ENGINEERING RESEARCH INSTITUTE THE UNIVERSITY OF MICHIGAN ANN ARBOR Final Report A STUDY OF POSSIBLE "PHOTOSENSITIZATION" OF THE HUMAN EYE BY WHITE LIGHT H. Richard Blackwell O. Thomas Law Vision Research La]boratories-. ERI Project 2455 BUREAU OF SHIPS, DEPARTMENT OF THE NAVY CONTRACT NO. Nobs-72038 WASHINGTON, D. C. June 1958

"T Z3; C_ e u? I

2455-9-F The experiments reported here represent the basis for the doctoral dissertation "The Effect of Background Luminance on Brightness Discrimination" by O0 Thomas Law University of Michigan, 1952

The University of Michigan 245 Ti * Engineering Research Institute;5-9 -F,P CONTENTS,tie i iit Io HIo III List of Figures List of Tables Introduction Apparatus and Procedures Results Discussion References iv 1 3 12 154 15 1 ii

- The University of Michigan T Engineering Research Institute I 2 3 8 10 11 2455 -9-rF LIST OF FIGURES Title Artist's conception of the basic test facility~ Optical schematic draing of the Threshold data for observer OTLS Threshold Threshold Threshold Threshold Threshold Threshold Threshold Values of Data for Observer AK; Data for Observer HF; Data for Observer LP* Data for Observer OTL Data for Observer AK; Data for Observer BHF Data for Observer LPo a / u =a 1 minute psychophysics target projector o 1 mi nute 1 minute 4 ~5 minutes: 45 minutes cH45 minutes a =45 minutes 12 Values of /R; 'a -5 mimnutes iii

IT l I- 1: — C KA L A A:- - r Y f - -~L Y= Y 1 CB JI LI IPI 9) Y = L ~ ~ a ine university o1 Ivllcnigan * Cnglle~erllg nKIetrci2455-9-F LIST OF TABLES tI ~ Refractive Deta from the Oculometer II Tkhreshold Data for Observer OTL; C- = I minute III Thresho1:d Data fo~r Obsekver AK;:- = 1 minute IV Threshold Data for Observer HF; '- = I minute V Threshold D:.'ta for Observer LP; = 1 minute VI Threshold Data for Observe;w OTL; ~ = 45 minutes VII Threshold Data for Observer AK; = 45 minutes VIII Threshold Data for Observer HF; a 45 minutes IX Threshold Data for Observer ILP4 a= 45 minutes X Differences in NR I I ELILUL. I iv

The University of Michigan T Engineering Research lnstitute 2455-9-F The value of the threshold luminance increment has been measured for each of a number of background luminance levels and for a totally dark backgroun-d, Target and background were white light, of color temperature 2850 Ko Targets subtended I and 45 minutes of arc, with a 001.second exposure duration. It was found that the threshold luminance increment increases smoothly and continuously as background luinance was increasedo The threshold was equivalent for a totally dark background and for low levels of background luminance, These data fail to exhibit ay evidence that the huma eye ca be '"photosensitized" with white light. (Photo sensitization would occur if the addition of a small amount of light superimposed over a lighted target and its totally dark background made the target more visible.) The data from the present study suggest that the classical distinction between the "absolute threshold" and the "difference t-hreshold" is not meaningful since there was no discontinuity in the data obtained with a totally dark background and those obtained with low levels of background luminaace, 1 V

L The University of Michigan * Engineering Research Institute -2455-9-F Io INTRODUCTION The capability of the human eye for the detection of targets differing from their background in luminance has been frequently studied during more than a century. The usual experiment may be described in the following wayo A large area, referred to as the adaptation or background luminance, or simply as the background, is uniformly illuminated to a predetermined level over its entire surface. The luminance of the background may be symbolized by the letter B. An additional quantity of lightp, AB, is added to some porticon of the large area for a brief duration0 The precise amount of this quantity is varied until a response is evoked from the observer indicating that he has detected a lack of uniformity in the luminance of the background for the brief duration of the increment. The value of AB which evokes the signalling response 50O, of the time is referred to as the brightness difference threshold, AB. 'The relations between the brightness difference threshold and various aspects of the stimulus situation have been extensively studied. The level of the luminance of the background is known to have an important effect on the detection capability of the eye. The sensitivity sof the eye has usually been descriebed in terms of the value of the contrast, C, which can just be detected where C =A~BB B It is well-known that the threshold value of C decreases steadily and smoothly as B is increased. Experiments involving a finite value of B have been customarily referred to as measurements of the "difference threshold". In the special case where B=O, it is customary to measure sensitivity by the value of 0AB. In these cases, it is customary to refer to t the data as measurements of the "absolute threshold". The usual definitions of the difference and the absolute thresholds make it impossible to compare them. For this reason, one of us (Ref. 1) in 1947 first plotted threshold values of A0B for all values of B. Use of this measure of visual detection senstivity makes it possible to compare values of the absolute threshold and values of the difference threshold for low levels of background luminanceo Although these comparisons were made in the 1947 study, there were few data available at various low levels of background luminanceo In recent years, there has been considerable interest in the development of theories of visual detection. Quantitative relations between difference and absolute thresholds are of crucial importance to several of these theories. It is imperative that the receptor population be identical for measurements 1

I 7 - The University of Michigan ~ Engineering Research Institute 2455-9-F relating these quantities, Most existing data which could be plotted in terms of nAB were collected under conditions in which cone photoreceptors were responsible for the threshold for the high values of B whereas rod photo-receptors were responsible at low values of B. Suitable data for theoretical purposes would require that either a pure cone or a pure rod population be utilized. It is, of courset, considerably easier to isolate a pure cone than a pure rod population, since cones exist alone in the normal foveal retinae whereas rods do not exist entirely alone at any retinal location. The present study was undertaken to provide data on the variations in he iation i he threshold 1umnance increment for the foveal retina as a function of a wide ranggeof values of the luminance- of the backgrond c including a value of zero0 Simple theoretical arguments would lead to the prediction that the threshold value of B would increase steadily and smoothly as. B was increased, and that there would not be a discontinuity in the data for B=O, From this'point o'f"vti&e:th& taiimm' valueof:,iD " should occur ~or =O Verification or rejection of this theoretical relationship seemed to be a straightforward experimental problemE The present study was undertaken for an additional reason0 Crozier (Ref0 2) has presented data suggesting that A B has a minimum value for B>)O this result has been taken to represent "photosensitization" of the eye at low levels of background aluinanceo There are conceivable mechanisms of the visual detection process which would lead to this result; indeed, some detectors of electromagnetic radiation have' this property0 It was decided that a concerted effort would be made to verify or reject the possibility of photosensitization of the human eye. by white light. It should be apparent that photosensitization of the human eye could have important military usefulness in connection with improving the night vision capabilities of night observers. To take military advantage of photosensitization, it would be necessary only to flood the entire retinae of the eyes with a uanform veil of dim light, which would cover the images of targets and their backgrounds existing in real space0 The photosensitizatiaton mechanism would result in an increase in the visibility of targets under these conditions0 Crozier suggested that only one specific value of dim light produced the effect, other values re$ulting in impaired visibilityo Since a particular value presumably had to be used to elicit it, the existence of photosensitization could not be rejected out of hand on the basis that it was not confirmed by ordinary experience in night observingo It was decided to study detection thresholds at a number of values of background luminance in addition to zero to attempt to identify I 2

