CERTAIN textures and materials that are used on a

Transcription

1 HSI 2011 Yokohama, Japan, May 19 21, 2011 Physical and sensoy factos of textues that appeal to human touch Hikau Nagano, Shogo Okamoto, and Yoji Yamada Nagoya Univesity, Nagoya, Japan Abstact. Some textues appeal to human touch, the easons fo which have not been investigated. We investigate the elationship between the physical and sensoy factos of 24 clay textues. Expeimental esults eveal that 70% 80% of the textues appeal to human touch can be explained on the basis of these factos. Keywods: affinity to textue, sensoy evaluation, textual dimension. I. INTRODUCTION CERTAIN textues and mateials that ae used on a daily basis appeal to human touch. Well-polished metal sufaces and finely woven clothes ae examples of such mateials. Discoveing the eason fo this affinity could enable us to design poducts that attact customes. In this study, we investigated the popeties of mateials that appeal to human touch. To the best of ou knowledge, human affinity to such mateials has not been investigated. We assumed that ou affinity is elated to the physical and sensoy chaacteistics of the textues. In pevious studies, tactile sensations had been explained by the sensoy factos of textues [1], [2], [3], [4]. Othe studies have investigated the elationship between the physical and sensoy chaacteistics of textues [5], [6]. Fom the viewpoint of poduct design, Winako et al. studied the effect of the textiles physical chaacteistics on thei sensoy evaluation [7]. In addition, the physical and sensoy chaacteistics of some specific mateials wee investigated. Fo example, Chen et al., Kim et al., and Matsuoka et al. and Picad et al. investigated the chaacteistics of wapping pape [8], sleeping wea [9] and ca sheets [10], [11], espectively. Kawabata et al. poposed a method to estimate sensoy factos fom the physical factos of clothing fabics [12]. Howeve, the elationship between the physical and sensoy factos of the textues and thei appeal to human touch has not been epoted thus fa. The objective of this study is to investigate to what extent and how the physical and sensoy factos of textues coelate thei appeal to human touch. When we see a textue, ou affinity to them is induced via its appeaances. Theefoe, the physical factos of the textues may diectly induce human affinity. On the othe hand, the affinity to textues may be detemined though human sensoy pocessing. We expeimentally veify the hypothesis that the affinity is attibutable to the physical and sensoy factos of the textues. This affinity may depend on pesonal pedispositions of paticipants, but we aveage individual esponses to investigate a potential tend in the elationships between the factos of the textues and thei appeal to human touch. II. DEFINITIONS OF APPEAL TO HUMAN TOUCH In this study, we assume that the textues visual appeaance contibute to thei appeal to human touch. Theefoe, in the expeiment, paticipants evaluate textues without touching them. In ode to pevent the vaiations in cultual backgound fom affecting the intensity of the affinity, the sample textues should not be associated with something specific. To this end, in this study, we use simple clay plates with textued sufaces. The paticipants ank all the sample textues in the ode of the intensity of thei affinity to the textues, and we quantitatively assess the intensity of the affinity fom the ank. Finally, touch does not imply pushing, o holding an object; it simply implies stoking the suface of textues. III. STIMULATION: PHYSICAL FACTORS OF TEXTURE Rathe than finely investigating the effect of single physical facto, we ae going to see the possible influence of some physical factos. We made 24 diffeent textues a) Suface pattens Gid Stipe b) Samples displayed to paticipants 50 mm 50 mm Sample textue White plastic plate = 0.5/1.0/2.0 mm Top view Side view Top view Side view Side view Plastic plate This wok was patly suppoted by MEXT KAKENHI Fo addess, see Fig. 1. Clay sample textue /11/$ IEEE 324

