UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Complex color stimuli and emotional responses Rasmussen, Per Gorm 1979

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-UBC_1979_A1 R38.pdf [ 22.37MB ]
Metadata
JSON: 831-1.0076993.json
JSON-LD: 831-1.0076993-ld.json
RDF/XML (Pretty): 831-1.0076993-rdf.xml
RDF/JSON: 831-1.0076993-rdf.json
Turtle: 831-1.0076993-turtle.txt
N-Triples: 831-1.0076993-rdf-ntriples.txt
Original Record: 831-1.0076993-source.json
Full Text
831-1.0076993-fulltext.txt
Citation
831-1.0076993.ris

Full Text

COMPLEX COLOR STIMULI AND EMOTIONAL RESPONSES by PER GORM RASMUSSEN B.A., Simon Fraser University* 1972 M.A. (Ed . ) , Simon Fraser Un ivers i ty , 1973 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES In terd i sc ip l inary (Psychology - Education - Architecture) We accept th i s thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA October 1979 (_) Per Gorm Rasmussen, 1979 In present ing t h i s t h e s i s i n p a r t i a l f u l f i l m e n t of the requirements f o r an advanced degree at the U n i v e r s i t y of B r i t i s h Columbia, I agree tha t the L i b r a r y s h a l l make i t f r e e l y a v a i l a b l e f o r re fe rence and s tudy. I f u r t h e r agree tha t permiss ion f o r ex tens ive copying of t h i s t h e s i s f o r s c h o l a r l y purposes may be granted by the Head of my Department or by h i s r e p r e s en t a t i v e s . I t i s understood tha t copying or p u b l i c a t i o n of t h i s t h e s i s f o r f i n a n c i a l ga in s h a l l not be a l lowed wi thout my w r i t t e n pe rm i s s i on . Department of Psychology, Education, Architecture The U n i v e r s i t y of B r i t i s h Columbia 2075 Wesbrook P lace Vancouver, Canada V6T 1W5 Date October 15, 1979 11. ABSTRACT The primary purpose of the present study was to examine the larger issue of color and emotional responses and, within that framework, to explore ways of specifying complex color displays. Several steps were involved in this investigation. First, a total of 80 color displays representing five levels of hue, two levels of value, two levels of chroma and four levels of motif were constructed. These were unique in that they accurately and systematically sampled the Munsell color space, and in the fact that they contained large numbers of color elements which were colorimetrically specifiable and which at the same time were arranged in such a way that they resembled color pictures. They thus bridged the gap between stimuli used in single color experiments which could be colorimetrically specified, and experiments with unspeci-fiable color pictures. Secondly, an emotional response measure employing the three dimensions of pleasure, arousal and dominance, was used to assess the effects of the display dimensions of hue, value, chroma and motif and the subject variable of sex. In addition, a verbal measure of information rate was used to assess the extent to which the display motifs influenced subjects' non-affective (i.e., cognitive responses), and subjects' ability to recognize the display motifs was assessed as wel1. Thirdly, the problem of stimulus specification was approached through the application of a three-step procedure involving increasing stimulus specificity. These approaches dealt with the specification in terms of (1) the individual color elements making up a display, (2) the quantity of these individual color components, and (3) the distribution or location of these elements across the display surface. The latter specification s'cheme, which was termed "distribution specification", made use of 24 procedures—some based on accepted artistic views and others of a more abstract nature—for calculating the relationship between the color elements in the displays. The mea-sures which these procedures resulted in were subseqently assessed against subjects' responses on the dimensions of pleasure, arousal, dominance and information rate. Initially, a pilot study with 20 subjects and 16 of the 80 displays was conducted to test the general performance of the response measures and to test whether the displays could beppresented in the form of projected slides. The results of this study showed that the general experimental procedure was acceptable but that the projection iv. technique distorted the colors of the displays excessively. Based on the conclusions of the p.ilot study, a larger study using 82 subjects and the displays as originally constructed was conducted. The results were surprising to the extent that complex color stimuli did not differevery substantially from those elicited by single color stimuli: the color dimension of value influenced the emotional responses to the greatest extent, chroma to a somewhat lesser extent, and hue very little. The motif of the displays, on the other hand, was found to make a substantial difference to the way subjects felt abojit a display, and the way they assessed it in terms of information rate. Also, it was found that the verbal measure of information rate was a good predictor of how well sub-jects would recognize a motif. The results of the analysis of stimulus specification in terms of the 24 distribution measures was^particularly interesting and gratifying in that several of the measures emerged as strong pre-dictors of responses to the emotional measures and information rate. In particular, the artistically common-sense notions of top-bottom and left-right pictorial balance were prominent, as was the specially constructed measure of contrasts within small sampling areas of the displays. It was concluded, first, that the study had reinforced the findings of many past studies dealing with color and affect, and V. that it had thrown some new light on some of the controversial and contradictory findings of the past. Secondly, the study had moved the investigation of emotional responses to color pictures and works of art a substantial step closer to realization. Finally, the study had suggested new and promising avenues to follow in the further investigation of colorimetric specification of complex color stimuli. v i . TABLE OF CONTENTS PAGE Abstract i i L i s t of tables x i i i L i s t of f igures x v i i i Acknowledgements xxv Chapter I INTRODUCTION 1 Chapter II COLOR PREFERENCE STUDIES, PICTURE SPECIFICATION SCHEMES, INFORMATION RATE AND EMOTIONAL RESPONSES 10 I COLOR PREFERENCE STUDIES 11 Studies using simple co lor s t imu l i 13 Summary 59 II THE PROBLEM OF COLOR SPECIFICATION OF COMPLEX COLOR STIMULI 62 III INFORMATION RATE AND THE MEASURE OF EMOTIONAL RESPONSES 79 The information rate concept 80 The emotional responses 89 Chapter III HYPOTHESES AND RESEARCH QUESTIONS 95 I HYPOTHESES 95 II RESEARCH QUESTION 97 Chapter IV I COLOR DISPLAYS FOR USE AS STIMULI 98 v i i . PAGE General descr ipt ion 98 The Munsell co lor notation system 100 Composition of the displays 105 Hue 105 Value and chroma 109 Motif 114 The C L E . system of co lor spec i f i ca t i on 119 Color imetr ic measurements of ind iv idua l d isp lay components 132 II SPECIFICATION OF THE COLOR SURFACE 135 Component spec i f i ca t i on 135 Quantity spec i f i ca t i on 136 D is t r ibut ion spec i f i ca t i on 141 The Judd-Hunter co lor di f ference formula 151 A measure of co lor di f ferences wi th in displays 153 A measure of perceived temperature of d isplay colors 155 Four surface d i v i s i on schemes 157 Figure/background d i v i s i on 159 Top/bottom balance 164 Le f t / r igh t balance 167 Average of small-area contrasts 173 V111. PAGE Chapter V A PILOT STUDY 186 The sample 188 The s t imul i 188 Response measures 193 P i l o t study administration and d i rect ions 195 Scoring 196 S t a t i s t i c a l analyses 196 Results 197 1. Conversion of displays to s l i des 197 2. Ef fect of recognit ion of motifs 198 3. E f fect o f viewing distance 199 4. Order of presentation e f fec t 201 5. E f fect of mothue, value, chroma and sex 201 Discussion, conclusions and modif icat ions to the experimental procedure 205 Chapter VI MAIN STUDY (EXPERIMENTS WITH AFFECTIVE AND COGNITIVE RESPONSES) 208 The sample 208 The experimental s i tua t ion 208 The s t imu l i 209 Response measures 215 Test administrat ion and d i rect ions 217 1x. PAGE Scoring 218 S t a t i s t i c a l analyses 219 Chapter VII RESULTS 222 I REPORTING OF THE RESULTS 221 Display descr ipt ion based on mean scores f o r pleasure, arousa l , dominance and Information rate 221 EMOTIONAL MEASURES AND INFORMATION RATE 225 Results ofi the analysis of variance with the d isplay var iables of hue, value, chroma, moti f and the subject var iab le sex as independent var iables and pleasure, arousal , dominance and Information rate as dependent var iables 225 S ign i f i can t main e f fects 230 Hue 230 Value 230 Chroma 233 Motif 234 Sex 236 Interact ion of main e f fec ts 237 Hue by motif Interact ions 239 X. PAGE Value by motif interact ions 243 Motif by sex interact ions 246 Hue by sex Interact ion 248 MOTIF RECOGNITION 252 Results of the analysis of variance with the d isp lay var iables of hue, value, chroma, motif and the subject var iable sex as independent var iables and motif recognit ion as the dependent var iab le 252 S ign i f i can t main e f fec ts 254 Hue 254 Value 255 Mot i f 255 Interact ion of main e f fec ts 256 Hue by motif in teract ion 256 Value by motif Interact ion 256 MOTIF RECOGNITION AND INFORMATION RATE 259 REGRESSION ANALYSES WITH DISPLAY COMPONENT RELATIONSHIPS 263 Equations fo r pleasure 264 Equations f o r arousal 270 Equations fo r dominance 271 x l . PAGE Equations for Information rate 273 II INTERPRETATION OF THE RESULTS 276 THE EMOTIONAL MEASURE AND INFORMATION RATE HYPOTHESIS 277 The ef fects of hue 278 The e f fec ts of value 280 The ef fects of chroma 281 The e f fec ts of motif 281 The e f fec ts of sex 283 THE HYPOTHESIS ABOUT REPRESENTATIONAL MOTIFS AND THE ABSTRACT 283 THE MOTIF RECOGNITION HYPOTHESIS 286 THE HYPOTHESIS ABOUT MOTIF RECOGNITION AND INFORMATION RATE 288 THE RESEARCH QUESTION 290 D is t r ibut ion spec i f i ca t i on and pleasure 291 D is t r ibut ion spec i f i ca t i on and arousal 300 D is t r ibut ion spec i f i ca t i on and dominance 309 D is t r ibut ion spec i f i ca t i on and information rate 309 Chapter VIII DISCUSSION 316 Introduction 316 x i i . PAGE BIBLIOGRAPHY Appendix A Appendix B Appendix C Appendix D The hypothesis dealing with the d isplay var iables of co lor , motif and the subject var iable sex versus the emotional response measures and Information rate 319 The hypothesis deal ing with the representational motifs of face, landscape and bui ld ings versus the abstract 329 The motif recognit ion hypothesis dealing with the d isp lay var iables of hue, value, chroma, motif and the subject var iab le of sex versus motif recognit ion 330 The hypothesis about motif recognit ion and information rate 334 The research question: 24 d i s t r i bu t i on spec i f i ca t i on var iables versus pleasure, arousal , dominance and Information rate Conclusion P i l o t study questionnaire P i l o t study, tes t booklet cover Main study tes t booklet cover Summary analysis of variance tables for motif recognit ion versus the four emotional response measures 357 334 344 346 351 353 355 xi11. LIST OF TABLES PAGE TABLE 1 Hue, value and chroma levels of colors used in hue test TABLE 2 Hue, value and chroma levels of colors used in value test. TABLE 3 Hue, value and chroma levels of colors used 1m chroma test TABLE 4 Preference order for hue test TABLE 5 Preference order for value test TABLE 6 Preference order for neutral value test TABLE 7 Preference order for chroma test TABLE 8 Colorimetric specifications for the second hue test TABLE 9 Three sets of colors used for the chroma test TABLE 10 "W" coefficients and preference order for first hue test TABLE 11 "W" coefficients and preference order for second hue test TABLE 12 Partial regression coefflcents on three diemnslons of color TABLE 13 25 colors used as backgrounds 16 18 19 22 22 22 22 27 30 31 31 47 54 x i v . TABLE 14 Pleasantness rat ings fo r 5 i l luminants TABLE 15 Interpretat ion and spec i f i ca t ion of in terest areas TABLE 16 Factor ia l composition of the information rate scales TABLE 17 Items and factors of the information rate scale expressed as functions of emotional states TABLE 18 Rotated factor matrix of the f i n a l set of emotional response scales TABLE 19 L i s t of the 70 colors used in the displays together with the i r Munsell notations and perceived temperature TABLE 20 Number and color of components of the 80 displays TABLE 21 Color imetr ic spec i f i ca t ions of 70 colors used in the displays TABLE 22 Quantity spec i f i ca t i on of hue elements in d isp lay no. 1 TABLE 23 Quantity spec i f i ca t ion of value elements in high and low value displays TABLE 24 Quantity spec i f i ca t i on of chroma elements 1n high and low chroma displays TABLE 25 Color imetr ic spec i f i ca t i on for majors and minors 1n displays and the i r average 4 E , value, chroma and temperature measurements PAGE 57 68 85 86 92 113 117 133 137 138 139 154 XV. TABLE 26 Figure-background proportions of AE and value f o r 80 displays TABLE 27 Figure-background proportions of chroma and temperature for 80 displays TABLE 28 Top-bottom proportions of AE and value fo r 80 displays TABLE 29 Top-bottom proportions of chroma and temperature fo r 80 displays TABLE 30 Le f t - r i gh t proportions of AE and value fo r 80 displays TABLE 31 Le f t - r i gh t proportions of chroma and temperature f o r 80 displays TABLE 32 Adjacent-difference measures fo r At, value, chroma and temperature TABLE 33 Adjacent-variance measures fo r AE, value, chroma and temperature TABLE 34 Color lmetHc data f o r selected colors of o r i g ina l and projected displays used in the p i l o t study TABLE 35 Overview of d isplay numbers used in p i l o t study TABLE 36 S ign i f i can t e f fects for motif recognit ion analysis Table 37 S ign i f i can t e f fects f o r analysis with pleasure, arousa l , dominance and information rate x v i . PAGE TABLE 38 Color imetr ic data for d isplay colors with and without "Colorkey" 211 TABLE 39 Means and standard deviations fo r 80 displays 222 TABLE 40 Means, standard deviat ions and corre lat ions f o r pleasure, arousal , dominance, information rate and motif recognit ion 224 TABLE 41 Summary of analysis of variance fo r pleasure 226 TABLE 42 Summary of analys is of variance fo r arousal 227 TABLE 43 Summary of analysis of variance fo r dominance 228 TABLE 44 Summary of analysis of variance for information rate 229 TABLE 45 Size of F-stat1st ic due to main e f fec ts 231 TABLE 46 Rank ordering of hue and motif combinations according to pleasure 240 TABLE 47 Rank ordering of hue and motif combinations according to dominance 242 TABLE 48 Rank ordering of value and motif combinations according to pleasure 245 TABLE 49 Rank ordering of value and motif combinations according to information rate 247 TABLE 50 Rank ordering of motif and sex combinations according to pleasure 249 TABLE 51 Rank ordering of hue and sex combinations according to dominance 251 x v i i . PAGE TABLE 52 Summary of analysis of variance with motif recognit ion as dependent var iable TABLE 53 Rank ordering of hue and motif combinations 1n terms of decreasing motif recognit ion TABLE 54 Rank ordering of value and motif combinations in terms of decreasing moti f recognit ion 2 TABLE 55 Normalized coe f f i c i en t s , R , standard errors and overa l l F-probabl l i t ies fo r the l i nea r regression equations TABLE 56 Normalized coe f f i c i en t s , R . standard errors and overa l l F-probabi l i tes fo r the quadratic regression equations TABLE 57 Pa r t i a l corre lat ions and per cent variance accounted fo r by l i nea r and quadratic equations f o r pleasure TABLE 58 Pa r t i a l corre lat ions and per cent variance accounted for by l i nea r and quadratic equations fo r arousal TABLE 59 Pa r t i a l corre lat ions and per cent variance accounted ... fpr by l i nea r and quadrat ic equations fo r dominance TABLE 60 Pa r t i a l corre lat ions and per cent variance accounted for by l i nea r and quadratic equations fo r information rate 253 257 258 265 266 269 272 274 275 XV111. LIST OF FIGURES PAGE FIGURE 1 Preference order for value 23 FIGURE 2 Preference order fo r chroma 25 FIGURE 3 Design of b i pa r t i t e f i e l d used by Granger fo r h is second hue test 28 FIGURE 4 Average a f fec t i ve values re lated to brightness of neutra ls , with t he i r cu rv i l i nea r regressions 38 FIGURE 5 Regression l ines re la t ing a f fec t ive value to hue 38 FIGURE 6 Isohedonic charts f o r R, YR and Y 39 FIGURE 7 Isohedonic charts for GY, G and BG 40 FIGURE 8 Isohedonic charts for B, PB and P 41 FIGURE 9 Isohedonic chart fo r RP 42 FIGURE 10 Spectral energy d i s t r i bu t i ons of f i ve l i g h t sources 55 FIGURE 11 The paint ing "Madame Le Brun and her daughter" used by McAdory fo r analysis 66 FIGURE 12 McAdory's graphic analysis of the paint ing "Madame Le Brun and her daughter" 69 FIGURE 13 Hue plan and graphic analysis of p icture with l i gh t pattern dominating dark background 72 FIGURE 14 Hue, value and chroma Intervals or contrasts 73 x1x. PAGE FIGURE 15 General arrangement of Harmon's scanning procedure 76 FIGURE 16 The "Mona L i sa " reconstructed from d i g i t i z ed data 77 FIGURE 17 Mehrabian and Russe l l ' s proposed framework for studying the environment in terms of pleasure, arousa l , dominance and information rate 81 FIGURE 18 I n i t i a l set of adject ive pairs for information rate measures 84 FIGURE 19 Mehrabian and Russe l l ' s measure of information rate 88 FIGURE 20 Display no. 1 (face) in i t s i n i t i a l stage of construction 99 FIGURE 21 The Munsell co lor s o l i d 102 FIGURE 22 The Munsell hue c i r c l e 102 FIGURE 23 The Munsell value scale ^ 102 FIGURE 24 Munsell chroma scale 104 FIGURE 25 Munsell PB-Y hue plane 104 FIGURE 26 Select ion of major, adjacent and complementary hues used 1n the displays 107 FIGURE 27 Graphic representation of proportion of major, adjacent and complementary co lor elements in the displays 110 FIGURE 28 Basic pattern of major, adjacent and complementary hues in d isplay number 1 112 XX. PAGE FIGURE 29 Value and chroma patterns in the major-complementary hue plane fo r displays no. 1, 2, 3 and 4 115 FIGURE 30 Displays nos. 1-20 with face as moti f 120 FIGURE 31 Displays nos. 21-40 with landscape motif 121 FIGURE 32 Displays nos. 41-60 with bui ld ings as motif 122 FIGURE 33 Displays nos. 61-80 with an abstract motif 123 FIGURE 34 C L E . 1931 equal energy d i s t r i bu t i on curves 124 FIGURE 35 Trist imulus values f o r 2° and 10° f i e l d s izes 124 FIGURE 36 The C L E . 1964 chromat idty diagram 126 FIGURE 37 Graphic i l l u s t r a t i o n of der ivat ion of t r i s t imu lus values for a colored surface 127 FIGURE 38 Typical spectral energy d i s t r i bu t i on curves fo r a var iety of i l luminants 129 FIGURE 39 P lot of color l o c i (x,y) in the C L E . chromat idty diagram 131 FIGURE 40 70 colors used in the displays p lotted i n the C L E . chromat idty diagram 134 FIGURE 41 Visual representation of quantity spec i f i ca t i on of value and chroma content of high and low displays 140 FIGURE 42 Color element numbers in d isplay number 1 ( face, BG-R, high value, high chroma) 142 FIGURE 43 Visual representation of major versus adjacent and complementary hues 144 xx1. PAGE FIGURE 44 High value motifs represented by three leve ls of value 145 FIGURE 45 Low value motifs represented by three leve ls of value 146 FIGURE 46 High chroma motifs represented by three leve ls o f chroma 147 FIGURE 47 Low chroma motifs represented by three leve ls of chroma 148 FIGURE 48 Majors and minors of f i r s t 20 displays 156 FIGURE 49 Conceptual scheme for assigning perceived temperature values to Munsell colors used in the displays 158 FIGURE 50 Figure-background boundaries for 3 motifs 160 FIGURE 51 Le f t - r i gh t p i c t o r i a l balance e f fec t 172 FIGURE 52 Part of 16 x 16 pasted-up display with 32 x 32 aperture matrix and 2 x 2 sampling areas 180 FIGURE 53 16 x 16 f i lm mask used 1n p i l o t study 189 FIGURE 54 S l ide of d isplay no. l 189 FIGURE 55 Color sh i f t s of selected colors and white due to s l i de conversion 192 FIGURE 56 Ef fects of order of presentation on pleasure, arousa l , dominance and information rate 202 FIGURE 57 Color s h i f t of three colors and white due to "Colorkey" overlay 213 XX11. PAGE FIGURE 58 Spectral ref lectance curves f o r white sample with and without "Colorkey" overlay material 214 FIGURE 59 Display no. 1 (face) with 32 x 32 matrix mask 216 FIGURE 60 S ign i f i can t interact ions of main ef fects on pleasure, arousal , dominance and Information rate 238 FIGURE 61 P lo t of expected res iduals versus the res idual 1n the motif recognition-Information rate regression equation 261 FIGURE 62 P lot of motif recognit ion means versus information rate means, and regression l i ne 262 FIGURE 63 P lo t of expected residuals versus residuals f o r l i nea r and quadratic regression equations 267 FIGURE 64 Approximate regression l i ne f o r the top/bottom-value component of the l i nea r equation f o r pleasure 293 FIGURE 65 Approximate regression Hne fo r the top/bottom-chroma component of the l i nea r equation fo r pleasure 294 FIGURE 66 Approximate regression Hne f o r the adjacent/variance-chroma component of the l i nea r equation fo r pleasure 296 FIGURE 67 Approximate regression curve for the quadratic form of left/r ight-chroma versus pleasure 298 FIGURE 68 Approximate regression curve fo r the quadratic form of adjacent/variance-temperature versus pleasure 299 xx1i1. PAGE FIGURE 69 Approximate regression l i ne fo r the adjacent/ dif ference-value component in the l i nea r equation fo r arousal 301 FIGURE 70 Approximate regression l i ne for the l e f t / r i gh t -va lue component 1n the l i nea r equation f o r arousal 303 FIGURE 71 Approximate regression l i ne f o r the adjacent/ difference-temperature component in the l i nea r equation f o r arousal 304 FIGURE 72 Approximate regression curve f o r the quadratic form of top/bottom-chroma versus arousal 306 FIGURE 73 Approximate regression curve for the f igure/ : Background-chroma component in the quadratic equation for arousal 307 FIGURE 74 Approximate regression curve for the adjacent/ variance-AE component of the quadratic equation f o r arousal 308 FIGURE 75 Approximate regression l i ne for the adjacent/ difference-chroma component of the l i nea r equation f o r Information rate 311 FIGURE 76 Approximate regression l i ne for the l e f t / r i g h t -chroma component of the l i nea r equation fo r information rate 312 xx iv . PAGE FIGURE 77 Approximate regression Hne for the average-value component of the l i nea r equation fo r Information rate 313 FIGURE 78 Approximate regression curve fo r top/bottom-value 1n the quadratic equation for information rate 314 FIGURE 79 Overview of var iables entering Intofcthe l i nea r and quadratic regression equations 336 XXV. ACKNOWLEDGEMENTS The wr i t e r would Wee to express h is sincere appreciation and grat itude to Dr. R. Lakowski f o r his unt i r ing encouragement and support during the lengthy thesis research; to Dr. J .B . Co l l i ns for his imaginative c r i t i que and suggestions, and to Dr. R. Corteen, Dr. E.G. F ied le r , Mr. W. Gerson, Mr. S. Black and Dr. J .A. Russell for the i r general guidance and help which they provided so generously. A lso, the wr i te r i s indebted to the Canada Council fo r f i nanc ia l support over a period of more than three years, and to Vinay Kanetkar who contributed so much to making the display surface computations poss ib le . 1. CHAPTER I INTRODUCTION The connection between co lor and human emotional responses has intr igued researchers for many years. Espec ia l ly since the tremendous accelerat ion in the var iety of pigments produced over the l as t hundred years or so--when natural colorants were re-placed by chemical pigments—and since the general increase in psychological cu r i os i t y covering about the same time frame, has the subject of co lor and responses been a focus of study. Before the ear ly psychologists of the l a s t century developed an in teres t in the top i c , a r t i s t s had struggled with the problem from a more prac t i ca l point of view. Their job had always been to portray, as accurately as poss ib le, s i tuat ions which among other things (such as form and s t ruc tura l realism) would have included emotional expressions e i ther inherent in the portrayed s i tuat ion or intended to resu l t from the port raya l . Color has a l l along in one way or another been suspected of 2. g iv ing r i se to par t i cu la r moods or emotions, and a r t i s t s probably never for a moment doubted t h i s , but the comprehensive e luc ida-t ion of the connection between the two--predsely how the one influences the other—has not yet been accomplished. As f a r as the pract i s ing a r t i s t 1s concerned, he s t i l l has to re ly to a large extent upon rules of thumb or upon h is own i n tu i t i on in cases where i t 1s desirable to e l i c i t spec i f i c emotions through spec i f i c works of a r t . This i s not to say that the voluminous work done in the f i e l d of co lor aesthetics i s of no use to the a r t i s t at a l l . In f a c t , many of the rules o f thumb he ava i l s himself of are Indeed the resu l ts of the f indings of these s tud ies . The general problem can probably be traced to two d i s t i n c t areas of discrepancy be-tween what the psychologist does and what the a r t i s t has to do. The most prominent of the problems concerns the nature of the psychologist 's s t imul i versus the a r t i s t ' s p ic ture . The picture i s usual ly a complex assemblage of numerous colors and shapes, whereas in most psychologica l s tud ies, the s t imul i have e i the r been s ing le colors or pairs of colors at most. (Norman and Scott , 1952; B a l l , 1965; Valent ine, 1962; P ick ford , 1972) I t stands to reason that only the most inexperienced a r t i s t can be w i l l i n g to abdicate h is i n tu i t i ona l judgment and construct h is complex work on f indings based on only a few component elements. The discrepancy 3. which th is problem deals with has nothing to do with the d i s -crepancy between the "atomist ic" and the "Gestal t" views as described by Granger (1955c). I t simply addresses I t s e l f to the fac t that the a r t i s t has not In the psychologist 's s t imu l i been able to recognize the type of a r t i s t i c products which he deals w i th . The second problem concerns the response measures which have been employed in studies o f aesthet ics . The vast majority of these have been of the "preference" type, u t i l i z i n g adjec-t ives such as " l i k e - d i s l i k e " or "pleasant-unpleasant". (Norman & Scot t , 1952) Although these types of measures deal d i r e c t l y with essent ia l properties of the aesthet ic , they do not by any means begin to cover the multitude of emotional responses which the a r t i s t struggles w i th . A l so , a good case can be made fo r the fac t that the term "aesthet ic" i t s e l f may have more cognit ive than a f fec t ive connotations. An aesthet ic judgment may be a "meta-judgment" in the sense that i t i s a judgment based on the summation of a l l the responses—cognitive as wel l as a f f e c t i ve— rather than a judgment which has a deta i l ed , prac t i ca l app l i ca-t ion to the t r i a l and error process of a r t production. I t 1s of course possible that the a r t i s t aims at a par t i cu la r overa l l aes-thet i c expression both before and during the production, but in terms of actua l ly l i nk ing moods and fee l ing with the physical manipulations which he undertakes as the work progresses, the 4. overa l l aesthet ic expression may not be the most useful measure. Two r e l a t i v e l y recent developments have raised the poss i -b i l i t y of spec i fy ing a c learer and more exact re lat ionsh ip between colors and emotions. F i r s t , developments in the f i e l d of color imetry, inc luding developments in colorimetry instrumentation, have made i t possible to speci fy a co lor with a high degree of accuracy. Furthermore, automatic colorimeters with attached mini-computers have speeded up the determination of co lor spec i f i ca t ions to such an extent that a large number of co lor samples can be measured f a i r l y qu i ck ly . This i s 1n contrast to the older methods which used v isua l matching techniques and elaborate charts, and Involved a considerable amount of hand ca lcu lat ions before the f i n a l spec i f i ca t ion was obtained. The s en s i t i v i t y and r e l i a b i l i t y of current instruments i s also noteworthy since these properties ensure that minute nuances (to the t h i r d decimal place in many Instances) can be taken into account, and that measurements done on one occasion can be repeated with great accuracy on another. Secondly, through the recent work of Mehrabian and Russel l (1974), i t appears that the multitude of emotional responses one can associate with an aesthet ic experience have been successfu l ly reduced to a set of three r e l a t i v e l y independent response dimensions: 5. pleasure, arousal and dominance. These dimensions are the resu l t of factor analyses of a very large number of possible responses, and 1n the i r f i n a l (questionnaire) form they consist of a small number of sub-dimensions which i nd i v i dua l l y make sense t o , and are of eminent prac t i ca l u t i l i t y to the a r t i s t . Furthermore, follow-up research to the development o f these dimensions has l inked responses to pleasure, arousal and domin-ance to behavioral responses such as approach-avoidance, thus expanding the usefulness of the measures considerably. (Russe l l , 1974) While much progress has been made in the two areas mentioned, there s t i l l remain a number of problems to be solved. As f a r as the stimulus i s concerned, colorimetry i s pa r t i cu l a r l y wel l su i ted to the spec i f i ca t i on of an ind iv idua l co lo r , but as soon as a stimulus consists of more than one co lo r , the system runs into d i f f i c u l t i e s . The co lor imetr ic spec i f i ca t i on system presently used (the system of the Commission Internationale d 'Ec la i rage, for short , the C L E . system) produces a three-dimensional spec i f i ca t ion of a co lo r , and these dimensions can be p lo t ted , with a su i tab le change of the dimensional values, in a two-dimensional space. When another color i s introduced Into the st imulus, another p lo t in th i s two-dimensional space i s the r e su l t . The problem 6. here essen t i a l l y consists in der iv ing some form of quant i tat ive measure which takes both of these co lor imetr ic spec i f i ca t ions into account. Since the C L E . system i s essen t i a l l y a mathemat-i c a l model, although derived from experiments with mixtures of colored l i g h t s , (radiant energies), and taking Into account certa in v isua l a t t r ibutes of the human observer, i t spec i f i es colors in a euc l id ian space, while (the human perceptual system seems to arrange colors in a non-eucl idian space. Thus r e l a t i on -ships between colors derived from the C L E . space do not necessar i ly agree with how combinations of these colors w i l l ac tua l ly appear to an observer. Since most p i c tures , as was mentioned, consist of a large number of Individual co lo rs , and since i t i s prec ise ly the interact ion or combination of a l l of these colors which properly ought to be character ized, i t i s easy 5 to see the l im i ted app l i c ab i l i t y of the C L E . co lor spec i f i ca* t ion system to complex s t i m u l i . The complex color s t imu l i of necessity have to be spec i f ied s ince , without that , one side of the stimulus-response equation would be incomplete and furthermore, i f the stimulus i s not spec i f i ed , i t 1s not possible to vary i t systemat ica l ly to the extent which a comprehensive experiment would require. There are also some problems associated with the response measures mentioned. Since the function of these measures i s to 7. assess emotional responses, i t i s important that they spec i -f i c a l l y and exc lus ive ly measure t h i s . However, i t 1s quite poss-i b l e that the measures at hand confound emotional responses with cognit ive ones, j u s t as the responses in the preference studies may have done. A l so , i n cases where the stimulus i s a p i c tu re , cognit ive factors such as recogni t ion, f am i l i a r i t y and meaning may e i ther Influence the emotional response or promote a cognit ive rather than an emotional response. In an attempt to overcome the d i f f i c u l t i e s out l ined above, the present research proceeds in the fo l lowing way: 1. The s t imul i used are complex insofar as they consist of a large number of ind iv idua l co lo rs . They are thus more c lose ly re lated to actual p i c t o r i a l mater ia l . Atfethe same time, t he i r composition i s planned in such a way that each component co lor i s c l ea r ly d i s t i n c t and spec i f i ab l e . The cost of th i s l a t t e r move has been a reduction in rea l i sm, i . e . , they are not " rea l " co lor p ic tures , although i t i s hoped that they resemble them. The term "d isp lays" i s used to describe the s t imu l i to indicate t he i r hybrid nature or t he i r locat ion i n the continuum between simple (s ing le or duplex) color s t imu l i and " r e a l " (complex) color p i c tu res . 8. 2. Since i t i s desirable to be able to vary stimulus dimensions i n a systematic way, and with equal percep-tual steps, colors fo r the displays are i n i t i a l l y chosen from the Munsell co lor space. This spec i f i ca t ion system has the advantage that i t i s a purely perceptual system in contrast to the C L E . system, and that i t s at t r ibutes or dimensions of hue, value and chroma have equal ly spaced steps with in each dimension. In addit ion to the systematic var ia t ion of the three co lor dimensions, a dimension of "motif" tits added in order to make the displays more c l ose ly resemble p i c t o r i a l mater ia l . The motifs selected fo r analysis are a face, a landscape, bu i ld ings , and an abstract des i gn. 3. Subsequent to the construction of the d isp lays , the component colors are measured and spec i f ied according to the C L E . system. 4. Both the Munsell and the C L E . spec i f i ca t ions of the component colors in the displays are used in an extensive analysis of the color combinations used in the d i sp lays . Various ways of describing and ca lcu lat ing th^eolfer. content in thei d i s p l a y sa r e employed^ and a mult ip le a. regression analysis performed to assess which of these descr ipt ive procedures accord best with the resu l ts of the dependent measures of emotional responses. 5. A verbal measure of information rate (also developed by Mehrabian and Russe l l , 1974), together with a specia l motif recognit ion measure, are f i n a l l y employed to assess the extent to which cognit ive factors may inf luence the emotional responses to pleasure, arousal and dominance. The primary focus of the present thes i s ; i s , In summary, on the construct ion of s t imu l i whldh are nei ther s igg le colors or pairs of colors which can be f u l l y spec i f i ed , nor complex color pictures which cannot be spec i f ied with any degree of accuracy. The measures of emotional responses which have been selected are 1n a sense yet another way of spec i fy ing the s t imul i and they are used, in addit ion to t he i r ro le as dependent measures of responses to the d isp lays , as a datum l ine against which other and more sophist icated spec i f i ca t i on schemes of the complex d i s -plays are assessed. 10. CHAPTER II COLOR PREFERENCE STUDIES, PICTURE SPECIFICATION SCHEMES, INFORMATION RATE AND EMOTIONAL RESPONSES The color s t imul i employed in the present research d i f f e r considerably from any which in the past have been used to i n v e s t i -gate the re lat ionsh ip between color and emotional responses, and as a r e su l t , the f indings of most past research deal ing with color and a f fec t are of questionable value here. At tthe same time* i t i s evident.from many of the s ing le color preference studies that there i s a general be l i e f among researchers in the "atomist ic" approach, i . e . , that s ta r t i ng with responses to s ing le coflor s t i m u l i , one can gradually make the s t imul i more complex un t i l eventual ly responses to very complex s t i m u l i are arr ived a t . In the sense that co lor imeter ic measurements are based on spec i f i ca t ions of ind iv idua l d isplay elements, the present study adheres to th i s view. However,, i t i s also rea l i zed 11. that the spec i f i ca t ion of ind iv idua l elements i s r e l a t i v e l y useless unless at the same time some ways of spec i fy ing the overa l l co lor imetr ic character i s t i cs of the displays are developed. This chapter deals with three d i s t i n c t l y d i f f e ren t top ics , a l l of which to a greater or lesser extent, form the basis fo r the present research strategy. 1. Major co lor preference studies using both one and two colors as well as color p ictures as s t i m u l i . 2. The problem of spec i fy ing complex co lor s t imul i and rudimentary attempts at th i s type of spec i f i c a t i on . 3. The emotional and cognit ive response measures developed by Mehrabian and Russell (1974) which the present experiments employ. I COLOR PREFERENCE STUDIES Studies which have invest igated the re lat ionsh ip between" co lor and emotions have in most cases consisted of tests in which 12. s ing le co lor swatches or pairs of co lor swatches have been pre-sented under r e l a t i v e l y contro l led condit ions. Response measures, however, which various invest igators have employed have ranged from scales of l i k e - d i s l i k e , pleasant-unpleasant to highly sophis-t i cated mult ip le semantic d i f f e r en t i a l scales which subsequently have been factor analyzed. The focus of many of these l a t t e r studies hasebeen on the explorat ion of psychological response dimensions rather than on d i rec t connections between spec i f i c at t r ibutes of co lor and, in some cases, the spec i f i ca t i on of colors used has not been as e x p l i c i t as might have been hoped fo r from the present point of view. However, most of the recent, major studies dealing with these typesl of st imul i—which here are termed "simple" in contrast to the "complex" ones used in the present thes is—are reported here regardless of which response measures were employed, the c r i t e r i on for se lect ion being that the s t imul i they used have been (or could have been) co lo r imet r i ca l l y spec i f i ed . There also ex i s t s a large number of studies which have employed actual p i c t o r i a l mater ia l . These have been extensively reviewed by Valentine (1962) and by Pickford (1972). Since, however, the color pictures used in these experiments have not been c o l o r i -metr i ca l l y spec i f i ed , i t i s d i f f i c u l t to assess t he i r impact on the present research. Many of the studies are of great in te res t i n the e luc idat ion of more general aesthet ic preferences to pictures 13. and to works of a r t , and some of the experiments—if the i r s t imul i could be co lo r imet r i ca l l y spec i f ied at a l a t e r date--would be of great value indeed. The wr i t e r sees no reason, fo r instance, why colon'me t r i e measurement techniques of complex color s t imu l i , perhaps based on some of the procedures developed in the present thes i s , could not resu l t in a re-evaluation of the resu l ts of some of the more recent experiments with co lor p i c tu res . Studies using simple co lor s t imu l i Owing to the lack of precise co lor spec i f i ca t ion schemes used in the past, as has been repeatedly pointed out in the recent l i t e r a tu re (Granger, 1955a, B a l l , 1955; Jacobs and Suess, 1975), as wel l as perhaps the perceived unimportance of these schemes to experiments which emphasized psychological response dimensions rather than e x p l i c i t stimulus cha rac te r i s t i c s , only a number of recent, major studies are reported here. Even many of these recent studies neglect to speci fy the s t imu l i used adequately with the resu l t that i t i s d i f f i c u l t or impossible to assess t h e i r f indings or to base hypotheses for further research on them. For instance, one f a i r l y recent major study (Eysenck, 1941) found consistent preference rankings of colors as fo l lows: "blue, red, green, v i o l e t , orange, ye l low, a l l f u l l y saturated; 14. green, red and orange tints;* and a yel low shade." From th i s verbal descr ipt ion i t i s impossible to ascertain exact ly what colors were used, espec ia l l y which t i n t s and shades and, in add i t ion , the viewing conditions and the l i gh t source werertnot spec i f i ed so that , as a r e su l t , even i f the actual stimulus material used were ava i l ab l e , I t could not be establ ished what the s t imu l i ac tua l l y looked l i k e at the time of t h e i r presenta-t i o n . The f i r s t major studies of co lor preferences using wel l spec i f i ed co lor s t imu l i are those of Granger (1955a, 1955b, 1955c, 1955d). Most o f the studies dealt with employ the Munsell co lor notation system. This system i s described in de ta i l in chapter IV. "An experimental study of colour preferences", (Granger, 1955a). Although Granger's object in th i s study i s pa r t l y to show that responses to co lor corre late with responses to other aesthet ic material (he uses the Maitland Graves tes t of Design Judgment [ c f . Graves, 1951]), i t i s possible to extract from th i s study the part which deals only with colors and preference statements. Granger states three general research questions perta in ing to th i s problem? 15. (a) Is there a general order of preference fo r each a t t r ibute of colour, at d i f f e rent leve ls of the colour so l id? (b) I f there i s , does the general order of preference fo r any one a t t r ibute of colour remain i n -var iant under change in level of the other two at t r ibutes? (c) Is the general order of preference dependent on inherent stimulus propert ies? (p. 5) As the plan was fd r s t to tes t the three dimensions of hue, value and chroma independently, three groups of s ingle co lor s t imul i were prepared: 1. For the hue t e s t , the 10 Munsell hues of 5R, 5YR, 5Y, 5GY, 5G, 5B, 5PB, 5P and 5RP were selected fo r t e s t i ng . Within these 10 hues as many leve ls of value and chroma as were ava i lab le were selected so that , as a resuiht, hue could be varies while the a t t r i -butes of value and chroma were kept constant. Table 1 sgows the 10 Munsell hues used, and the resul tant experimental sets of colors produced at constant value and chroma l eve l s . In addit ion to the 24 sets se lected, one addit ional set cons ist ing of the maximum chromas wi th in the 10 hues was included. There were a to ta l of 214 co lor chips in the hue t e s t . 2. In se lect ing colors fo r the value t e s t , the object was to keep hue and chroma constant, and a se lec t ion which was as comprehensive as possible with in the avai lab le pigments resulted in a maximum of 16. TABLE 1 Hue, value and chroma leve ls of colors used i n hue tes t . Set leve l 5R 5YR 5Y 5GY 5G 5BG 5B 5PB 5P 5RP 1 2/2 X X X X X X X X X X 2 2/4 X X X X X 3 3/2 X X X X X X X X X X 4 3/4 X X X X X X X X X 5 3/6 X X X X X X 6 3/8 X X X X 7 4/2 X X X X X X X X X X 8 4/4 X X X X X X X X X X 9 4/6 X X X X X X X X 10 4/8 X X X X X X 11 5/2 X X X X X X X X X X 12 5/4 X X X X X X X X X X 13 5/6 X X X X X X X X X X 14 5/8 X X X X X X X 15 6/2 X X X X X X X X X X 16 6/4 X X X X X X X X X X 17 6/6 X X X X X X X X X X 18 6/8 X X X X X X X 19 7/2 X X X X X X X X X X 20 7/4 X X X X X X X X X X 21 7/6 X X X X X X X X X 22 7/8 X X X X X 23 8/2 X X X X X X X X X X 24 8/4 X X X X X X X X 25 "maxima" X X X X X X X X X X 17. 7 value steps for each set (some had as few, as 5) covering the f u l l range of hues, and with chroma ranging from 2 to 6. An addit ional set of achromatic colors (neutrals) ranging in value from 1 to 9 was included. The to ta l number of colors included in the value tes t was 160, as shown in table 2. 3. Due to the l imi ted a v a i l a b i l i t y of many value l eve l s , at a l l chroma l eve l s , of the hues selected fo r the hue and value t e s t s , the 7.5 hue leve l was used here. 6 hues at rather i r r egu la r value leve ls (as table 3 shows) were selected for test ing at chroma leve ls ranging from 2 to 14. A to ta l of 59 colors were used in the chroma tes t . Granger's subjects numbered 25 men and 25 women ranging in age from 19 to 36, and representing varying occupations and in te res t s . Both male and female subjects were included since a study by Eysenck (1941) had raised the p o s s i b i l i t y that sex dif ferences may account f o r di f ferences in co lor preferences. Eysenck himself did not f i nd th i s to be the case, except in the case of yel low and orange, where men preferred orange to yel low and women yel low to orange. His corre la t ion for preferences between the two sexes was .95. One espec ia l ly in terest ing aspect of Granger's study was 18. TABLE 2 Hue, value and chroma leve ls of colors used in value t e s t . Set Hue/chroma leve l 1 2 3 Value notation 4 5 6 n 8 1 5PB/2 X X X X X X X 2 5PB/4 X X X X X X 3 5PB/6 X X X X X X 4 5B/2 X X X X X X X 5 5Y/2 X X X X X X X 6 5Y/4 X X X X X 7 5YR/2 X X X X X X X 8 5YR/4 X X X X X X 9 5RP/2 X X X X X X X 10 5RP/4 X X X X X X X 11 5RP/6 X X X X X X X 12 5P/2 X X X X X X X 13 5P/4 X X X X X X X 14 5P/6 X X X X X X 15 5BG/2 X X X X X X X 16 5BG/4 X X X X X X 17 5G/2 X X X X X X X 18 5G/4 X X X X X X 19 5GY/2 X X X X X X X 20 5GY/4 X X X X X X 21 5R/2 X X X X X X X 22 5R/4 X X X X X X X 23 5R/6 X X X X X X X X X 19. TABLE 3 Hue, value and chroma leve ls of colors used in chroma t e s t . Hue/value C h r o m a n o t a t 1 o n Set leve l 7 4 6 8 10 12 14 1 7.5PB/3 X X X X X X X 2 7.5PB/5 X X X X X X X 3 7.5Y/8 X X X X X 4 7.5YR/7 X X X X X 5 7.5RP/3 X X X X X 6 7.5RP/5 X X X X X 7 7.5RP/6 X X X X X 8 7.56Y/6 X X X X X 9 7.5GY/7 X X X X X 10 7.5R/3 X X X X X 11 7.5R/6 X X X X X 20. the pre-test screening of subjects fo r co lor v i s ion de f i c i enc i es . He wanted to make certa in that a l l h is subjects possessed normal co lor v i s i o n , and fo r screening purposes he used the Ishihara tes t ( Ish ihara, 1948), the Rabkln tes t (1939) and the Farnsworth-Munsell 100-hue tes t (Farnsworth, 1943). A l so , the testppresentation was care fu l l y cont ro l led . An i l luminant "A" l i gh t source with a Macbeth f i l t e r resu l t ing in a color temperature of 6500° K was used to i l luminate the samples, and the samples were viewed in a spec ia l booth l ined with neutral gray paper. Only the leve l of i l luminat ion on the tes t ing surface i s not mentioned. The object of the experiment was fo r subjects to arrange the co lor chips with in each of the 60 co lor sets in order of preference, and the spec i f i c ins t ruct ion to subjects was that , i i i , t h e y should judge the colours according to the i r preferences at the present moment. I t was emphasised that they should not judge the colours according to t he i r asso-c iat ions with objects or people, nor in terms of t he i r use f o r various prac t i ca l purposes; the colours should be con-sidered only in terms of t he i r appearance under the condi-t ions of the experiment, (p. 8) The resu l ts of the study showed that , f i r s t , there i s a general order of preference which can be expressed Independently fo r the three Munsell dimensions of hue, value and chroma. 21. For hue, th i s order 1s shown in table 4. Using Kendal l 's measure of concordance, W, the coe f f i c i en t f o r the group as a whole reached a s ign i f i cance level of .01. Granger further noted: In general, the hues of shorter wavelength are preferred to those of longer wavelength; the blues and greens are pre-ferred to the fe l l ows , oranges and reds. No precise r e l a t i on -ship can be establ ished, in view of the small number of co lours, and the imposs ib i l i t y of inc luding the extra-spectra l purples. However, i f [only] the e ight spectral hues are considered, i t can be shown that the i r rank order, according to increasing wavelength, agrees f a i r l y c lose ly with the general order of preference. . . . (p. 14) Secondly, Granger found that for value preferences, the rank ordering of judgments appeared to produce roughly an Inverted U-shaped function in which value 4 i s most preferred, and values e i ther l i gh te r or darker are less preferred. Table 5 shows the preference orders for value, and f igure 1 shows the preference order graphica l ly represented. A problem arose when the colors of value 5 were viewed against the neutral gray background. Granger speculates that the background Interfered with the preference judgments in such a way that when for Instance, colors of value 3 and 7 were compared, the equal perceptual steps between each of these colors and the background confused the subject to such an extent that no object ive d iscr iminat ion of the two tes t colors could be made. In table 5 i t can be seen that i f value 5 i s ignored, I 22. TABLE 4 Preference order fo r hue t e s t . Munsell notation 5B 5PB 5BG 5G 5P 5RP 5R 5GY 5YR 5Y Rank order 1 2 3 4 5 6 7 8 9 10 TABLE 5 Preference order f o r value t e s t . Munsell notation 2 3 4 5 6 7 8 Rank order 5 4 1 2 3 6 7 TABLE 6 Preference order fo r neutral val lue t e s t . Munsell notat ion 1 2 3 4 5 6 7 8 9 Rank order 8 6 3 1 4 2 5 7 9 TABLE 7 Preference order f o r chroma t e s t . Munsell notation 2 4 6 8 10 Rank order 5 3 2 1 4 23. 3 4 5 6 7 8 Value leve ls FIGURE 1 Preference order for value. (Dotted l i ne Indicates order with value 5 ignored.) 24. the preference order decreases on e i ther side of value 5 rather than value 4, and the dotted l ine in f igure 1 shows th is new! curve. Granger's resu l ts of the achromatic colors by themselves are shown in table 6. Th i rd ly , for the chroma test i t was found that preference increased with increasing saturat ion, up to chroma 8, whi le fo r the higher chroma of 10, preference decreased. Again, as in the case of the value t e s t , anclinverted U-function emerged. Table 7 sShows th is re la t ionsh ip , and the preference order i s shown graph-i c a l l y in f igure 2. F i n a l l y , Granger found no differences between co lor pref-erences of males and females. "Out o f 60 corre lat ions between the orders of preference for men and women in the hue, value, and chroma t e s t s , 48 reached s ign i f i cance at the 5 per cent l e v e l . . . . " (p. 13) "An experimental study of colour harmony", Granger, •(i-9S5b);-.-;4ii--W:s study, Granger increased the complexity of h is stimulus material from one to two co lors . Based on previous research by Eysenck (194fls), which did not speci fy the colors used very accurately, and by Clarkson, Davies and V ickers ta f f (1950), Granger proposed the general hypotheses that preference should increase with increasing s i ze of hue in terva l and with decreasing s i ze of value i n t e r va l . As 25. 