The University of Michigan Engineering Research Institute 2455-9-F the precise luminance required for the photosensitization process to occur, IIo Apparatus and Procedures As noted above, the present study is intended to provide evidence on the fundamental nature of the relation between threshold,luminance increment,, AB, and background luminance, 3, for the foveal retina. The extent of our knowledge of the functioning of the visual system is s$fficiently advanced so that it is possible to specify a number of experimental conditions which must be met if meaningful data are to be obtained, These conditions may be listed briefly as follows: 1o Control of the location of the target on the approximate center of the foveal retinas 2Q Control of the pupillary apertures 3 0 Control of the refractive state of the eyes ~4 Control of the background luminance; and 5 Control of the size, luminance, and duration of the target The ways in which these contlrols were mLintained are described in detail in the following sections0 AO OBSERVERS Four observers were used throughout the experiments so that the results would have reasonable generality0. Two of the observers were adult females, one (LP) an emmetrope, and one (HF) with corrected low-order myopia and astigmatismo The other two were adult males, one (oTL) a corrected myope of moderately high-order, the other (AK) with corrected low-order myopia. All observers were thoroughly trained prior to recording the experimental data. Three of the observers were graduate students in experimental p$syhology, whereas the fo urth was a laboratory technician. Motiviation seemed well sustained throughout the long and tedious series of experiments$ Because of individual differences in the sensitivity of the observers, it was necessary at times to collect data on groups of two observers at a time; at other times, because of the nature of the orientation light system used, it was necessary to collect data on observers singly0 In generalj, all observers worked at a given luminance level before they all proceeded to the next l uminance level0 The possible influence of practice effects due to increases in the observers' familiarity with the conditions and procedures, and their facility in response, was 'in great part obviated by a protract ed period of practice (about two weeks) in the observing situation, and further mitigated by the order of the experimental sessions as described below,, Observers' were practiced in the situation until their threshold curves showed I I 3

L The University of Michigan * Engineering Research Institute I 2455 9-F no further improvement Be APPARATUS The apparatus used to produce the background and target luminances is illustrated schematically in Figure 1. The observers viewed the far wall of a large'cube, which served as the background luminance,- through a large opening in the near wallo The far wall was lighted by a series of incandescent lamps, located on the near wall of the cube around the opening, which were shielded from the observers' eyes. The target appeared from time to time in the center of the far wall, as a luminance incremento The target was produced by a projection system located behind the far wallo The larger portion of the far wall was covered with a translucent plastic screen. The target was produced by translluminating the screen over a restricted area. The screen was located an average of 1008 feet from the eyes of the observers. High background luminances were produced by four arrays of 100l watt lamps, one array on each of the sides of the opening in the near waill Reduction in the number of these lamps was used to produce luminances in the range from 3 to 100 ft.-Lo A level of 1 fto-L4 was produced by three 40-watt incandescent lamps, one at each of the sides and top of the near wall. In each case, direct illumination of the plastic screen by the lamps was prevented by disc deflectors placed on the lamps. For all values below 1 fto-L., the luminance of the plastic screen was provided by two small integrating light boxes containing 6-volt, 32-candle-power incandescent lamps. Each light-box had an opal glass screen, These screens were not used to illuminate the plastic screen directly, but were used to illuminate the near wall of the cube directly, the plastic screen being illuminated entirely by reflection from the near wallo These two boxes were mounted against the sides of the cube walls at a point which prevented the observers from seeing the opal glass surfaces. The luminance of the plastic screen was adjusted by use of Wratten neutral filters, placed over the opal screens of the light-boxeso The target projector was based upon a special tungsten ribbon-filament lamp developed by the General Electric Company, the output of which was equivalent to a 1000-watt monoplane projection lamp0 The ribbon-filament lamp was operated at 6-volts AC. Two sets of Wratten neutral filters were used to adjust target luminanceo "Fixed" filters reduced the target luminance to the threshold range. These filters were fixed during any given experimental se$siomn0 Additional adjustment of the target luminance was obtained by the "psychophysical" filters. These filters adjusted tlhe target luminance to the values desired within the psychophysical range. Five psychophysical filters were used, mounted on an electrically-driven filter-selector which could positiontf one or anothper of them in the pr1ojection beam. Transmittance values for these filters were: 1.000, 0.752, 0 450, 0,287, and 0172,

The University of Michigan * Engineering Research Institute - 2455-9-F The presentation of the target was controlled by the operation of a flag-type shutter0 Whenever a solenoid was activatedl, the shutter was removed from the projection beam. The optical system of the projector is shown in Figure 2o The target was presented through the rear of the translucent plastic screen (I). A metal plate with a circular hole was pressed tightly against the rear of the screen, and this aperture defined the size of the target The projector was used therefore, merely to ilb:lluminate the metal plate uniformly, An image of the filament (9) was formed by the condensor lenses (6) in the plane of the shutter wheel (3). The lens (2) imaged the plane of uniform luminance (7') on the metal aperture. The aperture (8) was placed in a lamp house which prevented unfocused light from the lamp from illuminating the screen. The target projector output was reduced by the fixed filters (5) and the psychophysical filters (4)o Different focal length lenses (2) were used to illuminate the plates used to produce the two target sizes studied, The side of the, plastic screen facing the observers.was covered with an extremely thin layer of white sphere paint, designed to eliminate the specularity of the plastic screen without introducing spectral selectivity. The color temperatures of the target and of the background luminance were set for 2850~ Ks The timing of the target presentation was arranged and controlled by a timer comprising two discs mounted on a single shaft, one rotating at seven times the speed of the other, and designed to juxtapose two adjustable slots in the perimeters of the discs and the projection beam. The timer wheels intercepted the projector beam in the plane of a filament image0 Variation in the chord length of the slots can be utilized to produce continuous variation in the duration of the pulse of light from about 0o001 to 0,03 second0 The time required for the target to come to full luminance was 0.0001 second. (A ribbon-filament was used in order to minimize the onset time0) In the present experiment, an exposure duration of 0o01 second was used throughout0 This duration was selected to be shorter than the critical duration (Ref. 3) at all values of B. Experiments were conducted with each of two target sizes, representing 1 and 45 minutes of arc, These two sizes represent the extremes of a point source and an extended source which covers nearly the entire rod-free area of the central foveaQ C. EXPERIMENTAL PROCEDURES 1. Psychophysical Method The temporal forced-choice variant of the method of constant $timuli as described in detail by Blackwell (Ref. 4) was used in this study0 Essentially, the temporal forced-choice method I I 5

