Journal of Gerontology 1978, Vol. 33, No. 5, 702-710
Age-related Differences in Binocular Backward Masking with Visual Noise1 Robert E. Till, PhD2
ESEARCH with a variety of experimental tasks has indicated that a slowing of perR ceptual processes is associated with aging. An important question for such research concerns the locus of these age-related changes. Some researchers, for example, have suggested that the aging process may contribute to an increased span of the psychological moment, or unit of central scanning time (Eriksen et al., 1970). Visual perception data presented by Eriksen et al. and by Walsh (1976) are consistent with this hypothesis and suggest a substantial age difference in central, or cortical, processes in perception. At the same time, Axelrod and his colleagues (Axelrod & Eisdorfer, 1962; Axelrod et al., 1968) have argued for increased stimulus persistence with aging, a phenomenon that need not be of central origin. For example, recent work discussed by Pollack (1978) suggests that the mechanism of stimulus persistence may be in the retina. If so, perceptual slowing with age may be attributable to changes at several levels in the visual system. Obviously, the relative importance of "central" and "peripheral" factors in perceptual slowing is not yet understood. In addition, there is considerable confusion in the literature concerning the use of the terms "central" and "peripheral." Some studies seem to have adopted the traditional neurophysiological distinction between "peripheral 'This research was supported in part by a National Science Foundation Institutional Grant for Science (7002873-88-15). 2 Requestsfor reprints should be sent to Robert E. Till, Dept. of Psychology, Southern Methodist Univ., Dallas 75275.
702
sensory" changes and "central cortical" changes in the nervous system (e.g., Birren & Botwinick, 1955; Birren & Wall, 1956; Kline & Szafran, 1975). Others have used an information processing distinction, proposed by Turvey (1973), which considers as peripheral the retina, lateral geniculate nucleus, and striate cortex (e.g., Walsh, 1976; Walsh et al., 1978). The latter approach has an advantage in that clearly defined criteria are established for the central-peripheral distinction. While some may argue that the approach does not guarantee the exclusive isolation and study of central or peripheral mechanisms, it is at least possible to reliably distinguish between studies of "central" and "peripheral" processes based on the behavioral characteristics outlined by Turvey. Recent investigations by Walsh (1976) and Walsh et al. (1978) have utilized a well-established paradigm in visual research, namely backward masking, in an attempt to study central or peripheral factors (as defined by Turvey) in perceptual slowing. These studies suggest that age-related differences exist in both central and peripheral visual processing and that perceptual slowing at these two levels of processing differs in nature. Backward masking occurs when two brief visual stimuli are separated in time by a short interval. The first stimulus (the target) is masked by the second stimulus (the mask), and the subject is unable to identify the target. Obviously, target processing may or may not be complete at the time of the mask onset, and it is an empirical task to determine the critical interstimulus interval (ISIc) necessary
Downloaded from http://geronj.oxfordjournals.org/ at University of Birmingham on August 29, 2015
Peripheral processing, as denned by Turvey (1973), was investigated in a study of binocular backward masking involving random visual noise. For young (mean age 20.3 years) and old (mean age 55.4 years) adults, processing time (represented by stimulus onset asynchrony, SOA) was characterized as a power function of target energy (TE) (i.e., TE& x SOA = K). Overa wide range of target energy levels, processing time showed a similar rate of decline (similar exponent, b) for both age groups with increasing target energy. However, old subjects processed targets more slowly at all target energy levels (different constant, K), and this age difference was constant across all target energy levels. Results were independent of variables such as sex, education, response criterion, unmasked threshold, and method of determining stimulus onset asynchrony.
AGE DIFFERENCES IN BACKWARD MASKING
possible that masks of intermediate targetmask similarity would produce less distinctive evidence for "central" or "peripheral" masking. In this regard, it should be noted that there are exceptions to the generalizations embodied in Turvey's criteria. For example, Kinsbourne and Warrington (1962a, b) have reported dichoptic masking with a visual noise mask. Furthermore, Turvey himself has found dichoptic masking with the visual noise mask under certain conditions (1973). Using Turvey's (1973) view of central processing and stimuli like those of Turvey, Walsh (1976) examined age differences in backward masking. The masking obtained with both young and old adults closely fit the criteria of central masking suggested by Turvey. Walsh found a reliable age difference, suggesting a central locus for perceptual slowing with age. Furthermore, this slower processing appeared as a constant difference in processing time between old and young at each TE level investigated. In contrast, Walsh etal. (1978) recently studied peripheral processing, using Turvey's criteria, and found an age difference in processing time that was not constant; an interaction between age and TE level was obtained. Specifically, the age difference in peripheral processing time was reduced as TE increased. This is seen in the power functions presented by Walsh et al. to describe the relationship between TE and processing time. The power functions for old and young were equations of the form TEb x SOA = K. (It was suggested that SOA is a more appropriate measure of peripheral processing time than is ISIc which was used by Turvey.) The exponent b was similar for both age groups, suggesting that processing time declines at the same rate for both groups as TE increases. However, the values of K were different, reflecting the fact that one curve (for the old) was seen graphically as above the other (for the young). Thus, it appears that there is an age difference in peripheral processing, which is independent of central perceptual slowing and that the magnitude of the age difference in peripheral processing is a function of TE. Walsh et al. suggested that the maximum speed of peripheral neural net operation may not change with advancing age, although the speed of operation for equivalent-energy targets may be less.
