Perceptr~aland Motor Skills, 1377, 48, 771-798. @ Perceptual and Motor Skills 1977
SIGNAL DETECTION ANALYSIS 01: EFFECT OF WHITE NOISE INTENSITY O N SENSITIVITY T O VISUAL FLICKER DAN W. HARPER University of Munjtoba' Sumnzary.-Rating scale estimates of sensitivity to visual flicker were obtained from three subjects under 10 different intensities of auditory stimulation. Results indicated reliable "sawtooth"-like changes in sensiriviry as a function of increasing intensity of white noise. N o systematic and reliable changes were found in estimates of response bias. Theory and future research are discussed with reference to the possible contribution of cortical arousal.
There has long been evidence that the Critical Flicker Frequency (CFF) may be altered by concomitant auditory stimulation (Von Shiller, 1935; Kravkov, 1935; Allen & Schwartz, 1940; Knox, 1945a, 1945b; Ogilvie, 1956a, 1956b; Walker & Sawyer, 1961; Maier, Bevan, & Behar, 1961). Additional evidence suggests that different intensities of the auditory stimulation may differentially affect the CFF (ICruger, 1962; Miller, 1963) although this may not be the case when all of the values sampled are of very high intensity ( McCroskey, 1958). Taken together the parametric assessments of the effect of auditory stimulus intensity suggest an initial rise in CFF values with increased intensity. A peak in CFF is observed after which further increases in the intensity of auditory stimulus appears to reduce the CFF relative to that peak. At very high intensities CFF values typically approach or even fall below no-stimulation baseline values. Such an inverted-U-shaped function suggests an arousal mediator of the sensory interaction. As the intensity of the auditory stimulus increases there may be increased cortical arousal, presumably through the Ascending Reticular Activating System (ARAS). Increased arousal should lead to enhanced visual resolution up to some critical arousal level by "priming the cortex" (e.g., see Lindsley, 1961). Excessive cortical activation would make resolution more difficult by becoming noise and obscuring the signal. Some empirical evidence for the role of the Ascending Reticular Activating System in mediating such changes in temporal acuity is presented by Fuster (1958; Fuster & Uyeda, 1962). Monkeys receiving moderate electrical stimulation to the mid-brain reticular formation demonstrated an enhanced ability co resolve a double flash of light. The present invesrigation addressed questions which remain concerning the effect of intensity of auditory stimulation on the CFF. The first of these 'Department of Psychiatry, 770 Bannatyne Ave., Winnipeg. Manitoba R3E OW3.
792
D. W. HARPER
concerns the levels of auditory stimulus intensities sampled. In previous studies intensities of less than 70 dB (SPL) have not been employed. If an adequate sampling from the lower intensity range was included would the inverted-Ushaped function still be found? Another question concerns the fact that all of the research outlined above has employed estimates of CFF based on classical psychophysical techniques, notably the method of limits. Is it possible that the reported results reflect, ac least in part, response bias systematically associated with the auditory intensities rather than changes in visual temporal acuity? Clark and his associates (Clark, Rutschmann, Link, & Brown, 1963; Clark, 1966; Clark, Brown, 8r Rutschmann, 1967) have shown that the CFF is subject to response bias and that relatively bias-free estimates of ability to resolve a flashing light may be obtained by employing methods based on the Theory of Signal Detectability. The present study was aimed toward gaining better understanding of the effect of intensity of auditory stimulation on visual temporal acuity by employing several auditory intensities representing a wide range of values and by employing the rating-scale signal-detection procedure to obtain simultaneous but independent estimates of sensitivity and response bias.