—..,a, r a ao I. min o I. - - fl L - r % de% r% rp ft Lf I "Av)~ A + e I I The University ot MIcnhgan * tnglneering Kese:arc IIstLILtutI 2455-9-F presents the observer with four successive, aurally delineated; 2-second time intervalso In one of these only, as determined randomly by the presentation sequence. there appears a target of ramdomly selected luminanceo After a cycle of four temporal intervals is completed, the observer is allowed eight seconds to press one of four button$ located on his arm rest indicating in which of the four intervals he' believes the stimulus to have appeared0 These responses are automatically tallied in a distant room on electrical counters and punched on record cards for permanent reference0 Although ten targets of the same luminance succeed each other, each block of ten is randomly arranged with respect to all other blocks of ten in terms of luminance. Five such blocks, ranging in difficulty from a target visible nearly 100% of the time to a value virtually never visible are presented in one experimental sessiono After the presentation of fifty targets, observers are permitted a five nute break while the equipment is re-set0 A fifteen inute break for refreshment customarily follows the third b ock of s timul o Automatic presentation and recording equipment was used, designed for use ath this method and described by Blackwel Pritchard, and o hmart (Ref0 5)o The amassing of the great amounts of data necessary to this study would have been impossible without this equipment o The basic experimental data were percentages of correct choice for each of five target luminance incrementso Analysis of the data begins by eliminating the effect of chance successes by means of the relationo 75 where p = corrected proportion, and p raw proportiono The corrected proportions were analyzed by a variant of the probit analysis developed by Kincaid and Blackwell (Refo 6), based on the probit method of Finney (Ref 7). asically the probit method fits a theoretical curve to the data to satisfy the maximum likelihood criterona 0 In this case, the theoretical curve was the normal ogiveo Analysis of the data in this manner yields the value of the threshold, the standard error of the threshold, the slope of the ogive, the standard error of the slope, and an estimate of the goodness of fit determined by the Chi-square test0 2 Photometric Procedures The basi$ of all photometry was the Macbeth Illuminometer, calbrated against tandar8 d aps and standard reflectance surfaces certified by the Electrical Testing Laboratories, New York0 The 6 I

The University of Michigan 5* Engineering Research Institute calibration of the Illumirometer and filters was checked several times during the experiments, High luminances of the screen were measured directly with the Illuminometer, fitted with a lens which imaged the screen in the photometric cube of the deviceo Low luminances, produced by the light boxes described above&, were photometered indirectly. The ratio between the luminances of the opal screens of these light boxes and the resulting screen luminance was measured at maximum output of the light-boxeso When the output of the light boxes was reduced by filters to produce low luminances of the screens the luminances of the opal screens of the light boxes were measured and the ratio used to compute the screen luminance0 This procedure is entirely adequate since the optics of the light sources were not altered by the use of the filters 0 The luminance of the large target was measured directly with the Illuminometer9 with all fixed and psychophysical filters removed from the projector~ The luminances of the targets actually used in the experiments -were computed from the luminance without filters and the transmittances of the filters, The transmittances of the filters were determied by standard photometric procedures, with an optical bench photometer. The luminance of the small target was more difficult to photometer~ The basic measurement involved what may be called a "candle-power box"0 A closed metal box was made with an opal disc at one end and a small aperture at the othero The aperture was fitted snugly against the screen so that the transilluminated target lay entirely within ito The target thus became a source for 'illumination of the opal disc at the other end of the box, Baffles were placed within the box to eliminate interreflections0 The luminance of the opal disc was measured with the Illeumanometer. From the transmittance of the opal disc and the inverse square law, the intensity of the small target could be determined, The luminance of the small target was computed from the measured size of the target. The measurement with the candlepower box is somewhat tediouso Accordingly, occasional measurements were made in this way and more frequent measurements were made with a photoelectric telephotometer, calibrated in terms of the candlepower box measurementso The telephotometer consisted of a telescope which imaged the small target on the cathode of a 931 photomultiplier tube0 The relative luminance of the target was determined by the emf required to compensate the photo-electric current produced by the target. 3o Order of Experiments All experimental measurements with the 1 minute target..1 7

The University of Michigan ~ Engineering Research Institute 2455-9-F were first made and then the measurements were made with the 45 minute targeto Studies with the 1 minute target are designated Experiment I; those with the 45 minute target are designated Experiment IIl Experiments were begun, for EJxperiment I with a background luminance of 10 fto-Lo which was then reduced in one log unit steps.n successive sessions to a terminal background value of "lzerol" Keeping all other conditions constant, background luminance was then increased from 3 3 x 1-06 ft -Lo in one log unit steps to 33 fto-Lo and finally to 1000 ft0-L0 and back to 100 ft -Lo No differential effects of the order of the experimental conditions were observed o Experiment II began at a background luminance of zero and increased in one log unit steps to 1 x 10~4 fte-L3 Experiments were then conducted at i0 ft.-Lo oand the background luminance was decreased in one log unit steps to 1 x 10 >- ft.-L.,, followed by experiments at 100 fto-L, and 1000 fto-L. 4o Other Experimental Procedures To insure stimulation of the same region of the foveal retina under all conditions, the observers' heads were fixed in position by means of hardened wax dental impressions in which the observers' teeth were clamped during all stimulus presentations. These dental impressions in turn were bolted to supports fastened to the arm rests of the observation chairs. Constant fixation and accommodation of the eye at the point of eventual appearance of the stimulus was required. Unfortunately, no single method of producing this result could be discovered which would serve satisfactorily throughout the extensive range of background luminance values. Consequently, three different systems were employed which produced equivalent orientation lightso In each case, the system produced four spots of light, each of approximately one minute of arc diameter, surrounding the stimulus spot and equidistant from its edges at a constant distance of eighteen minutes of arc. These four spots were arranged in the form of the terminal points of the arms of a cross, with the point of stimulus appearance located at the intersection of an imaginary line through each of the vertical and horsizontal pairs0 The orientation spots were maintained at a luminance value approximating ten times the average threshold of the group of observers for the prevailing luminance, a level found in previous studies to be comfortable and effective over long periods, The system used for producing the orientation lights for, background values ranging from 1 fte-Lo "'to 1 x 10-3 ft,-Lm consisted of four projectors using the images of the filaments of 6-volt, 32-candlepower tungsten lamps. These projectors were mounted on each of the four sides of the back of the cube and projected the orientation lights directly on the screen. I 8