Downloaded from http://geronj.oxfordjournals.org/ at University of Birmingham on August 29, 2015
for avoidance of masking. Such a measure is likely to be influenced by several variables including characteristics of the target and mask,, as well as subject characteristics such as age. Turvey (1973) has examined the functions obtained with different kinds of masking stimuli, and his results suggest that backward masking may result from experimental conditions of two distinctly different kinds. These conditions provide the basis for the criteria Turvey uses to establish the probable locus of interference in visual masking research. In one series of experiments, Turvey (1973) used capital letters as targets with random visual noise (a 50% black, 50% white, random arrangement of small-grained contours) as the mask. When target and mask were presented monoptically or binocularly, an energy-sensitive masking function was observed. Such masking, which is said to arise peripherally, is defined by four characteristics: (a) The masking occurs only when mask energy exceeds target energy; (b) the relationship between target energy (TE) and ISIc is described by a power function (i.e., TE x ISIc = K, a constant); (c) figural features of the mask are noncritical for masking; and (d) such masking characteristics are not obtained with dichoptic presentation of target and mask. Central masking is quite different, according to Turvey (1973), and is determined by the time separating the onset of the target and the onset of the mask (stimulus onset asynchrony, SOA). The distinctive characteristics are: (a) A target may be masked by a mask of lower energy; (b) an addition rule relates target duration (TD) and ISIc, that is, TD + ISIC = a constant = SOA; and, (c) masking of central origin is dependent on the similarity of figural characteristics of target and mask. Thus, a pattern mask, constructed of line segments similar to the target, produced strong dichoptic masking while the random noise mask usually had no effect. It is important to note that Turvey's (1973) central-peripheral distinction depends upon the different patterns of results obtained with two very different masking stimuli. Thus, it is the similarity of the figural characteristics of target and mask that is the important determinant of the masking effects one observes (cf. Haber, 1970). Since Turvey's masking stimuli represent two distant points on a continuum of target-mask similarity, it seems quite
703
704
TILL SO As" could be determined and used to provide estimates of processing time equated for percentage of accuracy of identification. It was expected that coarse SO As would reflect high points on the psychophysical function — overestimates, to some extent, of peripheral processing time. Adjusted SO As, on the other hand, would be less likely to contain error introduced by the use of the ascending limits method and were expected to provide a more precise assessment of age differences in processing time. METHODS
Subjects Ten old (3 female, 7 male, range 54-56 years) and ten young (4 female, 6 male, range 18-23 years) served as subjects. The old subjects were unpaid volunteers recruited from staff personnel and from the community. Their mean education level was 16.3 years (range 12-20 years). The young subjects were undergraduate students who received course credit for participation. Their mean education level was 14.0 years (range 12-16 years). All subjects were screened to insure a minimum visual acuity of 20/30 with both eyes (including subjects with corrected vision) as tested with a Snellen chart. The TD required for correct identification of four out of six targets in the absence of the mask was determined under binocular presentation. All subjects were successful at a TD of 4 msec or less with the target luminance set at the 3.8 cd/m2 level. (One old subject was initially successful only when TD was increased to 6 msec. A second test given after the masking trials showed his unmasked threshold to be 2 msec.) Thus, all subjects were successful with unmasked targets shown at the lowest level of luminance used in the study. Materials and Apparatus A dark field with a faint point of red light at the center was used as a fixation field. This fixation point was generated internally by the tachistoscope and was set at a luminance of less than .1 cd/m2. Four symmetrical letters of the alphabet, composed of straight line strokes, served as target stimuli (i.e., T, V, X, Y). Each letter was reproduced 25 times using 24-point Helvetica capital letters (Quik Stik
Downloaded from http://geronj.oxfordjournals.org/ at University of Birmingham on August 29, 2015
The present experiment was designed to extend and clarify the findings of Walsh et al. (1978). While the procedures were similar in most respects, there were notable exceptions. For example, binocular presentation was used for reasons of convenience and on the assumption that no important differences arise between monoptic and binocular masking with visual noise (cf. Turvey, 1973). (Informal pilot data seemed to confirm this.) On the other hand, target stimuli and the visual noise mask were kept very similar to those of Walsh et al. in order to facilitate a comparison of the results of the two studies. A major feature of the present research was the inclusion of a wider range of TE levels over which to examine age differences in peripheral processing. While the Walsh et al. (1978) study used only three TE values at rather low levels, the present study examined seven TE levels, three within the range used by Walsh et al. and four at considerably higher TE values. It was expected that power functions computed over the seven TE levels would provide similar, but more reliable, estimates of the TE x SOA relationship for old and young than were obtained in the Walsh et al. study with only three TE values. Furthermore, it was expected that increases in TE would continue to produce proportional changes in the speed of peripheral processing for the two age groups studied, even at high TE levels. In addition to the wider range of TE values, the present study introduced a variation in the method of estimating processing time (SOA) designed to allow greater accuracy of measurement. Thus, two sets of results were obtained. One set of data could be directly compared with the data of Walsh et al. (1978). The other set of data consisted of adjusted SOA scores calculated to reflect estimates of processing time that were equivalent with respect to their point on the psychophysical function relating percentage of accuracy of identification to SOA. Specifically, a method of ascending limits like that of the Walsh et al. study was used to establish a "coarse SOA" (four successive correct identifications). Following the determination of this coarse SOA, the percentage of accuracy of identification was assessed at several SOA values ranging from values 10 msec smaller than the coarse SOA up to the original coarse SOA itself. Thus, "adjusted
AGE DIFFERENCES IN BACKWARD MASKING
Procedure Unmasked trials. — After visual acuity was tested, subjects were seated in the darkened laboratory and were allowed 5 min to adjust to the low ambient room illumination (produced by filtered light from a 15W fluorescent bulb located out of the subject's view). The fixation field, a target letter, and the random noise mask were each shown for 20 sec to familiarize subjects with the materials. All stimuli were presented binocularly. On each trial, subjects were forced to report one of the four possible target letters. The 1.5 sec presentation of the fixationfieldwas initially followed by a 1-msec presentation of a target; a series of six targets was presented under these conditions. For each successive series, the TD was increased
by 1 msec. The experimenter determined the unmasked TD necessary for a subject to correctly identify four of the six targets in a series. Masking trials. — After the determination of the unmasked threshold, masking trials were begun. Again, all stimuli were presented binocularly. Each trial consisted of a 1.5 sec presentation of the fixation field followed by a target letter at one of seven TE levels. The target was followed by a dark interstimulus interval (blank channel) which ended with the 50 msec, 30.8 cd/m2 mask presentation. Using a method of ascending limits, the experimenter determined the ISIc needed for avoidance of masking. The initial interstimulus interval was set at 2 msec and was increased in 2-msec steps each time an error was made until the subject correctly identified four consecutive targets all at a given ISI. This method provided a measure of coarse SOA (ISIc plus 4-msec TD = SOA). Following this, the subject rested for 30 sec and was then given six series of masking trials, each containing six targets. The first series included six trials with SOA set at 10 msec less than the just-determined coarse SOA. The second series presented trials with SOA set at 8 msec less than the coarse SOA. The remaining four series approached the coarse SOA value by 2-msec steps with the last series being presented at the coarse SOA itself. The six series provided two kinds of information: (a) the percentage of accuracy of identifications at six SOA values ranging from the coarse SOA down to a value likely to be below the threshold for identification and (b) the shortest SOA at which four of the six targets in a series were correctly identified (i.e., the "adjusted SOA"). The coarse and adjusted SOAs were determined in the manner described above for each of the seven TE levels. Ten different sequences of TE levels were randomly determined, and a particular randomized sequence was used with one subject from each age group. The pace of the experimental session was determined primarily by the experimenter through the use of rest periods, but also by the subject with respect to trials needed to reach initial coarse SOA at any TE level. Thus, the entire session lasted between 70 and 90 min (approximately). Because of their higher coarse SOA values, old subjects tended to have slightly longer sessions than young subjects.
