Aging and the Inhibition of Spatial Location

Two experiments compared the performance of older and younger adults on a task assessing suppression (or negative priming) for location of distractors. A 3rd experiment compared the 2 age groups on suppression for location and identity of distractors such that location was irrelevant to selection and response. Older and younger adults showed location suppression across all experiments. In Experiment 3, identity suppression was found for younger but not older adults. In addition, younger adults revealed an additive effect for suppression of identity and location. Consistent evidence of inhibition of return was not found for either age group. The findings are discussed in terms of the Hasher and Zacks (1988) theory of reduced inhibitory efficiency in the elderly and in terms of neurophysiological evidence that inhibition of identity and location may function separately within the 2 cortical visual systems.

Some attention theorists have argued that selection of a target may entail active inhibition of concurrent distractors (e.g., Keele & Neill, 1978 ; Navon 1989a , 1989b ; Neumann, 1987 ; Tipper, 1985 ). One task in which such ative inhibition has been shown conforms to a suppression or "negative priming" procedure (e.g., Tipper, 1985 ); in this task, subjects are required to select a target stimulus (such as a letter, picture, or word) across a series of trials and to identify it (e.g., by naming it) while ignoring a distractor stimulus of the same class. On critical experimental trials, the target item was the item that served as the distractor on the immediately preceding trial. Subjects are slower to name the target on such experimental trials than they are on control trials in which the current target did not appear on the preceding trial. This delay in identifying a just-selected-against stimulus term is the major empirical marker of the existence of an active inhibitory process that operates in selective attention. ¹

Most of the research on inhibitory effects in selective attention has looked at suppression directed toward the identity of a distractor. College students reliably demonstrate this phenomenon under a variety of circumstances and for a variety of materials, including words, letters, pictures, numbers, Stroop items, and unfamiliar forms (e.g., Allport, Tipper & Chmiel, 1985 ; DeSchepper & Treisman, 1991 ; Driver & Tipper, 1989 ; Lowe, 1979 ). The importance of this mechanism is substantial. For example, Hasher and Zacks (1988) outlined a general theory in which inhibitory mechanisms of selective attention play a critical role in a range of behaviors including remembering, discourse comprehension, and speech production (see also Gernsbacher, 1990 ). Recent work ( Stoltzfus, Hasher, Zacks, Ulivi, & Goldstein, 1993 ) suggests that inhibition of distractors may play a critical role in creating and maintaining coherent streams of thought and action directed toward targets.

In this light, one intriguing finding is that older adults do not demonstrate suppression under the same circumstances as do younger adults ( Hasher, Stoltzfus, Zacks, & Rypma, 1991 ; Kane, Hasher, Stoltzfus, Zacks, & Connelly, 1993; McDowd & Oseas-Kreger, 1991 ; Stoltzfus et al., 1993 ; Tipper, 1991 ). Indeed, to our knowledge, older adults do not demonstrate a reliable suppression effect in selection tasks that require a response to the identity of a target. These findings are taken as evidence consistent with the view of Hasher and Zacks (1988) that there is an age-related deficiency in inhibitory attentional mechanisms involved in the selection process. One question raised by that theory concerns the generality of the inhibitory deficit.

Recent evidence suggests that inhibition operates not just on the identity of a distractor item but also on the location that a distractor occupies in a visual display ( Tipper, Brehaut, & Driver, 1990 ; Tipper & McLaren, 1990 ; Tipper, Weaver, Kirkpatrick, & Lewis, 1991 ; see also Wolfe & Pokorny, 1990 ). These two inhibitory effects, addressed to the identity versus the location of distractors, may not represent the same underlying system; indeed, neurophysiological studies suggest the presence of at least two separate inhibitory systems, as there are at least two cortical visual pathways that send information to frontal cortex ( Ungerleider & Mishkin, 1982 ). One of these specialized pathways, the ventral or occipito-temporal pathway, passes through the inferior temporal lobe and processes the identity of the objects. Neurons in inferior temporal cortex may serve working memory by attenuating responses to stimuli that have previously been presented, whereas responses to new stimuli are not so affected ( Miller, Li, & Desimone, 1991 ). The responses of these neurons are specific to an object's shape, independent of its location in the visual field ( Desimone, Albright, Gross & Bruce, 1984 ). The other pathway, the dorsal or occipitoparietal pathway, passes through the posterior parietal area and processes spatial location.

Evidence that supports the hypothesis of a dual nature of the visual system comes from several rather disparate literatures, including assessments of clinical populations and neural modeling. Farah, Brunn, Wong, Wallace, and Carpenter (1990) noted that damage to the dorsal pathway causes the neglect syndrome; patients with this syndrome can identify stimuli presented contralateral to their lesions but are unable to attend to them. Researchers who have tried to model the visual system have found that it is computationally more efficient to segregate visual processing into two separate streams on the basis of identity and location than it is to combine object and location information in a single processing system ( Rueckl, Cave, & Kosslyn, 1989 ). Because parietal lobe neurons have differing locations of maximal response and large, overlapping receptive fields ( Motter & Mountcastle, 1981 ; Motter, Steinmetz, Duffy, & Mountcastle, 1987 ), they can be modeled to produce a system that effectively encodes spatial location ( O'Reilly, Kosslyn, Marsolek, & Chabris, 1990 ). In contrast, neurons in the inferior temporal lobe respond maximally to stimuli that fall on the fovea ( Andersen, Asanuma, Essick, & Siegel, 1990 ); as such, they are well suited to identifying and encoding objects but not to encoding their spatial location.