r, t r - -. - - n - - - - -L li a a e ILA w-I - - Ihe University ot Michigan E tngineering Kesearcn Instrirure 2455-9-F The system used for background values less than 1 x 10-3 ft -Lo involved use of orientation light boxes. A 32 candlepower, 6-volt lamp housed in a light-tight metal case with four small holes provided the orientation spots for Experiment Io Because of the necessarily greater separation of the orientation lights, these were replaced by four 2o2 volt lamps (GoEo No0 222) with small spherical lenses for Exp4riment IIo These light-boxes were mounted at right angles to the observersu line of sight, at an optical distance equal to that of the distance from the observer to the plastic screen, A cover glass- mounted at forty-five degrees before the observers eye reflected the four spots into the observers$eyes so that they appeared to come from the screen. (The values of fx and of 3 were of course corrected for the transmission of the cover glasso) This system eliminated the existence of unintentional illumination of the light cube by the orientation lights The orientation lights used at higher luminance levels produced an unacceaptable amount of background luminanee from reflections within the light cube. In order to provide sufficient intensity for use at background luminance levels greater than 1 ft.-L0, orientation spots were produced by transillnmiation through the rear of the plastic screen. An orientation light template was made, consisting of four pieces of shimstock with holes made with a number 64 drill, which were fixed to the side of the glass plate bearing the target apertureo A light box containing one 2.2 volt lamp was mounted behind each of these, so that the orientation spots appeared on the face of the screen by transillumination from behind0 The luminance of these spots was controlled by means of a variable density 16 mmo photographic film, placed between each light source and the screen. Different positions of the film strip produced different luminances of the individual orientation spots0 Throughout the experiments$ monocular vision was employed, using the right eye of each observer. An artificial pupil was used in order to elimnate variations in retinal llumi nation with changes in the diameter of the natural pupil attendant on luminance changes of the background$s, and to make it unnecessary to determine for these experimental conditions the extent of the Stiles-Crawford effect The artificial pupil was fixed Oo2 to 003 inches before the eye of each observer, and on line with the optic axis0 Initially a 2 mmo artificial pupil was selected for these experiments, but it was noticed that when viewing an extended surface of uniform luminance through this aperture, the surface appeared to have several distinct areas differing in apparent luminance The appeae appearance of the field visible through the artificial pupil may be described as follows: There was a fuzzy penumbra region-just inside the circular limit of the visible field0 There was an annulus of greater apparent luminance and then a central area of lower apparent luminance0 The central 7 9

The University of Michigan ~ Engineering Research Institute 2455-9-F area of lower apparent luminance -v.... ompaQ a-t:ivelyl s!lta!!:Wih:a Ld the 2 mmo pupil but was enlarged by use of a 1 mmo artificial pupil. The 1 mm0 pupil was selected for use in the experiment, since it was considered expecially undesirable for the boundary between the central dark area and the brighter annulus to fall near the test target 'With the e1 mm. pupil. this boundary was separated from the nearest edge of the test target by at least 10 degreeso However, the target definitely fell within the central zone of lower apparent luminance0?Tpiis effect of an artificial pupil has apparently never been reported and considerable exploratory experimentation was required to detereine the basis for the observed effect. It was found that the effect could be duplicated in an entirely physical 9way with an ordinary spherical lens being used in place of the eye. The phenomenon was shown to be absent with.an aperture placed at the first principal point of the lens (:paralleling the case of the natural pupil of the eye)D -: was present with the aperture placed about four to six inches before this (paralleling the case with the artificial pupil)o Kincaid and Blackwell (Ref. 8) have shown that the effect is apparently due to the presence of spherical aberration in the eye0 The annulus of light surrounding the darker inner disc represents an excessive accumullation of light due to the fact that these rays are more nearly in focus than are than he rays arriving at other points on the retinao The refractive condition of the eye of each observer was determined under the conditions of the experiment by a method outlined by Ogle (Ref 9)~ An "oculometer" was used to measure dynamic refr, e-z:cti on with the stigmatoscopic technique0 Measurements were made with this apparatus under conditions identical with those of the experiment proper, with the exceptions that a stigma, or point of light, appeared in the apparent location usually occupied by the target0 The stigma was seen by reflection from a half-silvered mirror located in the oculometer The principal oculometric measurements were made with a 2 mmo artificial pupil to increase the precision of measurement, since the depth of focus of the eye is considerable with a 1 mm, pupil0 There is no a ro_ reason to expect there to be a difference between refractive error with I and 2 0mm pupils$ Moreover, a crude check was made by obtaining oculometric readings first with the I and then with the 2 mmo pupil0 No systematic differences were foundo Oculometric readings were taken at each of two background luminances$ 10 ft o-Lo and zeroo There is no a iori. reason to expect differences in refractive cnndition as a function of background luminance, and indeed none were found to exist0 I.1 10