Downloaded from http://geronj.oxfordjournals.org/ at University of Birmingham on August 29, 2015
792) and placed on translucent-white Kodak Ektagraphic slides. All letters were centrally located in the visual field and subtended a visual angle of 1.11 degrees vertically by .91 degrees horizontally. The width of the strokes composing the letters subtended a visual angle of approximately .20 degrees. Letters were black against a white background. Apparent viewing distance was 680 mm. Targets were presented in a randomized order and appeared for durations of 4 msec. Seven luminance levels were used: 3.8, 4.8, 9.6, 12.3, 19.2, 30.8, and 38.5 cd/m2. The combination of these luminance levels with the 4-msec TD produced seven TE levels: 15.4, 19.2, 38.4, 49.2, 76.8, 123.2, and 154.0 cd/m2 x msec units. Random visual noise served as the mask and was a 50% black, 50% white array spread over the entire visual field. This mask was identical to the one used by Walsh et al. (1978), and it consisted of horizontal and vertical line segments subtending a visual angle of about .04 degrees in width and .10 degrees in length. In all cases, mask luminance was 30.8 cd/m2 and was presented for a 50-msec duration. An automatic three-channel tachistoscope (Scientific Prototype Model N-1000/A) with automatic stimulus changers was used. The unit provided three channels of binocular stimulus presentation in addition to a fourth channel containing only the internal fixation point. Luminance levels were measured with a Spectra Spotmeter Model UBD-10 with an internal calibration source (Photo Research). However, Kodak neutral density filters were used for reliable variation of stimulus luminance.
705
TILL
706
Table 1. Stimulus Onset Asynchrony (SOA) Required to Escape Masking (in msec) As a Function of Target Energy, Threshold Method and Age.
Target Energy (c/m2 x msec)
Coarse SOA Method Young Old
Adjusted SOA Method Young Old
18.00 4.32
26.60 9.71
14.20 2.74
21.80 6.07
123.2 Mean SOA SD
18.00 4.99
27.60 9.56
14.40 4.79
23.00 9.53
76.8 Mean SOA SD
24.60 5.89
32.00 10.24
19.80 4.56
26.40 9.56
49.2 Mean SOA SD
27.40 6.80
43.40 9.98
23.40 6.74
35.40 9.75
38.4 Mean SOA SD
32.80 8.07
43.40 12.00
27.00 6.48
38.00 10.79
19.2 Mean SOA SD
40.80 7.25
53.00 18.09
36.80 5.09
48.00 18.40
15.4 Mean SOA SD
49.60 10.01
63.00 12.97
41.80 9.45
56.20 11.90
RESULTS
An analysis of the TDs required for subjects to identify four out of six unmasked targets indicated a significant age difference between young and old subjects, /(18) = 2.923,p < .01. The mean durations were 1.20 msec for the young (with SD = 0.42) and 2.70 msec for the OLD 65 old (with SD = 1.57). YOUNG DATA FROM Table 1 presents the means and standard 55+ WALSH ET AL. deviations for the results of the masking trials; SOA values are shown as a function of age, 45 method of SOA determination, and TE level. When the standard deviations associated with coarse SOA measurement are compared with those obtained under the adjusted SOA meth25od, a small but reliable difference is found (sign test,p < .001). Although the average difference 15is only 1.00 msec, it appears that the adjusted SOA method consistently produces less vari-t—i—i—i—i—i—i—i—i—i—i—i—i—i—^ ° 20 4 0 60 80 100 120 140 ance among subjects' SOA scores than does TARGET ENERGY [Cd/m 2 x MSEC] the coarse SOA method. The effect is evident for 13 of the 14 pairs of standard deviations. Fig. 1. Comparison of present data with those of Walsh A 3-way analysis of variance was performed et al. (1978, Exp. 2). Mean SOAs for old and young are on the SOA values summarized in Table 1. plotted as a function of target energy. Data points for Each of the main effects was significant: for the present study are based on the adjusted SOA method.
Downloaded from http://geronj.oxfordjournals.org/ at University of Birmingham on August 29, 2015
154.0 Mean SOA SD
age, F(l, 18) = 12.23, p < .005; for method of SOA determination,F(l, 18) = 402.9,/? < .001; and for TE level, F(6, 108) = 62.29, p< .001. None of the interactions were significant (p > .05). In particular, the age x TE level interaction reported by Walsh et al. (1978) was completely missing from the present data, F(6,108) = 0.78. In order to pursue this discrepancy, an alternative 3-way analysis of variance was performed. Rather than using all seven TE levels, the analysis included only the three lowest levels since these fall within the range examined by Walsh et al. This alternative approach mirrored the results of the previous analysis. In particular, no age x TE level interaction emerged. Thus, one is led to question the generality of the interaction suggested by the data of Walsh et al. and to consider the alternative that the age difference in peripheral processing is constant across TE levels as it is in central processing (cf. Walsh, 1976). The evidence for an interaction (from the Walsh et al. study) as well as evidence for noninteraction (from the present study) is illustrated in Fig. 1 and will be discussed later. Since seven TE levels were used during the masking trials, it was possible that practice and/or fatigue effects might be confounded with TE levels. Although TE levels were presented in a randomized sequence, it was important to analyze for a possible order effect since the two age groups might be differentially
AGE DIFFERENCES IN BACKWARD MASKING
707
log y = -b (log x) + log k i
I
504030-
TE=15.4 o OTHER TE LEVELS -10 -8 -6 -4 -2 0 MSEC BELOW COARSE SOA
Fig. 2. Percentage of accuracy of target identification for each target energy level (averaged over age groups). Data points in a curve represent accuracy at specified SOAs below the coarse SOA. The curve for the lowest energy level is labeled separately; the other six are not since they intersect considerably.
Converted to logarithmic form, the present data would approximate a straight line relationship. Thus, linear regression analysis was performed on the (log TE, log SOA) pairs in order to determine the line of best fit. The slope of the regression line (represented by -b in the logarithmic form of the equation) is the best estimate of the exponent in the power function. Linear regression analysis of the 70 data points for each age group indicated similar exponents in the best-fit equations. For the young, TE-486 x SOA = 165.2, and for the old, TE-420 x SOA= 183.5. The fit of these equations is quite good, as indicated by the correlation between log TE and log SOA: r = -.854, p < .001 for the young, and r = -.711, p < .001 for the old.
Downloaded from http://geronj.oxfordjournals.org/ at University of Birmingham on August 29, 2015
susceptible to the effects of practice or fatigue. from coarse SOA. None of the interactions An additional analysis of variance was per- proved significant, nor was there an age difformed, therefore, to provide information ference. Distance from coarse SOA was a sigabout the adequacy of the randomization of nificant factor reflecting the improvement TE levels and to check for order effects. The in accuracy that occurred when the SOA inanalysis was based on the previously analyzed terval was increased from a time 10 msec bedata of Table 1, except that SOA values were low coarse SOA up to the coarse SOA value itrearrranged to reflect the order of testing of a self, F(5,90)i = 63.57,p .05. Power functions for old and young show similar rates of exponential decline. Since age differences in peripheral processing were not reflected in exponent differences, it is reasonable to suppose that the age difference is indicated by different constants (K) in the power functions for old and young. Indeed, Fig. 1 shows clearly that the constants differ for the two groups, on the average, since the curve for the old group is above the curve for the young group. However, statistical comparison of the constants for the two groups cannot be done as directly as it was for the exponents of the two groups. This is because the constant computed for an individual is dependent on the value of the computed exponent; slight changes in exponent value typically produce large changes in the constant. Thus, variance among the constants computed for a group is large. For example, the standard deviation for the 10 constants computed for the young group was 80.3; the value for the old subjects was 79.6. As an alternative method for examining age differences in constants, subjects in one group were compared with counterparts in the other age group who had data described by similar exponents. Matching was possible for seven pairs of subjects for which the average absolute value of the difference between exponents in a pair was only .008. The computed constants for these seven matched pairs were then examined with a /-test which indicated a significant difference, f (6) = 2.922, p < .03. The mean constant for the seven young was 137.2 (withSD = 31.2), and for the seven old, 209.2 (withSD = 63.4). Furthermore, the constant for the old subject was higher than that for the young subject in every matched pair.