Behavioral data support the view that there are separate inhibitory systems for identity and location selection. Tipper, Borque, Anderson, and Brehaut (1989) demonstrated that although second-grade children fail to show suppression to the identity of a distractor, they do indeed show suppression to its location ( Tipper & McLaren, 1990 ). Given the importance accorded inhibitory mechanisms in recent empirical and theoretical work (e.g., Gernsbacher, 1990 ; Hasher & Zacks, 1988 ), we considered the possibility that older adults, who have yet to demonstrate suppression toward the identity of a distractor, may well be able to show suppression for the location of a distractor. Indeed, evidence from selective attention research in cognitive gerontology might be taken to suggest the sparing of location suppression for older adults: Age differences in selective attention are minimized under circumstances in which targets and distractors are in predictable as compared with unpredictable locations (e.g., Madden, 1983 ; Nissen & Corkin, 1985 ; Plude & Hoyer, 1985 ). One value of considering attention developmentally is analogous to that derived from studies that compare cognitive functions, such as memory and speech comprehension and production, across normal and neurologically impaired groups (e.g., amnesics, aphasics): Theories may be built or constrained by comparative findings (e.g., Schachter, Kaszniak, & Kihlstrom, 1991 ). The data reported here may serve just such a function.

In two experiments we assess the presence of location suppression in older adults using a task first introduced by Tipper and McLaren (1990) . By also testing college-aged adults, we are able to assess the existence of age differences in location suppression. In the present experiments, older adults consistently show location suppression and its magnitude is not detectably different from that shown by younger adults. ² These findings stand in sharp contrast to others showing substantial age differences for identity suppression. In a third experiment, we include a direct, within-subject comparison of identity versus location suppression for younger and older adults.

As a secondary question, two experiments also include conditions to assess the presence of an effect variously called inhibition of return ( Posner & Cohen, 1984 ) and attended repetition ( Tipper et al., 1990 ). This is an effect in which subjects are slower to respond to a target on one trial that reappears in the same location it occupied on the immediately preceding trial. The inhibition of return mechanism is thought to play an important role in maintaining spatial selectivity ( Maylor & Hockey, 1985 ; Posner & Cohen, 1984 ) by biasing the subject against returning to locations from which target information has already been extracted. The location inhibition mechanism may, by analogy, function to constrain attention away from locations in which distractors have recently appeared. Together, these two locational inhibitory mechanisms may orient the viewer toward extracting information from new locations.

Following Tipper and McLaren (1990) , the experiment involved a paired trials procedure, consisting of a prime followed by a test or probe trial. On all trials, subjects indicated the location of a target (an "O") appearing in one of four locations on a monitor by pressing a key on the computer keyboard that corresponded to the target's location. A distractor (a plus sign) occurred on most prime trials and all probe trials.

The target and a distractor, if present, could occur in four possible positions, arranged in a broad, flat V-shape (see Figure 1 ). The four locations were underscored by white stickers centered on a monitor such that the widest horizontal distance between two positions (the top of the "V") measured 45 mm and subtended approximately 9.46° of visual angle. The smallest horizontal distance measured 17 mm (subtending 3.6°), and the vertical distance measured 12 mm (subtending 2.54°). The stimuli themselves (a white "O" and a plus sing, in 40-column text mode) subtended 1.49° and 1.17° vertically and 0.85° and 1.17° horizontally, respectively. A small white cross (subtending 1.17° horizontally and vertically) was centered in the middle of this display and used as a fixation point. The background color for all displays was black. The display of all the stimuli (except for the four location markers that were affixed to the monitor) and the timing of responses were both controlled by programming done for an IBM-compatible 286 computer, with an Enhanced Graphics Adaptor (EGA) monitor.

In all instances, the critical dependent measure was the amount of time that elapsed before subjects correctly identified the target's location. Subjects responded on the computer's keyboard using four response keys (D, C, K, M), which mapped spatially onto the four stimulus locations. The keys were covered with blank, white stickers.

Across pairs of prime and probe trials, three types of effects were observed: interference, suppression, and inhibition of return. Interference was measured by comparing two types of prime trials, one on which both the target and distractor appeared (distractor-present trials), and one on which only the target appeared (target-only trials). The difference in reaction time between these two types of trials was a measure of the degree to which the presence of a distractor interfered with responses to the location of the target.

Inhibition of return and suppression effects were measured on probe trials. Inhibition of return was measured on those probe trials that followed target-only prime trials. On critical probe trials, the target reappeared in the same location on the probe trial as it had occupied on the prime trial (location-repeat trials). On location-control probe trials, both the target and the distractor reappeared in locations that were not occupied on the preceding prime trial. The difference in response times between these two conditions was a measure of the degree to which responding to the location of the target was disrupted by its reappearance in the same location across two successive trials.

Suppression was measured on those probe trials that followed distractor-present prime trials. On critical probe trials (suppression trials), the target occupied the same location as the prime trial's distractor. On suppression-control trials, the target and distractor appeared in a location not used on the previous trial. The difference in response times between these two conditions was a measure of the degree of suppression tied to a distractor's location during the selection of a target.

The entire experimental series consisted of 384 paired trials randomly presented to a subject. There were 96 target-only prime trials and 288 distractor-present prime trials. The target-only trials were followed by 24 location repeat trials and 72 location control trials, such that the probe trial target had an equal (25%) chance of appearing in any of the four possible target locations during a target-only prime trial.

The same probability relationship was true of distractor-present prime and probe trial pairs, that is, given a distractor-present prime, the probe target would be equally likely to appear in any of the four possible locations. There were 48 pairs of each type of critical probe trial (location suppression and control). Following the reaction time task, vocabulary skills were measured with a difficult multiple-choice test, the Extended Range Vocabulary Test (ERVT; Educational Testing Service, 1976 ).

Subjects were tested individually. They were seated in front of the computer at a distance of approximately 27 cm from the screen. Subjects were shown the stimulus display and the keyboard and were instructed how to map their key presses to the visual display to indicate the location of the target. For all subjects, the left and right index fingers were positioned over the C and M keys, while the left and right middle fingers were positioned over the D and K keys. Subjects were informed that every display would contain only one target but that some would also contain a distractor in one of the three remaining locations. Participants were instructed to ignore the distractor as it was irrelevant to the task and to try to be as quick and as accurate as possible in responding to the location of the target.

