Click for next page ( 320


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 319
APPENDI XES -

OCR for page 319

OCR for page 319
APPENDI X A: TESTING N IGHT VI ST Ob THE MILITARY EXPERIENCE DURING AD FOLLOWING WO=D - R II INTRODUCTI ON Night vision tests have been proposed, produced, anct employed in many places and under many c ircumstances. Their number and var. iety are astonishing. A major source of experience in testing of night vision occurred in the military during and just after World War IT. Since the circumstances that were obtained in the 1940s are unlikely to recur, ana large-scale night vision testing based upon the same goals and criter ia may never again be essayed, the lessons of that t one are summarized in this review. Night vision testing aimed at evaluating large nurr.bers of inaivio- uals and categorizing them according to levels of performance was a development of wartime needs. highs blindness was of course a well- known clinical syndrome, but there were no recognized standards that could be applied to detect mild cases or to compare normal individuals. The safety of a naval vessel could not be left in the trust of a look- out whose night vision was even moderately impaired, and men selected for night reconnaissance flights should certainly be ~r.en whose visual senses were as keen as could be found. What was wanted was a test that would give each man a score or rating that could be referred to when duties called for night observations. Work was carried out in labora- tories connected with various branches of the military service in a number of the warring nations. Dozens of devices were developed, hundreds of papers written, and a complete history would include accounts of parallel attempts to determine the value of the different devices. The knowledge gained, however, can be elucidated with a few c hoice examples. Th is appendix has been excerpted by Jo Ann K 1nney f ram the draf t of a chapter for an unpublished book on night vision prepared by the late Par ker Johnson. 321

OCR for page 319
322 THE TYPES OF TE STS Early emphasis in the military was on the development of ~one~shot, single-score measures of general capabil ity or capacity, for use in military selection or screening. Some envisioned only a rough test to call out the small pe rcentage of personnel who were ser iously night- blind. Others hoped for something more: for a test that would grade men on some sort of scale . Appropr late night-vision standards could then be set for var ious military duties. The task was approached with the hope that night vision would be a stable, unitary trail-quickly and reliably measurable, and valid as a predictor of any assignable night vision duty. It was assumed that the test could be administered, requiring only a modest expenditure of time and effort, somewhere in the induction and training process. The score would then be inscribed in the inductee's record, along with his height and weight, to be ready as needed to predict night performance capabilities. As we shall see, however, it never worked out that simply. The tests that were developed may be divided, by and large, into two categories: tests of light sensitivity and forms perception tests. In addition, there were a number of field tests made; examples of each of these will be g iven. Tests of Light Sensitivity The course of dark adaptation had been the object of study for some time; the Gullstrand Photoptometer of 1905 and the bagel Adaptometer (1907) were two early instruments. As research continued, showing the two segments, rod and cone, of the dark adaptation curve, the word "adaptometer" came to be assoc iatea with the measurement of dark-adapted thresholds for light. Because adaptation slows and ceases after about 30 minutes in the dark, it seemed that the f inal, steady level might be taken as representative of the indi~ridual's night vision capability. In fact, the individual who totally lacks scotopic vision is readily identif fed by such tests. He needs more than 100 times as much light to see as does the normal, a degree of difference that is difficult to overlook. The seriously night-blina may thus be screened f rom a popu- lation with even the least sophisticated of devices. To discriminate meaningfully among members of the large, n nor~r,al" population is much harde r. The Hecht-Shlaer Adaptometer The instrument, originally used for research, appeared suitable for screening service personnel and was offered to Britain's Flying Personnel Research Committee in 1941. The apparatus u see a circular test area with a diameter of 3 degrees, exposed periodically, at a point 7 degrees above the fovea. A dim red light was provicea for f ixation. The light was violet to ensure the least possible photopic involvement. Wratten neutral density f liters and a wedge of f liters