METHOD Strb jects Three male graduate students were paid for their participation in this research. Each had normal vision when corrected by lenses and normal von Bekesy audiograms. The subjects received extensive training in the rating-scale procedure and enough practice at the cask so that performance in distinguishing "flickering" from "fused" light consistently produced d' values between 0.5 and 1.0. Appamtrts
All testing took place in a double-unit sound-attenuated chamber consisting of a (2.24 m x 3.24 m X 2.44 m high) test room and a (2.05 m X 2.84 m x 1.83 m high) experimental room (Model 1405-aACT, Industrial Acoustics Corp). T h e noise- and flicker-generating apparatus was housed in the test chamber which was separated by a 27.96-cm air space. The stimuli generating apparatus was connected to the test room by a " j a c k panel constructed so as t o preserve the acoustical characteristics of the chamber. The test-room portion of the chamber was of single-wall construction, while the experimental room consisted of a room within 3 room separated by a 9.32-cm air space. The floor of the inner room was floated o n rubber vibration-insulated rails to ensure maximum elimination of structurally borne sounds. Additional characteristics of the experimental room were: rwo 9.32-cm thick sound-attenuated doors, a silent ventilation system, and a sound reduction level of 8 1 d B for frequencies greater than 600 cps. Finally, ambient noise level with the subject i n place was found to be about 45 d B (SPL). T h e visual stimulus consisted of a white light presented monocularly to the preferred eye by a cold cathode modulating lamp (Sylvania, Type R1131c; crater diameter 0.236 mrn) mounted at the rear of a viewing chamber (Lafayette, Model 1 2 0 2 ~ ) . The subject was required to centrally fixate the stimulus as it was presented through a 1.25 m m diameter Plexiglas diffuser. T h e stimulus-to-eye distance was 36.24 cm and the visual
NOISE IN VISUAL FLICKER
793
angle subtended equalled ZO1O', a value assuring full foveal stimulation. The inside of the viewing chamber was lined with dull black material to eliminate reflectance. T h e front of the chamber was constructed of molded rubber which closely fit the subjen's face thus eliminating extraneous light. The flicker-generating apparatus (GrasonStadler, Model EGGG) was set at a light-dark ratio of 0.50 and a lamp luminance during the "on" phase was approximately 35 cd/ma. T h e auditory stimulus was produced by a white noise-generator housed in a von Bekesy audiometer (Grason-Stadler, Model E800). The spectrum produced by this device falls off only above 10,000 Hz. The intensity range of the generator allows for discrete settings from 20 to 120 d B (SPL), adjustable in 5-dB (SPL) steps. The stimulus was presented binaurally by means of one pair of calibrated air-conduction earphones (Grason-Stadler. M X 4 l / A R ) factory calibrated to be used with the above described audiometer. T h e intensity level produced by the audiometer was recalibrated by means of a Vu meter and calibration controls located o n the audiometer control panel before each block of trials. T h e presentation of the stimulus was controlled by two Hunter timers set for simultaneous "on"- and "off"-set and a constant intertrial interval. During testing in the "Headphones and Earplugs" condition a pair of air cushioned earplugs were worn (Willsan "sound silencer," Model EP-100). These earplugs provided an average attenuation of about 3 0 d B (SPL) between 125 and 8.000 Hz. Subjects responded to the stimuli by means of a hand-held "responding device" wired to the audiometer. When the button on this device was depressed, it initiated a loud "click" i n the audiometer in the adjacent room. T h e operation of this device, however, was inaudible to the subject. Testing Procedure Testing was completed for each subject in 10 testing sessions held one a day over a 14-day period. Sessions took place in the morning and began at the same time each day for each subject ( & one h o u r ) . In each test session 500 trials were given as five blocks of 100 trials. Each block was characterized by one of 1 0 auditory conditions: "Headphones and Earplugs." "Headphones." or 40, 50, 60, 70, 80, 90, 100, 110 dB (SPL) of white noise presented via headphones.' The noise levels were quasi-randomly assigned to blocks so that the same level did not appear twice in the same session but did appear a total of five times. A further restriction was that each noise level had to appear ar least every other test day. Blocks were separated by a 2- to 10-min. rest. the exact elapsed time being subjectcontrolled. Time was also provided for the dark adaptation of the subjects before each block of trials. This was necessary as the subject's chamber was dimly lit during the rest but very dark during testing. T o ensure that the subject began testing for each block at about the same level of dark adaptation a small "light leak" between the subject and test room was provided. This " l e a k was such that it was visible only after several minutes in the darkened room. Subjects were individually consistent i n the duration of this dark adaptation period which averaged 5 min. Dark adaptation during testing was. of course, a consistent function of the experimental visual stimulation and the darkened viewing chamber. The visual stimulation in a random 5 0 % of the trials of each block was "fused" cps). O n the other 5 0 % the presentation was "flickering" (presented at the (150 on-off rate determined to reliably produce d' values between 0.5 and 1.0 for each subject during preliminary testing). During the 6-sec. intertrial interval the subject was required to decide his confidence that a flickering light had been presented i n accordance with a 4-point rating scale ("1" indicating certainty or near certainty the light had been
+
794
D.W. HARPER
"fused," "2" that it probably was "fused," "3" that it probably was "flickering," and "4" that he was certain or almost certain that the light presented was "flickering"). The subject was then to depress the response button an appropriate number of times. T h e experimenter used this interval to reset (when necessary) the on-off frequency of the visual stimulus to be presented o n the next trial and to record the response of the subject to the preceding trial.
RESULTS Double probability plots of the Receiver Operating Characteristic (ROC) curves of each subject for each auditory condition were constructed. Examination of these indicated that each was well fit by a straight line but that the slopes varied considerably from unity, tending to take values of less than one. This last attribute was taken as an indication that the underlying distributions were not consistently of equal variance (see Green & Swets, 1966, pp. 96-97) and therefore that the sensitivity index d' and the bias under p were inappropriate. Accordingly, the sensitiviry index d,' (Egan & Clarke, 1966) which avoids assumptions concerning the sizes of variances of the underlying distributions was calculated. McNichol's (1972, pp. 93-96) over-all response-bias index (here called Bh,,) which corrects for any slope deviations from unity in the ROC curve double probability plot was also calculated. In Fig. 1 sensitivity changes from the most quiet baseline condition for each subject are plotted as a function of the intensity of the concomitantly
-
SUBJECTS 1.0
0.8
0.6 Q)
u 0.4
0
'0.C'
-
Y 'B.S.