4 6 8 10 Chroma leve ls FIGURE 2 Preference order for chroma. 26. fa r as chroma was concerned, no previous research was ava i lab le on which to base any hypotheses. Two tests of hue were conducted. The f i r s t employed a se lec-t ion of 20 Munsell hues, the complete set of a l l 5 and 10 ser ies co lors , both with value and chroma leve ls of 6. The second hue tes t used colors ident i ca l to those used by Clarkson, Davies and V ickers ta f f (1950). The complete set con-s i s ted of 24 colors selected espec ia l ly fo r t he i r high chroma content, and these are l i s t e d in table 8. I t i s in terest ing to note that , whereas the hues of the f i r s t tes t were selected with constant value and chroma (a lack of ava i lab le Munsell colors predicated t h i s , but Granger notes that based on h is previous resu l ts of the invariance of preferences at d i f f e rent leve ls of the co lor space, " . . . i t i s u n l i k e l y . . . that any ' laws' of colour harmony operating at Value 6, Chroma 6, would f a i l en t i r e l y to operate at other l e ve l s . " (p. 24) ) , the second hue tes t em-ployed colors with var iat ions in both hue, value and chroma l eve l s . A lso , i t i s noteworthy that a special b i pa r t i t e f i e l d was used fo r the second hue t e s t . This design (shown i n f igure 3) was supposed to present the two tes t colors in a way which made the l i ne of contact pa r t i cu l a r l y &ong. The value tes t employed s i x sets of hues: 5P, 5R, 5B, 5G, 5Y and 5RP—all at chroma leve l 2 except the 5RP which had a chroma 27. TABLE 8 Color imetr ic spec i f i ca t ions for the second hue t e s t . Chromatidty Hue coordinates Brightness Munsell notation no. X y W Hue Value Chroma f .530 .293 18.5 3.1R 15.3 4.9 2 .569 .316 18.6 5.8R 15.6 4.9 3 .576 .329 19.7 7.OR 15.3 5.0 4 .576 .346 23.9 8.2R 15.4 5.4 5 .562 .378 31.7 0.5YR 15.3 6.1 6 .555 .380 37.2 0.9YR 15.8 6.6 7 .491 .419 53.4 6.5YR 12.4 7.7 8 .441 .458 64.1 4.1Y 10.5 8.3 9 .393 .491 61.4 3.56Y 10.3 8.1 10 .344 .505 46.8 7.9GY 11.0 7.3 11 .305 .520 34.6 0.1G 11.7 6.4 12 .277 .511 26.0 1.3G 11.5 5.6 13 .207 .471 21.5 4.7G 13.2 5.2 14 .173 .357 18.1 2.9BG 11.4 4.8 15 .169 .334 15.7 4.3BG 10.6 4.5 16 .174 .265 12.6 1 .IB 8.2 4.1 17 .168 .212 11.3 7.5B 8.9 3.9 18 .171 .161 14.3 5.0PB 13.1 4.3 19 .180 .136 8.5 6.7PB 13.2 3.4 20 .202 .119 6.0 9.1PB 13:6 2.9 21 .269 .141 6.1 4.4P 13.0 2.9 22 .341 .204 17.5 0.7RP 9.6 3.2 23 .426 .259 10.1 7.6RP 8.6 3.7 24 .512 .289 14.2 2.2R 13.1 4.3 28. FIGURE 3 Design of b i pa r t i t e f i e l d used by Granger fo r his second hue t e s t . 29. level of 4—with a l l leve ls of value from 2 to 8. Table 9 shows the three sets of colors selected for the chroma t e s t . As in the previously mentioned study by Granger, the tes t presentation was care fu l l y contro l led: a booth l ined with neutral gray paper and an i l luminant "A" with a Macbeth f i l t e r to produce a co lor temperature of 6500° K were used. The f i r s t hue tes t was administered in such a way t ha t ! a l l the 20 hue samples were arranged in a c i r c u l a r pattern, and one hue at a time was s ingled out fo r comparison with the remaining 19 hues. There was a to ta l of f i ve hues s ingled out to act as "standards": 5R, 5Y, 5G, 5B and 5P. The resu l ts of th i s tes t showed that preference did Indeed tend to Increase with Increasing hue dif ferences between component colors and. . . . when a preference ranking i s de r i ved . . . [table 10] as a whole, on one side of the complementary the general order o f preference corre lates per fec t ly with the rank order of hue i n te rva l s , while on the other side there i s only one displacement in rank pos i t i on , (p. 31) Table 10 shows the hue in terva ls and preference rankings f o r the f i r s t hue tes t as wel l as the corre la t ion coe f f i c i en t fo r the f i ve colors used as standards. 30. TABLE 9 Three sets of colors used fo r the chroma t e s t . Value Chroma 2 4 6 8 10 12 Set 1 4 7.5R and 7.5PB Munsell ser ies Set 2 4 7.5R and 7.5PB Munsell ser ies Set 3 6 2.5YR Munsell ser ies TABLE 10 MWM coe f f i c i en t s and preference order fo r f i r s t hue t e s t . Hue Interval W 1 2 3 4 5 6 7 8 9 10 9 8 7 6 5 4 3 2 1 .647 16 17 15 13 11 10 8 7 6 .354 13 11 8 10 9 7 6 3 2 .630 13 12 10 9 7 5 2 1 3 .457 8 7 5 3 2 1 4 6 12 .700 12 15 14 13 18 19 17 16 10 5 4 2 1 3 9 12 14 18 19 1 4 5 12 17 18 16 15 14 19 4 6 8 11 14 15 18 19 17 16 16 18 19 17 15 13 14 11 10 9 2 1 5 3 4 7 6 8 9 11 TABLE 11 "W" coe f f i c i en t s and preference order fo r second hue t e s t . Standard W Hue Interval 5.8R 4.1Y 4.76 5.0PB .569 .537 .520 .524 14 16 20 18 17 12 11 8 4 1 2 3 6 12 11 10 8 7 6 5 3 2 1 4 9 13 18 21 19 22 20 23 17 16 12 9 7 3 1 19 20 22 23 21 11 12 14 15 10 5 3 1 5 15 2 2 7 9 10 13 19 22 23 21 15 18 17 19 21 22 23 20 16 14 4 5 6 8 10 11 14 13 15 4 6 7 8 9 13 17 16 18 32. The second hue test made use of the b i pa r t i t e f i e l d design shown in f igure 3. Four co lors: 5.8R, 4.1Y, 4.7G and 5PB were used here as "standards", and a l l the other colors were 1n turn assessed against these. However, in th i s t e s t , a black instead of a neutral gray background was used for the t es t i ng . Although not as pronounced, the resu l ts of th i s second hue test agree f a i r l y wel l with those of the f i r s t t e s t . Granger further notes that the general increase in pleasure with i n -creasing hue in terva l seems to ho ld , even though value and chroma are allowed to vary as w e l l . Since the background as well as the shape of the s t imul i was d i f f e ren t (the f i r s t hue tes t employed rectangular co lor swatches), he further concludes that'.these pref-erence judgments seem to be independent also of background and shape. Table 11 shows the hue i n t e rva l s , the preference rankings and the corre la t ion coe f f i c i en ts fo r the second hue t e s t . The value t e s t , which used samples of value 2 to 8 for s i x hue l eve l s , was presented in a manner s im i l a r to that of the f i r s t hue t e s t . The co lor chips of value 2 acted as the standard for a l l s i x hue sets and, for some unspecif ied reason, value 8 was in addit ion used as the standard fo r the 5P hue se t . Granger reports that the resu l ts of th i s tes t are compar-able to those he had found e a r l i e r (Granger, 1955a), i . e . , that a decrease in value contrast resu l ts i n an increase in pleasure. 33. However, he does not report h is resu l ts in any further d e t a i l . The chroma tes t was also conducted in much the same way as the f i r s t hue t e s t . Two chromas, 2 and 12, were used as stan-dards, and a l l chromas within the f i ve hue sets were assessed against these. Here, preference tended to decrease with increas-ing chroma i n te rva l s , although " . . . the re la t i on between chroma preference and colour interva l i s somewhat obscure." (p. 32) F i n a l l y , based on a comparison of Kendal l 's coe f f i c i en t of concordance for ind iv idua l rankings and the general order of preference fo r the d i f fe rent colors used as standards, Granger concludes that , Preference for in terva ls of hue, l ightness , and saturat ion are independent of the pa r t i cu l a r colours used as standards and preferences for hue in terva ls are also to some extent independent of s ize and shape of the component colours, the mode of combining them, and background l i g h t ne s s . . . . A score on one set (or t e s t ) . . . has considerable pred ict ive value with regard to other, s im i l a r sets or tests{ i t also has predic-t i ve value with regard to tests Involving preferences for s ingle co l ou r s . . . . " (p. 34) "Aesthetic measure applied to colour harmony: an experimental  tes t " (Granger, 1955c) and "The predict ion of preference for  colour combinations" (Granger, 1955d). Three papers by Moon and Spencer (1944a, 1944b, 1944c) develop the idea that , based on B i rkho f f ' s formula fo r "aesthet ic measure" (B i rkhof f , 1933), i t 34. i s possible to predict responses to color combinations from re lat ions between component co lo rs . Granger, in an experiment employing s i x sets of binary color combinations and one t r i a d , however, concluded that the Moon and Spencer formula d id not have any pred ic t ive value. (Granger, 1955c) Granger's view—which i s also shared by the present w r i t e r— i s that an empirical ("atomist ic") approach rather than one based on a speculat ive formula of aesthet ic measure i s the more sensible route to fol low in research with complex co lor s t imul i although, as Granger also notes " . . . the e f fec t of the components... [might] decrease systemat ica l ly as the complexity increased; the more complex the colour arrangement, the greater would be the importance of the re lat ions between the colours and the less would be the importance of the colours themselves." (Granger, 1955d) The f i n a l study by Granger to be reported here (Granger, 1955d) hypothesized tha t , . . . any formula for predict ing preferences for binary combinations of colours must take in to account two sets of factors at least: (a) the component preferences and (b) factors re la t ing to the combination of such." (p. 217) In th i s experiment, Granger used a se lect ion of Munsell colors of value 6 and chroma 6 only, but a l l the hues of the 5 and 10 ser ies 35. were included. Subjects ranked both s ing le co lor samples and a l l combinations in order of preference, using the standard exper i -mental arrangement as in the previous studies of Granger. The author reported the fol lowing three corre la t ions: (1) fo r ranking derived from the component preferences and the observed order of preference for the combinations, .631; (2) fo r the co lor in terva l rankings and the order of preference fo r the combina-t ions , .554; and (3) between the two derived rankings, .004. According to these f ind ings, Granger concluded that both the com-ponent preferences and the in terva l re lat ions wereriof importance for the predict ion of co lor preferences. He notes that " . . . a formula which Includes both factors [of component and in terva l preferences] would account for about 70 per cent of the to ta l var iance." (p. 219) "A system of color-preferences" (Gui l ford and Smith, 1959). The ch ief merit of th i s study derives from the fact that i t 1s ex t ra-o rd ina r i l y comprehensive, being the summary of several past studies (Gu i l fo rd , 1934; Gu i l f o rd , 1939; Gu i l f o rd , 1940; Gu i l f o rd , 1949) as wel l as the resu l ts of experiments carr ied out espec ia l l y for th i s study. The colors used were from the Munsell Book of Color (1929), and the se lect ion was probably the most complete one used in any 36. research up un t i l that t ime. The se lect ion included 10 equal ly spaced hues (except in the green and blue sectors where colors of intermediate hues were included because previous research had indicated that preference changes might not fol low those of the other sectors) , Munsell values of 2, 4, 6 and 8, as wel l as a l l ava i lab le chroma l eve l s . In add i t ion, 21 achromatic (neutral) colors were inc luded, In a l l , the to ta l number of co lor samples used amounted to 316. The color papers (2" squares) were viewed in an enclosed booth, through an opening, and the I l luminat ion in the booth consisted of four 60 watt " frosted b lu ish Mazda bulbs". The s t imu l i were each exposed for 5 seconds. The subject 's task was to assess the pleasantness of the color presented, and to score i t on a 10 point s ca l e , ranging from 0 = "most unpleasant imaginable" to 10 = "most pleasant imaginable". The ^instruction read in part: You are to judge the pleasantness of the color presented, being careful to judge i t as a co lo r . Do not think of i t in connection with any object inppar t i cu la r , but rather take i t j u s t as a sensat ion, (p. 489) In contrast to Granger's f ind ings, Gui l ford had found evidence of sex d i f ferences, so his sample consisted of 20 males and 20 females. A l l subjects were co lor v i s ion tested before the experiment, but 37. Gui l ford does not mention which tes t or tests he used fo r th i s purpose. Gu i l ford 's resu l ts are most c l ea r l y reported in the form of graphs, and reproductions of his own graphs are therefore shown in f igures 4, 5, 6, 7, 8 and 9. In the main, these graphs are s e l f -explanatory. Referring to the f igures , the fol lowing general points may be made: 1. Most of the achromatic colors (value 2 through 8) were found to be unpleasant. Correlat ions of .87 and .94 for males and females, respect ive ly , were found f o r the f i t between observed and expected values. Females tended to rate a l l samples as s l i g h t l y less pleasant than males ( f igure 4 ) . 2. Females found the green and blue hues most pleasant, and espec ia l ly so fo r the darker (lower) values. The least preferred hues were in the green-yellow sector ( f igure 5 ) . 3. Colors at value leve ls at which they can be most satur-ated tended to be preferred. 4. Pleasure rat ings " . . . are usual ly at neither the lowest 38. > . , i _ i 1 i u B 1 * \ « - —^» 3 •|uEN •WOME Brightness FIGURE 4 Average a f fec t ive values related to brightness of neutra ls , with t he i r cu rv i l i nea r regressions. Hue FIGURE 5 Regression l ines re la t ing a f fec t i ve value to hue. 39. 4 0 . Saturation Saturation FIGURE 7 Isohedonic charts fo r GY, G and BG. 41. Saturation Saturation FIGURE 8 Isohedonic charts for B, PB and P. 42. Saturation Saturation FIGURE 9 Isohedonlc chart for RP. 43. leve l of saturat ion nor at the lowest leve l of brightness and th i s ru le holds for both sexes and at most hues." (p. 495) 5. In general, " . . . the best predict ions can be made in the region from green to purple-blue and. . . the poorest predic-t ions occur for purple, green-yellow, and ye l low-red." (p. 500) Predict ions were better fo r men (corre lat ion among observed and predicted pleasure rat ings for hues was .93) than for women (corre lat ionsof .88). 6. Although there was general ly high agreement between males and females, there were spec i f i c instancesoof d i s -agreement. In pa r t i cu l a r , in the red-purple region, where the same corre lat ion as above was as low as .27, females preferred the highly saturated colors whereas males p r e -ferred the less saturated ones. In summarizing his experiments, Gu i l ford warns against using his graphs ("isohedonlc charts" , as he ca l l s them) in s i tuat ions which deviate markedly from those employed tin the experiment described. In general, he recommends that more information about the pa r t i cu l a r app l icat ion of co lor ( e . g . , i t s texture, g loss , shape, brightness level of viewing background and whether the c 44. color i s the color of a pa r t i cu la r object) be obtained before attempting to make predict ions about preferences based on h is data. "The meanings 6f co lor" (Wright and Rainwater, 1962). While the studies reported so f a r have used preference measures such as rank order of l i k i n g and pleasantness, th i s study was unique in that i t employed a set of 48 adject ives arranged in the manner of a semantic d i f f e r en t i a l quest ionnaire. The object was thus not so much the explorat ion of the connection between pa r t i cu l a r at t r ibutes of co lor and predetermined aes thet i ca l l y oriented responses (which the vast majority of previous research had occ ip ied i t s e l f w i t h ) , but rather the extract ion by means of pr inc ipa l component analyses, from the 48 questionnaire items, of s i gn i f i c an t factors which would begin to explain connotative responses to co lor . As a resu l t of th i s emphasis on response measures, the study did not deal e x p l i c i t l y with the spec i f i c a -t ion of colors used as s t i m u l i . The only information the authors supply about the 50 co lor samples used i s that they i'covered the gamut of hue, l i g h t -ness and saturat ion." (p. 90) The Munsell notation system was used, and hue had a mean of 31 and a standard deviat ion of 23; value a mean of 5 and a standard deviat ion of 2.1; while chroma had a mean of 9 and a standard deviat ion of 3.8. A rather unusual scoring of 45. the Munsell system was used in th i s study. The hue c i r c l e was treated as a continuous scale from 0 to 100 and i t appears from subsequent references to wavelength (used here as synonymous with hue) that the scale mentioned was placed with zero at the blue end of the spectrum and 100 at the red end. I t , however, i s not c lear which of the Munsell hues were chosen as the s ta r t ing point and which as the f in i sh ing point of the sca le . The subjects consisted of 955 males and 2705 females, with ages ranging from 16 to 65 years . Since i n i t i a l analysis o f the resu l ts indicated that no age or sex dif ferences ex i s ted , the ent i re sample of 3660 subjects was treated as one group. Each subject rated only one of the 50 co lor samples, arid only on a questionnaire made up of ha l f of the 48 semantic d i f f e r en t i a l sca les . One of the more in terest ing resu l ts from th i s study i s the fact that the f i r s t three factors extracted conform surpr i s ing ly wel l with the dependent measures of pleasure, arousal and domin-ance used in the present thes i s , and described in deta i l l a t e r in th is chapter. The three fac tors , with t he i r mult ip le co r re l a -t ion coe f f i c i en ts in brackets, were: 1. Happy, young, f resh, c l ea r , soc ia l and grace fu l . The authors termed th i s factor "happiness". (.67) 46. 2. Outstanding, showy and ex c i t i ng . This factor was ca l led "showiness". (.87) 3. Strong and f o r c e f u l . This factor was termed " forcefu lness". (.84) A f a i r l y strongfourth fac tor , "warmth" (with mult ip le co r re la -t ion of .62), i s fo r the present purpose l e f t out of t h i s d i s -cussion. The three main factors of in terest here together with t he i r loadings,are shown in table 12. From th i s table i t can be seen how chroma, in pa r t i cu l a r , accounts for the responses on a l l the three main fac tors . I t contributed most to "showiness", but also con-s iderably to the other (two. Value was the second most important in f luence. I t contributed pos i t i ve l y to "happiness" and "showiness" but negatively to " forcefu lness" . Hardly any e f fects of hue were evident from the resu l t s . In general, " . . . the l i gh te r or more saturated i s a co lor , the more 'happiness' i t connotes" (p. 95), and " . . . the darker . . . [and] more saturated i s a co lor , the more i t connotes " forcefu lness". (p. 95) The lack of inf luence of hue, the authors noted, was curious since i t i s that dimension which most often i s equated with co lor . They further suggested that t he i r resu l ts in 47. TABLE 12 Pa r t i a l regression coef f i c ients on three dimensions of co lo r . Factors Hue Value Chroma 1. Happiness .014 .194 .102 2. Showlness .034 .118 .262 3. Forcefulness .017 -.190 .142 48. respect to chroma were somewhat su rp r i s i ng , and they speculated that perhaps there i s some bas ic , but as yet undiscovered, r e l a -t ionship between chroma and connotation. "A pr inc ipa l component analysis of semantic d i f f e r en t i a l judgments  of s ing le colors and color pa i r s " (Hogg, 1969). In th i s study both s ing le colors and pairs of colors were used as tes t s t i m u l i . The Munsell c61ors of 5R, 5Y, 5G, 5B and 5P were used at value leve ls 2, 4 and 6 and at chroma leve ls 4, 6 and 8. These leve ls were combined into f i ve groups of the fol lowing composition: 1. 5R 6/6 5R 6/2 5R 4/4 5R 8/4 2. 5Y 6/6 5Y 6/2 5Y 4/4 5Y 8/4 3. 5G 6/6 5G 6/2 5G 4/4 5G 8/4 4. 5B 6/6 5B 6/2 5B 4/4 5B 8/4 5. 5P 6/6 5P 6/2 5P 4/4 5P 8/4 To further expand the range of values and chromas, the fo l lowing two groups were added: 6. 5R 6/8 5Y 6/8 5G 7/6 5B 4/8 5P 7/6 7. 5R 4/10 5Y 5/4 5G 3/4 5B 7/2 5P 4/10 49. From these groups were constructed two-color combinations with a l l possible combinations of hue, value and chroma. A to ta l of 8 com-binations were thus produced. The presentation was made against a gray background of value 3, and with an i l luminant "A" l i g h t source. The sample con-s i s ted of 20 males and 30 females, and the scoring was carr ied out by means of the fol lowing 12 semantic d i f f e r en t i a l sca les: The resu l ts of the s ing le color experiment showed that four main factors (accounting for 93.44% of the variance) emerged. These were, with perccent of to ta l variance accounted for in brackets, a "potency plus a c t i v i t y " fac tor (70.56%), a "usual-obvious" fac tor (12.24%), a "pleasantness" factor (5.87%) and a "hot-warm-exc l t i ng " factor (4.77%). In respect to the e f fec t of co lor on ind iv idua l fac to rs , i t was found that the more saturated a co lor was the more potent or act ive was i t perceived to be. There were 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. blatant-muted hot-co Id pleasant-unpleasant hard-soft act ive-passive lush-austere strong-weak excit ing-calming obvious-subtle sharp-dul l v i b r a n t - s t i l l usual-unusual 50. no apparent effects of either hue or value on this factor. Load-ings on the second factor showed that red and yellows of value levels less than 8 were unusual and "subtle, while at value levels equal to or greater than 8, they became obvious, Greens and blues were generally usual and obvious, whereas the purples tended to be neutral. The author does not mention the effect of chroma. The effects of color on the third factor, that of pleasantness, ... indicate the preference here is similar to that found in other experiments on color preference: namely, that there is a general preference for colors in the B-P region, low preference in the Y-G region, with R occupying an inter-mediate position. However, this is only a general order, and some reds are preferred to some blues. No systematic relation between brightness and the individual color weight-ings 1s seen in the present results, but there is a tendency for colors in the mid-saturation region to be most preferred: i.e., when C • 6. This holds for R, Y, G and B, while the most preferred saturation for P is C = 4. (p. 133) Finally, the effects of color on the Hh6t-lush-excit1ng" factor supported the conventional findings that red is aa"warm" and blue is a ecool" color. |p. 134) The resDlts of the experiment with color pairs showed that four factors, accounting for 94.36? 6IF the variance, emerged: an 51. " a c t i v i t y " factor (with 60.90% of the to ta l var iance), a "pleasant-ness" factor (20.48%), a "warm-lush" factor (9.10%) and a "usual-obvious" factor (3.88%). The fac tor " a c t i v i t y " was re lated to co lor i n two ways: (1) the greater the overa l l saturat ion of the two colors in the stimulus pa i r , the more act ive the pa i r i s perceived to be, and (2) the greater the value contrast , the more act ive the combina-t ion i s . Loadings on the factor of "pleasantness" showid the fo l low-ing in terest ing resu l t s : The c learest stimulus of the pa i r weightings on th i s fac tor i s hue. Pairs in which both colors are blue, or blue i s present, are highly pleasant; and a f ter these, those in which both are green, or green and blue combined, are judged pleasant. The presence of blue or green with red, ye l low, or purple tends to show a weighting ind i ca t ive of r e l a -t i ve pleasantness. Combinations of two reds, two'yel lows, or two purples tend to be r e l a t i ve l y unpleasant. Tentat ive ly , we would suggest that preference 1s not dependent here upon separation in the hue c i r c l e , as many pleasant pairs are made up of the same hue. (p. 134) The fac tor termed "warm-lush" was found to mainly be related^ to hue: Re lat ive ly greater warmth and lushness i s dependent upon there being a contrast in hue between the colors in the pa i r . Secondly, hue i t s e l f contributes to the judgment, red and yel low in the pairs tend to lead to high warmth/ 52. lushness judgements, though one curious anomaly emerged. Pairs in which both hues are red tend to be judged cool and austere. G, B, and P tend to contribute to coolness and auster i ty , (p. 134) F i na l l y , the fourth factor of "usual-obvious" was related to color as fo l lows: . . . pairs in which hue i s not contrasted, the Y pairs are unusual and to a lesser extent are G and B pa i r s , whi le P pairs are usual and red pairs intermediate in th i s respect. In considering the hue contrast pa i r s , certa in combinations of hues are general ly usual , such as red and yel low, red and purple, g*?een and purple, and to a lesser extent green and ye l low, while blue and green pairs are unusual. (p. 135) "The role of spectral energy of source and background color in  the pleasantness of object co lors" (Helson and Lansford, 1970). In several of the previous studies reported, the wr i ters noted the l i gh t source and the background against which the co lor samples were seen. This study i s an extension of these previous studies except that here l i gh t sources as wel l as backgrounds are var ied. A to ta l of 125 colors on 25 backgrounds and 5 sources o f i l luminat ion were tested. The colors were a l l of the 5 and 10 ser ies Munsell hues, the value/chroma combinations were a l l of the 2/2, 5/2, 8/2 se r i e s , about ha l f of the 2/6 se r i e s , samples 53. of value 5 in combination with chromas ranging from 6 to 10, value 8 samples ranging in chroma from 4 to 8, a maximum chroma ser ies covering a l l hues except 10G, and three neutrals o f values 2, 5 and 8. The Munisell colors making up the backgrounds for the experiment are shown in table 13. The f i ve l i gh t sources used 1n the experiment were: (1) C.I.E. i l luminant "A" (2854° K) , (2) standard f luorescent cool white (3000° K) , (3) Deluxe f luorescent cool white (4500° K) , (4) Deluxe f luorescent dayl ight (6500° K) , and !(»5) Macbeth f i l t e r e d incandescent (6800° K) . Their spectral energy d i s t r i bu t i on curves are shown in f igure 10. F i na l l y , the subjects used fo r the study consisted of 5 males and 5 females who, altogether, made a tota l of 156,250 judgments on a semantic d i f f e r en t i a l scale ranging from "very, very un-pleasant" (1) , to "very, veryi:pleasant" (9 ) . The resu l ts of the study, which i s rather extensive because of the large number of fac to rs , are presented by Helson and Lansford in a ser ies of graphs and tab les . They w i l l be reported here in point form: 1. The general resu l ts showed that both the spectral energy of the l i gh t sources and the color of the back-grounds were highly s i gn i f i c an t in determining rat ings on pleasure. 54. TABLE 13 25 colors used as backgrounds. Hue Value/chroma 5R 2/2 5/2 8/2 5/4 2.5YR 5/14 5Y 2/2 5/2 8/2 5/4 5GY 7/10 5G 2/2 5/2 8/2 5/4 2.5PB 2/2 5/2 8/2 5/4 10PB 3/10 10RP 4/12 N 1/ 5/ 10/ 55. 180 - E 3 0 O O K f l u o r e s c e n t warm white EI ( s t a n d a r d ) •1500K f luorescent c o o l white —E (deluxe equivalent) 160 6 5 0 O K f luorescent d a y l i g h t ( d e l u x e e q u i v a l e n t ) 6 8 0 0 K f i l t e r e d incandescent ( M a c b e t h ) 140 — I 2 8 5 4 K C I E i l luminant A 7\ x / i i i i I i i i i I i i i i I i i i 1.1 i i i i ! i i i i 400 500 600 Wavelength (nm) 700 FIGURE 10 Spectral energy d i s t r ibut ions of f i ve l i gh t sources. 56. 2. As table 14 shows, the "best" l i g h t source turned out to be the 4500° K deluxe f luorescent cool white, whereas the worst was the Macbeth f i l t e r e d 6500° K source. 3. "While co lor o f [ l i g h t ] source inf luences pleasantness rat ings to a highly s i gn i f i c an t degree, background co lor i s s t i l l more important s ince co lor contrast may d r a s t i c a l l y a l t e r the appearance of object co lo rs . " (p. 1538) 4. Sex differences were found to be s i gn i f i c an t i n the co lor preferences: " . . . the warm colors (R, YR, and Y) were rated higher by women i n almost a l l sources, while, the cool colors (B, PB, P, and RP) were rated higher by men in most of the sources.: (p. 1538) 5. Value contrast was found to be that dimension o f co lor which had the most pronounced e f fec t on the pleasantness rat ings: the greater th i s contrast , the greater the pleasure rat ing o f the combination o f tes t co lor and background. 6. The e f fec ts of chroma contrasts were not as c lear as the ef fects of value, although there appeared to be 57. TABLE 14 Pleasantness rat ings f o r 5 i l lura lnants. Color temperature Pleasantness Source (°K) rat ings Incandescent tungsten 2854 5.90 Fluorescent standard warm white 3000 5.78 Fluorescent cool white (Deluxe equivalent) 4500 5.94 Fluorescent dayl ight 6500 5.85 F i l t e red incandescent 6500 5.69 58. a tendency for higher pleasure rat ings with greater chroma contrasts. For small chroma contrasts between test co lor and background, the response rat ings could be e i the r pleasant or unpleasant. 7. For hue contrasts, there was a tendency fo r higher pleasure rat ings the greater the contrast was. However, these resu l ts were not very c lear e i the r , since " . . . i t was found that e i ther the same, c lose ly re la ted , or complementary hues may enter into e i the r good or poor color combinations." (p. 1539) As long as some value contrast i s preserved, for pattern v i s ion to occur, any hue combination may be judged pleasant. "On the meaning of colours" (S i v i k , 1974). Although pr imar i ly concerned with colors in arch i tectura l se t t ings , I t 1s of in teres t here because i t i s a continuation of the t rad i t i ona l "atomist ic" meifehod of e l i c i t i n g responses to ind iv idua l colors and, from these resu l t s , to general ize about other and more complex s i t u a -t i ons . The general izat ions to colors in architecture are not very c l ea r , however, in view of the fact that the author attempts to re late responses to s ing le color s t imul i to colors on actual bu i ld ings . His hope 1s,v!»ery simply put, that i f a pa r t i cu la r test 59. co lor resul ts In a judgment of , say, happiness, that same co lor , i f painted on a bu i ld ing , w i l l resu l t i n inducing a state of happiness i n people who encounter the bu i l d ing . Two resul ts of th i s study are of pa r t i cu la r Interest here. F i r s t , from a factor analysis of an unspecif ied number of semantic d i f f e r en t i a l sca les , four factors which resemble those found by Hogg (1969) and reported e a r l i e r , were found. These were: (1) "g la r ing-exc i t ing" , (2) "pos i t i ve -beaut i fu l - tas te fu l " (eva luat ive) , (3) "potency" and (4) "temperature". Secondly, there was no evidence that hue had any s i gn i f i c an t e f fec t on the response measures, whereas both value and chroma d id . These resu l ts agree with most of the resul ts reported here so f a r . Summary As can be seen from the research reviewed so f a r , there i s a good deal of disagreement among the f indings of the various researchers who have used s ing le colors or co lor pairs as s t i m u l i , as e v i -denced by the reviews of Norman and Scott (1952) and Burnham, Hanes and Bartleson (1963). In pa r t i cu l a r , Norman and Scott note that the var iety of types of s t imu l i used ( i . e . , varying s i z e s , co lors , e tc . ) as wel l as the tendency to prematurely generalize ffirom the resu l ts of responses to co lor swatches to color in general may account fo r the discrepancies. They recommend that 60. " . . . the prac t i ca l usefulness of the resu l ts had best be tested 1n a prac t i ca l s i t ua t i on , rather than Inferred from 2-1nch squares." (p. 192) This was in fact done in the study by Helson and Lansford 01970) in which they tested pleasantness rat ing against sources of i l luminat ion for butter , raw beef, apple, l e t tuce , tomato and male faces. They found that the l i g h t source which produced the most pleasing resu l ts was the 2854° K incandescent tungsten source, whi le the worst l i g h t source turned out to be the 3000° K standard f luorescent warm white. These resu l ts c on f l i c t with those reported e a r l i e r — i n which the 4500° K deluxe f luorescent cool white was found to be the best and the 6500° K Macbeth f i l t e r e d incandescent source the worst—and th is seems to suggest that Indeed Norman and Scott may be r i gh t . Another source of potent ia l disagreement pointed out by Norman and Scott (1952) i s re lated to the response measures. They wr i te: Most of such studies have e x p l i c i t l y or i m p l i c i t l y defined "a f fec t " i n terms of a "pleasantness-unpleasantness" con-tinuum, and interpret high preference, to mean that the color i s h ighly pleasant and low preference to mean the opposite. On the one hand th is simple "operat ional" de f i n i t i on may be c r i t i c i z e d for not inc luding enough of what Is commonly referred to by the term "a f fec t ive value." I f the word i s used in the same sense as "emotional tone," i t would prob-ably be appropriate to consider not only "pleasantness" and "unpleasantness," but also other qua l i t i e s of a f fec t indicated by soch words as "happy," "sad," "depressing," "stimulating 1,"' "soothing," e t c . (p. 95) 61. In summary, there seem to be three important aspects which must be taken into account in research dealing with colors and emotional responses: 1. The s t imul i must be spec i f ied in accurate, co lor imetr ic terms. 2. The response measures must be constructed, in such a way that they to the greatest extent possible assess emotional states of the subject. 3. They must be";truly representative of the s t imul i or physical phenomena the researcher wishes to make inferences about. The f i r s t requirement i s discussed In greater de ta i l in the next sec t ion , and a large port ion of chapter IV i s devoted to actual co lor imetr ic spec i f i c a t i ons . The second requirement i s dealt with in sect ion I I I of t h i s chapter, and the t h i r d require-ment w i l l be discussed further in chapter VIII in which the resu l ts of the present experiments are evaluated. 62 . II THE PROBLEM OF COLOR SPECIFICATION OF COMPLEX COLOR STIMULI As the studies with "simple" co lor s t imul i showed, i t i s r e l a t i v e l y easy to speci fy a stimulus co lor in co lor imetr ic terms. Any one of a number of "co lor systems" may be used fo r th i s purpose, and in the present research, the wr i t e r has chosen to use the Munsell as wel l as the C L E . systems. However, when i t comes to spec i fy lngsst imul i which contain more than two com-ponent co lors , the task goes beyond that of ca lcu lat ing in terva ls between colors or addit ive re la t ionsh ips . Both the spec i f i ca t ion of component colors as wel l as t h e i r i n te r - re la t i ons are important in the spec i f i ca t ion of complex color s t imul i (Granger, 1955d), and i t i s se l f -ev ident that the in ter- re la t ionsh ips cannot be stated without f i r s t spec i fy ing the components. Thus the f i r s t requirement of the co lor imetr ic spec i -f i c a t i on of a complex co lor stimulus i s the l i s t i n g of spec i f i c a -t ions of ind iv idua l components. The next question one would l i k e to ask about the complex stimulus i s the extent to which these component colors par t i c ipate 1n the st imulus. This may simply involve s tat ing the number of reds, the number of ye l lows, the number of colors of value 5, and so on. ifhe f i na l statement of stimulus "content" i s a quant i tat ive expression in the form of "how many". 63. F i na l l y , since the various quant i t ies of colors can appear in innumerable locat ional configurations in the stimulus (the number depending on the number of component co l o r s ) , i t i s des i r -able t o , in some way, express th i s d i s t r i bu t i on of colors across the stimulus surface. Once the fol lowing three spec i f i ca t ions have been made, a complex co lor stimulus can be sa id to have been spec i f ied f u l l y : 1. Component spec i f i ca t i on A spec i f i ca t i on of the color at t r ibutes of the ind iv idua l co lor elements making up the st imulus. 2. Quantitat ive spec i f i ca t ion A quant i tat ive assessment of the number of these elements i n the st imulus. 3. D is t r ibut ion spec i f i ca t i on The locat ion or d i s t r i bu t i on across the stimulus surface of the co lor elements. In the studies with s ing le colors or pairs of colors the second and th i rd requirements fo r a complete spec i f i ca t ion were c l ea r l y not re levant. In Granger's studies with a b i pa r t i t e f i e l d (Granger, 64. 1955b), for instance, i t was mentioned that the "standard" was to the l e f t and the tes t co lor to the r i gh t during the presenta-t i o n . However, he does not indicate whether a reversal of the colors might make a change to h is r e su l t s . The th i r d requirement of locat ion was not appl icable to his research. In terms of ease with which the three requirements can be met, the t h i r d i s by far the most d i f f i c u l t . The present thesis attempts in several ways to meet the th i rd requirement (described in deta i l in chapter IV) but these are not the f i r s t attempts at th i s type of spec i f i c a t i on . I t i s i ns t ruc t i ve to examine three methods which In very d i f f e ren t ways suggest ways of accomplishing a d i s t r i bu t i on spec i f i c a t i on . Neither of these procedures, i n c i den ta l l y , bad the present purpose of s p e c i f i -cat ion as t he i r object ive . "Color in Pa int ing" (McAdory, 1926). In th i s study, which was intended to ins t ruc t teachers in how to teach co lor in schools, McAdory used examples of eight famous French and Dutch paintings to demonstrate various ways of describing them. Each reproduction was accompanied by a verbal descr ipt ion of the general theme and, which IsS of pa r t i cu la r in te res t here, by an analysis—both verbal and graphic—of the co lor content of the pa in t ing . As an example, the verbal analysis of the paint ing "Madame Le Brun and her daughter" 65. by Le Brun ( f igure 11) reads as fo l lows: Many areas are composite, as 1n the ch i l d ' s dress, where parts are Purple-blue, others Blue or Green. A careful study of the hues has been made and the resu l ts are I l l u s t r a t ed 1n [ f igure 11 ( a ) ] . This f igure also shows the dif ferences 1n chroma. Most of the hues f a l l wi th in the Red to Yellow sequence but contrast i s afforded through the use of Blue, [ f igure 11 (b)] shows the value and chroma re lat ionsh ip of a l l the colors o f the pa in t ing . The general impression of in teres t areas 1s given in [ f igure 11 (c)] with a pa ra l l e l l i s t i n g of colors with the percentage area of each. A small graph i s drawn 1n r e l a -t ionship to these proport ions. . . . These small blocks of co lor may be arranged in value sequence, in hue sequence, In chroma sequence, or even in general area sequence as i l l u s t r a t e d by the small drawings. Table 15 shows McAdory's in terpretat ion of in teres t areas, t he i r Munsell notat ion, per cent of these areas and the descr ipt ion both of the co lor and of i t s object . In add i t ion, a bar graph adjoins the items In the table to v i sua l l y indicate the propor-t ion of the colors used for each item. The method, in i t s s imp l i c i t y , has the advantage that i t i s very d i r e c t , and that only comparison samples of Munsell co lor chips are needed to carry out the co lor matching and determination of the major color areas i n the pa in t ing . I t 1s very d i r e c t In the sense that no involved measuring procedures are needed. The f i r s t step in McAdofcy's method i s to es tab l i sh "areas of in teres t " in the picture and the i r average or predominant co lo r . A graphic representation of the resu l ts of th i s "analys is" i s FIGURE 11 The p a i n t i n g "Madame Le Brun and her daughter" used by McAdory f o r a n a l y s i s . 67. shown in f igure 12 ( c ) . The next step i s to map out the colors l i s t e d in table 15 in the Munsell co lor space. This i s done in f igure 12 (a) fo r hue and chroma and in f igure 12 (b) for value and chroma. In add i t ion, a very simple graphic representation of the estimate of per cent area of each co lor in the picture i s shown added to the items in table 15. F i n a l l y , these l a t t e r areas are rearranged into a set of three bar diagrams (figure 1 12 (d) , (e) and ( f ) ) o f fer ing yet another v isua l in terpretat ion of amounts of hue, value and chroma in the p i c ture . Comparing McAdory's procedures with the three require-ments for a f u l l spec i f i ca t ion described e a r l i e r , i t i s evident that at leas t the f i r s t requirement (of component spec i f i ca t ion) and the second requirement (of quantity spec i f i ca t ion) have been met. Furthermore, as a step in the process of component s p e c i f i -cat ion, she has to some extent f i l l e d the th i rd requirement {of d i s t r i bu t i on spec i f i ca t ion) as w e l l . The resu l t of th i s i s the sketch shown in f igure 12 ( c ) . Granted that v i s ua l l y i t represents only a small change from the o r i g ina l painting--and researchers who have worked w1th"unspecified" color pictures might with some j u s t i f i c a t i o n claim that had they shown repro-ductions of t he i r s t imu l i , the resu l t would have been the same type of "spec i f i cat ion%-but in an essent ia l aspect o f information 68. TABLE 15 Interpretat ion and spec i f i ca t i on of Interest areas. Interest Munsell % grouping notation area Description 1 5YR7/3 8 f l esh 2 5R8/1 5 mother's dress " 5B3/2 14 ch i l d ' s dress 7R4/5 4 red band and sash 3 < Y3/2 3 ha i r 5BG3/1 2 cushion . 5Y5/3 14 mother's wrap 4 5YR5/1 50 background Proportion of colors used I I I I I 1 I 1  I I I I I I TTT1 TJ3 69 . 3/ 5/ 7/ 8/ Dark Middle Light FIGURE 12 McAdory's graphic analysis of the paint ing "Madame Le Brun and her Daughter". 70. about p icture content i t seems to represent a subtle s h i f t from the unanalyzed representation to one which graphica l ly at l eas t , and cer ta in ly ana l y t i c a l l y , shows the d i s t r i bu t i on or locat ion of component p icture par ts . There are two problems, however, inherent in McAdory's procedure. The f i r s t concerns the method by which the major in terest areas in a picture are extracted. Presumably she herse l f decided which areas were of i n te res t , but a better procedure might have been to have a number of subjects do the judging and then average the i r responses. This way the se lect ion would have been somewhat more object ive . Secondly, the determination of per cent area was probably based on a method of v isua l est imat ion. I f an area measuring device such as a planimeter had been used to measure the areas] the resu l ts might have been more accurate. Th i rd ly , and most importantly, the graphic representation of the d i s t r i bu t i on spec i f i c a t i on , although in terest ing in i t s e l f , i s not very useful fo r further ana lys i s . That I s , 1f the object had been to compare the d i s t r i bu t i on spec i f i ca t ion to , say, a var iety of responses, a way of numerically speci fy ing the d i s t r i bu t i on would have been needed rather than a graphic representat ion. "The Ar t of Color and Design" (Graves, 1951). The purpose of McAdory's study and presentation was to enable teachers to better promote the 71. study of co lor and the appreciation of a r t . Graves' work, on the other hand, i s s pe c i f i c a l l y aimed at helping the a r t i s t produce better pictures by exposing him to the var ie t i es of design elements such as l i n e , d i r e c t i on , shape, s i z e , texture and co lo r . Further-more, each of these design elements i s analyzed separately in terms of the i r possible range, in order that the best possible re la t i onsh ip , both with in the design elements and between them, can be achieved. The way Graves describes ways of determining the hue, value and chroma content of a p icture i s what i s of pa r t i cu la r in te res t here. Figure 13 shows to the l e f t the p icture to be analyzed, and in the middle the determination of the dominant hue (0) , the adjacent hue (A) , and the contrast ing (but not complementary) colors (W) of the background and (Z) of the contrasts i n the background. Below the hue c i r c l e are l i s t e d the chroma leve ls of each of these four co lors: co lor (D) - chroma 2; co lor (A) 9 chroma 8; colors (W) and (Z) - chroma 6. To the r ight in f igure 13 are shown separately the four component colors of the p icture together with t he i r Munsell notaions. A l so , i n th i s "plan of co lor areas" i s shown graph ica l ly the proportions of these four co lo rs . Figure 14 shows, to the r i gh t , the locat ion of the four colors A, D, W and Z on the value scale (ranging from value 2 to 8 ) , while to the l e f t Graves shows the s i x possib le hue in terva ls which determine the type or c l a s s i f i c a t i o n 6"f co lor scheme used in the FIGURE 13 Hue p lan and g raph i c a n a l y s i s o f p i c t u r e w i t h l i g h t p a t t e r n dominat ing dark background. D COLOR CHORD FOR COMPOSITION VIII The hue, value, and chroma plans when combined produce the total color intervals or contrasts. This diagram illustrates the order of sizes of the six un-equal color intervals that form the D color chord for Composi-tion VIII. M I N O R . ( S M A L L E S T ) I N T E R V A L OR . W E A K E S T C O L O R C O N T R A S T . w A Y% GY w E5x Y% i | B - P B . ^ w VALUE PLAN VI D A 4 A Y% M A J O R ( G R E A T E S T ) I N T E R V A L OR. A S T R O N G E S T C O L O R C O N T R A S T . FIGURE 14 Hue, value and chroma in terva ls or contrasts. 74. p i c tu re . In th i s case a r e l a t i v e l y l i gh t pattern of value 6 dom-inates a darker pattern of value 3, chroma 6 i s predominant and, as f a r as hue i s concerned, §GY (together to some extent with 5Y) dominates 5PB and 10PB. The proportions of the four colors might, i f the co lor scheme i s fo r a work in preparat ion, have been determined from some aesthet ic measure such as the "golden mean" which, in the Fobonacci se r i e s , fol lows the numerical sequence of 8, 13, 21, 34, 55, . . . (Graves, p. 237). I f Graves* method i s used fo r ana ly t i ca l pur= poses, the areas might have been measured with an area measuring device, and the same sort of spec i f i ca t ion schemes as described arr ived a t . Graves' method does not d i f f e r substant ia l l y from that of McAdory's mentioned e a r l i e r . Essent ia l l y these spec i f i ca t i on schemes do not advance" beyond the second stage of spec i f i ca t i on (of quant i ty) , although the inc lus ion of the "color chord" spec i -f i ca t i ons in Graves' procedure ( f igure 14) 1s a refinement in the quant i tat ive spec i f i ca t i on over McAdory's method. "The recognit ion of faces" CHarwron, 1973). This work, although pr imar i ly concerned with the recognit ion of f a c i a l features, i s o f in terest here because i t employs a method of handling p icture content which i s used in the present thes i s . The method in question 75. i s one of d iv id ing the picture surface Into a large number of small squares which in turn can be dealt with as separate compon-ents. In Harmon's study, a method of scanning an object p icture or a s l i de and reconstructing i t by e lec t ron ic means i s employed as shown in f igure 15. The main advantage of th i s procedure i s the fact that the content of the object p icture can be manipulated e l e c t ron i ca l l y before the f i n a l reproduction i s produced. Figure 16 shows one example of Harmon's reconstruction technique. I t was o r i g i na l l y produced in co lor , by a method which Harmon does not describe In d e t a i l . Harmon determined that a minimum matrix s i ze of 16x16 was necessary for the recognit ion of faces, and that i t made l i t t l e di f ference whether a gray scale with eight leve ls or one with s ixteen levels was used. This information w i l l be referred to l a te r 1n chapter IV, when the construction of the research s t imu l i i s described. The f i r s t point then to be made, as a resu l t of Harmon's study, i s that by d iv id ing a picture surface into small component elements, a log i ca l way has been prepared fo r the spec i f i ca t i on of ind iv idua l p icture components. To ar r ive at value leve ls f o r these components, Harmon measured 1024 points in each p icture and averaged the values for sub-groups of 64 po ints . This produced unique value leve ls fo r each of the 256 squares in the 16x16 FIGURE 15 General arrangement of Harmon' s scanning procedure. 77 FIGURE 16 The "Mona L i s a " r e cons t r u c t ed from d i g i t i z e d d a t a . 78. matr ix. F i na l l y , these 256 unique value leve ls were compared to the 8-level value scale mentioned above, and whichever value on th is sca le was c losest to the unique leve l of each square was chosen to represent that square. This procedure does not d i r e c t l y apply to the present re-search since here, pictures are constructed rather than analyzed as in Harmon's case. By the same token, the question of perceptual impl icat ions of the averaging procedure need not be ra ised here. The second point of Interest resu l t ing from Harmon's study concerns the sequential recording of the p icture content. This method can in some ways be construed as a form of d i s t r i bu t i on spec i f i c a t i on , thus f i l l i n g the requirement of the t h i r d type of stimulus spec i f i ca t i on out l ined e a r l i e r . The method i t s e l f w i l l be discussed more f u l l y i n chapter IV when various procedures for d i s t r i bu t i on spec i f i ca t i on are developed. In summary, i t appears that of the three studies reviewed which have dealt with the various degrees of stimulus spec i f i c a t i on , Harmon's study Is the most success fu l . Although i t did not e x p l i c i t l y speci fy the components of the p i c tu re , nor the quant i t ies o f the components, they were nevertheless accounted fo r and, furthermore, i t was the only one of the three studies which produced any h int at a numerical method of speci fy ing the d i s t r i bu t i on across the p i c -ture surface of ind iv idua l co lor elements. 