0028-3932’90S3.00+0.00 Pcrgarnon Prer., plc
N~uropsgckolog~o. Vol. 28, p30.3, pp. 291-301. 1990 Pnnted I” Great antarn.
VISUOSPATIAL
ATTENTION: EFFECTS OF AGE, GENDER, SPATIAL REFERENCE
AND
DAVID LEEROBINSONand CAROLINEKERTZMAN Section on Visual Behavior, Laboratory of Sensorimotor Research, National Eye Institute, of Health, Building 10, Rm IOClOl, Bethesda, MD 20892, U.S.A.
National
Institutes
(Received 31 January 1989; accepted 28 September 1989) Abstract-Visual attention is remarkably stable when spatial cuing is used, but non-spatial cues lead to slowing among females and older subjects. Non-spatial cues are associated with poorer performance during the middle stages of the menstrual cycle. Motivation increased overall response speed but not attentional measures, whereas increasing age was associated with generalized slowing and directional asymmetries. Right-eye dominance was correlated with slow responses lo downward targets. These data suggest that attentional performance is modified by age, gender, and endocrine status when spatial reference is not present.
INTRODUCTION ALL ORGANISMS are faced with numerous forms of sensory stimulation and must respond to some and ignore others. Attention is one mechanism whereby the central nervous system chooses among stimuli, and this selection can be among modalities as well as within one modality. Selections can be made from many characteristics, and the choice of location in visual space has received much study recently [15, 22, 26, 34, 35, 38,411. Many of the studies of spatial attention have used a visually cued, reaction time paradigm [26]. Cues are presented, and these modify reaction times, presumably because the cues control the direction of attention.Reaction times are faster when visual targets appear at cued locations. Damage to different regions of the brain leads to unique response patterns on these tasks [22,27,28,3 1,331. In addition, we have found that a diffuse, nonspatial cue which lacks spatial reference evokes very slow responses in patients with parietal lesions [22]. Previous studies have shown that although increasing age is associated with generalized slowing, visuospatial attentional performance is not significantly changed [20]. Others have found improvements in exogenously directed attention with age [30]. These studies indicate that older individuals are more dependent on cues to shift attention. However, little is known about the influences of normal variations on attention when spatial reference is altered. It is also unknown if visuospatial attention is influenced by stages of the menstrual cycle. Several studies have shown gender-related and cycle-related differences on some tests of spatial ability as well as on reaction time tasks [3,5, 17, 18, 19, 361. In the present studies, we were interested in determining the effects of age, gender, motivation, and status of menstrual cycle on attentional performance. 291
292
DAVID LEE Roer~so~
and CAROLIVE KERTZMAN
METHOD Suh~ect selection
Twenty-four males between the ages of 26 and 49 (mean = 35 t 6 years) volunteered as subjects for our study. Fourteen men were right-eye dominant. eight were left-eye dommant. and the remaining two showed rmxed dominance. Twenty-one females between the ages of 19 and 45 (mean = 32 _+7 years) served as subjects. Twelve women were right-eye dominant. six were left-eye dominant. and three were mixed. Finally. four males and three females ranging in age from 57 to 71 years (mean=64+5 years) were tested: these constituted our older group of subjects. Five were right-eye dominant and two were left-eye dominant. The majority of the subjects had strong right-handedness; two males were clearly left-handed. All subjects had full visual fields as determined by the Humphrey 76 point screening test and visual acuity corrected to no worse than 20,X in each eye.
The procedures and apparatus used in the present study were similar to those described previously [?I?]. The subjects were seated in a darkened room in front ofa screen. Their task was to fixate on a spot of light and to press a button as soon as a target light appeared on the horizontal meridian either to the right or left of the central fixation point. The button was 9 cm long and 3 cm dia. was held freely in the dominant hand. and the subject used the thumb of that hand to depress the button. No correlation was found between the dominant hand and any measure. A cue stimulus was flashed briefly on the screen immediately preceding the presentation of the target: the Interval between cue onset and target onset varied pseudorandomly in five equal steps. The cue was a hexagon projected. in the horizontal direction, above and on the same side (valid cue) (Fig. IA, B) or on the opposite side (invalid cue) (Fig. IC, D) of the subsequent target position. For testing in the vertical direction. the cue was projected on the vertical meridian at the same location as the subsequent target or at the mirror-symmetric location in the opposite hemifield. Some!imes we presented weak illumination of the entire field of view (diffuse cue) (Fig. 1E. F). Thus. a trial consisted of the presentation of a cue followed by the presentation of the target after a variable cue -target interval; during each trial the subject gazed at the fixation point.