*--a
'W.T*
-
-
-
0
H~ P
H " 40
50 ~
60" 70 ' 8 0
9"0
100' 110 ~
dB (SPLI-WHITE NOISE FIG. 1 . Flicker sensitivity (d.') as a function o f intensity of auditory condition
795
NOISE IN VISUAL FLICKER
presented noise. The complex functions clearly vary for each subject in terms of degree of changes produced by the various noise conditions. What is remarkably similar, however, is that there is a greac deal of agreement between subjects on the general shape of the curves and on the points where sensitivity is at a peak or low poinr. A Friedman two-way analysis of variance for ranked "white noise" data confirmed the impression of significant variability in sensitivity over the noise conditions (X,.' = 17.2, df = 7, 9 < .02). The visual impression of consistency between subjects was corroborated by an average rank intercorrelation of 0.83.
-
SUBJECTS
t--.'D.C' ' 6 .S' ,---a 'W.T'
01
E
P
0
H P
H
40
50
60
70
80
90
100
110
dB (SPL)
FIG. 2. Response bias (BM.) as a function of intensity of auditory condition
Fig. 2 illustrates changes in response bias due to noise level. Little change in bias is evidenced by two of the subjects whose BAI, values hover around the value of 1. The third subject shows sporadic bias of a conservative nature. A Friedman analysis on these data showed no significant variation over noise level. The observed bias then is best characterized as idiosyncratic.
DISCUSSION The present findings of systematic variation in visual temporal acuity with changes in intensity of a concomitantly presented auditory stimulus invite comparison with previous research. In the studies reviewed in the introduction there appeared to be a peak in sensitivity at moderately intense levels of a tone
796
D. W. HARPER
or noise (77 to 90.5 dB, SPL). Further increases brought a lowering of sensitivity, relative to that peak and perhaps (Miller, 1963) relative to a condition of minimal auditory stimulation. In the present investigation data also yielded a peak in sensitivity at an intermediate value of auditory intensity, 70 dB (SPL). As in the previous research further increases in incensity brought a decrease in sensitivity-but in the presenc results this was true only up to a point. Sensitivity at 100 dB (SPL) was in the present cases again high, while in much of the previous research the downward trend with increased intensity continued. Kruger (1962), however, reported that one of his two subjects showed a slight increase in sensitivity at 96 dB (SPL) from more moderate intensities. Munro (reviewed by Miller, 1963) reported a reversal towards increased sensitivity when over-all auxiliary stimulation was increased. The present research was the first to assess the effect of low levels of intensity of the auditory input on flicker sensitivity, viz., 40, 50, 60 dB (SPL). This "arm" of the curve changes radically the inverted-U or oscillatory function described by previous authors. The data from auditory intensities of 50 through 90 dB (SPL) generates a curve comparable to previous descriptions. What is added are two other peaks at or near the extremes of intensity of white noise input. One such peak was not expected on the basis of previous research ( 100 dB). The other was in an incensity region not previously explored ( 4 0 dB). Previous related studies made no attempt to control for response bias. However, in the present study there was no evidence to indicate any responsebias effects that were systematic and common to the three subjects. These data, therefore, do not add to the understanding of the findings obtained by means of the traditional psychophysical procedures. Their value lies in the fact that any response bias, regardless of its idiosyncracy has not entered into the sensitivity data. Undoubtedly this has contributed to the high level of agreement in sensitivity data among subjects in this study. It is difficult to interpret the present results as supporting the theoretical speculations presented in the introduction. It was noted there that those speculations were based on insufficient and possibly confounded data. The present research has sampled from a wider range of auditory intensities and has demonstrated considerable agreement while employing superior methodology. It seems reasonable, therefore, to examine those theoretical speculations in light of this "best evidence." It is clear chat no simple theory could predict the complex relationship between visual flicker sensitivity and auxiliary auditory intensity observed in the presenc research. The suggestion that the effect is based on simple arousal is not tenable in light of the present resulcs. Of course the presenc findings in no way indicate that arousal is not involved in the sensory interaction observed here.