OCR for page 319
323 were used to effect changes in luminance. The luminance of the test light was gradually reduced, the observer simply stated when he could see the light, and the time noted. The Hecht-Shlaer Adaptometer was widely used and numerous studies prov ide norms and data on its reliability and validity (Berry, 19491. Var ious modif ications were made to it, and in the end it became a Family of devices" rather than a single instrument. Test-retest correlations of various models, in reported studies that u sea 100 or more men in the sample tested, ranged from .42 to .78 with a median value of .64, levels not overly encouraging for mass testing. Tests Using More Complex Forms Some of the test developers felt that simple light sensitivity might not be well correlated with successful night observation and favored tests involving some deg ree of form perception. Thus the A.R.L. Adaptometer of Britain's Royal Navy required location of the position of a 45-degree, wedge-shaped, dark sector missing from an otherwise circular lighted area; the test was given after a 30-minute per iod of afar k adaptation. The S.A A. (School of Aviation Medicine) N ight Vision Tester rem quired an even greater degree of form perception since it presented a Landolt ring. Other tests similarly employed a T in different posi- tions. In each case, the figure remained the same but the or ientation was different. The luminance level was systematically droppea through a succession of fixed steps and the " score" determined by either the lowest level with which an individual could cope or by the total number of correct position identif ications for the whole series. Early ver- sions of the Army N ight Vision Tester were similar in design. Outdoor Courses Another sort of test that had some appeal was the outdoor observa- tion course, in which men were conducted over a prescribed course under natural light conditions, locating and identifying as many as possible f rom an assortment of real ob jects that lay along the way. But since one cannot hold constant the deg ree of darkness desired, such testing was not capable of being standardized for routine categor ization of men. Such tests, however , had a g reat deal of ~ face-validity.: they appeared to be sampling exactly the sort of skill required. Whenever c ircumstances smi led on an outdoor range and a number of men were run through the course under what seemed to be reasonably constant cond i- tions of lighting, motivation, and experience, their scores were eager- ly tabulated and compared with scores by the same men on the more pre- cisely controlled indoor tests, tests that had them doing more artif i- cial things like detecting dim lights, determining the orientation of standard f arms, or reading letters.

OCR for page 319
324 PROBLEMS OF TESTING NIGHT VI SION A few examples will be given here to illustrate the problems encoun- tered during attempts to test night vision during World War II. Both indoor tests and the exper fences in an outdoor range are described. The Livingstone Rotating Hexagon Night Vision Test 4 This test was used by the Royal Air Force, the Royal Australian Air Force, and the Royal Canadian Air Force in the early 1940s. In an effort to save time in the testing procedure, the apparatus was built with six faces and could be administered to six men- at a time using one machine and one technician. The task was one of recognition of letters and of familiar forms such as styli zed sh ips and airplane silhouettes. Several n night" luminance levels were employed; testing was started at the higher luminance ~ .005 foot Lamberts) and progressed in steps down to .0003 ft.L. The score was simply the number of correct recognitions out of a possible 32. Instruction was given in the use of of f-center vision and sub jects were allowed to scan f reely dur ing each exposure. Setting aside questions of reliability and validity ~ in which the Hexagon, adequately maintained and administered, revealed neither 9 reat super for ity nor infe rior ity), a wide gamut of problems were encounterea in its use which are inst ructive for night vision testing in general. First, engineering to ensure adequate control of the scotopic lumi- nance of the light stimulus is both essential and difficult. Varia- tions in amount of light f rom one face of the Hexagon to another, f roar, one instrument to another, and f ram one time to another occurred due to aging, voltage fluctuations, homogeneity of lamp output, dirt, damage, etc. Both the or iginal setting of the luminance and corrective ad just- ments require considerable technical skill. Changes in the voltage, for example, change the color temperature of the light which in turn can produce marked deviations in the scotopic luminance. The wide range of nighttime luminances requires adjustment over many log units of inten- sity, again requiring considerable skill in the means of adjusting ano maintaining such a range. Second, not only must the instrument be standardized, but also the testing procedures. There are several levels of concern here: (1) to see that both the administrator of the test, who may be only a techni- cal assistant, and the people being tested follow instructions to the best of their ability; (2) to demand that these instructions be clear and comprehensible to preclude lack of understanding as a reason for individual variation; and (3) to ascertain that the auxiliary para- phernalia, including the arrangements for communicating responses, not require irrelevant skills. The various forms of noncooperativeness--aisinterest, disdain, malingering, cheating--cannot be ignored and the test must be designed with them in mind. In the dark, men could and aid shorten the viewing distance between them and the target to improve their chances. For res- ponding, each man was provided with a clipboard and answer sheet aividea into areas for response that had to be located, in the dark, by running