79. I l l INFORMATION RATE AND THE MEASURE OF EMOTIONAL RESPONSES As Norman and Scott (1952) point out in t he i r review of experiments dealing with color and a f f ec t , the use of a response dimension such as "pleasant-unpleasant" may not capture enough of what they re fer to as "emotional value". They suggest that words l i k e happy, sad, depressing, st imulat ing and soothing may be included in order to obtain a more appropriate measure of a f fec t i ve value. I t has also been pointed out by many researchers in the f i e l d of environmental psychology—to which the present wr i t e r considers appl ied aesthetics to be c lose ly re l a ted—(e .g . , Cra ik, 1970; I t t e l son , et a l . , 1974) how the lack of an overa l l paradigm or theory ty ing together the numerous but diverse f indings has hampered the development of the f i e l d to a great extent. The problem has essen t i a l l y been one of stimulus spec i f i ca t i on and of f ind ing appropriate measures for use In in terpret ing the ef fects of the s t i m u l i . One proposal which shows promise of consol idat ing a wide range of response measures into four major dimensions which are both theore t i ca l l y reasonable and quite simple to use, i s Mehrabian and Russe l l ' s (1974) measures of pleasure, arousa l , dominance and 80. information rate . (Since extensive reference w i l l be made to t he i r book, An Approach to Environmental Psychology, "M&R" w i l l be used for short to re fer to th i s work.) These four measures are pr imar i ly psychological ly or iented since they emphasize the behavioral responses and the intermediate, "primary emotional responses" of pleasure, arousal and dominance (as the diagram in f igure 17 shows), and i t s main merit l i e s i n the f a c i l i t y with which i t can re late measures on the dimensions of pleasure, arousa l , dominance and information rate to concrete behavioral responses such as approach-avoidance, e t c . ( e . g . , Russe l l , 1974). The problem of stimulus spec i f i ca t i on i s , as the present wr i te r has conceptualized 1t , a physical measurement problem which 1s unique and can be dealt with qui te separately from that dealt with by Mehrabian and Russe l l . The Information rate concept I t was pointed out in previous sections how i t has always been the asp i rat ion of researchers 1n the f i e l d s of perception and aesthetics ( e . g . , Valent ine, 1962) to a r r ive at a spec i f i c a -t ion of the st imul i employed or found in a complex p i c t o r i a l or environmental s i t ua t i on . Most experimenters to date, however, have had to content themselves with the spec i f i ca t ion of one o r , at the most, a very l imi ted number of s t i m u l i . The main reason 81. THE ENVIRONMENT Sense modality variables (e.g., color and temperature) Information rate (characterizing the spatial and temporal relationships among the stimulus com-ponents of an environment) Characteristic emotions associated with PERSONALITY PRIMARY EMOTIONAL RESPONSES Pleasure Arousal Dominance BEHAVIORAL RESPONSES Approach-avoidance (which includes physical approach, exploration, affiliation, performance, or other verbal and non-verbal communications of preference) FIGURE 17 Mehrabian and Russe l l ' s proposed framework fo r studying the environment i n terms of pleasure, arousa l , dominance and information ra te . 82. fo r th i s i s , as was shown i n sect ion I I , that as the stimulus f i e l d gets more complex, the more impossible the stimulus spec i -f i c a t i on becomes. Furthermore, the desire toeemploy "real l i f e " s t imul i such as color pictures or environmental displays or s i tuat ions has made th is a pa r t i c u l a r l y aggravating problem. Information theory concepts have shown some promise in the resolut ion of th i s problem. Procedures of counting elements (b i ts ) together with mathematical formulations of uncertainty have found use i n research with s t imu l i which are neither too complex fo r "counting" nor too simple to f a l l into the category of simple s t imu l i , (e.g, Atneave, 1957; Chipman, 1977) However, fo r the appl icat ion to r ea l l y complex s t i m u l i , such as co lor pictures or Cra ik ' s (1970) "environmental d i sp lays" , information theory as a mathematical measure has not so fa r proved i t s e l f use fu l . SfThere have been no attempts to measure the information of the complex s t imu l i in everyday environments, such as v isua l displays of houses or people." (M$R, p. 80) The notion of Information i s re lated to the cer ta in ty-uncertainty continuum, and i t has to do with the p robab i l i t i e s or r e l a t i ve frequencies of events. (Berlyne, 1971) In other words, "what one knows" i s re lated to "what one does not know" or to "what there i s to know". As the concept 1s exp l i cated, i t i s conceptualized along a set of dimensions such as complex, 83. random, j a r r i n g , heterogeneous, var ied , novel , and so on, a l l of which are ind icat ive of a high rate of information, as opposed to dimensions such as simple, patterned, harmonious, homogeneous, redundant, and f am i l i a r , a l l of which indicate a low information ra te . Furthermore, a d i s t i n c t i on can be made between information rate and amount of information " . . . because th i s d i s t i n c t i on allows one to understand the change in a person's . . . reactions to a s t a t i c stimulus conf igurat ion ( e . g . , a paint ing) over t ime." (M&R, P. 94) The concept of information ra te , as expressed in th i s fashion—in terms of conceptual or verbal dimensions—is now amenable to further ana lys i s , and the development of a verbal measure of information rate i s the f i n a l step in th i s process of accounting fo r physical charac ter i s t i cs of the complex stimulus s i t u a t i on . Figure 18 shows the I n i t i a l questionnaire used by Mehrablan and Russel l fo r the e l i c i t i n g of information ra te . (M&R, p. 90) Factor analyzed, the scores from th is questionnaire reduce to f i ve factors (table 16) which, a f te r further analysis using scores already obtained on the pleasure, arousal and dom-inance diimehslions, to be described sho r t l y , produced the regres-sion equation shown in table 17. (M&R, p. 93) Since factors one and f i ve turned out to " . . . character ize aspects of environments 84. Instructions to Subjects Please use the adjective pairs below to describe situation [the situation identification number was entered here]. Each of the following adjective pairs describes the situation or the rcbtion among the various parts of the situation. Put a check somewhere along the line (Example: — -: ) to indicate what you think is an appropriate description. (+) simple : : : : : :• - - -: : complex (+) patterned : : : : : : : : random (-) novel : : : : : : familiar (+) meaningful : : : :- - - -: :- - • -: : meaningless (+) small scale : : — - : ; : : : : large scale (-) immediate : : : : : : distant (+) good form : : : : : :- - - -: :- - - - bad form (-) varied : : : : : : : : redundant (-) dense : : : : : : : : sparse (+) common : : : : : : : : rare (-) heterogeneous : : : : : : : : homogeneous (-) intermittent : :?—: : : : : : continuous (-) crowded : : : : : : : uncrowded (+) usual I : : : : : : : : surprising (-) man-made : : : : : : : : natural (+) harmonious : : : : : : : : jarring (-) asymmetrical : : : : : : : : symmetrical (+) similar : : : : : :-.-•>.. . contrasting (+) rural — uioan (+) consonant - - - - - dissonant (—) indoor — r : : : : : : : outdoor *The signs preceding each adjective pair indicate the scoring direction. FIGURE 18 I n i t i a l set of adject ive pairs fo r information rate measure. 85. TABLE 16 Factor ia l composition of the Information rate sca les . Scoring d i rec t ion and Item Factor loadings Factor 1 (+) good form-bad form +0.84 (+) harmonious-jarring +0.74 (+) meaningful-meaningless +0.69 (+) consonant-dissonant +0.64 Factor 2 (+) common-rare +0.80 (-) nove l- fami l iar -0.77 (+) usual-surpr is ing +0.71 Factor 3 (+) small sca le- large scale +0.74 (+) simple-complex +0.69 (-) dense-sparse -0.63 (-) crowded-uncrowded -0.48 (-) Immediate-distant -0.-34 Factor 4 (-) intermittent-continuous -0.67 (-) heterogeneous-homogeneous -0.64 (-) asymmetrical-symmetrical -0.61 (+) s imi lar-contrast ing +0.56 (+) patterned-random +0.50 (-) varied-redundant -0.41 Factor 5 (-) Indoor-outdoor -0.83 (+j rural-urban +0.78 (-) man made-natural -0.74 86. TABLE 17 Items and factors of the information rate scale expressed as functions of emotional s ta tes .* Itern-total Regression Mu l t ip le cor re la t ion Item equation cor re la t ion 0.51 good form-bad form -0.78P 0.78 0.68 harmonious-jarri ng -0.62P+0.32A 0.72 0.29 meaningful-meanlngless -0.61P-0.15A 0.62 0.56 consonant-d1ssonant -0.42P+0.20A 0.49 Factor 1 -0.77P+0.13A 0.80 0.21 common-rare +0.14P+0.18A 0.22 0.33 fami l iar-novel +0.09P+0.28A-0.09D 0.29 0.32 usual-surpr is ing +0.13P+0.31A 0.33 Factor 2 +0.14P+0.32A-0.10D 0.33 0.16 small sca le- large scale +0.20P+0.32A-0.12D 0.36 0.59 simple-complex -0.16P+0.49A 0.53 0.42 sparse-dense +0.27A 0.27 0.63 uncrowded-crowded -0.36P+0.37A 0.55 0.11 distant-Immediate +0,11P+0,20A 0.22 Factor 3 -0.11P+0.53A 0.55 0.34 continuous-intermittent +0.11A 0.11 0.50 homogeneous-heterogeneous -0.09P+0.24A 0.26 0.57 symmetrical-asymmetrical -0.23P+0.22A 0.34 0.51 simi lar-contrast1ng +0.34A 0.34 0.26 patterned-random +0.13A+0.08D 0.17 0.14 redundant-varied +0.44P+0.29A 0.50 Factor 4 +0.37A 0.37 0.35 outdoor-Indoor -0.34P-0.13A+0.14D 0.36 0.47 rural-urban -0.32P+0.10A 0.35 0.50 natural-man made -0.42P +0.100 0.43 Factor 5 -0.44P +0.12D 0.44 * In these equations, P=pleasure, A=arousal and D=dominance. 87. that pr imar i ly affected p leasure . . . " (M&R, p. 92) the items cons t i -tut ing these two factors were deleted from the f i n a l question-naire as shown in f igure 19. (M&R, Appendix D) One pa r t i cu l a r l y in terest ing aspect of th i s verbal measure of informationrrate i s that " . . . ind iv idua l di f ferences in past encounters with environments and dif ferences in the d iscr imina-t ion of patterning are automatical ly taken into account." (M&R, p. 81) This i s of specia l importance since "redundance" or repeat-a b i l i t y .is such a prominent part of the information concept. Purely mathematical methods of assessing Information rate would have great d i f f i c u l t i e s dealing with redundancy sincd 1t i s re lated to the experiences of the observer. While th i s questionnaire thus captures some subject ive aspects, the ob j e c t i v i t y o f the procedure cannot be ca l l ed into question as i t rests f i r m l y on consensual ground. The scoring of the information rate measure i s very simple. The 14 items of the measure ( f igure 19) are scored from -4 to +4 with (+s) or (-) signs (which are not shown to the subject) i nd i ca -t ing which i s the pos i t ive and which the negative end of the sca le . F i n a l l y , the 14 scores thus obtained are added together and d iv -ided by 14 to obtain one value. 88. Instructions to Subjects Please use the following adjective pairs to describe the situation shown (or described). Each of the following adjective pairs helps define the situation or die relation among the various parts of the situation. Put a check mark somewhere along the line (Example: : - - v - - : ) to indicate what you think is an appropriate description. redundant complex familiar large-scale contrasting sparse continuous surprising homogeneous crowded symmetrical distant rare random *ln actual administration, the scoring direction signs to the left of each scale are omit-ted. Assign a score of -4 to checks placed in the farthest left space, -3 to the space next to it, on to +4 to checks placed in the farthest right space. To obtain a total score, change the signs of responses to the negatively signed items, and then sum over all responses. With this procedure for obtaining total scores, our sample yielded: Mean =-2.2, and standard deviation = 15.7. (-) varied (+) simple (-) novel (+) small-scale (+) similar (-) dense (-) intermittent (+) usual (-) heterogeneous (+) uncrowded (-) asymmetrical (-) immediate (+) common i (+) patterned FIGURE 19 Mehrabian and Russe l l ' s measure of information rate . 89. The emotional responses The three emotional dimensions making up Mehrabian and Russe l l ' s emotional response measures are: pleasure, arousal and dominance. They are defined as fo l lows: 1.. Pleasure The interpretat ion of pleasure i s close to the common sense meaning of i t except that , to avoid high leve ls of cor re la t ion ( i . e . , orthogonal ity) with the other two dimen-s ions, i t i s d ist inguished from " . . . preference, l i k i n g , pos i t ive reinforcement, approach-avoidance." (M&R, p. 18) 2. Arousal Although evidence ex is ts fo r physio logica l corre lates of the psychological concept of arousa l , (M&R, pp. 14-15), Berlyne's (1960) notion 6f arousal as a " . . . "feel ing state varying along a s ing le dimension ranging from sleep to f r an t i c excitement, . . ." (M&R, p. 18) i s adopted. I t i s further noted that arousal corre lates with information rate (information rate = 0.57 arousal state [M&R, p. 94]) , a fact which i s apparent from table 17> which shows that in a l l but two cases, arousal was a s i gn i f i c an t component of informa-t ion rate . (M&R, p. 92) 90. 3. Dominance This dimension i s contrasted to "submissiveness", and re lates to an i nd i v i dua l ' s fee l ing of the extent to which he i s res t r i c ted or free to act in a var iety of ways accord-ing to the s i tuat iona l charac te r i s t i c of the environment. (M&R, p. 19) The way these three dimensions are defined shows them to be decidedly a f fec t ive in nature. However, since the de f in i t i ons are of the oper-at ional kind (although to a large extent based on previous research), i t i s quite possible that a lack between them and the behaviohejihey are supposed to character ize e x i s t s . To answer th i s possib le objec-t i o n , Mehrabian and Russell compare the i r three dimensions to Osgood's factors of evaluat ion, potency, oriented a c t i v i t y , s t a b i l -i t y , tautness, novelty, recept iv i ty and aggressiveness. (Osgood, et a l . , 1957) These wr i ters found that the per cent common variance at t r ibutab le to these eight factors was ( in the order mentioned): 38.00, 16.54, 11.22, 8.20, 6.20, 7.37 and 6.24. These resu l ts are from the "Thesaurus sampling" study, and another study y ie lded four factors of which three were espec ia l l y prominent. These three were (with per cent of to ta l variance shown in brackets): evaluative (33.78), potency (7.62), and a c t i v i t y (6.24). The fourth factor only accounted for 1.52% of the to ta l variance. 91. I t 1s thus with a good deal of j u s t i f i c a t i o n that Mehrabian and Russell draw on the evidence of Osgood et a l . , (1957) In support of the i r three a f fec t ive dimensions. "These studies [of Osgood et a l . ] defined arousal (the a c t i v i t y f a c t o r . . . ) and pleasure (the evaluative factor) as basic responses to s t i m u l i , and also suggested a th i rd dimension (potency)." (M&R, p. 16) In the development of the measure for the three emotional dimensions, questionnaires s im i l a r to that shown In appendix A were employed. This pa r t i cu l a r questionnaire 1s the version which the wr i t e r used in the p i l o t study (described in deta i l In chapter V) . A f te r the var iables from several i n i t i a l studies were analyzed, Mehrabian and Russell found the factors of pleasure, arousal and dominance to emerge quite strongly ( c f . table 18). The resu l ts of analyses using the f i n a l version of the measure showed that the to ta l variance accounted fo r was 642. This was d i s t r ibuted among the three emotional dlmensionsnas fo l lows: Pleasure - 27% Arousal - 23% Dominance - 14% Mehrabian and Russell further note tha t , " . . . no e f f o r t was made in our i n i t i a l se lect ion of emotional descriptors used in Study 1 to Include a 11st of adject ive pairs that would exhaustively des-cr ibe the great d ivers i ty of human emot ions. . ." , and " . . . rather, 92. TABLE 18 Rotated factor matrix of the f i n a l set of emotional response sca les . Factor 1 Factor 2 Factor 3 Emotional response pleasure arousal dominance Happy-unhappy 0.92 0.01 0.01 Pleased-annoyed 0.91 -0.09 -0.02 S a t l s f i ed-unsat1sf1ed 0.92 -0.04 -0.01 Contended-melanchol1c 0.85 0.01 0.02 Hopefu1-des pa1ri ng 0.79 0.02 0.09 Relaxed-bored 0.84 0.08 -0.05 Stlmulated-relaxed -0.29 0.75 0.05 Exc1ted-calm -0.11 0.82 0.01 Frenzied-sluggish r0.04 0.80 0.05 J i t t e r y - du l l 0.04 0.77 -0.04 Wide awake-sleepy 0.24 0.79 0.00 Aroused-unaroused 0.06 0.80 -0.03 Control11ng-control1ed 0.11 -0.06 0.76 Dominant-submissive -0.01 0.28 0.67 Inf1uenti al-1nf1uenced 0.02 -0.01 0.79 Important-awed 0.00 0.02 0.46 Autonomous-guided 0.03 -0.09 0.69 In control-cared fo r -0.12 -0.02 0.68 Percent variance 27 23 14 93. we proceeded d i r e c t l y to construct scales that would most d i r e c t l y and uniquely measure each of the three fac to rs . " (M&R, p.-27) The hypothesized weak re lat ions of the three emotional response dimensions were confirmed by the data. Across the three studies conducted, the in ter-cor re la t ions ranged from -0.07 to 0.26. In two of the s tud ies , however, there were s i gn i f i c an t pos i t ive corre lat ions between pleasure and dominance. These low i n te r corre lat ions among the three emotional response dimensions provide support f o r our assert ion that they const i tute a par-simonious base for the descr ipt ion of the great d i ve r s i t y of emotional responses that occur in everyday s i t ua t i ons . (M&R, p. 27) As table 18 shows, there are s i x ind iv idua l semantic d i f f e r en t i a l scales associated with the emotional dimensions of pleasure, arousal and dominance. In the f i n a l version of the quest ionnaire, these are scrambled and Items reversed as shown in appendix A. Subjects are asked to con-s ider the semantic d i f f e r en t i a l sca les , one at a time, and to place a check mark along the l i ne between the two extremes. The f i r s t Item of the quest ionnaire, fo r example, might be checked o f f l i k e t h i s : HAPPY UNHAPPY +4 +3 +2 +1 0 -1 -2 -3 -4 The scoring of the Items fol lows that adopted for the information rate measure In which scales range from -4 to +4. Thus, in the above example, since "happy" i s re lated pos i t i ve l y to pleasure, th i s semantic d i f f e r -en t i a l scale would be scored as +2. Scores on items belonging to 94. pleasure, arousal and dominance, respect ive ly , are extracted from the "scrambled" presentation order, and signs are changed according to whether the pos i t ive or the negative end of the scale i s to the l e f t or to the r i gh t . Item to ta l s fo r the dimensions are added up and divided by s i x to obtain the f i na l measure of pleasure, arousal and dominance. The resultant expression for the emotional and information rate responses to appart icu lar stimulus s i tua t ion thus consists of four items: Response + c-jP; c 2A; CjD; c^I where P, A, D and I re fer to pleasure, arousal , dominance and information rate , and c-. to cA are average scores on these dimensions. 95. CHAPTER III HYPOTHESES AND RESEARCH QUESTION" As a resu l t of the theoret ica l considerations of chapter I and the evidence of past research presented in chapter I I , the fol lowing four testable hypotheses and one general research question are advanced. I HYPOTHESES 1. I t i s hypothesized that the var iables of hue, value, chroma, motif and sex w i l l s i gn i f i c an t l y inf luence the way subjects rate the displays in terms of pleasure, arousa l , dominance and information ra te . 2. I t i s hypothesized that subjects ' responses on 96. pleasure, arousal , dominance and information rate w i l l d i f f e r s i gn i f i c an t l y according to whether they are presented with a representational moti f ( I . e . , the face, the landscape or the bui ld ings) or the abstract . In pa r t i cu l a r , i t i s expected that the representational motifs w i l l resu l t i n a lower information rate score than the abstract moti f . 3. I t i s hypothesized that the var iables of hue, value, chroma, motif and sex w i l l s i gn i f i c an t l y inf luence the extent to which subjects w i l l recognize the motif of a d isp lay . 4. I t i s hypothesized that subjects ' responses on the dependent measure of information rate w i l l d i f f e r s i gn i f i c an t l y according to the ease with which they recognize the motif of the d isp lay . That i s , the eas ier i t 1s to recognize a moti f , the lower the information rate w i l l be. While hypotheses 1 and 3 are testab le , the d i rec t ion of t he i r outcome i s not pred ictab le. At best, predict ions based on past studies w i l l be Incomplete or even inaccurate because of the lack 97. of consistency of previous f ind ings . At worst, past f indings do not apply at a l l to the present research since the s t imul i used in the past were s ing le colors or pairs of co lors , while those employed here are complex. II RESEARCH QUESTION In an attempt to develop ways of speci fy ing the locat ion or d i s t r i bu t i on of color elements in the d isp lays , 24 ca lcu la t ion procedures, based on operat ional ly defined re lat ionships between ind iv idua l elements in the d isp lays, are performed (c f . chapter IV) . The resu l ts of these ca lcu lat ions are used as pred ictor var iables and regressed against the four dependent measures of pleasure, arousal , dominance and information rate in order to assess which of these operat ional ly defined re la t ionsh ip measures best describe the d i s t r i bu t i ona l character i s t i cs of the d isp lays . Since one can only speculate about the possible outcome of such a procedure, i t can best be termed "hypothesis c reat ing" , and the general research question pertaining to i t formulated as fo l lows: Do any or a l l of the numerical expressions of the component re lat ionsh ips predict the outcome of pleasure, arousa l , dominance and information rate scores for the displays? 98. CHAPTER IV THE STIMULI I COLOR DISPLAYS FOR USE AS STIMULI General descr ipt ion A tota l of 80 co lor displays were constructed, each consist ing of 256 1/4 inch squares of glossy Munsell paper supplied by the Munsell Color Company. The co lor chips were f i t t e d t i g h t l y together into a 16x16 matrix pattern and pasted onto a piece of cardboard. The s i ze of the f in ished display matrix was 4"x4". Figure 20 shows display no. 1 as i t appeared when pasted up. The 80 displays were each supplied with a black aperture mask to block out i r regu la r paste-up seams, and to i so la te ind i vidual colors from each other so that e f fec ts of simultaneous contrast were minimized. There were two types of masks used, but these d i f f e red considerably in the two experiments to be described l a t e r . The actual construction of the masks 1s there-fore described in deta i l in chapters V and VI, respect ive ly , i n 99 FIGURE 20 Display no. 1 (face) in Its i n i t i a l stage of construct ion. 100. the context of the experiments. There were two main requirements which had to be taken into account in the construction of the d isp lays . The f i r s t was that they be as varied as possible with respect to the main Independent var iab le , co lo r . The second requirement was that these var iat ions be created in such a way that a systematic re lat ionsh ip existed among them. Both of these requirements would be met by the use of the Munsell co lor system. In addit ion to co lo r , the notion of a representational "moti f", as an addit ional independent var iab le , was introduced. This was pr imar i ly done to make the displays more p i c t u r e - l i k e . The motffs w i l l be d iscussed! later in th i s sec t ion . The Munsell co lor notation system The Munsell system was developed by A.H. Munsell as an aid to the conceptual izat ion and spec i f i ca t i on of co lo r . A r t i s t s had up un t i l quite recent ly found i t d i f f i c u l t to communicate to others t he i r perceptions about co lo r . A type of physical spec i f i ca t i on system had ex i s ted , at least in theory, since Newton's time, and 1n the l as t century—when the spectrum was studied—an approach toward a precise co lor spec i f i ca t i on system was under way. However, a r t i s t s did not f i nd a workable system unt i l the appearance of the Munsell notation around the beginning of the present century. 101. (Munsel l, 1913; Nickerson, 1940) The system assumes that co lor 1s perceived 1n terms of the three dimensions of hue, value and chroma, and i t u t i l i z e s a three-dimensional "co lor s o l i d " or co lor space to vis 'ual ly represent these three dimensions (see f igure 21). Hue i s that dimension Which dist inguishes between, for example, red and blue or yel low and green, and i t 1s indicated in the co lor space as a number of planes rad iat ing from the ve r t i ca l a x i s . The "hue c i r c l e " i s divided in to f i ve pr inc ipa l hues, R (red) , Y (yel low),6 (green), B (blue) and P (purple) as shown in f igure 22, and between each of these p r i n -c ipa l hues are the f i ve intermediary hues of RP, YR, GY, BG and PB. A further subdiv is ion of the Intervals between these hues into ten parts , numbered from 1 to 10Q makes i t possible to speci fy up to a to ta l of 100 hues. I t i s noteworthy that the hues are supposed to be arranged in such a way that d iametr ica l ly opposed colors are complementaries. That i s , 5YR and 5B, for example, are supposed to be complementaries, provided that febe "neutra l" center (or i l luminant , 1n fact) remains in the same pos i t i on . I f that moves, the supposed pattern of com-plementaries no longer e x i s t s . The ve r t i ca l axis 1n the co lor space represents the l ightness or value scale (see f igure 23). The f u l l scale ranges from 0 (a per-fect black) to 10 (a perfect whi te) , although as far as surface co lor 102. FIGURE 22 The Munsell hue c i r c l e . FIGURE 23 The Munsell value sca le . 103. samples are concerned, only the range from 1 to 9 i s of p rac t i ca l importance. One in terest ing and very important aspect of the value scale 1s that i t i s divided into subject ive ly equal steps. This means that , 1n theory at l eas t , a di f ference between, say, step 2 and 3 on the scale should be perceived to be of the same magnitude as a d i f f e r* ence of one step anywhere else on the sca le . The th i rd dimension 1n the Munsell co lor space represents the colorfulness or Intensity of the co lor , and i t 1s termed chroma. As with value steps, chroma steps are designed in such a way that differences between steps are perceived to be equal, and the i r mag-nitude increases as the distance from the center of the s o l i d Increases. The ve r t i ca l ax i s , being the neutral gray sca le , i s designated as having a chroma of zero (see f igure 24). To speci fy any color in the Munsell system i t i s necessary only to wr i te the hue, the value and the chroma. Thus PB 5/6 i n d i -cates a color with the hue "purple-blue", of value 5 and with a chroma of 6 . . One complete hue plane i s shown In f igure 25. A r t i s t s and ar t educators have used the Munsell system exten-s i ve l y since i t s in t roduct ion. Maitland Graves (1951), for instance, has used i t to describe the pr inc ip les of co lor design in a r t . In his book, The Art of Color and Design, he develops and explains the various notions associated with hue, value and chroma schemes. 104. 10 8 Chroma Chroma FIGURE 24 Munsell chroma sca le . White 5/10 5/8 5/6 5 / 4 5 / 2 5/4 5/6 FIGURE 25 Munsell PB-Y hue plane Black 105. Although perhaps not very appl icable to present-day a r t production or c r i t i c i s m , the various schemes do exemplify to a great extent how a r t i s t s t r ad i t i o na l l y have approached the subject of t o l o r , and many of h is prescr ipt ions fo r "good* hue, value and chroma schemes are s t i l l quite acceptable as such to the lay person. I t 1s par t ly with the general acceptab i l i ty of these schemes in mind, and par t ly because i t seems reasonable to assume that a r t i s t s t r ad i t i ona l l y have not been en t i re l y wrong in the develop-ment of the i r co lor schemes for representational p i c tures , that the fol lowing deta i l s of the construction of the displays re ly heavi ly on these time-honored methods of handling hues, values and chromas. Composition of displays Hue Since hue was the most prominent of the Independent v a r i a b l e s -hue i s probably that dimension which most often Is equated with " c o l o r "— i t was desirable to tes t as wide a var iety of hues as poss ib le . At the same time, the hues had to be chosen in such a way that the value and chroma var iat ions wi th in these hues could be dupl icated exact ly with in the Munsell co lor space. A f te r a number of t r i a l s , a pattern centered around the hues 5BG, 5PB, 5RP, 5YR and 5YG 106. was se t t l ed on. This se lect ion provided a systematic and compre-hensive sampling of the hue c i r c l e . The Munsell space was s u f f i -c i en t l y f i l l e d with samples in these f i ve hue planes, and there were a l so su f f i c i en t colors ava i lab le to use the complementaries of these f i ve major hues: 5R, 5Y, 56, 5B and 5P. One of the major prerequis i tes of the s t imul i was that they be " r e a l i s t i c " as wel l as complex—i.e. , bear reasonably close resemblance to works of a r t , co lor photographs or natural scenery, as well as consist of a r e l a t i v e l y large number of colors—and i t was therefore decided to construct them in such a way that they not only contained the spec i f ied hue (the 5BG for instance) but a lso had some color chips of the complementary or minor hue (5R in th i s case).as wel l as some of the adjacent hues (5B and 5P here). The resu l t was that the par t i cu la r d isplay mentioned as an example would be designated as "predominantly 5BG" rather than as simply "5BG". Figure 26 shows graphica l ly the se lect ion of major, comple-mentary and adjacent hues. The choice of th i s procedure had both drawbacks and advan-tages. Displays could not be designated as s ing le hues as mentioned. This resulted in considerably more complex co lor imetr ic measurements and ca l cu la t ions . Furthermore, the responses to these complex st imul i could not be compared d i r ec t l y to those of the co lor preference l i t e r a t u r e . On the pos i t ive s ide , the displays could t ru ly be claimed 107. So l i d l ines are major hues. Dotted l ines are complementary as wel l as adjacent hues. FIGURE 26 Select ion of major, adjacent and complementary hues used in the d i sp lays . 108. to be more r e a l i s t i c , i . e . , more c lose ly resembling co lor pictures or nature, at least in the i r co lor composition, since they were both complex and embodied a dominant color scheme. In assigning the proportion of the 256 color chips to major, complementary and adjacent hues, i t was assumed that the type of t rad i t i ona l a r t i s t i c interpretat ion mentioned had some structure with in I t whifich could be u t i l i z e d in the construction of the d isp lays . This assumption made i t possible to introduce a mathematical r e l a -t ionship into the se lec t i on . The "golden mean" re lat ionsh ip had enjoyed widespread use, at least among t rad i t i ona l a r t i s t s ( c f . Fechner, 1876; Arnheim, 1966; Graves, 1951; Granger, 1955; Valent ine, 1962), and 1t was judged that th i s measure could be used as at least a s ta r t i ng point for the construction of the d isp lays . A number of tes t d isp lays, using the golden mean proportion (1.618:1 or 1:1.618 = .618), were constructed. I t was soon found, however, that certa in types of co lor groupings were better than others. For Instance, i f the 256 chips were divided among major and complementary hues according to the 1.618:1 measure, the resu l t was that there were f a r too few major hues (160) in re la t i on to complementaries (96). In other words, the preponderance of comple-mentary hues was so pronounced that the major hue did not stand out very w e l l . Th is , of course, ran counter to the intent ion in the present study of placing the most emphasis on the major hue. 109. I t was f i n a l l y decided to group color chips according to major conceptual d i f ferences, and to use only an approximation to the golden mean?9and then only as a s ta r t i ng po int . Figure 27 shows the -four major groupings of co lor chips according to con-ceptual d i f ferences. In th i s f igure , M 1s the major hue, Al and A2 are the two hues adjacent to the major, and C 1s the comple-mentary to the major hue. The f igures in brackets ind icate the number of co lor chips which went into each of the four groups, and the resu l t ing proportions of these groups are as fo l lows: (1) M / M + Al + A2 + C = 0.672 (2) M / Al. + A2 + C - 2.048 (3) M// C = 7.167. Value and chroma To keep the number of displays with in manageable l i m i t s , i t was decided that two leve ls of value and two (levels of chroma were su f f i c i en t to obtain an adequate var ia t ion of value and chroma. These would be termed "high value - low value" and "high chroma -low chroma", respect ive ly . In a manner s im i l a r to that used in the se lec t ion of hues, these high-low leve ls would not contain only high and low values (or chromas), but would be "predominantly high" or '•predominihtly low" values and chromas. The a r t i s t i c terms for n o . M * Major hue Al and A2 = Adjacent hues C s Complementary hue FIGURE 27 Graphic representation of proportion of major, adjacent and complementary co lor elements in the d isp lays . 111. these types of d isplay would be " l ight-dark" or "high key - low key" in the case of value, and "high saturat ion or in tens i ty - low saturat ion or in tens i ty" in the case of chroma. In order fo r the already selected major, adjacent and complementary hues to be grouped in a fashion that re f lec ted these value and chroma d i s t i n c t i ons , a specia l process of co lor chip se lect ion had to be resorted to . Since some values and some chromas in the Munsell space are not as ava i lab le in some hue areas as in others, the finajTlpattern seledted had to He f a i r l y close to the neutral axis (chroma 6 being as far away from the axis as the colors could go ) . The disadvantage of th i s procedure was that the displays were not quite as "codorful" ( i . e . , as highly saturated or intense) as perhaps could have been hoped f o r . A f te r a number of t r i a l s , the pattern shown in f igure 28 was se lected. This f igure shows the pattern of values and chromas for hue plane 5BG-5R only; the other four hue planes would have a s im i l a r value and chroma s t ructure . A l l the colors thus selected were ava i lab le from the Munsell Color Company with the exception of one co lo r , the 5GY 2/4. This was made up by the wr i te r to v i sua l l y occupy the pos i t ion in the Munsell space of the missing co lo r . Its chromaticlty coordinates are recorded in table 21. A f te r the addit ion of the one missing co lor ch ip, the selected pattern could be rotated around the neutral ax i s , covering a l l of the f i ve hues. 112. axis FIGURE 28 Basic pattern of major, adjacent and complementary hues in display number 1. 113. TABLE 19 L i s t of the 70 colors used in the displays together with the i r Munsell notations and perceived temperature.* ;olor Munsell Perceived Color Munsel1 Perceived no. notation temperature no. notation temperature 1 5BG7/6 11 36 5YR2/4 2 2 5BG5/6 11 37 5YR7/2 4 3 5BG3/6 11 38 5YR5/2 4 4 5BG8/4 10 39 5YR3/2 4 5 5BG7/4 10 40 5B7/4 10 6 5BG5/4 10 41 5B5/4 10 7 5BG3/4 10 42 5B3/4 10 8 5BG2/4 10 43 5RP7/6 3 9 5BG7/2 8 44 5RP5/6 3 10 5BG5/2 8 45 5RP3/6 3 11 5BG3/2 8 46 5RP8/4 4 12 5R7/4 2 47 5RP7/4 4 13 5R5/4 2 48 5RP5/4 4 14 5R3/4 2 49 5RP3/4 4 15 5GY7/6 6 50 5RP2/4 4 16 5GY5/6 6 51 5RP7/2 5 17 5GY3/6 6 52 5RP5/2 5 18 5GY8/4 6 53 5RP3/2 5 19 5GY7/4 6 54 5G7/4 8 20 5GY5/4 6 55 5G5/4 8 21 5GY3/4 6 56 5G3/4 8 22 5GY2/4 6 57 5PB7/6 9 23 5GY7/2 6 58 5PB5/6 9 24 5GY5/2 6 59 5PB3/6 9 25 5GY3/2 6 60 5PB8/4 8 26 5P7/4 6 61 5PB7/4 8 27 5P5/4 6 62 5PB5/4 8 28 5P3/4 6 63 5PB3/4 8 29 5YR7/6 1 64 5PB2/4 8 30 5YR5/6 1 65 5PB7/2 7 31 5YR3/6 1 66 5PB5/2 7 32 5YR8/4 2 67 5PB3/2 7 33 5YR7/4 2 68 5Y7/4 4 34 5YR5/4 2 69 5Y5/4 4 35 5YR3/4 2 70 5Y3/4 4 *See p. 155 for a f u l l explanation of perceived temperature. 114. Referring to f igure 29, the color numbers which would appear in the high value, high chroma 4BG-5R display were: 1, 2, 4, 5, 6, 8, 11, 12, 13, 14 in addit ion to 54, 55, 56 and 40, 41, 42 of the adjacent hues. The high value, low chroma display contained: 4, 5, 6, 9, 10, 3, 8, 12, 13, 14 $ i i i addit ion to 54, 55, 56 and 40, 41, 42 of the adjacent hues. This 1s shown 1n f igure 29 (b)). The low value, high chroma d isp lay, shown 1n f igure 29 ( c ) , contained: 2, 3, 6, 7, 8, 4, 9, 12, 13, 14 and 54, 55, 56, 40, 41, 42 of the adjacents. The low value, low chroma d isp lay , shown in f igure 29 (d) , contained: 6, 7, 8, 10, 11, 1, 4, 12, 13, 14 and adjacent hues 54, 55, 56, 40, 41, 42. There were 14 colors used in each of the f i ve hue planes, making a grand tota l of 70 color ch ips. These were numbered in con-secutive order fo r each hue (as shown in f igure 28), s ta r t i ng with hue 5BG and moving clockwise through the hue c i r c l e . Table 19 l i s t s the 70 colors used and the i r Munsell notat ions. Motif I t was suspected that the way the co lor chips were arranged— i . e . , the pattern they made—was important to the outcome of the responses to co lor . In addit ion to the already mentioned var iables of hue, value and chroma, four "motifs" were therefore selected for tes t ing: a face, a landscape, bui ld ings and an abstract. The object of th i s se lect ion was to arr ive at as varied a range of motifs 115. Majors Complementaries Majors Complementarles I 4 5 O9 | # 1 2 - O 2 6 " 1 0 30 O7 p u y 13 14 (a) O (b) High value, high chroma O O o (c) Low value, high chroma High value, low chroma O O V (d) Low value, low chroma FIGURE 29 Value and chroma patterns in the major-complementary hue plane fo r displays no. 1, 2, 3 and 4. 116. as poss ib le . This pa r t i cu la r se lect ion included: 1. A human motif (a face) 2. A natural motif (a landscape) 3. Man-made structures (bui ld ings) 4. A non-representational motif (an abstract) This conten£/no-content d i s t i n c t i on was of interest because through i t an assessment could be made of the Influence 6~f a representational motif (a face, landscape or bui ld ings) versus the inf luence of a decidedly non-representational, abstract mot i f . I t was hoped that th i s d i s t i n c t i on was c lear enough to throw l i g h t on the possibe inf luence 6f a cognit ive factor (recognit ion) which might be confounded with the emotional dependent variables of pleasure, arousal and dominance. A l l four motifs were constructed in a l l combinations of the 5 hues, 2 values and 2 chromas. Table 20 shows which colors were used (numbered and notated according to Munsell across the top) , the number of co lor chips used ( in the body of the tab l e ) , and which display numbers (shown at the l e f t ) contained the indicated co lors . The four columns at the l e f t furthermore indicate the motifs involved: face (column 1 ) , landscape (column 2 ) , bui ld ings (column 4 ) . Thus, face no. 1, landscape no 21, bui ld ings no 41 and abstract no. 61 TABLE 20 Number and co lor of components of the 80 d isp lays . Munsell notat ion and c o l o r nunber M*jor hue Display nunber >OU)lO<r««««NNN lOUIf l«»«»»NNN_ « I O « « « t « V N N N lfl<OU*«W»«NNN_ ^U>0«*»««NNN 1 2 3 4 5 6 7 8 9 11 12 13 14 15 16 17 18 19 20 21 22 2 3 24 25 2 6 27 2 8 29 30 31 32 33 34 3 5 36 37 3 8 3 9 40 41 42 4 3 44 4 5 46 47 48 49 50 51 52 53 5 4 5 5 56 5 7 5 8 5 9 60 61 62 63 64 I 1 21 41 61 32 32 32 32 32 6 6 8 8 8 10 10 10 10 10 10 5BG 3 22 42 62 32 32 6 32 32 32 6 8 8 8 10 10 10 10 10 10 23 43 61 6 32 32 32 6 32 32 8 8 8 10 10 10 10 10 10 4 24 44 64 6 6 32 32 32 32 32 8 8 8 10 10 10 10 10 10 5 25 45 65 6 26 46 66 7 27 47 67 8 28 48 68 32 32 32 6 6 8 8 3 2 6 32 32 32 6 8 8 6 32 32 32 6 32 32 8 8 6 3 2 3 2 3 2 3 2 3 2 8 8 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 9 29 49 69 10 30 50 70 11 31 51 71 12 32 52 72 10 10 10 10 10 10 10 10 10 10 10 10 32 32 32 32 32 6 32 32 6 32 32 32 6 6 32 32 32 6 32 32 6 6 32 32 32 32 I 8 8 8 8 8 8 10 10 10 10 10 10 10 10 10 10 10 10 13 33 53 73 14 34 54 74 15 35 55 75 16 36 56 76 17 37 57 77 I S 38 5B 78 19 39 59 79 20 40 60 80 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 32 32 32 32 32 6 32 32 6 32 32 32 6 32 32 32 6 : 6 6 32 32 32 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 32 32 32 32 3c 32 32 6 32 32 32 6 5 6 6 32 Si 32 32 32 32 118. a l l contained exact ly the same number and kind of co lor ch ips, namely 32 of colors no. 1, 2, 4, 5, 6; 6 of colors no. 8 and 11; 8 of colors no. 12, 13, 14 and 10 of colors no. 40, 41, 42, 54, 55 and 56. Once the sets of 256 co lor chips had been assigned to the 80 d isp lays, the motifs were constructed. Certain co lor schemes natura l ly lent themselves better to par t i cu la r motifs than others ( i . e . , certa in colors made pa r t i cu l a r motifs more natural look ing), so the face was constructed using the co lor chips assigned to face no. 9 (5YR-5B), the landscape was landscape no.25 (5GY-5P), the bui ld ings were d isplay no. 41 J5BG-5R), and the abstract , no,77 (5PB-5Y). A f ter the motif patterns had been estab l i shed, they were rep l icated in a l l the f i ve hue, two value and two chroma combinations as described e a r l i e r . The actual "form" of the four motifs was arr ived at through a t r i a l and er ror method, and depended to a large extent upon the wr i t e r ' s a r t i s t i c judgment. C r i t e r i a fo r the con-st ruct ion were the prerequis i te of a form (or f igure) which stood out enough against the background to be recognizable without too much d i f f i c u l t y , as wel l as the time-honored pr inc ip les of strong form and balance (Arnheim, 1974). In the case of the abstract, co lor chips were simply scattered across the surface and then rearranged to»avoid c lusters of s im i l a r chips or obvious patterns. A f ter com-p l e t i on , the abstract displays were Informally tested on a number of ind iv idua ls to ensure that no d iscern ib le patterns ex i s ted . 119. Figures 30, 31, 32 and 33 show the complete set of the 80 d isp lays . The C L E , system of co lor spec i f i ca t i on The Munsell spec i f i ca t ion of the displays described so f a r was one which allowed fo r a maximum of la t i tude in the se lect ion of colors and in the construction of the d i sp lays . I t was found to be pa r t i c u l a r l y useful because i t could express the gross conceptual ideas embodied in the s t i m u l i . For psychologists, how-ever, who were more Interested in the workings of the v isua l system and in spec i fy ing s t imul i in very precise terms than 1n spec i fy ing colors fo r everyday s i tuat ions the way Munsell had done i t , a system which was more c lose ly re lated to v isual phenomena was needed. Kflnig and Abney had started invest igat ing observers' spectral s en s i t i v i t y functions as a prel iminary step toward the development of a universal co lor spec i f i ca t ion system, and when Wright's measures of the la te 'twenties were combined with Gui ld 's spectral s e n s i t i v i t y curves, the resu l t was the establishment of the Commission Inter-nationale de 1'Eel ai rage ( C L E . ) 1931 equal-energy d i s t r i bu t i on curves fo r a 2° viewing f i e l d (Wright, 1969). The underlying assump-t ion for th i s work was that any co lor could be matched by a su i tab le mixture of three "primary" co lors , and f igure 34 shows the curves for these pr imaries: red ( x ) , green (y) and blue ( z ) . S t i l e s l a t e r 120. 17. 5PB HV/HC 18. 5PB LV/HC 3. 5BG HV/LC 4. 5BG LV/LC 7. 5GY HV/LC 8. 5GY LV/LC 11. 5YR HV/LC 12. 5YR LV/LC 15. 5RP HV/LC 16. 5RP LV/LC 19. 5PB HV/LC 20. 5PB LV/LC FIGURE 30 Displays nos. 1-20 with face as mot i f . HV - high value, LV • low value, HC • high chroma, LC • low chroma. 121. 29. 5YR HV/HC 30. 5YR LV/HC 31. 5YR HV/LC 32. 5YR LV/LC 33. 5RP HV/HC 35. 5RP HV/LC 36. 5RP LV/LC 37. 5PB HV/HC 38. 5PB LV/HC 39. 5PB HV/LC 40. 5PB LV/LC FIGURE 31 Displays nos. 21-40 with landscape as motif, HV • high value, LV • low value, HC • high chroma, LC • low chroma. 122, rr-— till*!-* »n Ml}:! : :m»n • i JS i i 41. 5BG HV/HC 42. 5BG LV/HC 4 3 . 5BG HV/LC 44. 5BG LV/LC 46. 5GY LV/HC 47. 5GY HV/LC 48. 5GY LV/LC 49. 5YR HV/HC 50. 5YR LV/HC 51. 5YR HV/LC 52. 5YR LV/LC 53. 5RP HV/HC 54. 5RP LV/HC 55. 5RP HV/LC 56. 5RP LV/LC 57. 5PB HV/HC 58. 5PB LV/HC 59. 5PB HV/LC 60. 5PB LV/LC FIGURE 32 Displays nos. 41-60 with bui ld ings as mot i f . HV • high value, LV • low value, HC • high chroma, LC * low chroma. 123. 77. 5PB HV/HC FIGURE 33 78. 5PB LV/HC 79. 5PB HV/LC 80. 5PB LV/LC Displays nos. 61-80 with an abstract mot i f . HV - high value, LV • low value, HC • high chroma, LC » low chroma. 124. 400 500 600 700nm Wavelength FIGURE 34 C L E . 1931 equal energy d i s t r i bu t i on curves. 700nm Wavelength FIGURE 35 Tr ist imulus values f o r 2° and 10° f i e l d s i z e s . 125. invest igated the tr ichromatic matching functions for a 10° viewing f i e l d , and as a resu l t , a revised set of C L E . curves was pro-duced in 1964 (Wright, 1969). For pract i ca l purposes, however, there i s not a great deal of di f ference between the two sets of curves as f igure 35 shows. The o r i g ina l curves were based on color matching data, and they involved some negative coordinates ind icat ing " . . . that desaturation of the spectral colours was necessary before they could be matched by a pos i t ive mixture of the instrumental s t i m u l i . " (Wright, 1969, p. 104) In 1931, the C L E . decided that , for p rac t i ca l .reaScirts.ij an a l l - po s i t i v e system would be a more appropriate system, and the o r ig ina l red, green, blue coordinate system was transferred to the present X, Y, Z system. In addit ion to being an a l l - po s i t i v e system, the X and Z values were chosen i n such a way that y ( in f igure 34) was equal to the standard observer's photometric curve V .^ This means that , as the X, Y, Z system i s used in i t s two-dimensional form (see f igure 36 where x • X/X+Y+Z and y = Y/X+Y+Z), the Y value expresses d i r e c t l y the luminosity value in per cent of a given reference white. To show graphica l ly the ca lcu la t ion of the t r i s t imu lus spec i f i ca t ion of a colored surface, the fol lowing example i s ins t ruc-t i v e . (Wright, 1969) To begin w i th , one determines the spectral r e f l e c t i on curve of the surface (see f igure 37). This can be done by 126. 0.8 0.6 0.4 0.2 \jj>40 0 boo \j580 \600 \620 .^770 m n 480«! 47oV 0.0 0,2 0.4 0.6 0.8 FIGURE 36 The C L E . 1964 chromaticity diagram. 127. FIGURE 37 Graphic i l l u s t r a t i o n of der ivat ion of t r i s t imulus values for a colored surface. 128. using a colorimeter ( e . g . , a Zeiss colorimeter which photo-e l e c t r i c a l l y measures re f l ec t i on values through a rotat ing f i l t e r wheel at 13 nm interva ls across the spectrum from 400 to 700 nm), a spectrophotometer, or a Lovibond Tintometer (which, using f i l t e r s , establ ishes quant i tat ive measures by means of v isua l matches). A spectral ref lectance curve from the colored surface might look something l i k e (a) in f igure 37. Next, the energy d i s t r i bu t i on of the i l luminant which f a l l s on the surface i s measured. This i s important since the surface co lor w i l l be a function of the co lor of l i gh t f a l l i n g on i t . Figure 38 shows the energy d i s t r i bu t i on curves of a number of common i l luminahts . In the present case, the energy curve might look something l i k e (£) in f igure 38. Since the curve begins to reach a maximum around 450 nm, one may assume that the i l luminat ion might be close to sun l ight . The next step i s to introduce the s en s i t i v i t y curves of the C L E . standard observer and to re late them to the i l luminant in f igure 37.(b) . The product of t h i s (b) curve and the t r i s t imu lus curves i s shown in f igure 37 ( c ) . F i n a l l y , by re la t ing the curve of the surface (a) to those of the standard observer + the i l luminant ( c ) , one can a r r i ve at the curves fo r the t r i s t imulus values for the colored surface shown in f igure 37 (d) . 129. 400 450 500 550 600 650 700nm Wavelength FIGURE 38 Typical spectral energy d i s t r i bu t i on curves fo r a var ie ty of i l luminants . The curves are based on re l a t i ve energies and bear no re la t ion to each other. 130. Having arr ived at these X, Y and Z values one can now calcu late the x and y coordinates fo r p lo t t i ng in the C L E . diagram. This p lo t 1s shown 1n f igure 39. (See also Lakowski, 1969, fo r the procedure out l ined so f a r . ) In the way of a summary of the procedure, the fol lowing information can be extracted from the diagram in f igure 39: 1. The diagram shows the pos i t ion of the i l luminant (C) . 2. The co lor l oc i of the surface (x, y) i s shown located in pos i t ion (S) . 3. One can Hcead of f" the dominant wavelength ( X c ) of the surface co lor by drawing a s t ra ight l i ne from (C) through (S) un t i l i t intersects the spectrum locus. The approx-imate dominant wavelengthe( A n ) i s i n th i s case 480 nm. This means that the predominant co lor of the sample 1s blue. 4. The complementary co lor to the one which was tested (Xq) can be approximated by continuing the l i ne from (S) through (C) un t i l i t intersects the locus on the opposite s ide . In th i s case, the complementary color would probably be a greenish-yel low. The procedure fo r determining the complementary co lor here i s only approximate, and more w i l l be sa id about th i s problem 131. FIGURE 39 P lo t of co lor l o c i (x,y) In the C L E . chromat idty diagram. 132. in the next sect ion. 5. The saturat ion ( i . e . , the spectral pur i ty of the co lo r , P e ) i s indicated by the proportion a/b as shown. In th i s case the color i s quite saturated. 