The University of Michigan ~ Engineering Research Institute 2455-9-F The basic oculometric procedure may be described as followso the observer fixates the orientation spots situated 10.8 feet from the eye and observes changes in appearance of the bright stigma centered among them. The assembly which houses a tungsten lamp, stigma, and reduction filters in the oculometer is movable along the optic axis parallel to a scaleo The distance from the eye to the field lens is made equal to the focal length of the latter, making possible a linear scale calibrated in diopters whose modulus is determained by the dioptric strength of the field lenso (The zero point of the scale corresponds to the point at which the stigma-to-lens distance is equal to the focal length of the field lens and the image is at infinityo) Using the method of limits, the observer adjusts the position of the lamp housing until the stigma appears to have minimum size~ A mean of many such settings, translated into diopters by the affixed scale, reveals the refractive state of the eye0 Accommodation on the stigma would give a spUrious result. This possibility is precluded by causing the lamp to flash interumittently, the "off" period being sustained considerably longer than the "on", The discrepancy between the mean of these oculometric settings and the theoretical "normal" refractive SUate;: —ei6f'tl'he eye (-0o33 diopters) at this distance represents the refractive "error" of each observer, From refractive errors~, it is possible to compute refractive correctionso Oculometric measurements were repeated after corrections were made. This procedure was continued until the refractive state of the eye as measured by this method closely approximated -0.33 diopterso Results of the oculometric measurements Are presented in Table I, We note that one observer obtained negligible refractive errors with corrections. The three observers used the indicated corrections during-the experiments. Although the method described can be used to give a measure of the astigmatic error, it was not considered sufficiently sensitive for specification and correction of the cylindrical error noted in one observer. Accordingly, the method of Tait', using a Stenopaic slit was employed in this instance, This method involves determining the reft'active error in each of several meridians by a subjective technique. The observer reports on the least positive correction which gives clear imagery in each meridian and the astigmatic error is computed from the difference between the most differentnce between the momert different merans I 11

The University of Michigan * Engineering Research Institute 2455-9-F III, RESULTS Threshold values resulting from the probit analysis of the experimental proportions are available for each observer for each of the two targetso The data for the 1 minute target are presented in Tables II, III, IV, and V; data for the 45 minute target are presented in Tables VI, VII, VIII, and IX. In each case, the symbols have the following meanings. Values of A~B represent target luminances for which pI =.50o Values of aAOB represent the standard errcrs of the values of AB,0 Values of a represent the standard deviations of the normal frequency distributions upon which the ogives were basedo 'Values of a represent standard errors of the values of a. Values of X2 have the usual statistical meaning,, as do values of dofoo Values of p(x2) represent the probabilit-Las that values of X2 as large as those obtained, or larger, could have occurred by chance. Values of VR, the "variability ratios", are obtained by dividing a by AeBo Values of aVR represent standard errors of these quotients0 The data represent in all more than 38 000 observations0 The threshold values are presented in Figure 3 - 10 as functions of log background luminanceo There is a separate graph for each observeris data obtained with each target size. An horizontal line has been used in each figure to fit the experimental data for the smallest values of background l.uminance and for zero, Fit of the data to these lines represents the fact that AL-B is equivalent for zero and for various small values of background luminance0, Such a relation means that there is no photosensitization by white light0 It is apparent that the eight sets of data fit the horizontal lines adequately over a considerable range of values of background luminance0 Hence, there is considerable evidence against the postulation of a photosensitization effect0 Values of a/A~B or VR are -presented in Figures 11 and 12R as a function of log background luminance. A separate graph is presented for each target sized representing the averages of the data from the four observerso An horizontal line has been drawn in on Figures 11 and 12 to facilitate evaluation of the extent to which a/AOB has the same value at all values of background luminanceo The data points are quite scattered and it is difficult to detect a trend with respect to background ltuminance There $is a suspicion that the value of a/AOB decreases at the highest background luminance levels, The hypothesis that a/AOB does not depend upon the background luminaxnce was subjected t9o statistical test by means of a X2 test based on measures weighted for reliability, The results of this test are presented in Table X, The probability of X2 values in the table represent the expectation that a value of X2 this large or larger will occur by chance alone0 It may be seen that, with two or possibly three exceptions, the data may mot be' represented as arising from a single universe. On a purely statistical baesis, then, we must reject the hypothesis that a/A B is constanto I -— i 12

The University of Michigan ~ Engineering Research Institute - 2455-9-F It must be added that the test applied in this case is not entirely fair to the hypothesis; it represents an extremely conservative estimate of the probability, since the deviations of ao/a B from their central value may presumably vary due to day-to-day variability of threshold datao The fact remains that values of a/AOB are remarkably constant as background luminance varies over a considerable range, Values of the ratio also appear to be equivalent for the two target sizes, This fact confirms the result reported elsewhere (Ref0 10) by one of us. The apparent constancy of VR is of considerable practical and theoretical significance o We may inquire to what extent the probit analysis was justified by the goodness of fit of the data, Data for each observer were tested for the goodness of fit of each set of data to its normal ogive, using the X2 test of conformity. These values of X2 appear summed at the bottom of the individual data tables, The probabilities of values of X2 this large or larger for each of the eight sets of data are as follows: *19, o59, 47, o023, o075 J19,g 87, o074, Considering the severity of this test, we may consider the agreement of experimental data to theoretical ogives to be excellent and the probit analysis to be entirely justified. 13

Tlr II. 1.....:,. AA_~ 1.L: __ C. i; aftDL 11 IP 'a 1o -% % S^ L I;+ Ine university or IviIcnrgan * cllleerlilg K adrcn[ IIns51tute 2455-9-F IV, DISCUSSION The main experimental data of the present study have revealed that the threshold luminance increment is equivalent for total darkness and for a variety of low background luminance levels Thus$ there is no evidence that the human fovea' can be photosensitized by small amounts of white lights There is no conceivable basis for expecting to improve foveal vision on dark nights by the addition of a small amount of veiling white light covering the entire retinae. These results also suggest that there is absolutely no basis for the usual assumption that the "absolute threshold" differs qualitatively from the "difference threshold",,. The eye operates as a detector in the same manner when the background is totally dark as when it has a very low luminance. Furthermore, the threshold luminance difference increases as background luminance is increased in a smooth and continuous manner, This result will prove useful in the development of a general theory of visual detection, i