2 contolled by fou physical factos: suface colos, suface gloss pattens, suface shape-types, and suface idge and goove widths. We adopted light clay (Heaty Soft White, PADICO, Tokyo, Japan) as the textue mateial. The clay was mold into 55.0 mm 55.0 mm 5.0 mm flat plates using aluminum fame pais. The suface colo was eithe blue o oange; the clay was mixed with paint colos befoe molding. The colo vaiation of blue and oange is a complementay elationship. These colos should cause a lage deviation in sensoy evaluations. The blue (Phthalocyanine Blue, Liquitex, Ohio, USA) and oange paints (Scalet Red, Liquitex, Ohio, USA) wee (4.0BP, 1.5, 7), and (8.0R, 5.0, 13), espectively, in the Munsell colo pesentation. A mixing atio fo the clay and paints was 100 g clay to 1.25 ml paint. Fo the glossiness of the textues, glossy and glossless textues wee pepaed. To make the textues glossy, the plates wee vanished (SEALER Supe Gloss, PADICO, Tokyo, Japan) afte being coloed, mold, and died. The JIS specula gloss of the textues wee 2.4%, 94.2%, 1.7%, and 85.1% fo the glossless and glossy blue and glossless and glossy oange, espectively. As shown in Fig. 1, two shape types wee used: gid and stiped. The goove and idge widths wee 0.5, 1.0, o 2.0 mm. In total, 24 types of textues wee pepaed (2 colos 2gloss pattens 2 shape types 3 idge and goove widths). The physical factos of the textues wee quantified by physical facto scoes. The physical facto scoes wee nomalized values with a mean of 0 and vaiance of 1 fo each of the fou physical factos: suface colo (1: blue, -1: oange), gloss (1: glossless, -1: gloss), shape type (1: stipe, -1: gid), idge and goove width (-1.07: 0.5 mm, -0.27: 1.0 mm, 1.34: 2.0 mm). IV. EXPERIMENT In Task 1, the paticipants evaluated textues one afte the othe using the semantic diffeential (SD) method. Using the scoes given by the paticipants, we quantified the sensoy popeties of each textue. We then applied facto analysis to these popeties to obtain the independent sensoy factos. Afte Task 1, the paticipants anked all 24 textues in ode of the intensity of thei affinity to the textues (Task 2). Based on the anks of the textues, we calculated the degees of affinity to each textue. Finally, we investigated the elationships between the degees of affinity and the physical and sensoy factos using multiple egession analysis. A. Task 1: sensoy evaluation of textues In ode to quantify the sensoy popeties of the textues, we conducted sensoy evaluations by using the SD method. The paticipants evaluated the textues using five-point scales in tems of a pai of adjective tems, such as ough-smooth. The adjective tems used in the expeiments wee chosen in efeence to studies on visual and haptic peception. In the expeiments, we showed both English and Japanese tems on evaluation sheets. We conducted peliminay expeiments to emove and mege the candidate tems in tems of thei appopiateness fo the textues in this study. Fo example, we emoved the tems whose scoes did not deviate, such as thick-thin. In addition, we meged adjective tems with simila meanings such as shiny-matte and glossy-dull. In the end, 16 adjective pais wee left, as shown in Table I. The paticipants wee 12 voluntay students in thei 20s fom the authos laboatoy. As shown in Fig. 1b, a lage white plate with a 50 mm 50 mm squae window was placed on the sample textue so that the paticipants would see only the textued sufaces and not the sides of the samples. The paticipants wee instucted to attempt to keep thei head positions fixed in ode to etain the elative position between the head and textue samples. The ode in which the textues and adjective tems used to sot them wee pesented to each paticipant andom. We conducted expeiments in a laboatoy envionment whee the tempeatue was appoximately 24 C, and the textues wee placed at appoximately 460 lx. B. Facto analysis: data analysis of Task 1 We assigned values of 1 5 on the five-point adjective scales obtained in Task 1. In ode to educe the effects of individual diffeences in sensoy evaluation, the evaluation values wee nomalized to within a single paticipant. To