6. The Y component of the t r i s t imu lus equation indicates the luminance fac tor , and thus i s re lated to the l ightness of the surface. This i s indicated by the value of Y, which i s Indicated in per cent of a standard white such as MgtV,. Judging from (a) i n f igure 37, the aver-age l ightness might be around 65%. Color imter ic measurements of ind iv idua l d isplay components. The 70 co lor chips which went to make up the 80 displays were measured by means of a Zeiss RFC-3 color imeter. The aperture was 5 mm, and a gloss trap was used. (See Lakowski and Sharpe, 1977, fo r a deta i led descr ipt ion of the performance of th i s instrument.) The t r i s t imulus values and the chromat idty coordinates, i n re la t ion to i l luminant"C", are l i s t e d i n table 21. A graphic representation of these colors i s shown in the C L E . diagram, f igure 40. In add i t ion, the dominant wavelengths ( Ap) and the exc i ta t ion or co lor imetr i c pur i t i e s (P e ) were calculated using the procedure described by Judd (1933). These are also shown in table 21. (The measurements were a l l made with a specia l mask, described in deta i l i n chapter VI , super-imposed over the co lors .) TABLE 21 Colorimetric specif ications of 70 colors used in tha displays. c Indicates complementary wavelength of non-spectral stimulus co lor . Color AD pe Color AD Pe 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 27.178 11.795 4.378 39.515 28.666 13.351 4.738 3.362 30.511 14.938 6.239 36.668 18.298 8.136 28.568 13.380 5.049 41.994 29.823 13.899 5.602 4.005 31.043 14.964 5.679 35.969 17.221 6.974 38.028 18.237 7.663 50.829 36.743 16.419 6.973 35.531 15.969 6.141 47.812 34.936 16.504 5.964 4.018 34.515 16.797 6.956 33.361 16.052 6.824 34.099 16.368 6.270 47.585 34.040 16.090 6.501 4.559 33.712 16.342 6.203 34.188 15.861 6.103 34.061 15.935 6.413 48.045 34.490 15.111 6.162 42.563 20.316 8.829 56.000 41.162 20.476 8.132 5.526 39.862 20.095 8.783 32.508 15.452 6.729 16.939 7.647 3.320 31.439 22.702 10.569 4.569 3.356 29.158 14.692 5.735 48.788 24.414 10.963 20.051 8.581 3.851 36.493 26.520 10.607 4.442 .2581 .2453 .2262 .2756 .2736 .2652 .2515 .2605 .2908 .2882 .2838 .3576 .3674 .3751 .3588 .3577 .3446 .3470 .3445 .3426 .3359 .3360 .3305 .3253 .3223 .3024 .2995 .2901 .4127 .4265 .4274 .3754 .3758 .3896 .3966 .3375 .3321 .3174 .3335 .3334 .3279 .3166 .3113 .3290 .3240 .3164 .3253 .3223 .3146 .4283 .4377 .4282 .3932 .3932 .3967 .3899 .3824 .3589 .3552 .3521 .2874 .2758 .2538 .3696 .3727 .3577 .3549 .3528 .3586 .3505 .18 .23 .31 .12 495.90 493.90 491.12 496.99 496.64 .12 494.08 .15 491.03 488.51 499.08 .06 495.19 .08 491.01 .10 599.38 .15 604.97 .17 620.01 .17 566.93 .43 565.83 .45 563.42 .39 568.05 567.25 566.14 564.68 .27 565.82 .25 568.03 .17 565.79 .15 564.38 .13 555.73c.11 555.73c.16 555.73c.25 586.83 .42 587.79 .46 591.98 .42 584.95 .28 585.87 .27 586.49 .33 590.44 .32 .22 .19 .30 .30 .30 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 3.960 34.270 16.378 6.944 29.868 13.908 5.624 40.322 19.703 8.528 50.550 37.488 17.826 8.049 5.221 34.469 16.680 6.982 28.255 13.347 5.277 29.613 15.838 6.900 38.241 33.414 16.452 7.209 5.990 33.567 16.430 7.316 33.375 16.353 7.289 3.467 33.768 15.977 6.598 34.064 16.069 6.566 34.256 15.999 6.419 46.951 34.202 15.864 6.813 4.523 33.653 15.818 6.388 34.357 16.477 6.561 30.650 16.165 6.901 40.233 34.274 16.990 7.292 6.093 34.539 16.480 7.449 34.794 16.873 7.422 2.932 31.819 15.187 6.436 48.285 24.556 11.421 40.292 19.874 8.941 54.829 39.849 19.417 8.867 6.219 38.867 18.910 8.228 33.106 16.262 6.869 57.630 31.425 15.858 63.321 52.527 27.709 13.525 10.407 44.736 23.246 11.247 21.692 10.365 5.305 .3822 .3431 .3444 .3475 .2661 .2550 .2381 .3510 .3545 .3570 .3318 .3361 .3356 .3391 .3270 .3221 .3244 .3232 .2951 .2896 .2821 .2511 .2497 .2326 .2696 .2779 .2701 .2572 .2663 .2974 .2907 .2812 .3714 .3751 .3641 .3346 .3381 .3360 .3302 .3035 .2946 .2780 .2982 .2878 .2687 .3082 .3066 .2987 .2871 .2833 .3145 .3076 .2957 .3589 .3575 .3507 .2599 .2548 .2326 .2837 .2851 .2774 .2601 .2709 .3060 .2979 .2863 .3872 596.04 .24 583.46 .15 585.38 .14 591.31 .14 485.73 .27 485.73 .22 483.47 .30 495.07c.14 497.83c.19 502.23c.27 495.57c.07 495.07c.09 497.15c.11 502,23c.17 515.53c.16 698.00c.03 497.15c.06  510.50c.ll 525.07 .08 517.65 .08 506.90 .09 478.45 .28 475.65 475.65 480.68 .18 475.65 .15 475.65 475.65 475.65 480.69 478.66 475.65 .14 575.61 .35 ,28 .38 .18 .26 .21 .06 .09 .3870 576.46 .36 .3707 577.52 .32 134. I «59 .23 -| 1 1 1 1 r 1 1 I i r I .21 .23 .25 .27 .29 .31 x .33 .35 .37 .39 .41 .43 FIGURE 40 70 colors used in the displays p lotted in the C L E . chromaticity diagram. Insert i l l u s t r a t i o n shows locat ion of p lo t in overa l l C L E . diagram. 135. II SPECIFICATION OF THE COLOR SURFACE In an e f f o r t to examine in greater de ta i l the spec i f i ca t ion problem associated with the content and d i s t r i bu t i on of colors in the d isp lays , a var iety of ca lcu lat ions and analyses were per-formed. The general problem, as i t was stated 1n chapter I I , could be divided into three parts: Component spec i f i ca t ion A spec i f i ca t i on of the propert ies of the ind iv idua l co lor elements making up the d isp lay . Quantity spec i f i ca t ion An assessment of the quantity of these elements in each d isp lay . D is t r ibut ion spec i f i ca t ion The locat ion or d i s t r i bu t i on across the display surface of these co lor elements. Component spec i f i ca t i on The se r i a l l i s t i n g of co lor elements which was shown e a r l i e r in th i s chapter, or which for example was used by Harmon (1973) in his studies of feature recognit ion in faces, i s typ ica l of spec i -f i ca t ions of th i s f i r s t type. The numerical values l i s t e d may be 136. in the Munsell notation or they may be C L E . t r i s t imu lus values or chromaticity coordinates. In a l l cases, any one of these l i s t i n g s 1s su f f i c i en t to enable a subsequent reconstruction of the st imulus, provided that the pattern of s e r i a l i z a t i on i s also ind icated. That i s , as long as the number of items per " l i n e " 1n the se r i a l i z ed s t r i ng of items i s known. Quantity spec i f i ca t i on In order to go one step further and account for the quantity of co lor elements used in a d i sp lay , a se r i a l type of l i s t i n g i s again necessary. Table 22 shows, fo r instance, two ways of present-ing the hue content of display no. 1, while tables 23 and 24 show the value and chroma content of high/low value displays and high/low chroma d isp lays . In addit ion to the s e r i a l l i s t i n g of these quant i t i es , an ind icat ion of per cent content may be l i s t e d , or a graphic repre-sentation may be used fo r ease of in te rpre ta t ion . Thus the s izes of the areas shown e a r l i e r in f igure 27 are proportionate to the number of elements belonging to them, and the l a s t two tables may be converted into bar graphs as shown in f igure 41. Although no new information 1s added in the process of con-vert ing the data to diagrams, the graphic representations do o f f e r an a l ternat ive method of spec i fy ing the quantity of co lor elements in d i sp lays . 137. TABLE 22 Quantity spec i f i ca t ion of hue elements in d isplay no. 1. Number of elements Hue spec i f i ca t i on Percent of major, adjacents and contrasts 32 32 32 32 32 10 10 10 10 10 10 6 6 5BG7/6 5BG5/6 5BG8/4 5BG7/4 5BG5/4 5B7/4 5B5/4 5B3/4 5G7/4 5G5/4 5G3/4 5BG2/4 5BG3/2 Major (Major + adjacents) 90.62% CO OO 00 5R7/4 5R5/4 5R3/4 Contrasts 9.38% 256 100% Hue Number % area 5BG Major 172 67.19 5R Contrasts 24 9.38 5B Adjacents 30 11.72 5G Adjacents 30 11.72 138. TABLE 23 Quantity spec i f i ca t ion of value elements in high and low value d isp lays . High value displays Number of Value Percent elements spec i f i ca t i on of to ta l 6 2 2.34 34 3 13.28 92 5 35.94 92 7 35.94 32 8 12.50 256 100.00 Low value displays Number of Value Percent elements spec i f i ca t ion of to ta l 32 2 12.50 92 3 35.94 92 5 35.94 34 7 13.28 6 8 2.34 256 100.00 139. TABLE 24 Quantity spec i f i ca t ion of chroma elements in high and low chroma d isp lays . High chroma displays Number of Chroma Percent elements spec i f i ca t ion of to ta l 6 2 2.34 186 4 72.66 64 6 25.00 256 100.00 Low chroma displays Number of Chroma Percent elements spec i f i ca t ion of tota l 64 2 25.00 186 4 72.66 6 6 2.34 256 100.00 uo. Value r i 2 3 5 7 8 High value displays 2 3 5 7 8 Low value displays Chroma 2 4 6 High chroma displays 2 4 6 Low chroma displays FIGURE 41 Visual representation of quantity spec i f i ca t i on of value and chroma content of high and low d isp lays . Dist r ibut ion spec i f i ca t i on Depending on what prec ise ly i s meant by the surface d i s t r i b u -t ion of co lo rs , there are a number of ways in which the require-ments of the t h i r d spec i f i ca t ion c r i t e r i on may be met. Ea r l i e r in th i s chapter, the so lut ion to the problem was to speci fy the d i s t r i bu t i on of colors as making up a face, landscape, a bu i ld ing • i or an abstract . This i s a quite legit imate answer s ince , had the colors been arranged d i f f e r en t l y , they would not have produced these mot i fs . Somehow, however, t h i s type of spec i f i c a t i on , s ince i t i s mainly based on an overa l l v isua l impression, lacks the prec is ion which a more mathematical formulation would provide. Another way to specify the d i s t r i bu t i on of colors i s to simply l i s t the co lor numbers (according to the l i s t of 70 colors shown in table 20)-!in an array such as the one shown in f igure 42. This method i s very precise in that the numbers In the array can provide the basis on which to reconstruct the exact display of which i t i s a record- However, i t lacks the featurewwhlch dist inguished the previous method in that i t cannot t e l l much about what perceptual ly i s contained in the display. Some of the v isual material contained in the numerical d isp lay ( f igure 42) can be extracted fn ways other than through reconstruc-t ion of the complete co lor d i sp lay . For instance, alphameric charac-ters or specia l graphic symbols may be inserted in to the display instead 142. 01 01 05 40 55 06 42 42 06 06 02 06 02 41 41 40 01 05 06 06 06 56 56 06 42 42 42 06 02 02 05 40 05 05 02 06 04 04 04 05 01 01 06 06 02 02 05 05 05 06 06 01 04 04 04 04 01 01 05 06 02 02 06 02 05 55 55 04 04 04 04 04 01 01 05 06 02 06 06 05 01 55 01 06 56 06 55 02 55 56 56 02 05 05 12 40 01 06 02 11 56 56 04 02 56 56 11 02 41 41 40 40 12 13 06 05 05 05 04 55 04 04 04 01 41 41 40 40 55 12 01 01 05 04 04 01 04 04 01 01 02 41 41 12 54 55 06 01 05 04 04 04 01 04 01 04 02 41 13 12 54 55 08 06 11 02 02 06 01 01 01 02 41 12 13 40 54 05 08 04 56 42 11 42 11 01 02 02 42 05 13 40 54 06 06 08 04 01 04 04 02 02 42 42 02 05 05 05 05 01 06 13 06 04 04 06 08 08 08 02 02 02 05 05 01 01 13 14 11 06 14 14 12 12 54 54 02 05 05 05 01 14 14 14 14 13 14 13 54 54 54 01 54 02 02 05 FIGURE 42 Color element numbers in display number 1 (face, BG-R, high value, high chroma). 143. of the numbers. Thus the symbol M may be subst i tuted for a l l * "co lors" of the major hue, in the numerical d isplay (while the contrast ing hues may be l e f t blank) and a v isua l representation of major/contrasting hue d i s t r i bu t i on across the d isp lay surface obtained. Figure 43 shows the four motifs displayed in th i s manner. Other p o s s i b i l i t i e s according to th i s method include v isua l rep-resentations of high value/low value and high chroma/low chroma. Figures 44, 45, 46 and 47 show these p o s s i b i l i t i e s . This procedure has the advantages of both high accuracy and v isua l reference to the d isp lay, but i t lacks conciseness. In summary, these seem to be the three c r i t e r i a for an accptable spec i f i ca t ion of the co lor d i s t r i bu t i on in a d isp lay: 1. Reference to what i s seen in the d isp lay, 2. Mathematical accuracy, and 3. Conciseness of formulat ion. E a r l i e r , the spec i f i ca t ion of "what was in the d isp lay" was brought to the point where proportions of spec i f i c hues (major and complementary), spec i f i c values (high and low) and spec i f i c chromas (high and low) were s tated. Referring now to the above c r i t e r i a , i t may be possible to carry th i s spec i f i ca t i on one step further by averaging the proportions for hue, value and chroma. 144. MMMMMM MM MMMMMMMMMM MMMMMMMMMM MM MMMMMMMM MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM MM MMMMMMMMMMMMMMMMMMMMMMMMMM MMMMMMf f MM MM MMMMMM MMMMMMMM MMMM MMMM MMMMMMMMMM MMMMMMMM MMMMMMMMMMMMMMMMMMMMMM MMMMMMMMMMMMMMMMMMMMMM MMMMMMMMMMMMMMMMMMMM MMMMMM MM MMMMMMMM MM MMMMMMMMMMMMMMMMMM MMMMMMMM MMMMMM MMMMMMMMMMMMMMMMMMMMMMMM MMMM MMMM MMMMMMMM MM MM MMMMMM Face MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM MM MMMMMM MMMMMMMM MMMMMMMMMM MMMMMMMM MMMMMMMMMM MMMMMM MMMMMMMMMMMMMM MMMMMMMMMMMMMMMMMM MMMMMMMMMMMMMMMM MMMMMMMMMMMMMMMM MMMMMMMMMMMMMMMM MM MMMMMMMMMM MMMM MMMM MMMMMMMMMMMM MMMMMMMM MMMMMMMMMMMMMM MMMMMMMM MMMMMMMMMMMMMMMM MMMMMMMM Bui ldings MMMMMMMMMMMM MMMMMMMMMMMMMMMMMM MMMMMMMMMMMM MMMMMMMMMMMMMM MMMM MMMMMMMMMMMMMMMM MMMMMMMM MMMMMMMMMMMM MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM MMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMM MM MMMMMMMMMMMMMMMMMMMMMMMMMMMM MMMM MMMMMMMMMMMMMMMMMMMMMMMM MMMM MMMMMMMMMMMMMMMMMMMMMM MMMM MMMMMM MMMMMMMMMMMMMMMM MMMMMM MM MM MMMMMM MMMM MMMMMM MMMMMMMM MMMMMM MMMMMM MM MMMMMM MM MMMM MMMM MMMM MMMMMM Landscape MMMMMM MMMMMM MMMM MMMMMM MM MMMM MM MMMM MMMMMMMMMM MMMM MMMM MMMM MMMM MM MMMMMMMM MMMM MMMM MM MMMMMM MMMMMMMM MMMMMMMMMMMM MM MMMMMM MM MMMM MMMM MM MMMMMMMM MMMM MMMMMMMMMMMM MM MM MMMMMMMM MMMM MMMMMM MMMMMMMM MMMMMM MMMMMMMMMMMMMM MMMMMMMM MM MMMM MM MM MM MM MM MMMMMMMMMMMMMM MMMM MM MMMMMMMMMMMM MMMMMM MMMMMMMMMM MMMMMM MMMMMM MM MMMM MM MM MMMMMM MMMM MM MMMMMM MMMMMM MMMM MMMM MMMMMMMMMM MMMM MMMM Abstract FIGURE 43 Visual representation of major versus adjacent and complementary hues. M = major hues; blank • adjacent or complementary hues. 145. ::::111IMMMMI1111111111111 IIIIIIMMMMIIMMMMMMIIIIII:: I I I I : : : : : : : : : : : : I I I I I I I I : : I I I I : : : : : : : : : : : : : : : : I I I I I I I I I I I I I I : : : : : : : : : : : : : : : : I I I I I I I I II::IIMMIIIIIIIIMMMMII:::: IIIIMMMMMM::IIMMMMMMIIIIII I I I I : : : : : : : : I I : : : : : : : : I I I I I I : : : : : : : : : : : : : : : : : : : : : : I I I I I I I I I : : : : : : : : : : : : : : : : : : I I I I I I IIMMIIMMIIIIII::::::IIII::II ::MM::MMMMMMMMMM::1111MM::11 IIIIMM::::::::IIIIMMMMII:: ::IIIIII::::IIMMMMMMIIIIII ::IIMMMMIIMMMM::II::::II:: MMMMMMMMIIMMII:::::::::: I I I I Face I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I II I I I I I I I I I I : : I I I I I I IIIIIIMMIIIIII :IIMMII::II IIMM::II ::MMII ::MM:: II II I I I I MMMMMM MMMMIIMMMMMMI11111111IMMMMMMIIMM MMMMMMIIMMMMMMMMMMMMIIMMIIIIMMII I I I I I I I I I I I I :: I I I I I IMM II I I I I I I I I III111IMMI111 IIIIIIIIIIIMM :IIIIMMMMMMMM Landscape I I I I I I I I I I : : : I I I I I I : : : ::MMMMMMMMMMII::::IIIIII: MMMMMMMMIIIIIIIIIIIIIIII: MM::::::II II II I I I I I I I I I I I I I I IIMMMMMM MMMMMMMMIIIIIIIIII MMMMMMMMIIIIIIIIII MMMMMMMMIIIIIIIIII MMMMMMMMIIIIIIIIII I I I I I I I I I I : : : : : : : : I I I I : : : : I I : : : : : : : : I I I I : : : : : : : : I I I I : : I I I I : : : : : : : : I I I I I I I I I I I I : : : : : : : : I I I I I I :IIIIIIIIMMMM::MMIIMMMM MM::MMIIIIMM MMMMIIIIII MMMMIIIIII MMMMIIIIII Bui ldings I I : I I : I I : • • • ::MM ••ITTI*•••TT« * • • • MM::::IIIIMM::II: II::MM MMII MM:: I I : : ::MMIIMM: ::IIMMII: IIMM::II:: : : I I I I : : : : II::MMIIII MMIIII::MM I I : : : : I I I I : : : : I I I I : : IIII::::MMIIII::MMII MM::IIMMIIMMMM::II::::II::II:::: I I I I : : : : : : : : : : I IMM::I I : : : : MM::II::MMII::::::II::::II ::IIMMIIIIIII I I I: II I I : : MM:: : :II ii::MM::II::IIMM IIIIIIMM::II::II ::::::::IIMMII:: MMIIMM I I : : I I : : :::II : I I I I ::::MMII: I I I I : : I I : MM: rilMMMMIIII :IIIIMMII :::::::MM :IIIIMMII : l i s sII • i i i i IIMM ::II Abstract FIGURE 44 High value motifs represented by three leve ls of value. ":" =values 7 - 8 ; " I "* value 5; "M" = values 2 - 3 . 146. MMMMMMIsIIIIMMMMIIIIIIIIIIIIII:: MMMMIIIIIIMMMMIIMMMMMMIIIIIIMM:: MMMMI111MMMMMMMMMMMM11111111MMMM MMIIIIMMMMMMMMMMMMMMMMIIIIIIIIII MMI111MMMMMMMMMMMMMMMM1111111IMM MMIIMMIIMMIIIIIIIIMMMMIIMMMM:::: MMIIII::MMMMMMIIMMMM::IIIIII:::: ::IIIIMMMMMMMMIIMMMMMMMMIIII:::: II::MMMMMMMMMMMMMMMMMMMMIIIIII:: ::1111MMMMMMMMMMMMMMMMMM111111:: :: IIIIIIIHIKIMMMMMMI I I I : I I I : : ::MM::MMMMMM::MM::MMIIIIMMMMII:: ::IIII::MMMMMMMMIIIIMMMMIIMMMMMM MMMMIIIIIIMMMMII::::::IIIIIIMMMM MMMMIIMM::IIMMMM::::::::IIMMMMMM MMMMMMMMMMIIMMII::::::MM::IIIIMM Face I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I : : I I I I I I I I I I I I MMMMMMMMMMMMMM::::::::IIII::MMMM MMMMMMMMMMMMMMMM::IIIIMMMMMMMMMM ::::::::::::::::::::MMMMMMMMMMMM MMMMMMIIIIMMMMMMMMMMMMMMMMMMMMMM MMI1111111MMMMMMMMMMMMMMMMMMMMMM II::IIIIIIMMMMMMMMMMMMMMMMMMMMMM IIIIIIMMIIIIIIMMMMMMMMMMMMMMMMMM MMI IMMI I:: IIIiMMMMMMMMMMI I MMMMMMMM MMMMIIMMMMIIMM::IIMMI1111111MMMM MMMMMMMMII::::::IIIIIIIIIIMMIIII MMMMMMMM::::::IIIIIIIIIIIIIIII:: ::::::MM::::::::::::IIIIMMMM:::: MMMMIIMMMMMMI111111111MMMMMMII:: MMMMMMIIMMMMMMMMMMMMIIMMIIII::II Landscape MMMMMMMMMMMMIIMMMMMMMMMMMMMMMMMM MMMMMMMMIIIIIIIIIIMMMMMMMMMMMMMM MMMMMMMMMMIIIIIIMMMMMMMMMMMMMMMM. MMMMMMMMMMMMIIMMMMMMMMMMMMMMMMMM MMMMMMMMMMMMIIMMMMIIIIIIMMMMMMMM MMMMMMMMI11111111111111IMMMMMMMM MM::::::IIMMMMIIIIIIIIII::MMMMMM ::::::IIMMMMMMIIIIMMMMII:::::::: ::::::IIMMMMMMIIIIMMMMMMMMIIII:: ::11111IMMMMMMI11IMMMMMM::111111 IIIIIIIIMMMMIIIIIIMMMMMM::IIIIII IIMMMMMMMMIIIIIIII::::MMMMIIMMMM MMMMMMMMIIIIIIIIII::::MMMMIIIIMM MMMMMM::IIIIIIIIII::::::::IIIIII MMMM::::IIIIIIIIII::::::::IIIIII MM:: : : : : I I I I I I I I I I : : : : : : : : I I I I I I Bui ldings MM::MMIIII::MMIIMM::MMIIMMMMIIMM II::MMMMMMI11IMMMMII::MMI111MMMM IIMMII::MMMMMMMMIIMMMMII::MMI111 II::MM::IIMMIIII::IIMM::IIIIMMMM MMMMMMMM::MMIIMMMMIIIIMMMMIIIIMM MMMMIIIIMMMMIIII::MM::IIIIMMMMII MMMMII::IIMMMMMMII::MMIIMMIIMMMM IIIIMM::MMMMMMIIMMMMII::MMMMII:: ::MMIIMMMMIIMMMM::IIMMMMIIMMMM:: MMII::IIIIIIIIIIMMMMMMIIMMMMIIII IIMM::MMMM::MMMMIIMMMMMMMMIIMMII ::MM::MM::111IMMIIMMIIMM::MMMMMM MMIIMMMMMMMMMMII::MMIIIIMMIIIIMM IIMMMM::II::IIMM::IIII::II::MMMM IIIIIIMMMMIIMMIIMMMMMM::MMMMIIMM ::MMMM::II..IIMMMMIIII::II::MMII Abstract FIGURE 45 Low value motifs represented by three leve ls of value. ":" = values 77- 8; " I " = value 5; "M" = values 2 - 3 . 147. MMMMIIIIIIIIIIIIIIIIMMIIMMIIII MMIIIIIIIIIIIIIIIIIIIIIIMMMMII 111IMMI11111111IMMMMI11IMMMMII IIIIIIMMIIIIIIIIMMMMIIIIMMMMIIMM IIIIIIIIIIIIIIIIMMMMIIIIMMIIII MMIIMMIIIIIIIIMMIIIIIIMMI11111 MMIIMM::IIIIIIMMIIII::MMIIIIII I I I I I I I I I I I I I I I I I I I I I IMMI I I I I I 111IMMMMI1111IMMI11IMMMMMMI111 IIIIIIMMIIIIIIIIMMIIMMIIMMIIII IIIIIIII::MMMMIIMMMMMMMMIIIIII IIIIIIIIIIII::II::MMMMMMMMIIII 111111111IMMI11IMMMMI11IMMI111 IIMMIIIIIIIIIIIIIIIIIIMMMMMMII MMMMIIII::1111111111111IMMI 111 MMIIIIIIIIIIIIIIIIIIIIMMIIMMMM Face I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I MMMMMMMMMMMMMMIIIIIIIIIIIIIIIIII I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I IMMMMIIII I I I I I I I I I I I I I I I I I I IIMMMMMMMMIIIIIIIIIIIIIIIIIIIIII 111IMMIIMMMMMMI11111111111111111 11111111MMMMMMMMMMMMMMMMMMMMMMMM IIMMIIIIMMMMMMMMMMMMMMMMMMMMMMMM 1111111IMMMMMMI1111IMMMMMMMMIIMM MMIIIIIIMMIIIIIIIIIIMMIIMMIIMMMM MMI111111111111IMMMMMMMMIIMMMMII : : I I : : I I : : I I : : : : : : I I I I I I I I I I I I I I I I I I I I I I I I I I I I I IMMI I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I IMMI I I I Landscape MMMMMMMM MMMM IIIIMMMMMMMMMMMM IIMMMMMMII I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I MMIIIIIIIIII MMIIMMMMIIII MMI I MMMMMMMM MMIIMMMMMMII MMMMIIMMMMMMII MMMMMMII::::MMII MMMMMMMMIIII::MMII MMMMMMMMIIIIIIII:: MMMMMMMMIIIIIIII:: MMMMMMMMIIIIIIII:: Bui ldings MM MMMM MMMM MMMM M I I I I I I I I I I I I I I I IMMI I I I I I I I I I I I I I IIIIMMIIIIIIMMIIIIIIIIMMIIIIIIII I I I I I I I I I I I I I IMMIIIIMM::III I I I I I MMI 11111111 IMMMMI 11 IMMU'MMMMI 111 MMMMMMMMMMI111111111111IMMMMMMM IIIIII::IIIIIIMMMMIIMMIIMMI111II 111IMMI1111IIIMMIIIIIIIIIIIIIIII IIMMMMIIIIMMMMIIIIMMMMMMIIIIMMMM MMII::MMIIIIIIIIMMIIIIMMIIIIIIII IIMMIIIIMMIIMMIIIIIIIIMMIIMMIIMM IIIIIIIIIIIIIIIIIIMMIIIIIIMMIIMM MMIIIIIIIIIIIIMMIIIIIIIIIIMMIIMM IIIIIIIIMMIIMMIIIIIIMM::IIIIMMMM IIIIMMIIMMMMMMIIIIIIIIIIIIIIIIII IIIIMMIIII::II111IMMII::MMIIIIII Abstract FIGURE 46 High chroma motifs represented by three leve ls of chroma. ":" = chroma 2; " I " = chroma 4; "M" = chroma 6. 148. : I I I I I i : I I I I I I I I : : i I : : I I I I ] [ 1111:: 11 [II I I I I I I] I I : : I I I I ] II::MMIi: ; I I I I I I I I ; [ I I : : : : I i : [IIII::II] [ I I I I I IMM: I I I I I I I I ; I I I I I I I I : : I I I I I i : ::IIIIMM] I I I I I I I I ] I I I I I I I I : s i l m i n i u m I I I I : : : : I I I I I I I I : : : : I I I I I I I I : : : : I I I I I I : : I I I I I I : : I I : : IIIIMM:: I I I I I I I I I I : : I I I III I I I I : : I I ..» T T » • • » * i A • • • • MMIIMM:: I I I I : : : : I I I I I I I I I I I I I I : : I I I I I I I I I I I I I I I I I I I I I I : : II I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I • • • • •• T T T T • • * • • • • • • • • • > i i i i I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I : : I I : : II 111111 : : I I I I I I I I I I I I I I I I : I I I I I I I I r l l l l l l l l : : : : : I I I I I: : : : : : I I I I r l l l l l l l l I I I I I I I I I I I I I I : : MMIIMMIIMMIIMMMMMM I I I I I I I I I I I I I I : : I I I I I I I I I I I I I I I I : 1111 n i [m i i n R I I I I I I I [ i i i I I i i : I I I I I I I [ I I I I I I I [ I I I I I I I I I : : : : [I::::II [ I I I I I I I [ I I I I I I I : i : : I I I I Face Landscape I I I I I I I I I I : : : : : : : : I I I I I I I I : : : : : : I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I : : I I I I I I I I I I I I I I : : : : I I I I I I I I : : : : : : : : I I I I : : : : : : I I I I I I : : : : : : I I I I IIMMMM::IIII IIIIMM::IIII IIIIIIIIMMII IIIIIIIIMMII IIIIIIIIMMII: : : I I I I I I I I I I I I I I I I I I I I : : I I I I I I I I I I I I I I : : I I : : I I I I MMIIIIII : I I I I I I I I •TTTT*••• • i l l ! • • • • MM::IIIIII I I I : : I I : : I I I I I I I I I I I I I I I I I I I I I : : I I : : : I I : : : : : : :IIIIMMII I I I I I : : I I I I : : I : : I I I I I I I I I I I I I I I : : I I I I I I III::MMIIIIII I I I : : I I : : : : I I I I I I I I I : : : : I I : I I : : I I : : I I I I I I I I I I I I I I I I I I : : : : : : I I I I : : : I I I I : : I I I I I I I I I I I : : I I : : I I I s : I I I I I I : :II I I I I I I I I I : : I I III::MMIIII:: I I I I I I I I I I I I I I::IIMM::IIII Bui ldings FIGURE 47 Abstract Low chroma motifs represented by three leve ls of chroma. ":" » chroma 2; " I " = chroma 4; "M" = chroma 6. 149. Following th is l i ne of (thought, these types of ca lcu la t ion may now be made: 1. Hue. The c losest to the idea of an average hue for a d isp lay 1s the spec i f i ca t i on of the major Munsell co lo r . Adjacent and complementary hues would techn ica l ly have to modify the major hue, but since the 100-step scale on the hue c i r c l e does not have a meaningful zero po int , th i s type of ca lcu la t ion cannot be done. Thus, "predominant hue" describes most accurately the preponderance of one pa r t i c -u lar (major) hue. 2. Value. The Munsell dimension of value, on the other hand, can be treated quite d i f f e r en t l y . Ea r l i e r i t was shown that high value displays contained 6 elements of value 6, 34 of value 3, 92 of value 5, 92 of value 7 and 32 of value 8. ( c f . table 23) By mult ip ly ing the values by the number of co lor elements, summing and d iv id ing by 256 (the to ta l number of co lor elements), the average value of 5.76 i s obtained. S imi lar ca lcu lat ions for the low value display produces the average of 4.24. This di f ference between high and low value displays i s h ighly s i gn i f i c an t : with of 88.96 and 4 degrees of freedom, the probab i l i ty i s less than .0001. The value averages fo r a l l the displays are l i s t e d 1n table 25. 3. Chroma. Calculat ions s im i l a r to those above, but fo r chroma, 150. show that the average fo r a high chroma display of 6 elements of chroma 2, 156 of chroma 4 and 64 of chroma 6 y i e l d a f igure of 4.45. ( c f . table 24) For the low chroma d isp lay, th i s f igure was 3.55. This di f ference between high and low chroma displays i s also highly s i gn i f i c an t . With X ? of 96.12, and 2 degrees of freedom the probab i l i ty i s less than .0001 (c f . table 25). These averages cer ta in ly f i l l the c r i t e r i a of mathematical accuracy and conciseness of formulat ion, but they f a l l short of the c r i -ter ion of v isua l reference. In the case of value, fo r Instance, to State the average value as 5.$6 i s s im i l a r to suggesting that i f a l l the co lor elements had been of th i s value, the d isplay would have e l i c i t e d the same responses as the o r ig ina l d i sp lay . The main problem with a statement of th i s kind i s that there i s no empir ical evidence presently ava i lab le to suggest t h i s . On the other hand, i t might be argued that the spec i f i ca t i on of an average value does not necessar i ly pretend to re fer to any perceptual f a c t s . Rather, the ca lculat ions themselves may be con-ceptual ly sound, and the generation of the hypothesis, f o r instance, that responses may be the same, i s the primary purpose of the present study. Furthermore, the v i a b i l i t y of these hypotheses may be Increased by regressing the averages against the c r i t e r i on variables of pleasure, arousal , dominance and Information rate as out l ined in chapter I I I . The conceptual j u s t i f i c a t i o n for the 151. ca lcu lat ion procedures was based on the fact that a r t i s t s do ta lk about overa l l co lor features of a p ic ture: t he i r ent i re terminology ( I . e . , majors, minors, predominant, etc . ) re fer to gross charac-te r i za t ions of the d i s t r i bu t i on of colors across the picture surface. Although the averaging procedures did re fer conceptually to ways of speaking about the picture surface, they dealt with the overa l l p icture surface rather than with features with in I t . The c r i t e r i on of d i s t r i bu t i on mentioned e a r l i e r was therefore only p a r t i a l l y met. In order to a t ta in more s p e c i f i c i t y , a number of ana ly t i c procedures fo r deal ing with features wi th in the p icture surface were considered. Since many of these procedures deal with d i f f e r -ences of co lor between various parts with in the p icture surface, a b r i e f account of a colorlmetry method of handling co lor d i f f e r -ences w i l l be given. The Judd-Hunter co lor di f ference formula The prec is ion of the C L E . system for co lor spec i f i c a t i on ! 1s c l ea r l y an advantage over other systems such as the Munsell nota-t i o n . However, as fa r as describing what colors actua l ly look l i k e , the C L E . system i s iisimited to " . . . a predict ion of whether two given co lor s t imul i of d i f fe rent spectral power d i s t r ibut ions w i l l 152. be perceived to have the same co lo r . " (Wyszecki, 1972) In other words, the C L E . co lor space i s based on a mathematical model and only to a l im i ted extent does i t r e f l e c t what a co lor in th i s space w i l l look l i k e . This problem i s compounded when attempts are made toddescribe di f ferences between two co lors . As a r e su l t , various formulae for changing the C L E . co lor space into a space with equal perceptual steps fo r a l l three dimensions of co lor ( i . e . , hue, saturat ion and brightness) have been proposed (c f . Judd and Wyszecki, 1963). The formula used in the present study i s one developed by Judd and Hunter (Hunter, 1942). I t i s described in deta i l in Color in  Business, Science, and Industry. (Judd and Wyszecki, 1963). I ts formulat ion, as adapted for computer use, 1s: Al = fg ((221 ( Y a ) ) 1 / 4 ( dE 2 + d D ) 1 / 2 ) 2 + (10T) 2) ^ 2 where: Al = co lor di f ference expressed 1n NBS units fg = Ya/(Ya + 2.5) 221 * sca l ing factor to adjust fo r s i ze of NBS unit Y a = ( Y l + Y 2 ) / 2 T = (Y1)V2 - (Y2)V2 E = (.511x + 1.2447y - O.57O8)/0<+ 2.2633y + 1.1054) D - (2.4266X - 1.3631y - 0.3214/(x + 2.2633y + 1.1054) 10 = factor re la t ing in importance the l ightness saturat ion and hue sca les . In th is case i t i s adjusted for surface color samples, x = X/(X + Y + Z) y = Y/(X + Y + Z) 153. As w i l l be seen from th is formula, once the chromaticity coordinates (x, y) or the t r i s t imu lus values (X, Y, Z) of the two samples are known, the color di f ference (AE) can be ca lcu la ted. Although the co lor space which resu l ts from the formula i s non-euclidian (Judd and Uyszecki, 1963), i t does to a large extent produce a space with equal percep-tual steps fo r the three dimensions of co lo r . Furthermore, once a co lor di f ference has been calculated for two co lo rs , i t can be a lgebra ica l l y manipulated in conjunction with other color d i f f e r -ences since the space i s uniform. A measure of co lor di f ferences with in displays One feature of the displays was that major, adjacent and contrast ing hues made up groups of co lor elements which were c l ea r l y d is t ingu ishab le. Conceptually, the displays could also be viewed as composed of major hues.(which would include the adjacent hues as wel l ) and minor hues. Since majors were located on one side of the i l luminant in the C L E . space, and minors on the other, an average of theetr ist imulus values fo r these two groupings could be obtained. In add i t ion, the co lor di f ference between majors and minors in each d isp lay could be ca lcu lated, using the Judd-Hunter color di f ference formula, and a type of d i s t r i bu t i on spec i f i ca t i on arr ived a t . Table 25 l i s t s the t r i s t imu lus Values, chromaticity coordinates and the AE values for majors and minors in the d isp lays , 154. TABLE 25 Color imetr ic spec i f i c a t i on f o r majors and minors in d isp lays and t h e i r average <4E, value, chroma and temperature measurements. Majors Minors Display no. X Y z X Y z 1 21 41 6 1 21.0314 26. 0033 31.3472 21.3330 18. 9849 13. 4722 2 22 42 62 11. 1626 13. 7643 17.2854 21.3330 18. 9049 13. 4722 3 23 43 63 21.8765 25. 9619 30.9454 21.3030 18. 9849 13. 4722 4 2'4 44 64 11. 7669 14. 0170 17.318S 21.33 30 13. 9849 13. 4722 5 25 45 65 22.321 1 25. 7332 16.5608 20.0545 13. 7172 23. 3549 6 26 46 66 12.1354 13. 9864 9.6399 20.0546 18. 7172 28. 0549 7 27 47 67 22.8647 25.6779 19.1554 20.0546 18. 7172 28. 0549 8 28 43 68 12. 3765 13. 983 1 10.6291 20.0546 18. 7172 28. 0549 9 29 49 69 27.5974 25. 6252 18.3442 16. 5715 19. 0193 23. 2706 10 30 50 70 14. 8284 13. 6730 9.9799 16. 571 5 19. 0193 23. 2706 11 31 51 71 26.8412 25. 5358 20.8113 16.5715 19. 0190 23. 2706 12 32 52 72 14.4451 13. 5957 10. 8703 16. 5715 19. 0193 28. 2706 13 33 53 73 28.5428 25.4693 30.4244 15.5252 19. 1315 13. 7456 14 34 54 74 15.7284 13. 3027 17. 1714 15. 6262 19. 1315 13. 7456 15 35 55 75 27.3 591 25. 3622 30.1131 15.6252 19. 1315 18- 7456 16 36 56 76 15.2498 13. 7891 16.9772 15.6252 19. 1315 18. 7456 17 37 57 77 23.4367 24. 2233 39.8351 19.0055 19. 6962 12. 4539 18 38 53 78 13.8678 14. 2214 23. 79 82 19.0355 19. 6962 12. 4539 19 39 59 79 24. 0 561 24. 846 1 37.0123 19.3055 19. 6962 12. 4539 20 40 60 80 13.9070 14. 2925 22.3324 19.0055 19. 6962 12. 4539 Majors 1 Minors Average x y x y AE Value Chroma Temperature 0.2683 0.3318 0. 3625 0. 3 231 49. 4253 5. 7573 4. 4531 9. 2188 0. 2644 0. 3261 0. 3625 3. 3231 44- 7400 4. 2422 4. 4531 9. 2188 0.2777 0. 3295 0. 3625 3. 3231 44. 1430 5. 7578 3. 5469 8. 5391 0.2730 0. 3252 0. 3625 3. 3231 40. 7220 4. 2422 3. 5469 3. 5391 0.3454 0. 3933 0. 3001 0. 2801 35. 2860 5. 7578 4. 4531 6 . 0000 0. 3393 0. 3911 0. 3001 0. 2801 29. 9343 4. 2422 4. 4531 6. 0000 0. 3377 0. 3793 0. 3001 0. 2801 30. 4600 5. 7578 3. 5469 6. 0030 0.3346 0. 3 73 0 0. 3001 0. 2301 26. 8333 4. 2422 3. 546 9 6. 0030 0.3856 0. 3581 0. 2595 3. 2973 53. 3180 5. 7573 4. 4 531 2. 7313 0.3853 0. 3553 0. 2595 0. 2973 47. 77 10 4. 2422 4. 4531 2. 7813 0. 3665 0. 3 49 3 0. 2 595 0. 2973 46. 0730 5. 7573 3. 5469 3. 4 60 9 0. 3712 0. 3494 0. 2595 3. 2 973 42. 9400 4- 2422 3. 5469 3. 4609 0.3 3 80 0. 3016 0. 2921 0- 3575 38. 8340 5. 7578 4. 4531 4. 1434 0.3368 0. 2955 0. 2921 3. 3575 36. 1950 4. 2422 4. 4531 4. 1434 0. 330 3 0. 3062 0. 2921 3. 3576 33. 850 3 5. 7573 3. 5469 4. 6016 0. 3314 0. 29 97 0. 2921 3. 3575 32. 8220 4. 2422 3. 5 46 9 4. 6016 0. 2673 0. 2769 0. 3715 0. 3350 43. 2550 5. 7578 4. 4531 7. 3516 0.2673 0. 2741 0.3715 0. 3853 40. 0340 4. 2422 4. 4531 7. 3516 0. 2800 0. 2892 0. ,3715 0.3350 37. 7763 5. 7578 3. 5469 7. 3934 0.2 75 2 0. 2328 0. 3 715 0. 3853 36. .5950 4. 2422 3. 5469 7-3934 155. and f igure 48 shows the chromaticity coordinates p lotted in the C L E . space for the f i r s t 20 d isp lays . As table 25 shows, displays 21-40, 41-60 and 61-80 had AE values ident i ca l to the f i r s t 20 d isp lays, since these four groups contained the same co lor elements, d isp lay fo r d i sp lay . A measure of perceived temperature of d isplay co lo rs . One of the co lor d i s t inc t i ons most frequently used by a r t i s t s i s that of warm and cool co lors , and i t i s general ly recognized that red hues for instance are warm and blue-green coo l . However, there i s general disagreement about the perceived co lor temperature of most other hues. Arnheim (1974) wr i tes: The pure fundamental primaries can hardly be ca l l ed e i the r warm or co ld . Is a pure red c l ea r l y more warm than a pure blue of equal saturation? Is a pure yel low cold or warm? But temperature qua l i ty seems to be more meaningful when applied to the admixture of a co lor . A b lu ish yel low or red tends to look co ld , and so does a ye l lowish red or b lue. On the contrary, a reddish yel low or blue seems warm. My suggestion i s that not the main co lor but the color to-ward which i t deviates may determine the e f f e c t . This would lead to the perhaps unexpected resu l t that a reddish blue looks warm whereas a b lu ish red looks co ld . (p. 369) Arnheim further refers to the work of Ifcten (1961) in which red-orange and blue-green are designated as opposing temperature poles. This approach has been followed in the present study, and hues in the Munsell co lor space have been assigned values according to t he i r 156. .41 -5 • .39 - • 6 ft 8 • Y .37 -• G ft .35 -11 12 10 .33 -1 • 3 • y *2 N " C " • © R .31 - 15 • .29 -i © B 19 e 1 4 .27 -17 * 2 ° ® P • • 18 .25 - I 1 1 1 1 1 1 .25 .27 .29 .31 x .33 ,35 .37 .39 .41 • Major colors e Minor colors FIGURE 48 Majors and minors of f i r s t 20 d i sp lays . 157. proximity to the warmest hue. Referring to f igure 49, i t can be seen how 5YR and 5R have the same temperature according to th i s scheme, and how 5BG and 5B have the same (and lowest) temperature. For the purpose of fur ther ana lys i s , a l l the 70 colors used in the displays were scored on a scale from 1 (warmest) to 11 (coo les t ) . Figure 49 further shows the scoring of these colors according to t he i r chroma. The center of the c i r c l e represents neutral (or zero) chroma, and the three rings surrounding the center represent chroma leve ls 2, 4 and 6. The complete l i s t i n g of the perceived temperature fo r these colors i s included in table 19, and the overa l l averages f o r the displays are shown in table 25. Four surface d iv i s i on schemes For ana ly t i ca l purposes, i t i s convenient to div ide a display surface into parts which conceptually make sense. Four surface d i v i s i on schemes were adopted, and the i r respective boundaries and areas were operat ional ly def ined. These were: 1. Figure/background d iv i s i on 2. Top/bottom balance 3. Le f t / r igh t balance 4. Average of small-area contrasts 158. FIGURE 49 Conceptual scheme fo r assigning perceived temperature values to Munsell colors used in the d isp lays . 159. Figure/background d iv i s ion From the time of the Gestalt psychologists ( e . g . , Katz, Kohler), th i s surface d iv i s i on has played a major ro le in perceptual experiments. I t has been claimed that the f igure tends to appear to l i e in front of the background, that a surrounding surface tended to be seen as a background, and that the r e l a t i v e l y smaller area in a picture tended to be seen as a f i gu re . I t i s also noticeable that the simpler area (often the convex area), the l i gh te r area and the one associated with the warm colors tended to be regarded as the f i gu re . (Arnheim, 1974) For the present purpose* the figure/background boundaries were defined by the experimenter fo r the face, landscape and bu i ld ings . The abstract motif did not, as was mentioned, have any d iscern ib le patterns or f igures in i t . Figure 50 shows the three motifs and the i r associated figure/background boundaries. The face conformed most c lose ly with the above observations about the ideal figure/background re la t ionsh ip . The landscape, on the other hand, s t i l l maintained the s i ze requirement but i t could be argued that the background was more convex than at least the bottom part of the f igure . The bui ldings motif occupied by f a r the largest area while i t , on the other hand, stood out strongly as an integra l f igure from the background. 160. FFFFFF FFFFFF FFFFFFFF FFFFFFFF FFFFFFFF FFFFFFFF FFFFFFFFFF FFFFFFFF FFFFFFFFFF FFFFFFFF FFFFFFFFFF FFFFFFFFFF FFFFFFFFFF FFFFFFFFFF FFFFFFFFFF FFFFFFFFFF FFFFFFFF FFFFFFFF FFFFFFFF FFFFFF FFFFFF FFFF FFFF FFFFFF FFFFFFFF FFFFFFFFFF FFFFFFFFFFFF FFFFFFFFFFFFFF Face FF FFFFFFFFFF FFFFFF FF FFFFFFFFFFFF F FFFFFFFFFFFFFFFFFFF FFFFFFFFFFFFFFFFFFF FFFFFFFFFFFFFFFFFFF FFFFFFFFFFFFFFFFFFF FFFFFFFFFFFFFFFFFFF FFFFFFFFFFFFFFFFFFF FFFFFFFFFFFFFFFFFFF FFFFFFFFFFFFFFFFFFF FFFFFFFFFFFFFFFFFFF FFFFFFFFFFFFFFFFFFF FFFFFFFFFFFFFFFFFFFI FFFFF FFFFF FFFFFFF FFFFFFFFFFFF! FFFFFFFFFFFF FFFFFFFFFFFFI FFFFFFFFFFFF FFFFFFFFFFFFI FFFFFFFFFFFF FFFFFFFFFFFF FFFFFFFFFFFFI FFFFFFFFFFFF Bui ldings FF FFFFFFFFFFFFFF FFFFFFFF FFFF FFFFFFFF FFFFFFFFFF FFFFFFFFFFFFFF FFFFFFFFFF FFFFFFFF FFFF FF FF FFFFFFFF FFFFFFFFFFFFFFFF FF FFFFFFFFFFFFFFFFFF FFFFFFFFFF FFFFFFFFFFFFFFFFFFFF FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF F a Flgure Background i s blank, Landscape FIGURE 50 Figure-background boundaries f o r 3 motifs. 161. 1. Color di f ference measurements. To obtain a measure of the co lor di f ference between the f igure and background fo r a l l the 80 d i sp lays , the average t r i s t imulus values fo r the major and minor hues were calculated for both f igure and background, and the co lor di f ferences between the major and minor for the f igure and the major and minor for the background found. Since these color dif ferences conceptually may be interpreted to indicate the level of d i spa r i t y , tension or contrast between major and minor hues, the proportion between the AE f o r the f igure and the AE f o r the background was taken to indicate the extent to which the f igure predominated over the background. A value greater than 1.0 would indicate that the f igure was predominant, a value less than 1.0 would indicate a lack of any predominance. Since, as mentioned, the abstract motif did not display any f i gu ra l pre-dominance, values of 1.0 were assigned to the d isplay numbers 61 to 80. Table 26 l i s t s the AE values fo r majors and minors, as wel l as the proportion AEM AE1 fo r a l l 80 d i sp lays . 2. Calculat ions of value, chroma and perceived temperature based  on the figure/background d i v i s i o n . As adjuncts to the overa l l average values, chromas and perceived temperatures already worked out and reported e a r l i e r , the average values, chromas and perctved temperatures fo r f igure and background were ca lcu lated. In add i t ion, the respective proportions of measures f o r f igure and background 162. TABLE 26 Figure-background proportions of AE and value fo r 80 d i sp lays . Figure Backgr. Display Figure Backgr. value value VI/V2 no. 2IE1 AEZ AE1/AEZ VI V2 1 53. 1510 49.3550 1. 0769 6, 0256 5. 5324 1. 0891 2 40. 1480 48.5720 0. 8249 3. 6325 4. 7554 0.7639 3 48. 241 0 4 3.7310 1. 101 9 6. 0256 5. 5324 1.0891 4 33. 0520 43.3550 0. 87 7 7 3. 6325 4.7554 0.7639 5 4 1. 2 53 0 33.6250 1. 2259 6. 0256 5.5324 1.0891 6 27. 6010 32.5810 0. 84 7 2 3. 6325 4. 7554 0.7639 7 36. 9440 28.6010 1. 291 7 6. 0256 5. 5324 1.0891 8 26. 5560 2d.2760 0. 9392 3. 6325 4.7554 0.7639 9 57. 5180 52.6390 1. 0917 6. 0256 5. 5324 1.089 1 10 4 3. 4530 51.5430 0. 84 3 3 3. 6325 4. 7554 0.7639 11 50. 797 0 45.1090 1. 1261 6. 0256 5. 5324 1.0891 12 41. 2670 44.8140 0. 92 0 9 3. 6325 4.7554 0.763 9 13 '4 0. 9120 40.4690 1. 0139 6. 0256 5.5324 1.0891 14 30. 7950 40.7230 0. 7562 3. 6325 4. 7554 0.7639 15 36. 4240 35.3130 1. 0315 6. 0256 5.5324 1.0891 16 29. 1230 36.1330 0. 836 1 3. 6325 4.7554 0. 7639 17 44. 3120 43.4130 1. 0322 6. 0256 5.5324 1.0891 18 35. 9320 43.6120 0. 8250 3. 63 25 4. 7554 0.763 9 19 3 9. 7890 3 7. 8790 1. 0534 6. 0256 5.5324 1.0891 20 34. 3870 33.7620 3. 837 1 3. 6325 4.7554 0.7639 21 4 9. 9010 4 9.2120 1. 0143 5. 8462 5.6835 1.0286 22 44. 0600 45.6570 0. 9550 4. 2906 4.2014 1.0212 23 43. 5180 44.8370 0. 9736 5. 8462 5. 6835 1.0286 24 39. 1320 42.3630 0.9237 4. 2906 4. 2014 1.0212 25 36. 0590 34.7130 1. 0333 5. 8462 5.6835 1.0286 26 29. 4 79 0 30.5710 0. 9643 4. 2906 4. 2014 1.0212 27 30. 3070 30. 69 10 0. 9875 5. 8462 5.6835 1.0286 28 26. 8970 29.4350 0. 9147 4. 2906 4.2014 1.0212 29 54. 4220 52.8960 1. 3283 5. 8462 5.6835 1.0286 30 46. 8450 48.3930 0. 9530 4. 2906 4. 2014 1.0212 31 45. 3340 4b.3530 0. 9576 5i 8462 5.6835 1.0286 32 41. 3030 44.6070 0. 9259 4. 290 6 4.2014 1.0212 33 39.1620 38.7450 1. 0103 5. 846 2 5.6835 1.0286 34 3 5. 4140 3 7.2190 0. 951 5 4. 2906 4.2014 1.0212 35 33. 1350 34.5960 0. 9592 5. 8462 5. 6835 1.0286 36 31. 2770 34. 4730 0. 9373 4. 2906 4.2014 1.0212 37 43. 8690 43.0210 1. 0197 5. , 8462 5. 6835 1.0286 38 39. 9500 43.5870 0.93 4 3 4. 290 6 4.2014 1.0212 39 3 7. 3740 38.5150 0. , 9704 5. 8462 5. 6835 1. 02 86 40 35. ,7280 37.8360 0. ,9443 4. , 2906 4.2014 1.0212 continued 163. Table 26 continued Figure Backgr. Display Figure Backgr. value value no. AE1 AEZ AIM Att. VI V2 V1/V2 41 4 7. 7380 51. 3080 0. 9334 5. 3661 6. 0625 0.8851 42 44. 4320 50. 1 860 0. 83 5 3 4. 3839 3. 7431 1. 3048 43 46. 0550 44. 7560 1. 0290 5. 3661 6. 0625 0. 885 1 44 42. 6260 43. 0700 0. 9397 4. 8839 3. 7431 1. 3048 45 34. 9060 3b. 5780 0. 954 3 5. 3661 6. 0625 0.8851 46 31. 3560 34. 4300 0. 9107 4. 8339 3. 7431 1.3048 47 31. 74 7 0 30. 5340 1. 03 97 5. 3661 6. 0625 0.885 1 48 29- 8890 29. 6330 1. 0386 4. 8839 3. 7431 1. 3048 49 50. 9140 55. 4050 0. 913 9 5. 3661 6. 0625 0.8851 50 4 7. 0 25 0 53. 8320 0. 3727 4. 3839 3. 7431 1. 3048 51 46. 0380 46. 3910 0. 9318 5. 36 61 6. 0625 0.8851 52 44. 7940 45.5860 0. 9326 4. 8839 3. 7431 1.3048 53 37. 2440 40. 5080 0. 9194 5. 3661 6. 0625 0.8851 54 34. 5740 42. 5490 0. 8126 4. 3339 3. 7431 1.3048 55 34. 0270 34. 2540 0. 9934 5. 3661 6. 0625 0. 885 1 56 33. 0430 36. 9150 0.8951 4. 8839 3. 7431 1. 3048 57 38. 4480 47. 0290 0. 8175 5. 3661 6. 0625 0;835 1 58 35. 8260 47. 5930 0. 7528 4. 8839 3. 7431 1. 3048 59 35. 1780 39. 9360 0. 8793 5. 3661 6. 0625 0.8851 60 34. 3770 41. 5860 0. 8266 4 . 8839 3. 743 1 1.304 8 61 i i 1. 0000 1.0000 62 1. 0000 1.0000 63 1. 0000 1.0000 64 1. 0000 1. 0000 65 1. 0000 1.0000 66 1. 0000 1.0000 67 1. 0030 1.0030 68 1. 0000 1.0000 69 1. 0000 1.0000 70 1. 0000 1.0000 71 1. 0000 1.0000 72 1. 0000 1.0000 73 1. 0000 1.0000 74 1. 0000 1.0000 75 1. 0000 1.0000 76 1. 0000 1.0000 77 1. 0000 1.0000 78 1. 0000 1.0000 79 1. 0030 1.0030 80 1. 0000 1.0000 164. were computed. Tables 26 and 27 show the averages fo r f igure and background as wel l as t he i r proportion fo r value, chroma and perceived temperature. ( I t i s noted that , again, the abstract mot i fs , number 61 to 80, have 1.0 assigned to the proport ions.) Top/bottom balance The balance between the top and the bottom halves of a p icture has always played an important ro le to a r t i s t s , since the fee l ing of balance in a p icture has to a large extent depended on whether i t was top or bottom "heavy". Valentine (1962) describes experiments in which bottom heavy diagrams are preferred to top heavy ones because of t he i r fee l ing of s t a b i l i t y , and Arnheim,(1974) points out how a p i c ture , to be seen as evenly balanced, must be constructed in such a way that the bottom part predominates. A picture which looks per fect ly balanced w i l l thus look top heavy i f turned upside down. The top/bottom boundary for the present displays was defined as the hor izontal l i ne d iv id ing the display Into two equal halves. 1. Color di f ference measurements. The top and bottom parts of the displays had the i r average t r i s t imu lus values fo r majorssand minors, AE f o r top and bottom par ts , and the proportion of the top AE and the bottom AE ca lcu lated, exact ly as described for the figure/background d i v i s i on above. The value of 1.0 fo r the 165. TABLE 27 Figure-background proportions of chroma and temperature for 80 d i sp lays . Figure Backgr. Figure Backgr. splay chroma chroma temp. temp. no. Cl C2 C1/C2 TI T2 T1/T2 1 4 . 4103 4 . 4892 0 . 9 8 2 4 9 . 0 4 2 7 9 . 3 6 6 9 0 . 9654 2 4 . 4 103 4- 4892 0 . 9 8 2 4 9 . 0 4 2 7 9 . 3 6 6 9 0 . 9654 3 3 . 5897 3 . 5108 1 . 0 2 2 5 8 . 4 2 7 4 8 . 6 3 3 1 0 . 9762 4 3 . 5897 3 . 5108 1 .0225 8 . 4 2 7 4 3 .6331 0 . 9762 5 4 . 4103 4- 4892 0 .9824 5 . 7 4 3 6 6 - 2 1 5 8 0 . 9240 6 4- 4103 4 . 4892 0 . 9 8 2 4 5 . 7 4 3 6 6 . 2 1 5 3 0 . 9240 - 3 . 5897 3 . 5108 1 .0225 5 . 7 4 3 6 6 . 2 1 5 8 0 . 9240 8 3 . 5897 3 . 5108 1 . 0 2 2 5 5 . 7 4 3 6 6 . 2 1 5 8 0 . 9240 9 4 . 4 103 4 . 4892 0 . 9 8 2 4 2 . 7 0 0 9 2 . 3 4 8 9 0 . 9481 10 4 . 4103 4 . 4892 0 . 9 8 2 4 2 . 7 0 0 9 2 . 8 4 8 9 0 . 9481 11 3- 5897 3 - 5108 1 .0225 3 . 3 1 6 2 3 . 5 8 2 7 0 . 9256 12 3 . 5897 3- 5108 1 .0225 3 . 3 1 6 2 3 . 5 8 2 7 0 . 9256 13 4 . 4103 4 . 4892 0 . 9 8 2 4 4 . 4 6 1 5 3 . 3 8 4 9 1- 1484 14 4 . 4 103 4 . 4892 0 . 9 8 2 4 4 . 4 6 1 5 3 . 3 8 4 9 1. 1484 15 5897 3 . 5108 1 . 0 2 2 5 4 . 8 7 1 8 4 . 3 7 4 1 1. 1138 16 3 . 5897 3 . 5108 1 .0225 4 . 8 7 1 8 4 . 3 7 4 1 1. 1138 17 4 . 4103 4 . 4892 0 . 9 8 2 4 8 . 0 5 1 3 7 . 6 9 0 6 1. 0469 18 4 . 4103 4 . 4892 0 . 9 8 2 4 8 . 0 5 1 3 7 . 6 9 0 6 1. 0469 19 3 . 5897; 3 . 5108 1 .0225 7 . 6 4 1 0 7 . 1 8 7 0 1. 0632 20 3 . 5897 3 . 5108 1 . 0 2 2 5 7 . 6 4 1 0 7 . 1870 1 . 0632 21 4 . 5470 4 . 3741 1 .0395 8 - 8 6 3 2 9 . 5 1 8 0 0 . 9 312 22 4 . 5470 4 . 3741 1 .0395 8 . 8 6 3 2 9 . 5 1 8 0 0 . 9312 23 3 , 4530 3 . 6259 0 . 9 5 2 3 8 . 0 4 2 7 8 . 9 5 6 3 0 . 8979 24 3 . 4530 3 . 6259 0 . 9 5 2 3 8 . 0 4 2 7 8 . 9 5 6 8 0 . 8979 25 4 . 5470 4 . 374 1 1 .0395 5 . 8 6 3 2 6 . 1 1 5 1 0 . 9583 26 4 . 5470 4- 3741 1 .0395 5 . 8 6 3 2 6* 1151 0 . 9588 27 3 . 4530 3. 6259 0 . 9 5 2 3 5 . 8 6 3 2 6 .1151 0 . 9588 28 3 . 4530 3 . 6259 0 . 9 5 2 3 5 . 3 6 3 2 6 .1151 0 . 9538 29 4 . 5470 4 . 3741 1 .0395 2 . 9 1 4 5 2 . 6 1 1 5 1 . 1160 30 4 . 5470 4 . 3741 1 .0395 3 . 9 1 4 5 2 . 6 1 1 5 1. 4989 31 3 . 4530 3 . ,6259 0 . 9 5 2 3 3 . 8 2 0 5 3 . 1 5 8 3 1. 20 97 32 3. 4530 3 . 6259 0 . 9 5 2 3 3 . 8 2 0 5 3 . 1 5 8 3 1. 20 97 33 4 . 5470 4 . 3741 1 .0395 4 . 3 7 6 1 3 . 9 5 6 8 1. , 1060 34 4 . 5470 4 . 3741 1 .0395 4 . 3 7 6 1 3 . 9 5 6 8 1. , 1060 35 3 . 4530 3 . 6259 0 . 9 5 2 3 4 . 9 2 3 1 4 . 3 3 0 9 1. 1367 36 3 . 4530 3 . 6259 0 . 9 5 2 3 4 . 9 2 3 1 4 . 3 3 0 9 1. ,1367 37 4. 5470 4 . ,3741 1 .0395 7 . 8 9 7 4 7 . 8 1 2 9 1 .0108 38 4 . 5470 4 . ,3741 1 .0395 7 . 8 9 7 4 7 . 8 1 2 9 1 .0103 39 3 . ,4530 3. .6259 0 . 9 5 2 3 7 . 3 5 0 4 7 .4101 0 . 9 9 1 9 40 3. .4530 3 . 6 2 5 9 0 . 9 5 2 3 7 . 3 5 0 4 7 .4101 0. , 9919 . . . continued 166. Table 27 continued Figure Backgr. Figure Backgr. Display chroma chroma temp. temp. no. Cl C2 C1/C2 TI T2 T1/T2 41 4 . 3214 4 . 5556 0 . 9 4 8 6 9 . 1250 9 . 2 9 1 7 0 . 9 8 2 1 42 4 . 3214 4 . 5556 0 . 9 4 8 6 9 . 1250 9 . 2 9 1 7 0 . 9 8 2 1 43 3. 6786 3. 4444 1 . 0 6 8 0 8 . 6 4 2 9 8 . 4 5 8 3 1 . 0 2 1 8 44 3 . 6786 3- 4444 1 . 0 6 8 0 8 . 6 4 2 9 8 . 4 5 8 3 •1.0218 45 4 . 3 2 1 4 4 . 5556 0 . 9 4 8 6 6 . 1786 5 . 8 6 1 1 1 . 0 5 4 2 46 4 . 3214 4 . 5556 0 . 9 4 8 6 6 . 1786 5 . 8 6 1 1 1 . 0 5 4 2 47 3. 6 7 8 6 3 . 4444 1 . 0 6 8 0 6 . 1786 5 . 8 6 1 1 1 . 0 5 4 2 48 3. 6786 3 . 4444 1 . 0 6 8 0 6 . 1786 5 . 3 6 1 1 1 . 0 5 42 49 4 . 3214 4 . 5556 0 . 9 4 8 6 3 . 0 5 3 6 2 . 6 0 4 2 1. 1726 50 4 . 3214 4 . 5556 0 . 9 4 8 6 3 . 0 5 3 6 2 . 6 0 4 2 1 . 1 7 2 6 51 3 . 6786 3 . 4444 1 , 0 6 8 0 3 . 5 3 5 7 3 . 4 0 2 8 1 . 0 3 9 1 52 3. 6786 3 . 4444 1 . 0 6 8 0 3 . 5 3 5 7 3 . 4 0 2 8 1 .0391 53 4 . 3 2 1 4 4 . 5556 0 . 9 4 8 6 4 . 0 5 3 6 4 . 2 2 2 2 0 . 9 6 0 1 54 4 . 3 2 1 4 4- 5556 0 . 9 4 8 6 4 . 0536 4 . 2 2 2 2 0 . 9 6 0 1 55 3. 6786 3 . 4444 1 . 0 6 8 0 4 . 3750 4 . 7 7 7 8 0 . 9 1 5 7 56 3. 6786 3. 4444 1 . 0 6 8 0 4 . 3 7 5 0 4 . 7 7 7 3 0 . 9 1 5 7 57 4 . 3 2 1 4 4 . 5556 0 . 9 4 8 6 7 . 5 8 9 3 8 . 0 5 5 6 0 . 9 4 2 1 58 4 . 3214 4 . 5556 0 . 9 4 8 6 7 . 5893 8 . 0 5 5 6 0 . 9 4 2 1 59 3 . 6786 3 . 4444 1 . 0 6 8 3 7 . 2 6 7 9 7 . 5 0 0 0 0 . 9 6 9 1 60 3 . 6786 3 . 4444 1 . 0 6 8 3 7 . 2 6 7 9 7 . 5 0 0 0 0 . 9 6 9 1 61 1 . 0 0 0 0 1 . 0 0 0 0 62 . i 1 . 0 0 0 0 1 . 0 0 0 0 63 1 . 0 0 0 0 1 . 0 0 0 0 64 1 . 0 0 0 0 1 . 0 0 0 0 65 1 . 0 0 0 0 1 . 0 0 0 0 66 1 . 0 0 0 0 1 . 0 3 0 0 6 7 1 . 0 0 0 0 1 . 0 0 0 0 68 1 . 0 0 0 0 1 . 0 0 0 0 69 1 . 0 0 0 0 1 . 0 0 0 0 70 1 . 0 0 0 0 1 . 0 0 0 0 71 1 . 0 0 0 0 1 . 0 0 0 0 72 1 . 0 0 0 0 1 . 0 0 0 0 73 1 . 0 0 0 0 1 . 0 0 0 0 74 1 . 0 0 0 0 1 . 0 0 0 0 75 1 . 0 0 0 3 1 . 0 0 0 0 76 1 . 0 0 0 0 1 . 0 0 0 0 77 1 . 0 0 0 0 1 . 0 0 0 0 78 1 . 0 0 0 0 1 . 0 0 0 0 79 1 . 0 0 0 0 1 . 0 0 0 0 80 1 . 0 0 0 0 1 . 0 0 0 0 167. proportion meant that the d isplay was equal ly balanced, a value greater than 1.0 that i t was top heavy, and a value less than 1.0 that i t was bottom heavy. Table 28 l i s t s the AE values for majors and minors, and the proportion /JE-j/zJEg for the 80 d isp lays . 2. Calculat ions of value, chroma and perceived temperature based  on the top/bottom d i v i s i o n . As in the case of the figure/background d i v i s i o n , the average values, chromas and temperatures for the top and bottom parts of d i sp lays , as wel l as the proportion of these values and chromas, were ca lcu lated. Tables 28 and 29 show these averages fo r both top and bottom par ts , and the resu l t ing proport ions. Le f t / r igh t balance Although not as important to the balance of a p icture as the top/bottom d i v i s i o n , the l e f t / r i g h t d i s t r i bu t i on s t i l l plays a large part in the construction of a p i c tu re . Arnheim (1974) notes how, i f both l e f t and r igh t parts of the picture are of equal "weight", the p icture w i l l seem to be right-heavy and, in order to remedy t h i s , i t i s the pract ice of a r t i s t s to emphasize the l e f t s ide to compensate for th i s imbalance. Figure 51 shows an example from Arnheim demon-s t ra t ing th i s po int . As in the previous d iv i s ion of the picture surface, the l e f t / r i g h t boundary was defined as the ve r t i ca l l i ne 1 6 8 . TABLE 28 Top-bottom proportions of AE and value f o r 80 d i sp lays . Top Bottom Display Top Bottom value value no. AE) A E2 AE}/ AE2 VI V2 V1/V2 1 49. 9370 49. 9920 0 .9989 5. 8047 5. 7109 1.0164 2 49. 6 190 44. 2280 1 .1219 4. 0 70 3 4, 4141 0.9 221 3 44. 9733 44. 0720 1 .0204 5. 8047 5. 7109 1.0164 4 45. 1470 40. 4470 1 . 1162 4. 0703 4. 4141 0.9221 5 34. 6930 36. 4343 0 .9521 5. 8047 5. 7109 1.0164 6 3 4. 0170 30. 1970 1 .1255 4. 0703 4. 4141 0.9221 . 7 33. 1103 31. 0653 0 .9693 5. 8047 5. 7109 1.0164 8 33. 84 9 0 27. 3393 1 . 1263 4. 0703 4. 4141 0.9221 9 5 4. 0263 53. 9713 1 .0313 5. 8047 5. 7109 1. 0164 10 52. 93 40 47. 1440 1 . 1239 4. 0703 4. 4141 0.9221 11 47. 1343 45. 8370 1 .0283 5. 8047 5. 7109 1.0164 12 47. 777 0 42. 6140 1 . 1159 4. 0703 4. 4141 0.9221 13 43. 5120 33. 9313 1 .0436 5. 8047 5. 7109 1.0164 14 42. 8730 34. 7250 1 .2346 4. 0703 4. 4141 0.9221 15 35. 8090 33. 4090 1 .0718 5. 8047 5. 7109 1.0164 15 39. 2130 31. 5550 1 .2427 4. 0703 4. 4141 0.9221 17 45. 7280 43. 8553 1 .0427 5. 8047 5. 7109 1.0164 18 45. 7440 39. 972 3 1 . 1444 4. 0703 4. 4141 0.9 221 19 40. 3930 37. 8290 1 .0678 5. 8047 5. 7109 1.0164 20 41. 8653 36. 74 4 3 1 . 1334 4. 0703 4. 4141 0.9221 21 53. 2290 49. 1930 1 .0210 6. 4922 5. 0234 1.2924 22 43. 0 3 30 44. 2770 1 . 0849 4. 0391 4. 4453 0.9086 23 4 7. 5980 40. 553 0 1 . 1738 6. 4922 5. 0234 1.2924 24 44. 9430 39. 9270 1 .1256 4. 0391 4. 4453 0.9086 25 35. 1180 36. 4540 0 .9634 6. 4922 5. 0234 1.2924 26 32. 