Each prime—probe trial pair began with the screen display "Ready?" positioned 16 mm above the topmost location markers. The subject then pressed the space bar to remove the "Ready?" signal from the screen and to initiate the prime trial. A fixation cross appeared 1,500 ms after the offset of the prompt and remained onscreen for 500 ms. Immediately after fixation offset, the prime display appeared and remained on-screen for 150 ms, after which the screen turned blank and remained so until the subject responded. This was immediately followed by the appearance of another fixation cross on-screen for 500 ms, which was then replaced by the probe display for 150 ms. Again, the screen blanked until the subject responded, at which time the "Ready?" display appeared again and the sequence was repeated. The testing session began with 48 pairs of prime—test practice trials, and the vocabulary test was administered after the subject completed the attention task.

The data from one older male subject were dropped from analysis because a recent stroke prevented the subject from responding with both hands. The remaining 19 older subjects had more years of education ( M = 15.68) than did the 19 younger subjects ( M = 12.90), t (36) = 4.75. Older and younger subjects did not differ in their vocabulary scores ( M s = 32.57 and 27.41 of 48, respectively), t (36) = 1.70.

Throughout the experiment, the basic dependent measure was each subject's median reaction time on correct trials in each condition. If an error was made on either member of the prime—probe trial pair, the reaction time for both members was deleted from consideration. Furthermore, the basic analysis of variance (ANOVA) used throughout was a mixed 2 (age: younger vs. older) × 2 (condition), with repeated measures on the appropriate conditions. Throughout, the p value was set at .025. The critical conditions for measuring interference occur on prime trials when the target and distractor are both present versus when only the target is present. These values, along with errors, are shown in Table 1 . Older subjects were slower to respond than younger subjects, F (1, 36) = 35.74, MS _e = 13,729.16. The interference effect was reliable, with subjects slower when the distractor was present than when it was absent, F 1, 36 = 33.89, MS _e = 115.54 . There was no interaction, indicating that the interference effect for younger adults (15 ms) did not differ from that for older adults (14 ms), F < 1.

Suppression effects were measured on probe trials by comparing younger and older subjects' reaction times on suppression and suppression control trials. Again, older subjects responded more slowly than younger subjects, F 1, 36 = 43.61, MS _e = 13,706.93 . The suppression effect was reliable, with slower responses occurring during suppression trials than during control trials, F 1, 36 = 39.52, MS _e = 386.55 . What is particularly notable here is the absence of an Age × Condition effect ( F < 1). Both older and younger adults show reliable location suppression and they do so to an equivalent extent, 28 ms for each group. It is also worthy of note that neither older nor younger subjects showed a significant correlation of interference and suppression, with r s of .07 and -.32, respectively.

The inhibition-of-return effect was examined by considering response times on location-repeat trials and location-control trials. ³ Again, older subjects were slower than younger subjects, F 1, 36 ) = 45.66, MS _e = 11,242.25 . There was a significant effect of condition, F 1, 36 = 10.35, MS _e = 685.24 , although an inspection of the means reveals that this effect is not inhibitory in nature (see Table 1 ). The interaction of age and condition approached significance, F 1, 36 = 3.85, MS _e = 685 .24, p = .057, as a result of a large and significant facilitatory effect shown by older adults (31 ms; F 1, 18 = 7.91, MS _e = 1,161.83 . The younger adults' inhibition-of-return effect was not significant ( F < 2.6).

Error data are treated as the reaction time data: separate 2 (age) × 2 (condition) mixed ANOVAs, with repeated measures on conditions, were conducted for each of the three effects of interest, interference, suppression, and inhibition of return. For the interference data, only one effect, that of age, F 1, 36 = 8.39, MS _e = 0.0002 , was significant. Surprisingly, older adults actually made fewer errors on prime trials than did younger subjects.

There was only one significant effect for the suppression data, the interaction of age and condition, F 1, 36 = 4.13, MS _e = 0.000112 . Further analysis revealed that this resulted from the tendency of younger adults to make more errors on suppression than on suppression control trials, F 1, 18 = 4.01, MS _e = 0.000182, p = .061 , whereas there was no difference in the number of errors older adults made in those two conditions, F < 1. The analysis of the attended repetition error data revealed no significant effects. These findings suggest that interpretations of the reaction time data are not compromised by speed/accuracy trade-offs.

Both younger and older adults clearly show a location suppression effect; targets in locations that were just occupied by a distractor were indicated more slowly than targets in locations that were unoccupied. Thus, both older and younger adults appear to suppress responding to the position in which a distractor was located in the course of selecting the target, identifying its location, and responding with its position. Indeed, the extent of suppression seen here for older adults is not discriminably different from that seen for younger adults. This finding contrasts strongly with the absence of identity suppression effects seen elsewhere for older adults (e.g., Hasher et al., 1991 ; Kane et al., in press ; Stoltzfus et al., 1993 ). Thus the results of this experiment suggest the presence of a spared inhibitory function, that addressed to the location of distractors.

We failed to demonstrate the existence of an inhibition of return effect for either younger or older adults. With respect to younger adults, the lack of an inhibition of return effect seen here is consistent with two reported failures to find it, both involving the present task (Experiments 1b and 4 of Tipper et al., 1990 ). It is worth noting that we were also unable to produce reliable inhibition of return effects in two pilot studies; indeed, when reliable effects were obtained, they were facilitatory in all but one case. S. P. Tipper (personal communication, November 10, 1992) pointed out that the pattern of data obtained both in our lab and in his own ( Tipper et al., 1990 ) could be taken to suggest that older adults have less inhibition of return than do younger adults, as older adults consistently have larger facilitatory effects than we have obtained with younger subjects. Of course, it also is possible that older adults are slower to remove attention from the attended location. At present, the data do not discriminate between these two hypotheses.