OCR for page 319
325 one's finger over raised guides. His responses were to be written in the proper area, but writing in the dark is a sadly neglected skill in our society and those who had to score the sheets later found their powers of judgment and ingenuity strained. Another procedural problem is the amount of learning or practice involved in taking the test. The Hexagon, with its unfamiliar scanning procedure, showed signif icant learning over the f irst trials, with some individuals catching on at a more rapid rate than others. Third, a great deal can be learned by considering the scoring prom cess used in the Hexagon and its meaning. The very decision to Reek a single score for night vision performance reflects a belief either: (1) that night vision is a unitary skill or ability, or (2) that, though distinguishable components may exist and vary independently, the test draws on a sufficient number of these to be representative of per- formance in real-life tasks. The Hexagon, operating over a range of luminances and employing a variety of targets, appeared to be based upon the second assumption. However, it spread people on a scale whose units were obscure; they indicated neither the variation in form per- ception or acuity among observers at one, standard light level nor the amount of light required to perform a standard task. Rather, an arbi- trary composite of the two made up the individual scores. Moreover, when using scores to grade man, there is always the implication that the man who scores 24 is somehow better, even twice as god, as the men who scores 12; in the ca se of the Hexagon, there was no reason to believe this was true. The scoring process may be summer ized as follows : ( 1) Scores that are rationally related to specific performance variables are much more meaningful than arbitrary scores reported on scales that reflect chiefly the vagar ies of the testing procedure; and .~2) unless someone has a sped c if ic reason for emphasizing some other aspect of performance, a score that translates directly in "effective visual range" is probably best. A Range Test for Validation In order to determine whether the scores f roar, a specif ic test related to on-the- job performance, there were many validation. stuaies run in the 194 Os. Most proceeded by setting up an artif icial f ield situation, outdoors, with natural nighttime illumination and employing test targets for detection or recognition that were considered repre- sentative of real military targets. For an adequate comparison among subjects, conditions must be standardized; that is, each man should see the same ob jects against the same backs round, under the same illun~ina- tion. Lack of control of illumination level, cloud cover, and viewing background; differences in the size and configuration of the targets making them of unequal cliff iculty; obtaining data on enough men at once to be meaningful; and controlling differing attention and motivation in the sub jects are just a few of the innumerable cliff iculties in such an endeavor. A validation study run by the Navy in Long Island Sound illustrates the difficulty. In setting up the validity trials, great care had been

OCR for page 319
326 given to maintenance of realistic viewing conditions. The men whose vision was being assessed were lined up along the rails of a large ves- sel, all at the same height above the water. Their job was to watch a specific section of the horizon for the approach of a smaller vessel and to signal when they had it in view. The plan was simple. The target ship would proceed to a position beyond the view of even the best and then make a run toward the lookout ship. All men would be alerted to start looking at the beginning of the run. The target ship would continue toward the lookout ship until it had been seen by all the ~lookouts.. Arrangements were made to record the time and distance scores for each man as he called in his sighting. The target ship would then retreat beyond view and come in again. It was hoped that several such runs would provide performance measures for comparison with scores generated from the same men on a dozen or so night vision tests that had been administered back at the base. It was not easy for a small group of night vision research people to commandeer this much equipment and to get everything together for such a trial. Everything had to go right. It didn't. The night turned out to be one of the blackest of which the Sounc] was capable. On its f irst run in, the target ship approached until its commander stopped the run, fearing he would ram the lookout ship. lio one had seen him at all. Being encouraged to try again, he circled around ano approached, until a military ai rcraft, not associated with the enter- prise, "checked things out" by turning on the beam of an airborne searchlight. Instantly, every lookout being tested signaled n target in view." Such hard luck stories were almost the predictable outcome of trials attempting to incorporate high degrees of naturalness and realism. In examining the material on tests and validations from the World War II literature, Berry (1949) summarized: In short, it appears that the reliability of the tests was not too impressive, varying for the most part from .60 to .80 with occasional unreproduced examples of higher coefficients. They correlated with each other either indifferently or badly . . . the test results did not furnish any satisfactory basis for prediction and there were far too many incongruities, published and unpublished, of service men who n failed" the tests and yet performed altogether adequately. POST-WO=D WAR ~ I DEV~OP~NI S The tests developed during the war gradually fell into disuse, for a variety of reasons, including Berry's {1949) comprehensive assessment, and in the decade that followed only two new tests appeared on the scene (Kinney, 1962~. Both were grounded in the earlier experiences and on additional new research. One of them, developed at the naval Medical Research Laboratory (the NM*L Test) (Kinney et al., 1960), sampled the sub ject's c~ark-adapted sensitivity at various locations throughout his visual f ield. The rationale was that the previous poor