;nl I :
_
A
B
C
D
E
. F
Fig. I. Effects of prior instruction on attentional performance. The vertical axis represents reaction times in msec. The horizontal axis shows the six cuing conditions tested. These values are for targets which followed the onset of the cue by 180 msec. Open bars show the performance with standard instructions and those with striping are from trials where the subjects were directed to respond as fast as possible. These two blocks of trials were run serially in the same test session. Data in A and B are Jar validly cued targets to the left and right respectively, C and D for invalidly cued targets to the left and right, and E and F for diffusely cued. When left. right. up. or down are used, they refer to the location of the target. not the cue The brackets on each bar represent the standard error ofthe mean. Subjects were six of the males tested originally.
VISUAL
ATTENTION;
IN NORMALS
293
Data were collected in blocks of 160 trials each; within a test block, 70% of the trials had valid cues. 15% of the trials had invalid cues, and 15% had diffuse cues. Subjects received information about the distribution of cues and were given practice trials prior to testing. A complete experimental session consisted of four test blocks, with a brief rest between blocks. After the second block of testing, subjects completed the Edinburgh Handedness Inventory [21]. Then their ocular dominance was assessed [25]. Subjects were instructed to fixate on a target viewed through an aperture. Each eye was subsequently occluded, and the eye which continued to view the target during this monocular viewing was considered the dominant eye. Subjects received three blocks of trials along the horizontal meridian and one block along the vertical meridian within a single session. Five different testing schedules were used that presented the blocks in various orders. and the testing order was counterbalanced among the subjects. The women were tested over three sessions within a 1-3 month period. They were tested on either day 10.11. or 12 of their menstrual cycle (day IO situation): day 25.26. or 27 (day 25 situation); and one non-specific day that did not coincide with the above times (random situation). In the normal female cycle elevated levels of FSH and increasing estrogen levels occur during the day 10 situation, while high levels of progesterone occur during the day 25 situation. The subjects filled out a survey concerning the regularity of their cycle; women on birth control pills were not included. Data from the I4 females who completed all three sessions were used to evaluate the effects ofendocrine status.and data from the first session ofeach of21 female subjects were used in the evaluation of sex differences on this task.
Apparatus Subjects were seated in a comfortable, formed chair. A tangent screen made of lucite faced the subjects at a distance of 76 cm, and all images and background illumination were rear-projected [6]. Background illumination was in the photopic range at I cd/m2. Cues and targets (1.2 log units above background) were projected into a beam splitter and then onto a pair of mirrors mounted on galvanometers whose positions were under computer control. Targets which were 0.5 log units brighter than these evoked reaction times from 17 to 30 msec faster. For the data reported here, all cues and targets were 20” in the periphery. When we placed them at 5’ eccentricity. the reaction times were 20 to 33 msec faster. Button release responses were associated with reaction times from IO to 40 msec faster. None of these faster reaction times altered the differential effects of the cues. The diffuse cue (0.4 log units above background) and fixation point were projected directly onto the screen. All subjects wore earphones. and masking noise was used to eliminate auditory cues. All stimuli and data collection were controlled by an on-line digital computer system [14].