AGE DIFFERENCES IN BACKWARD MASKING
difference appears similar to that obtained in Walsh's (1976) study of age differences in (supposedly) central perceptual processing. Additional research will be needed to determine whether the constant age differences in these experiments are distinctly different, or whether both are reflections of an age change with a single locus. It is interesting to compare the power functions computed for old and young with those of the Walsh et al. (1978) study. The exponents are surprisingly similar despite procedural differences, sample differences, the use of binocular presentation, and the wider range of TE values over which the exponents were computed. Peripheral processing time is clearly affected by such variables, or at least it can be — a fact reflected in the different exponents of power functions fitted to the data presented by Turvey (1973). Some of Turvey's exponents approach the values of the present study; others approach a value of 1.00 (cf. Walsh etal., 1978). It is possible that the age difference in the present study is due to constant age-related changes in the operating speed of the slowest peripheral nets outputting data on the target (cf. Turvey, 1973), while changes at earlier stages of peripheral processing produce no appreciable age differences. Alternatively, the age difference observed here may have a central locus while the minimal processing time (as seen in younger subjects) may be a peripherally determined variable dependent on target energy and other features of the stimulus. SUMMARY
Age differences in visual perception were examined by means of a binocular backward masking technique. Backward masking occurs when a target stimulus is followed so closely by a second stimulus that the subject does not have time to process the target. As the separation between target and mask is increased, a critical separation (stimulus onset asynchrony, SOA) is reached at which subjects have enough uninterrupted time to process, and therefore identify, the target. In the present study, old and young differed in terms of the SOA interval required, with older subjects needing more processing time. For young and old, processing time was characterized as a power function of target
Downloaded from http://geronj.oxfordjournals.org/ at University of Birmingham on August 29, 2015
subjects' SOA scores is smaller than it is under the coarse method. Second, information is provided about the approach to the threshold SOA (i.e., the accuracy of subthreshold values of SOA). It appears that accuracy during the approach to threshold criterion (four out of six correct) does not vary with age, but does vary with TE (especially at low TE levels). An important purpose of the present investigation was the examination of peripheral processing over a wide range of TE levels. The results obtained here replicate those of Walsh et al. (1978) insofar as main effects were found for TE and age. As illustrated in Fig. 1, however, the present data do not agree with the findings of Walsh et al. (Exp. 2) indicating a TE x age interaction. For the present data, no interaction was found across a wide range of TE values, nor was there an interaction when only the lowest three TE levels were examined (those within the range investigated by Walsh et al.). Furthermore, data from Experiment 1 of Walsh et al. were obtained with a method of SOA determination requiring seven out of eight correct identifications, and they do not indicate an age x TE interaction. One can argue that the latter results, as well as those of the present experiment, were obtained under conditions insuring the equivalence of SO As with respect to accuracy of identification on the psychophysical function relating accuracy to SOA values and that the results, therefore, involve less error variance. Of course, it is possible that an age x TE interaction occurs only at low TE levels (i.e., below the 15.4 level used here). But there is also more chance of error in this range since SOAs are long and coarse methods of SOA determination may produce threshold values that are very high on the accuracy function (as found for the 15.4 level of the present study). Given the lack of an age difference in power function exponents and the lack of a TE x age interaction (here, and in Exp. 1 ofWalshetal., 1978), it seems most parsimonious to suppose that the age difference in peripheral processing is constant across a wide range of TE values. Faced with the constant age difference in processing time, one is forced to consider that aging does not produce a change in the degree of attenuation of target energy (as indicated by similar power function exponents). Rather, it adds a constant value to the processing time measured by SOA. Furthermore, the constant
709
710
TILL
REFERENCES
Axelrod, S., & Eisdorfer, C. Senescence and figural aftereffects in two modalities. Journal of Genetic Psychology, 1962, 100, 85-91. Axelrod, S., Thompson, L., & Cohen, L. Effects of senescence on the temporal resolution of somesthetic stimuli presented to one hand or both. Journal of Gerontology, 1968,25, 191-195. Birren, J., & Botwinick, J. Age differences in finger, jaw and foot reaction time to auditory stimuli. Journal of Gerontology, 1955,10, 429-432. Birren, J., & Wall, P. Age changes in conduction velocity, refractory period, number of fibers, connective tissue space and blood vessels in sciatic nerve in rats. Journal of Comparative Neurology, 1956, 104, 1-16.
Eriksen, C , Hamlin, R., & Breitmeyer, R. Temporal factors in visual perception as related to aging. Perception and Psychophysics, 1970, 7, 354-356. Haber, R. Note on how to choose a visual noise mask. Psychological Bulletin, 1970, 74, 373-376. Kinsbourne, M., & Warrington, E. The effect of an aftercoming random pattern on the perception of brief visual stimuli. Quarterly Journal of Experimental Psychology, 1962, 14, 223-234. (a) Kinsbourne, M., & Warrington, E. Further studies on the masking of brief visual stimuli by a random pattern. Quarterly Journal of Experimental Psychology, 1962, 14, 235-245. (b) Kline, D., & Szafran, J. Age differences in backward monoptic visual noise masking. Journal of Gerontology, 1915,30, 307-311. Pollack, R. A theoretical note on the aging of the visual system. Perception and Psychophysics, 1978, 23, 94-95. Turvey, M. On peripheral and central processes in vision: Inferences from an information-processing analysis of masking with patterned stimuli. Psychological Review, 1973, 80, 1-52. Walsh, D. Age differences in central perceptual processing: A dichoptic backward masking investigation. Journal of Gerontology, 1976,31, 178-185. Walsh, D., Till, R., & Williams, M. Age differences in peripheral perceptual processing: A monoptic backward masking investigation. Journal of Experimental Psychology: Human Perception and Performance, 1978, 4, 232-243. Weale, R. On the eye. In A. Welford & J. Birren (Eds.), Behavior, aging and the nervous system. Charles C Thomas, Springfield, IL, 1965.
Downloaded from http://geronj.oxfordjournals.org/ at University of Birmingham on August 29, 2015
energy (TE), that is, TEb x SO A = K. Over a wide range of TE levels, processing time showed a similar rate of decline (similar exponent, b) for both age groups with increasing TE. As noted above, old subjects processed targets more slowly at all TE levels (different constant, K), and this age difference was constant across TE levels. While the locus of the observed age difference in processing time has not been determined, it appears that the age difference reflects changes in processing that are not a simple result of attenuation of target energy in the aging eye.