In Experiment 1, older adults showed reliable suppression effects in the course of selecting a target and identifying its location. Indeed, the amount of suppression shown by younger and older subjects was nearly identical. These findings contrast sharply with those in earlier studies in which suppression was assessed for the identity of a distractor ( Hasher et al., 1991 ). Several researchers reported that older adults do not suppress the identity of distractors ( McDowd & Oseas-Kreger, 1991 ; Tipper, 1991 ). However, differences between experiments that measure suppression of distractor identity and those that measure suppression of distractor location are considerable. First, there are differences in materials. For example, both the number of different items used and the meaningfulness of those items vary. In location experiments, subjects typically choose between two items, with the target and distractor remaining the same across all trials. In identity experiments, a minimum of about nine different items are used (e.g., Yee, 1991 ), and the same items serve as both targets and distractors.

There are procedural differences as well. One particularly notable procedural difference between Experiment 1 of our study and most identity studies lies in the response mode, with manual, spatially mapped responses used here and in other location tasks ( Tipper & McLaren, 1990 ; Tipper, Weaver, Cameron, Brehaut, & Bastedo, 1991 ), whereas vocal (typically naming) responses have been used in identity tasks ( Hasher et al., 1991 ). Our first concern, then, was with the generality of location suppression effects across response modes; thus, in a second experiment, we translated the standard manual procedure for assessing suppression of location to a vocal one. In fact, when we started this experiment, there was no evidence as to whether suppression of location would be seen outside of the spatially mapped manual response mode. Since completing this study, such evidence has become available; Tipper, Weaver, Kirkpatrick, & Lewis, (1991) have shown that for young adults at least, the suppression effect occurs when a target occupies the location of the previous distractor, even when a vocal response is used. The effects reported by Tipper, Weaver, Kirkpatrick, and Lewis (1991) are, however, variable. In one instance (Experiment 1), suppression or negative priming was not found with a vocal response, although in another (Experment 3) it was. Even in the latter case, however, the effect was small and unstable across blocks of trials. Thus, the present experiment reassesses their findings with younger adult participants and extends the question of the persistence of suppression of location using a naming response to older adults. We note particularly that older adults have hitherto failed to show suppression when a vocal response mode was used. Of course, in all such vocal response studies, the identity of the target, not its location, was the focus of the response ( Hasher et al., 1991 ; Kane et al., in press ; McDowd & Oseas-Kreger, 1991 ; Stolzfus et al., 1993 ; Tipper, 1991 ). In the present study, it is the location of the target to which the older subject must make a vocal response.

Eighteen younger adults (range = 18—21 years, M = 18.83) and eighteen older adults (range = 65—78 years, M = 70.67) participated. Subjects were recruited as in Experiment 1, and no individual served in that experiment. The data from one younger participant were dropped because of mechanical problems during testing.

Experiment 2 was similar to Experiment 1, except that the subject was required to indicate location using a vocal response. The stimuli were similar to those used in the first experiment, save for a few aspects of the screen display. In the present experiment, a green equal sign, rather than the white plus sign used in Experiment 1, was used to indicate the fixation point. The new fixation point subtended an angle of 1.06° horizontally and 0.64° vertically.

Numbers were placed on the white stickers that marked the four possible target locations. These numbers were arranged such that the top left location was labeled 1, the bottom left was labeled 2, the bottom right was labeled 3, and the top right was labeled 4. Subjects were instructed to say aloud the number that corresponded to the location of the target. This procedure is similar to that independently introduced by Tipper, Weaver, Kirkpatrick, and Lewis (1991) . Subjects were also told that a green equal sign would appear in the center of the display shortly before a trial was to begin and that they could improve their performance by focusing on it. The sequence of events on each trial was otherwise identical to that in Experiment 1 with the exception of training on the use of the voice key and on naming of the numbers of the corresponding locations.

The entire experimental series consisted of 152 paired trials. There were 56 target-only prime trials and 96 distractor-present prime trials. The target-only trials were followed by two types of test trials, 20 of which were location-repeat trials and 36 of which were location-control trials. The 96 distractor-present prime trials were followed by two types of test trials, 48 each of suppression and suppression-control trials. ⁴ The different sorts of trials were combined randomly for each subject. The experimental session began with 20 pairs of prime—test practice trials, and the vocabulary test was administered after the subject completed the attention task.

Older subjects had more years of education ( M = 15.28) than younger subjects ( M = 12.59), t (33) = 6.29. The two groups did not differ in terms of vocabulary performance ( M s = 32.47 and 29.40, for older and younger adults, respectively), t (33) = 1.21.

This was examined on prime trials. As can be seen in Table 2 , older adults were slower overall than younger adults, F 1, 33 = 17.81, MS _e = 16,502.13 . Subjects were faster on target-only trials than on distractor-present trials, F 1, 33 = 47.22, MS _e = 447.59 . The Age × Condition interaction was not significant, F = 1.48. Both older and younger subjects were slowed by the presence of distractors in a prime array.

Again, older adults were slower than younger adults, F 1, 33 = 17.66, MS _e = 17,222.29 . Subjects were slower on suppression trials than on control trials, F 1, 33 = 6.22, MS _e = 316.30 . The interaction between age and condition was not significant, F < 1, indicating that the suppression effect for younger adults (9 ms) did not differ from that for older adults (12 ms). These effects, like those reported by Tipper, Weaver, Kirkpatrick, and Lewis (1991) for vocal responses, are smaller than those seen in Experiment 1 and in our pilot studies (see ^{Footnote 4} ). Nonetheless, the overall suppression effect was reliable. Once again, as in the first study, the correlation between interference and inhibition failed to reach significance for either younger ( r = -.11) or older ( r = .12) subjects.