Data analysis Data were analyzed through the multivariate techniques of Hotelling’s TZ and profile analysis [13]. The profile analysis consisted of three statistically independent tests that are analogous to the different components of a twofactor repeated-measures ANOVA. Subjects’ reaction times were measured from target onset to button press under six combinations of cue type (valid, invalid, diffuse) and target location [(right and left) or (up and down)]. We excluded reaction times less than 100 msec and greater than 1050 msec from all subsequent analyses: the former were considered anticipatory in nature. and those longer than the cutoff. outliers. Less than 5% of the data were excluded by this process. A mean reaction time for each subject in each condition was calculated at each cue-target interval. We conducted a multiple profile analysis to examine differences in how the three groups responded to the various cue types and target locations. and we used the single-sample version of Hotelling‘s T* for repeated-measure designs to compare the sessions of the cycle study [lo. 131. The within-subject variables are represented by the stimulus conditions shown in Fig. 1. The parallelism test for group differences was carried out separately for the effect of cue type, target location, and the interaction of the two. The mean validity effect for a group ofsubjects was calculated as follows: for each subject we substracted the mean valid reaction time from the mean invalid reaction time. and the mean of these scores was computed to arrive at a group value. Validity effect values were obtained for only the earliest cue-target interval (180 msec)and left/right or up,‘down values were combined. Diffuse effects were calculated in a similar manner using mean valid and diEuse reaction times. A linear least-squares regression analysis was used to calculate correlation coefficients between reaction times and other variables. Probability values for two-tailed t-tests and standard deviation values are reported.
RESULTS Attentional
performance
of male,female
and older subjects
Initially, we compared the reaction times of male, female, and older subjects to assess any major effects on visual attentional performance. Table 1 shows the mean reaction times for targets presented on the horizontal and vertical meridians; the performance of the three groups was remarkably similar. The results for both meridians were similar, except for an
294
DAVID LEE ROBINSONand CAROLINEKERTZMAN
up/down asymmetry present in all groups; therefore, only the horizontal results are addressed here. These data come from the trials in which the target followed the onset of the cue by 180 msec, an interval that yields maximum efficiency of target detection at cued locations [27]. Validly cued targets were responded to faster than invalidly cued ones, F (1, 49) = 30.6, P60.001, in agreement with previous studies showing that detection of a visual target is faster if attention previously had been directed to that region of space [22,26]. There were no significant differences among the validity values (valid/invalid differences) for the three groups, F (2,49)= 1.09, P=O.34. Reaction times to validly cued targets were significantly faster than those to diffusely cued targets, F (1,49) = 122.6, P40.001. The males had a smaller valid/diffuse difference than either the females or the older subjects, F (1,49) = 4.43, P=O.O4, and F (1,49)= 19.52, P
N~uropsgckolog~o. Vol. 28, p30.3, pp. 291-301. 1990 Pnnted I” Great antarn.
VISUOSPATIAL
ATTENTION: EFFECTS OF AGE, GENDER, SPATIAL REFERENCE
AND
DAVID LEEROBINSONand CAROLINEKERTZMAN Section on Visual Behavior, Laboratory of Sensorimotor Research, National Eye Institute, of Health, Building 10, Rm IOClOl, Bethesda, MD 20892, U.S.A.
National
Institutes
(Received 31 January 1989; accepted 28 September 1989) Abstract-Visual attention is remarkably stable when spatial cuing is used, but non-spatial cues lead to slowing among females and older subjects. Non-spatial cues are associated with poorer performance during the middle stages of the menstrual cycle. Motivation increased overall response speed but not attentional measures, whereas increasing age was associated with generalized slowing and directional asymmetries. Right-eye dominance was correlated with slow responses lo downward targets. These data suggest that attentional performance is modified by age, gender, and endocrine status when spatial reference is not present.