The University of Michigan T Engineering Research Institute 2455-9-F REFERENCES 1 Blackwell, Ho R,, The inter-relations of contrast, area, and adaptation brightness in human binocular vision, PhD, dissertation,' University of Michigan, 1947, 2 Crozier, W, J., Respired oxygen and visual photosensitization, Minutes and Proceedings of the Armed Forces - NRC Vision CowmiG tee, 27th Meeting, C10e-11 November 1950, 3 Graham, C o H and Kemp,, E, H. Brightness discrimination as a function of the duration of the increment in intensity, Jo Gen. Physiol, 21, 635-650 (1938), 1. Blackwell, H, R,, Psychophysical tlhresholds: Experimental studies of methods of measurement, University of Michigan, Engng, Res, Bull, No. 36, p'i 227 (1953). Blackwell, H. R,, Pritchard, B. S., and Ohmart, J. G., Automatic apparatus for stimulus presentation and recording in visual threshold experiments, J. Opt. So_. Amer,, 4,4 322:;326 (1954)o 6 Kincaid, W, M. and Blackwell, H. R., Application of probit analysis to psychophysical data Io Techniques for desk computation, University of Michigan, Engineering Research Inst, Report 2144-283-T (in press). 7 Finney, D J., Probit Analysis, Cambridge: Cambridge University Press, 1947%- - po 256o 8 Kincaid, W. MN and Blackwell, H, R, Irregularities of retinal illumination resulting from an artificial pupil, (Abstract), J, Ar 42, 872 (1952). 9 Ogle, K. N,, Private communication 10 Blackwell, H0 R,, Brightness discrimination data for the specification of quantity of illumination, Illum Ens. N0Y. 47, 602609 (1952) 15

The University of Michigan * Engineering Research Institute 2455-9-F TABLE I Refractive Data from the Oculometer (Diopters ) Observer AK OTL HF LP Uncorrected Error -o 66 4. 46 -0.76 Corrected Error -0o33 -0.37 -0 31 Strength of,Op ha. Th..i c- Lens -0 375 -4.oo - 4o375-0.75 at 1650 0 e375-1;1- 075 at 165 16

_ _I ~ _I_~~~~~~~~~~~~~~~~ t TABLE II Threshold Data for Observer OTL;. a I minute Ordrer B (ft,-,Lo) AB aAB a a df. P(X2) VR aVR 1 1o0o5 406, 22, 1 125 o 36.2 1 09 3 078.307.0839 2.908 182, 11.3 85.5 14,4 2.78 3.42.469 oo0906 3.0972 171. 11,3 79,3 12,6 4.45 3.22.464.0892 4 1.12 x 10-2 122, 7.27 53~9 9024 2.99 3.39 ~440.0855 5 1,27 x 10 36 b 8.34 56j7 9 06 4 70 3.20.418 o0811 6.921 x I0n5 131, 8.19 54.9 8.70 6o35 3.094.420.0800 7.977 x 10o6 125, 7.99 52.3 8.31 2.67 3.46.418.0800 8 0.0 141. 9 33 73.1 15o4.865 3.83.520.111 9 2 62 x 10=6 130, 5.92 38.0 6,77 7.89 3.050 o291.0588 10 2.86 x 10-5 113, 7513 53,6 8.84 2.53 3.49.475.0911 11 3.00 x 10o4 115. 18,7 62.5 9.60 2 41 3.49.543.106 12 2.57 x 10-3 152, 10Q2 77 0 13,1 1o75 3.64.508.101 13 1o65 x 10-2 153o 8,93 68,4 11.5 5036 3.14.446.0835 14.295 132, 9c26 64,1 14.0 5.15 3.16.487 0e941 15 3.17 236, 12,4 84 9 14,7 1 79 3 o61.359.0701 16 30 2 556. 27.2 183, 33.3 6.28 3.10.330 x0659 17 301. 2320 o 516o 1270. 20400 5. 06 3 o17.549.108 18 1210. 5400, 3730. 2510, 390. 6,07 3.11.465.0893 19 1019 847, 62.2 403. 63.1 10.4 _.o019 476.0940 Total 80,4 57 p(X2) o,23 -4:3 m m -I _. 0 C: 3 a Q) P* Report 2455-9-F

TABLE III Threshold Data for Observer AK; a = I minute Order B (ftoLo) ~A cA B; ara X2 dof p(X2) VR c(VR 1 1 0,5 292. 10 8 57 0 1 ol 2,30 3.51.511.0409 2 1,67 159o 12o1 68.9 10o9 1,35 3.73.432.0818 3.0932 91.4 6o06 41,7 6,53 3031 3.35.459.0871 4 1o07 x 10=2 73~5 5,21 32,3 5 20 >318 3 o96 o440 o0875 5 1.37 x 10=3 87~4 6.42 47.2 9 79 6.33 3.097.540.103 6 1.20 x 10-4 97.7 6~35 47.5 7.76 2.36 3.49 o486 o0933 7.915 x t0-5 98 3 4,97 30,1 5.19 4 88 3.19.307 I0603 8.814 x 10-6 90o2 5~38 39,0 6,46 4,23 3,24.433.0827 9 0,0 102, 6,.23 45 6 9.09 s583 3.90 ~448 ~0992 10 3.61 x o106 8404 5.45 38.3 6005 1.91 3.59 p454.0863 11 3.09 x 10o5 95 5 5 36 36,6 6o05 5 31 3 o16.383.0731 CC) 12 3.00 x 1084 84.2 5o80 42 8 6,,74 4.66 3.20.508.0972 13 2 58 x 10-3 104. 4.o39 26.5 4 o98 1 o66 3 o64.255.0522 14 1.65 x i0o2 87o9 7.20 48.4 9.33 9.26 3.027.550.a27 15.295 902 5,10 29,5 5,12 1,55 3.68 327.0663 16 3 17 160. 7,62 57.5 12.0 2.80 3 42 o361.0706 17 30 o2 432 16o8 97.3 20o8 4 55 3.20.225.0505 18 301. 2050 o 77.3 434, 84 o5 2.60 3.46 o212 o04.42 19 1210 4660. 259, 1800o 304, 4.89 3 19.387.0744 20 101 801o, 46.8 338~ 57-3 5.16 1 6 ~423 0817 Total 70.0 60 p(X2)-.19 "9:am_ o m CgQ:s 0 s #A Am m1 Report 2455-9-F