Older adults were slower on these trials than were younger adults, F 1, 33 = 16.05, MS _e = 17,912.81 . No other effects were significant ( F s < 1). Once again, we failed to find a reliable inhibition-of-return effect.

Neither the interference nor the suppression effects reported above were compromised by speed/accuracy trade-offs. On trials designed to assess interference, older adults made more errors than younger adults, F 1, 33 = 5.55, MS _e = 0.0002 . Errors were not different across the two trials on whiich interference was assessed, nor was there an interaction between conditions and age of participants, largest F < 1. On trials used to assess suppression, the pattern of results was similar. Older subjects, tended to make more errors than younger subjects, F 1, 33 = 3.102, MS _e = .001, p = .087 , but there was neither an effect of trial type, nor an interaction between age and type of trials (all F s < 1.26). When errors made on inhibition of return and their control trials were examined, older adults made more errors than younger adults, F 1, 33 = 6.25, MS _e = .001 . Once again, there was neither a main effect of condition, nor an interaction between age and condition, largest F 1, 33 = 3.03, p = .09, MS _e = 0.041 .

The data from this experiment clearly demonstrate that young adults can show reliable inhibition to location even when tested with a vocal response. These findings thus strengthen those independently reported by Tipper, Weaver, Kirkpatrick, and Lewis (1991) , which show a large location suppression effect for younger adults who are tested manually, and a small suppression effect for those tested vocally.

The data from the present experiment also clearly show that the findings of preserved inhibition of spatial location in older adults are not simply an artifact of response modality, with older adults showing suppression only when responding in a manual mode. Instead, it seems clear that older adults can show suppression for a distractor's location whether tested manually or vocally. Thus, the failure of older adults to demonstrate identity suppression in previous studies that required a vocal response was probably not due to the particular response mode used in those studies. The present finding, then, is consistent with the suggestion that inhibition to location and to identity may well be separate mechanisms following different time courses in development.

Across two experiments we now have evidence that older adults are able to suppress location as well as younger adults can. These findings may be compared with others (e.g., Hasher et al., 1991 ; Stoltzfus et al., 1993 ) in which older adults, unlike younger adults, appear to be unable to suppress the identity of irrelevant stimuli. Such findings suggest the possibility of a dissociation between identity and location suppression. Stronger evidence of a dissociation is provided in the present experiment, in which the same subjects are tested on both identity and location suppression. In this study, subjects named a target letter (on the basis of its color, red or green), which (a) appeared where a distractor had been located on the prime trial, and also shared that distractor's identity; (b) appeared where the distractor had been, but did not share its identity; (c) shared identity but not location with the previous distractor; or (d) shared neither location nor identity with the preceding distractor.

This experimental procedure enabled us to address two additional empirical questions: First, what happens when the two types of inhibition co-occur, that is, when a distractor's location and its identity are both reused by the new target? Evidence for independent mechanisms might suggest that the two effects should at least be additive. Second, will location suppression be shown when location is largely irrelevant to the task? Note that in the task used in Experiments 1 and 2 (and also elsewhere in the literature on location suppression), subjects selected the target on the basis of its identity and responded on the basis of its location. Thus the location of stimuli was a salient attribute of the task. In the present experiment, the targets and distractors were red and green letters and subjects selected the target on the basis of its color; when the target was green, the distractor was red, and vice versa. Location was not critical for selection or response, because subjects named the letter, not its location. To our knowledge, this is the first experiment examining location suppression in which location is not a relevant aspect of either selection or response. It was conceivable that the location of distractors might be suppressed only when the task makes it relevant to either selecting or responding. Experiment 3 addresses this possibility and provides a within-subject test of the hypothesis that, for older adults, there is a dissociation between identity and location suppression.

Twenty-four younger adults (age range = 18—21, M = 18.71) and 23 older adults (age range = 60—76, M = 65.87) participated in this study. Subjects were recruited from the same sources as in previous experiments; no subject served in more than one study.

As in the previous experiments, the present study involved a paired prime-trial/probe-trial procedure. Targets and distractors appeared above one of four sticker locations arranged on the computer screen exactly as in Experiments 1 and 2. On each prime or test display, subjects named aloud one of two letters (selected from the set of A, B, C, D, E, J, K, N, O, S, T, and V used previously by Hasher et al., 1991 ), rather than making a keystroke response to a target location. The letters were green and red; when the target letter was red the distractor was green, and vice versa. The stimuli themselves subtended 0.96° horizontally and 1.27° vertically. As in Experiment 1, a white fixation cross subtended 1.17° horizontally and vertically.

The critical dependent measure was the time required to name the target letter correctly. Half of the subjects named the green letter, half of them the red. Three types of effects were observed in Experiment 3: location suppression, identity suppression, and the combined effects of location and identity suppression. Location suppression was examined by comparing reaction times on control probe trials (in which letters used on prime and probe trials shared neither identity nor location) to reaction times on location suppression trials in which the target on the probe trial appeared in the location occupied by the distractor on the prime trial, and no letter appeared more than once in each prime—probe pair.

The suppression of identity effect was calculated by comparing reaction time on control probe trials (in which letters used on prime and probe trials shared neither identity nor location) with reaction time on identity suppression trials. Identity suppression trials consisted of prime—probe pairs in which the distractor on the prime trial became the target on the probe trial, and no location was occupied on both prime and probe trials.

The last effect of interest, the combination of location and identity suppression, was measured by comparing reaction time on control probe trials to reaction time on combined identity and location suppression trials. On this type of trial, the target on the probe trial and the distractor on the prime trial have the same identity and occupy the same location (see Figure 2 ).

