BRIGHTNESS AND BRIGHTNESS RATIO: AS FACTORS IN ATTENTION VALUE. Thesis for the Degree of Ph. D. ‘ MICHIGAN STATE UNIVERSITY RICHARD FAY PAIN 1968 mega insist} :0}! 3:“. m g.) I T" . v In. -_. Lie-31‘5“ 51 as??? This is to certify that the thesis entitled Brightness and Brightness Ratio as Factors in Attention Value presented by Richard Fay Pain has been accepted towards fulfillment of the requirements for Ph.D. degree in Psychology Major professor Datefl/fl/t/ /Z ///6 57/ 0-169 ABSTRACT BRIGHTNESS AND BRIGHTNESS RATIO AS FACTORS IN ATTENTION VALUE by Richard Fay Pain The attention gaining characteristics of highway signs have been studied recently and mathematical models were de- veloped. These models were based on either brightness ratio or percent contrast and therefore implicitly hypothesized that contrast (the object-background brightness relationship) determined attention value. In the literature on animal and human learning and attention it was found that stimulus in- tensity had frequently been varied while background intensity remained invariant. This meant that stimulus brightness and contrast had always been confounded. ‘ Contrast had been studied at threshold levels and in terms of brightness matching. The effect of stimulus bright- ness and its possible interaction with contrast on attention value at suprathreshold levels had not been exhaustively studied. Four hypotheses were developed. 1. The highest available brightness ratio would be reported as being seen "best and quickest" when stimulus bright- ness was held constant (B = K condition). tive were Richard Fay Pain The highest available brightness would be seen "best and quickest" when brightness ratio was held constant (R = K condition). If brightness and brightness ratio were varied together brightness ratio would be the predominant determiner of attention value (B and R # K condition). The higher available brightnesses would enhance the ef- fect of the higher available brightness ratios. In other words brightness ratio would increase in attention value in an accelerating curve as stimulus brightness increased. Eye movement measures (stimulus fixated first, time on each stimulus, and number of fixations per stimulus) would give essentially the same results as subjects subjective judgement of which stimulus was seen "best and quickest." The hypotheses were tested for both positive and nega- contrast directions. Using a polymetric-polaroid eye movement camera, pictures taken and subjective reactions recorded for stimuli in a paired comparison design. Ten college student S's saw the B = K and R = K conditions in counterbalanced order for the positive contrast phase. Ten different S's saw the B and R # K condition. The procedure was repeated using twenty Richard Fay Pain additional S's for the negative contrast phase. Stimuli were generated using various combinations of Munsell neutral gray matte chips and backgrounds. Three stimulus bright- nesses and three brightness ratio levels were used in each phase. The positive contrast and negative contrast phase background brightnesses were in the mesopic (night driving) and the low photopic (twilight) ranges respectively. An adapting light of .24 ft. L. was used in both phases. All data were reduced to proportions and transformed into paired comparison scale values. Hypothesis one and two were unreservedly confirmed. With regard to the third hypothesis, brightness ratio pre- dominated very strongly in the positive contrast phase but was not as conclusively predominant in the negative con- trast phase. Support for the fourth hypothesis came'mainly from the negative contrast phase where the highest bright- ness ratio increased in scale value in an accelerating curve as stimulus brightness increased. The positive con- trast phase gave little reliable evidence for the hypothe- sis. Eye movement measures gave results substantially dif- ferent from the subjective reaction measure for the B = K and R = K conditions. There was a high correlation between all the dependent variables for the B and R # K condi- tion. This result was discussed in terms of the effect of Richard Fay Pain a uni as compared to a multidimensional stimulus on attention value. The highway traffic sign models from the previous study were successfully applied to selected stimuli of the present study and the brightness ratio model gave the closest fit. Addition of a factor considering high relative brightness in combination with high relative brightness ratio would refine the‘model. 6"“. Approvedzjzgzééghéigp /QZ%Z/Z?4Z§57 Chairman Date Thesis Committee T. W. Forbes, Ph.D. Chairman T. M. Allen, Ph.D. Paul Bakan, Ph.D. BRIGHTNESS AND BRIGHTNESS RATIO AS FACTORS IN ATTENTION VALUE By Richard Fay Pain A THESIS Submitted to IMichigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Psychology 1968 f 7 / vr/J ./..2 ACKNOWLEDGMENTS I would like to extend thanks to Dr. Terrance Allen and Dr. Paul Bakan for their time and help while serving on both guidance and thesis committees. Dr. M; R. Denny gave considerable financial support for which I am most grateful. The help of my wife in motivating me, reducing data, and typing was extensive. Thank you Linda. MW'deepest appreciation and thanks to Dr. T. W. Forbes for several years of guidance, encouragement, patience, and consideration. ii TABLE Section Introduction . . . . . . . Problem . . . . . . . . . . Method . . . . . . . . . . Experimental Design . Stimuli . . . . . . . Subjects . . . . . . . Apparatus . . . . . . Procedure . . . . . . Scoring and Analysis . Results . . . . . . . . . . Discussion . . . . . . . . OF CONTENTS Considerations Pertinent to the Hypotheses Relation of Fourth Hypothesis Results to Blackwell's Data . . . Implications of the Eye Movement Results Application of the Results . . Conclusions . . . . . . . . References . . . . . . . ,iii Page 14 16 16 18 19 20 21 22 25 38 38 44 47 48 52 54 Table 10 11 12 LIST OF TABLES Experimental Design . . . . . . . . . . Summary of Stimulus Conditions for C-Positive and C-Negative Phases . . . . . . Chip and Card Brightness and the Resultant Brightness Ratios . . . . . . . . Age and Characteristics of C-Positive Phase Subjects . . . . . . . . . . . . . . . Age and Characteristics of C-Negative Phase Subjects . . . . . . . . . . . . . . . B = K.Condition Paired Comparison Scale Values for Each Dependent Variable; C-Positive Phase . . . . . . . . . . . . . B = K.Condition Paired Comparison Scale Values for Each Dependent Variable; C-Negative Phase . . . . . . . . . . . . . . R = K.Condition Paired Comparison Scale Values for Each Dependent Variable; C-Positive Phase . . . . . . . . . . . R = K.Condition Paired Comparison Scale Values for Each Dependent Variable; C-Negative Phase . . . . . . . . . B and R i K Condition Paired Comparison Scale Values for Each Dependent Variable; C-Positive Phase . . . . . . . . . . . . . . B and R i K Condition Paired Comparison Scale Values for Each Dependent Variable; C-Negative Phase . . . . . . . . . . . . . . Rank Order Correlations Between Dependent Variables; C-Positive Phase . . . . . . . iv Page 61 62 63 64 64 65 66 67 68 69 69 70 Table 13 14 15 Page Rank Order Correlations Between Dependent Variables; C-Negative Phase . . . . . . . . . 71 Individual Response Patterns for the B - K and R = K Conditions; C-Positive Phase, Subjective Reaction Measure . . . . . . 72 Individual Response Patterns for the B and R # K.Condition; C-Positive Phase, Subjective Reaction Measure . . . . . . 73 LIST OF FIGURES Figure Page 1 Hypothesized Relation Between Brightness and Brightness Ratio (Hypothesis 4) . . . . . 74 2 Layout of Stimulus Configurations . . . . . . 75 3 Outline Diagram, Polymetric Eye Movement Recorder . . . . . . . . . . . . . . . . . . 76 4 B = K.Condition Paired Comparison Scale Values Plotted for Each Dependent Variable; C-Positive Phase . . . . . . . . . . 77 5 B = K.Condition Paired Comparison Scale Values Plotted for Each Dependent Variable; C-Negative Phase . . . . . . . . . . 78 6 R = K.Condition Paired Comparison Scale Values Plotted for Each Dependent Variable; C-Positive Phase . . . . . . . . . . 79 7 R = K.Condition Paired Comparison Scale Values Plotted for Each Dependent Variable; C-Negative Phase . . . . . . . . . . 8O 8 B and R # K Condition Paired Comparison Scale Values Plotted for Each Dependent Variable; C-Positive Phase . . . . . . . . . . 81 9 B and R # K.Condition Paired Comparison Scale Values Plotted for Each Dependent Variable; C-Positive Phase . . . . . . . . . . 82 10 The Subjective Reaction Measure Paired Comparison Results for the C-Positive Phase, B and R # K Condition Graphically Summarized . . . . .~. . . . . . . . . . . . . 83 11 B and R # K Condition Paired Comparison Scale Values Plotted for Each Dependent Variable; C-Negative Phase . . . . . . . . . . 84 vi Figure 12 13 14 15 16 17 18 19 20 vii B and R i K Condition Paired Comparison Scale Values Plotted for Each Dependent Variable; C-Negative Phase . . . . . . . . . Subjective Reaction Measure Paired Comparison Results for the C-Negative Phase, B and R # K.Condition, Graphically Summarized . . . . . . . . . . . Stimulus Fixated First Measure Paired Comparison Results for the C-Positive Phase, B and R # K.Condition, Graphically Summarized . . . . . . . . . . . Stimulus Fixated First Measure Paired Comparison Results for the C-Negative Phase, B and R # K Condition, Graphically Summarized . . . . . . . . . . . Time on Each Stimulus Measure Paired Comparison Results for the C-Positive Phase, B and R i K Condition, Graphically Summarized . . . . . . . . . . . Time on Each Stimulus Measure Paired Comparison Results for the C-Negative Phase, B and R # K Condition, Graphically Summarized . . . . . . . . . . . Number of Fixations Mbasure Paired Comparison Results for the C-Positive Phase, B and R f K Condition, Graphically Summarized . . . . . . . . Number of Fixations Measure Paired Comparison Results for the C-Negative Phase, B and R # K Condition, Graphically Summarized . . . . . . . . . Comparison of Results Observed Experi- mentally and Theoretical Results Calculated from Brightness Ratio and Contrast Models; C-Positive Phase . . . . . . . . . . . . . . Page 85 86 87 88 89 9O 91 92 93 Figure 21 viii Comparison of Results Observed Experi- mentally and Theoretical Results Calculated from.Brightness Ratio and Contrast Models; C-Negative Phase . . . . . . . . . . . . . Page . . 94 Appendix I II III LIST OF APPENDICES Page Instructions to Subjects . . . . . . . . . . . 58 Response Reversal Critique . . . . . . . . . . 59 Comparison of Different Methods of Calculating Contrast . . . . . . . . . . . . . 6O ix INTRODUCTION From a systems point of view the highway sign serves as a communication channel. However, signs are always located in the midst of a tremendous amount of irrelevant (to the driver) as well as relevant visual stimulation. For the com- munication between driver and sign to be effective the driver must first be aware that a sign is present and second, re- cover the information from the sign. These two factors sug- gest three major problems in the design of effective signs. 