6643 31. 2743 1 .0444 4. 0391 4i 4453 0.9086 27 32. 69 80 28. 850 3 1 . 1334 6. 4922 5. 0234 1.2924 28 33.2013 29. 3310 1 .0279 4. 0391 4. 4453 0.9086 29 53. 6680 54. 64 8 0 0 .9821 6. 4922 5. 0234 1.2924 30 53. 2460 47. 4393 1 .0592 4. 0391 4. 44 53 0.9086 31 50. 0350 41. 9490 1 . 1923 6. 49 22 5. 0234 1.2924 32 46. 5350 42. 4370 1 . 0935 4. 0391 4. 4453 0.9086 33 43. 0220 37. 6310 1 .0635 6. 4922 5. 0234 1-2924 34 41. 6623 33. 3763 1 . 2483 4. 0391 4. 4453 0.9086 35 37. 4730 29. 7930 1 .2579 6. 4922 5. 0234 1-2924 36 3 9. 0240 29. 674 0 1 .3151 4. 0391 4. 4453 0. 9086 37 4 4. 2330 42. 0310 1 .0524 6. 4922 5- 0234 1.2924 38 42. 4 313 38. 4413 1 . 1038 4. 0391 4. 4453 0.9086 ^39 41. 3360 33. 6160 1 .2297 6. 4922 5. 0234 1.2924 40 3 9. 7330 34. 6770 1 . 1472 4. 0391 4. 4453 0.9086 . . . continued 169. Table 28 continued Top Bottom Display Top Bottom value value no. A El A E2 21E1/^E2 VI V2 V1/V2 41 50. 7630 43. 4440 1.0479 6. 4375 5-0781 1.2677 42 52. 5750 45. 1380 1. 1395 3. 5938 4. 8906 0.7348 43 45. 5740 42. 5360 1.0973 6. 4375 5. 0781 1. 2677 44 49. 0643 41. 8330 1. 1727 3. 5938 4. 8906 0.7348 45 36. 2100 34. 1250 1.0611 6. 4375 5. 0781 1. 2677 46 33. 2833 31. 6423 1.2103 3. 5938 4. 8906 0.7348 47 32. 4060 27. 3140 1.1864 6. 4375 5. 0781 1.2677 48 36. 3453 28. 1153 1.2927 3. 5938 4. 8906 0.7348 49 55. 1440 51. 8470 1.0636 6. 4375 5, 0781 1. 2677 50 5 3. 7150 48.6330 1.1653 3. 5938 4. 8906 0.7348 51 49. 9170 41. 5560 1.2012 6. 4375 5. 0781 1.2677 52 52.8290 43. 3750 1.2133 3. 5938 4. 8906 0.734 8 53 40. 7290 38. 4300 1.0598 6. 4375 5. 0781 1. 2677 54 45. 6593 36. 2663 1.2866 3. 5938 4. 8906 0.7348 55 3 6. 7550 31. 5270 1. 1658 6. 4375 5- 07 81 1.2677 56 43. 9090 32. 5470 1 .349 1 3. 5938 4. 8906 0.7348 57 43. 4 160 40. 5450 1.1912 6. 437 5 5. 0781 1. 2677 58 51. 3753 38. 8490 1.3224 3. 5938 4. 8906 0.7348 59 43. 7330 33. 5470 1.3036 6. 4375 5. 0781 1.2677 60 43. 3080 35.1560 1.3741 3. 5938 4. 8906 0.7348 61 43. 7760 53. 3170 0.9694 5. 7344 5. 7813 0. 9919 62 45. 0370 44. 4340 1.0136 4. 20 31 4. 2813 0.9817 63 4 4. 4630 44. 2430 1.0053 5. 7344 5. 7813 0. 9919 64 41. 13 10 40. 3760 1.0187 4. 2031 4. 2813 0.9817 65 34. 6953 36. 2320 0.9584 5. 7344 5. 7813 0. 9919 66 30. 3710 29. 4300 1.0299 4. 203 1 4. 2813 0.9817 67 30. 6213 30. 4493 1.0056 5. 7344 5. 7813 0.9919 68 27.3660 26. 4500 1.034 6 4. 2031 4, 2813 0.9 817 69 52. 7133 54. 103 0 0.9744 5. 7344 5. 7813 0.9919 70 43. 4030 46. 8650 1.0329 4. 2031 4. 2813 0.9817 71 46. 3270 45. 6 94 0 1.0248 5. 7344 5. 7813 0.9919 72 43. 6330 42. 1680 1.0359 4. 2031 4. 28 13 0. 9817 73 33. 2433 39. 4940 0.9685 5. 7344 5. 7813 0.9919 74 36. 6930 35. 5060 1.0334 4. 2031 4. 2813 0. 9817 75 34. 1970 33. 7240 1.0143 5. 7344 5. 7813 0.9919 76 33. 4390 32. 0590 1.0430 4. 2031 4. 2813 0.9317 77 43. 0270 43. 3140 0.9934 5. 7344 5. 7813 0.9919 78 40. 8750 38. 9370 1.0506 4. 2031 4, 2813 0„9817 79 33. 64 30 37. 0770 1.0424 5. 7344 5. 7813 0.9919 30 37. 5940 35. 4223 1.0613 4. 2031 4. 2813 0.9817 170. TABLE 29 Top-bottom proportions of chroma and temperature fo r 80 d i sp lays . Top Bottom Top Bottom Display chroma chroma temp. temp. no. CI C2 C1/C2 Tl T2 T1/T2 1 4. 4375 4. 4688 0. 9930 9. 7656 8. 6719 1.1261 2 4. 4375 4. 4688 0- 9930 9. 7656 3 . 6719 1. 1261 3 3-5625 3. 5313 1. 0088 9. 1094 7. 9688 1. 1431 4 3. 5625 3. 5313 1. 0088 9. 1 094 7. 9683 1. 14 31 5 4. 4375 4. 4688 0. 9930 6. 0469 5. 9531 1.0158 6 4. 4375 4. 4688 0. 9930 6. 046 9 5. 9531 1.0158 7 3. 5625 3. 5313 1. 0088 6, 0469 5. 95 3 1 1.3158 8 3. 5625 3. 5313 1-0088 6. 0469 5. 9531 1.0158 9 4. 4375 4. 4688 0. 9933 2. 2813 3. 2813 0.6952 10 4. 4 3 75 4-4638 0-9930 2. 2313 3. 2313 0.6952 11 3. 5625 3. 5313 1. 0088 2. 9375 3. 9844 0.7373 12 3. 5625 3. 5313 1. 0088 2-9375 3. 9844 0.7373 13 4. 4375 4. 4688 0. 9933 3. 8281 4. 4688 0.8566 14 4. 4375 4. 4688 0. 9930 3. 3281 4. 4638 0.8566 15 3. 5625 3. 5313 1. 0088 4. 2556 4. 9375 0.8639 16 3. 5625 3. 5313 1- 0083 4. 2656 4. 9375 0.3639 17 4. 4375 4. 4688 0. 9933 8. 0859 /., 6250 1.0604 13 4. 4375 4. 4688 0. 9930 8. 0859 7. 6250 1.0604 19 3. 5625 3. 5313 1. 0088 7. 6328 7. 1563 1.0666 20 3. 5625 3. 5313 1. 0088 7. 6 328 7. 1563 1 .0666 21 4. 2656 4. 6406 0. 9192 9. 3672 9„ 0703 1.0327 22 4. 2656 4. 6406 0. 9192 9. 3672 9. 0703 1.0327 23 3. 7344 3-3594 1. 1116 8. 9688 8. 1094 1. 1060 24 3. 7344 3. 3594 1. 1116 3. 9688 3. 1094 1.1060 25 4. 2656 4. 6406 0. 9192 6. 2969 5-7031 1. 1041 26 4. 2656 4. 6406 0. 9192 6. 2969 5. 7031 1. 1041 27 3. 7344 3. 3594 1. 1116 6. 2969 5-7 031 1. 1041 23 3. 7344 3. 3594 1. 1116 6. 2969 5. 7031 1.1041 29 4. 2656 4. 6406 0» 9192 2. 9297 2. 5547 1. 1468 30 4. 2656 4. 6406 0. 9192 2. 9297 2. 554 7 1.1468 31 3. 7344 3. 3594 1. 1116 3. 3281 3. 5938 0.9261 32 3. 7344 3. 3594 1. 1116 3. 3281 3. 5938 0.9261 33 4„ 2656 4. 6406 0. 9192 3. 9453 4. 3516 0.9066 34 4. 2656 4. 6406 0, 9192 3. 9453 4. 3516 3.9066 35 3. 7344 3-3594 1. 1116 4. 2109 4. 9922 0.8435 36 3, 7344 3. 3594 1. 1116 4. 2109 4. 9922 0.3435 37 4. 2656 4. 6406 0. 9192 7. 4 60 9 8. 2422 0.9052 33 4. 2656 4. 6406 0. 9192 7. 4609 3 . 2422 0.9052 39 3. 7344 3. 3594 1- 1116 7. 1641 7. 6016 0.94 24 40 3. 7344 3. 3594 1. 1116 7. 1641 7. 6016 0.9424 . . . continued 171. Table 29 continued Top Bottom Top Bottom Display chroma chroma temp. temp, no. Cl C2 C1/C2 - TI T2 T1/T2 41 4-3438 4.5625 0.9521 9.5469 8. 8906 1.0733 42 4.3438 4.5625 0.9521 9.546 9 8. 8906 1.0738 43 3-6563 3.4 37 5 1.06 37 9.0313 8. 0459 1.122 3 44 3. 6563 3.4375 1.0637 9.0313 8. 0469 1.1223 45 4.3438 4.5625 0.9521 5.7969 6. 20 31 0.9345 46 4.3438 4.5625 0.9521 5.7969 6. 20 31 0.9345 47 3.6563 3.4375 1.0637 5.7969 6. 2031 0.9345 43 3.6563 3.4375 1.0637 5.7969 6. 2031 0.9345 49 4.3438 4.5625 0.9521 2.2891 3. 3125 0.6910 50 4.3438 4.5625 0.9521 2.2891 3. 3125 0.6910 51 3. 6563 3.4375 1.0637 2.7656 4. 1563 3.6654 52 3. 6563 3.4375 1.0637 2.7656 4. 1563 0.6654 53 4.3438 4-5625 0.9521 4.1875 4. 1094 1.0190 54 4.3438 4.5625 0.9521 4.1375 4. 10 94 1.0190 55 3.6563 3.4375 1.0637 4.5313 4. 6719 0.9699 55 3.6563 3.4375 1.0637 4.5313 4. 6719 0.9699 57 4.3438 4.5625 0.9521 8.2188 7. 4844 1.0931 53 4.3438 4.5625 0.9521 3.2188 7. 4844 1.0981 59 3.6563 3.4375 1.0637 7.8750 6. 9219 1. 1377 60 3.6563 3.4375 1.0637 7.8750 6. 9219 1. 1377 61 4.3750 4.5313 0.9655 8.9844 9. 4531 0.9504 62 4.3750 4.5313 0.9655 8.9344 9. 4531 0.9504 6 3 3. 6250 3.4688 1.0450 8.4219 8. 6563 0.97 29 64 3. 6250 3.4688 1.0450 8. 4219 8. 6563 0.9729 65 4.3750 4.5313 0.9655 5.9219 6. 0781 0.9743 66 4.3750 4.5313 0.9655 5.9219 6. 0781 0.9743 67 3.6250 3-4688 1.0450 5.9219 6. 0781 0.9743 68 3.6250 3.4688 1.0450 5. 9 219 6. 0781 0.9743 69 4.3750 4.5313 0.9655 2.9375 2. 6250 1. 1190 70 4.3750 4.5313 0.9655 2.9375 2. 6250 1. 1190 71 3.6250 3.4688 1.0450 3.5000 3. ,4219 1.0228 72 3.6250 3.4688 1.0450 3.5000 3. 4219 1.0228 73 4.3750 4.5313 0.9655 4.3594 3. 9375 1.1071 74 4.3750 4.5313 0.9655 4.3594 3. 9375 1. 1071 75 3. 62 50 3.4688 1.0450 4.7344 4. ,4638 1.0594 76 3.6250 3.4688 1.0453 4. 7344 4. ,4688 1 .0594 77 4.3750 4.5313 0.9655 7.7969 7. .906 3 0.9862 78 4.3750 4.5313 0.9655 7.7969 7.906 3 0.9862 79 3.6250 3.4688 1.0450 7. 4219 7, , 3750 1.0064 80 3.6250 3.4688 1.0450 7-4219 7. ,3750 1.0064 172. Balance Lack of balance FIGURE 51 Le f t - r igh t p i c t o r i a l balance e f f e c t . 173. d iv id ing the display into two equal halves. 1. Color di f ference measurements. Average t r i s t imulus values, co lor differences between major and minor, and proportion of Z lE( le f t ) / idE( r igh t ) were ca lcu lated. These are l i s t e d in table 30. 2. Calculat ions of value, chroma and perceived temperature  based on the l e f t / r i g h t d i v i s i o n . Average values, chromas and temperatures fo r the two parts of the displays were calculated as before, and the proportions of these measures computed. Tables 30 and 31 show these values. Average of small-area contrasts When looking at a p icture surface, i t i s apparent that some areas are more contrasty than others. In a landscape, for instance, a sample area of the sky may hardly contain any var ia t ion at a l l , while a sample area of face, for example, may contain very dark shadows around a f a c i a l feature as wel l as some l i g h t , uniform area of s k i n . In the l a t t e r case, the s i ze and frequency of these smal l -area contrasts would be of considerable magnitude. In order to carry out an analysis of average small-area contrast i n the present d isp lays , a "small-area" was defined as an area consist ing of a 2 x 2 matrix of co lor elements. These small areas consist ing of four elements each were then used to systematical ly 174. TABLE 30 Le f t - r i gh t proportions of AE and value f o r 80 d i sp lays . Le f t Right DisDlay Le f t Right value value no. JE1 AE1 4E1 /4E2 VI V2 V1/V2 1 5 0 . 302 0 5 0 . 6353 1„ 3023 5 . 7500 5. 7656 0 .997 3 2 4 1 . 4060 4 9 . 7243 0 . 8327 3 . 9063 4 . 5781 0 . 8 5 3 3 3 4 7 . 1320 4 3 . 4970 1. 0336 5 . 7500 5. 7656 0 . 9 9 7 3 4 3 9 . 7980 4 3 . 5350 0 . 9142 3. 9063 4 . 5781 0 . 8 5 3 3 5 3 3 . 3140 3 5 . 0900 1 . 0919 5 . 7500 5 . 7656 0. 9973 6 2 8 . 2390 3 3 . 7600 0 . 8379 3 . 9063 4 . 5781 0 . 8 5 3 3 7 3 5 . 0450 2 3 . 5350 1. 2 281 5 . 7500 5 . 7656 0 . 9 9 7 3 S 2 7 . 3600 2 9 . 0740 3 . 94 10 3. 9 0 6 3 4 . 5781 0 . 8 5 3 3 9 5 4 . 6 9 7 0 5 4 . 6780 1 . 0003 5 . 7500 5 . 7656 0 . 9 9 7 3 10 4 4 . 6750 5 2 . 6303 0 . 8494 3 . 9063 4 . 5781 0 . 8 5 3 3 11 4 9 . 6650 4 4 . 8110 1. 1083 5 . 7500 5 . 7656 0 .9973 12 4 2 . 3200 4 5 . 0430 0 . 9505 3 . 9063 4 . 5781 0 . 8 5 3 3 13 3 3 . 6770 4 1 . 8820 0 . 9235 5 . 7500 5. 7656 0 . 9 9 7 3 14 3 1 . 5260 4 2 . 5723 0 . 7405 3. 9063 4 . 5781 0 . 8 5 3 3 15 3 5 . 2300 3 5 . 1370 1. 0341 5 . 7500 5. 7656 0 . 9 9 7 3 16 3D. 2230 3 7 . 3520 0 . 8091 3 . 9 0 6 3 4 . 5781 0 . 8 5 3 3 17 4 2 . 9510 4 5 . 9343 0 . 9361 5 . 7500 5. 7656 0 . 9 9 7 3 18 3 6 . 6510 4 5 . 4410 0 . 8066 3. 9063 4 . 5781 0 . 8 5 3 3 19 3 9 . 2100 3 9 . 4320 •). 0232 5. 7500 5. 7656 0 . 9 9 7 3 20 3 5 . 3 8 1 0 3 9 . 9200 0 . 8863 3 . 9063 4 . 5781 0 . 8 5 3 3 21 4 8 . 4640 5 0 . 4690 0 . 9603 5 . 6719 5 . 8438 0 . 9 7 0 6 22 4 3 . 8190 4 5 . 9930 0 . 9527 4 . 3594 4. 1250 1 .0568 23 4 3 . 5080 4 4 . 3500 0 . 9701 5 . 6719 5. 8438 0. 973 6 24 4 1 . 0610 4 3 . 3760 1. 0170 4 . 3594 4 . 1250 1 .0568 25 3 4 . 0270 3 6 . 6430 0 . 9286 5 . 6719 5. 8438 0 . 9 7 0 6 26 2 8 . 7 1 4 0 3 1 . 5120 0 . 9112 4 . 3594 4 . 1250 1 -0568 27 2 9 . 4 92 0 3 1 . 5553 0 . 9346 5 . 6 719 5 . 8438 0 . 9 7 3 6 2 8 2 7 . 3720 2 3 . 7400 0 . 96 9 3 4 . 3594 4 . 1250 1 .0568 29 5 2 . 5710 5 4 . 7290 0 . 9606 5 . 6719 5 . 8438 0 . 9 7 0 6 30 4 5 . 8860 4 9 . 6640 0 . 9239 4 . 3594 4 . 1250 1 .0568 31 4 5 . 1650 4 7 . 0530 0 . 9599 5 . 6719 5. 8438 0 . 9 7 3 6 32 4 2 . 9720 4 2 . 9590 1 . 0003 4 . 3594 4 . 1250 1 .0568 33 3 9 . 2200 3 9 - 5170 0 . 9672 5 . 6719 5 . 8438 0 . 9 7 0 6 34 3 5 . 4 120 3 7 . 2300 0 . 9512 4 . 3594 4 . 1250 1 .0568 35 3 3 . 5320 3 4 . 2290 0 . 9796 5. 6719 5. 8438 0 . 9 7 3 6 36 3 3 . 1 2 4 0 3 2 . 5170 1 . 0187 4 . 3594 4 . 1250 1 .0568 3 7 4 2 . 2310 4 4 . 3070 0- 9531 5 . 671 9 5 . 8438 0 . 9736 33 3 8 . 6 8 3 0 4 1 . 7770 0 . 9259 4 . 3594 4 . 1250 1 .0568 39 3 7 . 1420 3 3 . 6200 3 . 9617 5 . 6719 5 . 8438 0. 9736 40 3 6 . 4270 3 6 . 9520 0 . 9958 4 . 3594 4. 12 50 1 .0568 . . . continued 175. Table 30 continued Le f t Right Display Le f t Right val ue val ue no. 1^E1 AEZ AE]/ AEZ VI V2 V1/V2 41 52. 9330 50 .6160 1 .04 59 5. 3906 6. 1250 0.880 1 42 44. 9610 43 .8730 1 .0248 4. 2031 4. 2813 0-9817 43 50. 7270 45 .0930 1 . 1248 5. 390 6 6. 1250 0.8801 44 40. 63 3 0 41 . 7860 0 .9716 4. 20 3 1 4. 2813 0.9817 45 4 1 . 8560 37 .1340 1 . 1256 5. 3906 6. 1250 0.8801 4b 33. 2220 30 . 5440 1 .0877 4. 2031 4. 2313 0-9817 47 38. 2780 32 .0370 1 . 1929 5. 3906 6. 1250 0.8801 43 29. 6120 29 .5720 1 .001 4 4. 203 1 4. 2813 0.9817 49 57. 0480 54 , 5670 1 .0455 5. 3906 6. 1250 0.8801 50 4 8. 3110 4 7 .3100 1 .0212 4. 2031 4. 2813 0.9817 51 51. 2940 47 .4670 1 .0306 5, 3906 6. 1250 0.8801 52 4 2. 5180 45 .5410 0 .9335 4. 2031 4. 2813 0.9817 53 40. 8920 39 .0840 1 .0463 5. 3906 6. 1250 0.8801 54 34. 7980 34 .7150 1 .0024 4. 2031 4. 2313 0.9817 55 37. 0890 33 .9020 1 .0940 5. 3906 6. 1250 0.8801 56 30. 8380 3 3 . 1550 0 .9301 4. 203 1 4. 2313 0.9817 57 37. 3230 46 .0.530 0 .8104 5. 3906 6. 1250 0.880 1 58 3 1 . 3 95 0 42 . 1103 0 .7455 4. 2031 4. 2813 0.9817 59 33. 5970 40 - 2260 3 .8352 5. 3906 6. 1250 0.8801 60 27. 5070 40 .6140 3 .6773 4. 203 1 4. 2813 0.9817 61 49. 2980 49 . 6260 0 .9934 5. 6 79 7 5. 8359 0.9732 62 43. 9850 45 .6520 0 .9635 4. 2578 4. 2266 1.0074 63 44. 2650 44 . 1230 1 .0032 5. 6797 5. 8359 0-9732 64 40. 5 19 0 41 .0 570 0 .9369 4. 2578 4. 2266 1.0074 65 3 5 . 3070 35 . 3620 0 .9984 5. 6797 5. 8359 0.9732 66 2 9 . 2900 30 .7110 0 .9537 4. 2578 4. 2266 1.0074 67 33. 6 180 30 .0820 1 .0173 5. 6797 5. 8359 0- 9732 68 26. 7090 27 .1730 0 .9829 4. 2578 4. 2266 1.0074 69 5 3 . 0300 53 .5770 0 .9879 5. 6797 5. 8359 0-97 3 2 7 0 45. 6770 43 .7570 0 .9571 4. 2578 4. 2266 1.0074 71 46. 1380 46 .0850 1 .0011 5. 6797 5. 8359 0.9732 7 2 42. 6390 43 . 2390 0 .9350 4. 2578 4. 2266 1.0074 73 3 3. 3890 39 .3310 0 . 9760 5. 6797 5. 8359 0-9732 74 34. 9570 3 7 .5710 0 .9304 4. 2578 4. 2266 1.0074 75 33. 6560 34 . 1030 0 . 9869 5. 6797 5. 8359 0.9732 76 32. 0260 33 .7370 0 -9493 4. 2578 4. 2266 1.0074 77 42. 8 09 0 43 .6810 0 .9800 5. 6797 5. 8359 0. 9732 78 39. 0590 41 .1120 3 .9501 4. 2578 4. 2266 1.0074 79 37. 6640 33 . 0240 0 .9905 5. 6797 5. 8359 0.9732 80 36. 0910 37 . 2530 3 . 9688 4. 2 57 8 4. 2266 1.0074 176. TABLE 31 Le f t - r i gh t proportions of chroma and temperature f o r 80 d i sp lays . Le f t Right Le f t Right Display chroma chroma temp. temp. no. Cl C2 C1/C2 TI T2 T1/T2 1 4. 2813 4.6250 0. 9257 8.3750 9.5625 0. 9281 2 4. 2813 4.6250 0. 9257 8.8750 9.5625 0. 9231 3 3. 7188 3.3750 1. 1019 8.4531 8.6250 0. 9801 u 3. 7188 3.3750 1. 1019 8.4531 8.6250 0. 93 01 5 4. 2813 4. 6250 0. 9257 5.7813 6.2188 0. 9296 6 4. 2813 4.6250 0. 9257 5.7813 6.2188 0. 9296 7 3. 7188 3.3750 1. 1019 5.7813 6.2188 0. 9296 8 3. 7198 3.3750 1. 1019 5.7813 6.2188 0. 9296 9 4. 2313 4.6250 0. 9257 2.9063 2.6563 1. 0941 10 4. 2813 4.6250 0. 9257 2.9063 2.6563 1. 0941 11 3. 7188 3.3750 1. 1019 3.3281 3.5938 0. 9 261 12 3. 7188 3.3750 1. 1019 3.3281 3.5938 0. 9261 13 4. 2813 4.6250 0. 9257 4.5469 3.7500 1. 2125 14 4. 2813 4.6250 0. 9257 4.5469 3.7500 1. 2125 15 3. 7188 3.3750 1. 1019 4.8281 4.3750 1. 1036 16 3. 71 88 3.3750 1. 1019 4.8281 4.3750 1. 1036 17 4. 2813 4.6250 0. 9257 7.8906 7.3203 1. 0090 18 4. 2313 4.6250 0. 9257 7.8906 7.3203 1. 009 0 19 3. 7183 3.3750 1. 1019 7.6094 7. 1797 1. 0598 20 3. 7188 3.3750 1. 1019 7.6094 7.1797 1. 0598 21 4. 4063 4.5000 0. 9792 9.4453 8.9922 1. 0504 22 4. 4063 4.5000 0. 9792 9.4453 8.9922 1. 0504 23 3. 5938 3,5000 1. 0268 8.3359 8.2422 1. 0720 24 3. 593 8 3.5000 1. 0 268 8. 8359 3.2422 1-0720 25 4. 4063 4.5000 0. 9792 6.0469 5.9531 1. 0158 26 4. 4063 4.5000 0. 9792 6„0469 5-9531 1. 0158 27 3. 5338 3.5000 1. 0268 6. 0469 5.9531 1. 0158 28 3. 5938 3.5000 1. 0268 6.0469 5.9531 1. 0158 29 4. 4063 4.5000 0. 9792 2. 5234 2.9609 0. 3522 30 4. 4063 4.5000 0. 9792 2.5234 2.9609 0. 3522 31 3. 5938 3.5000 1. 0268 3.2109 3.7109 0. 3653 32 3. 5933 3.5000 1. 0263 3.2109 3.7109 0. 3653 33 4. 4063 4.5000 0. 9792 3. 9 37 5 4.3594 0. 9032 34 4. 4063 4.5000 o, 9792 3.9375 4.3594 0. 9032 35 3. 5938 3.5000 1. 0268 4.3438 4.359 4 0. 89 39 36 3. 5938 3.5000 1. 0268 4.3438 4.8594 0. 39 39 37 4. 4063 4.5000 0. 9792 7.9688 7.7344 1. 0303 38 4. 4063 4.5000 0. 9792 7.9688 7.7344 1. 030 3 39 3. 5938 3. 5000 1. 0268 7.5625 7.2031 1. 0499 40 3. 5938 3.5000 1. 0268 7.5625 7.2031 1. , 0499 . continued 177. Table 31 continued Lef t Right Lef t Right Display chroma chroma temp. temp. no. Cl C2 C1/C2 TI T2 T1/T2 41 4.5313 4.3750 1.0357 9. 9219 8.5156 1. 1651 42 4.5313 4.3750 1.0357 9.9219 8.5156 1. 1651 43 3.4688 3.6250 0.9569 9.1250 7.9531 1. 1474 44 3.4688 3-6250 0.9569 9. 1250 7.9531 1. 1474 45 4.5313 4.3750 1.0357 6.3125 5.6875 1 . 1099 46 4.5313 4.3750 1.0357 6.3125 5.6375 1. 1099 47 3.4688 3.6250 0.9569 6.3125 5.6875 1. 1099 48 3.4688 3.6250 0.9569 6.3125 5.6875 1. 1099 49 4.5313 4.3750 1.0357 2. 3906 3.2109 0. 7445 50 4.5313 4.3750 1.0357 2.3906 3.2109 0. 7445 51 3.4688 3.6250 0.9569 3.1875 3.7344 0. 3536 52 3.4688 3.6250 0.9569 3.1875 3-7344 0. 3536 53 4.5313 4.3750 1.0357 3.5156 4.7813 0. 7353 54 4.5313 4.3750 1.0357 3.5156 4-7813 0. 7353 55 3.4688 3.6250 0.9569 4.0469 5.1563 0. 7848 56 3.4688 3.6250 0.9569 4.0469 5. 156 3 0. 7843 57 4.5313 4.3750 1.0357 7.3594 7.8438 1. 00 20 53 4.53 13 4.3750 1.0357 7.8594 7.3438 1. 0020 59 3. 4688 3.6250 0.9569 7„3281 7.4638 0- 99 12 60 3.4688 3.6250 0.9569 7.3 281 7.4688 0. 9312 61 4.4219 4.4844 0.9851 9. 1406 9.2969 0. 9832 62 4.4219 4,4844 0.9861 9.1406 9.2969 0. 9332 63 3.5781 3.5156 1.0178 3. 5078 3.5703 0. 9927 64 3.5781 3.5156 1.0178 8.5078 3.5703 0. 9927 65 4.4219 4.4844 0.9861 5.9531 6.0469 0. 9845 66 4.4219 4.4844 0.9361 5.9531 6.0469 0. 9845 67 3.5781 3.5156 1.0178 5.9531 6.0469 0. 9845 68 3. 57 81 3.5156 1.0178 5.9531 6.0469 0. 9845 69 4.4219 4.4844 0.9861 2.8125 2-7500 1. 0227 70 4.4219 4.4844 0.9861 2.8125 2.7500 1. 0227 71 3.5781 3.5156 1.0173 3.4453 3.4766 0. 9910 72 3.5781 3.5156 1.0178 3.4453 3.4766 0. ,9910 73 4.4219 4.4844 0.9861 4.2109 4.0859 1. ,0306 74 4.4219 4.4844 0.9861 4.2109 4.0859 1. ,0306 75 3.5781 3.5156 1.0178 4. 6328 4.5703 1. ,0137 76 3. 5781 3.5156 1.0173 4.6328 4.5703 1. ,0137 77 4.4219 4.4844 0.9861 7.8828 7.8203 1.0030 78 4.42 19 4.4844 0.9861 7.8828 7.3203 1, ,0080 79 3. 5781 3.5156 1.0178 7.4609 7.3359 1. ,0170 80 3. 5781 3.5156 1.0178 7. 4609 7.3359 1. .0170 178. sample the ent i re surface of the d isp lays . Two d i s t i n c t l y d i f f e rent methods of ca lcu lat ing the "contrast" wi th in each of these small areas were used. F i r s t , the absolute numerical differences among a l l s i x combinations of the four color elements were calculated using, success ive ly, values of AE, Munsell value, chroma and perceived temperature. The average of these s i x dif ferences was regarded as a numerical expression of the contrast wi th in that par t i cu la r small area. F i n a l l y , an overa l l average for the ent i re display was obtained by averaging a l l the averages of the small areas. This average was thought of as expressing the s i ze and frequency of a l l the contrasts in the d isp lay. The term given to th i s measure was "adjacent-difference" since i t dealt with (1) co lor elements which were adjacent to one another and (2) perceptual dif ferences between color elements. The second method of ca l cu la t ing the average small area contrasts, was exact ly l i k e the f i r s t method except that instead of absolute numerical d i f ference, the sum of squares about the mean of the values of the four elements was used. In contrast to that of the f i r s t method, th i s measure was intended to express deviat ion from uniformity or monotony. That i s , the higher the numerical ... value of the va r i a t i on , the less uniform the display was, and the more contrast there was present. This method was termed "adjacent-variance." 179. I t i s noted here that these ca lcu lat ions were only carr ied out in conjunction with the displays used in the main study, which w i l l be described shor t l y . Since these displays employed a 32 x 32 aperture mask superimposed on the o r ig ina l 16 x 16 pasted-up s t imu l i , the placement of the 2 x 2 small-area matrices for sampling of contrasts was c r u c i a l . In order to capture a l l possible con-t rasts ex i s t ing in the o r ig ina l s t imu l i , the 2 x 2 sampling mat* r i ces were placed in such a way that they always covered four o f the co lor chips! in the 16 x 16 s t i m u l i . Figure 52 shows part of a 16 x 16 stimulus and the placement of 2 x 2 sampling matr ices. 1. Color di f ference measurements. Referring to f igure 52, the elements of the upper l e f t small-area matrix have been numbered from 1 to 4. To obtain the average adjacent/d i f ference-AE value fo r th i s small area, the fol lowing formula was used: .|-1E1-4E2|+| „2-4E3|+| 4E3-4E4| + | _E4-4El| +| 4El- iE3| +UE4-Z1E2 Average AE =;  6 Once a l l the co lor dif ferences for a l l the small areas of the display were obtained, they were averaged to y i e l d one overa l l AE value for that pa r t i cu la r d isp lay. Table 32 l i s t s the adjacent/difference--4 E values for the 80 d isp lays . To obtain the average adjacent/variance-/! E fo r the same 180. roolooioo 1 loB'StOlolo o o o o o o 1 L \ j. _ oooooo pppoooH___ I Dotted l ines indicate 16x16 matr ix. So l i d l ines Indicate small sampling areas, FIGURE 52 Part of 16x16 pasted-up d isp lay with 32x32 aperture matrix and 2x2 sampling areas. 181. small area shown above, the fo l lowing formula from Ferguson (1971) was used: V(4El+/4E2+dE3+ZlE4) 2 - 4 AEZ ; The AE values thus obtained fo r a l l the small areas of a display were f i n a l l y averaged to y i e l d one overa l l AE value. Table 33 l i s t s the adjacent/var iance-4E values for the 80 d isp lays . 2. Calculat ions of small-area contrasts with value, chroma.and  perceived temperature. The procedures for obtaining the average adjacent-difference and adjacent-variance values as out l ined above were also carr ied out for value, chroma and perceived temperature. Tables 32 and 33 l i s t these overa l l averages. 182. TABLE 32 Adjacent-dif ference measures f o r AE, value, chroma and temperature. Display Adjacent-dif ference no. AE Value Chroma Temperature 1 16.3040 1. 3126 0.7141 1.3467 2 15.2003 1. 2444 0.714 1 1.3467 3 16.7292 1. 3126 0.714 1 1.4489 4 14.6172 1. 2444 0.714 1 1.4489 5 16.2546 1. 3126 0.7141 0. 5467 5 13.9515 1. 2444 0.714 1 0.5467 7 15.7374 1. 3126 0. 7141 0.5467 8 13.4221 1. 2444 0.714 1 0.5467 9 18. 1685 1. 3126 0.714 1 1.2800 10 15.7244 1. 2444 0.7141 1.2800 11 17.5095 1. 3126 0.714 1 1. 3343 12 15.0214 1. 2444 0.7141 1.3348 13 16.3517 1. 3126 0.714 1 1. 1022 14 14.3572 1. 2444 0.7141 1. 1022 15 15.9233 1. 3126 0.714 1 1. 1141 16 13.2757 1. 2444 0.714 1 1.1141 17 15.5263 1. 3126 0.7141 1.0933 18 14.3352 1. 2444 0.714 1 1.0933 19 15.633 1 1. 3126 0.714 1 1. 0563 20 13. 1339 1. 2444 0.7141 1.0563 21 14.4923 0. 9081 0. 54 81 1.3526 22 14.6386 1. 1452 0. 543 1 1.3526 23 14.0628 0. 9081 0. 543 1 1.351 1 24 13.9388 1. 1452 0.5431 1.3511 25 14.0233 0. 908 1 0. 5431 0.5644 26 13.3616 1. 1452 0.5431 0.5644 27 13.5757 0. 908 1 0. 54B1 0.5644 28 15.000b 1. 1452 0.5431 0.5644 29 15. 3266 0. 908 1 0. 543 1 1. 3215 30 1 5.454 1 1. 1 452 0. 543 1 1.3215 31 15.1956 0. 9081 0. 54 8 1 1.5096 32 14.9628 1. 1452 0. 54 31 1.5096 33 13.971 3 0. 9081 0. 548 1 1.1274 34 14.1040 1. 1452 0.5481 1.1274 35 13. 101 1 0. 9081 0. 543 1 1.0385 36 12.8340 1. 14 52 0.5431 1.0335 37 13.6014 0. 9081 0. 54 3 1 1.1067 38 14.00 9 6 1. 1452 0. 543 1 1.1067 39 13.5901 0. 9081 0. 543 1 1. 1244 40 1 3. 2060 1. 1452 0. 54 8 1 1.1244 . . . continued 183. Table 32 continued Display Adjacent-dif ference no. A E Value Chroma Temperature 41 10.7206 0.8548 0.4104 ' 3. 8541 42 9.6598 0.35 19 0.4104 0. 354 1 43 1 1.067 9 0.0543 0.4104 0.93 81 44 9.0239 0.351 9 0. 4104 0. 9081 45 10.2944 0.8548 0.4134 0. 2904 46 8.7245 0.9519 0.4134 0. 2904 47 9.9217 0.8548 0.4104 3. 2934 48 8.3715 0.8519 0.4104 0. 2904 49 1 1.6855 0.3548 0. 4 03 9 0. 33 33 50 9.9283 0.8519 0.4104 0. 8030 51 1 1. 2432 0. 8548 0.4104 0. 831 1 52 9.5260 0.8519 0-4104 0. 33 11 53 10.5631 0.8543 0.4134 0. 6822 54 9. 1742 0.85 1 9 0.4104 0. 6822 55 10.0055 0. 8543 0.4104 3. 6481 56 3.2705 0.851 9 0.4104 0. 6431 57 9.808J 0.8543 0.4104 0. 6970 58 9.1933 0.85 19 0.4104 0. 6970 59 9.8070 0.8543 0.41 J 4 3. 6452 60 8.2528 0.8519 0.4104 0. 6452 61 2 6 . 1306 1. 9503 0.3622 2. 19 55 62 2 3.530 7 1. 9533 0.3622 2. 1955 63 2 4.8532 1.9503 0.8622 2. 1373 64 2 1 . 6 3 6 3 1. 9533 0. 36 2 2 2. 1378 65 2 5 . 3294 1.9503 0.U622 3. 85 67 66 21.4824 1.9533 0.8622 3. 8667 67 2 3 . 7 7 1 5 1.9503 0.8622 3. 866 7 68 20. 1333 1. 9533 0.8652 , o. 8667 69 2 8.0819 1.9503 0.8622 2. 2296 70 2 4 . 6 9 0 3 1. 9533 0. 8622 2. 2296 71 2 5.9982 1.9503 0.3622 2. 1 155 72 22.5717 1. 9533 0. 8622 2. 1155 73 24.7558 1.9503 0.8622 1. 7630 74 22.096 1 1.9533 0.8622 1. 7600 75 2 3. 4738 1.9503 0.8622 1. 6933 76 19.9635 1.9533 0.8622 1. 6933 77 23.8724 1.9503 0.8622 1. 802 3 70 22-3551 1. 9533 0.3622 1. 8029 79 23.2666 1.9503 0.8622 1. 6300 80 20. 1339 1. 9533 0.8622 1. 6300 184. TABLE 33 Adjacent-variances measures f o rZ lE , va lue, chroma and temperature. Display Adjacent-variance no. 21E Value Chroma Temperature 1 102.2471 0. 9141 0. 4215 2. 1423 2 104.6796 0. 7348 0. 4215 2. 1423 3 100.2476 0. 9141 0. 4215 1. 9007 4 97.9981 0. 734 8 0. 4215 1. 9007 5 75.1907 0. 9141 0. 4215 0. 3693 6 70.8490 0. 7343 3. 4215 0. 3693 7 69.4391 0. 9141 0. 4215 0. 3693 3 71.0641 0. 7348 0. 421 5 0. 3693 9 108.9407 0. 9141 0. 4215 2. 1939 10 117.9906 0. 7343 3. 4215 2. 1939 11 106.0977 0. 9141 0. 4215 2. 0099 12 108.7509 0. 7348 3. 4215 2. 0399 13 82.0837 0. 9141 0. 4215 0. 7607 14 30.7522 0. 7343 3. 4215 0. 7607 15 33. 1604 0. 9141 0. 4215 0. 7423 16 78.1335 0. 7348 3. 4215 0. 7423 17 75.9908 0. 9 1 4 1 0. 4215 0. 6799 18 84.0619 0. 7343 3. 4215 0. 6799 19 80.0822 0. 9141 0. 4215 0. 6479 20 '78.4214 0. 7348 3. 4215 0. 64 79 21 105.0022 0. 5404 0. 336 0 3. 0129 22 123.8399 0. 859 6 0. 3360 3. 0129 23 98.5407 0-54 04 0. 3860 2. 6419 24 110.3455 0. 8596 0. 3860 2. 6419 25 66.2103 0. 5404 0. 3860 0. 4602 26 89.0313 0. 8596 0. 3860 0. 4602 27 61.0141 0. 5404 0. 3860 0. 4602 28 73.8736 0. 8596 0. 3860 0. 4632 29 108.4547 0. 5404 0. 3860 2. 7830 30 141.7114 0. 8596 3. 3860 2. 7830 31 105.2046 0. 5404 0. 3860 2. 6170 32 125.0230 0. 8596 0. 3 86 0 2. 6170 33 75.2395 0. 5404 0. 3860 1. 0072 34 96.7920 0. 8596 r\ <J . 3860 1. 0072 35 71.5365 0. 5404 0. 3860 0. 3975 36 90.1653 0. 8596 0. 3360 0. 8975 37 83.9730 0. 5404 0. 3860 1. 0930 33 106.1044 0. 8595 3. 3360 1. 0980 39 80.9588 0. 54 0 4 0. 3860 1. 0633 40 96.9386 0. .8596 0. 3360 1. 0633 • • * continued 185. Table 33 continued Display Adjacent-vari ance no. AE Value Chroma Temperature 41 105.6332 0.7635 0. 338 1 2.0941 42 114.3405 0.71 14 0.3881 2.0941 43 104.0240 0.7685 0.388 1 1.6370 44 89.6794 0.7114 0.3881 1.6370 45 86.2655 0.7635 0.3381 0.3256 46 79.2033 0.71 14 0. 338 1 0.3256 47 74.4042 0. 7685 0.3801 0.3256 48 71.4099 0.7114 0.3881 0.3256 49 125.4822 0.7685 0.3381 2. 1217 50 126.3509 0.7114 0. 338 1 2.1217 51 100.8875 0.7685 0.3881 1.6821 52 99.7114 0.71 14 0.3881 1.6321 53 38.7464 0.7685 0. 383 1 0.7700 54 91.8218 0.71 14 0.3831 0.7700 55 83. 2894 0.7635 0.3831 0.6366 56 75.5602 0.71 14 0.3381 0.6366 57 88.1833 0.7635 0.3881 0.7999 58 99.0704 0.71 14 0. 338 1 0.7999 59 3 1. 4636 0.7685 0.3381 0.6532 60 77.5276 0.71 14 0.3388 0.6532 61 .76.0759 0.4246 3.3583 2.9764 62 !78.0857 0.4269 0.3588 2.9764 63 65.2502 0.4246 0.3538 2.3573 6 4 64.1470 0.4269 0.3388 2.3573 65 36.6646 0.4246 3.3588 0.4533 66 44.6597 0.4269 0.3583 0.4533 67 31.2563 0.4246 0.3588 0.4533 68 4 1.8220 0.4269 0.3583 0.4533 69 90.5369 0.4246 0.3588 2.9229 70 91.1584 0.4269 0.3588 2.9229 71 71.3975 0.4246 0.3588 2.3633 7 2 73.9229 0.4269 0.3588 2.3630 73 43.9556 0.4246 0.3588 0.6709 74 54.5803 0.4269 0.3583 0.6709 75 37.5305 0.4246 0.3538 0.5653 76 48.2197 0.4269 0.3588 0.5653 77 52.7065 0.4246 3.3588 0.6877 78 61.3872 0.4269 0.358 8 0.6877 79 43.8652 0.4246 0.3583 0.6004 80 . 52.464 1 0.4269 0.3588 0.6004 186. CHAPTER V A PILOT STUDY A p i l o t study was carr ied out to tes t a number of construction features of both the displays and the response measures. The purpose of the study was s pe c i f i c a l l y to invest igate and evaluate the fol lowing f i ve po ints. 1. I t was the o r i g ina l intent ion to convert the displays described in chapter IV to co lor s l ides and to show these to a r e l a t i v e l y large number of subjects in a s ing le s i t t i n g . However, the extent to which the Munsell colors used in the displays might change due to the s l i d e conversion process was unknown. Part of the object of the p i l o t study was therefore the actual s l i de conversion, the subsequent project ion of the display s l i d e s , and the i r co lor imetr ic ana lys i s . 2. As mentioned in chapter IV, masks-were to be superimposed on 187. the displays in order to cover the i r regu la r paste-up l i n e s . A 16 x 16 matrix pattern was chosen for the masks, and i t was now desirable to f ind out how a mask of th i s fineness would a f fec t the recognit ion of the mot i fs . For reasons stated prev ious ly, subjects were to wr i te down on the response form what they thought they saw i n the d i sp lays . This amounted to asking subjects whether they in fact recognized the intended moti f . Since the working out of the wording of th i s pa r t i cu la r question had caused some problems, i t was decided to tes t one version in th i s study. 3. Due to the physical arrangement of subjects in an experiment with projected s t imu l i , subjects would have to s i t at varying distances from the project ion screen. Part of th i s prel iminary study was to ascertain whether distance (6r , as1 a resu l t of distance, v isual angle) made any dif ference to the recognit ion of the mot i f s . 4. In experiments with s e r i a l presentations, an e f fec t due to order often manifests i t s e l f . This e f f ec t might be the resu l t for instance of learning e f fec t or due to fat igue. Whether th i s was the case here would be checked in the p i l o t study.. 5. F i na l l y , the p i l o t study was intended to supply an i n i t i a l ind icat ion 188. of the e f fects due to hue, value, chroma, motif and sex. The sample The p i l o t sample consisted of 19 evening class students enro l led 1n an introductory psychology course at The Univers i ty of B r i t i s h Columbia. There were 7 males and 12 females. The overa l l mean age was 27.6 years (range 18 to 38 years) , the mean age fo r males was 24,3 years (range 18 to 31) years and for females 29.6 years (range 21 to 38 years ) . The st imul i The s t imu l i were the pasted-up displays described in chapter IV. These were photographed onto Ektachrome s l i d e f i l m , a mask made of Kodak Ortho type 3 f i lm (as shown in f igure 53) was superimposed onto the 35 mm s l i d e f i l m , and the ent i re assembly mounted i n Gepe s l i d e mounts. A f in ished s l i d e i s shown in f igure 54. The mask consisted of a 16 x 16 matrix of apertures--correspond1ng to the 16 x 16 matrix of co lo r chips—since th i s "fineness sca le" was thought to produce recognizable motifs (Harmon, 1973). Before i t s reduction to 35 mm s l i de s i z e , the black mask had apertures of 7/32" diameter and spaces between the apertures of 1/32". The to ta l area of the display was 4" x 4" (16 sq. i n . ) with 9.62 sq . i n . made up of the co lor f i l l e d apertures, and 6.38 sq . i n . made up of the black 189. it::::::::::?::: »**•••*••••*•*** **••••••••••#••» **•••*••*•*••*•• :::::::::::::::: ................ FIGURE 53 16x16 f i lm mask used in p i l o t study, (Actual s i z e . ) FIGURE 54 S l i de of d isplay no. 1. 190. mask. A f ter the s i ze reduction, the proportion of co lor to to ta l area remained unchanged at 60.1%. I t was very l i k e l y , as mentioned, that the pasted-up Munsell co lor displays might have changed due to the project ion conversion process. The l i s t of possible sources of change were*. (1) the s l i de f i lm conversion, (2) the project ion lamp, (3) the heat absorbing f i l t e r , (4) the project ion lens, and (5) the project ion screen. As many precautions as possible had been taken to minimize th i s change. For instance, the s l i de f i lm used was the best qua l i ty Ektachrome ava i l ab le , and a l l the s l ides were processed simultaneously in the same developing batch by Kodak and the project ion lamp was mew, which meant that yel lowing from aging had not begun as ye t . The e f fects of the optics of the projector (a Kodak Carousel) and the qua l i ty of the project ion screen could not be cont ro l l ed . To check the overa l l e f fec t of these possible changes, a number of selected colors were measured in a se t t ing s im i l a r to the experimental one with a Spectroradiometer. These readings were converted to C L E . chromaticity coordinates and p lotted in the C L E . color space, table 34 shows Munsell values, o r ig ina l coordin-ates, and project ion chromaticity coordinates for f i ve sample colors and white. Figure 55 shows these samples p lotted in the C L E . space. In studies by Mehrabian and Russell (1974), which used wr i t ten descr ipt ions of s i tuat ions as s t i m u l i , two minutes were allowed 191. TABLE 34 Color imetr ic data for selected colors of o r ig ina l and projected displays used in the p i l o t study. Color Munsell Or ig inal Projected no. notation X y X y 15 5GY7/6 .3588 .4283 .4240 .4529 23 5GY7/2 .3305 .3589 .4133 .4277 26 5P7/4 .3024 .2874 .4046 .3845 32 5YR8/4 .3754 .3549 .4336 .4172 42 5B3/4 .2381 .2780 .3227 .3796 White i l l . "C" .3101 .3163 .3839 .3906 192. .50 FIGURE 55 Color sh i f t s of selected colors and white due to s l i d e conversion. 193. for the subject to "get into the mood" of the s i t ua t i on . Since at least one minute would be required for f i l l i n g out the question-naire i t was judged that, in the present s i t ua t i on , a to ta l of three minutes would be su f f i c i en t for both the viewing of the stimulus as wel l as the paper work. As a consequence, a to ta l of 16 stimuli) were tested during the a l l o t t ed data co l l e c t i on per iod . . The 80 displays described in chapter IV represented 5 leve ls of hue, 2 leve ls of value, 2 leve ls of chroma and 4 leve ls of mot i f . In the present, l im i ted se lect ion of 16 d isp lays , both leve ls of value and both leve ls of chroma were tested, while hue and motif were examined together in the form of a composite var iable ca l l ed "mothue". The four leve ls of th i s new var iable were BG-R face, GY-P landscape, YR-B bui ldings and RP-G abstract . Table 35 shows the display numbers selected as wel l as the i r associated var iables and l e ve l s . Response measures The pleasure, arousal , dominance and information rate measures as described in chapter II were used in th is study, together with the question: . . . complete th i s sentence for whatever you bel ieve about th i s s l i d e : "This i s a s l i de of ." The complete questionnaire fo r one display i s shown in appendix A. The formulation of the l a t t e r question was somewhat problematic J 194. TABLE 35 Overview of d isplay numbers used 1n p i l o t study. Hue Value Chroma Face Landscape Bui ldings Abstract BG-R High Low High Low High High Low Low 1 2 3 4 GY-P High Low High Low High High Low Low 25 26 27 28 YR-B High Low High Low High High Low Low 49 50 51 52 RP-G High Low High Low High High Low Low 73 74 75 76 PB-Y High Low High Low High High Low Low 195. in that i t should not ask d i r e c t l y about what the subject saw in the s l i de since th i s would imply that there was indeed "something" to be seen. The present wording seemed to leave open the p o s s i b i l i t y of answering: "a face", "a landscape", e t c . , as wel l as "nothing". P i l o t study administrat ion and d i rec t ions . The data co l l ec t i on took place in a classroom with l i gh t s dimmed so that there was j u s t su f f i c i en t by which to read and f i l l out the questionnaire booklet. The s l i des were projected in such a way that the stimulus on the project ion screen measured 36" x 36", and the order of presentation was the fo l lowing (with numbers re ferr ing to display numbers): 28, 76, 4, 52, 3, 75, 27, 51, 2, 26, 50, 74, 49, 1, 25 and 73. This order was worked out in such a way that no two motifs were seen sequent ia l ly . The subjects were seated in rows varying 1n distance from the screen. There were f i ve rows in a l l , with respective distances of 102", 138", 174", 210" and 246" and corresponding v isual angles of 19°, 15°, 12°, 10° and 8°. The test was introduced as an experiment 1n perception, rather than as an experiment with colors since the mention of co lor might bias the responses in favor of hue at the expense of value, chroma and moti f . The instruct ions appearing on the front cover of the test booklet (see appendix B) were read aloud and any questions about the 196. f i l l i n g out of the semantic d i f f e r en t i a l scales were answered. The subjects were f i n a l l y to ld that only three minutes would be allowed fo r each s l i d e , and that they would have to work fas t and as spon-taneously as poss ib le . Scoring The procedure used by Mehrabian and Russell (1974) and described in chapter II was used to score the pleasure, arousal , dominance and information rate measures, while for the "comment" or "motif recognit ion" quest ion, a specia l procedure was adopted. Subjects' comments were treated as a dichotdmy between (1) seeing the motif the experimenter intended, and (2) not seeing th i s intended moti f . The dichotomy wassconstructed as fo l lows: " r ight " answers were treated as one category, and "no answer" and "wrong" as the other. The response "nothing" was c l a s s i f i e d as "wrong" when appl ied to the face, landscape and bu i l d i ng , but as " r ight " when applied to the abstract. S t a t i s t i c a l analyses The design to test the ef fects of recognit ion of motif on mothue, value and chroma was a 2 x 4 x 2 x 2 between-within subject analysis of variance using recognit ion comment as the dependent var iab le . The factors were sex (2) , mothue (4) , value (2) and chroma (2) . 197. A 5 x 16 f a c to r i a l between-within subject mul t ivar iate analysis design was used to test the e f fec t of viewing distance. Pleasure, arousal , dominance and Information rate were the dependent measures, and the factors were distance (5) and displays (16). The means from the above analysis for the dependent var iables of pleasure, arousa l , dominance and information rate were plotted fo r the 16 displays to v i sua l l y ascerta in any order trend. The design to tes t the e f fects of mothue, value, chroma and sex was a 2 x 4 x 2 x 2 f a c to r i a l between-within mul t ivar iate analysis of variance design with pleasure, arousal , dominance and information rate as dependent var iab les , and the factors of sex (2 ) , mothue (4) , vaiTue (2) and chroma (2) . Results 1. Conversion of displays to s l i des Although the s t imu l i when projected did not look pa r t i cu l a r l y d i f fe rent from the or ig ina l pasted-up d isp lays, f igure 55 shows a marked s h i f t of the tes t co lo rs . Without exception, the c o l o r s -inc luding the white—moved from the purple-blue corners of the C L E . space towardsthe yel low-red area. I t i s noticeable too from examining the curve of the Planckian black body rad iator how the 198. color temperature of the white changed from approximately 6500° K to approximately 3600° K. This indicated an overa l l yel lowing e f f e c t . To estimate the re1dab11ity of the mean scores for pleasure, arousa l , dominance and information ra te , a procedure suggested by Winer (1962, pp. 124-132) was used. The formula for the r e l i a b i l i t y coe f f i c i en t was: r k M Sbetween people - M S r e s i dua l ^between people A sub-sample of 10 subjects was randomly extracted from the main sample and analyzed. R e l i a b i l i t y coe f f i c ients of .51 fo r pleasure, .75 fo r arousa l , .74 for dominance and .82 for information rate were obtained. These r e l i a b i l i t y coe f f i c ients were interpreted as an estimate of the corre la t ion of the mean scores fo r the four depen-dent variables which one would obtain in another experimental s i t u a -t ion f o r the same 16 displays but with d i f fe rent groups of subjects. 2. E f fect of recognit ion of motifs Although subjects* recognit ion scores discr iminated s i g n i f i -cantly among the four mothues (p=.0004), with means of 1.645, 1.855, 199. 1.934 and 1.789 respect ive ly, ("r ight" answers were coded as "I" while "wrong" answers were coded as " 2 " ) , and between the two levels of value (p=.0012), with means of 1.717 and 1.895 respect ive ly, subjects in general d id not recognize the motifs very readi ly ( c f . table 36). In descending order of recognit ion, the mothues were (with per cent of " r igh t " responses shown in brackets): BG-R face (35%), RP-G abstract (21%), GY-P landscape (14%) and YR-B bui ld ings (7%). The high value displays were most eas i l y recognized (at 28%) while the low value displays were the most d i f f i c u l t (10%). There were not any s i gn i f i c an t dif ferences between the two leve ls of chroma. Males did s i gn i f i c an t l y better (p = .0110) than females: 31% of males' responses indicated recognit ion of motifs (mean 1.696) while only 13% of the females' did (mean 1.870). Overa l l , 21% of the responses indicated recognit ion of the mot i fs . The best set of responses fo r any subject showed 50% recognit ion, while there were 4 subjects who did not recognize any of the mot i fs . 3. Ef fect of viewing distance The projected displays were viewed from f i ve distances: 102" - 19° (2 subjects); 138" - 15° (4 subjects); 174" - 12° (4 subjects); 210" - 10° (6 subjects) and 246" - 8° (5 subjects) . The analysis of variance resul ts showed that there were no e f fec ts on pleasure, arousa l , dominance or information rate scores due to the distance 200. TABLE 36 S ign i f i cant e f fects for motif recognit ion ana lys i s . Source SS df MS F P Sex 2.1260 1 2.1260 8.0568 .0110 Subj. error 4.4859 17 .2639 Mothue 3.4309 3 1.1436 7.3848 .0004 Mothue x Subj. error 7.8981 51 .1586 Value 2.3980 1 2.3980 15.1984 .0012 Value x Subj. error 2.6821 17 .1578 Residual error 3.7314 51 .0732 Total 47.5493 303 201. from which the subject viewed the s t i m u l i . 4. Order of presentation e f fec t Since the order of presentation of the 16 displays was the same fo r a l l 19 subjects, the standard deviations fo r pleasure, arousa l , dominance and information rate f o r th is sequence of displays were plotted in order to assess i f any order e f fec ts ex i s ted . Figure 56 shows the plots of these values. From a vlisual inspection of the curves, there does not appear to be any e f fec t s due to order of presentat ion. Regression analyses with pleasure, arousa l , dominance and information rate mean scores respect ive ly , as pred ic tor var iab les , and presentation order as c r i t e r i on var iable further confirmed th i s f i nd ing . No s i gn i f i c an t di f ferences between predictor and c r i t e r i on var iables were found. 5. Ef fects of mothue, value, chroma and sex Although the analysis of variance with mothue, value, chroma and sex was intended only as an i n i t i a l Indicat ion of what sortcdf resu l ts might be expected from a larger and more complete study, the present resu l ts are of considerable Interest 1n themselves. Only the main e f fec ts are described here, although a number o f i n t e r -actions were found to be s i g n i f i c an t . These are l i s t e d in table 37. 3 . 0 Pleasure Arousal Order e f fect on subject v a r i a b i l i t y Dominance (Displays 1n order o f presentation) Information rate FIGURE 56 Effects of order of ^ r e & n t a i i b V ^ arousal .dominance arid information rate. 203. Mothuet There were s i gn i f i c an t dif ferences among the four leve ls of mothue on pleasure (p = .0495) and on information rate (p = .0072) on ly. (See table 37). In descending order of pleasantness, the mothues were: YR-B bui ld ings (mean 0.350), GY-P landscape (mean -0.040), BG-R face (mean -0.630) and RP-G abstract (mean -0.710). In descending order of information rate , the mothues were: RP-G abstract (mean 0.395), GY-P landscape (mean 0.263), BG-R face (mean -0.106) and YR-B bu i l d -ings (mean -0.265). Value. There were only s i gn i f i c an t dif ferences between the two leve ls of value on information rate (p = .0398). The high value ( l i gh te r ) displays were scored low on information rate (mean -0.031) while the low value (darker) displays were scored high on informa-t ion rate (mean 0.174). Chroma. There were s i gn i f i c an t dif ferences between the two leve ls of chroma on arousal only (p = .0413). High chroma displays resulted in higher arousal scores (mean -0.022) than displays of low chroma (mean -0.321) although both types resulted in negative means. Sex. Difference of sex did not appear to have any s i gn i f i c an t e f fects on pleasure, arousal , dominance or information rate . 204. TABLE 37 S ign i f i can t e f fects fo r analys is with pleasure, arousa l , dominance and information ra te . Source SS df MS F P Pleasure Mothue 57.6764 3 19.2255 2.6847 .0495 Mothue x Subj. error 365.2165 51 7.1611 Residual error 83.1230 51 1 .6299 Total 1047.5011 303 Arousal Mothue x value 33.4849 3 11.1616 6.0845 .0014 Mothue x value x Subj. error 93.5567 51 1 .8344 Chroma 6.7830 1 6.7830 4.7724 .0413 Chroma x Subj. error 24.1619 17 1.4213 Residual error 66.8691 51 1.3112 Total 713.0082 303 Domi nance Value x chroma 6.1271 1 6.1271 6.9623 .0166 Value x chroma x Subj. error 14.9605 17 .8800 Residual error 38.1863 51 .7488 Total 445.5266 303 Information rate Mothue 21.7486 3 7.2495 4.4997 .0072 Mothue x Subj. error 82.1663 51 1.6111 Value 3.1830 1 3.1830 4.8523 .0398 Value x Subj. error 11.1516 17 .6560 Mothue x chroma 4.8718 3 1.6239 3.7442 .0165 Mothue x chroma x Subj . error 22.1197 51 .4337 Residual error 27.0732 51 .5308 Total 348.6579 303 205. Discussion, conclusions and modif ications  to the experimental procedure I t was c lear from the co lor imetr i c measurements o f the displays before and a f te r the s l i de conversion that the method of project ing the s t imul i was unacceptable. The marked s h i f t of the colors toward the yel low-red part of the color space (as shown in f igure 55) d i s -torted the colors to such an extent that the meticulous spacing of the Munsell colors around i l luminant "C" had been l o s t . From the general lack of recognit ion of the mot i fs , i t might be suspected that the 16 x 16 aperture matrix was too coarse. Even in t he i r o r ig ina l s ta te , the 16 x 16 displays were probably not easy to recognize—the experimenter (and others associated with the experiment) thought the motifs f a i r l y c l ea r , but th i s was most l i k e l y due to f am i l i a r i t y with the purpose of the experiment and with the s t imul i content—and the greater magnif ication of the matrix which resulted from the project ion possib ly made the motifs even more d i f f i c u l t to recognize. The resu l ts of varying the distance from the project ion screen, on the other hand, showed that recognit ion was not influenced by e i ther distance or viewing angle over the range tested. There did not seem to be any ef fects due to order of presentat ion, and i t was concluded that any order was acceptable as long as i t 206. followed the procedure of arranging the displays in such a way that no motifs were seen sequent ia l ly . In sp i te of the fact that the sample was quite smal l , and the desigh was incomplete ( in the sense that motifs and hues were confounded), i t was surpr i s ing how many of the main ef fects and interact ions were s i gn i f i c an t . Three resul ts were of pa r t i cu la r i n te res t . F i r s t , the abstract motif c l ea r l y rated higher on i n f o r -mation rate than did the three other mot i fs . This was to be expected since a high information fate indicates uncertainty, complexity, asymmetry, randomness, e tc . which i n fact the abstract was meant to convey. Secondly, there were s i gn i f i c an t dif ferences between high and low value d isp lays, again on information ra te . This confirmed that the "negative" versions of the motifs resulted in much more uncertainty than did the "pos i t ive" vers ions, which were in accordance with the way the motifs were constructed, A lso , th i s resu l t complements the motif recognit ion resul ts insofar as the well recognized version of the motifs also were found to be the ones which resulted in the lower information rate scores. Th i rd ly , chroma made a s i gn i f i can t di f ference to the way subjects scored on arousal . The resu l ts showed that the higher the chroma (or saturat ion) , the higher the arousal score. Since chroma may be interpreted as co lor fu lness, i t i s no surpr ise that 207. the more co lor fu l a display was, the more arousing i t was found to be. The p i l o t study gave the wr i te r an opportunity to tes t a number of ideas pr imar i ly re lated to the mechanics of s t imu l i and presentat ion, and to modify the main study in accordance with the resu l t s . I t was concluded that , f o r the subsequent experiment, the project ion mode of presentation should be abandoned, and that the aperture matrix somehow be made f i n e r . On the pos i t ive s ide , the question re la t ing to recogn izab l l i ty seemed to work, the procedure for constructing the order of presentation was acceptable and, judging from the responses to the main e f fects of mothue, value and chroma, there was su f f i c i en t encouragement to proceed with a rev ised, more complete experiment. 208. CHAPTER VI MAIN STUDY (EXPERIMENT WITH AFFECTIVE AND COGNITIVE RESPONSES) The sample The sample consisted of 82 students enro l led in a second year psychology course at The Univers i ty of B r i t i s h Columbia. There were 41 females and 41 males. The overa l l mean age was 21.7 years (range 18 to 39 years) , the mean age fo r males was 21.2 years (range T8 to 28 years), and fo r females 22.1 years (range ?8 to 39 years ) . The experimental s i tuat ion The experiment took place on October 10, 1978, in a large classroom with windows facing north, The weather outside was s l i g h t l y overcast but br ight , and the corrected color temperature fo r the experimental area was approximately 5500° K. Only dayl ight was used fo r the experiment. 209. The subjects were aitflowed to seat themselves at random 1n the seven rows of chairs c losest to the windows. The l i gh t leve l var ied from 860 lux (lumen/m ) near the windows to 215 lux far thest away from the windows. This spread was reduced to between 365 and 590 lux by ins t ruc t ing the subjects to hold the co lor s t imul i at a spec i f ied angle to the l i gh t : perpendicular to the l i g h t close to the windows, and progressively facing the l i gh t more as the distance from the windows increased. The co lor temperature and the l i g h t leve ls were measured with both an "EEL" (Evans Electroselenium Ltd.) and a "Spectra" model FC-200 photometer (made by Photo Research). The s t imul i The s t imu l i consisted of the 80 displays described i n chapter IV with aperture masks superimposed. Conclusions from the p i l o t study necessitated a change in (1) the mode of presen-tat ion and (2) the aperture mask which was placed over the o r ig ina l pastedpup d i sp lays . F i r s t , the displays were presented in t he i r o r ig ina l s ta te , as "cards", rather than as projected s t imu l i , since the color sh i f t s due to the project ion process were simply too great to be acceptable. That i s , the careful and systematic placement of the Munsell samples around the "C" i l luminant—which was one of the main v i r tues of the construction of the displays—was mostly l os t through the project ion process. 