ability can be forearm. After some piloting, the authors found that using

just two temperatures (either warm 35◦C or cold 29◦C) published in the literature. Therefore, this overview will was most effective in guiding the participant through the help researchers working on thermal devices to make good maze. As in a familiar game, “warm” signalled that they choices for types of stimulation (e.g., warming, cooling, were moving the cursor in the right direction, and “cold” rate of change of temperature), and the body regions most signalled a wrong direction. Although thermal stimulation suitable for their intended applications. is rather slow, most participants already noticed cooling or From what we found, it is clear that thermal devices warming of the device and they interpreted this information should not focus on conveying complex messages. Human correctly. thermal resolution is rather poor and thermal summation The three very different studies described in this sub- and thermal referral will complicate and thus limit the in- section showed that thermal stimulation might be of use in formation detail that can be transmitted within a given time. some applications. Probably not surprisingly, warm stimu- However, some of the studies on thermal communication lation had a relaxing effect. In a gaming setting, navigating (e.g., [55], [56], [58]) showed positive effects of thermal mes- a maze was possible with thermal feedback and in a com- saging and people seemed to like the thermal stimulation puter game, thermal feedback could potentially be used in over the (short) period of testing. In communication, warm combination with other types of feedback, such as visual or had positive associations, whereas cold was interpreted auditory feedback. as negative. It might be worthwhile to develop devices specifically for persons with combined visual and auditory disablilities. For these persons, communication is often hard 4 CONCLUSIONS and emotions of other persons go often unnoticed. Also, As could have been expected in advance, thermal phenom- children with deafblindness have no idea how far away ena like adaptation and spatial summation, that have been their caretaker is and how soon this person will come back. reported for hands, fingers and face, were also found for Possibly, such information could be conveyed by means of other body parts. However, an important fact that becomes a thermal device. clear from this overview is that thermal sensitivity varies Finally, there were some positive findings in gaming widely over the body, with mouth, face and fingers very sen- studies. Navigating a maze was possible with thermal feed- sitive, while foot and leg have poor sensitivity [2]. Moreover, back and in a video game, thermal feedback in combina- the same stimulus presented to different body parts will be tion with other types of feedback improved performance. perceived as different in intensity [34]. Humans are more Adding thermal feedback to video games might therefore sensitive to cooling stimuli than to warming stimuli, but if increase the feeling of immersiveness. Only very few studies asked for preferences in, for example, communication, they investigated these aspects, so there is much room for further prefer the warm stimulation. A stronger thermal sensation research. We conclude that the potential interest of thermal can be obtained by either increasing the area of stimulation devices lies in the areas of both gaming and communication. or increasing the intensity of the stimulation [37]. As the intensity of the stimulation is summated over the area (i.e. thermal summation), spatial resolution of thermal stimula- ACKNOWLEDGMENTS tion is quite poor. The authors have received funding from the European When using thermal stimulation in an application or Union’s Horizon 2020 research and innovation programme experiment, it is essential to take all following factors into under grant agreement No 780814, project SUITCEYES. account, as all of these will influence the results: rate of temperature change, size of stimulated area, and body location. One should also be aware of possible influences REFERENCES of thermal illusionary effects, such as thermal referral (e.g. [48]), where thermal stimulation at one location influences [1] L. A. Jones and H.