1. How to design a sign that will be seen or visually acquired by drivers - attention or target value of the sign. 2. How to design a sign that after being acquired, most effectively communicates information - legibility or readability of the sign. 3. What information to place on the sign -usefulness or meaningfulness of the message. The remainder of this paper will be limited to the first problem area, attention or target value. Attention value as a general concept might generally be thought of as how much attention a stimulus will receive rela- tive to other stimuli in the same setting. There are many broad statements like this that could be made about attention value. However, they are of little use. 2 Operationally, attention value as applied to highway signs was first described in a study conducted by Forbes (1939). Drivers were given directions in terms of route numbers and destinations and instructed to drive as if in a moderate hurry calling out all route, destination, and warning signs as soon as they read them. Trained observers rode with each driver and kept him.in conversation. They also recorded his action and responses. The results indicated two types of attention value: 1. Target value - the characteristics which make a sign stand out in competition with other signs and distrac- tions. 2. Priority value - qualities of a sign which result in one sign being read first, given equal target value. In general, little or no other research has been di- rected specifically toward attention value of highway signs. Much work has been done in relation to advertising displays but this involves relatively small displays attempting to do more than simple communication. Legibility has also been extensively studied but involves readability, not attention value of a sign. Detectability and visibility have also been studied for the highway situation but are oriented more to threshold considerations, e.g., how much light is neces- sary to detect an object X feet away from a driver (Blackwell, 3 Prichard, and Schwab, 1960) or which color is most readily detected under various lighting conditions (Grain and Siegel, 1962) (Abstracted in Forbes, Snyder, and Pain, 1964 and Forbes, Snyder, and Pain, 1965.) While not actually being concerned with attention value, visibility and detectability are closely related in that stimulus factors resulting in the lowest thresholds (detected with least light or at greatest distance) may very likely be the same factors which will have high attention value (will be perceived first) under non-threshold conditions. There- fore it would seem that detectability studies as well as ad- vertising attention studies would be a source of ideas or clues to the stimulus characteristics which most probably would be operative in the attention value of highway signs. Burtt (1957) summarized the attention getting stimulus variables studied in relation to advertising. Those mentioned were humor, pictures vs. words, color, novelty, position, iso- lation, contrast (color, size, brightness, temporal), motion, intensity, and size. Earlier research was reviewed by Lucas and Britt (1950) and they found basically the same character- istics to be favorable for gaining attention. Not all of these factors would be applicable in the highway situation, notably humor, novelty, and picture vs. words. 4 Detectability studies have primarily been concerned with color, object-background contrast, the effect of tinted glass, and sign brightness. (From papers abstracted in Forbes, Snyder, and Pain, 1964.) The term "contrast" has been mentioned and will be re- ferred to extensively throughout this study. Contrast in general means the difference in some dimension, i.e., color, brightness, between an object and its background. The method of calculating this difference varied from one author to an- other. It should be noted that contrast ratio refers to the ratio of the change or difference between an object and back- ground (AI/I) while brightness ratio represents the simple ratio of the object and background (Io/1b). For clarity therefore, the following definitions will be adhered to henceforth. B1 Brightness ratio = B; where: B1 = object or background brightness, whichever is brightest. B2 = object or background brightness, whichever is darker. Contrast ratio = B1 - B2 100 (Percent contrast) B1 x where: B1 and B2 were defined as above. Blackwell contrast g Bo - Bb (Positive contrast Bb direction) where: B0 = object if brighter than background. Bb = background. B - B Blackwell contrast = b 0 (Negative contrast Bo direction) where: B0 = object if darker than background. Bb = background. The most recent work on attention value of highway signs studied several of these variables in relation to sign target value (Forbes, Pain, Joyce and Fry, 1968). In part this was done in a series of laboratory studies using a simulation technique. The general procedure for these studies was to present tachistoscopically four simulated highway signs simul- taneously for 1 second against a continually present highway 6 scene. The subject was to indicate which of the presented signs was seen "best and quickest" by pushing the appropriate response button. To simulate the constant activity (visual and motor) involved in driving, the subject performed a sec- ondary task. In a matrix of twelve red lights 1, 2, 3, or 4 lights might go off. ‘8 had to relight the off lights by pushing the button corresponding to the number of unlit lights. Highway sign presentations were randomly inter- spersed among secondary task trials. In one experiment, four signs were used. Each was a different brightness of green and the signs were mounted two over the highway and one on each side of the road. The re- sults showed that the overhead signs were picked first. Also the brightest sign was picked first most frequently against a night background while the darkest sign was picked first most when a daytime background was used. The same finding was replicated at a later data. (Forbes, Pain, Fry, and Joyce, 1967). In later studies it was found that the bright- ness of the letters on the sign and the sign size modified the results of the first study. (Forbes, Fry, Joyce, and Pain, 1967). Several mathematical models based on logical assumptions were investigated for mathematically describing the results from the above studies. The model which gave the closest 7 approximation to the experimental data of four experiments was based on the sign to background brightness ratio with letter to sign brightness ratio added where appropriate. The summed ratios were then converted to percentages. Sign1 Sign2 Signn where: ”ml on H Nbdl w l—l Ud'bd no h‘ L + l 2 L x1 '- a x then: L2 1 le...n £1- x *2 L2 2 le...n 2;: x. x. L2 2x1. .n 2x1...n theoretical > = percent seen first = sign or background brightness whichever is brighter. = sign or background brightness whichever is darker. = letter or sign brightness whichever is brighter. letter or sign brightness whichever is darker. This model is obviously based on brightness ratios and the assumption that brightness ratio was the factor controlling targetnvalue. In the experiments where signs were blank, the brightness ratio model fitted well and therefore appeared to 8 account for the target value of each sign when more than one sign was present. A model based on the more commonly used percent contrast gave the next best fit. Since the bright- ness ratio model has the closest approximation to the actual data it was hypothesized that brightness ratios, not absolute intensity, controlled the attention value of the simulated signs. The model was expanded to include size but that is not of concern in this paper (Forbes, Pain, Joyce and Fry, 1968). Other reasons for considering the model as a hypothesis arose from practical experimental limitations such as: First, it was not possible to delve into the study of any one factor of target value extensively. Therefore only four brightnesses were used, these were selected by inspection, and they proved to be unequally distributed along the brightness continuum. Second, the different brightnesses were in a green hue and subject reports, certain data, and the known difficulty of independently varying brightness and saturation suggested that there may have been a just perceptible difference in saturation on one of the green simulated signs. Third, and most importantly, the simulated signs were always presented against a fairly homogeneous background scene. This meant that whenever the four signs were of different brightnesses, the sign-background brightness ratios were also different. 9 In short, stimulus brightness and sign to background bright- ness.ratio were always confounded. The confounding of these two variables does not appear to be unique to the above experiments. In the literature on visual stimulus intensity, whether concerned with animal or human learning or attention, the intensity of the stimulus was varied but background intensity was invariant. The stimulus-background ratio was therefore not held constant and it was impossible to tell which factor was responsible for the obtained results (for example, Razran, 1939). According to Hull (1949) stimulus intensity affected response strength. Experimental evidence found this to be the case. When stimulus-background contrast was proposed as an alternative hypothesis it was found to explain the data equally well (Bragiel and Perkins, 1954). When spe- cifically tested, the contrast interpretation of stimulus intensity dynamism tended to be supported (Perkins, 1953; Johnsgard, 1957). Gray (1965) has since criticized the con- trast interpretation with regard to other learning phenome- non. However this does not negate the evidence that con- trast may be equally as strong a behavioral determinant as stimulus intensity. 'More directly concerned with stimulus intensity and at- tention value, Berlyne (1950) studied varying intensity by 10 ‘ changing aperature size in front of light bulbs or varying bulb brightness. While the brighter light was picked for both conditions the results cannot be interpreted as a pure intensity effect because the background was invariant and the ratio of bulb-background brightness varied. Histori- cally similar results have been found but have always been confounded with brightness ratio. For example, Breese (1899) used a stereoscope and found that the brighter of two stimuli would dominate in a binocular rivalry situation. When Dallenbach (1923) simultaneously exposed two spots equidis- tant from a fixation point and asked which stood out with the most "clearness" the brighter was chosen significantly more than the dimmer. Both studies used invariant backgrounds. In the area of perception much study has been done using brightness contrast (not to be confused with simultaneous contrast phenomenon). iMost of the experimental work deals with either thresholds or brightness matching and does not bear directly on intensity or contrast as attention vari- ables. For instance, Blackwell (1946) reports on 450,000 observations. The dependent variable was whether or not the observer had seen a spot of light flashed on a screen. Vari- ables studied included stimulus size, brightness, and bright- ness of the background. One finding was that as the back- ground decreases in brightness the object-background contrast 11 (Blackwell formula) must increase to remain just perceptible. This indicates that brightness and contrast do not have ex- actly the same perceptual result. A study which held contrast ratio constant but varied over-all level of luminance was done by Jameson and Hurvich (1961). Five squares were made with five different fixed neutral density filters. All five and a separate matching field were viewed simultaneously and for each setting the observer had to match the brightness of each with the match- ing field. Three different levels of luminance were used but the contrast relations between the squares did not vary, the appearance of the contrast did vary with the different luminance levels. From these studies it can be concluded that brightness ratio and