The experiment proper consisted of 192 pairs of trials randomly presented to a subject. There were 48 pairs of each type of trial (location suppression, identity suppression, location plus identity suppression, location plus identity suppression, and control). Stimuli were arranged so that in each condition, each letter appeared only once as a target and once as a distractor in each of the four locations on the prime trial and on the probe trial. Interference was not measured in this experiment. Following the reaction time task, subjects were administered the ERVT, as in the previous experiments.

Subjects were tested individually and were seated in a dimly lit room about 27 cm away from the computer screen. Once seated, they were shown the stimulus display and were given instructions about using the voice key. They were then instructed to identify the (red or green) target letter as quickly and as accurately as possible and to ignore the distracting letter.

The timing of stimulus displays was identical to that of Experiment 1, except for stimulus duration, which was increased to 200 ms in the present experiment. ⁵ As in Experiment 2, the testing session began with 20 pairs of prime—test practice trials, and the vocabulary test was administered after the subject completed the attention task.

One younger subject was dropped from analysis because of an equipment failure during testing. Three younger subjects were excluded from further analysis because they indicated their awareness of the identity relationship between the distractors and targets on the identity suppression and on the identity and location suppression trials. Previous research ( Allport et al., 1985 ; Hasher et al., 1991 ; Neumann & DeSchepper, 1992 ) has shown that under such circumstances subjects occasionally use a very different strategy to perform the task, one that makes it inappropriate to compare their data to those of unaware subjects (cf. Neill & Valdes, 1992 ). The data from 3 older subjects were dropped because each had an error rate of greater than 20%.

The remaining 20 older subjects had more years of education ( M = 15.25) than did the remaining 20 younger subjects ( M = 12.55), t (38) = 4.35. Older and younger subjects did not differ in their vocabulary scores ( M s = 25.26 and 25.09 out of 48, respectively), t (38) < 1.

Analyses were conducted on each subject's median reaction time for correct trials in each condition. As in previous experiments, an error made on either member of the prime—probe trial pair resulted in the deletion of the reaction time for both members. The data, presented in Table 3 , were first subjected to an overall mixed 2 (age: younger vs. older) × 4 (condition) ANOVA, with repeated measures on within-subject conditions. This analysis revealed that older subjects were slower to respond than were younger subjects, F 1, 38 = 5.5, MS _e = 23,791.56 , and that there was a main effect of condition, F 3, 114 = 41.76, MS _e = 242.05 . This effect of condition was reliable within the two age groups, F 3, 57 = 22.27, MS _e = 185.93 and F 3, 57 = 21.46, MS _e = 298.16 , for younger and older subjects, respectively.

Because specific predictions were made about the nature of the differences among conditions for the different age groups, planned contrasts were used to explore the differences among condition means. We note, however, that the interaction of age and condition in the omnibus test did not reach significance, F 3, 114 = 1.78, p = .15, MS _e = 242.00 , (an unsurprising finding given the pattern of means in older and younger adults, i.e., the age groups looked similar in two of three conditions). These contrasts revealed that, for younger adults, there were significant effects of location suppression, F 1, 19 = 19.46, MS _e = 504.19 ; identity suppression F 1, 19 = 4.72, MS _e = 243.24 ; and identity and location suppression, F 1, 19 = 54.98, MS _e = 373.66 . The combination of identity and location suppression resulted in a larger suppression effect than did location or identity suppression alone, F 1, 19 = 6.91, MS _e = 283.59 , and F 1, 19 = 54.98, MS _e = 373.66 . We note that young adults showed a trend toward a correlation of location suppression with identity suppression ( r = .41, p = .077).

For older adults, a significant location suppression effect was evident, F 1, 19 = 23.08, MS _e = 656.67 . Consistent with earlier work, however, the identity suppression effect was insignificant, F < 1. Although older adults also showed a significant location and identity suppression effect, F 1, 19 = 37.55, MS _e = 520.15 , there was no difference between the effect of location suppression alone and the effect of location plus identity suppression, F 1, 19 < 1 . This latter finding is consistent with previous work on identity suppression, which suggested that older adults fail to suppress the identity of a distractor item. Furthermore, responses from older adults do not show a correlation between location suppresion and identity suppression, r = -.15.

The percentage of errors made on probe trials in each condition was tabulated for each subject, and these data were subjected to a mixed 2 (age) × 4 (condition) repeated measures ANOVA, as were the reaction time data. This analysis revealed that older subjects made more errors than did younger subjects, F 1, 38 = 31.37, MS _e = 0.002 , and that there was no effect of condition, F 3, 114 = 1.82, p = .148, MS _e = 0.001 , and no interaction of age with condition, F 3, 114 = 1.34, p = .265, MS _e = 0.001 . The results of this analysis, taken together with an inspection of the error rates presented in Table 3 , make it extremely unlikely that the reaction time data are comparomised by speed—accuracy tradeoffs.

The primary results from this experiment are quite clear: Younger adults show not only suppression of identity and suppression of location but also an effect of the combination of identity and location suppression that appears to be additive in nature. By contrast, older adults show suppression of location but, consistent with earlier findings (e.g., Hasher et al., 1991 ), exhibit no evidence of being able to suppress the identity of an item. For older adults, the magnitude of the effect of identity and location suppression is equivalent to the location suppression effect alone, thus providing firm, within-subject evidence that older adults can inhibit locations occupied by distractors but do not inhibit the identity of those distractors. Finally, the present experiment also clearly demonstrated that location suppression is obtained for both age groups, even though the location of an object was neither a selection nor a response dimension of the task.