INTRODUCTION ALL ORGANISMS are faced with numerous forms of sensory stimulation and must respond to some and ignore others. Attention is one mechanism whereby the central nervous system chooses among stimuli, and this selection can be among modalities as well as within one modality. Selections can be made from many characteristics, and the choice of location in visual space has received much study recently [15, 22, 26, 34, 35, 38,411. Many of the studies of spatial attention have used a visually cued, reaction time paradigm [26]. Cues are presented, and these modify reaction times, presumably because the cues control the direction of attention.Reaction times are faster when visual targets appear at cued locations. Damage to different regions of the brain leads to unique response patterns on these tasks [22,27,28,3 1,331. In addition, we have found that a diffuse, nonspatial cue which lacks spatial reference evokes very slow responses in patients with parietal lesions [22]. Previous studies have shown that although increasing age is associated with generalized slowing, visuospatial attentional performance is not significantly changed [20]. Others have found improvements in exogenously directed attention with age [30]. These studies indicate that older individuals are more dependent on cues to shift attention. However, little is known about the influences of normal variations on attention when spatial reference is altered. It is also unknown if visuospatial attention is influenced by stages of the menstrual cycle. Several studies have shown gender-related and cycle-related differences on some tests of spatial ability as well as on reaction time tasks [3,5, 17, 18, 19, 361. In the present studies, we were interested in determining the effects of age, gender, motivation, and status of menstrual cycle on attentional performance. 291
292
DAVID LEE Roer~so~
and CAROLIVE KERTZMAN
METHOD Suh~ect selection
Twenty-four males between the ages of 26 and 49 (mean = 35 t 6 years) volunteered as subjects for our study. Fourteen men were right-eye dominant. eight were left-eye dommant. and the remaining two showed rmxed dominance. Twenty-one females between the ages of 19 and 45 (mean = 32 _+7 years) served as subjects. Twelve women were right-eye dominant. six were left-eye dominant. and three were mixed. Finally. four males and three females ranging in age from 57 to 71 years (mean=64+5 years) were tested: these constituted our older group of subjects. Five were right-eye dominant and two were left-eye dominant. The majority of the subjects had strong right-handedness; two males were clearly left-handed. All subjects had full visual fields as determined by the Humphrey 76 point screening test and visual acuity corrected to no worse than 20,X in each eye.
The procedures and apparatus used in the present study were similar to those described previously [?I?]. The subjects were seated in a darkened room in front ofa screen. Their task was to fixate on a spot of light and to press a button as soon as a target light appeared on the horizontal meridian either to the right or left of the central fixation point. The button was 9 cm long and 3 cm dia. was held freely in the dominant hand. and the subject used the thumb of that hand to depress the button. No correlation was found between the dominant hand and any measure. A cue stimulus was flashed briefly on the screen immediately preceding the presentation of the target: the Interval between cue onset and target onset varied pseudorandomly in five equal steps. The cue was a hexagon projected. in the horizontal direction, above and on the same side (valid cue) (Fig. IA, B) or on the opposite side (invalid cue) (Fig. IC, D) of the subsequent target position. For testing in the vertical direction. the cue was projected on the vertical meridian at the same location as the subsequent target or at the mirror-symmetric location in the opposite hemifield. Some!imes we presented weak illumination of the entire field of view (diffuse cue) (Fig. 1E. F). Thus. a trial consisted of the presentation of a cue followed by the presentation of the target after a variable cue -target interval; during each trial the subject gazed at the fixation point.
;nl I :
_
A
B
C
D
E
. F
Fig. I. Effects of prior instruction on attentional performance. The vertical axis represents reaction times in msec. The horizontal axis shows the six cuing conditions tested. These values are for targets which followed the onset of the cue by 180 msec. Open bars show the performance with standard instructions and those with striping are from trials where the subjects were directed to respond as fast as possible. These two blocks of trials were run serially in the same test session. Data in A and B are Jar validly cued targets to the left and right respectively, C and D for invalidly cued targets to the left and right, and E and F for diffusely cued. When left. right. up. or down are used, they refer to the location of the target. not the cue The brackets on each bar represent the standard error ofthe mean. Subjects were six of the males tested originally.
VISUAL
ATTENTION;
IN NORMALS
293
Data were collected in blocks of 160 trials each; within a test block, 70% of the trials had valid cues. 15% of the trials had invalid cues, and 15% had diffuse cues. Subjects received information about the distribution of cues and were given practice trials prior to testing. A complete experimental session consisted of four test blocks, with a brief rest between blocks. After the second block of testing, subjects completed the Edinburgh Handedness Inventory [21]. Then their ocular dominance was assessed [25]. Subjects were instructed to fixate on a target viewed through an aperture. Each eye was subsequently occluded, and the eye which continued to view the target during this monocular viewing was considered the dominant eye. Subjects received three blocks of trials along the horizontal meridian and one block along the vertical meridian within a single session. Five different testing schedules were used that presented the blocks in various orders. and the testing order was counterbalanced among the subjects. The women were tested over three sessions within a 1-3 month period. They were tested on either day 10.11. or 12 of their menstrual cycle (day IO situation): day 25.26. or 27 (day 25 situation); and one non-specific day that did not coincide with the above times (random situation). In the normal female cycle elevated levels of FSH and increasing estrogen levels occur during the day 10 situation, while high levels of progesterone occur during the day 25 situation. The subjects filled out a survey concerning the regularity of their cycle; women on birth control pills were not included. Data from the I4 females who completed all three sessions were used to evaluate the effects ofendocrine status.and data from the first session ofeach of21 female subjects were used in the evaluation of sex differences on this task.