OCR for page 319
327 correlations along tests were due in part to the fact that each tested a different retinal area and that signif leant individual differences in the pattern of retinal sensitivity existed (Sweeney et al., 1959~. Moreover, leaving it up to the sub ject to f ind his own ~ best. area simply added another source of variation. Because area and brightness had previously been shown to be interchangeable (deGroot et al., 1952, 1953) in scotopic sensitivity, target size was varied, rather than tar- get luminance, making control easier. This meant that the scores could be related directly to minimum visual acuity and to visual range. Scores were based upon overall performance and shown to be reliable and normally distributed within the population of normal young Navy men. The idea behind the second new test, the Army Night Seeing Tester (ANST), a device developed within the Personnel Research Branch of the Adjutant General's Office (Uhlaner and Zeidner, 19611, was that testing carried out at higher mesopic levels, at which rod and cone vision were both involved, would be as useful as purely scotopic for the purpose of predicting performance at ~ low light levels.. It thus circumvented the need for time-consuming dark adaptation and did not require specialized dark rooms for testing. Resolution acuity was tested at a light level of about .005 ft.L. Extensive f ield tests were done in which men were scored on their ability to detect Enemies at night, against f told and forest backgrounds (Martinek and Mellinger, 1959~. Though the correla- tions between the indoor and the outdoor tests were not high, they were sufficiently impressive, given the rather meager reliability of the validation tests, to make clear that the ability being measured indoors was significantly involved in the outdoor exercises. At first blush, each of these tests looks like a continuation of the traditional search for a one-shot test, but in rather important ways, each represented a retreat from that goal. The justification offered for the ANST was not that a mesopic test could predict scotopic performance as well as a scotopic test; the argument was simply that for many applications, a good scotopic test takes more time and trouble to administer than it is worth. Admitting that a test carried out at Topic levels was testing a somewhat different function, the juagment was made that these higher, mesopic levels, besides being easier to test, were also useful to know about. Moreover, Kinney's (1968) work demonstrated the lack of correlation among measures of photopic and mesopic acuities ana the NELL scotopic sensitivity- test . The values ~ iven in Table A-1 were based on data f rom 100 healthy young men ~ ithout pathology. Scotopic sensitivity correlates poorly with either photopic or mesopic acuity, but it is obvious that the closer together the light level of the tests, the higher the correlations. LESSONS LEARNED FRO}: THE MI LITARY EXPERIENCE Sen s i t iv i ty and Form Pe rcept ion: One Func t ion or Two? The assumption of one, single Quality of night vision gave way to a m.ultiple-function hypothesis, involving at least sensitivity and form