210. Second, there were doubts about the recogn izab i l i ty of the motifs a f ter the resul ts of the p i l o t study. This, together with further informal test ing of the card s t imul i with 16 x 16 aperture masks showing that the motifs were not recognized unless they were seen from a distance of at least 6 f ee t , prompted a decision to increase the fineness of the aperture matrix from 16 x 16 to 32 x 32. Since i t was impract ical to reconstruct the pasted-up displays in a 32 x 32 pattern, a 32 x 32 mask was placed on top of the or ig ina l pasted-up d isp lays . This procedure, while reducing the matrix s i z e , did not substant ia l l y change the "color information" in the d isp lays . I t was found here, again through informal t e s t s , that in most cases motifs were ROW d iscern ib le i f the subject held the display at arms' length (approx. 24 inches). The revised matrix pattern therefore seemed sui tab le fo r the type of test ing proposed. The aperture mask i t s e l f was made from a sheet of g lossy, black 3M "Colorkey" material which had been exposed and developed with the required pattern of round apertures. To tes t the extent to which the "Colorkey" material might d i s t o r t the Munsell co lors , co lor imetr ic analyses of a numberfof colors were carr ied out. Table 38 shows the t r i s t imulus values and the chromaticity coordinates for a BaSO^ white, and the Munsell colors 5GY 5/4, 5B 5/4 and 5R 5/4, both with and without the "Colorkey" 211. TABLE 38 Color imetr ic data for selected display colors with and without "Colorkey". co lor Without "Colorkey" With "Colorkey II number X Y Z x y X Y Z x y White (BaS04) 97.783 100.301 116.741 .3105 .3185 84.721 86.962 99.447 .3124 .3207 5GY5/4 (#20) 16.225 19.159 11.297 .3475 .4104 13.790 16.063 10.393 .3426 .3991 5B5/4 (#41) 16.984 19.769 30.939.2509.2919 14.223 16.427 25.106.2550.2946 5R5/4 (#13) 23.077 20.132 19.266 .3693 .3222 19.105 16.770 16.180 .3670 .3221 Average 1 6 358 Average 193 34Zlx = .0003 ZIy = .0017 212. f i lm superimposed. The measurements were done on a Zeiss RFD-3 c o l o r i -meter using a 2 mm aperture and a gloss t rap. The values throughout are referenced to the "C" i l luminant . Figure 57 shows these four sets of measurements and the i r co lor sh i f t s plotted in the C L E . co lor space. To further i l l u s t r a t e the s h i f t , the ref lectance values of the white, with and without the "Colorkey" mater ia l , are graphica l ly represented in f igure 58. In order to quantify the extent of the c o l o r sh i f t , a co lor di f ference formula--the Judd-Hunter formula which was described in deta i l in chapter IV—was used to calculate the perceptible d i f f e r -ences between colors with and without the overlay mater ia l . The fol lowing dif ferences (AE) were obtained: The dif ferences fo r the three colors were found to be neg l i g i b l e , 3 units being regarded as a commercially acceptable d i f ference. Furthermore, th i s 3 unit d i f ference refers to ca re fu l l y contro l led laboratory condit ions, and in pract ice th i s resu l t w i l l vary. Figure 58 shows a s l i gh t loss 1n the blue part of the spectrum for ? Sample AE (NflB.S. units) white (BaSO.) 5GY 5/4 5B 5/4 5R 5/4 6.837 4.147 3.905 3.602 213. .42 .40 J .38 A .36 .34 -.32 .30 J .28 T5GY 5/4 White 5R 5/4 •^5B 5/4 .24 .26 .28 i r .30 .32 x .34 .36 .38 FIGURE 57 Color s h i f t of three colors and white due to "Colorkey" over lay. 214. l .cH-.2 H 400 500 600 700nm Wavelength White without "Colorkey" White with "Colorkey" FIGURE 58 Spectral ref lectance curves for white sample with and without "Colorkey" overlay mater ia l . 215. white as a resu l t of the f i lm over lay. However, th i s discrepancy i s very smal l . In general the co lor dif ferences were mainly due to a s l i gh t brightness reducing e f fec t of the f i l m , i . e . , the f i lm makes the colors look s l i g h t l y darker than the o r ig ina l Munsell samples, but th i s e f fec t isalmost evenly d is t r ibuted across the spectrum. As a r e su l t , a l l colors would be affected to the same extent. The s i ze of the matrix display was 4" x 4", the aperture diameter 3/32", and the distance between apertures 1/32". The tota l area of the display was 16 sq . i n . , of which 7.13 sq. i n . were taken up by the colors and 8.87 sq . i n . by the black mask. The proportion of co lor to mask was 80.4%* The s i ze of the overa l l stimulus ("card") was 4-7/8" x 5-7/8" (horizontal format), and in the upper r ight hand corner of the card was an arrow ind icat ing which edge was up, and also showing the number of the d isp lay, (see f igure 59). Response measures To obtain scores on the dimensions of pleasure, arousa l , dominance and information ra te , the exact same questionnaire as that used in the p i l o t study was used here. The question e l i c i t i n g responses about recognit ion of the motif was also unchanged except that the word "card" was subst i tuted fo r " s l i de " and instead of * Proportion of co lor to to ta l area was 44.6%. 216. 9 9 9 | | I I M • • • f • •• • •• ft I • ••• • ••• 9 9 9 9 0 • • • • • I • f) tt fl • • • • f 9 • •••• g ta S g • #••••< • •••••< • •••• 99 • • • I • I!1 • I • • • I • t 11 A w 999 999 • •• • • • • ••• • ••• • ••• • •••• • •••• « | | I • ' ? • • | | | | g g • • « • • ••• • ••• 9 'tf' 9) 9 9 9 0 ft u ii? • ••• • ••• • ••• • ••• >••••• • ••• • ••• • ••• • ••• • ••• • ••• FIGURE 59 Display no. 1 (face) with 32x32 matrix mask. 217. "This i s a s l i de of ", the sentence "This i s a card showing " was used. The word "showing" seemed a l i t t l e more open-ended than did "of"; "of" seemed to lead to a too obvious object conclusion. Test administration and d i rect ions The displays were held in such a way that dayl ight var iat ions were kept to a minimum, and subjects were instructed accordingly, row by row. Also the viewing distance of 24 inches or arms' length was mentioned. As in the p i l o t study, the instruct ions—which as a resu l t of comments during the p i l o t study had been c l a r i f i e d s l i g h t l y - -were read a loud, and any questions regarding the questionnaire answered. The fact that the displays embodied very subtle d i f ferences—that several of the displays might seem to look alike—was emphasized, as was the need to look at the displays as much as possible to avoid working from memory. F i na l l y , the time l i m i t of 3 minutes fo r each display was mentioned. Appendix C shows the t i t l e page of the tes t booklet. The 80 displays which had previously been arranged in random order were d is t r ibuted to the students. A f ter the f i r s t d isplay had been seen, the students handed the display to t he i r neighbor, and so on. Two of the students did not get any displays the f i r s t t ime, but the second time one of them saw one, and the t h i r d time 218. the second student saw one. In the meantime, students at the beginning o f the row were without d i sp lays . In a l l , each student ended up seeing an average of 12 d isp lays, with a few as low as 10 and as high as 13. The order of presentation was as fo l lows: 28, 76, 4, 52, 3, 75, 27, 51, 2, 26, 50, 74, 49, 1, 25, 73, 32, 80, 8, 56, 7, 79, 31, 55, 6, 30, 54, 78, 53, 5, 29, 77, 36, 64, 12, 60, 11, 63, 35, 59, 10, 34, 58, 62, 57, 9, 33, 61, 40, 68, 16, 44, 15, 67, 39, 43, 14, 38, 42, 66, 41, 13, 37, 65, 24, 72, 20, 48, 19, 71, 23, 47, 18, 22, 46, 70, 45, 17, 21 and 69. Scoring The same scoring procedure used in the p i l o t study was also used here. Scoring of the pleasure, arousa l , dominance and informa-t ion rate dimensions followed the pract ice of Mehrabian and Russell (1974) as described in chapter I I , while the "motif recognit ion" question was scored as " r igh t " (2) or as "wrong" (f|) depending on whether the answers f e l l in to one of two categories: " r i gh t " (or "nothing" fo r the abstract) ind icat ing that the subject had seen the motif which the experimenter intended and "wrong" (or "notfiitng" when applied to the face, landscape or bui ld ings) ind icat ing that the subject had not seen th is intended moti f . 219. S t a t i s t i c a l analyses In order to test the hypothesis pertaining to the display var iables of chroma, motif and the subject var iable of sex, as stated in chapter I I I , a mul t ivar ia te analys is of variance in the form of a completely randomized f i ve - fac to r design with f ixed ef fects was carr ied out. I t was noted that there was not s t r i c t independence nor s t r i c t depencence among the c e l l s , and that corre lat ions were expected to be pos i t i ve . This design thus ensured a conservative t e s t . The four dependent measures in the mul t ivar iate model were pleasure, arousal , dominance and information rate. In addit ion to the main ana lys is , selected contrasts among treatment means were calculated using a program option requesting mult ip le range t e s t s . The Duncan mutiple range tes t , at a spec i f i ed s ign i f i cance leve l of .05 was chosen. To tes t the hypothesis re la t ing motif recognit ion to the display variables of hue, value, chroma, motif and the subject var iable of sex, an analysis of variance of the same form as the main mu l t i -var iate analysis but with motif recognit ion as the sole dependent var iable was carr ied out. In order to probe the extent of correspondence between motif recognit ion comments and information rate scores, a regression analysis was performed with motif recognit ion mean scores f o r the 80 displays as the c r i t e r i on var iable and information fcate mean scores 220. for the 80 displays as jbhe predictor var iab le . Step-wise regression analyses were f i n a l l y carr ied out to Investigate l i nea r and quadratic re lat ionships among the c r i t e r i o n , variables of pleasure, arousal , dominance and information ra te , and the predictor var iables developed in chapter IV, sect ion I I . The complete set of predictor variables (together with the i r references) were: Average-A E (table 25) Average-value (table 25) Average-chroma (table 25) Average-temperature (table 25) Figure/background-A E (table 26) Figure/background-value (table 26) Figure/background-chroma (table 27) Figure/background-temperature (table 27) Top/bottom- AE (table 28) Top/bottom-value (table 28) Top/bottom-chroma (table229) Top/bottom-temperature (table 29) L e f t / r i g h t - dE (table 30) Lef t/r ight-va lue (table 30) Left/right-chroma (table 31) Left/right-temperature (table 31) Adjacent/di f ference-A E (table 32) Adjacent/difference-value (table 33) Adjacent/dlfference-chromaj(table 33) Adjacent/difference-temperature (table 33) Adjacent/variance-A E (table 32) Adjacent/variance-value (table 33) Adjacent/variance-chroma (table 33) Adjacent/variance-temperature (table 33) 221. CHAPTER VII RESULTS This chapter 1s divided Into two par ts . The f i r s t part reports the resu l ts of the various analyses only, while 1t does not assess them ; The second part interprets these resu l ts in l i g h t of the hypotheses advanced in chapter I I I . I REPORTING OF THE RESULTS Display descr ipt ion based on mean scores fo r pleasure, arousal  dominance and information rate . The pleasure, arousal dominance and information rate measures can, as mentioned in chapter I I , be used as yet another form of descr ipt ion of the tes t d i sp lays . (Mehrabian and Russe l l , 1974) The complete 11st of the mean scores as wel l as t he i r standard deviations for these four dimensions are therefore shown in table 39. The overa l l 222. TABLE 39 Means and standard deviations for 80 d i sp lays . Display no. Pleasure Arousal Mean Std . Dev. Mean Std . Dev. Dominance Information rate Mean Std . Dev. Mean Std . Dev. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 -0.657 -0.239 0.416 0.110 -0.247 -0.387 -0.369 -1.142 -0.002 -0.192 -1.157 0.112 -0.533 -0.737 -0.761 -0.713 -0.353 -0.454 -1.244 -0.215 0.432 0.247 -0.906 -1.052 1.253 1.332 0.558 0.310 1.243 -0.706 •0.130 0.262 0.025 0.667 0.329 0.537 1.248 0.282 -0.266 -0.104 2.688 1.449 2.059 1.883 1.818 1.278 1.366 1.417 1.717 1.903 1.820 2.181 1.729 1.442 1.911 1.816 1.534 2.054 1.402 1.838 1.852 1.513 1.579 1.636 2.032 2.153 2.151 2.170 1.831 1.402 1.741 2.372 2.382 2.017 2.322 1.780 2.025 2.533 1.937 1.811 0.343 0.742 0.324 0.045 0.374 -0.469 0.146 -0.278 0.036 -0.538 -0.394 -0.319 -0.653 0.635 0.602 0.812 0.086 0.559 0.279 -0.499 -0.355 -0.439 0.551 -0.880 -0.611 -0.455 -0.635 -0.313 -0.401 -1.039 -0.749 -0.367 -0.974 0.227 -0.631 -1.118 -0.694 -1.326 •0.882 -0.124 1.959 1.512 1.791 1.870 1.263 1.195 1.499 1.499 1.838 1.373 1.601 1.645 1.531 1.981 1.477 2.173 0.968 2.016 1.625 1.372 1.729 1.510 1.840 2.131 1.118 1.481 1.699 2.051 1.106 1.468 1.407 1.979 1.399 1.039 1.807 1.339 1.757 1.391 1.679 1.486 0.616 0.196 -0.147 -0.436 -0.359 -0.562 -0.801 -0.548 0.363 0.436 -0.452 -0.258 0.027 -0.274 0.054 -0.801 -0.035 0.001 -0.372 0.374 -0.345 -0.328 -0.578 -0.240 0.596 0.232 0.736 0.065 -0.050 -0.311 -0.588 -0.319 0.325 0.121 0.051 0.000 •0.115 0.312 -0.826 •0.189 1.720 1.560 1.755 1.381 0.689 1.346 1.336 1.229 0.965 0.964 1.201 1.274 1.696 1.042 1.253 1.376 0.886 1.797 1.138 1.285 1.067 0.831 0.581 1.153 1.043 1.250 1.479 1.545 0.776 1.114 1.242 1.839 1.183 1.402 1.351 1.161 1.471 1.413 1.547 0.846 -0.716 0.458 0.255 0.749 -0.496 -0.150 0.072 0.390 -0.290 -0.700 0.384 0.423 -0.059 -0.333 0.880 0.494 -0.127 0.513 0.257 -0.070 -0.243 -0.347 -0.003 0.669 -0.330 •0.067 -0.248 0.111 -0.387 0.271 -0.453 0.129 -0.147 -0.188 0.091 0.080 -0.576 0.028 -0.149 0.560 1.230 1.078 1.288 1.071 1.047 0.843 1.179 1.113 0.691 0.639 1.189 1.360 1.147 1.062 1.267 1.075 0.908 1.209 0.833 1.225 0.639 0.820 1.050 1.355 0.575 0.871 1.022 1.228 0.936 0.950 1.181 1.366 1.491 1.100 1.263 1.177 1.243 0.953 1.561 0.763 . continued 223. Table 39 continued Display no. Pleasure Mean Std . Dev. Arousal Mean Std . Dev. Dominance Information rate Mean Std. Dev. Mean S td . Dev. 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 1.146 0.179 0.105 •0.503 0.185 -0.240 -1.139 -0.399 1.211 -0.371 0.831 -0.478 0.522 0.162 0.229 0.172 1.082 •0.375 -0.635 -1.177 -0.530 •0.201 0.733 •0.543 0.878 •0.019 •0.375 0.016 -1.286 •0.479 0.046 •0.350 0.185 0.103 -0.394 -0.217 0.550 •0.456 0.934 0.618 1.329 1.711 2.129 2.103 1.681 2.088 1.707 1.884 1.867 1.636 1.351 1.793 1.967 1.962 1.758 1.936 1.973 0.865 1.802 0.951 1.833 2.066 1.955 1.485 1.802 1.613 2.349 1.909 1.142 1.850 2.115 1.942 1.459 1.676 1.822 1.852 1.767 1.697 2.282 2.057 0.027 0.008 -0.271 -0.176 0.136 0.510 0.383 -1.377 -0.140 0.096 -0.632 -0.139 0.038 -0.082 0.309 -0.432 -0.075 0.486 -0.441 -0.524 -1.103 0.222 -0.333 -0.001 1.321 0.185 0.673 0.514 0.832 0.221 -0.423 -1.144 0.292 0.259 0.929 0.537 0.465 0.537 0.238 0.710 1.189 1.294 1.143 1.862 1.958 1.277 2.178 1.052 1.651 1.496 1.766 1.380 1.851 1.244 1.780 1.431 1.789 1.473 1.252 1.543 1.345 1.895 2.102 1.314 1.244 1.828 2.367 1.707 2.143 2.163 1.448 1.425 1.629 1.449 1.579 1.843 1.591 1.413 2.019 1.700 -0.213 -0.349 -0.262 -0.245 -0.712 0.223 -0.231 -0.596 0.170 0.255 0.247 -0.071 0.335 0.500 0.480 -0.094 -0.032 -0.403 -0.207 -0.005 0.424 0.375 0.103 0.121 -0.617 0.279 -0.773 -0.388 -0.422 -0.209 0.390 -0.477 0.065 -0.231 0.305 -0.284 -0.066 -0.762 -0.193 0.149 1.270 1.419 1.976 1.264 1.368 1.161 1.757 1.575 1.325 1.659 1.953 1.542 1.204 1.434 1.407 1.164 1.156 1.188 1.112 1.339 0.841 1.077 1.239 1.014 1.151 1.037 0.663 1.616 1.040 0.851 0.813 1.082 1.089 1.354 1.378 1.446 0.761 1.071 1.620 0.818 -0.302 -0.644 -0.103 -0.273 -0.249 -0.269 -0.153 -0.684 -0.099 -0.247 -0.203 -0.179 -0.110 •0.401 -0.168 -0.113 -0.595 0.088 -0.345 0.204 -0.149 0.132 0.232 0.022 -0.158 -0.050 0.307 0.214 0.503 0.245 -0.141 0.033 0.243 0.568 0.147 -0.221 0.100 1.143 0.490 0.208 0.692 0.979 0.963 1.233 1.166 0.706 0.698 1.216 0.870 1.038 0.813 1.176 0.545 0.693 0.967 0.719 0.742 0.845 0.924 1.142 1.030 1.035 0.996 1.530 1.004 1.228 1.306 0.784 0.781 1.677 0.920 1.165 1.351 1.240 1.245 1.590 1.297 1.336 0.878 1.300 TABLE 40 Means, standard deviations and correlat ions f o r pleasure, arousa l , dominance, information rate and motif recognit ion. Correlations Overal1 Std . Information Motif mean dev. Pleasure Arousal Dominance rate R e c o g n i t i o n Pleasure -.0428 1.8857 1.0000 Arousal -.0827 1.6694 -.1252 1.0000 Dominance -.0999 1.2923 .3012 -.1469 1.0000 Information rate .0009 1.1080 -.2932 .2778 -.2640 1.0000 Motif recognit ion 1.5197 .4999 -.0583 .1768 .2252 -.3958 1.0000 225. means (which are used as the neutral point for the scales of pleasure, arousal and dominance) and standard deviations as wel l as co r re l a -tions among pleasure, arousal , dominance, Information rate and motif recognit ion are shown in table 40. EMOTIONAL MEASURES AND INFORMATION RATE Results of the analysis of variance with the d isp lay variables of  hue, value, chroma, motif and the subject var iable sex as independent  var iables and pleasure, arousal , dominance and information rate as  dependent var iab les . Tables 41, 42, 43 and 44 show the four analysis of variance tables for the four dependent var iables of pleasure, arousal , dominance and Information rate. 226. TABLE 41 Summary of analys is of variance fo r pleasure. Source SS df MS F P Hue (H) 1 .7893 4 .4473 .1315 .9674 Value (V) 22.3086 1 22.3086 6.5575 .0101* Chroma (C) 14.7206 1 14.7206 4.3270 .0351* Motif (M) 61.6585 3 20.5528 6.0414 .0005* Sex (S) 33.7593 1 33.7593 9.9234 .0018* HV 12.3772 4 3.0943 .9096 .4557 HC 6.4958 4 1 .6240 .4774 .7530 HM 85.7718 12 7.1476 2.1010 .0143* HS 5.4952 4 1.3731 .4036 .8063 VC 6.3663 1 6.3663 1.8713 .1665 VM 29.5111 3 9.8370 2.8915 .0332* VS 1.1586 1 1.1586 .3406 .5658 CM 18.2693 3 6.0898 1.7901 .1435 CS 1.8043 1 1.8043 .5304 .4719 MS 24.2137 3 8.0712 2.3725 .0482* HVC 15.1306 4 3.7827 1.1119 .3462 HVM 44.7129 12 3.7261 1.0953 .3543 HVS 23.1091 4 5.7773 1.6982 .1445 HCM 34.4867 12 2.8739 .8448 .5990 HCS 11.6544 4 2.9136 .8564 .4882 HMS 25.1878 12 2.0990 .6170 .8262 VCM 2.8465 3 .9488 .2789 .8404 VCS 2.2216 1 2.2216 .6530 .7320 VMS 13.0336 3 4.3445 1 .2770 .2775 CMS 8.4717 3 2.8239 .8301 .4771 HVCM 34.2572 12 2.8548 .8392 .6050 HVCS 23.4701 4 5.8675 1.7247 .1386 HVMS 42.8074 12 3.5673 1.0486 .3954 HCMS 55.6982 12 4.6415 1.3643 .1725 VCMS 3.0083 3 1.0028 .2948 .8293 HVCMS 18.6012 12 1.5501 .4556 .9382 Error 2650.1672 779 3.4020 Total 3334.5615 938 * Indicates s ign i f i cance . 227. TABLE 42 Summary of analysis of variance for arousa l . Source SS df MS F P Hue (H) 14.9339 4 3.7335 1.3775 .2290 Value (V) 4.0432 1 4.0432 1.4917 .2150 Chroma (C) 1.1351 1 1.1351 .4188 .5209 Motif (M) 87.9644 3 29.3215 10.8181 .0000* Sex (S) 1.3165 1 1.3165 .4857 .4886 HV 8.4633 4 2.1158 .7806 .5297 HC 10.8003 4 2.7001 .9962 .3984 HM 41.6487 12 3.4707 1.2805 .2083 HS 1.3302 4 .3325 .1227 .9702 VC 4.2312 1 4.2312 1.5611 .2043 VM 9.1116 3 3.0372 1.1206 .3305 VS .0831 1 .0831 .0307 .8368 CM .2176 3 .0725 .0267 .9891 CS 1.6981 1 1.6981 .6265 .4297 MS 7.6157 3 2.5386 .9366 .4145 HVC 14.7618 4 3.6905 1.3616 .2345 HVM 35.3139 12 2.9428 1.0857 .3495 HVS 6.5759 4 1.6440 .6066 .6522 HCM 39.9310 12 3.3276 1.2277 .2415 HCS 6.8830 4 1.7208 .6349 .6315 HMS 50.7159 12 4.2263 1.5593 .0805 VCM 2.7903 3 .9301 .3432 .7916 VCS 1.1013 1 1.1013 .4063 .5273 VMS 7.7668 3 2.5889 .9552 .4053 CMS 5.8190 3 1.9397 .7157 .5374 HVCM 31.9981 12 2.6665 .9838 .4431 HVCS 8.3740 4 2.0935 .7724 .5353 HVMS 15.1123 12 1.2594 .4647 .9301 HCMS 11.0318 12 .9193 .3392 .9796 VCMS 3.8646 3 1.2882 .4753 .6968 HVCMS 56.4364 12 4.7030 1.7352 .0587 Error 2111.4403 779 2.7104 Total 2604.5094 938 * Indicates s ign i f i cance . 228. TABLE 43 Summary of analys is of variance fo r dominance. Source SS df MS F P Hue (H) 6.5340 4 1.6335 .9771 .4013 Value (V) 6.5345 1 6.5345 3.9092 .0421* Chroma (C) .3920 1 .3920 .2345 .6282 Motif (M) 1.0055 3 .3352 .2005 .8900 Sex (S) 18.2962 1 18.2962 10.9440 .0010* HV 1.6645 4 .4161 .2489 .9043 HC 7.4325 4 1.8581 1.1114 .3313 HM 52.4943 12 4.3745 2.6166 .0014* HS 16.2445 4 4.0611 2.4292 .0399* VC .0701 1 .0701 .0419 .8172 VM 5.4325 3 1.8108 1.0831 .3398 VS .3291 1 .3291 .1969 .6710 CM .2746 3 .0915 .0547 .9767 CS .4662 1 .4662 .2789 .5981 MS 4.4404 3 1.4801 .8853 .4342 HVC 9.0052 4 2.2513 1.3466 .2331 HVM 12.8321 12 1.0694 .6397 .7895 HVS 8.6677 4 .2167 .1296 .9667 HCM 17.2485 12 1 .4374 .8598 .5576 HCS 6.7504 4 .1688 .1010 .9777 HMS 28.2659 12 2.3555 1.4090 .1340 VCM 2.8537 3 .9512 .5690 .6265 VCS .0842 1 .0842 .0504 .8045 VMS 3.2565 3 1.0855 .6493 .5731 CMS 9.6015 3 3.2005 1.9144 .1135 HVCM 7.9980 12 .6665 .3987 .9587 HVCS 4.1310 4 1.0327 .6177 .6376 HVMS 4.9557 12 .4130 .2470 .9945 HCMS 12.1333 12 1.0111 .6048 .8219 VCMS 4.7593 3 1.5864 .9489 .4017 HVCMS 24.5881 12 2.0490 1 .2256 .2315 Error 1302.3537 779 1.6718 Total 1566.5623 938 * Indicates s ign i f i cance . 229. TABLE 44 Summary of analys is of variance for information r a te . Source SS df MS F P Hue (H) 5.0279 4 1.2570 1.0497 .3607 Value (V) 9.6558 1 9.6558 8.0633 .0040* Chroma (C) 5.2969 1 5.2969 4.4233 .0306* Motif (M) 25.8279 3 8.6093 7.1894 .0001* Sex (S) .3778 1 .3778 .3155 .5746 HV 3.7452 4 .9363 .7819 .5201 HC 8.6837 4 2.1709 1.8129 .1103 HM 6.1419 12 .5118 .4274 .9454 HS 5.1020 4 1.2755 1.0651 .3527 VC 1.0664 1 1.0664 .8905 .3388 VM 15.7481 3 5.2494 4.3836 .0038* VS 1.0080 1 1.0080 .8418 .3529 CM 4.5899 3 1.5300 1 .2777 .2636 CS .2772 1 .2772 .2315 .6298 MS 6.1115 3 2.0372 1.7012 .1504 HVC 4.2654 4 1.0664 .8905 .4504 HVM 15.6316 12 1.3026 1.0878 .3321 HVS .0432 4 .0109 .0091 .9991 HCM 12.8360 12 1.0697 .8933 .5192 HCS 10.8651 4 2.7163 2.2683 .0517 HMS 17.0744 12 1.4229 1.1882 .2545 VCM 3.5140 3 1.1713 .9781 .3862 VCS .4746 1 .4746 .3963 .5289 VMS 1.3059 3 .4353 .3635 .7724 CMS 2.3251 3 .7750 .6472 .5732 HVCM 6.8074 12 .5673 .4737 .9203 HVCS 3.8227 4 .9557 .7981 .5093 HVMS 22.8613 12 1.9051 1.5909 .0728 HCMS 6.2141 12 .5178 .4324 .9430 VCMS 7.5228 3 2.5076 2.0940 .0543 HVCMS 2.7445 12 .2287 .1910 .9983 Error 932.8383 779 1.1975 Total 1149.8099 938 * Indicates s ign i f i cance . 230. S ign i f i can t main e f fects Table 45 presents an overview of the var iables tested in the analysis i n terms of the s i ze of the F-stat is t1c of the main e f f e c t s . Aster isks indicate that s ign i f i cance was obtained. HUE I t was hypothesized that the f i ve leve ls of hue would show s i gn i f i c an t l y d i f f e rent mean scores on pleasure, arousa l , dominance and information ra te . However, no such dif ferences were found for any of the four dependent var iab les . VALUE The mean scores fo r value on pleasure, arousal , dominance and information rate were: Value Pleasure Arousal Dominance Information rate High .111 -.017 -.017 -.100 Low -.197 -.148 -.183 .102 Pleasure To tes t the e f fec ts of value on pleasure, the two value leve ls of "predominantly high" and "predominantly low" were presented. 231. TABLE 45 Size of F-stat is t1c due to main e f f e c t s . Information Variable/Levels Pleasure Arousal Dominance rate Hue (5) .1315 1.3775 .9771 1.0497 Value (2) 6.5575* 1.4917 3.9092* 8.0633* Chroma (2) 4.3270* .4188 .2345 4.4233* Motif (4) 6.0414* 10.8181* .2005 7.1894* Sex (2) 9.9234* .4857 10.9440* .3155 * Indicates s ign i f i cance . 232. I t was found that the two leve ls of value did d i f f e r s i gn i f i c an t l y (p = .0101), and that predominantly high value ( l i gh te r ) displays resulted in greater pleasure (mean .111) than predominantly low value (darker) displays (mean - .197). Arousal The two leve ls of value did not s i gn i f i c an t l y inf luence the subjects ' scores on arousa l . Dominance There were s i gn i f i c an t dif ferences between the e f fec ts of the two value configurations on dominance (p = .0421). Predominantly high value displays resulted in subjects fee l ing more in control of the s i tua t ion (mean -.017) than did the predominantly low value displays (mean -.183). However, in both s i tuat ions the means were lower than the overa l l mean fo r dominance of -.0999. Information rate The two value leve ls d i f fe red s i gn i f i c an t l y on information rate scores (p = -0040K Predominantly high value displays resulted in lower rates of information (mean -.100) while predominantly low value displays resulted in higher rates of information (mean .102). 233. CHROMA The mean scores for chroma on pleasure, arousal , dominance and information rate were: Chroma Pleasure Arousal Dominance Information rate High .083 -.048 -.079 -.075 Low -.167 -.117 -.120 .076 Pleasure Two levels of chroma, predominantly high and predominantly (low, were tested to assess the extent of the e f fects on pleasure. I t was found that these two leve ls of chroma did s i gn i f i c an t l y inf luence scores on pleasure (p = .0351). Predominantly high chroma (saturated) displays resulted in higher pleasure rat ings (mean .083) than low chroma (desaturated) displays (mean - .167). Arousal There were no s i gn i f i c an t differences due to chroma on arousal . Dominance There were also no s i gn i f i c an t differences due to chroma on dominance. 234. Information rate The two chroma leve ls resulted in s i gn i f i c an t l y d i f f e rent scores on information rate (p = .0306). Predominantly high chroma displays resulted in low mean rates of information (-.075) while predominantly low chroma displays resulted in high mean rates of i n fo rmati on.(.076). MOTIF The mean scores for motif on pleasure, arousa l , dominance and information rate were: Motif Pleasure Arousal Dominance Information rate Face -.437 .092 -.144 .099 Landscape .274 -.560 -.069 -.055 Bui ldings .036 -.121 -.068 -.241 Abstract -.043 .256 -.118 .198 Pleasure To test the e f fects which motif might have on subjects ' responses on the dimension of pleasure, four motifs were used: a face, a landscape, bui ldings and an abstract . I t was found that there were s i gn i f i c an t differences among the motifs on pleasure (p = .0005). Furthermore, Duncan's mult ip le range tests showed the fo l low-ing homogeneous subsets among the means (p<.05) . In decreasing order 235. of contr ibut ion to the overa l l s ign i f i cance of F they were: face-landscape, face-bui ldings and face-abstract. A lso, the contrast between the face and the other three motifs was s i gn i f i c an t . However, no s i gn i f i c an t di f ference was found between the three representa-t iona l motifs (face, landscape and bui ld ings) and the abstract . In decreasing order of pleasantness, the moti f means were: landscape .274, bui ld ings .036, abstract -.043 and face -.437. Arousal There were s i gn i f i c an t dif ferences on the arousal scores due to the four motifs (p<.0000), and Duncan's mult ip le range tests showed homogeneous subsets among the fol lowing f i ve contrasts, in order of decreasing contr ibut ion to the s ign i f i cance of the overa l l F (p<.05): landscape^abstract, face-landscape, landscape-buildings and bui ld ings-abstract . Also the contrast between the representa-t iona l motifs of face, landscape, bui ld ings and the abstract was found to be s i gn i f i c an t . In decreasing order of arousal , the means of the four motifs were: abstract .256, face ,092, bui ld ings -.121 and landscape -.560. Dominance There were no s i gn i f i c an t dif ferences among the four motifs on dominance. 236. Information rate The information rate scores were s i gn i f i c an t l y inf luenced by the four motifs (p = .0001), and Duncan's mult ip le range tests showed homogeneous subsets among f i ve contrasts (P>< .05). These were, jn decreasing order of contr ibut ion to the s ign i f i cance of the over-a l l F: bu i ld ings-abstract , face-bui ldings and landscape-abstract. A s i gn i f i c an t di f ference was also found between the three representa-t iona l motifs ( face, landscape, bui ld ings) and the abstract . In decreasing order of information ra te , the four motif»means were: abstract .198, face .099, landscape -.055 and bui ldings - .241. SEX The mean scores for male and female subjects on pleasure, arousa l , dominance and information rate were: Sex Pleasure Arousal Dominance Information rate Males -.235 -.121 .041 .021 Females .145 -.046 -.238 -.019 Pleasure There were s i gn i f i c an t di f ferences between responses of males and females on the dimension of pleasure (p = .0018). Females gen-e ra l l y found the displays more pleasing (mean .145) than the males (mean - .235) . 237. Arousal Sex differences didrmot s i gn i f i c an t l y account for any differences in arousal scores. Dominance Males and females showed s i gn i f i c an t l y d i f fe rent scores on the dimensions of dominance (p = .0010). Males f e l t more in control of the s i tua t i on (mean .041) than did females (mean - .238). Information rate There were no s i gn i f i c an t differences between the scores of males and females on information rate . Interact ion of main ef fects A tota l of s i x interact ions between the dependent and independent variables were found to be s i gn i f i c an t , and f igure 60 shows which dependent variables were s i gn i f i c an t as a resu l t of these in te rac t ions . 238. Hue Value Chroma Moti f Sex Hue Pleasure Dominance Domi nance Value Pleasure Information rate Chroma Motif Pleasure Sex FIGURE 60 S ign i f i can t interact ions of main ef fects on pleasure, arousa l , dominance and information ra te . 239. HUE BY MOTIF Pleasure The mean scores for pleasure in the hue by motif in teract ion were: Face Landscape Bui ldings Abstract BG-R -.092 -.349 .232 -.118 GY-P -.546 .863 -.359 .115 YR-B -.292 .150 .298 -.529 RP-G -.686 .387 .269 -.078 PB-Y -.571 .290 -.276 .397 I t was found e a r l i e r that hue by i t s e l f had no s i gn i f i c an t e f fec ts on pleasure but that motif d i d . In decreasing order of pleasantness, the four motifs were: landscape, bu i ld ings , abstract and face. The overa l l in teract ion e f fec t of hue and motif was found to be s i gn i f i c an t (p * .0143). Table 46 shows, in descending order of pleasantness, the mean pleasure ratings for a l l hue-motif combina-t i ons . The most pleasant combinations are at the top o f the table while the least pleasant ones are at the bottom. The hor izontal l i ne d iv id ing the table into two parts indicates that combinations above the l ine were found to be pleasant (with means above the grand mean of -.0482) whereas the combinations below the l i ne were rated as unpleasant (with means less than the grand mean). Duncan's mult ip le range tests (at p<.05) showed further a 240. TABLE 46 Rank ordering of hue and motif combinations according to pleasure. Mean Hue-motif combination .863 GY-P landscape .397 PB-Y abstract .387 RP-G landscape .298 YR-B bui ldings .290 PB-Y landscape .269 RP-G bui ldings .232 BG-R bui ld ings .150 YR-B landscape .115 GY-P abstract -.078 RP-G abstract -.092 BG-R face - .118- BG-R abstract -.276 PB-Y bui ld ings -.292 YR-B face -.349 BG-R landscape -.359 GY-P bui ld ings -.529 YR-B abstract -.546 GY-P face -.571 PB-Y face -.686 RP-G face 241. homogeneous subset of means for the representational motifs ( face, landscape, bui ld ings) and the abstract when the hues YR-B and PB-Y were shown, but not for BG-R, GY-P and RP-G. Dominance The mean scores for dominance in the hue by motif i n t e r -action were: Face Landscape Bui ldings Abstract BG-R .058 -.374 -.267 .255 GY-P -.567 .407 -.361 -.383 YR-B .032 -.317 .150 -.191 RP-G -.248 .127 .296 -.031 PB-Y -.007 -.205 -.162 -.222 While neither the e f fec t of hue nor motif were s i gn i f i c an t on dominance, the interact ion of these two main ef fects was s i gn i f i c an t (p = .0014). Table 47 shows, in descending order of subjects fee l ing in cont ro l , the mean dominance rat ings for a l l hue-motif combina-t i ons . The combination which the subjects scored as the least domin-at ing ( I . e . , that which they perceived themselves to be most in Control of) i s at the top, while the most dominating combination Is at the bottom. The d iv id ing l i ne represents, as before, the grand mean (-.0999}innthis case) so that the combinations above the l i ne are judged not to be dominating while those below the l i ne are. 242. TABLE 47 Rank ordering of hue and motif combinations according to dominance. Mean Hue-motif combination .407 GY-P landscape .296 RP-G bui ld ings .255 BG-R abstract .150 YR-B bui ld ings .127 RP-G landscape .058 BG-R face .032 YR-B face -.007 PB-Y face -.031 RP-G abstract -.162 PB-Y bui ld ings -.191 YR-B abstract -.205 PB-Y landscape -.222 PB-Y abstract -.248 RP-G face -.267 BG-R bui ld ings -.317 YR-B landscape -.361 GY-P bui ld ings -.374 BG-R landscape -.383 GY-P abstract -.567 GY-P face 243. In add i t ion, Duncan's mult ip le range tests (at p < .05) showed a homogeneous subset consist ing of means for the representational motifs (face, landscape, bui ld ings) and the abstract when subjects were looking at BG-R hues, but not when they saw any of the other four hues. VALUE BY MOTIF INTERACTIONS Pleasure The mean scores fo r pleasure in the value by moti f i n t e r -act ion were: Face Landscape Bui ldings Abstract High value -.382 .601 .359 -.128 Low value -.492 -.051 -.289 .043 I t was found e a r l i e r that ithe main e f fec t of value was s i gn i f i c an t by i t s e l f — t h a t high value displays were judged more pleasant than the low value displays—and that motif a lso had a s i gn i f i c an t e f f ec t on pleasure scores, the preferred motifs being, in order: landscape, bu i ld ings , abstract and face. The in teract ion of value and motif was also found to s i g n i f i -cantly inf luence scores on the dimension of pleasure (p = .0332). 244. Table 48 shows, in descending order of pleasantness, the mean pleasure ratings f o r a l l value-motif combinations. The hor izontal l i ne d iv id ing the table into two parts indicates that combinations above the l ine were judged to be pleasant (with means above the grand mean for pleasure of -.0482) while those below the l i ne were judged to be unpleasant (with means less than the grand mean for pleasure). Duncan's mult ip le range tests (at p<.05) showed a homogeneous subset of means fo r the representational motifs ( face, landscape, bui ld ings) and the abstract when high value displays were shown. Information rate The mean scores fo r information rate in the value by motif in teract ion were: Face Landscape Bui ldings Abstract High value -.188 -.195 -.283 .262 Low value .386 .083 -.199 .134 Ea r l i e r both the main ef fects of value and motif were seen to be s i gn i f i c an t dif ferences in information rate scores: high value displays resulted in low information rate scores, low value displays in high information rate scores and, in decreasing order of informa-t ion ra te , the four motifs were: abstract, face, landscape and bu i ld ings. 245. TABLE 48 Rank ordering of value and motif combinations according to pleasure. Mean Value-motif combination .601 High value landscape .359 High value bui ldings .043 Low value abstract -.051 -.128 -.289 -.382 -.492 Low value landscape High value abstract Low value bui ld ings High value face Low value face 246. The interact ion of value and motif was found to s i gn i f i c an t l y inf luence subjects ' information rate scores (p - .0038). The mean information rate scores are arranged in descending order of informa-t ion rate (or "uncertainty") in table 49. In contrast to the scales of the other dependent measures, which could be dichotomized fo r instance in terms of pleasant-unpleasant, the scale of information rate i s continuous from "maximum uncertainty" to "maximum cer ta in ty " . Thus, the overa l l mean fo r information rate (of .0009) i s not of pa r t i cu la r importance here. Results from Duncan's mult ip le range tests (at p < .05) showed a homogeneous subset of means fo r the representational motifs ( face, landscape, bui ld ings) and the abstract for high value d i sp lays , whereas th is was not the case when low value displays were presented. MOTIF BY SEX INTERACTION Pleasure The mean scores for pleasure in the motif by sex i n t e r -action were: Males Females Face -.375 Landscape .023 Buildings -.147 Abstract -.455 -.497 .523 .232 .319 Mean -.235 .145 247. TABLE 49 Rank ordering of value and motif combinations according to information ra te . Mean Value-motif combination .386 Low value face .262 High value abstract .134 Low value abstract .083 Low value landscape -.188 High value face -.195 High value landscape -.199 Low value bui ld ings -.283 High value bui ld ings 248. Ea r l i e r i t was found that both the main e f fects of motif and sex s i gn i f i c an t l y infjuenced subjects ' scores on pleasure: in descend-ing order of pleasantness, the motifs were found to be landscape, bu i ld ings, abstract and face. In terms of sex, females found the displays d i s t i n c t l y more pleasing than did the males. The interact ion of motif and sex was also found to s i g n i f i -cantly inf luence scores on pleasure (p = .0482). When mean pleasure scores for the interact ion were ranked from "most pleasant" to " least pleasant" 4 of the means indicated pleasant stimulus combinations ( i . e . , they were above the hor izontal l i ne which represents the grand mean fo r pleasure, -.0428), and 4 of the mean scores indicated un-pleasant stimulus combinations ( i . e . , t he i r means were less than the grand mean for pleasure). Table 50 shows the rank ordering of these mean scores. HUE BY SEX INTERACTION Dominance The mean scores fo r dominance in the hue by sex interact ion were: 249. TABLE 50 Rank ordering of motif and sex combinations according to pleasure. Mean Motif-sex combination .523 Landscape--fema1es -.319 Abstract-females .232 Buildings-females .023 Landscape-males -.147 Bul l dings-males -.375 Face-males -.455 Abstract-males -.497 Face-females 250. Males Females BG-R GY-P YR-B RP-G PB-Y .022 -.136 .351 .102 -.085 -.197 -.308 -.441 -.024 -.223 Mean .041 -.238 I t was e a r l i e r reported that the main ef fects of hue did not s i g -n i f i c an t l y inf luence scores on the dimension of dominance, but that the main e f fec t of sex d i d . I t was found that males scored higher on dominance (mean .041) than did females (-.238). The interact ion of hue and sex was, however? found to s i g n i f i -cantly inf luence subjects ' scores on dominance (p = .0399). Table 51 shows the mean dominance scores arranged in order of decreasing dominance. The hor izpntal Une , representing the grand mean for dominance (-.0999), separates the treatment conditions in which subjects f e l t more in control (above the l i ne) from those in which they f e l t dominated (below the l i n e ) . In add i t ion, Duncan's mult ip le range tests (at p < .05) showed a homogeneous subset of means fo r males and females when they looked at the YR-B d isp lays , but th is did not hold for any of the other four hues. 251. TABLE 51 Rank ordering of hue and sex combinations according to dominance. Mean Hue-sex combination .351 YR-B males .102 RP-G males .022 BG-R males -.024 RP-G females -.085 PB-Y males -.136 GY-P males -.197 BG-R females -.223 PB-Y females -.308 GY-P females -.441 YR-B females 252. MOTIF RECOGNITION The subjects were asked to judge what they saw i n the displays and to complete the question "This i s a card showing_ ' on the quest ionnaire. For purposes of ana lys i s , subjects ' scores were treated as a dichotomy between (1) seeing the motif the experimenter intended ( i . e . , a face, a landscape, bui ld ings or an abst rac t ) , and (2) not seeing th i s intended mot i f . The dichotomy was constructed as fo l lows: " r ight " answers were treated as one category, (scored as "2") and an absence of an answer and "wrong" answers as the other (scored as " 1 " ) . The response "nothing" was c l a s s i f i e d as "wrong" when applied to the face, landscape and bu i ld ings , but as " r ight" when applied to the abstract. Results of the analysis of variance with the d isplay var iables of  hue, value, chroma, motif and the subject var iable sex as independent  var iables and motif recognit ion as the dependent var iab le . The analysis employed the f u l l model ( i . e . , with main e f fects and a l l in teract ion terms), and table 52 shows the summary table fo r the analysis of variance with the s i gn i f i can t e f fects onlyia The grand mean for motif recognit ion was 1.519 (which indicates an overa l l rec-ognit ion rate of 51.9% since perfect recognit ion would resu l t in a mean of 2.0 and a complete lack of recognit ion in a mean of 1.0). 253. TABLE 52 Summary of analys is of variance with motif recognit ion as dependent var iab le . Source SS df MS F P Hue (H) 2.9397 4 .7349 3.4098 .0091* Value (V) 9.5066 1 9.5066 44.1071 .0000* Chroma (C) .0002 1 .0002 .0008 .9267 Motif (M) 12.1669 3 4.0556 18.8166 .0000* Sex (S) .0807 1 .0807 .3745 .5483 HV 1.2370 4 .3093 1.4349 .2193 HC .8484 4 .2121 .9840 .4164 HM 7.7286 12 .6441 2.9882 .0005* HS 1.2556 4 .3139 1.4564 .2124 VC .0266 1 .0266 .1233 .7239 VM 7.5888 3 2.5296 11.7363 .0000* VS .1198 1 .1198 .5559 .4626 CM .4591 3 .1530 .7100 .5497 CS .4304 1 *4304 1.9968 .1540 MS .9382 3 .3127 1.4509 .2253 HVC .6968 4 .1742 .8082 .5223 HVM 2.0642 12 .1720 .7981 .6538 HVS .3892 4 .0973 .4514 .7740 HCM 3.8026 12 .3169 1.4702 .1293 HCS .5423 4 .1356 .6290 .6451 HMS 3.4099 12 .2842 1.3184 .2018 VCM .0698 3 .0233 .1079 .9508 VCS .0729 1 .0729 .3382 .5684 VMS .3317 3 .1106 .5131 .6776 CMS .3245 3 .1082 .5018 .6853 HVCM 2.8228 12 .2352 1.0914 .3637 HVCS .0762 4 .0190 .0883 .9829 HVMS 1.4921 12 .1243 .5769 .8624 HCMS 1.8434 12 .1536 .7127 .7410 VCMS .5965 3 .1988 .9225 .4312 HVCMS 2.6605 12 .2217 1.0286 .4200 Error 167.9020 779 .2155 Total 234.4239 938 * Indicates s ign i f i cance . 254. The overa l l standard deviat ion was .410. S ign i f i can t main ef fects Hue= The mean scores f o r hue on motif recognit ion were: Hue Means BG-R 1.457 GY-P 1.572 YR-B 1.553 RP-G 1.567 PB-Y 1.444= In test ing the ef fects of the f i ve leve ls of hue on motif recogni-t i o n , i t was found that , i r respect ive of mot i f , hue contributed s i gn i f i c an t l y to the recognit ion of motifs (p = .0091). In decreasing order of motif detectab i1 i ty , the hues which lent themselves best to recognit ion were: GY-P (57.2%) RP-G {56.7%) YR-B (55.3%) BG-R (45.7%) PB-Y (44.4%) 255. Value The mean scores fo r value on motif recognit ion were: I t was hypothesized that value would predict the degree of motif recognit ion, and th is was found to be the case (p < .0000). High value d isp lays, i r respect ive of moti f , were s i gn i f i c an t l y more often correct ly i den t i f i ed (61.9%) than low value displays (41.8%). Moti f The mean scores fo r motif on motif recognit ion were: Value Means High Low 1.619 1.418 Motif Means Face 1.585 Landscape 1.349 Bui ldings 1.652 Abstract 1.489 As was expected, the motifs as they were constructed had a s i g n i f i -cant inf luence on subjects ' scores on motif recognit ion (p < .0000). In descending order of de tec tab i l i t y , the motifs were: 256. Buildings 1 (65.2%) Face (58.5%) Abstract (48.9%) Landscape (34.9%) Interact ion of main e f fec ts Hue by motif in teract ion The mean scores fo r motif recognit ion in the hue by motif in teract ion were: Face Landscape Bui ldings Abstract B6-R 1.479 1.326 1.521 1.500 GY-P 1.652 1.652 1.614 1.408 YR-B 1.660 1.370 1.646 1.532 RP-G 1.688 1.255 1.756 1.574 PB-Y 1.447 1.167 1.729 1.435 I t was reported above that both hue and motif resulted in s i g n i f i -cant e f fects on motif recogni t ion. The overa l l in teract ion e f fec t of hue and motif was also s i gn i f i c an t (p = .0005). Table 53 shows the hue-motif combinations and the i r associated per cent recogni-t ion in order o f decreasing moti f de tec tab l l I t y . Value by motif in teract ion The mean scores fo r motif recognit ion in the value by motif in teract ion were: 257. TABLE 53 Rank ordering of hue and motif combinations in terms of decreasing motif recogn i t ion. % Recognition Hue-motif combination 75.6 RP-G bui ld ings 72.9 PB-Y bui ld ings 68.8 RP-G face 66.0 YR-B face 65.2 GY-P face 64.6 YR-B bui ld ings 62.5 GY-P landscape 61.4 GY-P bui ld ings 57.4 RP-G abstract 53.2 YR-B abstract 52.1 BG-R bui ld ings 50.0 BG-R abstract 47.9 BG-R face 44.7 PB-Y face 43.5 PB-Y abstract 40.8 GY-P abstract 37.0 YR-P landscape 32.6 BG-R landscape 25.5 RP-G landscape 16.7 PB-Y landscape 258. TABLE 54 Rank ordering of value and motif combinations 1n terms of decreasing motif recogni t ion. % Recognition Value-motif combination 83.1 High value face 68.4 High value bui ld ings 62.1 Low value bui ldings 50.8 High value abstract 47.0 Low value abstract 45.3 High value landscape 33.9 Low value face 24.6 Low value landscape 259. Face Landscape Bui ldings Abstract High value 1.831 1.453 1.684 1.508 Low value 1.339 1.246 1.621 1.470 Both the main ef fects of value and of motif were s i gn i f i c an t as shown e a r l i e r . High value displays were more eas i l y recognized than low value d isp lays, and the motifs were recognized in the fol lowing order: bu i ld ings, face, abstract 9and landscape. The interact ion of value and motif was also found to be s i gn i f i c an t (p < .0000). Table 54 shows the value-motif combinations and the i r associated per cent rec-ognit ion in order of decreasing detectab i1 i ty . MOTIF RECOGNITION AND INFORMATION RATE I t was hypothesized that subjects ' recognit ion of a motif would be better the lower ( i . e . , the more "certa in") the mean information rate score was for that d i sp lay . To tes t th i s a regression analysis using display mean recognit ion scores as the c r i t e r i on var iable was performed. The general form of the l i nea r equation tested was: Recognition = a x Information + b The mult ip le corre lat ion coe f f i c i en t (R ) was .157 which indicated 260. that the tota l variance accounted fo r was about 15.7% and, in general, the equation did not seem to f i t the observed data very wel l although the overa l l F-probabi l i ty was .0004. The resultant regression equa-t ion was: Recognition - -25.114 x Information + 51.954 The normalized coef f i c ients were -.396 and 2.204, respect ive ly , f o r the two terms. In order for a regression equation to have any pred ic t ive value beyond the actual data used in the ana lys i s , i t s standard errors must fol low the normal d i s t r i bu t i on . (Draper and Smith, 1966) To tes t t h i s , the standardized residuals were calculated as the residual (y - y) / standard error (Le and Tenesci, 1977), and the expected values of the residuals p lotted against the standardized res idua ls . I f normality e x i s t s , the plotted points should fol low a l i ne with a slope of 110:(sloping upward at about a 45° angle). As f igure 61 shows, the d i s t r i bu t i on of the standard errors seemed to fol low the normal d i s t r i bu t i on . The corre lat ion between the mean scores on motif recognit ion and information rate for the 80 displays was -.40 (that i s , the square root of R ) . This negative corre la t ion Is re f lec ted in f igure 62 which shows that , as motif recognit ion mean scores ( i . e . , 261. Standardized res iduals FIGURE 61 P lot of expected res iduals versus the residuals In the motif recognit ion -Information rate regression equation. 262. . I I i I 1 .75 -.35! .01 .39 .77 1.15 Information rate FIGURE 62 P lot of motif recognit ion means versus Information rate means, and regression Une . 263. de tec tab i l i t y ) increased, the mean information rate scores ( i . e . , "cer ta inty" decreased. The slope of the regression l i ne was -25.144 or , expressed in terms of the normalized coef f i cent , fo r every one uni t increase in motif recognit ion, there was a corresponding de-crease of .396 units of information rate mean scores. REGRESSION ANALYSES WITH DISPLAY COMPONENT RELATIONSHIPS I t was hypothesized that any or a l l of the d i s t r i bu t i on spec i -f i ca t ions fo r the 80 displays developed and described in chapter IV, section I I , may predict the scores on pleasure, arousal , dominance arid information ra te . To test t h i s , two types of step-wise regression analyses, one l i nea r and one quadrat ic, were carr ied out, using the predictor var iables shown on page 220. The two analyses were en t i r e l y independent of each other in the sense that i n the l i nea r analyses the program was given the option of se lec t ing any of the 24 var iables as s i gn i f i c an t terms i n the equa-t i ons . In the quadratic analyses, on the other hand, 48 var iables (the 24 l i near and 24 quadratic) were potent ia l s i gn i f i c an t terms in the equations. In other words, variance accounted for by the quadratic equations was not the sum of variance accounted for by the l i near 264. equation plus variance1 accounted for by the quadratic equation. For the purpose of providing an overview of the r e su l t s , the normalized coe f f i c i en ts associated with the various terms (which are ind i ca t ive of the amount of s i gn i f i c an t contr ibut ion to the regression equat ion), standard e r ro r , t o ta l var ia t ion accounted fo r (R ) , and overa l l F-probabil1ty are shown fo r the two analyses in tables 55 and 56. Referring to f igure 63, which shows plots of the standardized residuals versus the residuals of both l i nea r and quadratic equa-t ions for the four c r i t e r i on var iab les , i t i s apparent according to the c r i t e r i on mentioned e a r l i e r that the standard errors followed reasonably c lose ly the normal d i s t r i bu t i on . Thus the requirement fo r the equations to have pred ic t ive power in s i tuat ions other than the experimental one seems to have been met. In the fo l lowing descr ipt ion of the regression equations, abbreviations for the predictor var iables are used. Tables 55 or 56 provide an index of these abbreviations and the i r f u l l names. Equations fo r pleasure The mult ip le corre lat ion coe f f i c i en ts (R ) indicated that , in the l i near regression equation, 29.1% of the variance was accounted fo r while in the quadratic equation, 38.6% was accounted f o r . Thus, the quadratic equation seemed to f i t the data better 265. TABLE 55 Normalized coe f f i c i en t s , R , standard errors and overa l l F-probabi l i t ies fo r the l inear regression equations.* Predictor var iables Pleasure Arousal Dominance Information rate Average- AE Average-value Average-chroma Average-temperature AVG-D AVG-V AVG-C AVG-T -.197 Figure/background- At Flgure/background-value Figure/background-chroma Figure/background-temperature F/B-D F/B-V F/B-C F/B-T Top/bottom-AE Top/bottom-value Top/bottom-chroma Top/bottomttemperature T/B-D T/B-V T/B-C T/B-T .320 -.308 L e f t / r i g h t - A E Lef t / r ight-va lue Left/right-chroma Left/right-temperature L/R-D L/R-V L/R-C L/R-T -.248 -.264 -.249 .354 Adjacent/dif ference- AE ADD-D Adjacent/di fference-value ADD-V Adjacent/difference-chroma ADD-C Adjacent/difference-temperature ADD-T -.256 .672 -.341 .421 Adjacent/variance- AE Adjacent/variance-value Adjacent/variance-chroma Adjacent/variance-temperature ADV-D ADV-V ADV-C ADV-T -.367 K 2. Standard error F-probabl l i ty .291 .56 .000 .261 .50 .000 .123 .36 .007 .359 .36 .000 * Note that these are normalized coeff lcents f o r the l i nea r orthogonal components of the o r i g ina l var iab les , not the var iables themselves. The resu l ts in terms of the o r ig ina l var iables are shown in tables 57, 58, 59 and 60. 266. TABLE 56 Normalized coe f f i c i en t s , R , standard errors and overa l l F-probabil1t1es f o r the quadratic regression equations.