NOISE IN VISUAL FLICKER
797
At the biological level of explanation the present complex results cause another sort of problem, namely, of what possible use to the organism would be the functional organization of the sensory system reflected in the present data? It may be that there is some advantage to accentuating visual temporal it is not now clear what that adacuity at 40, 70, and 100 dB (SPL)-but vantage might be. It is possible that, rather than reflecting an adaptive mechanism, the interaction (at least in this case) is merely a reflect-ion of the organization of the nervous or information-processing system-an epiphenomenon If sensory interaction is such an epiphenomenon the importance of this type of research is not diminished as it may lead to a better understanding of the organization that does underlie adaptive processes. One line of research which may contribute to theoretical advances concerns the assessment of any arousal component in the present results. Specifically two studies are suggested, both employing the basic procedures of the research reported here. One would optimize the arousal value of the auditory stimulus by totally randomizing presentation of levels of white noise rather than presenting them in blocks of 100 trials. Another would minimize the arousal value of the same stimuli by having the auditory stimulation be a constant background to a block of trials rather than being simultaneously -presented with the visual stimulus. With all other factors held constant across investigations (including the present one), the resultant sensitivity curves could be assessed for the contribution of arousal. It seems that another important line of research would involve assessment of the effect of the "auxiliary stimulus" on the auxiliary modality. A method for assessing the sensitivity of two simultaneously stimulated modzlities has been developed in another context by Eijkman and Vendrick (1965). This elegant procedure also cakes advantage of the methodology of signal detection analysis. REFERENCES ALLEN, F.. & SCHWARTZ, M. The effect of stimulation of the senses of vision, hearing, taste, and smell upon the sensitivity of the organs of vision. Jotma1 o f General Physiology. 1940. 24, 105-121. CLARK,W. C. The psyche in psychophysics: a sensory-decision theory analysis of the effect of instructions o n flicker sensitivity and response bias. Psychological B7illerin. 1966, 65. 358-366. CLARK, W .C., BROWN, J. C.. & RUTSCHMANN,J. Flicker sensitivity and response bias in psychiatric patients and normal subjects. Journal o f Abnormal Psychology, 1967. 72. 35-42. CLARK. W. C., RUTSCHMANN.J., LINK, R.. & BROWN,J. C. Comparison of flickerfusion thresholds obtained by the methods of forced-choice and limits on psychiatric patients. Perceptual and Motor Skills, 1963, 16, 19-30. EGAN,J. P.. & CLARKE, F. R. Psychophysics and signal detection. In J. B. Sidowski (Ed.), Experimental methods and instrr~nzentation in psychology. New York: McGraw-Hill. 1966. Pp. 21 1-246. ErjmAN, E.. & VENDRICK,A. Can a sensory system be specified by its internal noise? Jownal of the Acoirrtical Society o f America, 1965, 37, 1102-1109. FUSTER, J. M. Effects of stimulation of brain stem o n tachistoscopic perception. Science, 1958, 127, 150.
798
D. W. HARPER
FUSTER.J. M., & UYEDA,A. A. Facilitation of tachistoscopic performance by stimulation of midbrain tegmental points in the monkey. Experimental Nerrrology, 1962, 6, 384-406. GREEN, D. M., & SWETS, J. A. Signal detection theory and psychophysics. New York: Wiley, 1966. KNOX, G . W. Investigations of flicker and fusion: 111. T h e effect of auditory stimulation on the visual CFF, ]ournal o f General Psychology, 1945, 33, 139-143. ( a ) KNOX, G . W. Investigations of flicker and fusion: 1V. The effect of auditory flicker o n the pronouncedness of visual flicker. Joursal o f General Psychology, 1945. 33, 145-154. ( b ) KRAVKOV,S. W. Action des excitations auditives sur la frequence critique des papillotemeuts limineux. Acta Ophthnlmologica, 1935, 13, 260-272. KRUGER,S. I. The effects of chlorpromazine on flicker fusion threshold with intersensory stimulation. Unpublished Ph.D. dissertation, Boston Univer., 1962. LTNDSLEY, D. B. Common factors in sensory deprivatio.n, sensory distortion, and sensory overload. In P. Solomon. P. Kubzanskv. P. Leiderman. T. Mendelson. R. Trumbull, & D. Wexler ( ~ d s . ' ) , Sensory deprivation. ~ a k b ; r i d ~ e~:a r v a r dUniver. Press, 1961. Pp. 174-194. MAIER,B.. BEVAN,W., & BEHAR,1. The effect of auditory stimulation upon the critical flicker frequency for different regions of the visual spectrum. American Jorrrnal o f Psychology. 1961, 74, 67-73. MCCROSKEY,R. L. The effect of specified levels of white noise upon flicker fusion frequency. U. S. Naval School of Aviation Medicine, Pensacola. Florida. Joint Research Project, N M 18 OZ 99 Subtask 1, Report No. 80, 1958. M c N r c ~ o L D. , A primer o f signal detection theory. London: Allen & Unwin, 1972. ~ I I L L E RH., L. T h e effects of auditory stimulation o n the critical flicker fusion frequency response. Unpublished Ph.D. dissertation, Univer. of Hawaii, 1963. OGILVIE,J. L. Effect of auditory flutter o n the visual critical flicker frequency. Canadian Journal o f Psychology. 1956, 10, 61-68. ( a ) OGILVIE,J. L. The interaction of auditory flutter and CFF, the effect of brightness. Canadian Iosrnal o f Psychology, 1956, 10, 207-210. ( b ) VON SCHTLLER,P. Interrelation of different senses in perception. British Jorrrnal o f Psychology, 1935, 25, 465-469. WALKER,E. L., & SAWYER.T. M. The interaction between critical flicker frequency and acoustic stimulation. Psychological R e c o ~ d ,1961, 11, 187-193. Accepted March 21, 1979.