OCR for page 319
328 TABLE A-1 Correlations Among Photopic Acuity, Mesopic Acuity, and Scotopic Sensitivity Measured on 100 Young Men Test Car relet ions Photopic acuity _ i ~75* ~ At .0009 ft.L. ~I .17 Me Topic ecu ity At .005 ft.L. 1 .37* Scotopic sens it iv ity *Significant at the .01 level Source: Kinney (1968! .15 _ ~ ~ ~.29*. perception, as soon as it was learned that adaptometer scores did not correlate well with low-brightness form perception measures. It was even argued that since sensitivity was presumed to be aided by maxi- miz~ng spatial summation while acuity demanded a fine mosaic of func- t toning receptor areas, the two functions or abilities might be con- sidered fundamentally incompatible. In 1955, Ogilvie et al. found that light thresholds were signif i- cantly and positively related to acuity measures at the very lowest scotopic luminance levels. This was conf irmed by Pirenne et al. (1957) who dete rmined cor relet ions between absol ute threshold and scores determined using various acuity criteria. Acuity was measured using Landolt rings which range from 20/6, 000 (a Landolt ring with a gap size of 295 minutes of arc) to 20/60 (3 min gap). ~ e corresponding light levels ranged from close to absolute threshold to well up in the meso- pic range. m ese correlations, determined for 22 subjects, are given in Table A-2. To be noted in these data are: {1) the remarkable high agreement between sensitivity and acuity measures at the lowest light level ; (2) the fact that the correlation was best at this light level, which argues age inst any incompatibility of the functions being measured at low scotopic levels; and (3 ~ the slow but ultimately precipitous de- cline in cor relation with absolute threshold as the photopic level is approached . These data, together with those of Kinney (1968), suggest that a major determiner of the number of essential functions to be tested is the overall light level, photopic and Scotopic vision being essentially independent.

OCR for page 319
329 TABLE A-2 Correlation Between Absolute Threshold and Acuity at Va rious Levels Gap size, minu tes Threshold log, Correlation mmL Coef f ic tents 295 2.67 .927 191 2.78 .903 96 3.08 .846 24 4. 02 .8 41 4.74 5 95 6.59 Source: Pirenne et al. ( 1957~ . Controlled Fixation Versus Free Movement? The question of whether eye f ixation should be controlled or f ree to roam is an interesting and complex one. The Hecht-Shlaer Adaptometer was typical of most light-threshold devices in providing a fixation point so that the test flash could always be presented to the same area of the fovea. Because of known, gross variation in sensitivity in d if- ferent parts of the retina, failure to provide such control would lead to such var lability as to make the determination of a threshold almost impossible. But having provided a f ixation, we must recognize the arbitrary nature of the choice; the best area for one subject is not necessarily the best for another. The NMRL test, by sampling a number of the regions of the retina, sought to provide an average sensitivity score for the eye as a whole (Kinney et al., 1960) . However, it can still be asked whether this average score is what is wanted; a person skilled in the use of his night vision will make his crucial observations using that part of his retina that is best under the prevailing conditions. And the best part. varies, even for the same individual, depending upon his degree of dark adaptation, on the amount of light available, and on the acuity requirements of the task (Brown, 1954 ~ . Nor can a naive sub ject be trusted to select the optimum peripheral angle for each condition. Thus in choosing a procedure, the tester must look ahead to consi- der the use to be made of the obtained information. Also, in any later