decease the numbe of vaiables used fo the late analysis, we applied a facto analysis on the evaluation values and extacted common factos as a synthesis of vaiables that wee stongly coelated. The evaluation values x i fo p adjective tem pais wee boken down into common factos f i and unique factos e i : x i = A p 1 f i p m m 1 + e i p 1 (i =1 n), (1) whee the facto loadings A explain the stength of the elationships between common factos and adjective tems. In this model, the coelation matix R of x can be epesented as R = A p p p m m p AT + D p p whee D =diag(d 2 1,,d 2 p) ae unique factos. Matix R, whee diagonal elements of the coelation matix R ae eplaced by the estimated communality h 2 j,is 2 h p 2 R 21 h p =..... (= R D). (3). 2 p1 p2... h p R is appoximated by (2) R = ˆA ˆA T (4) TABLE I ADJECTIVE TERMS hash - not hash uneven - flat glossy - glossless elegant - inelegant vague - clea dak - light comfotable - uncomfotable soft - had dy - wet vivid - cololess ough - smooth wam - cold slippey - sticky simple - complex shap - blunt pedictable - unpedictable 325

3 whee ˆA is the appoximation of A. ˆA is given by R s m lagest eigenvalues ˆλ 1,, m and thei coesponding eigenvectos ˆc 1,, m. ˆA is ˆA = [ ˆλ 1ˆc 1,, ˆλ mˆc m ]. (5) We extacted the potential factos that explain textues by applying facto analysis to 15 adjective pais excluding pedictable-unpedictable. We applied vaimax otation to the facto loadings ˆA to facilitate intepetation of the elationships between factos and adjective tems. C. Task 2: textue anking Afte Task 1, all sample textues wee simultaneously pesented to each paticipant. The paticipants anked the textues in tems of the intensity of thei affinity to the textues. He was allowed to give the same ank to a few diffeent textues if he felt it difficult to ank all of the textues without any duplicate anks. By being pesented with all 24 diffeent textues at once, paticipants could evaluate the elative diffeences in textues. D. Nomalized-ank appoach: data analysis of Task 2 Because the anks of the textues ae odinal scales, we conveted the anks to inteval scales using nomalizedank appoach [13]. We defined these inteval scales as the degees of affinity. The textues did not include any that might include intense affinity. Hence, we assumed that the population of degees followed a nomal distibution. The degee of the kth anked textue was assigned to the expected value of the kth lagest obsevation in samples of size n fom a standad nomal population. The degee of the kth anked textue was detemined by n! E(x k n )= (n k)!(k 1)! x a(x) b(x) φ(x)dx (6) a(x) = [ 1 2 Φ(x)] k 1 (7) b(x) = [ 1 2 +Φ(x)] n k (8) φ(x) = 1 2π exp ( 1 2 x2) (9) Φ(x) = x φ(z)dz. (10) 0 The degees of affinity computed fo each paticipant wee aveaged. V. RESULTS A. Results of facto analysis The facto loadings and each facto s eigenvalues and cumulative contibuting ates calculated though the facto analysis ae shown in Table II. Eigenvalues mean the size of the factos effects. Cumulative contibuting ate is a pecentage that epesents the degee to which the textues ae explained by the obtained factos. We adopted the fivefacto model (m =5) because the cumulative ate was almost satuated when m was 5. Cells showing 0.7 o lage absolute facto loadings wee painted gay. Because the adjective tems with lage facto loadings epesent the popety of the facto, we named facto 1 as the glossless, ough, and dy facto, facto 2 as the comfotable and elegant facto, facto 3 as the cold and dak facto, facto 4 as the complex facto, and facto 5 as the uneven facto. Facto 1 was assumed to be affected by the fine oughness of the suface, which is associated with dyness, glossiness, and slippeiness. In contast, facto 5 is affected by the maco/coase oughness of the suface, which is elated with the goove and idge widths of the suface. Accoding to Table II, the cumulative contibuting ate of the five factos is appoximately 0.87; theefoe, the dimension space of the textue sensations was well established. B. Degees of affinity to textues The degees of affinity