TABLE IV Threshold Data for Observer HF; a - minute Order 1 2 3 4 5 6 7 8 9 10 Io 11 12 3 14 15 16 17 18 19 20 B (f t.L.) 10.8.908.0972 00112 o00127 1.44 x io"4 ~921 x 105 0977 x lo06 0.0 2063 x 10o6 3.09 x 10o5 3.00 x 1034 2.82 x Io03 165 x 10-2.295 3817 30 2 301 1210, 101 o..... 393 1430 119. 127, 1370 1140 1150 1400 99 6 138, 1190 1050 116. 9906 9605 2050 503 o 20000 5890 8030 cA B3 20 0 9,71 7.08 9.20 5 89 6.41 5,72 9,70 5.18 1003 6.80 5.74 5~99 7.38 7o07 13i8 26 o lo6~ 3230 36 o1 a 137. 66.7 46.7 46.7 33.6 45.9 3882 6o.6 60,6F 32 o0 65 3 47.3 41,2 41.0 46.8 40 o3 1o6. 190o 4780 1930o 171. d.f, P(X2) VR aVR 16,3 10.3 7 o53 14.5 7.11 7.90 6.86 9.69 5 050 10 4 7.82 7.30 7.30 7.41 68~9 17.5 30 1 96o9 3320 33.1 Total 4.19 3.63 1 64 1 99 2,24.511 1.55 1.27 3 50 19 o0 271 1 51 o627 o831 5.13 2,87 1.02 ~0622 5,o09 57.6 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3.24 Y 31.66.59.51.91.68.73 33.0003.960.68 o90.85 17.42.80 099.17.87 349.468.393.367.246 401o.332.434.321 474 0399.391.352 0470.418.519.345.239.327 o213.0695.0892.0748.101 00546 0780.0660.0858 0o632.0950.0762 0791.0697.0935 0o874.100 0O674.0563.0655,0455 _.. -N i. < m, I's 1-< 0 i. 3o m 3 r11 so go -! _. 3 -1 3 I;% m~ IV _.n 60 p (X2) 59 Report 2455-9-F

TABLE V Threshold Data for Observer LP; a 1 minute Order B (ft,LL,) &'B cA~B a a X2 dfo P( X2) VR aVR I 10 o5 222, 10 o9 75 6 15.2 3o61 3.31 e340.0725 2 1.67 1l15 9.70 42.3 7.14 8,67 3.03 367.0706 3.0932 54.4 4.31 31 6 4.90 1.27 3.73.582.115 4 1.12 x 102 62 2.456 28,7 4,48 6.08 3.11.462.0875 5 1.37 x 10o3 75.8 4.1o 29.2 5 15 3a89 3 026.386 o0759 6 1L20 x 10 4 73.2 4o49 34.9 6.37 2,80 3.42 476.0954 7.921 x 10i5 59.6 5.1 32.8 5~23 161 3.66.925 115 8.969 x 10-6 71,6 4o13 27.6 4~50 2.21 3.53.386.0738 9 0,0 73 6 4 47 32,2 5 24,804 3.85.437.Q833 10 3.61 x lo06 60,2 4.48 30,6 4~87 1,31 3 o71.508.104 11 309 x 10~5 75 4 3 93 24,5 4a19 1,43 3 o71.324,0636 12 3,07 x 10-4 75 1 4 13 25~6 4.30 1,92 3.59.341 ~0668 13 2.82 x 1o03 59,9 4.45 313 4,79.709 3 087.523.101 14 1o65 x 10-2 49,0 4,84 27.8 5.55 8,69 3 o04 567.148 15.295 60.5 5 29 32,5 5,33 1 o02 3.80 o538.115 16 3,17 111. 6,30 33.9 6o08 3o18 3.36.304 o0634 17 30.2 340. 1805 131, 230o 1 30 3.73.386.0754 18 301 1470 214 380, 65 7 6 40 3 o 11 259 0e506 19 1200a 3990. 217o 1 310 2230 3,46 3.33.328 o0652 |20 2t101 J 622. 31,6 150. 30 O j.68.241.0548 Total 61 9 60 p(x2)=.47 e' m~ < m~ O m, m~ $ m m~ m~ C0 r+ O. 3 nn 7 go m.:: _. MIa go s _. m O~ mP Report 2455-9-F - --- -- ----- - ---

-4 Q1 m (r+ l< TABLE VI - - I- - 0~~~i Threshold Data for Observer OTL; cz 45 minutes order B (fto-L. ) a'OB aA'B a a Xa d,f. P(x2) VR aVR 1 040 o588.055.418.0683 1.02 3.80.491.0911 2 1,11 x 10o6.632.0509 *313 o0550 1.87 3 Q59.302.0599 3 1,21 x 10-5.682 o0522.376 o0652 6.o04 3.ol.380.0731 4 1,13 x 10'4.646.0642.488.0845 9.57 3,022.464.0907 " 5 2.32 x 10-4.532.0579.413 x0644 5.65 3.13.481.0931 6 10,1.933.232 1 o35.380 1.35 3 ~99.505.122 7.961.851.0734 e484,133 8.60 3.035 o390 a102 8.985 x 10-1 o821.0573.403.0919 4,45 3.22.391,0881 9.942 x 10 2 Q933.0486.349.0687 1.45 3.68.375.0780 10.981 x 10o3 ~903 o0473 o341 o0677 5 54 3 13.377.0791 11 104. 10o 7 o543 3.61.621 3,44 3.33.377.0659 12 1210. 83.4 4 D79 33.7 5.59 1.60 _.66,405.0773 Total 490 4 36 p(x2).074 0 4% nI;:r io 0) 0f rl co m( m n.Q OO m P+ VW+ ]6 rql CD Report 245-59-F ----------— L --- — --- —

TABLE VII Threshold Data for Observer AK; a 45 minutes Order 2 3 4 w 5 fO 6 7 8 9 10 11 12 B (ft.~L.) 0o0 1.22 x 10'6 1,14 x iO5 1o11 x 10l4 2,32 x 10=4 10.0.961.985 x l0o1.942 x 10o2.981 x 1io3 104. 1210, A'B.737,814.675.669.682 2,66.886 ~789 o528.524 12.4 120 o OA'B.0511.0387,o465.0513.0342.151.0583.0398.0291.00452.456 5,72.322.256.263.419.241 1.03,451.250.184.323 2.53 3540.0513.o466.0432.0719 0o44o.170.0757,0432.0307.0496.487 9,09 1.31 4 20 14,9 T7.79 3o15 3.50.463 1.27 5 o51 1,92 I o06 '~r& df, 3 3 3 3 3 3 3 3 3 3 3 36 p(X2) VR aVR 0 w 00 lw.73.24 o0018.050.36.33.92.73.14.59.78.24.437.315 0390 ~626.353.389 ~509.317.349.616.204.283 o0858.0630.0772.128.0700.0743.0992 0o626 ~o677.126.00420.0739 Total 49o2 p(X2 ),,075 m m~ M 3" 9:1 M _. f_ 0) 04 OO P* _m. C IP Report 2455-9-F