stimulus brightness have an interactive perceptual effect at threshold levels and that a constant contrast ratio does not have a constant perceptual effect at suprathreshold levels. Since these perceptual effects appear to be inter- active it would seem reasonable to generalize the idea of interactive effect between contrast and stimulus brightness to their functioning as determiners of attention. The work of Forbes et a1. mentioned earlier would lead to the idea that brightness ratio will be the predominant determiner of attention. The perceptual work of Blackwell and Jameson and 12 Hurvich lead to the idea that brightness will in some way influence the effect of brightness ratio. In the research done by Forbes and his group the mea- sure of attention value used was the subjective responses of the subject, e.g., which sign the subject reported seeing "best and quickest." In other studies on attention the de- pendent variable usually has been some overt verbal or motor response of the subject (Berlyne, 1960). Where such re- sponses were used it was possible for the subject to make position responses, i.e., continually select only one part of a display or use only one of several available response indicators. This would, of course, not be an adequate mea- sure of attention value. When subjects have been instructed to make a response in terms of some general dimension, i.e., fastest, most clear, best and quickest, brightest, etc., it is not known how much judgement was involved in the response. The experimenter hopes and tries to design conditions such that subjects will make an immediate response along the desired dimension with- out including considerations of which stimulus should "logi- cally" be attended to, which stimulus the experimenter has designated as having more of the attribute, which stimulus the subject "likes" best, or in general attending to some di- mension not mentioned in the instructions. 13 One way of circumventing the above problems would be to monitor the orientation of the sensory mechanisms involved and determine how they relate to the stimuli. In the case of of vision, eye movements can be recorded. By using this technique several measures which may reflect attention value can be made (stimulus fixated first, time spent fixating each stimulus, number of fixations on each stimulus). These mea- sures are not without difficulties. The physical orientation of the cornea does not always indicate the perceptual result. This was demonstrated by Mackworth, Kaplan, and Metley (1964) in a study of vigilance. They reported many signals were "looked at" by the eye but were never reported or acted on by the subject. Since the stimulus reported by the subject would be the one which controlled overt responding this must be considered the primary response measure of attention value. Because the eye must be in a position to receive stimulation from the stimulus it would seem logical that eye movement would reflect which stimulus draws the eye to it (is attended) and would in some way be correlated with subjective responses. PROBLEM The present study attempts to evaluate the effect on attention value of brightness when brightness ratio is held constant, of brightness ratio when brightness is held con- stant, and of stimulus brightness and brightness ratio when varied simultaneously. Specifically it is hypothesized that: (1) if stimulus brightness is held constant (B = K condition), the highest brightness ratio will be reported as being seen "best and quickest;" (2) if brightness ratio is held constant (R = K condition), the highest brightness stimulus will be reported as being seen "best and quickest;" (3) if brightness ratio and stimulus brightness are varied simultaneously (B and R i K condition), brightness ratio will determine which stimulus is reported seen "best and quickest." The study also attempts to determine if an interaction between brightness and contrast such as found by Blackwell (1946) at threshold levels, will also be found at supra- threshold levels with regard to attention value. The hypoth- esis is that: (4) higher relative brightness will enhance the attention value of any brightness ratio. In other words there will be an interaction between brightness and bright- ness ratio. This hypothesis is illustrated in Figure l. 14 15 With regard to the dependent variables, eye movement measures and subjective reactions, this study attempts to evaluate the relation between a primary and several alterna- tive types of attention value measures. It is hypothesized that: (5) eye movement measures will be highly correlated with and give essentially the same results as subjective re- sponses which are considered the primary measure of attention value. The hypotheses were tested for both positive (stimulus lighter than background) and negative (stimulus darker than background) contrast directions. METHOD Basically the hypotheses were tested by showing S's Munsell chips against Munsell cards in a paired comparison design. Various combinations of stimulus and background brightness were used to achieve the conditions of stimulus brightness (B = K) or contrast ratio (R = K) constant and both varied together (B and R i K). Eye movements and subject responses were recorded. The study was done in two phases. The first used stimuli-background brightnesses resulting in a positive contrast direction (C-pos) and the second tested negative contrast direction (C-neg). Each phase used different subjects but the same method. Experimental Design: A paired comparisons design was used. The three condi- tions were generated from a matrix such as the one in Table 1. In the first condition, brightness held constant (B = K), pairs of stimuli-background combinations in which only the brightness ratio varied were used. In Table 1 these are rep- resented by the containing x's. The next condition, bright- ness ratio held constant (R = K), utilized pairs of stimulus- background combinations in which only the stimulus brightness varied. In Table 1 these pairs are represented by the cells containing zeros. Both conditions were tested over three 16 17 levels of the variable being held constant, e.g., B1 was tested versus B2 at the Al, the A2, and the A3 level. The B = K and R = K regimes were considered control con- ditions since they would show the attention value of one variable with the other variable held constant. The third condition was considered the experimental con- dition since stimulus brightness and brightness ratio varied together (B and R # K). This is represented by the blank cells in Table 1. In no pair were either the two stimulus brightnesses or the two brightness ratios the same. The pairs used in each condition are listed in Table 2. Conditions B = K. and R = K were presented in counter- balanced order to the same ten subjects. 8'3 1, 3, 5, 7, and 9 saw the B = K condition first and R = K condition second. The other five S's saw the conditions in the reverse order. Ten different S's saw the B and R i K condition. The order of presenting the pairs of stimuli for each condition was randomized for each S. The left-right posi- tion of the two stimuli in each pair was also randomized for each S. In brief, each S saw 18 pairs of stimuli with the order of the pairs and the stimulus position within the pairs randomized. 18 Stimuli: The stimulus configuration consisted of two 5/8 x 7/8 inch (136' of arc subtended at the eye) neutral gray matte finish Munsell chips seen against two 3" x 5" Munsell back- ground cards. Figure 2 gives a diagram of the stimulus con- figuration. This configuration included a flat black metal edge (1" in width) which held the cards in place. Table 3a shows the brightness ratios resulting from the combining of the three chip brightnesses and various back- ground (card) brightnesses for the positive contrast phase of the study. The card brightnesses were in the middle to high mesopic range (.5-l.0 ft. L.). Due to photographic limitations and color temperature of the light source con- siderations, lower light levels were not feasible. The card brightnesses were well within the luminance range (.1-1 ft. L.) defined as night driving (Richards, 1966). Table 3b gives the values for the negative contrast direction. These ranged from .52 to 3.6 ft. L. and were more representative of twilight visual driving conditions. To obtain the B = K condition, the same brightness chips were seen against various card brightnesses. Three levels of chip brightness were tested. Each level was seen with all combinations of three brightness ratios. By re- ferring to Table 3a it is seen that this condition was 19 developed by using only the cards found in each column. The R = K condition was obtained by using different brightness chips seen against cards giving the same bright- ness ratio. Three levels of brightness ratio were each tested with all combinations of the three chip brightnesses. Table 3a shows this condition as the cards in a row along with the appropriate brightness. The B and R i K condition used the same chip-card combinations but each combination was seen paired with a chip-card combination from a different row or column. The three conditions are summarized in Tables 1 and 2. The chip and card brightness values for the negative contrast phase are given in Table 3b. The three conditions were developed in the manner described above. Subjects: College students from introductory psychology classes volunteered and were given extra credit for participating in the experimait. Ten S's were used for the control group and ten for the experimental group. This was repeated for the second contrast direction so there was a total N of 40. Tables 4 and 5 give the age and sex characteristics of the S's. The subject sample was restricted in two ways. First, S's who normally wore reading glasses could not participate. Second, due to equipment limitations, S's over 6'1" in 20 height could not participate. S's normally wearing contact lenses were allowed to participate but could not wear their lenses during experimentation due to halation of the marker light beam at the contact lens. Apparatus: A Polymetric Products eye movement recorder, Series V-ll64, was used to present the stimulus and record S's eye fixations. An overhead view outline diagram of the re- corder is given in Figure 3. The recorder operated by re- flecting a beam of light from the subject's left eye into an optical system. The light beam or marker light consisted of a bright spot with a dimmer tail extending out from the cen- ter. The marker light rotated enabling sequence to be de- termined from a still photograph. Maximum rotation speed, 7% RPM, ‘was used. The optical system.superimposed the marker light on the image of the stimuli being viewed and this was then recorded by a Polaroid camera. Type 47 (ASA 3000) Polaroid film was used and exposure time was 4 seconds i 1 second. The varia- tion in exposure time was found necessary to compensate for individual differences in the quality of pictures obtained among subjects. From preliminary trials it was found that better picture resolution was obtained with the stimuli 22 inches rather than 24 inches from the subject's eyes. 21 To keep S from seeing the stimuli to be presented on the next trial a flat black screen was devised to fit over the semi-transparent mirror (see Figure 3). With the screen in place S could not see the stimuli in the stage but he could see the black bottom part of the stage. This meant that the same light illuminating the stimuli was serving as a constant adaptation light for S. The brightness of the adapting field was .24 ft. L. This and all other bright- ness measurements were made with a Prichard Photometer, ‘Model 1970-PR. A black box with two push buttons was located on the table beneath the base beam of the eye movement recorder. The buttons activated a Lafayette Model 1041 event recorder. By pushing the appropriate (left or right) button 8 indi- cated which of the stimuli he saw "best and quickest." Procedure: Each S was tested individually for one hour i 15 min. Each test session began with an explanation of the eye move- ment recorder and its operation to S. While a bite plate was being prepared, S adjusted a stool to a comfortable height. The room lights were turned off and S was adapted to the experimental lighting for a minimum of five minutes. During this time the optical system was aligned and focused and S was given instructions. Complete instructions are 22 given in Appendix I. The stimuli were then presented in a paired comparison design. It was found that alignment and refocusing had to be adjusted several times during the 18 trials. S's could not be expected to continually bite the plate since they could not easily swallow when in that posi- tion. Therefore they sat back while the stimulus materials were being changed. If any pictures did not turn out prop- erly the trial was repeated after the first 18 trials. After all trials were finished S was questioned about his general state of rest and specifically how his eyes felt before participating in the experiment. 