-N. Ho, “Warm or cool, large or small? The the perception at another location, and the thermal grill (e.g. challenge of thermal displays,” IEEE Transactions on Haptics, vol. 1, no. 1, pp. 53–70, 2008. [52]), where a grid of warm and cold stimulations might [2] J. C. Stevens and K. K. Choo, “Temperature sensitivity of the body create a painful sensation. surface over the life span,” Somatosensory & Motor Research, vol. 15, Clearly, our knowledge of the potential use of ther- no. 1, pp. 13–28, 1998. mal stimulation on other body parts than hands or face [3] M. A. Garc´ıa-Perez,´ “Forced-choice staircases with fixed step sizes: asymptotic and small-sample properties,” Vision Research, vol. 38, is severely limited. Moreover, the studies that have been no. 12, pp. 1861–1881, 1998. performed with these devices are usually quite informal. [4] K. R. Alexander and M. G. Brigell, “Psychophysical techniques,” in The question is why this is the case. Is this due to technical Disorders of Visual Processing – Handbook of Clinical Neurophysiology, G. Celesia, Ed. Elsevier B.V., 2005, vol. 5, ch. 9, pp. 167–188. limitations, lack of interest by researchers, disappointing [5] L. Jones and M. Berris, “The psychophysics of temperature percep- pilot results or simply because this research field is rather tion and thermal-interface design,” in Proceedings 10th Symposium new? We do not know the answer to this question, but fact on Haptic Interfaces for Virtual Environment and Teleoperator Systems, is in any case that the research field is new. Therefore, we HAPTICS 2002, 2002, pp. 137–142. [6] L. A. Jones and H. Z. Tan, “Application of psychophysical tech- conclude that there is certainly room for further exploration niques to haptic research,” IEEE Transactions on Haptics, 2013. and more rigorous experiments. [7] H. Moskowitz, “Magnitude estimation: notes on what, how, when The present overview provides knowledge of human and why to use it,” Journal of Food Qualit., vol. 3, pp. 195–227, 1977. thermal perception such as detection thresholds, resolution, [8] D. G. Pelli and B. Farell, “Psychophysical methods,” in Handbook of Optics, 2nd ed., M. Bass, E. W. Van Stryland, D. R. Williams, and dependence on body parts and illusory effects. It also pro- W. L. Wolfe, Eds. New York: McGraw-Hill, 1995, vol. I, ch. 29, vides an overview of devices that have been tested and pp. 29.1–29.13.

[9] I. Darian-Smith and K. O. Johnson, “Thermal sensibility and [31] D. R. Kenshalo and H. A. Scott, “Temporal course of thermal thermal receptors,” Journal of Investigative Dermatology, vol. 69, pp. adaptation,” Science, vol. 151, no. 3714, pp. 1095–1096, 1966. 146–153, 1977. [Online]. Available: http://www.jstor.org/stable/1718420 [10] D. C. Spray, “Cutaneous temperature receptors,” Annual Review of [32] L. E. Marks, “Spatial summation in relation to the dynamics of Physiology, vol. 48, pp. 625–638, 1986. warmth sensation,” International Journal of Biometeorology, vol. 15, [11] X. Zhang, “Molecular sensors and modulators of thermorecep- no. 2, pp. 106–110, 1971. tion,” Channels, vol. 9, no. 2, pp. 73–81, 2015. [33] J. C. Stevens and L. E. Marks, “Spatial summation and the dynam- [12] D. R. Kenshalo, C. E. Holmes, and P. B. Wood, “Warm and cool ics of warmth sensation,” Perception & Psychophysics, vol. 9, no. 5, thresholds as a function of rate of stimulus temperature change,” pp. 391–398, 1971. Perception & Psychophysics, vol. 3, no. 2, pp. 81–84, 1968. [34] J. C. Stevens, L. E. Marks, and D. C. Simonson, “Regional