Apparatus Subjects were seated in a comfortable, formed chair. A tangent screen made of lucite faced the subjects at a distance of 76 cm, and all images and background illumination were rear-projected [6]. Background illumination was in the photopic range at I cd/m2. Cues and targets (1.2 log units above background) were projected into a beam splitter and then onto a pair of mirrors mounted on galvanometers whose positions were under computer control. Targets which were 0.5 log units brighter than these evoked reaction times from 17 to 30 msec faster. For the data reported here, all cues and targets were 20” in the periphery. When we placed them at 5’ eccentricity. the reaction times were 20 to 33 msec faster. Button release responses were associated with reaction times from IO to 40 msec faster. None of these faster reaction times altered the differential effects of the cues. The diffuse cue (0.4 log units above background) and fixation point were projected directly onto the screen. All subjects wore earphones. and masking noise was used to eliminate auditory cues. All stimuli and data collection were controlled by an on-line digital computer system [14].
Data analysis Data were analyzed through the multivariate techniques of Hotelling’s TZ and profile analysis [13]. The profile analysis consisted of three statistically independent tests that are analogous to the different components of a twofactor repeated-measures ANOVA. Subjects’ reaction times were measured from target onset to button press under six combinations of cue type (valid, invalid, diffuse) and target location [(right and left) or (up and down)]. We excluded reaction times less than 100 msec and greater than 1050 msec from all subsequent analyses: the former were considered anticipatory in nature. and those longer than the cutoff. outliers. Less than 5% of the data were excluded by this process. A mean reaction time for each subject in each condition was calculated at each cue-target interval. We conducted a multiple profile analysis to examine differences in how the three groups responded to the various cue types and target locations. and we used the single-sample version of Hotelling‘s T* for repeated-measure designs to compare the sessions of the cycle study [lo. 131. The within-subject variables are represented by the stimulus conditions shown in Fig. 1. The parallelism test for group differences was carried out separately for the effect of cue type, target location, and the interaction of the two. The mean validity effect for a group ofsubjects was calculated as follows: for each subject we substracted the mean valid reaction time from the mean invalid reaction time. and the mean of these scores was computed to arrive at a group value. Validity effect values were obtained for only the earliest cue-target interval (180 msec)and left/right or up,‘down values were combined. Diffuse effects were calculated in a similar manner using mean valid and diEuse reaction times. A linear least-squares regression analysis was used to calculate correlation coefficients between reaction times and other variables. Probability values for two-tailed t-tests and standard deviation values are reported.
RESULTS Attentional
performance
of male,female
and older subjects
Initially, we compared the reaction times of male, female, and older subjects to assess any major effects on visual attentional performance. Table 1 shows the mean reaction times for targets presented on the horizontal and vertical meridians; the performance of the three groups was remarkably similar. The results for both meridians were similar, except for an
294
DAVID LEE ROBINSONand CAROLINEKERTZMAN
up/down asymmetry present in all groups; therefore, only the horizontal results are addressed here. These data come from the trials in which the target followed the onset of the cue by 180 msec, an interval that yields maximum efficiency of target detection at cued locations [27]. Validly cued targets were responded to faster than invalidly cued ones, F (1, 49) = 30.6, P60.001, in agreement with previous studies showing that detection of a visual target is faster if attention previously had been directed to that region of space [22,26]. There were no significant differences among the validity values (valid/invalid differences) for the three groups, F (2,49)= 1.09, P=O.34. Reaction times to validly cued targets were significantly faster than those to diffusely cued targets, F (1,49) = 122.6, P40.001. The males had a smaller valid/diffuse difference than either the females or the older subjects, F (1,49) = 4.43, P=O.O4, and F (1,49)= 19.52, P