SIGNAL DETECTION ANALYSIS 01: EFFECT OF WHITE NOISE INTENSITY O N SENSITIVITY T O VISUAL FLICKER DAN W. HARPER University of Munjtoba' Sumnzary.-Rating scale estimates of sensitivity to visual flicker were obtained from three subjects under 10 different intensities of auditory stimulation. Results indicated reliable "sawtooth"-like changes in sensiriviry as a function of increasing intensity of white noise. N o systematic and reliable changes were found in estimates of response bias. Theory and future research are discussed with reference to the possible contribution of cortical arousal.
There has long been evidence that the Critical Flicker Frequency (CFF) may be altered by concomitant auditory stimulation (Von Shiller, 1935; Kravkov, 1935; Allen & Schwartz, 1940; Knox, 1945a, 1945b; Ogilvie, 1956a, 1956b; Walker & Sawyer, 1961; Maier, Bevan, & Behar, 1961). Additional evidence suggests that different intensities of the auditory stimulation may differentially affect the CFF (ICruger, 1962; Miller, 1963) although this may not be the case when all of the values sampled are of very high intensity ( McCroskey, 1958). Taken together the parametric assessments of the effect of auditory stimulus intensity suggest an initial rise in CFF values with increased intensity. A peak in CFF is observed after which further increases in the intensity of auditory stimulus appears to reduce the CFF relative to that peak. At very high intensities CFF values typically approach or even fall below no-stimulation baseline values. Such an inverted-U-shaped function suggests an arousal mediator of the sensory interaction. As the intensity of the auditory stimulus increases there may be increased cortical arousal, presumably through the Ascending Reticular Activating System (ARAS). Increased arousal should lead to enhanced visual resolution up to some critical arousal level by "priming the cortex" (e.g., see Lindsley, 1961). Excessive cortical activation would make resolution more difficult by becoming noise and obscuring the signal. Some empirical evidence for the role of the Ascending Reticular Activating System in mediating such changes in temporal acuity is presented by Fuster (1958; Fuster & Uyeda, 1962). Monkeys receiving moderate electrical stimulation to the mid-brain reticular formation demonstrated an enhanced ability co resolve a double flash of light. The present invesrigation addressed questions which remain concerning the effect of intensity of auditory stimulation on the CFF. The first of these 'Department of Psychiatry, 770 Bannatyne Ave., Winnipeg. Manitoba R3E OW3.
792
D. W. HARPER
concerns the levels of auditory stimulus intensities sampled. In previous studies intensities of less than 70 dB (SPL) have not been employed. If an adequate sampling from the lower intensity range was included would the inverted-Ushaped function still be found? Another question concerns the fact that all of the research outlined above has employed estimates of CFF based on classical psychophysical techniques, notably the method of limits. Is it possible that the reported results reflect, ac least in part, response bias systematically associated with the auditory intensities rather than changes in visual temporal acuity? Clark and his associates (Clark, Rutschmann, Link, & Brown, 1963; Clark, 1966; Clark, Brown, 8r Rutschmann, 1967) have shown that the CFF is subject to response bias and that relatively bias-free estimates of ability to resolve a flashing light may be obtained by employing methods based on the Theory of Signal Detectability. The present study was aimed toward gaining better understanding of the effect of intensity of auditory stimulation on visual temporal acuity by employing several auditory intensities representing a wide range of values and by employing the rating-scale signal-detection procedure to obtain simultaneous but independent estimates of sensitivity and response bias.