Age-related Differences in Binocular Backward Masking with Visual Noise1 Robert E. Till, PhD2
ESEARCH with a variety of experimental tasks has indicated that a slowing of perR ceptual processes is associated with aging. An important question for such research concerns the locus of these age-related changes. Some researchers, for example, have suggested that the aging process may contribute to an increased span of the psychological moment, or unit of central scanning time (Eriksen et al., 1970). Visual perception data presented by Eriksen et al. and by Walsh (1976) are consistent with this hypothesis and suggest a substantial age difference in central, or cortical, processes in perception. At the same time, Axelrod and his colleagues (Axelrod & Eisdorfer, 1962; Axelrod et al., 1968) have argued for increased stimulus persistence with aging, a phenomenon that need not be of central origin. For example, recent work discussed by Pollack (1978) suggests that the mechanism of stimulus persistence may be in the retina. If so, perceptual slowing with age may be attributable to changes at several levels in the visual system. Obviously, the relative importance of "central" and "peripheral" factors in perceptual slowing is not yet understood. In addition, there is considerable confusion in the literature concerning the use of the terms "central" and "peripheral." Some studies seem to have adopted the traditional neurophysiological distinction between "peripheral 'This research was supported in part by a National Science Foundation Institutional Grant for Science (7002873-88-15). 2 Requestsfor reprints should be sent to Robert E. Till, Dept. of Psychology, Southern Methodist Univ., Dallas 75275.
702
sensory" changes and "central cortical" changes in the nervous system (e.g., Birren & Botwinick, 1955; Birren & Wall, 1956; Kline & Szafran, 1975). Others have used an information processing distinction, proposed by Turvey (1973), which considers as peripheral the retina, lateral geniculate nucleus, and striate cortex (e.g., Walsh, 1976; Walsh et al., 1978). The latter approach has an advantage in that clearly defined criteria are established for the central-peripheral distinction. While some may argue that the approach does not guarantee the exclusive isolation and study of central or peripheral mechanisms, it is at least possible to reliably distinguish between studies of "central" and "peripheral" processes based on the behavioral characteristics outlined by Turvey. Recent investigations by Walsh (1976) and Walsh et al. (1978) have utilized a well-established paradigm in visual research, namely backward masking, in an attempt to study central or peripheral factors (as defined by Turvey) in perceptual slowing. These studies suggest that age-related differences exist in both central and peripheral visual processing and that perceptual slowing at these two levels of processing differs in nature. Backward masking occurs when two brief visual stimuli are separated in time by a short interval. The first stimulus (the target) is masked by the second stimulus (the mask), and the subject is unable to identify the target. Obviously, target processing may or may not be complete at the time of the mask onset, and it is an empirical task to determine the critical interstimulus interval (ISIc) necessary
Downloaded from http://geronj.oxfordjournals.org/ at University of Birmingham on August 29, 2015
Peripheral processing, as denned by Turvey (1973), was investigated in a study of binocular backward masking involving random visual noise. For young (mean age 20.3 years) and old (mean age 55.4 years) adults, processing time (represented by stimulus onset asynchrony, SOA) was characterized as a power function of target energy (TE) (i.e., TE& x SOA = K). Overa wide range of target energy levels, processing time showed a similar rate of decline (similar exponent, b) for both age groups with increasing target energy. However, old subjects processed targets more slowly at all target energy levels (different constant, K), and this age difference was constant across all target energy levels. Results were independent of variables such as sex, education, response criterion, unmasked threshold, and method of determining stimulus onset asynchrony.
AGE DIFFERENCES IN BACKWARD MASKING
possible that masks of intermediate targetmask similarity would produce less distinctive evidence for "central" or "peripheral" masking. In this regard, it should be noted that there are exceptions to the generalizations embodied in Turvey's criteria. For example, Kinsbourne and Warrington (1962a, b) have reported dichoptic masking with a visual noise mask. Furthermore, Turvey himself has found dichoptic masking with the visual noise mask under certain conditions (1973). Using Turvey's (1973) view of central processing and stimuli like those of Turvey, Walsh (1976) examined age differences in backward masking. The masking obtained with both young and old adults closely fit the criteria of central masking suggested by Turvey. Walsh found a reliable age difference, suggesting a central locus for perceptual slowing with age. Furthermore, this slower processing appeared as a constant difference in processing time between old and young at each TE level investigated. In contrast, Walsh etal. (1978) recently studied peripheral processing, using Turvey's criteria, and found an age difference in processing time that was not constant; an interaction between age and TE level was obtained. Specifically, the age difference in peripheral processing time was reduced as TE increased. This is seen in the power functions presented by Walsh et al. to describe the relationship between TE and processing time. The power functions for old and young were equations of the form TEb x SOA = K. (It was suggested that SOA is a more appropriate measure of peripheral processing time than is ISIc which was used by Turvey.) The exponent b was similar for both age groups, suggesting that processing time declines at the same rate for both groups as TE increases. However, the values of K were different, reflecting the fact that one curve (for the old) was seen graphically as above the other (for the young). Thus, it appears that there is an age difference in peripheral processing, which is independent of central perceptual slowing and that the magnitude of the age difference in peripheral processing is a function of TE. Walsh et al. suggested that the maximum speed of peripheral neural net operation may not change with advancing age, although the speed of operation for equivalent-energy targets may be less.