TABLE VIII Threshold Data for Observer HF; a 145 minutes Order B (ft.-L.) AB aAB a X2 d.fo p(x2) VR GVR i 0O0.551.0349.228.0354 5.21 3.16.396.0773 2 1.22 x 10'6.564.0394.280.0433.495 3.92.497.0950 3 1,11 x 10'5,490.0279,195.0323 3,77 3.28.398.0762 4 1.35 x 10 4.464.0340.275.0472 5.24 3.16;593 o120 ro 5 2 78 x 10=4.570.0328 ~247.0143.822 3.85.434.0866 " 6 10,0 2.64,138.860.147 1.,42 3.71.326.0638 7,961.842.0587.456.0750.999 3,80,541.106 8 o985 x 10j1 o956.0541.392.0796 8.15 3.042.4110 o864 9,9412 x 10o2.696.0455.330 o0521 2.66 3.46.437.0899 10 1,04 x 10o3.696.0439.304,0481 3,17 3.36.437.103 11 10o4 10,2.597 4.34.729 9,27 3.027.428,0822 12 1210. 82.8 5.47 42,7 7.23 2j ~5.146.516 o101 Total 43,7 36 p(X2).19 In M 0 -4K 0 rm 00 Sb w m In 00 tA rp p. p cn =0 omu Mr e, a" m Report 24 5-9-F - ~ ~ -~ -- I I~

1 TABLE IX Threshold Data for Observer LP; cz 45 minutes Order B (ft.-La) AB aA~B or df, p(x2) VR aVR Yb a -4' Ir r* 0 S% 0) 0 1 2 3 4 5 6 ~- b7 8 9 10 11 12 13 0,0 1.02 x 106.95 x 1oQ5 1.12 x 104 2.~78 x 104 10.1.961.985 x l0ll.942 x 10-2.981 x Io- 3 1o04. 1210 o 1,36 x 10-5.369.405.506.451.424 1.91.520.406.331.443 9.09 84 8.525.0275.0325.0304 ~0303.0330.103.043o.0281.0331.0356.427 49,g.0303.186.240.217.221,252.719.259.130.220.214 2.16 34.5,226.0294.o370.0353.0348.0397.126.0429.0299.0347.0353 o437 5.94.0394.397 14o6 1,80 1.47 1.19 3,00.676.327 3 18 1.23 3 37 6.25 2.66 3 3 3 3 3 3 3 3 3 3 3 3 _di_.94.0030.61.68.75.39.87.96.36.75.33.10 ~46.494 *593.428.491.594.376.498.320.665.483.238.407.430.100.118 o8in4.0934.0736.106.0421.148.0195.08o4.0835 Total 27.0 P(x2) =.87 39 rm IK F mb m m WQ u, 0) r+ P+ m Report 2455-9-F

The University of Michigan * Engineering Research Institute 2455-9-F TABLE X Differences in CVR Observer AK AK HF HF LP LP.-OTL OTL Experiment I II I II I II I II 52, 65 o2 27.6 7 30 50 o2 50.9 48~6 6.64 df, 19 11 19 11 19 12 18 11 P(X2).000oo131.000002.o834.799 00oo131.000001.000075 0799 25

Fig. 1. Artist's conception of the basic psychophysics test facility. Ie Fig. 2. Optical schematic drawing of the target projector. 26

4' I 4 iE. I 2 ' 50% FORCED CGOIOE DETECTION4 of! I MWrE OBSERVER OTL o o 9 0 -- I,o i -, -; I &A.m..6 - v.9 -4 -3 *2 r - -LOG -GROND LMI2 -I 0 LOG BACKGROUND LUWIAN~E (FOOT-LAM9ERTS) I 2 3 Fig. 35. Threshold data from Experiment I. GA 1212 41: I MINUTE OBSERVER A K i I I I 50% FORICID CHOICE. oETETIos: a~ a o- ~ 9 0 -. 1.,.. 0~~~~ -6 I.G -4 4 -2 I 0 LOG UACKGROUNP IUMIAN (F9OT-rLAMBERSTS) B 3 Fig. 4. Threshold data from Experiment I., 9 n~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

d I MINUTE OBSERVER F ~, I a 9I 0 9 LOG BACKGROUND LUMINANCE (FOOT-LAMBERTS) Fig. 5. Threshold data from Experiment I. 4, I Ia Q 4 3i I OA 1Sl d a I MINUTE OeSERVER LP 50% FORGED CHOICE DETECTIONS o J 0 0 * 0 0 9* I -_ * L 0 k -- A..m I.I I~ -e, ~ - 0 ~ i. ~.4 ~-.4 -.. -I 0O I A LOG ACKGROUN LUMINANCE FOOT-LAMBERTS) Fig. 6. Threshold data from Experiment I. S _J — a 28

CA 1210 a[ 50% FORCED CHOICE DETECTIONS at 45 MINUTES OBSERVER OTL.- -5 -4 -3 -4 - I 0 - LOG BACKGROUND LUMINANCE (FOOT-LAMBERTS) CA lt07 &I I I 2, a 50% FORCED CHOICE DETECTION8 o 45 MINUTES OBSERVER AK o I -I0 " -4 D'3 4 -4 0 LOG BACKGROUND LUMINANCE (FOOT-LAMBERTS) Fig. 8. Threshold data from Experiment II. 29

" I UW i I -JO I a 46 MINUTES. 00% FORCED ONGE KTGTIOTS * ~ -A, -A 1. t. -.~~~~~~~ -1. 5- ' -!4:e LOG BCKID LUMINANCE (FOOT-LAMLeRTS) I i, Fig. 9. Threshold data from Experiment II. GA 120l I 2 I % I I %.9 a 01 50% FORCED CHOICE DETECTIONS X 45 MINUTES EAVER L P/ -. 0 I a e, -I,. a _ t-,.. I.. _ m. ' 5 -4 -4 -2 -I O LOG BACKGROUND LUMINANCE (FOOT-LAMBERTS) I a I,, I e s Fig. 10. Threshold data from Experiment II. 30

GA toi ~ I MINUTE a.0.5 9 ~ 4 ~~ ~ ~ ~.5 - - a.4.: I'. 0 I 2 5 oK GROUN UINANCE OOT-AMBERTS) Fig. 11. Psychophysical parameters from Experiment I. ca 1205.E.1 i~.' b..,46 MINUTES 4 4.o..... 3.. -3 ACKGOUND LU OOT-I A LOG BAOKGROUND AUINANCEd FOOT-LAIAGTTSW Fig. 12. Psychophysical parameters from Experiment II. 31

UNIVERSITY OF MICHIGAN 30 1 1111111111111111111111111111111111111111 1111114 11111 3 9015 02514 8399