The data from this experiment speak to an issue raised in the literature on negative priming (e.g., Neill, Lissner, & Beck, 1990 ; Tipper, Weaver, Cameron, Brehaut, & Bastedo, 1991 ) and in the review process as well. This issue concerns the nature of inhibition, and has centered on whether inhibition accrues to an item (or location) or to the response made to that item (or location). Several sets of findings suggest that inhibition is not strictly associated to a particular response. For example, Neill et al. (1990) used a task required a same/different matching response and found that young subjects showed suppression for the identity of distracting stimuli even when the response (same or different) was different between two successive trials (see also DeSchepper & Treisman, 1991 ). Tipper, Weaver, Kirkpatrick & Lewis (1991) showed suppression when different response modes (vocal vs. manual) were used across two successive trials. Together, these findings suggest that some more central representation of the stimulus or of the preparation for response, rather than just the response, was suppressed.

Taken together, the data from Neill et al. (1990) , DeSchepper and Treisman (1991) , and Tipper, Weaver, Kirkpatrick, and Lewis (1991) support the idea that inhibition does not act on the response alone, but they can be seen as somewhat equivocal. In the task used by Neill and colleagues, supression of identity is tested, and although identity is not an explicit part of the response, it must be completely processed for subjects to respond correctly. The procedure used by Tipper, Weaver, Kirkpatrick, and Lewis (1991) showing suppression for location whether the response was manual or vocal, makes the strongest case for central, rather than response based suppression; nevertheless, even in this paradigm subjects responded to the location of stimuli and suppression of location was assessed. Yet, it is possible that suppression is response specific in the sense that suppression is only seen when the response required must access the stimulus attribute that has been suppressed.

The data from the present experiment support even more strongly the idea that inhibition operates on stimulus characteristics rather than on responses. In this study, subjects selected a target on the basis of its color and responded on the basis of its identity. Location was not made a salient dimension of the task (and was surely irrelevant to the response), yet both younger and older participants were significantly slowed in naming a target when it appeared where a distractor had occurred on the previous trial. Here then, is the first clear demonstration of suppression accruing to a stimulus-related attribute (location) when that attribute is not part of the response.

Across three experiments, one involving a manual, spatially mapped response and two involving a vocal response, the data on location suppression are quite clear: Subjects were slower to identify a location containing the target (in Experiments 1 and 2) or to identify a stimulus (in Experiment 3) in a particular location if that location had just previously contained the distractor. The system seems clearly biased away from an immediately preceding location, if irrelevant information were located there. The bias, the suppression effect, appears present to the same degree for older and younger adults across all three experiments. These results for the reuse of a selected location stand in stark contrast to findings from Experiment 3 and elsewhere, suggesting that older adults do not suppress the identity of distractors ( Hasher et al., 1991 ; Kane et al., in press ; McDowd & Oseas-Kreger, 1991 ; Stoltzfus et al., 1993 ; Tipper, 1991 ). That is, older adults do not seem to respond more slowly when the new, to-be-named target is the object that was previously the distractor, whereas for younger adults there is reliable slowing under these same circumstances. These findings take on special importance in light of a recent theory of age differences in cognition that posits that a reduction in the efficiency of inhibitory mechanisms controlling access to and modulating the contents of working memory may underlie patterns of spared and deficit functioning associated with aging ( Hasher & Zacks, 1988 ).

According to that theory, inhibitory mechanisms operate to prevent nonrelevant information from entering and being sustained in working memory. Associated with the diminished ability to inhibit irrelevant ideas are other findings that are consistent with theoretically predicted consequences of such an activation pattern. For example, in relation to younger adults, older adults show a broader range of at least momentarily activated ideas in the context of both sentence and discourse processing tasks ( Connelly, Hasher, & Zacks, 1991 ; Hamm & Hasher, 1992 ; Stoltzfus, 1992 ) and speech patterns in which sudden switches in topics are more likely to occur ( Gold, Andres, Arbuckle, & Schwartzman, 1988 ). Older adults also show activation of previously relevant ideas that persists for a longer duration than is the case for younger adults ( Hamm & Hasher, 1992 ; Hartman & Hasher, 1991 ). These effects are natural by-products of a mechanism that fails to gate out of working memory stimuli and thoughts that are not directly relevant to current goals. There is also evidence that a major consequence of broad initial activation of nonrelevant ideas by older adults is seen in these adults' ultimately impaired retrieval of target information: This retrieval deficit can be attributed to the resulting larger number of ideas that inefficient inhibition permits to become interconnected during the acquisition of information (or to the functional "fan" size; see Gerard, Zacks, Hasher, & Radvansky, 1991 ).

The present data, which show that older and younger adults suppress irrelevant locations with equal facility, stand in sharp contrast to data regarding the suppression of identity. It seems possible that multiple inhibitory systems operate and that not all such systems diminish in effectiveness with age. The Hasher and Zacks (1988) theory appearsto need modification to allow for the suggestion that those inhibitory mechanisms that diminish with age are ones that operate to prevent nonrelevant, semantic-meaning—bearing information from entering working memory. Inhibitory mechanisms that operate to prevent processing of irrelevant locations seem not to diminish with age ( D'Aloisio & Klein, 1990 ).

This behavioral dissociation between suppression of identity and location in elderly adults may be tied to the neuro-physiological underpinnings of selective attention. As Harter and Aine (1984) have suggested, the neural mechanisms of selective attention may be closely related to the two cortical visual pathways: the ventral pathway, which processes identity information, and the dorsal pathway, which processes spatial information. Direct evidence that selective attention does indeed affect neuronal processing in the visual system comes from research by Moran and Desimone (1985) . These authors have shown that when a monkey attends to a stimulus within the receptive field of a neuron in extrastriate cortex, responses to neighboring but ignored stimuli within that same field are greatly attenuated or suppressed, a process which may be similar to the one we hypothesize occurs when people select a particular target at a particular spatial location (i.e., selecting a target entails active inhibition of the location of distractors). Although very little physiological evidence directly addresses the notion that inhibitory mechanisms associated with the processing of location are preserved with age, the behavioral evidence is compelling. ⁶