Downloaded from http://geronj.oxfordjournals.org/ at University of Birmingham on August 29, 2015
for avoidance of masking. Such a measure is likely to be influenced by several variables including characteristics of the target and mask,, as well as subject characteristics such as age. Turvey (1973) has examined the functions obtained with different kinds of masking stimuli, and his results suggest that backward masking may result from experimental conditions of two distinctly different kinds. These conditions provide the basis for the criteria Turvey uses to establish the probable locus of interference in visual masking research. In one series of experiments, Turvey (1973) used capital letters as targets with random visual noise (a 50% black, 50% white, random arrangement of small-grained contours) as the mask. When target and mask were presented monoptically or binocularly, an energy-sensitive masking function was observed. Such masking, which is said to arise peripherally, is defined by four characteristics: (a) The masking occurs only when mask energy exceeds target energy; (b) the relationship between target energy (TE) and ISIc is described by a power function (i.e., TE x ISIc = K, a constant); (c) figural features of the mask are noncritical for masking; and (d) such masking characteristics are not obtained with dichoptic presentation of target and mask. Central masking is quite different, according to Turvey (1973), and is determined by the time separating the onset of the target and the onset of the mask (stimulus onset asynchrony, SOA). The distinctive characteristics are: (a) A target may be masked by a mask of lower energy; (b) an addition rule relates target duration (TD) and ISIc, that is, TD + ISIC = a constant = SOA; and, (c) masking of central origin is dependent on the similarity of figural characteristics of target and mask. Thus, a pattern mask, constructed of line segments similar to the target, produced strong dichoptic masking while the random noise mask usually had no effect. It is important to note that Turvey's (1973) central-peripheral distinction depends upon the different patterns of results obtained with two very different masking stimuli. Thus, it is the similarity of the figural characteristics of target and mask that is the important determinant of the masking effects one observes (cf. Haber, 1970). Since Turvey's masking stimuli represent two distant points on a continuum of target-mask similarity, it seems quite
703
704
TILL SO As" could be determined and used to provide estimates of processing time equated for percentage of accuracy of identification. It was expected that coarse SO As would reflect high points on the psychophysical function — overestimates, to some extent, of peripheral processing time. Adjusted SO As, on the other hand, would be less likely to contain error introduced by the use of the ascending limits method and were expected to provide a more precise assessment of age differences in processing time. METHODS
Subjects Ten old (3 female, 7 male, range 54-56 years) and ten young (4 female, 6 male, range 18-23 years) served as subjects. The old subjects were unpaid volunteers recruited from staff personnel and from the community. Their mean education level was 16.3 years (range 12-20 years). The young subjects were undergraduate students who received course credit for participation. Their mean education level was 14.0 years (range 12-16 years). All subjects were screened to insure a minimum visual acuity of 20/30 with both eyes (including subjects with corrected vision) as tested with a Snellen chart. The TD required for correct identification of four out of six targets in the absence of the mask was determined under binocular presentation. All subjects were successful at a TD of 4 msec or less with the target luminance set at the 3.8 cd/m2 level. (One old subject was initially successful only when TD was increased to 6 msec. A second test given after the masking trials showed his unmasked threshold to be 2 msec.) Thus, all subjects were successful with unmasked targets shown at the lowest level of luminance used in the study. Materials and Apparatus A dark field with a faint point of red light at the center was used as a fixation field. This fixation point was generated internally by the tachistoscope and was set at a luminance of less than .1 cd/m2. Four symmetrical letters of the alphabet, composed of straight line strokes, served as target stimuli (i.e., T, V, X, Y). Each letter was reproduced 25 times using 24-point Helvetica capital letters (Quik Stik
Downloaded from http://geronj.oxfordjournals.org/ at University of Birmingham on August 29, 2015
The present experiment was designed to extend and clarify the findings of Walsh et al. (1978). While the procedures were similar in most respects, there were notable exceptions. For example, binocular presentation was used for reasons of convenience and on the assumption that no important differences arise between monoptic and binocular masking with visual noise (cf. Turvey, 1973). (Informal pilot data seemed to confirm this.) On the other hand, target stimuli and the visual noise mask were kept very similar to those of Walsh et al. in order to facilitate a comparison of the results of the two studies. A major feature of the present research was the inclusion of a wider range of TE levels over which to examine age differences in peripheral processing. While the Walsh et al. (1978) study used only three TE values at rather low levels, the present study examined seven TE levels, three within the range used by Walsh et al. and four at considerably higher TE values. It was expected that power functions computed over the seven TE levels would provide similar, but more reliable, estimates of the TE x SOA relationship for old and young than were obtained in the Walsh et al. study with only three TE values. Furthermore, it was expected that increases in TE would continue to produce proportional changes in the speed of peripheral processing for the two age groups studied, even at high TE levels. In addition to the wider range of TE values, the present study introduced a variation in the method of estimating processing time (SOA) designed to allow greater accuracy of measurement. Thus, two sets of results were obtained. One set of data could be directly compared with the data of Walsh et al. (1978). The other set of data consisted of adjusted SOA scores calculated to reflect estimates of processing time that were equivalent with respect to their point on the psychophysical function relating percentage of accuracy of identification to SOA. Specifically, a method of ascending limits like that of the Walsh et al. study was used to establish a "coarse SOA" (four successive correct identifications). Following the determination of this coarse SOA, the percentage of accuracy of identification was assessed at several SOA values ranging from values 10 msec smaller than the coarse SOA up to the original coarse SOA itself. Thus, "adjusted
AGE DIFFERENCES IN BACKWARD MASKING
Procedure Unmasked trials. — After visual acuity was tested, subjects were seated in the darkened laboratory and were allowed 5 min to adjust to the low ambient room illumination (produced by filtered light from a 15W fluorescent bulb located out of the subject's view). The fixation field, a target letter, and the random noise mask were each shown for 20 sec to familiarize subjects with the materials. All stimuli were presented binocularly. On each trial, subjects were forced to report one of the four possible target letters. The 1.5 sec presentation of the fixationfieldwas initially followed by a 1-msec presentation of a target; a series of six targets was presented under these conditions. For each successive series, the TD was increased
by 1 msec. The experimenter determined the unmasked TD necessary for a subject to correctly identify four of the six targets in a series. Masking trials. — After the determination of the unmasked threshold, masking trials were begun. Again, all stimuli were presented binocularly. Each trial consisted of a 1.5 sec presentation of the fixation field followed by a target letter at one of seven TE levels. The target was followed by a dark interstimulus interval (blank channel) which ended with the 50 msec, 30.8 cd/m2 mask presentation. Using a method of ascending limits, the experimenter determined the ISIc needed for avoidance of masking. The initial interstimulus interval was set at 2 msec and was increased in 2-msec steps each time an error was made until the subject correctly identified four consecutive targets all at a given ISI. This method provided a measure of coarse SOA (ISIc plus 4-msec TD = SOA). Following this, the subject rested for 30 sec and was then given six series of masking trials, each containing six targets. The first series included six trials with SOA set at 10 msec less than the just-determined coarse SOA. The second series presented trials with SOA set at 8 msec less than the coarse SOA. The remaining four series approached the coarse SOA value by 2-msec steps with the last series being presented at the coarse SOA itself. The six series provided two kinds of information: (a) the percentage of accuracy of identifications at six SOA values ranging from the coarse SOA down to a value likely to be below the threshold for identification and (b) the shortest SOA at which four of the six targets in a series were correctly identified (i.e., the "adjusted SOA"). The coarse and adjusted SOAs were determined in the manner described above for each of the seven TE levels. Ten different sequences of TE levels were randomly determined, and a particular randomized sequence was used with one subject from each age group. The pace of the experimental session was determined primarily by the experimenter through the use of rest periods, but also by the subject with respect to trials needed to reach initial coarse SOA at any TE level. Thus, the entire session lasted between 70 and 90 min (approximately). Because of their higher coarse SOA values, old subjects tended to have slightly longer sessions than young subjects.