METHOD Strb jects Three male graduate students were paid for their participation in this research. Each had normal vision when corrected by lenses and normal von Bekesy audiograms. The subjects received extensive training in the rating-scale procedure and enough practice at the cask so that performance in distinguishing "flickering" from "fused" light consistently produced d' values between 0.5 and 1.0. Appamtrts
All testing took place in a double-unit sound-attenuated chamber consisting of a (2.24 m x 3.24 m X 2.44 m high) test room and a (2.05 m X 2.84 m x 1.83 m high) experimental room (Model 1405-aACT, Industrial Acoustics Corp). T h e noise- and flicker-generating apparatus was housed in the test chamber which was separated by a 27.96-cm air space. The stimuli generating apparatus was connected to the test room by a " j a c k panel constructed so as t o preserve the acoustical characteristics of the chamber. The test-room portion of the chamber was of single-wall construction, while the experimental room consisted of a room within 3 room separated by a 9.32-cm air space. The floor of the inner room was floated o n rubber vibration-insulated rails to ensure maximum elimination of structurally borne sounds. Additional characteristics of the experimental room were: rwo 9.32-cm thick sound-attenuated doors, a silent ventilation system, and a sound reduction level of 8 1 d B for frequencies greater than 600 cps. Finally, ambient noise level with the subject i n place was found to be about 45 d B (SPL). T h e visual stimulus consisted of a white light presented monocularly to the preferred eye by a cold cathode modulating lamp (Sylvania, Type R1131c; crater diameter 0.236 mrn) mounted at the rear of a viewing chamber (Lafayette, Model 1 2 0 2 ~ ) . The subject was required to centrally fixate the stimulus as it was presented through a 1.25 m m diameter Plexiglas diffuser. T h e stimulus-to-eye distance was 36.24 cm and the visual
NOISE IN VISUAL FLICKER
793
angle subtended equalled ZO1O', a value assuring full foveal stimulation. The inside of the viewing chamber was lined with dull black material to eliminate reflectance. T h e front of the chamber was constructed of molded rubber which closely fit the subjen's face thus eliminating extraneous light. The flicker-generating apparatus (GrasonStadler, Model EGGG) was set at a light-dark ratio of 0.50 and a lamp luminance during the "on" phase was approximately 35 cd/ma. T h e auditory stimulus was produced by a white noise-generator housed in a von Bekesy audiometer (Grason-Stadler, Model E800). The spectrum produced by this device falls off only above 10,000 Hz. The intensity range of the generator allows for discrete settings from 20 to 120 d B (SPL), adjustable in 5-dB (SPL) steps. The stimulus was presented binaurally by means of one pair of calibrated air-conduction earphones (Grason-Stadler. M X 4 l / A R ) factory calibrated to be used with the above described audiometer. T h e intensity level produced by the audiometer was recalibrated by means of a Vu meter and calibration controls located o n the audiometer control panel before each block of trials. T h e presentation of the stimulus was controlled by two Hunter timers set for simultaneous "on"- and "off"-set and a constant intertrial interval. During testing in the "Headphones and Earplugs" condition a pair of air cushioned earplugs were worn (Willsan "sound silencer," Model EP-100). These earplugs provided an average attenuation of about 3 0 d B (SPL) between 125 and 8.000 Hz. Subjects responded to the stimuli by means of a hand-held "responding device" wired to the audiometer. When the button on this device was depressed, it initiated a loud "click" i n the audiometer in the adjacent room. T h e operation of this device, however, was inaudible to the subject. Testing Procedure Testing was completed for each subject in 10 testing sessions held one a day over a 14-day period. Sessions took place in the morning and began at the same time each day for each subject ( & one h o u r ) . In each test session 500 trials were given as five blocks of 100 trials. Each block was characterized by one of 1 0 auditory conditions: "Headphones and Earplugs." "Headphones." or 40, 50, 60, 70, 80, 90, 100, 110 dB (SPL) of white noise presented via headphones.' The noise levels were quasi-randomly assigned to blocks so that the same level did not appear twice in the same session but did appear a total of five times. A further restriction was that each noise level had to appear ar least every other test day. Blocks were separated by a 2- to 10-min. rest. the exact elapsed time being subjectcontrolled. Time was also provided for the dark adaptation of the subjects before each block of trials. This was necessary as the subject's chamber was dimly lit during the rest but very dark during testing. T o ensure that the subject began testing for each block at about the same level of dark adaptation a small "light leak" between the subject and test room was provided. This " l e a k was such that it was visible only after several minutes in the darkened room. Subjects were individually consistent i n the duration of this dark adaptation period which averaged 5 min. Dark adaptation during testing was. of course, a consistent function of the experimental visual stimulation and the darkened viewing chamber. The visual stimulation in a random 5 0 % of the trials of each block was "fused" cps). O n the other 5 0 % the presentation was "flickering" (presented at the (150 on-off rate determined to reliably produce d' values between 0.5 and 1.0 for each subject during preliminary testing). During the 6-sec. intertrial interval the subject was required to decide his confidence that a flickering light had been presented i n accordance with a 4-point rating scale ("1" indicating certainty or near certainty the light had been
+
794
D.W. HARPER
"fused," "2" that it probably was "fused," "3" that it probably was "flickering," and "4" that he was certain or almost certain that the light presented was "flickering"). The subject was then to depress the response button an appropriate number of times. T h e experimenter used this interval to reset (when necessary) the on-off frequency of the visual stimulus to be presented o n the next trial and to record the response of the subject to the preceding trial.
RESULTS Double probability plots of the Receiver Operating Characteristic (ROC) curves of each subject for each auditory condition were constructed. Examination of these indicated that each was well fit by a straight line but that the slopes varied considerably from unity, tending to take values of less than one. This last attribute was taken as an indication that the underlying distributions were not consistently of equal variance (see Green & Swets, 1966, pp. 96-97) and therefore that the sensitivity index d' and the bias under p were inappropriate. Accordingly, the sensitiviry index d,' (Egan & Clarke, 1966) which avoids assumptions concerning the sizes of variances of the underlying distributions was calculated. McNichol's (1972, pp. 93-96) over-all response-bias index (here called Bh,,) which corrects for any slope deviations from unity in the ROC curve double probability plot was also calculated. In Fig. 1 sensitivity changes from the most quiet baseline condition for each subject are plotted as a function of the intensity of the concomitantly
-
SUBJECTS 1.0
0.8
0.6 Q)
u 0.4
0
'0.C'
-
Y 'B.S.