sensi- [13] L. Gray, J. C. Stevens, and L. E. Marks, “Thermal stimulus thresh- tivity and spatial summation in the warmth sense,” Physiology & olds: sources of variability,” Physiology & Behavior, vol. 29, no. 2, Behavior, vol. 13, no. 6, pp. 825–836, 1974. pp. 355–360, 1982. [35] J. C. Stevens and L. E. Marks, “Spatial summation of cold,” [14] G. A. Jamal, S. Hansen, A. I. Weir, and J. P. Ballantyne, “An Physiology & Behavior, vol. 22, no. 3, pp. 541–547, 1979. improved automated method for the measurement of thermal [36] A. J. Rosza´ and D. R. Kenshalo, “Bilateral spatial summation of thresholds. 1. Normal subjects,” Journal of Neurology, Neurosurgery, cooling of symmetrical sites,” Perception & Psychophysics, vol. 21, and Psychiatry, vol. 48, pp. 354–360, 1985. no. 5, pp. 455–462, 1977. [15] A. Pertovaara and I. Kojo, “Influence of the rate of temperature [37] L. Jones, “Thermal touch,” Scholarpedia, vol. 4, no. 5, p. 7955, 2009. change on thermal thresholds in man,” Experimental Neurology, [38] S. Song, G. Noh, J. Yoo, I. Oakley, J. Cho, and A. Bianchi, “Hot vol. 87, no. 3, pp. 439–445, 1985. & tight: Exploring thermo and squeeze cues recognition on wrist [16] D. Claus, M. Hilz, I. Hummer, and B. Neundorfer, “Methods of wearables,” in Proceedings of the 2015 ACM International Symposium measurement of thermal thresholds,” Acta Neurologica Scandinav- on Wearable Computers, ISWC ’15. New York, NY, USA: ACM, ica, vol. 76, no. 4, pp. 288–296, 1987. 2015, pp. 39–42. [17] M. J. Hilz, S. Glorius, and A. Beric,´ “Thermal perception thresh- [39] R. H. Taus, J. C. Stevens, and L. E. Marks, “Spatial localization of olds: Influence of determination paradigm and reference temper- warmth,” Perception & Psychophysics, vol. 17, no. 2, pp. 194–196, ature,” Journal of the Neurological Sciences, vol. 129, no. 2, pp. 135– 1975. 140, 1995. [40] W. S. Cain, “Spatial discrimination of cutaneous warmth,” Ameri- can Journal of Psychology, vol. 86, no. 1, pp. 169–181, 1973. [18] M. J. Hilz, S. E. Glorius, G. Schweibold, I. Neuner, B. Stemper, and F. B. Axelrod, “Quantitative thermal perception testing in [41] J. Tewell, J. Bird, and G. R. Buchanan, “The heat is on: A CHI preschool children,” Muscle & Nerve, vol. 19, no. 3, pp. 381–383, temperature display for conveying affective feedback,” in ’17 Proceedings of the 2017 CHI Conference on Human Factors in 1996. Computing Systems, 2017, pp. 1756–1767. [19] M. J. Hilz, B. Stemper, G. Schweibold, I. Neuner, F. Grahmann, [42] S. J. Lederman and L. A. Jones, “Tactile and haptic illusions,” IEEE and E. H. Kolodny, “Quantitative thermal perception testing in 225 Transactions on Haptics, vol. 4, no. 4, pp. 273–294, 2011. children and juveniles,” Journal of Clinical Neurophysiology, vol. 15, [43] A. Manasrah, N. Crane, R. Guldiken, and K. B. Reed, “Perceived no. 6, pp. 529–534, 1998. cooling using asymmetrically-applied hot and cold stimuli,” IEEE [20] M. J. Hilz, B. Stemper, F. B. Axelrod, E. H. Kolodny, and B. Ne- Transactions on Haptics, vol. 10, no. 1, pp. 75–83, 2017. undorfer,¨ “Quantitative thermal perception testing in adults,” [44] ——, “Asymmetrically-applied hot and cold stimuli gives per- Journal of Clinical Neurophysiology, vol. 16, no. 5, p. 462, 1999. ception of constant heat,” in 2017 IEEE World Haptics Conference [21] C. Madera, M. Hojatmadani, N. Crane, and K. Reed, “Thermal (WHC), 2017, pp. 484–489. Proceedings of the perception of skin using optical projections,” in [45] M. Hojatmadani and K. Reed, “Asymmetric cooling and heat- ASME 2017 International Mechanical Engineering Congress and Expo- ing perception,” in Haptics: Science, Technology, and Applications, sition IMECE2017, Volume 8: Heat Transfer and Thermal Engineering , D. Prattichizzo, H. Shinoda, H. Z. Tan, E. Ruffaldi, and A. Frisoli, 2017, p. V008T10A042. Eds. Cham: Springer International Publishing, 2018, pp. 221–233. [22] J.