8 was also asked if he had any comments about the stimuli. If not, he was asked to indicate if possible some characteristic of the stimuli picked as being "best and quickest." Sample eye movement records were shown and explained to S. Scoring and Analysis: Four dependent variables were used in this experiment. (1) Subjective reaction was the subjects choice of which stimulus in a pair was seen "best and quickest." These were scored in the usual paired comparisons fashion (case V) de- scribed by Guilford (1954), e.g., raw scores were placed in matrix form, transformed into the percentage that each stimu- lus was chosen over every other stimulus, transformed into normal deviates, and scale estimates for each stimulus made. 23 (2) The chip fixated first for each trial was scored by tracing the rotation of the marker light tail to its-most counterclockwise position on any particular picture. Which- ever chip was under that fixation was credited with being fixated first. Paired comparison scale values were then de- termined in the same manner as noted above. (3) Duration, or the amount of time spent fixating on each chip, was supposedly measureable. However, due to mechanical difficulties, e.g., the marker light could not be coaxed to revolve at more than 7% RPM ‘which was approximately half the advertised speed, reliable and accurate time measures were impossible. Instead, a very crude three category judge- ment was made by the experimenter. The amount of time spent fixating each chip was judged to be more, equal, or less than the time spent fixating the other chip of each pair. Crite- rion for judgement was the area, lightness, and halation of the marker light connected with each chip, e.g., longer fixa- tions resulted in a larger whiter spot. These data were then transformed into percent "more" and "less" ratings with "equal" rating ignored. The percents were then scaled in paired com- parison fashion as outlined above. (4) The number of fixations were counted for each chip and entered in a paired comparison matrix. The raw data were transformed into percents by using the total number of 24 fixations made by all S's for each pair of chips as the basis for determining the percent fixations for each chip. The resultant percents were then scaled according to the paired comparison method mentioned above. RESULTS The results from the two phases of the study will be discussed together for each dependent variable. The depen- dent variables themselves will then be discussed. Subjective Reactions: 1. B = K The C-pos phase values for the B = K condition are presented in Table 6. From the scale values and from the graph of those values (Figure 4a) it is apparent that the highest brightness ratio was chosen as being seen "best and quickest" for any given pair of stimuli. The exception to this finding occurs at the 2.0 ft. L. level where the 7.1 ratio was chosen most often over both the 3.8 and 9.7 ratios. For the C-neg phase Table 7a and Figure 5a give the results of the B = K condition. The highest brightness ratio was chosen first most often while the middle ratio was second. This was found for all brightness levels but the order of the brightness levels for each ratio was not as expected. _ In both phases it was interesting to note that the chip brightness which was in the middle of the scale of bright- nesses used did not reach nearly as high a scale value at the highest ratio level as the other brightnesses. Generally 25 26 the results of the two phases were highly similar. 2. R = K Figure 6a and Table 8 show the results of the C-pos phase. The highest brightness chip was picked as seen "best and quickest" for most stimulus pairs and for each ratio level tested. Again an exception to this result was found for the 2.0 ft. L. chip at the 7.1 ratio level. The C-neg phase results are in Table 9 and Figure 7a. The highest followed by the next brightness were seen "best and quickest" most often. However, the ratio had an ex- tremely large effect on the magnitudes of the scale values. At the .35 and .66 ft. L. brightnesses the -5.4 ratio was substantially higher in scale value than the -2.4 or -4.2 ratios. For both phases it was generally found that the highest brightness was chosen as "best and quickest" at each ratio level. The range or spread of scale values between the ratio levels was much greater for the C-neg phase than for the C-pos phase. 30 B and R + K Table 10 gives the results for the C-neg phase. These results were plotted in four different ways. In Figure 8a scale values were plotted as a function of brightness ratio for each of the three chip values. From this it can be seen that the 7.1 and 9.7 brightness ratios were chosen as 27 "best and quickest" most frequently. At the 3.8 and 7.1 brightness ratios, chip brightness had virtually no effect. There does appear to be some effect due to brightness at the 9.7 ratio. However, the order of the relation (2.8, 1.45, 2.0) was not as expected. When the scale values were plotted as a function of chip brightness (Figure 9a) for each of the three ratios it was evident that the 9.7 and 7.1 ratios were chosen most frequently at all brightness levels. At the 3.8 and 9.7 ratio levels the 2.8 ft. L. chip had a slight enhancing effect on the degree to which brightness ratios were chosen. The effect of the two variables is summarized graphi- cally in Figure 10. The brightness ratios were relatively equidistant. This made the finding that the 7.1 and 9.7 brightness ratios had approximately the same attention value inexplicable. When percent contrast was used instead of brightness ratio, as in Figure 10, it was evident that the two higher contrasts (85.9 and 89.7%) were chosen "best and quickest" enough to make them relatively indistinguishable in terms of scale values. For the C-neg phase (Table 11, Figures 11a, 12a, and 13b) it was apparent that both intensity and ratio effected the subjective reactions. For any chip brightness the higher the ratio the higher the scale value and vice versa, for any ‘ ratio the higher the chip brightness the higher the scale value. 28 In summary then, the results showed that for the sub- jective reaction measure of attention value, scale values increased as brightness increased when brightness ratio was constant; that scale value increased as brightness ratio increased when brightness was constant; and that contrast was the predominant determinant of the subjective reaction measure of attention value when brightness and brightness ratio were varied simultaneously. This occurred for the positive but not the negative direction. Chip brightness affected subjective reactions more at higher brightness ratios than at lower ratios in the C-neg phase. Chip Fixated First: 1. B = K The scale value results for the C-pos phase are pre- sented in Table 5 and shown graphically in Figure 4b. With brightness held constant the stimulus fixated first, in rank order of scale value, were the 7.1 ratio at the 2.8 and 2.0 ft. L. levels, the 9.7 ratio at the 2.8 ft. L. level, and the 3.8 ratio at the 1.45 ft. L. level. In the C-neg~ phase (Table 7, Figure 5b) the middle brightness ratio was fixated first for the .21 and .38 brightnesses. The highest brightness ratio was fixated first most often at the .66 ft. L. level. 29 In both phases the brightness ratio in the middle of the scale used was fixated first most frequently. 2. R = K Scale values and their graphic presentation are given in Table 8 and Figure 6b for the C-pos phase. The results indicate that at each ratio level different chip brightnesses were generally fixated first. For the 3.8 and 7.1 ratio levels the lowest and highest brightness chips were fixated first but for the 9.7 ratio level the middle chip bright- ness was fixated most often. From Table 9 and Figure 7b Showing the C-neg phase it was seen that the highest and lowest brightnesses were fix- ated first at the -2.4 and -4.2 ratio levels while the .66 ft. L. brightness was fixated first at the -5.4 ratio level. In a very general way the fixation patterns of both phases were the reverse or opposite of the patterns found in the B = K condition. 3. B and R # K Table 10 gives the scale values for the C-pos phase. Figure 8b shows the scale values plotted as a function of brightness ratios. Except for the 2.8 ft. L. chip at the 9.7 brightness ratio, chip brightness showed no effect on the stimulus fixated first at the different ratio levels. It can be seen that the 7.1 and 9.7 ratios were fixated 30 first most often. The predominance of the 7.1 ratio was contrary to expectations. The same data plotted in reverse (Figure 9b) confirm the above result. The 7.1 and 9.7 ratios, in that order, were fixated first most frequently while chip brightness had very little effect with exception mentioned above. The data of this condition were summarized graphically in Figure 14. The C-neg phase results were the same as those de- scribed for the C-neg phase subjective reactions. The mag- nitudes of the scale values were not as great however, and the differences not quite as pronounced (Table 11, Figures 11b, 12b, and 15). For both phases when one dimension was held constant the patterns of chip fixated first were not similar to the sub- jective reaction patterns. Also the patterns of the R = K condition were virtually the reverse of the B = K condition. Where B and R # K the response patterns showed brightness ratio to be the predominant factor and very similar to the subjective reaction response patterns. Time on Each Chip: 1. B = K Table 6 and Figure 4c present the results for the C-pos phase. The middle (7.1) ratio was fixated for the greatest duration at the 1.45 and 2.0 ft. L. levels. The 9.7 31 ratio had the most time spent on it at the 2.8 ft. L. For the C-neg phase the -5.4 ratio was generally fixated the longest. However, the order of the chip bright- ness levels was the reverse of that expected (Table 7, Figure 5c) . The two phases generally gave different results. When the chip was lighter than the card the middle brightness ratio was fixated longest but in the reverse condition the highest brightness ratio was fixated longest. 