OCR for page 319
330 interpretation of the results, it is important to look back and see what procedure was used . Furthermore, before the tested individual may be asked to use his own judgment, we must be sure he has had experience that will qualify him to make proper use of this freedom. The Role of Knowledge in Determining Performance Testers necessarily try to keep irrelevant factors, as flagging interest, shrewd guesses, series effects, cheating, etc., from influ- encing the test scores. But even as we strive to keep knowledge of the situation from influencing what is intended to be an assessment of visual endowment, we also recognize that in any real situation where night vision is exercised, where the job is to look, to see, and to report or interpret correctly, something more than a good pair of eyes is called for. Two additional factors are knowledge (pertinent to the s i tua t ion) and intell igence . During World War II, British Lancaster bombers regularly invaded the skies over Germany at night. Me tail gunners peered from the rear machine-gun tur ret, alert for closing German f ighter a ire rat t . It was not, however, a situation in which the tail gunner could risk shooting at any faint blur he saw, or thought he saw, since the air around him was filled with other British bombers. If he could count the number of engines in the approaching aircraft, he had what he needed to know; the bombers were large, four-engine machines while the German fighters had one or two engines. but when a plane came into view, approaching from a distance, it was only a vaguely distinguishable blur; there was no time to wait for engines to become distinguishable, since it it was a fighter, a burst of gunfire was imminent. To enable the gunner to reach the earliest possible decision, he was instructed not to look for engines at all, but to concentrate on the general form of the blur; that is, was it a compact spot (a single engine), an oval (twin engines), or a dark horizontal line or bar (four engines)? Not every nighttime task can be reduced to such a simple formula, but in talking of Good vision,. we must always remember that practical vision engages both the mind and the eye to an extent that varies with the task. If we wish to measure capabil ities of the eye, independent of such factors as motivation, experience, or intelligence, we fre- quently f ind ourselves severely challenged. And to the extent that we do succeed , we must recognize that we are ref. ining out factors that may be important for success in other situations. REFERENCES Berry, W. 1949 Review of Wartime Studies of Dark Adaptation, Night Vision Tests, and Related Topics. Armed Forces, National Research Cou nc i 1 Vi s ion Commi ttee . Wash ing ton, D. C .: Nat tonal Academy of Sc fence s.

OCR for page 319
331 Brown, J . L. 1954 Effect of different preadapting luminances on the resolution of visual detail dur ing dark adaptation. Journal of the Optical Society of America 44:48-55. deGroot, S .G., J.~. Dodge, and J.A. Smith 1952 Factors in night vision sensitivity: me effects of bright- ness. Report No. 194. Groton, Canny: Nairal Medical Research Laboratory, Sub Base. deGroot , S .G ., J. M. Dodge , and J .A. Smith 1953 Factors in night vision sensitivity: The interrelation of size, brightness, and location. Report No. 234. Groton, Bonn.: Naval Medical Research Laboratory, Sub Base. Fulton, am., D.G. Marquis, H.T. Perkins, and P.~. Hoff 1945 A Bibliography of Visual Literature 1939-1944, Supplement. Unpublished reports on vision from United Nation as civilian and military sources. Publication No. 11. New Haven, Conn.: Historical Library, Yale Medical Library. K inney, J .A. S. . 1962 Review of literature on night vision testing. Pp. 3-11 in M.A . Wh itcomb, ea., Visual Problems of the Armed Forces. Washington, D.C.: National Research Council. Kinney, J.A.S. 1968 Clinical measurement of night vision. Pp. 139-152 in M.A. Whitcomb, ea., me Measurement of Visual Function. Washington, D.C.: National Research Council. K inney , J .A. S ., E .J . Sweeney , and A . P . Ryan 1960 A net? test of scotopic sensitivity. American Journal of Psychology 73 :4 61-46 7. iSartinek, H., and J.J. Mellinger 1959 Field evaluation of ANST--fixed position observer. Memo No. 59-11. USA TAGO Personnel Res. Br. Res. Mart~nek, H., R.N. Tanck, and J.J. Mellinger 1959 Field evaluation of ANST--armored patrol. Memo No. 59-12. USA TAGO Per sonnet Res. Br. res. Ogilvie, J.C., J.E.B. Ryan, R.F. Cowan, and E.I. Querengesser 1955 Interrelations and reproducibility of absolute light threshold and scotopic acuity. Journal of Applied Physiology 7: 519-522. Pirrene, M.H., F.H.C. Marriott, and E.F. O'Doherty 1957 Individual differences in night~ision sensitivity. Serial No. 294. Medical Research Council Special Report. London: Her Ma j esty' s Stat lone ry Of f ice . Sweeney, E.J., J.A.S. Kinney, and A.P. Ryan 1959 Standardization of a scotopic sensitivity test. Report No. 3 08. Groton, Conn.: Naval Medical Research Laboratory, Sub Base. Uhlaner, J.E., and J. Zeidner 1961 Abe Army Night Seeing Tester--development and use. HERB Technical Research Report No. 1120. Human Factors Research Branch, TAGO, DA.

OCR for page 319