-Y. Lee, M. Saat, C. Chou, N. Hashiguchi, T. Wijayanto, H. Wak- [46] T. Oron-Gilad, Y. Salzer, and A. Ronen, “Thermoelectric tactile abayashi, and Y. Tochihara, “Cutaneous warm and cool sensation display based on the thermal grill illusion,” in Haptics: Perception, thresholds and the inter-threshold zone in Malaysian and Japanese Devices and Scenarios. EuroHaptics 2008. Lecture Notes in Computer males,” Journal of Thermal Biology, vol. 35, no. 2, pp. 70 – 76, 2010. Science, M. Ferre, Ed., vol. 5024. Berlin, Heidelberg: Springer, [23] V. Heldestad Lillieskold¨ and E. Nordh, “Method-of-limits; cold 2008, pp. 343–348. and warm perception thresholds at proximal and distal body [47] A. A. Young, “Thermal sensations during simultaneous warming regions,” Clinical Neurophysiology Practice, vol. 3, pp. 134–140, 2018. and cooling at the forearm: A human psychophysical study,” [24] D. Filingeri, H. Zhang, and E. A. Arens, “Thermosensory mi- Journal of Thermal Biology, vol. 12, no. 4, pp. 243 – 247, 1987. cromapping of warm and cold sensitivity across glabrous and [48] R. Watanabe, R. Okazaki, and H. Kajimoto, “Mutual referral of hairy skin of male and female hands and feet,” Journal of Applied thermal sensation between two thermal-tactile stimuli,” in IEEE Physiology, vol. 125, no. 3, pp. 723–736, 2018. Haptics Symposium 2014, 2014, pp. 299–302. [25] D. R. Kenshalo, T. Decker, and A. Hamilton, “Spatial summation [49] K. Arai, S. Hashiguchi, F. Shibata, and A. Kimura, “Analysis of on the forehead, forearm, and back produced by radiant and con- paradoxical phenomenon caused by presenting thermal stimula- ducted heat,” Journal of Comparative and Physiological Psychology, tion on three spots,” in HCI International 2017, C. Stephanidis, Ed. vol. 63, no. 3, pp. 510–515, 1967. Cham: Springer International Publishing, 2017, pp. 281–286. [26] G. Wilson, M. Halvey, S. A. Brewster, and S. A. Hughes, “Some [50] A. Singhal and L. A. Jones, “Space-time interactions and the per- like it hot: thermal feedback for mobile devices,” in Proceedings of ceived location of cold stimuli,” in 2016 IEEE Haptics Symposium the SIGCHI Conference on Human Factors in Computing Systems, CHI (HAPTICS), 2016, pp. 92–97. ’11. ACM, New York, USA, 2011, pp. 2555–2564. [51] A. Singhal and L. Jones, “Space-time dependencies and thermal [27] M. Halvey, G. Wilson, Y. Vazquez-Alvarez, S. A. Brewster, and perception,” in Haptics: Perception, Devices, Control, and Applica- S. A. Hughes, “The effect of clothing on thermal feedback perceptions, F. Bello, H. Kajimoto, and Y. Visell, Eds. Springer Inter- tion,” in Proceedings of the 13th International Conference on Multi- national Publishing, 2016, pp. 291–302. modal Interface ICMI’11, 2011, pp. 217–220. [52] A. D. Craig and M. C. Bushnell, “The thermal grill illusion: [28] F. Bolton, S. Jalaliniya, and T. Pederson, “A wrist-worn thermo- unmasking the burn of cold pain,” Science, vol. 265, no. 5169, pp. haptic device for graceful interruption,” Interaction Design and 252–255, 1994. Architecture(s) Journal (I×D&A), vol. 26, pp. 39–54, 2015. [53] B. G. Green, “Localization of thermal sensation: An illusion and [29] K. Janjeng and T. Leelanupab, “User experiences and perceptions synthetic heat,” Perception & Psychophysics, vol. 22, no. 4, pp. 331– of thermal feedback in the tropics,” in The Twentieth International 337, 1977. Symposium on Artificial Life and Robotics, 2015, pp. 237–242. [54] F. A. Geldard and C. E. Sherrick, “The cutaneous “rabbit”: a [30] M. Ketna and T. Leelanupab, “Evaluating a thermal icon for the perceptual illusion,” Science, vol. 178, no. 4057, pp. 178–179, 1972. enhancement of mobile feedback perception in noisy and bumpy [55] H. Fujita and K. Nishimoto, “Lovelet: A heartwarming communi- environments,” in 2017 IEEE World Haptics Conference (WHC), cation tool for intimate people by constantly conveying situation 2017, pp. 418–423. data,” in CHI 2004, 2004, p. 1553.