that esulted fom Task 2 ae shown in Fig. 2. The textues ae aanged in descending ode of degee. In tems of the suface gloss, the degee of glossless textues was lage than that of glossy textues. Textues with the top ten degees ae glossless textues. Suface gloss was assumed to geatly affect the degees compaed with the othe physical factos. In tems of the effects of the shape type, degees of stiped textues ae lage than those of gid textues. As to the suface colo, the degees of blue textues ae slightly lage than those of oange textues; fo the idge and goove width, degees slightly incease in popotion to the width. VI. RELATIONSHIPS BETWEEN DEGREES OF AFFINITY AND FACTORS OF TEXTURES To investigate the elationships between the degees of affinity and factos of textues, we pefomed multiple egession analysis. We pefomed analysis in two ways: objective vaiables ae degees, and explanatoy vaiables ae physical facto scoes of textues; and objective vaiables ae degees, and explanatoy vaiables ae sensoy facto scoes of textues. Multiple egession analysis was applied to the standadized values. In addition, we obtained the coelation coefficients between the physical and sensoy facto scoes and attempted to connect both factos using these coefficients. A. Relationships between degees of affinity and physical factos We conducted the multiple egession analysis with the objective and explanatoy vaiables being the degees of TABLE II RESULT OF FACTOR ANALYSIS: FACTOR LOADINGS OF ADJECTIVE TERMS Facto 1 Facto 2 Facto 3 Facto 4 Facto 5 Glossy Dy Rough Slippey Hash Comfotable Elegant Shap Dak Cold Complex Uneven Soft Vivid Vague Eigenvalue Cumulative Contibut. ates

4 (Blue, Glossless, Stipe, 2.0 mm) (Blue, Glossless, Stipe, 1.0 mm) (Blue, Glossless, Stipe, 0.5 mm) (Blue, Glossless, Gid, 1.0 mm) (Oange, Glossless, Gid, 1.0 mm) (Blue, Glossless, Gid, 2.0 mm) (Oange, Glossless, Stipe, 2.0 mm) (Oange, Glossless, Stipe, 1.0 mm) (Blue, Glossless, Gid, 0.5 mm) (Oange, Glossless, Gid, 2.0 mm) (Oange, Gloss, Stipe, 1.0 mm) (Oange, Glossless, Gid, 0.5 mm) (Blue, Gloss, Stipe, 0.5 mm) (Oange, Glossless, Stipe, 0.5 mm) (Blue, Gloss, Stipe, 1.0 mm) (Oange, Gloss, Stipe, 0.5 mm) (Blue, Gloss, Stipe, 2.0 mm) (Oange, Gloss, Stipe, 2.0 mm) Degee of affinity (Oange, Gloss, Gid, 2.0 mm) (Blue, Gloss, Gid, 2.0 mm) (Oange, Gloss, Gid, 1.0 mm) (Blue, Gloss, Gid, 1.0 mm) (Oange, Gloss, Gid, 0.5 mm) (Blue, Gloss, Gid, 0.5 mm) Fig. 2. Degees of affinity to textues. Eo bas ae standad deviations among the paticipants. affinity and physical facto scoes of textues, espectively. The standad patial egession coefficients ae shown in Table III. We found a significant negative coelation between the suface gloss and degees. In othe wods, being glossy deceases the degees of affinity. Next, we detemined the coelation between the shape type (stipe/gid) of textues and the degees: the degees fo stiped textues wee geate than the degees fo gid textues. In contast, thee was not much of a coelation between the degees and suface colo (blue/oange) o idge and goove width (0.5/1.0/2.0 mm). Theefoe, we estimated that the degees of glossless and stiped textues ae high values. This estimate is consistent with the esults shown in Fig. 2. In this case, R 2 was 0.72: theefoe, physical factos explain the degees of affinity as 72% of the vaiance in the data. B. Relationships between degees of affinity and sensoy factos We descibe the elationships between the degees of affinity and sensoy facto scoes of textues that esulted fom facto analysis. We conducted the multiple egession analysis with the objective and explanatoy vaiables being the fome and latte types of vaiables, espectively. Sensoy facto scoes ae degees of coelations between each facto and textue, as shown in Table IV. We show standad patial egession coefficients in Table V that esulted fom multiple egession analysis. We found stong coelations between factos 1 (glossless, ough, dy) and 2 (comfotable, elegant) and the degees of affinity. Slight coelations between factos 4 (complex) and 5 (uneven) and the degees wee obseved. Facto 