Downloaded from http://geronj.oxfordjournals.org/ at University of Birmingham on August 29, 2015
792) and placed on translucent-white Kodak Ektagraphic slides. All letters were centrally located in the visual field and subtended a visual angle of 1.11 degrees vertically by .91 degrees horizontally. The width of the strokes composing the letters subtended a visual angle of approximately .20 degrees. Letters were black against a white background. Apparent viewing distance was 680 mm. Targets were presented in a randomized order and appeared for durations of 4 msec. Seven luminance levels were used: 3.8, 4.8, 9.6, 12.3, 19.2, 30.8, and 38.5 cd/m2. The combination of these luminance levels with the 4-msec TD produced seven TE levels: 15.4, 19.2, 38.4, 49.2, 76.8, 123.2, and 154.0 cd/m2 x msec units. Random visual noise served as the mask and was a 50% black, 50% white array spread over the entire visual field. This mask was identical to the one used by Walsh et al. (1978), and it consisted of horizontal and vertical line segments subtending a visual angle of about .04 degrees in width and .10 degrees in length. In all cases, mask luminance was 30.8 cd/m2 and was presented for a 50-msec duration. An automatic three-channel tachistoscope (Scientific Prototype Model N-1000/A) with automatic stimulus changers was used. The unit provided three channels of binocular stimulus presentation in addition to a fourth channel containing only the internal fixation point. Luminance levels were measured with a Spectra Spotmeter Model UBD-10 with an internal calibration source (Photo Research). However, Kodak neutral density filters were used for reliable variation of stimulus luminance.
705
TILL
706
Table 1. Stimulus Onset Asynchrony (SOA) Required to Escape Masking (in msec) As a Function of Target Energy, Threshold Method and Age.
Target Energy (c/m2 x msec)
Coarse SOA Method Young Old
Adjusted SOA Method Young Old
18.00 4.32
26.60 9.71
14.20 2.74
21.80 6.07
123.2 Mean SOA SD
18.00 4.99
27.60 9.56
14.40 4.79
23.00 9.53
76.8 Mean SOA SD
24.60 5.89
32.00 10.24
19.80 4.56
26.40 9.56
49.2 Mean SOA SD
27.40 6.80
43.40 9.98
23.40 6.74
35.40 9.75
38.4 Mean SOA SD
32.80 8.07
43.40 12.00
27.00 6.48
38.00 10.79
19.2 Mean SOA SD
40.80 7.25
53.00 18.09
36.80 5.09
48.00 18.40
15.4 Mean SOA SD
49.60 10.01
63.00 12.97
41.80 9.45
56.20 11.90
RESULTS
An analysis of the TDs required for subjects to identify four out of six unmasked targets indicated a significant age difference between young and old subjects, /(18) = 2.923,p < .01. The mean durations were 1.20 msec for the young (with SD = 0.42) and 2.70 msec for the OLD 65 old (with SD = 1.57). YOUNG DATA FROM Table 1 presents the means and standard 55+ WALSH ET AL. deviations for the results of the masking trials; SOA values are shown as a function of age, 45 method of SOA determination, and TE level. When the standard deviations associated with coarse SOA measurement are compared with those obtained under the adjusted SOA meth25od, a small but reliable difference is found (sign test,p < .001). Although the average difference 15is only 1.00 msec, it appears that the adjusted SOA method consistently produces less vari-t—i—i—i—i—i—i—i—i—i—i—i—i—i—^ ° 20 4 0 60 80 100 120 140 ance among subjects' SOA scores than does TARGET ENERGY [Cd/m 2 x MSEC] the coarse SOA method. The effect is evident for 13 of the 14 pairs of standard deviations. Fig. 1. Comparison of present data with those of Walsh A 3-way analysis of variance was performed et al. (1978, Exp. 2). Mean SOAs for old and young are on the SOA values summarized in Table 1. plotted as a function of target energy. Data points for Each of the main effects was significant: for the present study are based on the adjusted SOA method.
Downloaded from http://geronj.oxfordjournals.org/ at University of Birmingham on August 29, 2015
154.0 Mean SOA SD
age, F(l, 18) = 12.23, p < .005; for method of SOA determination,F(l, 18) = 402.9,/? < .001; and for TE level, F(6, 108) = 62.29, p< .001. None of the interactions were significant (p > .05). In particular, the age x TE level interaction reported by Walsh et al. (1978) was completely missing from the present data, F(6,108) = 0.78. In order to pursue this discrepancy, an alternative 3-way analysis of variance was performed. Rather than using all seven TE levels, the analysis included only the three lowest levels since these fall within the range examined by Walsh et al. This alternative approach mirrored the results of the previous analysis. In particular, no age x TE level interaction emerged. Thus, one is led to question the generality of the interaction suggested by the data of Walsh et al. and to consider the alternative that the age difference in peripheral processing is constant across TE levels as it is in central processing (cf. Walsh, 1976). The evidence for an interaction (from the Walsh et al. study) as well as evidence for noninteraction (from the present study) is illustrated in Fig. 1 and will be discussed later. Since seven TE levels were used during the masking trials, it was possible that practice and/or fatigue effects might be confounded with TE levels. Although TE levels were presented in a randomized sequence, it was important to analyze for a possible order effect since the two age groups might be differentially
AGE DIFFERENCES IN BACKWARD MASKING
707
log y = -b (log x) + log k i
I
504030-
TE=15.4 o OTHER TE LEVELS -10 -8 -6 -4 -2 0 MSEC BELOW COARSE SOA
Fig. 2. Percentage of accuracy of target identification for each target energy level (averaged over age groups). Data points in a curve represent accuracy at specified SOAs below the coarse SOA. The curve for the lowest energy level is labeled separately; the other six are not since they intersect considerably.
Converted to logarithmic form, the present data would approximate a straight line relationship. Thus, linear regression analysis was performed on the (log TE, log SOA) pairs in order to determine the line of best fit. The slope of the regression line (represented by -b in the logarithmic form of the equation) is the best estimate of the exponent in the power function. Linear regression analysis of the 70 data points for each age group indicated similar exponents in the best-fit equations. For the young, TE-486 x SOA = 165.2, and for the old, TE-420 x SOA= 183.5. The fit of these equations is quite good, as indicated by the correlation between log TE and log SOA: r = -.854, p < .001 for the young, and r = -.711, p < .001 for the old.
Downloaded from http://geronj.oxfordjournals.org/ at University of Birmingham on August 29, 2015
susceptible to the effects of practice or fatigue. from coarse SOA. None of the interactions An additional analysis of variance was per- proved significant, nor was there an age difformed, therefore, to provide information ference. Distance from coarse SOA was a sigabout the adequacy of the randomization of nificant factor reflecting the improvement TE levels and to check for order effects. The in accuracy that occurred when the SOA inanalysis was based on the previously analyzed terval was increased from a time 10 msec bedata of Table 1, except that SOA values were low coarse SOA up to the coarse SOA value itrearrranged to reflect the order of testing of a self, F(5,90)i = 63.57,p .05. Power functions for old and young show similar rates of exponential decline. Since age differences in peripheral processing were not reflected in exponent differences, it is reasonable to suppose that the age difference is indicated by different constants (K) in the power functions for old and young. Indeed, Fig. 1 shows clearly that the constants differ for the two groups, on the average, since the curve for the old group is above the curve for the young group. However, statistical comparison of the constants for the two groups cannot be done as directly as it was for the exponents of the two groups. This is because the constant computed for an individual is dependent on the value of the computed exponent; slight changes in exponent value typically produce large changes in the constant. Thus, variance among the constants computed for a group is large. For example, the standard deviation for the 10 constants computed for the young group was 80.3; the value for the old subjects was 79.6. As an alternative method for examining age differences in constants, subjects in one group were compared with counterparts in the other age group who had data described by similar exponents. Matching was possible for seven pairs of subjects for which the average absolute value of the difference between exponents in a pair was only .008. The computed constants for these seven matched pairs were then examined with a /-test which indicated a significant difference, f (6) = 2.922, p < .03. The mean constant for the seven young was 137.2 (withSD = 31.2), and for the seven old, 209.2 (withSD = 63.4). Furthermore, the constant for the old subject was higher than that for the young subject in every matched pair.