[56] W. Lee and Y.-K. Lim, “Explorative research on the heat as an expression medium: Focused on interpersonal communication,” Personal and Ubiquitous Computing, vol. 16, no. 8, pp. 1039–1049, 2012. [Online]. Available: http://dx.doi.org/10.1007/s00779-011- 0424-y [57] ——, “Thermo-message: Exploring the potential of heat as a modality of peripheral expression,” in CHI 2010, 2010. [58] S. Lee and T. Schiphorst, “Warmth and affection: Exploring thermal sensation in the design of parent-child distant interaction,” in Human-Computer Interaction. Novel User Experiences, M. Kurosu, Ed. Cham: Springer International Publishing, 2016, pp. 3–14. [59] K. Suhonen, S. Muller,¨ J. Rantala, K. Va¨an¨ anen-Vainio-Mattila,¨ R. Raisamo, and V. Lantz, “Haptically augmented remote speech communication: A study of user practices and experiences,” in Proceedings of the 7th Nordic Conference on Human-Computer Interac- tion: Making Sense Through Design, NordiCHI ’12. New York, NY, USA: ACM, 2012, pp. 361–369. [60] D. Gooch and L. Watts, “Communicating social presence through thermal hugs,” in SISSI 2010, 2010, pp. 237–242. [61] J. Klamet, D. J. C. Matthies, and M. Minge, “Wearelaxable: A wearable system to enhance stress resistance using various kinds of feedback stimuli,” in Proceedings of the 3rd International Workshop on Sensor-based Activity Recognition and Interaction, ser. iWOAR ’16. New York, NY, USA: ACM, 2016, pp. 2:1–2:6. [62] V. Kotsev, A. Nikolev, K. Pawlak, and M. Lochtefeld,¨ “Investigat- ing the usage of thermal feedback as an active game element,” in Proceedings of the 16th International Conference on Mobile and Ubiquitous Multimedia, ser. MUM ’17. New York, NY, USA: ACM, 2017, pp. 91–95. [63] J. Tewell, J. Bird, and G. R. Buchanan, “Heat-nav: Using temperature changes as navigation cues,” in Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems, ser. CHI ’17. New York, NY, USA: ACM, 2017, pp. 1131–1135.

Astrid M.L. Kappers studied experimental physics at Utrecht University, the Netherlands. She received the PhD degree from Eindhoven University of Technology. From 1989 till Septem- ber 2012, she was with the Department of Physics and Astronomy, Utrecht University. From 2008–2012, she was head of the Human Per- ception Group of the Helmholtz Institute. In September 2012, she moved with her whole group to the Department of Human Move- ment Sciences, Vrije Universiteit Amsterdam, the Netherlands. October 2018, she moved to Eindhoven University of Technology, where she works in three groups: Dynamics and Con- trol, and, Control System Technology of the Department of Mechanical Engineering, and Human Technology Interaction of the Department of Industrial Engineering & Innovation Sciences. She was promoted to full professor in 2005. Her research interests include haptic and visual perception. In 2003, she won the prestigious VICI grant. She is/was a member of the editorial boards of Acta Psychologica (2006-present) and Current Psychology Letters (2000-2011) and is an associate editor of the IEEE Transactions on Haptics (2007-2011 and 2017 till present).

Myrthe A. Plaisier received her Masters degree in experimental physics in 2006 and her PhD in 2010, both from Utrecht University (The Netherlands). She performed her PhD research on haptic perception in the group of prof. Astrid Kappers. Her thesis received the thesis award from the Dutch Psychonomics Society. In 2011 she received a Rubicon grant from the Nether- lands organisation for scientiﬁc research (NWO) which allowed her to continue her research in the lab of prof. Marc Ernst at Bielefeld University (Germany). In 2013 she received a VENI grant from NWO to investigate haptic perception of objects at the Department of Human Movement Sciences at Vrije Universiteit Amsterdam (The Netherlands). In Nov. 2018 she joined Eindhoven University of Technology in the Netherlands.