3 (cold, dak) did not show any effects on the degees. Thus, textues that ae dy, glossless, comfotable, and elegant showed high degees of affinity. The degees of affinity wee not detemined only by the evaluation of comfot. In this egession analysis, R 2 was 0.78; theefoe sensoy factos explain the degees of affinity as 78% of the total. Inteestingly, facto 3 is the thid facto that contibutes to the ecognition of textues, while it does not contibute to the degees. In contast, while factos 4 and 5 ae mino factos fo textue ecognition, they moe stongly influence the degees of affinity. TABLE IV SENSORY FACTOR SCORES Facto 1 Facto 2 Facto 3 Facto 4 Facto 5 Blue, Gloss, Stipe, 0.5mm Blue, Glossless, Stipe, 0.5mm Blue, Gloss, Stipe, 1.0mm Blue, Glossless, Stipe, 1.0mm Blue, Gloss, Stipe, 2.0mm Blue, Glossless, Stipe, 2.0mm Blue, Gloss, Gid, 0.5mm Blue, Glossless, Gid, 0.5mm Blue, Gloss, Stipe, 1.0mm Blue, Glossless, Stipe, 1.0mm Blue, Gloss, Stipe, 2.0mm Blue, Glossless, Stipe, 2.0mm Oange, Gloss, Stipe, 0.5mm Oange, Glossless, Stipe, 0.5mm Oange, Gloss, Stipe, 1.0mm Oange, Glossless, Stipe, 1.0mm Oange, Gloss, Stipe, 2.0mm Oange, Glossless, Stipe, 2.0mm Oange, Gloss, Gid, 0.5mm Oange, Glossless, Gid, 0.5mm Oange, Gloss, Gid, 1.0mm Oange, Glossless, Gid, 1.0mm Oange, Gloss, Gid, 2.0mm Oange, Glossless, Gid, 2.0mm TABLE III REGRESSION COEFFICIENTS OF PHYSICAL FACTOR SCORES AND DEGREES OF AFFINITY Degees of affinity to textues Colo Gloss Shape Type Ridge and (+:Blue/ (+:Glossless/ (+:Stipe/ goove wid. -:Oange) -:Glossy) -:Gid) TABLE V REGRESSION COEFFICIENTS OF SENSORY FACTOR SCORES AND DEGREES OF AFFINITY Degees of affinity to textues Facto 1 Facto 2 Facto 3 (Rough, Facto 4 Facto 5 (Elegant, (Cold, Glossless, (Complex) (Uneven) Comfot.) Dak) Dy)

5 TABLE VI CORRELATION COEFFICIENTS BETWEEN PHYSICAL AND SENSORY FACTOR SCORES Facto 1 Facto 2 Facto 3 Facto 4 Facto 5 Colo Gloss Shape type R & G width Colo (+:Blue/-:Oange) Physical facto Gloss (+:Glossless/-:Glossy) Degee of Affinity Shape Type (+:Stipe/-:Gid) Ridge and Goove Width C. Coelations between physical and sensoy factos We show the coelations between physical and sensoy factos in Table VI. Facto 1 (glossless, ough, dy) is elated to the suface gloss and shape type. Glossy and stiped textues had highe values in tems of facto 1. Facto 2 (comfotable, elegant) is influenced by the shape type and idge and goove widths; this indicates that stiped textues with fine suface pattens ae peceived to be moe comfotable and elegant. The coelations between facto 3 (cold, dak) and the suface colo was stong, while the othe physical factos had little effect on this facto. Facto 4 (complex) was stongly affected by the idge and goove widths in that smalle widths cause highe complexity. Futhe, gid textues tended to be moe complex than the stiped textues. Finally, facto 5 (uneven) was dominantly impacted by the idge and goove widths of the textue sufaces. Fig. 3 shows the above mentioned elationships between the degees of affinity and physical and sensoy factos of textues. The line width epesents the stength of the elationships, which coespond to the absolute values of the standad patial egession coefficients and coelation coefficients. As shown by this figue, although it is limited to the sample textues used in this study, the elationships between the degees of affinity and physical and sensoy factos of textues wee quantitatively claified. VII. DISCUSSIONS A. Physical and sensoy factos of textues effectively explained ou affinity to the textues We found that the physical and sensoy factos of the textues explained the degees of affinity by 72% and 78%, espectively, which indicates that these factos effectively explain ou affinity to the textues. Although the acquied elationships between the intensity of the affinity and these factos wee limited to the sample textues used in this study, the esults suppot the agument that affinity to the textues is geneally detemined by thei physical and sensoy factos. Because the linea combination of the physical factos accounted fo 72% of