2. R = K The data for the C-pos phase are available in Table 8 and Figure 6c. The highest and lowest chip brightnesses were fixated the longest at the lower brightness ratio levels. The response pattern at the 9.7 ratio level took a differ- ent form. In the C-neg phase the highest brightness was fixated longest at all ratio levels. The .21 ft. L. chip was fix- ated almost as long as the .66 ft. L. chip at the -2.4 ratio level (Table 9, Figure 7c). Comparing the two phases the results were similar in that the low and high brightnesses were fixated for the greatest durations. 32 3. B and R # K Table 10 gives the scale values for the C-pos phase. When scale values were plotted versus chip brightness (Fig- ure 8c) it was evident that brightness had little effect on the scale values. Instead the scale values varied as a function of brightness ratios (Figure 9c). Of importance, however, was the result that the middle ratio had the great- est fixation time. Figure 16 summarizes this condition graphically. The C-neg phase were again adequately described by subjective reaction results (Table 11, Figures 11c, 12c, and 17). The general form of the response patterns were different for the two phases. C-pos showed highest scale values for the middle brightness ratio while C-neg scale values gener- ally increased as brightness ratio increased. Number of Fixations: 1. B = K Results are given in Table 6 and plotted in Figure 4d for the C-neg phase. The 9.7 ratio received the greater proportion of fixations except at the 1.45 ft. L. level. The 7.1 and 3.8 ratios followed in that order for all chip brightness levels. 33 For the C-neg phase (Table 7, Figure 5d) there were very small differences between ratios. In general the .66 . ft. L. level was fixated most often with the -5.4 ratio being fixated a hair more often. Interestingly, C-pos results were similar in form to subjective reactions while C-neg phase results showed very small differences and were not very similar to the subjective reaction measure. 2. R = K The C-pos phase results are found in Table 8 and are plotted in Figure 6d. Chip brightness had little effect. However, there was a tendency for the lowest and highest brightnesses to have higher scale values at the 7.1 and 3.8 ratio levels. The opposite result occurs at the 9.7 ratio level. For the C-neg phase the .66 ft. L. chip was fixated most frequently at all ratio levels (Table 9, Figure 7d). 3. B and R # K For the C-pos phase chip brightness had a negligible effect and the brightness ratios controlled attention (Table 10, Figures 8d and 9d). The rank order of the ratios in terms of scale values was 7.1, 9.7, 3.8 which was not ex- pected. The results of this condition were summarized graph- ically in Figure 18. 34 In the C-neg phase the same basic results described for subjective reactions apply but the scale value magnitudes were very small and differences tiny when compared with those of the other measures (Table 11, Figures 11d, 12d, and 19). Even though the differences between brightness and bright- ness ratios were small the response patterns were very similar to subjective judgement responses. It was again observed that brightness was an effective response determinant in the C-neg phase but not so in the C-pos phase. Relation Between Measures: 1. C-pos Phase To determine how alike the results were among the four dependent variables, Spearman rank order correlations (Walker and Lew, 1953) were performed between all combinations of the measures for each testing condition (see Table 12). It was found that some relation (mean R = 41.5) existed between the four measures for the B = K.condition. By comparing the four graphs of Figure 4 much of the relationships could be accounted for by the fact that very low scale values were associated with the 3.8 ratio for three of the four measures. The relation between measures was generally higher for the R = K condition than for the B = K condition. A moderate (mean R = .56) relation existed between subjective reactions and to the other measures, however the other three measures 35 correlated 83, 83, and .92 among themselves. Comparison of the four graphs of Figure 6 make this finding quite apparent. For the third condition (B and R # K) the correla- tions range between .52 and .78. From comparison of the four graphs of Figure 9 it was noted that the existing corre- lation was due to the ordering of the ratios, e.g., the two higher ratios being highest in scale value, the 3.8 ratio being consistently lowest. The correlations were not any higher due to the inconsistent ordering of the chip bright- nesses at different ratios and for different measures. An examination of the patterns of scale values between correlations reveals a rather interesting pattern. Subjec- tive reaction plots took the same form for the B = K and R = K conditions. However, the results for the other three measures show almost exactly opposite response patterns be- tween the B = K and R = K conditions. Even though S's generally ordered both brightness and ratio from lowest to highest in terms of scale value, the eye movement character- istics for the two independent variables were generally dia- metrically opposed. Since this occurred when one or the other independent variable was constant, it would be very inter- esting to see how the two opposing eye movement patterns were resolved when both variables were varying. Comparison of 36 Figures 4, 6, 8, and 9 revealed that the three measures showed a pattern similar to the B = K condition while there was little or no similarity between the eye movement characteristics of the R = K and B and R # K conditions. 2. C-neg Phase Spearman rank order correlations were also calculated for the C-neg phase. From Table 13 it can be observed that the measures were moderate for the B = K and R = K condi- tion and were quite high for the B and R i K condition. The different eye movement response patterns noted in the C-pos phase were also present in this phase but were not as consistent. The chip fixated first results show ex- actly opposite patterns for the two constant conditions (compare Figures 5b and 7b). The time on each chip measure showed similar patterns for both measures while the number of fixations measure was generally insensitive and showed no reliable patterns. When B and R i K the eye movement characteristics did not take the pattern of either constant condition. Subject Differences: l. C-pos Phase Table 14 shows the response pattern for each S for the B = K and R = K conditions. It should be noted that each 37 S saw both conditions but in counterbalanced order. The table does not indicate preferences, only which chips were chosen and how often. For the B = K condition subjects 4, 5, and 10 showed response patterns that were virtually opposite the patterns of the other seven S's. In the R = K condition the same three S's response patterns were har- monious with the majority response pattern. No clear cut re- versals occurred in the B and R # K condition (see Table 15). These reversals in response pattern were considered inter- esting in a theoretical sense but they attenuated the scale values. Rather than disregard the data it was decided to switch the direction of the reversed data to agree with the majority of the S's. Some precedence and related rationale supporting a manipulation such as this were found and are given in Appendix II. 2. C-neg Phase There were no cases of reversed responding in any condi- tion. Several subjects were not totally consistent but their response patterns were certainly not reversed. In general, subject comments indicated awareness of brightness and con- trast differences for both phases. DISCUSSION Considerations Pertinent to the Hypotheses: 1. For the B = K condition, the results strongly confirmed the hypothesis (#1) that the highest brightness would have the highest scale values. This was true for both phases. It should be noted that in a paired comparisons design, such as this, only the lowest ratio was never the highest in any configuration. Therefore, the middle brightness ratio could be the lowest presented in some pairs and highest in other pairs. Apparently the position a chip had within the scale of brightnesses affected subjects responses. This was demon- strated by higher scale values for the lowest and highest brightnesses. This indicates the likelihood that the same principles found operative in human judgement on various di- mensions (e.g., Johnson, 1955) will be relevant to the scal- ing of various parameters in terms of attention value. 2. The second hypothesis, the highest brightness will be seen "best and quickest," was strongly confirmed for both phases. Again the middle value in the scale of ratios was responded to somewhat differently than the highest and lowest ratio. This further emphasized the relevancy of judgement and scaling principles for attention value phenomena. 38 39 3. Support for the third hypothesis, brightness ratio will be dominant over intensity in determining which chip is seen "best and quickest," was very strong in the C-pos phase of the study with less unequivocal support from the C-neg phase. In the C-pos phase, comparison of the control conditions with the experimental condition (B and R # K) showed that while scale values increased as brightness increased in the R = K condition, there was no increase in scale value as a function of brightness in the B and R i K condition. The only differences in scale values were between the brightness ratios. Therefore brightness ratio was the controlling fac- tor in determining attention value. The less clear-cut evidence shown by the C-neg phase was due to a brightness effect which was different for the various brightness ratios. The support found was the com- parison of the R = K and B and R # K conditions. When brightness alone varied large scale values and differences between ratio levels were observed. However when intensity was not the only dimension varying, the differences in scale values at each ratio level were much smaller. Further it was clear that the two higher brightness ratios accounted for the greatest proportion of the high scale values while brightness showed low scale values at the low ratio levels. In short a high ratio resulted in high scale values regardless 40 of chip brightness but stimulus brightness resulted in high scale values only in combination with higher brightness ratios indicating that brightness ratio was the controlling factor. 