*--a
'W.T*
-
-
-
0
H~ P
H " 40
50 ~
60" 70 ' 8 0
9"0
100' 110 ~
dB (SPLI-WHITE NOISE FIG. 1 . Flicker sensitivity (d.') as a function o f intensity of auditory condition
795
NOISE IN VISUAL FLICKER
presented noise. The complex functions clearly vary for each subject in terms of degree of changes produced by the various noise conditions. What is remarkably similar, however, is that there is a greac deal of agreement between subjects on the general shape of the curves and on the points where sensitivity is at a peak or low poinr. A Friedman two-way analysis of variance for ranked "white noise" data confirmed the impression of significant variability in sensitivity over the noise conditions (X,.' = 17.2, df = 7, 9 < .02). The visual impression of consistency between subjects was corroborated by an average rank intercorrelation of 0.83.
-
SUBJECTS
t--.'D.C' ' 6 .S' ,---a 'W.T'
01
E
P
0
H P
H
40
50
60
70
80
90
100
110
dB (SPL)
FIG. 2. Response bias (BM.) as a function of intensity of auditory condition
Fig. 2 illustrates changes in response bias due to noise level. Little change in bias is evidenced by two of the subjects whose BAI, values hover around the value of 1. The third subject shows sporadic bias of a conservative nature. A Friedman analysis on these data showed no significant variation over noise level. The observed bias then is best characterized as idiosyncratic.
DISCUSSION The present findings of systematic variation in visual temporal acuity with changes in intensity of a concomitantly presented auditory stimulus invite comparison with previous research. In the studies reviewed in the introduction there appeared to be a peak in sensitivity at moderately intense levels of a tone
796
D. W. HARPER
or noise (77 to 90.5 dB, SPL). Further increases brought a lowering of sensitivity, relative to that peak and perhaps (Miller, 1963) relative to a condition of minimal auditory stimulation. In the present investigation data also yielded a peak in sensitivity at an intermediate value of auditory intensity, 70 dB (SPL). As in the previous research further increases in incensity brought a decrease in sensitivity-but in the presenc results this was true only up to a point. Sensitivity at 100 dB (SPL) was in the present cases again high, while in much of the previous research the downward trend with increased intensity continued. Kruger (1962), however, reported that one of his two subjects showed a slight increase in sensitivity at 96 dB (SPL) from more moderate intensities. Munro (reviewed by Miller, 1963) reported a reversal towards increased sensitivity when over-all auxiliary stimulation was increased. The present research was the first to assess the effect of low levels of intensity of the auditory input on flicker sensitivity, viz., 40, 50, 60 dB (SPL). This "arm" of the curve changes radically the inverted-U or oscillatory function described by previous authors. The data from auditory intensities of 50 through 90 dB (SPL) generates a curve comparable to previous descriptions. What is added are two other peaks at or near the extremes of intensity of white noise input. One such peak was not expected on the basis of previous research ( 100 dB). The other was in an incensity region not previously explored ( 4 0 dB). Previous related studies made no attempt to control for response bias. However, in the present study there was no evidence to indicate any responsebias effects that were systematic and common to the three subjects. These data, therefore, do not add to the understanding of the findings obtained by means of the traditional psychophysical procedures. Their value lies in the fact that any response bias, regardless of its idiosyncracy has not entered into the sensitivity data. Undoubtedly this has contributed to the high level of agreement in sensitivity data among subjects in this study. It is difficult to interpret the present results as supporting the theoretical speculations presented in the introduction. It was noted there that those speculations were based on insufficient and possibly confounded data. The present research has sampled from a wider range of auditory intensities and has demonstrated considerable agreement while employing superior methodology. It seems reasonable, therefore, to examine those theoretical speculations in light of this "best evidence." It is clear chat no simple theory could predict the complex relationship between visual flicker sensitivity and auxiliary auditory intensity observed in the presenc research. The suggestion that the effect is based on simple arousal is not tenable in light of the present resulcs. Of course the presenc findings in no way indicate that arousal is not involved in the sensory interaction observed here.