the deviations in affinity, the affinity may be intuitively diven by the visual appeaances of the textues. On the othe hand, the sensoy factos explain the affinity as well as the physical factos, which suggest that ou intenal sensoy evaluation pocess mediates the affinity. The emaining 20% 30% of the deviations that these factos could not descibe may be due to the physical and adjective tems that this study did not conside. Individual diffeences in decision citeia at the two expeimental tasks may also have led to the esiduals of the egession analyses. Facto 1 Glossless, Rough, Dy Colo (+:Blue/-:Oange) Sensoy facto Facto 2 Comfotable, Elegant Facto 3 Cold, Dak Gloss (+:Glossless/-:Glossy) Facto 4 Complex Shape Type (+:Stipe/-:Gid) Facto 5 Uneven Ridge and Goove Width Fig. 3. Relationships between sensoy factos, and physical factos of textues, and degees of affinity to textues B. Textual dimensions: compaison with elated studies In this study, five factos/axes wee identified as the textual dimensions. Seveal eseaches also attempted to extact these haptic dimensions in mateial o textue ecognition. Hee, we compae ou esults with those of the elated studies. Table VII shows a efeence table fo the studies on haptic-textual dimensions. Yoshida [1], [2] investigated the dimensions fo haptic peception of 50 mateials by multidimensional scaling (MDS). Inteestingly, he studied the diffeences of such dimensions between the cases of haptic-only, visual-only, and visuo-haptic conditions. Fo the visuo-haptic condition, he found that thee dimensions, heaviness-coldness, wetsmoothness, and hadness, well established the dissimilaity space of the mateials. Hollins et al. [3] concluded that the textual dimensions of mateials ae composed of smooth-ough, had-soft, and sticky-slippey axes using MDS. Shiado et al. [5] used a factoy analysis method and identified fou dimensions fo 20 mateials: oughness, coldness, moistue, and hadness. Yoshioka et al. [6] found thee dimensions using MDS fo haptic exploation by bae finges: oughness, hadness, and stickiness. Although it was limited to ca sheet mateials, Picad et al. [11] found fou textual dimensions: soft-hash, thin-thick, elief, and hadness. Tanaka et al. [14] found two dimensions using facto analysis: moist and downy-wam. Because the dimensions of oughness/hashness, softness/hadness, and wamth/coldness wee epoted most often, these thee factos should be majo factos in textual ecognition. Ou esults included the oughness and cold factos, while the softness/hadness facto was not 328

6 TABLE VII REFERENCE TABLE OF TEXTURAL FACTORS Autho Yea Method Textue Facto 1 Facto 2 Facto 3 Facto 4 Facto 5 M. Yoshida [1], [2] 1968 MDS 50 mateials Heavy, Cold Smooth, Wet Had M. Hollins et al. [3] 2000 MDS 17 mateials Smooth/Rough Had/Soft Sticky/Slippey H. Shiado et al. [5] 2007 Facto analysis 20 mateials Rough Cold Moist Had T. Yoshioka et al. [6] 2007 MDS 17 mateials Rough Had Sticky D. Picad 24 ca seat 2003 MDS et al. [11] mateials Soft/Hash Thin/Thick Relief (Uneven) Had Y. Tanaka et al. [14] 2006 Facto analysis 13 fabics Moist Downy,Wam H. Nagano Glossless 2011 Facto analysis 24 textues et al. Rough,Dy Comfotable,Elegant Cold,Dak Complex Uneven clealy obseved. This may be because in ou study, the sample textues wee made of the same mateials and just thei suface geometic pattens vaied. In addition, the paticipants did not touch the textues but visually judged thei popeties. These specific expeimental conditions may have hidden the facto of softness/hadness. Anothe majo diffeence between this study and elated studies was the pesence of comfotable-elegant (facto 2) and complex (facto 4). None of the elated studies epoted such textual dimensions. These two factos may have been found because we adopted the adjective tems of elegant-inelegant, comfotableuncomfotable, and complex-simple, which ae usually used fo explaining visual impessions of textues. Finally, in this study, two types of oughness wee extacted as diffeent dimensions: hashness and unevenness. Because hashness was coupled to dy