4. The above discussion is relevant not only to the third but also to the fourth hypothesis, i.e., higher brightness will enhance the attention value of any brightness ratio. Support for this hypothesis came primarily from the C-neg phase results. Lack of support would have been indicated by horizontal lines in Figure 12. While it was apparent that brightness ratio was the predominant variable, attention value, as measured by subjective reactions, did increase as chip brightness increased for a given brightness ratio. The effect was most pronounced for the two highest brightness ratios. C-pos phase support was very slight. Only at the 9.7 brightness ratio was there an evident increase in scale value related to a brightness increase. In general, the hypotheses related to brightness ratio and brightness were confirmed but with some qualification about the effect of brightness on ratio for the different contrast directions. In this study brightness ratio has been used to express the card-chip brightness relationship. The C-pos phase scale 41 values indicated that there was very little difference be- tween the 9.7 and 7.1 ratios. Yet these ratios were separated by 2.6 while the 7.1 and 3.8 ratios were only separated by 3.3 and were readily differentiated as reflected in scale values by the subjects. This finding was more meaningful when the card-chip relationship was expressed and plotted in percent contrast (see Appendix III for conversions). This was done in the three dimensional figures (see Result section) and it is obvious that the two upper ratios were very close (3.8 difference) when trans- formed to percent contrast. There was then a relatively large difference between the upper two and lowest contrast (12.3 difference). The implication was that percent con- trast was a more adequate description of the attention value of the chip-card relationship as perceived by human observers. The same finding was observed in the C-neg phase but it was not as pronounced. Three possible reasons for the difference in results between the B and R i K conditions for the two phases are: (A) the brightness ranges sampled in the two phases were not identical; (B) the relation between the three brightnesses used in each phase was not the same; and (C) there was a lsystematic relationship between the card and chip brightness when brightness ratio changed. 42 (A) In the C-pos phase, background and adapting brightness were in the mesopic range while in the C-neg phase, adapting and chip brightnesses were in the mesopic range but the card brightnesses ranged from high mesopic to photopic brightnesses (.52 to 3.6 ft. L.). Possibly the higher card brightnesses of the C-neg phase which were always associated with the highest chip intensity, seen after a mesopic adapting light were perceived as dispropor- tionately brighter than the lower brightness chip-card combination. (B) With regard to the second possibility the chip brightnesses used in the C-pos phase covered a limited range which was in the low photopic area. The relation of the three chip brightnesses was that they increased by steps of just under half the ft. L. value (S1 = X, S = % XT+ X, 2 S3 = % 82'+ SZ)‘ In the C-neg phase the chips were all in the mesopic range and covered only a small brightness range. But the relation between the brightnesses was that each was almost double the physical brightness of the next lower chip (s1 = x, s = 2X, S = 282). Chips with this relation in 2 3 combination with a lower range of contrasts resulted in the hypothesized effect while physically closer intensities associated with a higher range of contrasts did not show the effect. 43 Any conclusion about the fourth hypothesis must be made with the qualification that the range of brightness ratios and brightnesses was small; and the brightnesses relatively low while the percent contrasts were relatively high. This may have given the contrast dimension an edge in determining attention value. Much higher brightnesses in conjunction with the same percent contrast should be investigated before a firm conclusion can be stated. (C) Another possible source of difference is the fact that in both phases, as either brightness ratio or brightness of the chip increased there was a corresponding increase in the card brightness. The card backgrounds in the C-pos phase had a brightness range of .21 to .75 ft. L. This was neither a wide nor a very intense range of brightnesses. On the other hand the C-neg phase cards ranged from .52 to 3.6 ft. L. This was a considerable difference compared to the C-pos phase and it is possible S's reacted to one card compared with the other card without much weight being given the chip to card relationship. Three arguments mediate against such an interpretation. (a) Figures 9 and 12 give card brightnesses for each data point. If the card brightnesses were controlling the sub- jects' reactions, it would be expected that any given card brightness would have a higher scale value than all other 44 lower card brightnesses. This in fact was not necessarily the case. The 1.15, .92, 1.58, and 2.0 ft. L. cards each had higher scale values than a higher card brightness. The same result was found in the C-pos phase (refer to Figures 9 and 12). (b) S's were set to respond to the chip-card relationship by the instructions and demonstration given at the beginning of the run (Appendix II gives the instructions). (c) Comments from the majority of the subjects indicated that they were indeed following the instructions and re- sponding to the chip-card relationships. Only two of the forty S's said they were influenced by the brightness of the background. Relation of FourthdHypgghesiszesults to Blackwell's Data: The idea that brightness and brightness ratio would inter- act rather than relate linearly to scale values was suggested by the curvilinear relation between brightness and contrast found by Blackwell (1946) while studying contrast thresholds. It was considered appropriate, therefore, to plot the data of this study in the same way as Blackwell's (1946) data. An isometric or topological graph was constructed using log con- trast and log brightness. Scale values of roughly the same magnitude were connected in an attempt to develop a family of curves each describing a general level or rank of attention 45 value. Several problems arose with this type of comparison. First, there were only nine possible points on the graph, not enough to form a family of curves. Second, such a plot would have dubious meaning in that each of the points on Blackwell's curves represented a measurement with one brightness and con- trast, whereas the present study used pairs of stimuli which were always well above threshold. Any point on the curve was independent of the other points. The data from this study would be limited to the stimuli encompassed by the study because each chip would have a certain attention value in relation to the chip it had been compared with. Adding a chip of higher brightness or contrast would change the atten- tion value relations of all the chips and new, not necessarily comparable, families of curves would have to be formed. For the above reasons, comparison of the two studies was not feasible. 5. The fifth hypothesis, subjective reactions and eye move- lment measures will be highly related, was partially supported. The amount of correlation varied as a function of stimulus condition. So, for the B = K condition there were low to medium correlations for all measures in each phase. For the R = K condition subjective reactions and eye movement mea- sures had medium. (.50) correlations and the three eye move- ment measures were highly correlated (.80). In the B and 46 R i K condition correlations were medium to high for the C-pos phase and high for the C-neg phase. The above pattern of correlations may be an indication that eye movements are related to subjective reactions as a function of the number of parameters varying within the stimulus configuration. As more stimulus dimensions, e.g., brightness, contrast, vary concomitantly the more eye move- ments are predictive of subjective reactions. The rationale of the above suggestion is that: (a) In the B = K condition it was seen that the majority of eye movement first fixations and time on each chip were greatest for the middle ratio and the lower chip brightnesses. It has already been shown that two of the contrasts were very close in terms of both percent contrast and subjective reaction scale values. The middle ratio value was apparently utilized as a comparison stimulus; it was fixated and com- pared with either the higher or the lower ratio or bright- ness. Supporting this contention was the finding that the mid-scale peaking found in the C-pos phase was not as con- sistent or pronounced in the C-neg phase where the ratios were somewhat more separated. (b) In the R = K.condition, where the chips were more equi- distant physically, the eye movement measures were very simi- lar between conditions and correlated with each other quite 47 highly within each phase. They also showed more of a start and end spurt effect found in scaling work. (c) In the most complex situation, B and R # K, the eye movements were highly correlated with subjective reactions. In other words it seemed that with an increase in the number of parameters S's were not as able to identify a middle value for use as a comparison stimulus. Therefore responses were not as prone to be based on a scaling of stimulus param- eters. The result being that eye movement measures were more indicative of attention value as measured by subjective reactions. Implications of the Eye Movement Results: The above discussion leads to the important question of which dependent variable is the most appropriate measure of attention value. Obviously, none of the measures are atten- tion value: rather each is some kind of reflection of the internal process of selective responding. Subjective reac- tions were most consistent over conditions. From the above considerations it appears likely that eye movements were more subject to judgemental and scaling processes and were less direct reflectors of attention value. Although the opposite case could conceivably be argued, another finding made such an argument untenable. In examining the raw data of the number of fixations measure, it was noticed that three 48 subjects in both phases frequently had no fixation recorded for one of the stimuli. This implied that S's could make a quick glance of the entire stimulus field and have the information necessary to make a decision as to the chip seen "best and quickest" before actually fixating on one of the chips. If this was the case, the chip fixated first, the time spent looking at a chip, and the number of fixations would be irrelevant in terms of attention; attention value would have been determined before any of the eye movement characteristics were measured. In the highway sign situation there would be a problem in deciding which measure might be most important. For ex- ample, in a group of signs one might be thought to stand out the best but if it was not fixated first or longest this could mean that people were looking more at different signs. This study does not warrant the assumption that to have the most attention value will be looked at first, most often or longest, except where the brightnesses and contrasts involved are somewhat different among several signs. _Application of the Results: In applying the results of this study to the highway sign situation the applicability of the Forbes models will be considered. 