AGE DIFFERENCES IN BACKWARD MASKING
difference appears similar to that obtained in Walsh's (1976) study of age differences in (supposedly) central perceptual processing. Additional research will be needed to determine whether the constant age differences in these experiments are distinctly different, or whether both are reflections of an age change with a single locus. It is interesting to compare the power functions computed for old and young with those of the Walsh et al. (1978) study. The exponents are surprisingly similar despite procedural differences, sample differences, the use of binocular presentation, and the wider range of TE values over which the exponents were computed. Peripheral processing time is clearly affected by such variables, or at least it can be — a fact reflected in the different exponents of power functions fitted to the data presented by Turvey (1973). Some of Turvey's exponents approach the values of the present study; others approach a value of 1.00 (cf. Walsh etal., 1978). It is possible that the age difference in the present study is due to constant age-related changes in the operating speed of the slowest peripheral nets outputting data on the target (cf. Turvey, 1973), while changes at earlier stages of peripheral processing produce no appreciable age differences. Alternatively, the age difference observed here may have a central locus while the minimal processing time (as seen in younger subjects) may be a peripherally determined variable dependent on target energy and other features of the stimulus. SUMMARY
Age differences in visual perception were examined by means of a binocular backward masking technique. Backward masking occurs when a target stimulus is followed so closely by a second stimulus that the subject does not have time to process the target. As the separation between target and mask is increased, a critical separation (stimulus onset asynchrony, SOA) is reached at which subjects have enough uninterrupted time to process, and therefore identify, the target. In the present study, old and young differed in terms of the SOA interval required, with older subjects needing more processing time. For young and old, processing time was characterized as a power function of target
Downloaded from http://geronj.oxfordjournals.org/ at University of Birmingham on August 29, 2015
subjects' SOA scores is smaller than it is under the coarse method. Second, information is provided about the approach to the threshold SOA (i.e., the accuracy of subthreshold values of SOA). It appears that accuracy during the approach to threshold criterion (four out of six correct) does not vary with age, but does vary with TE (especially at low TE levels). An important purpose of the present investigation was the examination of peripheral processing over a wide range of TE levels. The results obtained here replicate those of Walsh et al. (1978) insofar as main effects were found for TE and age. As illustrated in Fig. 1, however, the present data do not agree with the findings of Walsh et al. (Exp. 2) indicating a TE x age interaction. For the present data, no interaction was found across a wide range of TE values, nor was there an interaction when only the lowest three TE levels were examined (those within the range investigated by Walsh et al.). Furthermore, data from Experiment 1 of Walsh et al. were obtained with a method of SOA determination requiring seven out of eight correct identifications, and they do not indicate an age x TE interaction. One can argue that the latter results, as well as those of the present experiment, were obtained under conditions insuring the equivalence of SO As with respect to accuracy of identification on the psychophysical function relating accuracy to SOA values and that the results, therefore, involve less error variance. Of course, it is possible that an age x TE interaction occurs only at low TE levels (i.e., below the 15.4 level used here). But there is also more chance of error in this range since SOAs are long and coarse methods of SOA determination may produce threshold values that are very high on the accuracy function (as found for the 15.4 level of the present study). Given the lack of an age difference in power function exponents and the lack of a TE x age interaction (here, and in Exp. 1 ofWalshetal., 1978), it seems most parsimonious to suppose that the age difference in peripheral processing is constant across a wide range of TE values. Faced with the constant age difference in processing time, one is forced to consider that aging does not produce a change in the degree of attenuation of target energy (as indicated by similar power function exponents). Rather, it adds a constant value to the processing time measured by SOA. Furthermore, the constant
709
710
TILL
REFERENCES
Axelrod, S., & Eisdorfer, C. Senescence and figural aftereffects in two modalities. Journal of Genetic Psychology, 1962, 100, 85-91. Axelrod, S., Thompson, L., & Cohen, L. Effects of senescence on the temporal resolution of somesthetic stimuli presented to one hand or both. Journal of Gerontology, 1968,25, 191-195. Birren, J., & Botwinick, J. Age differences in finger, jaw and foot reaction time to auditory stimuli. Journal of Gerontology, 1955,10, 429-432. Birren, J., & Wall, P. Age changes in conduction velocity, refractory period, number of fibers, connective tissue space and blood vessels in sciatic nerve in rats. Journal of Comparative Neurology, 1956, 104, 1-16.
Eriksen, C , Hamlin, R., & Breitmeyer, R. Temporal factors in visual perception as related to aging. Perception and Psychophysics, 1970, 7, 354-356. Haber, R. Note on how to choose a visual noise mask. Psychological Bulletin, 1970, 74, 373-376. Kinsbourne, M., & Warrington, E. The effect of an aftercoming random pattern on the perception of brief visual stimuli. Quarterly Journal of Experimental Psychology, 1962, 14, 223-234. (a) Kinsbourne, M., & Warrington, E. Further studies on the masking of brief visual stimuli by a random pattern. Quarterly Journal of Experimental Psychology, 1962, 14, 235-245. (b) Kline, D., & Szafran, J. Age differences in backward monoptic visual noise masking. Journal of Gerontology, 1915,30, 307-311. Pollack, R. A theoretical note on the aging of the visual system. Perception and Psychophysics, 1978, 23, 94-95. Turvey, M. On peripheral and central processes in vision: Inferences from an information-processing analysis of masking with patterned stimuli. Psychological Review, 1973, 80, 1-52. Walsh, D. Age differences in central perceptual processing: A dichoptic backward masking investigation. Journal of Gerontology, 1976,31, 178-185. Walsh, D., Till, R., & Williams, M. Age differences in peripheral perceptual processing: A monoptic backward masking investigation. Journal of Experimental Psychology: Human Perception and Performance, 1978, 4, 232-243. Weale, R. On the eye. In A. Welford & J. Birren (Eds.), Behavior, aging and the nervous system. Charles C Thomas, Springfield, IL, 1965.
Downloaded from http://geronj.oxfordjournals.org/ at University of Birmingham on August 29, 2015
energy (TE), that is, TEb x SO A = K. Over a wide range of TE levels, processing time showed a similar rate of decline (similar exponent, b) for both age groups with increasing TE. As noted above, old subjects processed targets more slowly at all TE levels (different constant, K), and this age difference was constant across TE levels. While the locus of the observed age difference in processing time has not been determined, it appears that the age difference reflects changes in processing that are not a simple result of attenuation of target energy in the aging eye.