** Predictor var iables Pleasure Arousal Dominance Information rate Average- AE Average-value Average-chroma Average-temperature AVG-D AVG-V AVG-C AVG-T -.225 Figure/background- AE Figure/background-value Figure/background-chroma Figure/background-temperature F/B-D F/B-V F/B-C F/B-T -.236 Top/bottom-AE Top/bottom-value Top/bottom-chroma Top/bottom-temperature T/B-D T/B-V T/B-C T/B-T .288 -.336 -.427* -.448* Le f t / r i gh t - AE Lef t / r ight-va lue Left/right-chroma Le f t / r i ght-tempera ture L/R-D L/R-V L/R-C L/R-T -.409* -.264 -.249 .345 Adjacent/dif ference- AE ADD-D Adjacent/difference-value ADD-V Adjacent/difference-chroma ADD-C Adjacent/difference-temperature ADD-T Adjacent/variance- AE Adjacent/variance-value Adjacent/variance-chroma Adjacent/variance-temperature ADV-D ADV-V ADV-C ADV-T -.291* -.223 R 2 Standard error F-probabi l i ty .386 .39 .000 .310 .31 .000 .123 .36 .007 .382 .38 .000 * Indicates quadratic form of predictor var iab le . ** Note that these are normalized coe f f i c ients for the l i nea r and quadratic orthogonal components of the o r ig ina l var iab les , not the var iables them-selves. The resu l ts in terms of the o r ig ina l var iables are shown 1n tables 57, 58, 59 and 60. 267. Standardized res iduals Quadratic equations Standardized res iduals FIGURE 63 P lot of expected res iduals versus res iduals for l i near and quadratic regression equations. 268. than the l i near one, since i t accounted for andadditional 9.5% of the variance. The resultant regression equations were: Linear equation: Pleasure = 1.189 x T/B-V - 3.345 x T/B-C - .032 x ADD-D - 10.646 x ADV-C + 6.763 Quadratic equation: Pleasure = 1.070 x T/B-V - 3.649 x T/B-C + 173.185 x L/R-C - 85.454 x (L/R-C) 2 + 1.103 x ADV-T - .349 x (ADV-T)2 - 85.544 In both of these equations, top/bottom-value and top/bottom-chroma predicted pleasure scores. However, in the l i nea r equation, adjacent/ d i f ference-Al and adjacent/variance-chroma were added, while in the quadratic equation left/r ight-chroma and adjacent/variance-temperature were included in the i r quadratic form. The addit ion of these two terms in the i r l i n ea r form was done to complete the equation, since whenever a quadratic component appeared in the equation, the l i nea r form of th i s component was automatical ly added. The pa r t i a l corre lat ions between pleasure and the terms in the two equations, together with the per cent variance accounted fo r by these terms, are shown in table 57. 269. TABLE 57 Pa r t i a l corre lat ions and per cent variance accounted for by l i near and quadratic equations f o r pleasure. Linear Quadratic Term Par t i a l cor re la t ion Per cent variance Pa r t i a l cor re la t ion Per cent variance T/B-V T/B-C AOD-D ADV-C L/R-C L/R-C 2 ADV-T ADV-T2 .33 -.29 -.25 -.26 11.1 7.7 4.8 5.5 .32 -.31 -.03 -.38 -.16 -.34 ) ) ) ) 8.1 8.1 14.5 7.9 Total 29.1 38.6 270. As th i s table shows, the associat ion between the s i gn i f i c an t predictor var iab les , entering into the quadratic equation, and pleasure was due mostly to the var iable le f t/r ight-chroma. The other three var iab les , top/bottom-value, top/bottom-chroma and adjacent/variance-temperature contributed about equal ly and to a lesser extent to the equation. Equations fo r arousal The mult ip le corre lat ion coe f f i c i en t (R ) showed that 26% of the variance was accounted fo r by the l i near regression equation and that th i s was increased to 31% in the quadratic equation. The two equations were as fo l lows: Linear equation: Arousal = -2.268 x L/R-V + .910 x ADD-V - .384 x ADD-T + 1.382 Quadratic equation: Arousal = -3.5 x F/B-C - .006 x ADV-D + 142.775 x T.B-C - 70.329 x (T/B-C) 2 - 68.321 The terms in these two equations d i f fe red considerably between the l i nea r and the quadratic vers ion. In the l i nea r model, the terms 271. l e f t / r i gh t - va l ue , adjacent/difference-value and adjacent/dif ference-temperature appeared, while i n the quadratic equation the terms were f1gure/background=chroma, adjacent/var iance-4E and top/bottom-chroma. Changing from the l i nea r model to i t s quadratic form containing the same variables thus did not provide as good a f i t as the quadratic form with the var iables which were f i n a l l y se lected. The pa r t i a l corre lat ion coe f f i c ients fo r arousal and the ind iv idua l terms in the two equations together wiith the per cent variance accounted fo r by these terms are shown in table 58. In the quadratic equation, which was a s l i g h t l y better predictor of arousal mean scores than the l i nea r one, the var iable top/bottom-chroma was by fa r the most prominent. The other two var iables of figure/background-chroma and adjacent/variance-^ E played a small part in comparison to the f i r s t var iab le . Equations for dominance The mult ip le corre la t ion coe f f i c i en t (R ) indicated that 12% of the variance was accounted fo r in both the l i nea r and the quadratic equations: Linear equation: Dominance • -1.945 x L/R-C - .892 x L/R-T + 2.736 272. TABLE 58 Pa r t i a l corre lat ions and per cent variance accounted fo r by l i near and quadratic equations for arousa l . Linear Quadratic Pa r t i a l Per cent Pa r t i a l Per cent Term corre la t ion variance corre la t ion variance L/R-V ADD-V ADD-T F/B-C ADV-D T/B-C T/B-C2 -.28 .37 -.27 7.0 13.4 5.7 -.26 -.25 -.12 ) -.46 ) 5.2 4.5 21.3 Total 26.1 31.0 273. The quadratic equation was Ident ica l to the l i nea r one. The corre lat ions with dominance of the two ind iv idua l terms in the equations and the per cent variance accounted fo r by these terms are shown in table 59. As can be seen from th is tab le , the tota l variance accounted for by both equations was due about equal ly to the var iables of left/r ight-chroma and lef t/r ight-temperature. Equations for information rate There was a s l i gh t increase in the mult ip le corre la t ion coe f f i c i en t (R ) from the l i nea r equation (accounting fo r 35.9% of the variance) to the quadratic equation (which accounted for 38.2% of the var iance). Thus the quadratic equation seemed to provide a s l i g h t l y better f i t than did the lUnear one. The two equations f o r information rate were: Linear equation: Information rate = -.096 x AVG-V + 2.573 x L/R-C + .912 x ADD-C - 2.682 Quadratic equation: Information rate = -.110 x AVG-V + 2.511 x L/R-C + 10.683 x T/B-V - 5.121 x (T/B-V) 2 - 7.390 274. TABLE 59 Pa r t i a l corre lat ions and per cent variance accounted for by l i near and quadratic equations fo r dominance. Linear Quadratic Term Par t i a l Per cent corre la t ion variance Pa r t i a l Per cent corre la t ion variance L/R-C L/R-T -.25 6.1 -.26 6.2 -.25 6.1 -.26 6.2 Total 12.3 12.3 275. TABLE 60 Pa r t i a l corre lat ions and per cent variance accounted for by l i near and quadratic equations for information ra te . Linear Quadratic Par t i a l Per cent Pa r t i a l Per cent Term corre la t ion variance corre la t ion variance AVG-V L/R-C ADD-C T/B-V T/B-V2 -.24 .39 .44 3.8 12.5 19.6 -.27 .39 -.19 ) -.46 ) 5.1 11.9 21.2 Total 35.9 38.2 276. I t i s noticeable from these equations how the variables of average-value and!left/right-chroma pers isted in both models. In add i t ion, the l i nea r equation also included the term of adjacent/dif ference-chroma while the quadratic equation included the term of top/bottom-value instead. The corre lat ion between information rate and the terms In the two equations, together with the per cent variance accounted fo r by these terms are shown in table 60. As can be seen from th is tab le , most of the variance accounted fo r in the l i nea r equation was a t t r i bu t -able to the term of adjacent/distance-chroma and somewhat less to that of l e f t / r i g h t chroma. Hardly any e f fects were due to the average-value var iab le . In the quadratic equation, however, by f a r the largest e f fec t was due to the term of top/bottom-value, and proport ionately less to those of left/r ight-chroma and average-value. II INTERPRETATION OF THE RESULTS This sect ion interprets and summarizes the f indings of the resu l ts reported on in sect ion I, and provides a basis fo r the more general discussion in chapter V I I I . The deta i l s of the resu l ts were reported in sect ion I i n terms of the four major s t a t i s t i c a l analyses performed: 277. 1. Mul t ivar iate analysis of variance re la t ing the display var iables of hue, value, chroma, motif and the subject var iable of sex to the emotional var iables of pleasure, arousal , dominance and to information rate. 2. Analysis of variance re la t ing the display var iables of hue, value, chroma, motif and the subject var iable of sex to the dependent var iable to motif recogni t ion. 3. Regression analysis re la t ing information rate to motif r e c i gn i t i on . 4. Mult ip le step-wise regression analyses re la t ing the 24 display component re lat ionships to the c r i t e r i on variables of pleasure, arousal , dominance and informa-t ion rate . In th i s sec t ion , on the other hand, the exposit ion w i l l fol low the sequence of the four hypotheses and general research question as stated in chapter I I I . THE EMOTIONAL MEASURE AND INFORMATION RATE'HYPOTHESIS The hypothesis dealing with the e f fects of hue, value, chroma, motif and sex on pleasure, arousa l , dominance and information rate was: I t i s hypothesized that the var iables of hue, value, chroma, motif and sex w i l l s i gn i f i c an t l y inf luence the way subjects rate the displays in terms of pleasure, arousa l , dominance and information rate . 278, The e f fects of hue Predominant hue as a main e f fec t did not have any s i gn i f i c an t e f fec ts on pleasure, arousal , dominance or information rate scores, but the in teract ion of hue and motif had s i gn i f i c an t e f fec ts on pleasure and dominance, and that of hue and sex on dominance. The preference order of hue-motif combinations ( c f . table 46) showed however that no systematic e f fects of hue existed fo r pleasure: Hue preference Motif order Face (landscape Bui ldings Abstract 1 BG-R GY-P YR-B PB-Y 2 YR-B RP-G RP-G GY-P 3 GY-P PB-Y BG-R RP-G 4 PB-Y YR-B PB-Y BG-R 5 RP-G BG-R GY-P YR-B Thus i t was assumed that hue played only a minimal part in th i s i n te rac t i on . The same type of non-systematic hue order fo r dominance (with the in terest ing exception of the complete reversal of hue order fo r the face and the landscape) emerged from the hue by motif in teract ion for dominance: 279. Hue dominance order Face Landscape Bui ldings Abstract Moti f 2 3 4 5 BG-R YR-B PB-Y RP-G GY-P GY-P RP-G PB-Y YR-B BG-R RP-G YR-B PB-Y BG-R GY-P BG-R RP-G YR-B PB-Y GY-P Since, however, motif by i t s e l f , l i k e hue, was not s i gn i f i c an t on dominance, i t was concluded that on th is emotional dimension hue seemed to play a s l i g h t l y larger part i n the i n te rac t i on . Based on the resu l ts of these two in terac t ions , i t was con-cluded that hue, although not s i gn i f i c an t by i t s e l f , d id play a ro le in determining preference order ( i . e . , pleasure) of d isplays and the extent to which subjects f e l t dominated by the d i sp lays . In other words, preferred and dominating hue-motif combinations were judged d i f fe rent from preferred and dominating motifs alone. While the ef fects of hue in these two interact ions were somewhat obscure, the ef fects of hue on dominance due to the i n t e r -act ion of hue and sex was c l ea re r . A reordering of the hue-sex combinations in table 51 shows that the sequence of RP-G, BG-R, PB-Y and GY-P was judged by both sexes to be increas ingly con-t r o l l i n g (decreasing in dominance), except that males found YR-B to be the least dominating hue, while females found th is hue to be the most dominating: 280. Hue dominance order Males Females 1 2 3 4 5 YR-B RP-G BG-R PB-Y GY-P RP-G BG-R PB-Y GY-P YR-B The e f fec ts of value There were s i gn i f i c an t dif ferences on the dimensions of pleasure, dominance and information rate depending on whether displays were predominantly high value ( l i gh t ) or predominantly low value (dark) . Predominantly l i gh t displays resulted in greater pleasure, more fee l ing of dominance and a higher degree of uncertainty ( i . e . , higher information rate mean scores) than predominantly dark d isp lays . Furthermore, the resu l ts from the s i gn i f i c an t value by motif in teract ion showed that predominantly l i g h t faces, flight land-scapes, l i gh t bu i ld ings , but dark abstracts were preferred to pre-dominantly dark faces, dark landscapes, dark bui ld ings and l i g h t abstracts. In the case of information ra te , the value by motif i n t e r -act ion showed that predominantly l i g h t faces, l i g h t landscapes, l i gh t bui ld ings but dark abstracts resulted in higher degrees of certa inty (lower information rate mean scores) than predominantly 281. dark faces, dark landscapes, dark bui ld ings and l i g h t abstracts . The e f fects of chroma The main e f fec t of chroma was s i gn i f i c an t on the dimensions of pleasure and information ra te , but no interact ions involv ing chroma were s i gn i f i c an t . Predominantly high chroma ( i . e . , saturated) displays were preferred to predominantly low chroma ( i . e . , desaturated) d isp lays , and the predominantly high saturat ion displays resulted in lower information rate mean scores than did the predominantly low satur-at ion d i sp lays . The e f fects of motif Motif had s i gn i f i can t e f fects on pleasure, arousal and i n f o r -mation rate and, in a dd i t i o n , l i t had s i gn i f i c an t e f fec ts with hue on pleasure and dominance, and with value on pleasure and informa-t ion ra te . I t was the only one of the main e f fects to resu l t in s i gn i f i c an t changes in arousal scores. F i r s t , motif seemed to have a most pers istent e f f e c t on pleasure scores. Landscapes, bui ld ings and abstracts were gener-a l l y found to be pleasant while faces were cons istent ly judged unpleasant. In add i t ion, certa in hue-motif combinations were pre-ferred to others, ( c f . table 46) but no systematic re la t ionsh ip 282. between hue and motif seemed to emerge. In the case of the value-motif i n te rac t i on , however, motifs were shown to be preferred la rge ly as a resu l t of whether they were predominantly l i gh t or dark as mentioned e a r l i e r . Second, some motifs were found to be s i gn i f i c an t l y more arousing than others. In decreasing order of arousal , these were: abstract , face, bui ld ings and landscape. This e f f ec t was fur ther-more very pronounced as the s i ze of the F - s t a t i s t i c shows (c f . table 45). Th i rd, i t was found that some combinations of motif and hue resulted in s i gn i f i c an t l y d i f f e rent scores on dominance than other combinations. These are shown in table 47. However, as with the s i gn i f i c an t hue by motif interact ions on pleasure, no d iscern-i b l e pattern emerged. Fourth, subjects ' scores varied s i gn i f i c an t l y on the information rate dimension depending on what motif they saw. The abstract was c l ea r l y the motif which resulted in the most uncertainty, and then followed the face, the landscape and the bu i ld ings . From the s i gn i f i c an t in teract ion of motif and Value i t could furthermore be seen how l i g h t faces and landscapes resulted in greater certa inty than dark faces and landscapes. At the same time, i t d id not seem to make any d i f ference to information rate scores whether subjects saw l i g h t or dark bui ldings.and abstracts. 283. The e f fects of sex Males and females d i f fe red s i gn i f i c an t l y in the i r judgements of pleasure and dominance of the d i sp lays . In general, females found the displays more pleasant than did the males. At the same time, males f e l t more in control ( i . e . , they scored higher on dominance) than did females who, o ve r a l l , judged that the displays dominated them. The hypothesis re la t ing the display var iables of hue, value, chroma and mot i f , and the subject var iab le of sex to pleasure, arousa l , dominance and information rate was p a r t i a l l y accepted. THE HYPOTHESIS ABOUT REPRESENTATIONAL MOTIFS AND THE ABSTRACT The hypothesis dealing with the e f fec ts of representational motifs versus the abstixact on pleasure, arousal , dominance and i n -formation rate was: I t i s hypothesized that subjects ' responses on pleasure, arousal , dominance and information rate w i l l d i f f e r s i g -n i f i c a n t l y according to whether they are presented with a representational motif ( I . e . , the face, the landscape or the bui ld ings) or the abstract . In pa r t i cu l a r , i t i s expected that the representational motifs w i l l resu l t in a lower information rate score than the abstract moti f . 284. I t was assumed that in cases where no representational motif was involved, one or more of the dimensions of co lor would inf luence subjects', scores on the emotional measures. However, when the var iable of motif was introduced, i t was suspected that the inf luence of mot i f— that i s , whether a display f o r instance represented a face or a landscape—would lead to the introduct ion of a cognit ive component into the response measures. The present hypothesis was therefore advanced to explore j u s t what the di f ference due to representational mot i f s , ( faces, landscapes, bui ld ings) versus non-representational motif (abstract) would be. The di f ferences which might be expected due to th i s motif d i s t i n c t i on , in terms of information rate scores, seemed se l f -ev ident , so the expected d i rec t ion of th i s di f ference was indicated in tbe hypothesis. I t was shown above that motif inf luenced responses on pleasure, arousal , dominance ( in the hue by motif interact ion) and information ra te . The present r e su l t s , based on the post hoc comparisons already mentioned in sect ion I , indicated that responses were determined to a large extent by whether a representational or an abstract motif was shown. This e f fec t manifested i t s e l f in several instances. The abstract was general ly found to be more arousing than the representational mot i fs , although the face was also somewhat arousing. The landscape was by fa r the least arousing mot i f . The abstract was general ly perceived as more uncertain than the 285. representational mot i fs , although also here the face was judged somewhat more uncertain thantthe landscape and bu i ld ings . Results of the comparisons for the hue by motif in teract ion showed that the representational motifs were judged more pleasing than the abstract i n cases where the hue was YR-B, while the abstract was preferred when the hue was PB-Y. There were no such preferences fo r the other three hues. The abstract was found to be more dominating than the repre-sentational motifs only when the hue was BG-R and not for any of the other hues. From the post hoc comparisons in the value by motif i n te rac t i on , i t was found that the l i g h t representational displays were general ly judged to be more pleasant than the l i gh t abstract . No such prefer-ence was found for the dark d i sp lays . However, as was shown e a r l i e r , faces were the least preferred regardless of whether they were l i g h t or dark. The l i gh t abstracts were perceived as more uncertain than the l i g h t representational mot i f s . The dark displays did not r esu l t in such d i f ferences. The hypothesis re la t ing representational motifs to the abstract was accepted in part . 286. IHE MOTIF RECOGNITION HYPOTHESIS The hypothesis dealing with moti f recognit ion, display and subject var iable was: I t i s hypothesized that the var iables of hue, value, chroma, motif and sex w i l l s i gn i f i c an t l y inf luence the extent to which subjects w i l l recognize the motif of a d i sp lay . I t was shown above that there were a number of general d i f f e r -ences in scores on pleasure, arousal , dominance and information rate between the recognizable motifs and the abstract . The present hypo-thesis was advanced in order to explore the extent to which the display components of co lor (hue, walue, chroma) and mot i f , and the subject var iable of sex, would inf luence the recognit ion of the motif in a d isp lay . The resu l t s showed that hue, value and motif were the var iables which mainly contributed to motif recogni t ion. Of these, value was by fa r the most prominent: motifs were more eas i l y recognized in l i g h t (high value) displays than in dark (low value) d i sp lays . A lso , from the s i gn i f i c an t value by motif i n te rac t i on , i t was apparent that , while l i g h t displays were more eas i l y recognized, i t was mostly the l i g h t face which accounted fo r th i s r esu l t . Both l i g h t and dark bui ld ings were more eas i l y recognized than any of the other value-motif 287. combinations except the l i g h t face and, in f a c t , the l i g h t landscape turned out to be rather d i f f i c u l t to recognize. The displays had been constructed to show par t i cu la r mot i f s , and i t was expected that these motifs would be recognized by the subjects. However, only informal test ing during the i r construction had indicated that they could be recognized. The resu l ts from the present analysis confirmed the i n i t i a l expectation and the motifs were recognized in the fol lowing order: bu i ld ings , face, abstract , and landscape. The resu l ts from hue by moti f in teract ion necessitated only a s ight qua l i f i c a t i on of th i s general resu l t to the e f fec t that the GY-P landscape was1 "part icu lar ly eas i l y recognized (easier than some of the bui ld ings and abst rac ts) , and that the BG-R bu i ld ings , BG-R face and the PB-Y face were d i f f i c u l t to recognize. Hue was the least prominent of the s i gn i f i c an t main e f f e c t s , yet i t showed a strong inf luence on motif recogni t ion. In decreas-ing order of motif recognit ion, the hues were: GY=P, RP-G, YR-B, BG-R and PB-Y. An examination of the means i n the hue by motif in teract ion further showed that th i s order general ly appl ied to a l l mot i f s . F i n a l l y , i t d id not make any di f ference whether a subject was male or female as f a r as motif recognit ion was concerned. Thus, the hypothesis that the display var iables of hue, value, chroma and moti f , and the subject var iable of sex would s i gn i f i c an t l y inf luence the extent subjects recognized the motifs was accepted. 288. THE HYPOTHESIS ABOUT MOTIF RECOGNITION AND INFORMATION RATE The hypothesis dealing with the re lat ionsh ip between motif recognit ion and information rate was: I t i s hypothesized that subjects ' responses on the measure of information rate w i l l d i f f e r s i gn i f i c an t l y according to the ease with which they recognize the motif of the d i sp lay . That i s , the eas ier i t 1s to recognize a moti f , the lower the information rate w i l l be. Subjects were asked to complete the question "This 1s a card showing ", and t he i r answers coded accord-ing to whether or not they recognized the motif intended by the experimenter ( i . e . , a face, landscape, bui ld ings or an abst rac t ) . The mean scores fo r these coded answers were compared with the subjects" information rate mean scores. The hypothesized re lat ionsh ip between motif recognit ion and information rate was advanced since i t was thought that the extent to which a motif was recognized would depend on the perceived degree of redundancy, s imp l i c i t y , f a m i l i a r i t y , e t c . , Inherent in the d isp lay. That i s , the more redundant, simple and f am i l i a r , for instance, a d isplay was judged to be, the eas ier i t should be to recognize the motif of i t . The overa l l F - s t a t i s t i c for the regression equation was highly 289. s i gn i f i c an t (p • .0004), but only 15% of the to ta l variance was accounted for by the l i nea r regression equation. At the same time, i t was also shown that the standard errors followed the normal d i s t r i bu t i on , and since th i s indicates p r ed i c t ab i l i t y , however weak, the associat ion between the two variables seemed to indicate a trend. As Hays (1963) points out: A s i gn i f i c an t value for t he . . . [F t es t ] indicates that 1t i s safe to conclude that some p red i c t ab i l i t y i s afforded b by a l i nea r r u l e . . . However, . . . [ th i s tes t does not guarantee] that there e x i s t s . . . a strong l i n e a r . . . r e l a -t ionship in the population sampled.... (p. 546) The p red i c t ab i l i t y of motif recognit ion in th i s case was indicated most c l ea r ly by the corre lat ion coe f f i c i en t of - .40. Based on the above considerat ions, the hypothesis about motif recognit ion and information rate was accepted with caut ion. Since motif recognit ion turned out to correlate with informa-t ion rate (at - . 40 ) , addit ional analyses of variance were carr ied out with motif r e c i gn i t i on , the four display variables and sex as the independent variables and pleasure, arousal , dominance and information rate as the dependent variables to check the extent to which motif recognit ion in general influenced scores on pleasure, arousal , dom-inance and information rate (c f . appendix D). The resu l ts showed that only dominance in addit ion to information rate was-einfluenced by motif recognit ion: the more eas i l y a motif was recognized, the more " in contro l " and "cer ta in" the subjects f e l t ( c f . table 40). 290. THE RESEARCH QUESTION The general research question dealing with the d i s t r i b u -t iona l spec i f i ca t ion of the d isp lays , the emotional measures and information rate was: Do any or a l l of the numerical expressions of component re lat ionsh ips predict the outcome of pleasure, arousal , dominance and information rate scores for the displays? The re lat ionships between the 24 expressions of d isp lay component re la t i onsh ips , which were explained and calculated in chapter IV, sect ion I I , and the response measures of pleasure, arousa l , dom-inance and information rate could not be hypothesized with any degree of confidence. Since previous f indings had been found to be somewhat con f l i c t i ng and the s t imu l i used in previous research d i f fe red great ly from the displays employed here, the inves t iga-t ion of the re lat ionships was© thought of as "hypothesis creat ing" rather than as hypothesis t e s t i ng . In order to explore these re lat ionships to the f u l l e s t extent, a ser ies of step-wise regression analyses were carr ied out. These used a l l the 24 d i s t r i bu t i on spec i f i ca t ion re lat ionships (as described in chapter IV) in the l i near analyses, and these same 24 var iables plus t he i r quadratic form in the quadratic analyses. 291. Pleasure, arousal , dominance and information rate were used, in turn, as c r i t e r i on var iab les . The variance accounted for by the regression equations were in a l l cases (except those for dominance)bbetween 26.1 and 38.6%. Although not very large in general terms, they f a l l wi th in the range of "moderately strong" in terms of the re lat ionsh ips they ind i ca te , as pointed out by Hays (1963). This est imate, that about 31 percent of the variance in Y i s accounted fo r by X . . . lends assurance that a moderately s t r ong . . . re lat ionsh ip ex i s ts in the popu-la t i on represented by the data. (p. 527) Furthermore, the equations seemed to conform reasonably wel l with the normal d i s t r i bu t i on , ind ica t ing that one main prerequis i te for making predict ions to s i tuat ions beyond the experimental one ex i s ted . D is t r ibut ion spec i f i ca t i on and pleasure Although the quadratic equation c l ea r l y accounted fo r more of the variance (38.6%) than the l i near one (29.1%), the l i nea r equation was in teres t ing because the main terms in i t turned out to be top/bottom-value and top/bottom-chroma, and because these two terms pers isted also in the quadratic equation. C lear l y , then, the notion of top-bottom surface d iv i s i on (at least insofar as value and chroma were concerned) could be seen as determining, in par t , the 292. outcome of mean scores on the dimension of pleasure. The r e l a t i on -ships were such that fo r an increase in top/bottom-value ( i . e . , as the average value in the top hafllf Increased), scores on pleasure increased and, conversely, as top/bottom-value decreased, pleasure rat ings decreased. That i s , d isplays which on the average were l i gh te r in the top ha l f than in the bottom ha l f were found more pleasant than displays which were darker at the top and! l ighter at the bottom. This re lat ionsh ip i s shown in f igure 64. While a pos i t ive corre lat ion existed between top/bottom-value and pleasure, the exact opposite was the case for chroma: pleasure scores decreased as top/bottom-chroma values increased. That i s , displays which were more saturated in the top ha l f than in the bottom ha l f were found less pleasant than displays which, on the average, had more saturated colors in the bottom ha l f than in the top ha l f . Figure 65 shows th i s re la t ionsh ip . The other two terms in the l i nea r equation were adjacent/ d i f f e rence-AE and adjacent/variance-chroma which accounted for approximately the same per cent of the to ta l variance. The measure of adjacent-di f ference, as was pointed out in chapter IV, assessed the average di f ference between colors in successive sampling areas across the display surface, and was thus thought to be one way of expressing the accumulated contrasts in the d isp lay . I t was mentioned that one way to conceptualize the Idea of th i s measurement was to 293. High ... pleasure Pleasure Low pleasure Low Equal High top top- top value bottom value (dark) value ( l i gh t ) Top/bottom-value FIGURE 64 Approximate regression Hne for the top/bottom-value component of the l i near equation fo r pleasure. 294. High pleasure Pleasure Low pleasure Low Equal High top top- top chroma bottom chroma chroma Top/bottom-chroma FIGURE 65 Approximate regression Une fo r the top/bottom-chroma component of the l i near equation fo r pleasure. 295. think of a uniformly colored surface as having no contrast at a l l , and a checker-board design which would exh ib i t the maximum contrast poss ib le. Yet another way to phrase th i s was to say that i t measured (or, rather, calculated) the frequency of contrast areas across the surface. In the present case, the corre lat ion was between adjacent/ 2 difference-ZLE and pleasure, but the rather low R value of 4.8 meant that the real contr ibut ion of th i s var iable was somewhat questionable. The l a s t var iable to enter into the l i nea r equation was adjacent/variance-chroma. Although i t only contributed 5.5% to the to ta l variance, i t correlated at -.26 w*th pleasure, thus exh ib i t -ing a rather strongttrend. Figure 66 shows the approximate d i rec t ion of th i s component. The adjacent-variance concept, as described in chapter IV, las-'verisimilar 7 'to-'--tlH|i^ of the*adjacent-difference concept, except that i t ca lculated the contrasts with in the small sampling areas of the displays in terms of the variance of the numerical d i f f e r -ences between ind iv idua l colors rather than simply in terms of the average of these distances. In both cases, the average d i f f e r -ences and the variances fo r a l l the sampling areas were averaged to obtain one overa l l numerical value for each d i sp lay . Figure 66 shows how subjects found displays with low co lor 296. High pleasure Pleasure Low pleasure Low chroma contrast High chroma contrast Adjacent/variance-chroma FIGURE 66 Approximate regression Hne for the adjacent/variance-chroma component o f the l i near equation for pleasure. 297. contrast ( i . e . , small average dif ferences between major and con-t ras t ing colors) more pleasant than displays with high co lor contrast. In the case of the quadratic equation, the most prominent term was the left/r ight-chroma measure, espec ia l ly in i t s quadratic form. Since the pa r t i a l corre lat ion coe f f i c i en t between pleasure and the squared left/r ight-chroma term was -.38 i t i s a downward s loping cu rv i l i nea r trend whiich describes the data. Thus, the more saturated ( i . e . , higher chroma) the l e f t ha l f of the display was, in comparison with, the r ight ha l f , the more unpleasant i t was, and the less saturated the l e f t partoof the display was, in comparison with the r ight ha l f , the more pleasant i t was found to be. Figure 67 shows th is approximate cu rv i l i nea r re la t ionsh ip . As in the l i nea r case, the two second most prominent var iables turned out to be top/bottom-value and top/bottom-chroma. The approxi-mate regression l ines for these var iables are shown in f igures 64 and 65. The f i n a l var iab le which contributed to the to ta l variance of the quadratic equation fo r pleasure was adjacent/variance-temperature. This var iable ( i n i t s combined l i nea r and quadratic form) contributed almost as much (7.9%) as1 each of the var iables top/bottom-value (8.1%) and top/bottom-chroma (8.1%), and the pa r t i a l corre lat ion between i t and pleasure was -.34. Figure 68 shows the approximate regression curve which resulted from th i s assoc iat ion. From th is cu rv i l i nea r r e l a t i on -ship can be seen how as perceived temperature contrasts increased, 298. High _ pleasure Pleasure Low pleasure Low Equal High l e f t l e f t - r i g h t l e f t chroma chroma chroma Left/right-chroma FIGURE 67 Approximate regression curve fo r the quadratic form of left/r ight-chroma versus pleasure. 299. High pleasure Pleasure Low pleasure Low temperature contrast High temperature contrast Adjacent/variance-temperature FIGURE 68 Approximate regression curve for the quadratic form of adjacent/variance-temperature versus pleasure. 300. arousal leve l general ly decreased, except fo r the most extreme contrast s i tuat ions in whichaarousal seemed to s ta r t increasing again. No conclusions could be drawn about whether the contrasts Involved warm or cool hues, except perhaps in the very high contrast end of the curve where i t must be assumed that the con-trasts were great enough to include both warm and cool co lo rs . D is t r ibut ion spec i f i ca t ion and arousal A specia l and interest ing circumstance arose in the case of arousal , where the l i near and quadratic equations used en t i r e l y d i f f e rent terms. The l i nea r equation selected the terms adjacent/di f ference-value, l e f t / r i gh t -va lue and adjacent/difference-temperature. Of these, the f i r s t two accounted for most of the variance (20.4% out of 26.1%). The pa r t i a l corre lat ion between adjacent/difference-value and arousal was .37, so as values of adjacent/distance-value increased, scores on arousal would also general ly increase. Figure 69 shows th i s re la t ionsh ip . That i s , the more value contrasts there were in a d i sp lay , the more arousing subjects would f i nd the display to be and, conversely, the fewer there were of these value contrasts, the less arousing that pa r t i cu la r d isplay would be. The second important term in the l i near equation was l e f t / r i g h t -value with a pa r t i a l cor re la t ion coe f f i c i en t of - .28 . This meant that 301. High arousal Arousal Low arousal Low value contrast High value contrast Ad jacent/di fference-value FIGURE 69 Approximate regression Hne for the adjacent/difference-value component 1n the l i near equation fo r arousa l . 3Q2. the higher the numerical value of th i s term was, the lowefc the arousal score would be. Or, i n other words, the l i gh te r the l e f t ha l f of the display was as compared to the r ight ha l f , the less arousing the display would be judged, and the darker the l e f t ha l f of the display was as compared to the r ight ha l f the more arousing i t would be. Figure 70 shows th is approximate assoc iat ion. F i na l l y , the term adjacent/difference-temperature played a role in the equation. I t was found that the higher the value of adjacent/difference-temperature was, the higher the arousal leve l would be. That i s , the more temperature contrast areas there were in a d isp lay, the higher the arousal leve l of that d isplay would be. Figure 71 shows th i s assoc iat ion. Since temperature contrasts were conceived as dif ferences of hue temperature rather than as absolute measures of hue temperature, i t i s not possible to speak of warm and cool colors in th is connection. Rather, the inc lus ion of th i s term in the equation seemed to iliend addit ional support to the fact that the adjacent-difference concept was f i rmly associated with arousa l . Most of the variance accounted fo r by the quadratic equation was due to the top/bottom-chroma term in i t s quadratic form. From the ana lys i s , an inverted U-function emerged with a peak around the point where the chroma of the top ha l f of the display was equal to the chroma of the bottom ha l f of the d isp lay. Since the pa r t i a l 303. High arousal Arousal Low arousal Low Equal High l e f t l e f t - r i g h t l e f t value value value (dark) ( l i gh t ) Lef t / r ight-va lue FIGURE 70 Approximate regression l i ne f o r the l e f t / r i gh t -va lue component 1n the l i near equation fo r arousa l . 304. High arousal Arousal Low arousal Low temperature contrast High temperature contrast Adjacent/di fference-temperature FIGURE 71 Approximate regression l i ne fo r the adjacent/d1fference-temperature component i n the l i nea r equation fo r arousa l . 305. corre lat ion coe f f i c i en t was - .46 , the inverted U-curve followed a d i rect ion which a l i near function with a negative corre lat ion coe f f i c i en t of th i s order would fo l low. Figure 72 shows the approxi-mate form o l i t h i s regression curve. Referring to th i s f igure i t 1s evident that , as the top ha l f of the display became increas ingly sat-urated ( in re la t i on to the bottom h a l f ) , a decreased arousal level resulted whi le , as the bottom ha l f of the d isplay became increas ingly saturated ( in re la t ion to the top h a l f ) , a decrease in arousal level also resu l ted. Only when the chroma or saturat ion leve ls of the top and bottom halves of the display were approximately equal did the subjects f ind the d isplay maximally arousing. Two other terms entered into the quadratic equation fo r arousal: figure/background-chroma and adjacent/var iance-AE. Both of these were l i nea r terms. The f i r s t showed that in cases where the f igure was of a higher chroma than the background, the resul tant arousal scores were low, whereas in cases where the background had a higher average chroma than the f i gure , the perceived', arousal was higher. Figure 73 represents graphica l ly th i s re la t ionsh ip . The l a s t term to enter into the quadratic equation fo r arousal was adjacent/variance-^E which accounted fo r 4.5% of the to ta l var iance. Figure 74 shows the approximate d i rec t ion of th i s component. From f igure 74 can be seen how, as the numerical value of adjacent/var iance-Al increased, the arousal level decreased. Since the re lat ionsh ip was constructed as a measure of the overa l l contrast 3 0 6 . High arousal Arousal Low arousal Low Equal High top top-bottom top chroma chroma chroma Top/bottom-chroma FIGURE 72 Approximate regression curve fo r the quadratic form of top/bottom-chroma versus arousa l . 307. High arousal Arousal Low arousal Low Equal High f igure f i gure- f igure chroma background chroma chroma Figure/background-chroma FIGURE 73 Approximate regression curve for the figure/background-chroma component 1n the quadratic equation fo r arousa l . 308. High arousal Arousal Low arousal Low color contrast High co lor contrast Adjacent/variance- AE FIGURE 74 Approximate regression curyeffor the adjacent/variance- AE component of the quadratic equation fo r arousa l . 309. wi th in a display ( i . e . , between major and complementary co l o r s ) , i t 1s evident that the greater th i s co lor contrast was, the lower the arousal response turned out to be. However, since the con t r i -bution of th i s var iable to the to ta l variance accounted fo r by the quadratic equation was sma l l , th i s conclusion i s only tenta t ive . D is t r ibut ion spec i f i ca t i on and dominance The overa l l F-stat1st1c was s i gn i f i c an t in both the l i nea r and quadratic case, but the low per cent variance accounted for Indicated only a very weak associat ion between lef t/r ight-chroma, l e f t / r i g h t -temperature and dominance. Both of these associations were i nd i ca -ted by negative pa r t i a l corre lat ion coef f i c ients o f almost equal magnitude (-.25 and - .26): the higher the chroma and perceived temperature of the l e f t ha l f of the displays were, the lower the mean dominance scores. That I s , the more" saturated and the "cooler" the l e f t ha l f of the displays were, in -relation- to the r igh t ha l f , the less subjects f e l t " in con t ro l " . D is t r ibut ion spec i f i ca t i on and information rate The l i near regression analysis f o r information rate selected three variables which, in descending order of contr ibut ion to the to ta l variance accounted fo r , were: adjacent/difference-chroma (19.6%), left/r ight-chroma (12.5%) and average-value (3.8%). The most prominent var iab le , adjacent/difference-chroma, had a pa r t i a l cor re la t ion with information rate of .44 which meant that as the overa l l average of 310. chroma contrasts Increased, information rate scores increased. That 1s, the higher (and more frequent) the chroma contrasts in a d isplay were, the more "uncertainty" the subjects perceived to be in the d isp lay. Conversely, the smal ler these chroma contrasts were, the more " ce r ta in " , f am i l i a r , redundant, homogeneous, e t c . , the sub-jects judged the d isp lay. Figure 75 shows th i s re la t ionsh ip . The second most Important var iable 1n the l i nea r equation was left/r ight-chroma with a pa r t i a l corre lat ion with information rate of .39. Thus, in cases where the l e f t ha l f of a display had a higher average chroma than the r igh t ha l f , the information rate was high, whereas when the l e f t ha l f of a display had a lower average chroma than the r igh t ha l f , the information rate score on that display was low. Figure 76 shows th is re la t ionsh ip . The t h i r d component in the l i nea r equation was average-value with a pa r t i a l corre lat ion of -.24 with information rate . The approxi-mate regression l i ne of th i s component i s shown in f igure 77. From th is f igure can be seen how overa l l l i g h t displays rated as more certa in ( i . e . , low in Information) than dark d i sp lays , and how dark displays rated as more uncair$at# (4«e.» high in Information). The most prominent of the terms in the quadratic equation fo r information rate was top/bottom-value. I t accounted fo r more than ha l f (21.2%) of the to ta l variance accounted fo r (38.2%). The approxi-mate curve re f l e c t i ng the pa r t i a l corre lat ion coe f f i c i en t of -.46 i s shown in f igure 78. Here, displays with average values larger in the 311. High Information rate Information rate Low Information rate Low chroma contrasts High chroma contrasts Adjacent/d1fference-chroma FIGURE 75 Approximate regression l i n e for the adjacent/difference-chroma component of the l i near equation fo r Information ra te . 312. High Information rate Information rate Low Information rate Low Equal High l e f t l e f t - r i g h t l e f t chroma chroma chroma Left/rlght-chroma FIGURE 76 Approximate regression l i ne fo r the left/r ight-chroma component of the l i near equation fo r information r a te . 313. High information rate Information rate Low Information rate Low average value High average value Average-value FIGURE 77 Approximate regression Une fo r the average-value component of the l i nea r equation fo r information ra te . 314. High information rate Information rate Low Information rate Low top value Equal top-bottom value High top value Top/bottom-value FIGURE 78 Approximate regression curve for top/bottom-value In the quadratic equation fo r Information ra te . 315. top ha l f of the display ( I . e . , l i gh te r ) than in the bottom ha l f resulted in lower information rate scores than displays with low average values ( i . e . , darker) in the top ha l f and high average values in the bottom ha l f . In other words, displays which were i l i gh t e r at the top resulted i n more certa inty than displays which were darker at the top. The two remaining components of the quadratic equation were left/r ight-chroma and average-value, and the i r character i s t i cs were p ra c t i c a l l y ident i ca l to those found for the same variables in the l i near model. Figures 75 and 77 show the associat ions with Information rate which these components resulted i n . 316. CHAPTER VIII DISCUSSION Introduction The present study has dealt with the larger issue of co lor and emotional responses and, wi th in that framework, examined in pa r t i cu la r the problem of complex s t imul i and the i r spec i f i c a t i on . I t was experimental in the sense that i t employed a type of st imu-lus which had not been used previously, and fo r which therefore no spec i f i c predict ions of emotional responses could be s tated. A unique approach which resulted in what was termed "hybrid s t imu l i " was adopted in the construction of the s t i m u l i . The cha l -lenge was essent i a l l y one of producing displays which on the one hand could be spec i f i ed in co lor imetr ic terms and which, on the other, were as much l i k e color pictures as poss ib le. The resu l t ing s t imu l i , or d isp lays, which f i l l e d these two requirements consisted of p i c tu re - l i ke assemblages of large numbers of ind iv idua l co lo rs . Accurately spec i f i ed Munsell colors were selected for use in 317. these displays with the resu l t that displays could be constructed 1n f i ve hue versions, two value versions and two chroma vers ions. The advantage of using the Munsell colors was pr imar i ly that the 20 color var iat ions which resulted sampled the co lor space in per-ceptual ly equidistant steps. In addit ion to the co lor var ia t i ons , each of the 20 displays were produced in four motif versions making a to ta l of 80 d isp lays. This was done in order to make the displays more p i c t u r e - l i k e , but i t introduced the p o s s i b i l i t y of a cognit ive inf luence on the emotional response measures due to motif recogni t ion. The problem of stimulus spec i f i ca t i on was formulated in terms of three requirements: 1. Component spec i f i ca t i on: A spec i f i ca t ion of the properties of the ind iv idua l co lor elements making up the d isp lay. 2. Quantity spec i f i ca t i on : An assessment of the quantity of these elements In each d isp lay. 3. D is t r ibut ion spec i f i ca t i on : The locat ion or d i s t r i -~ bution across the display surface of these co lor elements. Subsequent to the construction of the d i sp lays , the ind iv idua l co lor elements in the displays were measured according to standard c o l o r i -metric methods and spec i f ied in teirms of C L E . t r i s t imulus values. This, in addit ion to the Munsell notation of the colors used, provided 318. yet another very accurate check on the colors used 1n the d i sp lays . To meet the second requirement of spec i f i c a t i on , the number of ind iv idual hues, values and chromas in each display were ca lcu-la ted . F ina l l y to meet the t h i r d requirement of the spec i f i c a t i on , that 6f d i s t r i bu t i on or l o ca t i on , a var iety of approaches were proposed. In pa r t i cu l a r , 24 ways of ca lcu lat ing d i s t r i bu t i on re lat ionships with in the displays were developed. Although the emphasis throughout the study was on the displays and the i r spec i f i c a t i on , the emotional response measures of pleasure, arousal Band dominance and the measure of information rate played a prominent ro le as w e l l . F i r s t , the measures of pleasure, arousal and dominance provided a r e l i ab l e and convenient way of determining the emotional responses to the d isp lays . In add i t ion , the measure of i n -formation rate suppl ied the means fo r assessing the very d i f f i c u l t problem of cognit ive influences on the emotional responses. Secondly, the mean scores fo r the emotional measures and i n fo r -mation rate were presented as yet another way of spec i fy ing the d isp lays . Four hypotheses and a general research question were formu-la ted . The th i r d and fourth of the hypotheses were accepted, while the f i r s t and second were rejected in part . The research question led to a number of in terest ing conclusions which w i l l be discussed shor t l y . 319. The hypothesis dealing with the display var iables of co lo r , motif  and the subject var iable sex versus the emotional response measures  and information rate . Hue. Of the three dimensions of co lo r , hue has probably always been the most enigmatic. This i s par t ly because hue has been equated with "co lor" , as was pointed out in several instances, and as such has been expected to be the sole var iable responsible fo r emotional responses. The studies which have found s i gn i f i c an t e f fects due to hue have furthermore usual ly been of the preference-type ( i . e . , Granger, 1955a), and i t i s undoubtedly the case that when a subject i s presented with a se lect ion of hues (with value and chroma held constant), a s i tua t ion of forced se lect ion i s created i n which preference-orders 6*f some sor t are bound to emerge. The fac t that these preference-orders in many cases have shown a good deal of agreement among observers does not a l t e r an experimental s i tua t ion which invar iab ly guarantees responses to hue. In the present study, hue was replaced by the notion of "pre-dominant hue", and as a consequence of t h i s , d i rec t comparison of the f indings of previous preference studies with those 6f the present one must be done with caut ion. Predominant hue by i t s e l f did not inf luence scores on pleasure, arousa l , dominance or information rate. However, when hue was combined with mot i f , s i gn i f i can t e f fects on pleasure and dominance emerged. 