49 The model developed by Forbes et al. (1968) was not in- tended to be applied to repeated paired comparisons measures. A rough idea of how the model works for this data can be ob- tained by selecting stimuli from each phase and calculating the expected proportions. Four chip-background card config- urations from each phase were chosen because they were ana- logous to the brightness and contrast relations between the simulated signs used for day and night backgrounds in the Forbes et a1. (1968) project. The proportions of times a stimulus was picked as "best and quickest" was the observed value and the theoretical percent picked was calculated using both the brightness ratio and percent contrast models ex- plained in the Introduction. In Figures 20 and 21 the pro- portion each of the selected chips was picked as being seen "best and quickest" was plotted versus the brightness and brightness ratio characteristics of the two chips being com- pared. The physical characteristics of each selected chip and card were then used in the two theoretical models and plotted in the same fashion. It is quite apparent that the brightness ratio model provides a closer fit to the observed data than the percent contrast model. This verifies the Forbes et a1. (1968) finding. In both phases it was observed that whenever high chip brightness was associated with high brightness ratio there 50 was a 20 percent point difference between the experimental and the theoretical results. However, when low chip bright- ness was associated with high brightness ratio the difference between experimental and theoretical results was only 10 percentage points. Since the theoretical model does not take into account chip brightness, the calculated values reflect only bright- ness ratio. The larger discrepency between observed and calculated values in the presence of high but not low bright- nesses reflects the interaction or enhancement effect of higher brightnesses on high brightness ratios. The above demonstration leads to an alternative ex- planation for the consistent finding of Forbes, Pain, Joyce, and Fry (1968) that calculated values were roughly 10 per- centage points below observed values. This was attributed to the repeated measures technique utilized however the present study makes it evident that the effect of high brightness in combination with contrast could also account for the discrepency. In general, then, there appears to be evidence that a brightness factor used in.a model for predicting attention gaining characteristics would add considerable refinement to the model. While this study does not permit a precise 51 empirical determination of the quantitative value of such a factor there were indications that a higher brightness en- hances a high brightness ratio by roughly 10 percentage points. CONCLUSIONS 1. Either brightness or brightness ratio when present as the only variable dimension have equivalent attention gaining value. For each dimension, attention value increased in a rapidly accelerating manner as the dimension increased. 2. When brightness and ratio are present together brightness ratio predominated. However, a high relative brightness en- hances the attention gaining effect of brightness ratios. This was more true for the negative contrast direction (back- ground brightness levels in the night driving-twilight range) than for the positive contrast direction (night driving back- ground brightness levels). 3. Generally, the subjective reaction measure was most con- sistent. The three eye movement measures were highly corre- lated with each other but not with the subjective reaction measure. As the number of dimensions varying concomitantly increased, the correlation between eye movement and subjective reaction measures became quite high. 4. The Forbes et a1. models for predicting traffic sign attention value were successfully applied to data from dif- ferent methodology and subjects. The brightness ratio model gave a closer fit to empirical results than the percent 52 53 contrast model. It was suggested that addition of a factor giving consideration to high relative brightness when in combination with high relative brightness ratio would refine the models. REFERENCES Berlyne, D. E., Stimulus intensity and attention in relation to learning theory, Quarterly Journal of Experimental Psychology, 1950, 2, 71-75. Berlyne, D. E., Conflict, Apousa1,_and_Curiosity, N. Y.: McGraw-Hill, 1960. Blackwell, H. R., Contrast thresholds of the human eye, Journal of the Optical Sociepy of_America, 1946, 36, 624-643. Bragiel, R. M. and Perkins, C. C., Conditioned stimulus intensity and response speed, Journal of Experimental Psychology, 1954, 47, 437-441. Breese, B. B., On inhibition, Psychological Monographs, 1899, 3, #1. Burtt, H. E., Applied Psychology, Englewood Cliffs: Prentice- Hall, 1957. Dallenbach, K. Mg, Position vs. intensity as a determinant of clearness, American Journal of Psychology, 1923, 34, 282-286. Forbes, T. W., A method for analysis of the effectiveness of highway signs, Journgl of Applied Psychology, 1939, 23, 669-684. 54 55 Forbes, T. W., Snyder, T. E., and Pain, R. F., A Study of lpaffic Sigp Requirements: II An Annotated Bibliog: lggphy, E. Lansing: iMichigan State University Division of Engineering Research, 1964, 85. Forbes, T. W., Snyder, T. E., and Pain, R. F., Traffic sign requirements: I review of factors involved, previous studies, and needed research, Highway Regearch Record, 1965, #70. Forbes, T. W., Pain, R. F., Fry, J. P., and Joyce, R. P., -Effect of sign position and brightness on seeing simu- lated highway signs, Highwangesgarch Record, 1967, #164. Forbes, T. W., Fry, J. P., Joyce, R. P., and Pain, R. F., Letter and sign contrast, brightness, and size effects on visibility, Highway Reseapch_Record, 1967 (in press). Forbes, T. W., Pain, R. F., Joyce, R. P., and Fry, J. P., Color and brightness factors in simulated and full scale traffic sign visibility, Highway Research Record, 1968 (in press). Forbes, T. W., A study of traffic sign requirements: final report, Division of Engineering Research, Michigan State University, 1967. Graham, C., (ed.) Vision and Visual Perceptiop, N. Y.: John Wiley and Sons, 1966. 56 Gray, J. A., Stimulus Intensity Dynamism, Psychological Bulletin, 1965, 63, 180-196. Guilford, J. P., Psychometrlc Methods, N. Y.: McGraw-Hill, 1954. Hull, C. L., Stimulus intensity dynamism.(V) and stimulus generalization, Psychological Review, 1949, 56, 67-76. Jameson, D. and Hurvich, L. Mg, Complexities of Perceived Brightness, Science, 1961, 133, 174-179. Johnsgard, K.'W., The role of contrast in stimulus intensity dynamism (V), Journal of Experimental Psychology, 1957, 53, 173-179. Johnson, D. M5, The Psychology of Thought and Judgemenp, N. Y.: Harper Bros., 1955. Lucas, D. B. and Britt, S. H., Advertising_Psyehology and Research, N. Y.: McGraw-Hill, 1950. Mackworth, N. H., Kaplan, I. T., and Metley, W., Eye move- ments during vigilance, Perceptual and Motor Skills, 1964, 18, 397-402. McCormick, E. J ., Human_Factors Engineering, N. Y.: McGraw- Hill, 1964. Perkins, C. C., The relation between conditioned stimulus intensity and response strength, Journal of Experimental Psychology, 1953. 46, 225. 57 Razran, G. H. 3., Studies in configural conditioning, III, the factors of similarity, proximity, and continuity in configural conditioning, Journal of Experimental Psychology, 1939, 24, 202. Richards, 0. W., Vision at levels of night road illumination XII, changes of acuity and contrast sensitivity with age, American Journal of Optometry, 1966, 43, 313-319. APPENDIX I Instructions to 3'3 In place of this target there will be two stimuli, one on the right and one on the left side (positions pointed out). Each trial will begin with this screen in place (screen over mirror demonstrated). You will be looking at the bottom of this frame (pointed out), I will say "ready" and then lift the screen. ‘We would like you to first look just above the center of the metal frame and there fixate briefly. Then look anywhere you like. At the same time we would like you to indicate which of the two stimuli you see "best and quickest." You do this by pressing the appropriate buttons on the box under the eye movement recorder (demonstration). While you are doing these tasks, a picture will be exposed for from 3 to 5 seconds. After the shutter closes and I say something, you may sit back (let go of the bite board) and relax until the next trial. Do you have any questions? 58 APPENDIX II First, in a paired comparison study with brightness as the dimension judged, Forbes et a1. (1967) found that S's could scale brightness from lowest to highest or vice-versa with equal facility. Second, when using verbal, subjective instructions such as "best and quickest" it was very easy for S's to ignore this and adopt some other criterion, 'e.g., most pleasant, liked best. Third, it was possible that when a subject was faced with an abrupt change in the stimulus dimension to be judged he changed criterion be- cause he was trying to second guess the hypothesis of the experiment. Fourth, the experiment was an attempt to assess the effect of the variables alone and when varied together. Since there were no clear cut cases of response reversal when both dimensions were varied (see Table 15) it was considered best to reverse the four cases of differ- ent response patterns and assume that a different sample of subjects would not have resulted in any reversed response patterns. 59 Appendix III Comparison of Different Methods of Calculating Contrast Negative Contrast Direction Brightness Ratio %Contrast Contrast 131 ‘ $4324 100 Bb-Bo BZ where _B1 BO where B1=brighter brightness Bis-brighter Bo=object/ darker B2=darker brightness B2=darker than background (McCormick, 1964) Bb=background —2. 48* -59. 8% . 593 -4.45 -76.5% . 765 -5.39 -81.4% - .814 Positive Contrast Direction Same Formula Same Formula BO -Bb Eb where Bo=object if brighter than background Bb=backgorund (Blackwell, 1946) 3.8 73.6% 2.79 7.13 85.96% 6.13 9.7 89.7% 8.73 *Negative sign indicates contrast direction. All numbers are positive. 60 61. Table 1. Experimental Design B1 ! 2 I 3 ' . A1 A2 A3 I A1 3 A3 A1 g__ A; A1 ‘ O O x x Bl A2 \ O x x A3 x x ”-1 A1 0 o x 32 O x. As x A1 0 o B; A2 0 A2 = CHIP BRIGHTNESSES O ___ B___K CONDITION A3 x = R=K CONDITION 13'1 32 = BRIGHTN ESS RATIOS BLANK 133 SQUARE = B AND R 7: K CONDITION 62 Table 2 Summary of Stimulus Conditions for C-positive and C-negative Phases A1 A2 A3 = Brightness levels B1 B2 B3 = Brightness ratio levels Brightness = K Brightness Ratio = K Brightness and Brightness Ratio at K (Bl-B2) /A1 (Al-A2) /31 (Bl-B3) /A1 (Al-A3) /31 A131 - A232 (32-33) IAI - (AZ-A3) /31 A132 - A231 A131 - A332 (Bl-B2) /A2 (Al-A2) /32 A132 - A331 (Bl-B3) /A2 (AI-A3) /32 A231 - A332 (32-33) IA2 (AZ-A3) /32 A232 - A3B1 (Bl-B2) IA3 (Al-A2) [33 A1B1 - A233 (Bl-B3) IA3 (Al-A3) /33 A133 - A231 (32-33) IA3 (AZ-A3) /33 A1B1 — A333 A133 - A331 A231 - A3B3 A233 - A331 A132 - A233 A133 - A232 A132 - A333 A1B3 - A332 A232 - A3B3 A2B3 - A3B2 63 Table 3 Chip and Card Brightnesses and the Resultant Brightness Ratios: Measurements Made in the Eye Movement Camera Apparatus with an Adapting Luminance of 0. 24 ft. L. (a) Positive Contrast Condition Brightness ' 8. 5 7.5 6. 5 (Munsell scale value) 2.8 2.0 1.45 (ft. Lamberts) Card necessary to achieve a given ratio Ratio BO BB 4. 75 4.0 3. 5 (Munsell scale value) 3.79 0.75 0.52 0.38 (ft. Lamberts) 7. 13 3. 5 3. 0 2. 5 0. 38 0.28 0.21 9.7 3.0 2.5- 2.0 0. 28 0. 21 0. 15 (b) Negative Contrast Condition 2.5. 3. 5 I 4.5 (Munsell scale value) 0.21 0.38 0.66 (ft. Lamberts) Card necessary to achieve a given ratio Ratio“ .32. B0 4.0 5. 25 7.0 (Munsell scale value) ~2.48 0.52 0.92 1. 7 (ft. Lamberts) -4.25 5.25 6.75 8 5 0.92 1.58 2.8 -5.39 6.0 7.5 9.5 1. 15 2. 0 306 *The brightness ratios in the negative contrast direction are positive numbers. The negative sign is used only to differentiate between positive and negative contrast direction brightness ratios. 64 Table 4 Age-and Characteristics of C-positive Phase Subjects Control Male Female Group # 3 7 mean age 19 19 age range 19 - 20 18 ~20 Experimental Group # 1 9 mean age 19 19 age range 18-20 Table 5 Age and Characteristics of C-negative Phase Subjects Control Male Female Group # 2 8 mean age 19 18.5 age range 18-20 18-20 Experimental Group # 2 8 mean age 22 18. 5 age range 21-23 .18 -20 65 Table 6 B = K Condition Paired Comparison Scale Values for Each Dependent Variable; C—positive Phase a ‘ I . (b) Subjective Stimulus fixated reaction , first Brightness Ratios level 3.8 7.1 9.7 3.8 7.1 9.7 1.45 0.000 1.782 2.482 0.652 0.517 0.000 2.0 """B'BBB ””” I .‘Jéimffiéi 5355""332'7m3'333 2.8 "MB-.633 """ 1' .‘sz'mz’fe'éZ 6365""3’38T'BTE7; (C) ((1) Time on each Number of stimulus - fixations Brightness Ratios level 3.8 7.1 9.7 3.8 7.1 9.7 1.45 . . 