and glossless in facto 1, hashness may indicate the fine suface oughness of the textues. On the othe hand, unevenness (facto 5) was coelated with the idge and goove widths of the sufaces. Hence, in this study, unevenness may have been ecognized as an adjective tem fo coase oughness. Fine and coase oughness wee peceived by diffeent dimensions. The same tendency was suggested by Picad et al. [11] as hash and elief factos. VIII. CONCLUSIONS In this study, we investigated the popeties of specified textues that appeal to human touch. To detemine the cause fo such affinity, we investigated the elationship between the physical and sensoy factos of the textues and thei appeal to human touch. The textues wee fabicated using clay, paints, textued molds, and seales which enabled us to contol thei physical and sensoy factos. The intensity of the affinity was measued by a anking system and nomalized-ank method. In ode to quantify the sensoy factos of the textues, we conducted sensoy evaluation using 15 adjective tem pais, and the esults wee analyzed via facto analysis. Consequently, five othogonal sensoy factos wee identified: glossless-oughdy, comfotable-elegant, cold-dak, complex, and uneven. Multiple egession analyses, evealed that the physical and sensoy factos effectively explained the degees of affinity by 72% and 78%, espectively. In tems of the physical factos of the textues, thei glossiness and suface shape pattens stongly affected the degees of affinity. In tems of the influence of the sensoy factos, glossless-ough-dy and comfotable-elegant wee obseved to stongly elate with the degees. The methodology of this study is useful fo detemining the best combinations among limited physical factos in tems of the textues appeal to human touch. In poduct design, limited physical factos ae available; hence, the poposed method can potentially enable us to design poducts o textues fo shops o amusement centes, which appeal to human touch. REFERENCES [1] M. Yoshida, Dimensions of Tactual Impessions (1), Japanese Psychological Reseach, vol. 10, pp , [2] M. Yoshida, Dimensions of Tactual Impessions (2), Japanese Psychological Reseach, vol. 10, pp , [3] M. Hollins, S. Bensmaia, K. Kalof and F. Young, Individual Diffeences in Peceptual Space fo Tactile Textues: Evidence fom Multidimensional Scaling, Peception and Psychophysics, vol. 62, pp , [4] W.M.B. Tiest and A.M.L. Kappes, Analysis of Haptic Peception of Mateials by Multidimensional Scaling and Physical Measuements of Roughness and Compessibility, Acta Psychologica, vol. 121, no. 1, pp. 1 20, [5] H. Shiado and T. Maeno, Modeling of Human Textue Peception fo Tactile Displays and Sensos, Poc. Wold Haptics Confeence, pp , [6] T. Yoshioka, S. J. Bensmaia, J. C. Caig and S. S. Hsiao, Textue Peception though Diect and Indiect Touch: An Analysis of Peceptual Space fo Tactile Textues in Two Modes of Exploation, Somatosensoy and Moto Reseach, vol. 24, pp , [7] G. Winako, C. J. Kim, and L. Wolins, Fabic Hand: Tactile Sensoy Assessment, Textile Reseach Jounal, vol. 50, pp , [8] X. Chen, F. Shao, C. Banes, T. Childs, and B. Henson, Exploing Relationships between Touch Peception and Suface Physical Popeties, Intenational Jounal of Design, vol. 3, pp , [9] C. J. Kim, and K. Piomthamsii, Sensoy and Physical Hand Popeties of Inheently Flame-Retadant Sleepwea Fabics, Textile Reseach Institute, vol. 54, no. 1, pp , [10] T. Matsuoka, H. Kanai, H. Tsuji, T. Shinya and T. Nishimatsu, Pedicting Textue Image of Coveing Fabic fo Ca Seat by Physical Popeties, Jounal of Textile Engineeing, vol. 54, no. 3, pp , [11] D. Picad, C. Dacemont, D. Valentin and A. Giboeau, Peceptual Dimensions of Tactile Textues, Acta psychologica, vol. 114, pp , [12] S. Kawabata, and M. Niwa, Fabic Pefomance in Clothing and Clothing Manufactue, Jounal of the Textile Institute, vol. 80, no. 1, pp , [13] H. L. Hate, Expected Values of Nomal Ode Statistics, Biometika, vol. 48, no. 1 and 2, pp , [14] Y. Tanaka, M. Tanaka and S. Chonan, Development of a Senso System fo Measuing Tactile Sensation, Poc. IEEE Sensos, pp ,