320. That i s , certa in hue-motif combinations were more pleasant and more dominant than others, and these combinations were quite d i f fe rent from those at t r ibutab le to motif alone. Two aspects of th i s resu l t are noteworthy. F i r s t , i t was evident that to speak of hue alone i s not very usefu l: hue never occurs in a vacuum. In the case of many previous s tud ies , hue was associated with colored pieces of paper while in the present study i t was shown that hue (apart from i t s obvious connection with the physical character i s t i cs of the s t imu l i ) was associated with mot i f . This f ind ing makes i n t u i t i v e sense also since the pleasure one would get from looking at a p icture of a red apple most l i k e l y would not match that which one would get from looking at a p icture of a green apple. The f ind ing also lends weight to Gui l ford and Smith's (1959) note of caution about taking the object with which the co lor i s associated into account. I t was surpr is ing to f ind that hue did not have any e f fec t on arousal scores since there i s some evidence to the e f fec t for instance that red is! more arousing than green. (Wilson, 1966) These resu l ts however were based on experiments with saturated hues, and i t i s possible that the e f fec ts were a resu l t of a combination of hue and saturat ion rather than of hue alone. In the present:case, furthermore, the hues were of medium saturat ion which might have prevented a s i gn i f i c an t in teract ion of hue and chroma to assert i t s e l f 321. on the arousal dimension. The lack of s ign i f i cance of hue on information rate scores, on the other hand, was not surpr is ing in view of the fact that adjec-t i ve pairs such as simple-complex, patterned-random, e t c . , were the sub-scales of th i s measure, There 1s no reason to bel ieve that hue should have anything to do with the assessment of these physical qua l i t i e s of the d isp lays . There was a strong e f fec t on dominance from the hue by sex in te rac t i on . However, males found YR-B to be the leastddominatlng while females found th is predominant hue scheme to be the most dominating, and the order of dominance for the remaining hues was the only one on which the scores d i f f e red . Value. The ef fects of value—the second dimension of color—emerged as pa r t i cu l a r l y strong from the resu l ts of the study. This f ind ing was in general agreement with much previous research ( e . g . , Helson and Lansford, 1970). Although predominantly l i g h t displays were gen-e ra l l y preferred to predominantly dark d i sp lays , the darker abstract was preferred to the l i gh te r abstract as opposed to the preference fo r the l i gh te r version of the face, landscape and bu i ld ings . This s i gn i f i c an t in teract ion of value and motif indicates that , as in the case of hue, value was associated with motif on the dimension of pleasure. 322. A s im i l a r type of re lat ionsh ip between value and motif was evident from the responses on information rate: l i gh t faces, land-scapes, and bui ldings were judged more "cer ta in" while dark abstracts wei?e judged more "uncerta in". For dominance, th i s connection between value and motif d id not manifest i t s e l f . However, value as a main e f fec t was s i gn i f i c an t and l i gh te r displays were found to be less dominating than darker d isp lays . Value did not a f fec t scores on the dimension of arousal . I t i s in teres t ing to consider possible explanations fo r why value and motif were associated on the dimensions of pleasure and Information rate . When the four motifs were created, they were made up using co lor chips selected fo r the predominantly high value ver-s ions, and these four "prototype" motifs were made as c l ea r l y recog-nizable (or unrecognizable in the case of the abstract) as the experi menter could make them. As predominantly low value versions of the motifs were subsequently constructed, no such control was poss ib le , and the recogn izab i l i ty of the motifs probably suffered as a r e su l t . In pa r t i cu l a r , the dark face did not seem as recognizable as the l i gh t face. I t would seem plausib le that certa in mot i fs , such as the face, would be more unpleasant in non-natural (rather than simply dark) value configurations than other mot i fs , and th is seems to be sup-ported by the motif recognit ion comments since several subjects 323. described the predominantly dark face as a mask. By the same token, information rate scores would r e f l e c t uncertainty in cases where the motif was not eas i l y recognized and, in add i t ion, uncertainty may have resulted because of the discrepancy between what a face should look l i k e and what the low value d isp lay—with d is tor ted values of the f a c i a l features—looked l i k e . On the other hand, the post hoc comparisons showed that there were only s i gn i f i c an t di f ferences on pleasure scores between high and low value displays when subjects looked at landscapes and bu i ld ings , and these motifs would not seem to be nearly as susceptible to the value d i s to r t i on mentioned as the face was. Thus, the pecu l i a r i t y of the construction technique was not the main reason for the value-motif assoc iat ion. The fac t that the abstract was neither pa r t i cu l a r l y l i ked nor d i s l i k ed according to value perhaps shows that i t i s d i f f i c u l t to say that something i s pleasant or unpleasant without also knowing what i t i s . I t was however a lso shown that the high-low value d i s t i n c t i on pers isted on the information rate dimension for faces and landscapes but not for bui ld ings and abstracts . T(jus the construction technique may in the case of information rate have contributed to the s i g n i f i -cant value by motif i n te rac t i on . The conclusion that high value displays were general ly more pleasant than <Dow value displays i s in agreement with f indings of 324. previous research with s ing le co lor s t i m u l i . Granger (1955a), f o r instance, found that moderately high value leve ls were preferred to low value l eve l s , and Wright and Rainwater (1962) found that high values rated high on the i r happiness dimension. Furthermore, the present f ind ing that l i gh t displays were less dominating than dark displays also agrees with Wright and Rainwater's (1962) resu l ts that high values e l i c i t lower rat ings on the i r forcefulness dimension. Chroma. The th i rd dimension of co lor—chroma—signi f icant ly inf luenced scores on pleasure and information rate only and no interact ions i n -volving chroma were s i gn i f i c an t . This lack of more widespread ef fects due to chroma contradicts to some extent the f indings of Wright and Rainwater (1962) which showed that of the three dimensions of co lo r , chroma was that which most pos i t i ve l y accounted fo r the responses on the i r three main factors of happiness, showiness and forcefu lness. However, these authors used chromas with a mean of 9 (I .e., of moderately high saturat ion) while in the present study, the chroma leve ls used were res t r i c ted to 2, 4 sand 6, with a mean for the low chroma displays of 3.55 and 4.45 fo r the high chroma d isp lays . Thus, the discrepancy between the two €1ridingsmay be due to the dif ference in saturat ion of the s t i m u l i . This conclusion is! further enhanced by the fact that some 325. subjects reported verbal ly to the experimenter that some of the displays looked the same. A c loser examination of these displays revealed that they were the same in a l l aspects except fo r d i f f e r -ences in chroma. The high-low chroma d i s t i n c t i on was c l ea r l y a very subtle one, and i t i s quite possible that s t imul i with stronger chroma differences would have changed the ef fects of chroma consider-ably. Unfortunately, the way the systematic sampling of the Munsell co lor space was conceived made i t impossible to use more extreme levels of chroma. In sp i te of the subtle chroma d i f ferences, then, i t was i n t e r -est ing to f ind that pleasure and information rate scores were s i g -n i f i c an t l y d i f f e rent due to th i s dimension of co lo r , and one might again speculate about the e f fects of greater chroma d i f ferences. Possibly highly saturated displays would be even more strongly pre-fe r red , although at extreme saturat ion levels pleasure may decrease, i f Granger's (1955a) results fo r s ingle colors apply here. As fa r as information rate scores are concerned, i t i s uncertain j u s t what more extreme chroma differences would resu l t i n . F i n a l l y , the same stimulus pecu l i a r i t y may account for the lack of s i gn i f i c an t e f fects on arousal and dominance. Tge research c i ted e a r l i e r , by Wilson (1966), would indicate that higher saturat ion levels would inf luence at least arousa l , and i t seems to make i n t u i t -ive sense that the higher the saturat ion, the more dominated subjects 326. would f e e l . These l a t t e r two suggestions would have to be i n v e s t i -gated byrft irther experiments. Moti f . Motif was c l ea r l y the one var iable which most pers i s tent ly influenced scores on the four dependent measures. I t was the only var iable a f fect ing arousal scores, and i t par t i c ipated in 5 of the 6 s i gn i f i c an t in terac t ions . The method of constructing the motifs from square co lor chips and superimposing an aperture mask may have been the main reason fo r the face being unpleasant and the landscape and bui ld ings being pleasant. The construction technique made a l l of the motifs some-what angular, but perhaps th is was found most objectionable in the case of the face since faces are usual ly made up of curves, rather than r ight angles. The abstract , which one should have thought was the motif least affected by the construction technique turned out to be neither pleasant nor unpleasant. The s i gn i f i c an t e f fects of motif on arousal are of pa r t i cu la r in te res t because (1) motif was the only var iab le a f fect ing arousal and (2) because these resul ts shed some further l i g h t on the ef fects of motif on pleasure. I t was in terest ing to f ind that motif was s i gn i f i c an t on arousa l , but one may wel l wonder why no other var iables were. The simplest and perhaps the most p laus ib le reason may be that the hue, value and chroma var iat ions in the displays were j u s t too subtle to 327. create much excitement and thus more extreme arousal scores. The or ig ina l response measure scales ranged from -4 to +4, yet the mean scores f o r the 80 displays never exceeded the range of -1.377 to 1.323 so, judging from the overa l l lack of extreme responses on a l l the response dimensions a l l the at t r ibutes of the displays were indeed very subt le . The abstract motif turned out to be the most arousing, as wel l as the most "uncerta in", which supports Meharabian and Russe l l ' s (1974) claim that arousal scores corre late pos i t i ve l y with information rate scores. The face was the second most arousing as wel l as the second most "uncertain" moti f , again supporting th i s c la im. Also in decreas-ing order of arousal , the representational motifs were: face, bu i l d -ings and landscape, and th i s order i s prec ise ly the reverse of that found f o r pleasure. In general, the more arousing a moti f was, the less pleasant i t was. This f i nd ing , however, contradicts Mehrabian and Russe l l ' s (1974) idea of independence of the emotional measures, but i t may also explain why the face was judged so unpleasant. I f the independence hypothesis i s correct i t means that the arousal measure in the present case may have assessed feel ings of i r r i t a t i o n , f rus t ra t ion and perhaps confusion rather than simply arousal or excitement. The f indingsthat the abstract was scored highest on informa-t ion rate was g ra t i f y ing since i t showed that indeed a display 328. without representational motif causes some fee l ing of uncertainty. A l so , th i s f ind ing i s supported by a study by Chipman (1977) in which she f inds that lack of structure resul ts in greater perceived complexity. On the other hand, the lack of a motif in the abstract , at the same time evidently enables some subjects to read the i r own motif into the d isp lay , and th is may have contributed to the mean pleasure rat ing fo r the abstract being at least not negative. The resu l t of the motif by sex in teract ion on pleasure was interest ing mainly to the extent that i t showed that both males and females agreed about the unpleasantness of the face. Sex. The present study found s i gn i f i c an t overa l l di f ferences between males and females on pleasure and dominance scores but not on arousal or information rate scores. The lack of s ign i f i cance in general on arousal scores was discussed e a r l i e r , while the lack of dif ferences on information rate scores may be due to the fact that th is was the most object ive (and the least a f fect ive) of the four measures. The question of emotional dif ferences due to sex i s a very d i f f i c u l t one to deal w i th . Some wr i ters have found differences in co lor preferences due to sex while others have not, and i t has been claimed from time to time that females—perhaps because of p r i o r condit ioning—are more sens i t ive to colors than males. The present resu l ts c l ea r l y showed 329. strong sex differences on pleasure^and dominance, judging from the s i ze of the F - s t a t i s t i c (c f . table 45), and possibly the same would have been the case for arousal , had the displays been more exc i t ing as mentioned. I f i t i s true that females are more sens i t ive to co lor , the conclusions that they found the displays more pleasing and more dominating than the males d id make sense. A person who i s sens i t ive to color would be expected to show more in teres t in the d isp lays , to show more exploratory behavior, as Saklofske (1975) points out ,and, as a resu l t , f i nd the displays more p leas ing. A lack of in te res t i n the d isp lays , which would characterize the behavior of a person who i s insens i t i ve to co lor , at the same time might resu l t in an at t i tude of not being subjected to the aesthet ic message of the d isp lay, but of being " in contro l " of i t . The hypothesis dealing with the representational motifs of face,  landscape and bui ld ings versus the abstract . I t was predicted in the above hypothesis that motif in general would inf luence scores on pleasure, arousal , dominance and information rate. The present hypothesis predicted, that the representational motifs of face, landscape and bui ld ings would e l i c i t responses s i g -n i f i c an t l y d i f fe rent from those of the abstract . The reason fo r the inc lus ion of the present hypothesis was that by the introduct ion of motifs a cognit ive component would probably be 330. added to subjects ' a f fec t ive responses. That i s , i t was assumed that non-representational designs such as the abstract would e l i c i t re-sponses which were as devoid of cognit ive content as one could possibly expect but that , when a recognizable motif was added, the subject would be affected by such cognit ive components as meaning, recognit ion, f am i l i a r i t y with the moti f , e t c . The conclusions reported e a r l i e r showe'd c l ea r l y that in a l l but one case in which motif was s i gn i f i c an t l y involved did subjects ' scores d i f f e r according to whether they saw a representational motif or the abstract . The one exception was the e f fects due to motif on pleasure, and an examination of the mean scores fo r pleasure on the four motifs showed that th is exception was probably due to the extreme negative means Score for the face. Although the introduct ion of motif thus changes the responses from purely a f fec t ive ones to S f fec t ive-cogn i t i ve , i t appears that the pleasure, arousa l , dominance and information rate dimensions are s t i l l quite capable of character iz ing the responses. The motif recognit ion hypothesis dealing with the display var iables  of hue, value, chroma, motif and the subject var iable of sex versus  motif recogni t ion. This hypothesis was advanced s ince, i f a cognit ive component of recognit ion and meaning inf luenced responses on pleasure, arousal , 331. dominance and information rate, i t would be highly informative to ask what the subjects saw in the d i sp lays . Furthermore, since what the subjects sa id they saw in the displays presumably was inf luenced by the display variables of hue, value and motif as wel l as perhaps by sex, a separate analysis with these var iables and motif recog-n i t i on scores was carr ied out. I t was surpr is ing and somewhat disappointing to f ind that the overa l l recognit ion rate reached only 51*9%. This seems part iCui lar ly Tow in view of the specia l care which had been taken to construct inecognizable mot i fs . The motif of bui ld ings was the one most eas i l y recognized, perhaps because of the re l a t i ve i n s en s i t i v i t y of square structures to the rather specia l construction technique, while land-scapes rated very poorly in terms of recognit ion. Previously, re f -erence was made to the fact that faces might have been found un-pleasant because of the angular treatment, y,et that motif scored second highest on motif recognit ion and at the same time, land-scapes were found the most pleasant mot i fs . I t seems therefonei/inot reasonable to a t t r ibute lack of recognit ion of landscapes to the construction technique. One aspect of motif which has not been mentioned so fa r i s that of sca le , and i t i s possible that th is factor may have been respons-i b l e fo r the low landscape recognit ion scores. The face seemed to f i t wel l with in the frame of the display in the sense that the features 332. necessary fo r the recognit ion of a face were a l l there, and as f a r as the bui ldings were concerned, a few large rectangles would su f f i ce to give the impression that th i s motif represented bu i ld ings. The abstract, of course did not encounter th i s problem of sca le . The motif of landscape, however, may have t r i ed to show more features than the l im i ted matrix s ize could handle. Perhaps i f instead of a t ree , a cloud, sky, ground and a lake, only a tree and sky and ground had been represented, the landscapemmight have been c loser to the scale of the d isp lays, and recognit ion might have improved. The hypothesis, as i t was s tated, was accepted because the resul ts showed substant ia l e f fects of value, motif and hue on motif recogni t ion. Chroma and sex dif ferences however did not s i gn i f i c an t l y inf luence motif recognit ion. I t was not surpr is ing to f ind that value and motif inf luenced motif recognit ion. For one th ing , these var iables were both strong influences on the a f fec t ive measures and information rate. They were c l ea r l y the two display character i s t i cs which stood out the most,1in contrast to hue and chroma, and they seem in general to have demanded the most a t tent ion . For another, i t was reasonable to suspect that motif would contribute to motif recognit ion since th is var iable was s pe c i f i c a l l y manipulated to achieve th is purpose. In some ways, the l a t t e r point appl ies to the var iable of value as w e l l . Although not de l iberate ly designed to enhance recognit ion, i t i s apparent that only 333. with great d i f f i c u l t y can one construct a recognizable motif with hue or chroma determining the motif while value i s l e f t randomly d is t r ibuted across the display surface. In other words, value seems to be that dimension of co lor which in the main determines the motif. This fact i s also 'supported by prac t i ca l experience, since f i r s t , when the experimenter created the s t imu l i , no thought was given to par t i cu la r dimensions of co lor , yet in order to achieve motif recog-n i t i o n , value seemed natura l ly to assume i t s place as the major co lor var iable in motif recognit ion. Figures 44 and 45 (chapter IV) i n d i -cates graphica l ly the extent to which value contributed to the motif patterns. Secondly, th is s i tua t ion i s probably typ ica l of most p i c -t o r i a l mater ia ls . For instance, in pr inted co lor pictures the black plate i s usual ly the key to the sharpness of the moti f , and in co lor te lev i s ion pictures i t i s the brightness components of the three primaries which determine the motif , ( e . g . , Hunt, 1967). Hue and chroma do not seem to contribute to the motif pattern in the ways value does. I t seems feas ib le of course that motifs could be constructed in which value i s constant and e i ther hue or chroma i s varied separately. In the present case, chroma was not a s i gn i f i c an t contr ibutor to motif recognit ion, but hue was. A v isua l inspection of f igures 43, 46 and 47 (chapter IV) gives andideaaof how hue and chroma part ic ipated in th i s task. The chroma patterns do not resemble the motifs at a l l whereas the hue patterns (major versus 334. adjacent and complementary hues) indicate that, without intending to control th i s var iab le , some weak patterns resembling the four motifs emerged. Perhaps these underlying hue patterns account for the s ign i f i cance of hue on motif recognit ion. The hypothesis about motif recognit ion and information rate. I t was hypothesized that the overa l l measure of information rate (representing at least some of the ind iv idua l items of the Information rate questionnaire such as famiit iar-novel, usua l-surpr is ing, common-rare and patterned-random) would predict the extent of motif recognit ion. The resul ts showed th is to be true and i t was found that there was a negative corre lat ion between the two. Since th is re lat ionsh ip emerged i t i s reasonable to assume that , given the r ight circumstances, an even higher degree of associat ion may be shown to ex i s t , perhaps through recourse to a higher order regression equation. At the same t ime, there would c l ea r l y be a l im i t to the extent of associat ion since perhaps only the four items from the questionnaire mentioned would a f fec t motif recognit ion, while the remaining 10 might be quite insens i t ive measures of t h i s . The-research question: 24 d i s t r i bu t i on spec i f i ca t ion var iables  versus pleasure, arousa l , dominance and information rate . 335. Although only a r e l a t i ve l y small number of the d i s t r i bu t i on spec i f i ca t i on variables turned out to be s i gn i f i c an t l y associated withtthe three emotional measures and information rate , i t was gra t i f y ing to f i nd that at least some of them were. In pa r t i cu l a r , i t was rewarding to f ind that those which did associate, exhibited a considerable degree of s t rength, thus providing a sound base on which to construct hypotheses for future research. Figure 79 shows an overview of the var iables entering into e i ther the l i nea r or quadratic regression equations for pleasure, arousal and information ra te . Dominance i s included here, but the per cent variance accounted for was too small to resu l t in any pos i t ive conclusions. The var iables with an aster isk Indicate that these were variables which contributed most strongly to the s i g n i f i -cant associat ion in the Individual equations. As can be seen from th is f igure , of a to ta l of 24 variables (48 in the case of the quadratic equations) there were 14 which pre-dicted mean scores on pleasure, arousal.dominance and in fomat ion ra te , and .o l these, 5 were pa r t i cu l a r l y strong. Also; i t i s noticeable that a l l s i x types of d i s t r i bu t i on spec i f i ca t ions were represented. The average content of the displays was represented by the variables of average- 4E, average-value, average-chroma and average-temperature, of which only average-value turned out to be s i gn i f i c an t in i t s predict ion of information rate scores. This r esu l t , that 3 3 6 . Var iab le L inear equations Quadratic equations Average-value Information rate Information rate Fi gure/background-chroma Arousal *Top/bottom-value Pleasure Pleasure Information rate *Top/bottom-chroma Pleasure Pleasure Arousal Left/r1ght-value Arousal *Left/r1ght-chroma Information rate Dominance Pleasure Information rate Domi nance Le f t / r i ght-tempe ra tu re Dominance Dominance Adjacent/d i f ference- AE *Adjacent/d1fference-value Pleasure Arousal •Adjacent/difference-chroma Information rate Adjacent/d1fference-temperature Arousal Adjacent/var iance- AE Adjacent/variance-chroma Adjacent/variance-temperature Pleasure Arousal Pleasure * Indicates var iab les which contr ibuted most strongly to the regression equations. FIGURE 79 Overview of var iab les enter ing Into the l i n ea r and quadratic regression equat ions. 337. l i gh te r displays e l i c i t lower information rate scores, i s comparable to that found e a r l i e r when the ind iv idua l display variables were compared to the emotional response measures and informationrrate. Value was c lear ly a very strong var iable and, as mentioned e a r l i e r , chroma might have been prominent as wel l had the chroma contrasts been greater. I t i s also in terest ing to note that had the value contrasts been smal ler, the prominent ef fects of value might not have shown up, and i t i s even possible that there i s a minimal contrast level beyond which the type of resu l t found here might not occur. Further experiments would have to ascertain t h i s . The lack of associat ion between average-AE and average-temperature, and the dependent variables can probably be at t r ibuted to the fact that these measures were not appropriate fo r the assess-ment of d i s t r i bu t i on spec i f i ca t i ons . The concept of figure-background was not very successful as a predictor of scores on the dependent measures. Only the f igure/ background-chroma var iable associated s i gn i f i c an t l y with arousal , and then not very s t rong ly . The Hack of success i s possibly due to the d i f f i c u l t i e s of d iv id ing the surface into fdigure and background Although there was good agreement between the small group of observers who were used for th is assessment, the task of deciding which parts were f igure and which were background was very d i f f i c u l t due to (1) the d i f fuse nature of the out l ines and (2) the diverse parts which in the 338. case of the landscape and bui ld ings made up the f igure and background. The fact that the figure/background-chroma was s i gn i f i c an t may be due to a coincidence of chroma levels and figure-background areas. This coincidence, however, was not further explored. The top-bottom d i s t i n c t i o n , i i n contrast, was a pa r t i cu l a r l y good predictor of scores on pleasure, arousal and information rate: the top/bottom-value var iable predicted scores on pleasure and information ra te , while the top/bottom-chroma var iab le predicted scores on pleasure and arousal . The resu l ts re la t ing to the top/bottom-value var iable showed that the l i gh te r the top ha l f of the display was in re la t ion to the bottom ha l f , the more pleasant and "cer ta in" was that d isplay found to be. Arnheim (1974) points out that the more bottom heavy a picture i s , the more pleasant i t i s , and Alexander and Shansky (1976) found that apparent weight was a decreasing function of value. Thus, the present resu l t i s in perfect accord with both of these f indings as fa r as pleasure i s concerned. However, in attempting to explain why bottom heavy displays should also rate lower on information rate , one would have to re fer to the previously mentioned connection between pleasure and information rate: that the less uncertain a stimulus i s , the more pleasurable i t i s . Information rate and arousal have been shown to corre late (Mehrabian and Russe l l , 1974), and Berlyne (1971) has shown that an inverted U-function describes the re lat ionsh ip 339. between pleasure and arousal: the highest pleasure leve ls are found in neither the most nor the least arousing s i t ua t i ons . The present resu l ts confirmed the cu rv i l i nea r re lat ionsh ip between top/bottom-value and information ra te , but information rate was a decreasing monotonic function of top/bottom-value rather than an inverted funct ion. The resu l ts of the var iable top/bottom-chroma's associat ion with pleasure showed that the higher the chroma of the top ha l f of the display was, in re la t ion to the bottom ha l f , the lower the pleasure. While Arnheim (1974) only speaks of weight in general, withou| mentioning at t r ibutes of co lo r , Alexander and Shansky (19764) found that apparent weight was an increasing function of chroma. The present resu l t agrees with these f indings since high chroma in the top ha l f implies a heavier top ha l f and thus an unpleasant combination. The resu l ts further showed that arousal was an inverted U-function of top/bottom-chroma. The maximum leve l of arousal was attained for an approximately equal balance of top and bottom chromas, while for e i ther greater chroma at the top ( i . e . , top heavy) or lesser chroma at the top ( i . e . , bottom heavy) arousal leve ls decreased. The l e f t - r i g h t display surface d iv i s i on seemed as prominent a predictor as the top-bottom d i s t i n c t i o n . The l e f t / r i gh t -va lue var iable 340. was associated with arousal: the l i gh te r t h e l l e f t hand ha l f of the display was, in re la t ion to the r ight hand ha l f , the lower the arousal l e v e l . According to standard a r t i s t i c pract ice ( c f . Arnheim, 1974), the l e f t ha l f of a p icture must have a greater apparent weight than the r ight ha l f , in order to at ta in a sa t i s fac tory v isual balance, and in the present case i t was found that " left-heavy" displays resulted in high arousal leve ls whereas "right-heavy" displays re-sulted in low arousal l e ve l s . The present resu l ts thus lent c red i -b i l i t y to t rad i t i ona l a r t i s t i c pract i ces . The high arousal leve ls in th i s case were probably low enough to be posit ioned on the ascending part of the inverted U-function (as described by Berlyrie, 1971), so arousal would correlate pos i t i ve l y with pleasure. The resu l ts of greater pleasure attained through a v isual balance produced by a display which i s left-heavy in regard to value i s thus in accord with t rad i t i ona l a r t i s t i c p rac t i ce . The left/r ight-chroma var iab le , on the other hand, turned out to contradict standard a r t i s t i c prac t i ce . Left-heavy chroma displays were unpleasant and rated high on information ra te , while from the resu l ts of the top/bottom-chroma var iable one shoDld have expected the opposite resu l t . The fo l lowing questions would have to be con-sidered in a fufcther invest igat ion of th i s contrad ict ion: (1) Do a r t i s t i c rules of thumb only apply to the value dimension? (2) How sound are Alexander and Shansky's (1976) f indings when appl ied to 341. chroma? (3) There are as yet no un iversa l ly accepted explanations for "apparent weight", apart from speculations which involve synes-the t i c interact ions between v is ion and kinesthesis (Alexander and Shansky, 1976), and a thorough invest igat ion of apparent weight and d i rect ion wouild possibly have to be carr ied out before the present resu l t can be explained. While the d i s t r i bu t i on spec i f i ca t i on concepts dealt with so far make immediate sense and have prac t i ca l appl icat ions to various p i c t o r i a l s i tua t i ons , the concepts of adjacent-difference and adjacent-variance are more involved and only in grossly s imp l i f i ed terms do they make i n t u i t i v e sense. Their main purpose was to assess the extent and frequency of contrasts with in a display surface, and the i r calculationswas explained in deta i l in chapter IV. The adjacent-difference var iable par t i c ipated Inf four of the regression equations, while that of adjacent-variance part ic ipated in three. The extent of the involvement shows that these variables contributed to the d i s t r i bu t i on spec i f i ca t i on of the displays to a considerable degree. Both the adjacent/d i f ference-AE and the adjacent/variance-AE var iables emerged/from the analys is: the f i r s t associated with pleasure while the second associated with arousal . The fact that these two variables were involved i s in teres t ing since they were the ones which in the most general way possible represented the co lor 342. contrasts with in d isp lays . However, the variance accounted for by these var iables was quite smal l , and the resu l ts were thereforeccon-sidered to be of questionable value. The adjacent/difference-value var iable was found to be asso-ciated with arousal: the greater the overa l l value contrast, the greater the arousa l . This resu l t i s noteworthy since i t i s very s im i l a r to that of Wright and Rainwater (1962) for pairs of co lors . I t has been noted by Granger (1955d) that the more complex a stimulus becomes, the less the ind iv idua l components and the more the com-ponent re lat ionships w i l l a f fec t the responses. The present resu l t does not confirm t h i s , and i t i s perhaps better not to adhere quite so s t r i c t l y to th i s d i s t i n c t i on since a f te r a l l the component re lat ionships of necessity are based on the charac ter i s t i cs of ind iv idual components. Both the adjacent/difference-chroma and the adjacent/variance-ehroma variables were prominent in the regression equations. The former was associated with information rate , the l a t t e r with pleasure. As chroma contrasts increased, information rate scores increased but pleasure scores decreased. These resu l ts support one another to the extent that i f information rate corre lates pos i t i ve l y with arousal (as Mehrabian and Russell (1974) showed, and which the present study confirmedi c f . table 40)v pleasure should increase with^ increasing information rate scores, which i t does. . Furthermore, Helson and Lansford (1970) in t he i r studies found that greater chroma contrasts 343. resu l t in greater pleasure, so i t i s possible that i t i s the adjacent/ variance-chroma concept which i s at fault-. On the other hand, Granger (1955b) found that preference scores tended to decrease as the s i ze o f chroma interva l was increased. Thus, the present f ind ing concurs with that of Granger. Two points must be remembered however5when t ry ing to understand these contradictory f ind ings. F i r s t , the studies referred to dealt with color pairs rather than complex s t imul i and second, the concep-tual framework for the distance and variance var iables i s not the same. The f i r s t point means that comparisons between the present f indings and previous resu l ts may in some cases not be meaningful, and the second point means that the di f ference between the di f ference and variance concepts may need to be further developed before these con-cepts are used in subsequent experiments. In the meantime, i t appears that the problem l i e s with the adjacent/variance-chroma var iable since the variance i t accounted for was only 5.5%, whereas that of the adjacent/difference-chroma var iable was 19.6%/ The f i n a l variables which entered into the regression equations were adjacent/di-fference-..tempera ture and.adjaceht/vari ance^temperature. In the former case, pleasure scores were an inverted U-function of adjacent/variance-temperature whereas i n the l a t t e r , arousal scores increased with increasing temperature contrast. These resu l ts seem i n t u i t i v e l y r igh t since the highest pleasure 344. scores would be expected from the lowest temperature contrasts (i.e«?,s from the most "temperature-harmonious" d i sp lays) , the lowest pleasure scores from displays with neither weak nor strong temperature contrasts ( i . e . , a rather bland composition), and medium high pleasure scores from displays with extreme temperature contrasts. S im i l a r l y with adjacent/difference-temperature and arousal: displays with high temperature contrasts would be expected to be highly arousing while displays with low temperature contrasts would be expected to rate low on the arousal dimension. Conclusion The present study endeavored to examine the larger problem of co lor and a f fec t in terms of emotional responses to prec ise ly spec i -f i ed complex co lor s t imu l i . Special emphasis was placed on the spec i f i ca t ion of these s t imul i since they were conceived as forming a bridge between the t r ad i t i ona l l y used simple color s t imu l i and colored pictures as exemplif ied by works of art and co lor photographs. The f i r s t part of the study examined emotional responses to the complex s t imu l i in terms of hue, value, chroma, motif and sex. Four general hypotheses were advanced of which two were accepted while two were p a r t i a l l y accepted. However, the explorat ion of even the" p a r t i a l l y accepted hypotheses provided so much ins ight in to the top ic atthand that t he i r formulation was wel l worthwhile. 345. In general , emotional responses to complex color s t imul i d id not d i f f e r that much from responses to simple s t imu l i . The only real differences occurred as a resu l t of the motif var iable which previous studies had not made use of. At the same time, the introduct ion of motif created new problems which need to be assessed in future research. In pa r t i cu la r , two problems w i l l have to be dealt wi th: (1) the response measures would c l ea r l y have to take the cognit ive component of motif meaning into account, and (2) the pecu l i a r i t i e soo f the display construction technique would have to be re-evaluated. I t was pa r t i cu l a r l y encouraging to f ind that the explorat ion of various stimulus spec i f i ca t ion schemes, which formed the second part of the study, resulted in so much so l i d evidence on which to base future hypotheses. Although much of th i s evidence agrees with the rules of thumb which a r t i s t s and designers have used for many years, i t has added a great deal of deta i l to th i s i n t u i t i v e knowledge. Even in t he i r present tentat ive state these f indings may be of value to the art and design community, as wel l as to the f i e l d of applied aesthetics in general. 346. BIBLIOGRAPHY Alexander, K.R. and Shansky, M.S. "Influence of hue, value, and chroma on the perceived heaviness of co lors . " Perception &  Psychophysics, 1_9, 1976. Arnheim, R, Toward a Psychology of Art: Col lected Essays. London: Faber, WRT. Visual Thinking. Berkeley: Univers i ty of Ca l i f o rn i a Press, T559T Art and Visual Perception. Berkeley: Univers i ty of ~ " Ca l i f o rn i a Press, 1974. Atneave, F. "Physical determinants of the judged complexity of shapes." J . Exp. Psychol . , 53, 1957. B a l l , V.K. "The aesthetics of co lor: a review of f i f t y years of experimentation." J . Aesthet. and Art C r i t i c i sm , 23, 1965. Berlyne, D.E. Con f l i c t , Arousal and Cur i os i t y . New York: McGraw-Hill, 1960. ' • Aesthetics and Psychobiology. New York: Appleton-Century-Crofts, 1971. B i rkhof f , G.D. Aesthetic Measure. Cambridge, Mass.: Harvard Univers ity Press, 1933. Burnham, R.W., Hanes, R.M. and Barte lson, C . J . Color: a Guide to  Basic Facts and Concepts. New York: John Wiley & Sons, 1963. Chipman, S.F. "Complexity and structure 1n visual patterns." J . Exp. Psychol . , 3, 1977. Clarksdn, M.E., Davies, O.L. and V lckers ta f f , T. "Colour harmony?" Colour, I .C . I , pub l i ca t ion , 1950. Cra ik, K.H., et a l . "Environmental psychology." New Direct ions in  Psychology. New York: Ho l t , Rinehart and Winston, 1970. Draper, N.R. and Smith, H. Applied Regression Ana lys is . New York: John Wiley and Sons, 196C 347. Eysenck, H.J. "A c r i t i c a l and experimental study of colour preferences." Amer. J . of Psychol . , 54, 1941. Farnsworth, D. "The Farnsworth-Munsel1 100-hue and dlchotomous tests fo r co lor v i s i o n . " J . dpt. soc. Amer., 33, 1943. Fechner, G.T. Vorschule der Aesthet ik. Le ipz ig: Breitkopf und HaVtel, 1 8 7 6 7 ' ~ Ferguson, G.A. S t a t i s t i c a l Analysis 1n Psychology & Education. New York: McGraw-Hill, 1971. Granger, G.W. "An experiemntal study, of colour preferences." J . Gen. Psychol . , 52, 1955(a). "An experimental study of colour harmony." J . Gen. Psychol . , 53, 1955(b). "Aesthetic measure applied to co lor harmony: an experimental t e s t . " J . Gen. Psychol . , 52_, 1955(c) ' "The predict ion o f preference f o r colour combinations." J« Gen. Psychol . , 52, 1955(d) Graves, M. The Ar t of Color and Design. New York: McGraw-Hill, 1951. Gu i l f o rd , J .P . "The a f fec t ive vakue of co lor as a function of hue, t i n t , and chroma." J . Exp. Psychol . , 17., 1934. • "A study in psychodynamlcs." Psychometrica, 4, 1939. "There i s system in co lor preferences." J . opt, soc. Amer?., 30, 1940. "System in co lor preferences." J . Soc. Motion P i c ; Eng., 5 2 , 1 9 4 9 . Gu i l f o rd , J . P . and Smith, P.C. "A system of color-preferences." Amer. J . Psychol . , 72, 1959. Harmon, L.D. "The recognit ion of faces." S c i e n t i f i c American, 229, 1973. ; Hays, W.L. S t a t i s t i c s f o r Psychologists. New York: Ho l t , Rlnehart and Winston, 1963. 348. Helson, H., and Lansford, T. The ro le of spectra l energy of source and background color In the pleasantness of object co lo r s . " Applied Optics, 9, 1970. Hogg, J . "A pr inc ipa l components analysis of semantic d i f f e r en t i a l judgments of s ing le colors and co lor pa i rs .? J . Gen. Psycho l . , 80, 1969. Hunt, R.W.G. The Reproduction of Colour. London: Fountain Press, 1957. Hunter, R.S. ' 'Photoelectric t r i s t imu lus colorimetry with three f i l t e r s . " Na t l . Bur. Std. C l r c . C429, 1942. Ishihara, S. Tests fo r Colour-Bl indness. London: Lewis, 1948. I t t e l son , W.H., et a l . An Introduction to Environmental Psychology. D. Dempsey (Ed.) New York: Ho l t , Rlnehart and Winston, 19747 I t ten , J . The Art of Color. New York: Van Nostrand Reinhold, 1961. Jacobs, K.W. and Suess, J . F . "Ef fects of four psychological primary colors on anxiety s t a te . " Percept, and Motor S k i l l s , 4j_, 1975. Judd, D.B. "The 1931 I .C . I , standard observer and coordinate system fo r co lor imetry." J . opt, soc. Amer., 23_, 1933. Judd, D.B. and Wyszecki, G. Color in Business, Science, and Industry. New York: John Wiley and Sons, 1963. Lakowskl, R. "Theory and pract ice of colour v i s ion testing:, a : \ review. , j B r i t ; J . Indstr . Med., 26, 1969. Lakowskl, R.• and SharpiL.fr.; Paper presented at 3rd Congress of the International Colour Assoc iat ion, Troy, New York, 1977. Le, C. and Tenesci, T. Tr iangular Regression Package. Vancouver: ^ Univers i ty of B r i t i s h Columbia Computing Centre, 1977. McAdory, M. Color in pa in t ing . Baltimore: The Munsell Color Co., 1926. Mehrabian, A. and Russe l l , J .A. An Approach to Environmental  Psychology. Cambrldge| 9Mass.; M.I.T. Press, 1974. Moon, P. and Spencer, D.E. "Geometric formularion of c l a s s i ca l co lor harmony," J . opt, soc. Amer., 34, 1944(a). 349. Moon, P. and Spencer, D.E. "Area in color harmony." J . opt, soc. Amer., 34, 1944(b). "Aesthetic measure applied to co lor harmony." J . opt, soc. Amer., 34, 1944(c) Munsel l , A.H. A Color Notat ion. Baltimore: The Munsell Color Co., 1913. Munsell Book of Color, Vo l . I , Neighboring Hues Ed i t i on . Baltimore: Munsell Color Co., 1950. Nlckerson, D. "History of the Munsell co lor system and i t s s c i e n t i f i c app l i ca t i on . " J . opt, soc. Amer., 30, 1940. Norman, R.D. and Scot t , W.A. "Color and a f fec t : a review and semantic eva luat ion." J . Gen. Psychdl,.°46,'il952. Osgood, C.E., Sus i , G.J. and Tannenbaum, B.H. The Measurement of  Meaning. Urbana: Univers i ty of I l l i n o i s Press, 1957. P ick ford , R.W. Psychology and Visual Aesthet ics . London: Hutchinson, 1972. Rabkin, E.B. Polychromatic Plates fo r Testing Colour V i s i on . Ukrainian Hlrschman Ophthalmic I n s t i t u t e , 1939. Russe l l , J.A. Approach-Avoidance Behaviors as Functions of the Emotion-El1clt ing Qua! i t ies of_Sett1ngs. PH.D. Thesis, Un1versity of Ca l i f o r n i a , Los Angeles, 1974. Saklofske, D.H. "Visual aesthet ics complexity, attract iveness and divers lve exp lorat ion." Percept, and Motor S k i l l s , 41 , 1975. S i v i k , L. "On the meaning of co lours ." Arch i tectura l Research  Review (Synthesis), 4, 1974. Valent ine, C.W. The Experimental Psychology of Beauty. London: Methuen, 19BX Wilson, G.D. "Arousal properties of red versus green." Percept, and  Motor S k i l l s , 23, 1966. Winer, B.J . S t a t i s t i c a l Pr inc ip les intExperimental"'Design. New York: McGraw-Hill, 1962. 350. Wright, W.D. The Measurement of Colour. London: Van Mostrand Reinhold, 1969. Wright B., and Rainwater, L. "The meanings of co l o r . " J . Gen. Psychol . , 67, 1962. Wyszecki, G. "Recent developments on color-d i f ference eva luat ions." Proc. Helmholtz Memorial Symposium on Color Metr ics, 351. Appendix A P i l o t study quest ionnaire. 352. S l i d e no. Try to determine how YOU FEEL as you look a t the s l i d e shown. HAPPY — : — - : : : — - UNHAPPY STIMULATED : :— ~ : : : : : - — : RELAXED CONTROLLING : : : : : : : — CONTROLLED ANNOYED — : : : : : : : PLEASED CALM : : : s : ; : : EXCITED INFLUENCED, — — : : : : : • - - : : INFLUENTIAL SATISFIED : — - : — - : : ; : — - : : UNSATISFIED FRENZIED : : : : : : SLUGGISH IN CONTROL : : : : : : : : CARED FOR MELANCHOLIC - — : : : : : : : : CONTENTED DULL : : : : i : : : JITTERY AWED : : : : : : : : IMPORTANT HOPEFUL : : : - — : : : — - : : DESPAIRING WIDE AWAKE : : : : : : : : SLEEPY DOMINANT : : : : : : : : SUBMISSIVE BORED — - : : : : : : : : RELAXED UfJAROUSED : : : : : : - — : : AROUSED GUIDED : : : :— « : — : : AUTONOMOUS Now, complete the ad j e c t i ve pa i r s below, except t h i s time t r y to make OBJECTIVE JUDGMENTS about the s l i d e shown. VARIED : : : : - — : : : : REDUNDANT SIMPLE : : : : : : : : COMPLEX NOVEL : : : : : : : : FAMILIAR SMALL-SCALE : : : : : : : : LARGE-SCALE SIMILAR : : : : : : : : CONTRASTING DENSE : : : : : : : : , SPARSE INTERMITTENT : : : : : : : : CONTINUOUS USUAL : : : : : : : : SURPRISING HETEROGENEOUS — - : : : : : : : : HOMOGENEOUS UHCROWDED : : : : : : : : CROWDED ASYMMETRICAL : : : : : : : : SYMMETRICAL IMMEDIATE : : : : : : : : DISTANT COMMON : : : : : : : : RARE PATTERNED : : : : : : : : RANDOM F i n a l l y , complete t h i s sentence f o r whatever you be l i eve about t h i s s l i d e : •This i s a s l i d e o f ." 353. Appendix B P i l o t study tes t booklet cover. 354. This 1s an experiment dea l ing w i th your percept ions . I t embodies some very subt le d i f fe rences i n a se r ies o f s l i d e s which you w i l l be shown s ho r t l y . I t ' l l be your job to respond to these subt le d i f f e r ences . The r e ' l l be 16 s l i d e s shown, and each w i l l appear on the screen fo r 3 minutes. I ' l l n o t i f y you 10 seconds ahead o f time before a s l i d e 1s changed. I t 1s abso lute ly e s sen t i a l tha t you keep look ing at the s l i d e . As a s l i d e i s f i r s t shown, study I t b r i e f l y . Then s t a r t f i l l i n g out the ques t l ona l r e . But keep r e f e r r i n g to the s l i d e wh i le w r i t i n g your answers. Don't work from memory. The sca les used are of the word p a i r type which you are probably already f a m i l i a r w i t h . One sca le might look l i k e t h i s : Good : : : : : : - — : : Bad. To record your answer, s imply put a check mark somewhere along the l i n e to Ind icate your choice: the s t ronger your f ee l i ngs co inc ide w i th one o f the words, the c l o s e r to that word should your check mark be. Each and every l i n e must have one (and only one) check mark. Some o f the word pa i rs may seem inappropr iate to you . However, y o u ' l l probably f e e l more one way than the o the r , so please do your best . Before going on, please complete t h i s l i n e : AGE' SEX (M/F) SEAT NO. 355. Appendix C Main study tes t booklet cover. 3 5 6 . This is an experiment dealing with your perceptions. It embodies some very subtle differences in a series of cards which you will be given shortly. It'll be your job to respond to these subtle differences. You'll be shown 16 cards in all, one at a time. The time allowed for each card is 3 minutes. At the end of this time pass the card on to the person next to you. It is absolutely essential that you look at the card as much as possible. As you first get the card, study it briefly. Then start filling out the questionnaire. But keep referring to the card while writing your answers. Don't work from memory. The scales used are of the word pair type which you are probably already familiar with. One scale might look like this: To record your answer, simply put a check mark somewhere along the line to indicate your choice: the stronger your feelings coincide with one of the words, the closer to that word should your check mark be. Some of the word pairs may seem inappropriate to you. However, you'll probably feel more one way than the other, so please do your best. Remember to put the number of the card you're looking at on each page. Before going on, complete this line: GOOD BAD EACH AND EVERY LINE MUST HAVE ONE (AND ONLY ONE) CHECK MARK. AGE _ SEX (M/F) 357. Appendix D Summary analysis of variance tables for motif recognit ion versus the four emotional response measures. 358. TABLE DI Summary of analysis of variance for pleasure. Source SS df MS F P Recognition (R) 4.1122 1 4.1122 1.2024 .2731 Hue (H) 1.1298 4 0.2824 0.0826 .9878 Value (V) 19.875 1 19.875 5.8115 .0161* Chroma (C) 17.088 1 17.088 4.9966 .0256* Motif (M) 66.827 3 22.276 6.5136 .0002* Sex (S) 32.406 1 32.406 9.4757 .0021* RH 23.692 4 5.9231 1.7319 .1235 RV 0.1285 1 0.1285 0.0376 .8464 RC 0.1670 1 0.1670 0.0488 .8251 RM 29.681 3 9.8938 2.8930 .0344* RS 3.4931 1 3.4931 1.0214 .3125 Error 3136.0 917 3.4199 Total 3334.6 938 * Indicates s ign i f i cance . 359. TABLE D2 Summary of analysis of variance fo r arousal . Source SS df MS F P Recognition (R) 0.9118 1 0.9118 0.3373 .5615 Hue (H) 14.883 4 3.7208 1.3766 .2401 Value (V) 3.7644 1 3.7644 1.3927 .2383 Chroma (C) 1.1192 1 1.1192 0.4141 .5201 Motif (M) 82.466 3 27.489 10.170 .0000* Sex (S) 1.2325 1 1.2325 0.4560 .4997 RH 8.5930 4 2.1483 0.7948 .4938 RV 0.4021 1 0.4021 0.1488 .6998 RC 2.5538 1 2.5538 0.9448 .3313 RM 9.7329 3 3.2443 1.2003 .3086 RS 0.2413 1 0.2413 0.0893 .7652 Error 2478.6 917 2.7030 Total 2604.5 938 * Indicates s ign i f i cance 360. TABLE D3 Summary analysis of variance for dominance. Source SS df MS F P Recognition (R) 9.4668 1 9.4668 5.7886 .0163* Hue (H) 8.1617 4 2.0404 1.2476 .2891 Value (V) 6.4176 1 6.4176 3.9242 .0488* Chroma (C) 0.1385 1 0.1385 0.0847 .7711 Motif (M) 1.4226 3 0.4742 0.2900 .8327 Sex (S) 18.182 1 18.182 11.117 .0009* RH 18.956 4 4.7390 2.8977 .0672 RV 0.1438 1 0.1438 0.0879 .7669 RC 1.0966 1 1.0966 0.6705 .4131 RM 2.8658 3 0.9553 0.5841 .6998 RS 0.0483 1 0.0483 0.0295 .8636 Error 1499.7 917 1.6354 Total 1566.6 938 * Indicates s ign i f i cance . 361. TABLE D4 Summary of analysis of variance for Information rate . Source SS df MS F P Recognition (R) 13.062 1 13.062 11.161 .0009* Hue (H) 3.1052 4 0.7763 0.6633 .6176 Value (V) 6.6676 1 6.6676 5.6973 .0182* Chroma (C) 4.6006 1 4.6006 3.9311 .0481* Motif (M) 24.078 3 8.0259 6.8580 .0001* Sex (S) 0.6934 1 0.6934 0.5925 .4416 RH 2.4704 4 0.6176 0.5277 .7154 RV 0.4770 1 0.4770 0.4076 .5234 RC 1.7546 1 1.7546 1.4992 .2211 RM 17.419 3 5.8063 4S9614 .0118* RS 2.2719 1 2.2719 1.9413 .1639 Error 1073.2 917 1.1703 Total 1149.8 938 * Indicates s ign i f i cance . 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0076993/manifest

Comment

Related Items