0.000‘ 0.946 0.165 0.135' 0.396 0.000 2.0 0.000' 1.143 0.023 0.000 0.293 0.951 2.3 0:000 0:040 0:577 0.000 0.800 1.025 66 Table 7 B = K Conditon Paired Comparison Scale Values for Each Dependent Variable; C-negative Phase (a) (b) Subjective Stimulus fixawd reaction Lips; Brightness Ratios level 2.4 4.2 5.4 0.21 0.38 0.66 0.21 0.000 1.222 1.964 0.000 0.645 0.125 0.38 0.000 1.005 1.530 0.000 1.530 0.525 0.66 0.000 0.472 1.829 0.000 0.645 0.525 (c) ' (d) Time on each Number of Stimulus Fixations Brightness Ratios level - 2.4 4.2 5.4 2.4 ' 4.2 5.4 0.21 0.073 0.000 0.922 0.025 0.000 0.100 0.38 0.000 0.660 0.660 0.000 0.013 0.062 0.66 0.213 0.000 0.511 0.000 0.250 0.275 it .llkvyutu n? .n' r..).. ,. . .. .fln . . . . .. .1. out llil“!fl 67 Table 8 R -- K Condition Paired Comparison Scale Values for Each Dependent Variable; C-positive Phase (a) . (b) Subjective Stimulus fixated reaction first Ratio Brightness in Ft. L. level 1.45 2.0 2.8 1.45 2.0 2.8 3.8 0.000 0.255 1.445 0.293 0.000 0.840 7. 1 0.472 0.000 1.829 0.652 0.000 0.517 9.7 0.000 .535 1.530 0.390 0.780 0.000 (0) ((1) Time on each Number of stimulus fixations Ratio Brightness in Ft. L. level 1.45 2.1' 2.8 1.45 2.0 2.8 3.8 0.782' 0.000 .946 0.203 0.000 0.214 7.1 1.167 0.000 .845 0.690 0.000 0.300 9.7 0.638 0.650 .000 0.145 0.243 0.000 68' Table 9 R = K Condition Paired Comparison Scale Values for Each Dependent Variable; C-negative Phase (a) (b) Subjective Stimulus fixated reaction first Ratio Brightness in Ft. L. level 0.21 0.38 0.66 0.21 0.38 0.66 2.4 0.000 0.632 1.669 0.262 0.000 0.524 4.2 0. 135 0.000 0.525 0.530 0.000 0.530 5.4 0.000 1.817 3.020 0.000 0.525 0.915 (0) ((1) Time on each Number of stgulus fixations Ratio Brightness in Ft. L. level 0.21 0.38' 0.66 0.21 0.38 0.66 2.4 0.752 0.000 0. 862 0. 124 0.000 0.212 4.2 0.000 0. 183 0.837 0.089 0.000 0.325 5.4 0.000 1.132 1.357 0.000 0.115 0.230 69 Table 10 B and R a! K Condition Paired Comparison Scale Values for Each Dependent Variable; C-positive Phase % contrast 73.6% 85. 9% 89. 7% Ratio (Brightness) 3.8 7 . 1 9. 7 Stimulus Brightness 1.45 2.0 2.8 1.45 2.0 2.8 1.45 2.0 2.8 Subject reaction 0.000 0.000 0.080 1.142 1.027 0.896 1.027 .7491.216 Stimulus fixated first 0.052 0.138' 0.000 0.525 0.656 0.653 0.767 0.330 0.925 Time on each stimulus 0.000 0.163 0.125 0.141 0.809 0.742 0.464 0.4410.387 Number of fixations 0.034 0.006 0.000 0.366 0.565 0.344 0.465 0.3840.342 Table 11 B and R a! K Condition Paired Comparison Scale Values for Each Dependent Variable; C-negative Phase % contrast 59.8% 76. 5% 81.4% Ratio 2.4 4.2 5.4 Stimulus . brightness 0.21 0.88 0.66 0.21 0.38 0.66 0.21 0.38 0.66 Subject A - reaction 0.000 0.330 0.348- 0.348 0.886 1.478' 1.308 1.4561.964 Stimulus fixated first 0.000 0.558 0.672 0.000 0.650 0.931 0.609 0.8671.210 Time on each ‘ stimulus 0.000 0.436 0.420 0.196 0.769 1.127 1.023 1.1871.611 Number of fixations 0.000 0.047 0.125 0.013 0.074 0.233 0.164 0.239‘0.318 Table 12 Rank Order Correlations Between Dependent Variable Condition Brightness constant (B=K) Ratio constant (R=K) Brightness and ratio a! K (Band R at K) Condition B = K R = K BandRstK C-Positive Phase Subjective reaction vs. Chip Fixed first R=0. 17 R=0.55 R=0. 77 Chip Fixated First vs. Time on Each Chip R=0. 80 R=0. 83 R=0.67 Subjective reaction vs. Time on Each Chip 0.88 0.55 0.52 Chip Fixated First vs . Number of Fixations 0.48 0.83 0.60 Subjective reaction vs. Number of Fixations 0.325 0.58 0.60 Time on Each Chip vs. Number of Fixations 0.35 0.92 0.78 Table 13 Rank Order Correlations Between Dependent Variables C-negative Phase Subjective reaction vs. Chip Condiflgg W Brightness constant (B=K) Rd). 3'? Ratio constant (R=K) R=0 . 50 Brightness and * ratio at K (Band R 7‘ K) R=0. 93 Chip Fixated First vs. Time on _ Each Condition Chip B = K R=0. 07 R = K ~ R=0. 65 Band's a! K R=0. 85 Subjective reaction vs. Time on Each Chip 0.63 0.58, 0.90 Chip Fixawd First vs. Nunb er of Fixations -0. 15 0.83 0.87 Subjective reaction vs . Number maidens 0.40 0.37 0.90 Time on Each Chip vs. Number of Fixations 0. _57 0. 87 0.88 72 Table 14 Individual Response Patterns for the B = K and R = K Conditions, C-positive Phase, Subjective Reaction Measure B=K _..3.8 7.1 9.7 R=K 6.5 7.5 8.5 1. 6.5 1 11 1. 3.8 1 11 7.5 1 11 7. 1 1 11 8.5 1 11 9.7 1 11 2. 1 1 1 2. 1 1 1 1 11 1 11 1 11 1 1 1 3. 1 1 1 3. 1 11 11 1 1 l! 1 1 11 1 1 1 4. 1 1 1 4. 1 11 1' 11 1 11 1 1 1 11 1 5. 11 1 ‘ z 5. 1 11 1 11 1 11 11 1 1 11 6. 1 11 6. 11 1 1 11 11 1 1 11 11 1 7. 1 11 7. 11 1 11 1 1 1 1 1 11 1 11 8. 1 11 8 1 11 11 1 11 1 1 11 1 11 9. 1 11 9. . 1 1 1 1 1 1 1 11 1 11 11 1 10. 11 1 10. 1 11 11 1 1 11 11 1 1 1 1 73 Table 15 Individual Response Patterns for B and R at K Condition, C-positive Phase, Subjective Reaction Measure _9_:_7_ 7. 1 ii 8.5 7.5 6.5 8.5 7.5 6.5 8.5 7.5 6.5 1. 111 11 111 111 11 11 1 11 2. 11 1 11 111 111 111 ' 1 1 11 3. 1111 11 11 11 111 111 11 4. 111 111 1111 11 g 11 1111 5. 11 111 1111 I 11 111 1111 6. 111 1 111 1 1 1111 1 1 111 7. 11 11 111 1111 11 11 11 1 8. 11 111 111 11 111 111 11 9. 11 11 1 1111 11 1111 11 10. 111 1111 111 1 111 111 1 74 Figure 1 Hypothesized Relation Between Brightness and Brightness Ratio (Hypothesis 4) High Brightness Ratio High Low Brightness Ratio Attention Value Low Low Brightness High 75 Figure 2 Layout of Stimulus Configurations 3n~ * 3" —’| 5/8'T k— —H 5/ " l4—— 7/8" 7/ n L——. . ___—u ,6 377 a. 76 Figure 3 Outline Diagram, Polymetric Eye Movement Recorder, Model V-1164 Stage Unit Marker Light Mirror Base Beam/ , ‘ Marker O 0 Light Unit Ocular - Camera Semi-Transparent Mounting, ’ . /0 Mirror Bracket / ' .4! , in O .4 ° \ n a" ‘5‘ Optical Head Rest System Bite Board Camera Eye Lens Recording Platform Scale Values Scale Values 77 Figure 4 B = K Condition Paired Comparison Scale Values Plotted for Each Dependent Variable; C-positive Phase (a) Subj ective Reactions 3.0- 2.5- / 1.45 Ft LV 2.0- ' I I l 3. 8 7'. 1 9. 7 Brightness Ratio (C) Time On Each Chip 1. 5 - ’4 l l 3.8 7.1 9.7 Brightness Ratio (b) Chip Fixated First ((0 Number of Fixations Scale Values Scale Values 78 Figure 5 B = K Condition Paired Comparison Scale Values Plotted for Each Dependent Variable; C-negative Phase (a) Subjective Reactions 2. 0 - . 21 Ft L / 66 Ft L\7’ 1.5 - ' .38 Ft L ’ ’/ 1. 0 " // / / .5 / / / 0 .— 1 1 J 1.5 -2. 4 -4. 2 -5. 4 Brightness Ratio (C) Time On Each Chip J -5.4 l l -2. 4 -4. 2 Brightness Ratio (b) Chip Fixated First / \ . 38 \ 21 / 66 )\ b. \ / / / J_ I -2.4 -4 2 -5.4 («0 Number of Fixations ” .66 .38 ” .21 l 1 J -2,4 -4 2 -5.4 79 Figure 6 R = K Condition Paired Comparison Scale Values Plotted for Each Dependent Variable; C-positive Phase (a) f (b) Subjective Chip Fixated Reactions First 2.0 - . / 6 L5- 97Rum / . g 3. 8 Ratio // ;> 1 0 _, 7.1 Ratio m O 5 a: . 5 .. 0 .- 1 1 J 1. 45 2. 0 2. 8 Chip Brightness, ft 1 (C) (C!) Time On Each Number of Chip ‘ Fixations l. 5 - _ ' CD 0 .3 1.0 ~ :> d) '3 . 5 - 0 m o - l 1 J 1. 45 2. 0 2. 8 Chip Brightness, ft 1 Scale Values Scale Values . 3.0 2.5 2.0 1.5 1.0 1.5 1.0 80 Figure 7 R = K Condition Paired Comparison Scale Values Plotted for Each Dependent Variable; C-negative Phase (a) (b) Subjective - Chip Fixated Reactions First -5. 4 Ratio _ / . / . -2. 4 Ratio r— h- r t >- #- 1 l l l l J .21 .38 .66 .21 .38 - .66 Chip Brightness, ft 1 (o) - (d) A Time On Each Number of Chip Fixations -4. 2 b '- -5. 4\ -2. 4E 25 ’ l L J .21 .38 .66 .21 .38 .66 Chip Brightness, ft 1 Scale Values Scale Values 81 Figure 8 B and R at K Condition Paired Comparison Scale Values Plotted for Each Dependent Variable; C-positive Phase (a) (b) Subjective Chip First Reactions Fixated l. 5 - _ /2.8 Ft L 1. 0 - />< 2. 0 Ft L 1 \ // ‘~ 1. 45 Ft L .5 ~ - 0 - >- I i I 1 1 1 3.8 7.1 9.7 3.8 7.1 9.7 Brightness Ratios (e) ((0 Time On Each Number of Chip . Fixations 1. 5 _ 2. 0 l. 0 i- 2. 8 \\ " 2. 0 \ 1. 4\ . 5 *- 1. 45 §\ - 5 ‘ / 2. 8 /_ __ _ y / 0 >- I l m 1 4 ~ 3.8 7.1 9.7 3.8 7.1 9.7 Brightness Ratios Scale Values Scale Values 1.5 1.0 1.5 1.0 Figure 9 B and R 1: K Condition Paired Comparison Scale Values Plotted for Each Dependent Variable; C-positive Phase (20 Subjective Reactions 7 9.7 Ratio 7. 1 Ratio .. \ ~~ “ (.15) / (38) (.21) ' 3.8Ratio 5.33) i (.52) (15) 1 l J 1. 45 2. 0 2. 8 Brightness, ft 1 (C) Time On Each Chip " 7.1 f~~—_ / u— // 907 3.8 F" __f fl l I J 1. 45 2. 0 2. 8 Brightness, ftl Note: Card Brightness in ( (b) Chip Fixated First T 5" .4 ((1) Number of Fixations >" \\ ‘Q 3.8 l L 4 1 45 2.0 2.8 ) 83 wouwamfiaam 5333920 625.80 M 1. m as m .5925 2558-0 85 .28 338m . comwuaafioo “Cohan c.3932 core—«om 33335 one 3 6.39m same A smog Scale Values Scale Values H 0' j—t O H 0 01 H C Figure 11 B and R at K Condition Paired Comparison Scale Values Plotted for Each Dependent Variable; C-negative Phase (a) Subj ective Reaction .66FtLy/ .38FtL I r- / / / / / l / / / .21 Ft L I / l _J I -2. 4 -4. 2 -5. 4 Brightness Ratios (0) Time On Each Chip ,— l I -2. 4 -4. 2 -5. 4 Brightness Ratios (b) Chip Fixated First ((0 Number of Fixations . 66 .38\’ . 21 / v’ 1 1 J ' . 7 5.10%}.7 fig MEI-5.... tun. .. x .1. Scale Values Scale Values 2.0 1.5 1.0 2.0 1.5 1.0 Figure 12 B and R :f K Condition Paired Comparison Scale Values Plotted for Each Dependent Variable; C-negative Phase (20 Subjective Reactions ' -5. 4 Ratio (2. 0)) / (1.58) / / / /:4. 2 Ratio -°, (. 92) (1.7) ' -2. 4 Ratio 1 I J . 21 . 38 . 66 Chip Brightness, ft 1 (C) Time On Each Chip I I I 2. 5 3. 5 4. 5 Chip Brightness, ft 1 (b) Chip Fixated First -5.4 / o- / ’ 4 2 / /’ -2 4 ll ’6 / I L J ((1) Number of Fixations Note: Card Brightness in ( ) 86 cm: 665.8885 3362920 603850 M a m “:8 m 6925 ozuewocuu 05 new Bgmom 22823800 “Cohen 9:6on conoeom mzuoonnam ma 95th to o In 0 o o H H C N sanre A 81203 fill-Sigiruflw a” .k' 4 ._ A _ Illillivl‘ll'l' 87 "83223285 mfiaoEnaao £33980 M 1. m Ea m .826 63:86-0 2: .82 333m 282.3860 602mm 92.5on «warm nofifih QED 3 enema same A smog 88 \,,\ m.\ umuwamfifinm 3333930 5038.80 M n m use. m .0925 ozuawocuo 9: you 333m 5252380 3th 9:5on 3.3m cousin Q20 3 enema o .4 ID 0 H o N' 881118 A areas 89 35.8885 bfiofinmuc £03350 M i m 98 m 6925 95609-0 93 a3 333m :omEamEoo 3th 3332 QED comm so 259 3 magma In I!) O .4 H 931mm areas o N' 633.2956 3303920 605980 M x m can m .mmanm ozfiwocuo 2: no“ 338m :omzaafioo 8th 9:532 QED comm co m8; 5 magma same A 61803 II II- . .-.-I:I'l 1W “Enid—555 bfimoEQauw £05980 M x m Ea m .395 2533-0 2: .8“ 315% comanEoo 6225 2:982 mcofiafih no “3852 3 ouswwh [D 0 CI!) .4 A sanre A 91803 N. .I1 I Ill 1" 92 6325885 3302920 £05980 M x m can m 6925 235390 05 .8“ mfismom comtmafioo cob—um c.3232 muoSafim no 39852 3 939““ «zoo H O N .4 same A 91203 93 Figure 20 Comparison of Results Observed Experimentally and Theoretical Results Calculated from Brightness Ratio and Contrast Models C-positive Phase 5; 100~ .3 oObS O 3 80- or 'O 5 60~ ad to 0) 2° 40- t: m 0) H t: (D 2 a) 0’ On Low Brightness Ratio High Low Chip Brightness , High Obs = Observed BR = Brightness Ratio Model %C = Percent Contrast Model 3;; C) if.) 80' OBS 5’ /BR '5 6O - ‘63 £3 40- 5 a, 20» U) Jul ‘1'. ‘1’ 0 O 3", Low Brightness Ratio High a. High Brightness Low 94 Figure 21 Comparison of Results Observed Experimentally and Theoretical Results Calculated from Brightness Ratio and Contrast Models C-negative Phase 3;; 90 p Q Obs .92 .2 a 70 P 0 BR '6 ’I” 5 50 p =-- ""‘ '%C H m G) ‘39 3o - s: d.) 0 g 10 r- c: Q) 0 S a. Low Brightness Ratio High Low Brightness High Obs = Observed BR = Brightness Ratio Model %C = Percent Contrast Model is g , 30 - o Obs g ,0 BR 60 .. 'g ’ / . ch a . *5; 4o » "' 2‘3 o’ 'c: 20» o O) m a o P Q) 2 3: Low Brightness Ratio High ' High Chip Brightness Low ___——