NOISE IN VISUAL FLICKER
797
At the biological level of explanation the present complex results cause another sort of problem, namely, of what possible use to the organism would be the functional organization of the sensory system reflected in the present data? It may be that there is some advantage to accentuating visual temporal it is not now clear what that adacuity at 40, 70, and 100 dB (SPL)-but vantage might be. It is possible that, rather than reflecting an adaptive mechanism, the interaction (at least in this case) is merely a reflect-ion of the organization of the nervous or information-processing system-an epiphenomenon If sensory interaction is such an epiphenomenon the importance of this type of research is not diminished as it may lead to a better understanding of the organization that does underlie adaptive processes. One line of research which may contribute to theoretical advances concerns the assessment of any arousal component in the present results. Specifically two studies are suggested, both employing the basic procedures of the research reported here. One would optimize the arousal value of the auditory stimulus by totally randomizing presentation of levels of white noise rather than presenting them in blocks of 100 trials. Another would minimize the arousal value of the same stimuli by having the auditory stimulation be a constant background to a block of trials rather than being simultaneously -presented with the visual stimulus. With all other factors held constant across investigations (including the present one), the resultant sensitivity curves could be assessed for the contribution of arousal. It seems that another important line of research would involve assessment of the effect of the "auxiliary stimulus" on the auxiliary modality. A method for assessing the sensitivity of two simultaneously stimulated modzlities has been developed in another context by Eijkman and Vendrick (1965). This elegant procedure also cakes advantage of the methodology of signal detection analysis. REFERENCES ALLEN, F.. & SCHWARTZ, M. The effect of stimulation of the senses of vision, hearing, taste, and smell upon the sensitivity of the organs of vision. Jotma1 o f General Physiology. 1940. 24, 105-121. CLARK,W. C. The psyche in psychophysics: a sensory-decision theory analysis of the effect of instructions o n flicker sensitivity and response bias. Psychological B7illerin. 1966, 65. 358-366. CLARK, W .C., BROWN, J. C.. & RUTSCHMANN,J. Flicker sensitivity and response bias in psychiatric patients and normal subjects. Journal o f Abnormal Psychology, 1967. 72. 35-42. CLARK. W. C., RUTSCHMANN.J., LINK, R.. & BROWN,J. C. Comparison of flickerfusion thresholds obtained by the methods of forced-choice and limits on psychiatric patients. Perceptual and Motor Skills, 1963, 16, 19-30. EGAN,J. P.. & CLARKE, F. R. Psychophysics and signal detection. In J. B. Sidowski (Ed.), Experimental methods and instrr~nzentation in psychology. New York: McGraw-Hill. 1966. Pp. 21 1-246. ErjmAN, E.. & VENDRICK,A. Can a sensory system be specified by its internal noise? Jownal of the Acoirrtical Society o f America, 1965, 37, 1102-1109. FUSTER, J. M. Effects of stimulation of brain stem o n tachistoscopic perception. Science, 1958, 127, 150.
798
D. W. HARPER
FUSTER.J. M., & UYEDA,A. A. Facilitation of tachistoscopic performance by stimulation of midbrain tegmental points in the monkey. Experimental Nerrrology, 1962, 6, 384-406. GREEN, D. M., & SWETS, J. A. Signal detection theory and psychophysics. New York: Wiley, 1966. KNOX, G . W. Investigations of flicker and fusion: 111. T h e effect of auditory stimulation on the visual CFF, ]ournal o f General Psychology, 1945, 33, 139-143. ( a ) KNOX, G . W. Investigations of flicker and fusion: 1V. The effect of auditory flicker o n the pronouncedness of visual flicker. Joursal o f General Psychology, 1945. 33, 145-154. ( b ) KRAVKOV,S. W. Action des excitations auditives sur la frequence critique des papillotemeuts limineux. Acta Ophthnlmologica, 1935, 13, 260-272. KRUGER,S. I. The effects of chlorpromazine on flicker fusion threshold with intersensory stimulation. Unpublished Ph.D. dissertation, Boston Univer., 1962. LTNDSLEY, D. B. Common factors in sensory deprivatio.n, sensory distortion, and sensory overload. In P. Solomon. P. Kubzanskv. P. Leiderman. T. Mendelson. R. Trumbull, & D. Wexler ( ~ d s . ' ) , Sensory deprivation. ~ a k b ; r i d ~ e~:a r v a r dUniver. Press, 1961. Pp. 174-194. MAIER,B.. BEVAN,W., & BEHAR,1. The effect of auditory stimulation upon the critical flicker frequency for different regions of the visual spectrum. American Jorrrnal o f Psychology. 1961, 74, 67-73. MCCROSKEY,R. L. The effect of specified levels of white noise upon flicker fusion frequency. U. S. Naval School of Aviation Medicine, Pensacola. Florida. Joint Research Project, N M 18 OZ 99 Subtask 1, Report No. 80, 1958. M c N r c ~ o L D. , A primer o f signal detection theory. London: Allen & Unwin, 1972. ~ I I L L E RH., L. T h e effects of auditory stimulation o n the critical flicker fusion frequency response. Unpublished Ph.D. dissertation, Univer. of Hawaii, 1963. OGILVIE,J. L. Effect of auditory flutter o n the visual critical flicker frequency. Canadian Journal o f Psychology. 1956, 10, 61-68. ( a ) OGILVIE,J. L. The interaction of auditory flutter and CFF, the effect of brightness. Canadian Iosrnal o f Psychology, 1956, 10, 207-210. ( b ) VON SCHTLLER,P. Interrelation of different senses in perception. British Jorrrnal o f Psychology, 1935, 25, 465-469. WALKER,E. L., & SAWYER.T. M. The interaction between critical flicker frequency and acoustic stimulation. Psychological R e c o ~ d ,1961, 11, 187-193. Accepted March 21, 1979.