It may be the case that inhibition of spatial location is a "first-in/last-out" process, as young children, learning-disabled children, and older adults show similar patterns of performance with respect to the inhibition of location versus identity, with the location system present early and spared with age and the identity system absent early and diminished with age ( Brown, Lorsbach, & Simpson, 1991 ; Stoltzfus et al., 1993 ; Tipper et al., 1989 ; Tipper & McLaren, 1990 ). Furthermore, the sparing of the location system ought to be written onto other behaviors, as the diminution of the identity system is. In the selective attention literature, data clearly show that the selective attention performance of older adults is spared when the location of either targets or distractors is predictable and is impaired when not (e.g., Madden, 1983 ; Madden & Plude, 1993 ; Plude & Hoyer, 1985 ). Thus, location cues may play a profound role in sustaining performance across the adult lifespan. These cues may be particularly important for older adults with profound (identity) memory problems. McEvoy (in press) recently demonstrated a remarkable improvement in the daily functioning of an Alzheimer's patient whose environment was changed by systematizing important locations (e.g., the laundry basket) in her life.

The inhibition of return or attended repetition data are equally straightforward. When a target reappeared in the same location, younger adults were not reliably slower to respond to that location, and older adults showed no tendency toward an attended repetition effect. Perhaps differences between the task used here and by Tipper et al. (1990) and the task more commonly used to assess inhibition of return did not allow us to see these effects consistently. For instance, as noted by Tipper et al. (1990) , in the typical inhibition-of-return experiment (e.g., Posner & Cohen, 1984 ; see also Kieley & Hartley 1990 ), subjects are presented with only one stimulus on any given trial. In the present experiments, subjects were presented with two stimuli on the test trial, and it could be that having to ignore a distractor at test has detrimental consequences for inhibition of return. Other investigators have also reported difficulty in obtaining an inhibition-of-return effect using nontraditional paradigms (Experiment 3 of Kieley & Hartley, 1990 ; Tipper et al., 1990 ).

The interference data are also clear. Interference from distractors was measured in Experiments 1 and 2. Reliable interference was found for both younger and older adults in both experiments, and the magnitude of these effects did not differ between two age groups. Furthermore, there were no reliable correlations between interference and suppression, consistent with other findings in the literature (see e.g., Driver and Tipper, 1989 ; Stolzfus et al., 1993 ; Tipper, Weaver, Kirkpatrick & Lewis, 1991 ), demonstrating that interference and suppression, or negative priming effects, do not necessarily covary. Initially, it was thought that the function of the inhibition indexed by the suppression effect was to reduce concurrent interference (e.g., Stoltzfus et al., 1993 ). Thus, when inhibition is large, interference should be small and vice versa. The mixed pattern of findings–with interference not necessarily tied to suppression–have yet to be explained. One possibility, suggested by Stoltzfus et al. (1993) , is that inhibition functions not to reduce activation of a concurrent distractor, but rather to enable the previously selected target to gain momentum toward establishing a coherent stream of thought or action by slightly retarding the development of excitation to the just-rejected distractor. In also showing a lack of correspondence between interference and location suppression effects, the present evidence suggests that such a function may serve both location and identity mechanisms, particularly because the patterns of performance shown by older adults on spatially cued selective attention tasks are similar to results produced by delaying onset of distractors in flanker tasks for young adults ( Eriksen & Schultz, 1979 ), thus experimentally controlling the temporal order in which items were processed: In both cases the effects of distracting materials are reduced.

The results of the three experiments reported here lead to the conclusion that inhibitory mechanisms addressed to location, unlike those addressed to identity, are spared with age. The results further suggest that for young adults, location and identity suppression effects can make separate contributions to slowing when these aspects of an old distractor are repeated in a current target.

1

The term inhibition is used widely and in a variety of literatures. We take inhibition toward distractors to be a theoretical mechanism. The operation of this mechanism can be indexed by a set of empirical operations from which one can observe the suppression or negative priming effect. Thus, we tend to use suppression or the suppression effect to refer to the operational outcome and inhibition to refer to the theoretical mechanism.

A related finding has been reported by Filion, McDowd, and Baylis (1992) .

The means and standard deviations for the conditions to be discussed are presented in Table 1 . Two other experimental conditions can also be used to assess inhibition of return. Both are variants of the standard procedure in which a target appears alone on one trial and then reappears in the same location but with a distractor. In these other conditions, a target and a distractor appear on the prime; on the test trial, either both the target and the distractor repeat in the same locations or only the target repeats in the same location. Performance on such trials can be compared with performance on control trials that follow distractor-present primes and in which no location is reused. All comparisons produce the same findings as those reported above, with no evidence of slowing when the target location is reused. Instead, reaction times are faster for reused locations, with older adults always showing more facilitation than younger adults. A similar pattern of results is obtained when distractor locations are reused by the distractor. Younger adults tend to show a nonsignificant facilitatory effect; older adults show reliable facilitation. This facilitation for older adults is consistent with the idea that they are able to suppress locations: Irrelevant information in an irrelevant location is less disruptive than irrelevant information in a previously unused location. By contrast, when the distractor appears where the target had been, neither older nor younger subjects respond differently than on control trials.

When the target and distractor switch places from prime to test trial, younger subjects respond significantly more slowly than they do during control trials. (This is not the case for older subjects.) However, the data for older adults is in the right direction, and, as a result of counterbalancing constraints, the analysis of the switch trials may have been somewhat compromised because this condition included only half the number of trials (24) than did most others. Note also that the effects for adults could be the product of location suppression combined with large facilitation shown when the distractor location is reused. (Younger subjects show less facilitation.)

It is important to note that the visual system is not the only candidate for the location of these suppression mechanisms ( Corbetta, Miezin, Dobmeyer, Shulman, & Petersen, 1991 ).