7 pages

Orienting in space and time: Joint contributions to exogenous spatial cuing effects

of 7
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
We examined whether the time course of exogenous spatial-cuing effects is sensitive to the allocation of attention in time. Expectation for a target within a particular time window following the cue was manipulated by varying the proportion of trials
  Psychonomic Bulletin & Review2003, 10 (4), 877-883 The spatial-cuing method has been used by many re-searchers to study the orienting of attention in space (e.g.,Posner & Cohen, 1984). Recently, a similar method has beenused to study the orienting of attention in time (Coull, Frith,Buchel, & Nobre, 2000; Miniussi, Wilding, Coull, & Nobre,1999). Although space and time might reasonably bethought of as orthogonal dimensions in which attentioncan be oriented, to our knowledge little research has beendirected to this issue. 1 The interaction between spatial andtemporal orienting was the focus of the experiment de-scribed below. Exogenous Spatial and Endogenous Temporal Orienting The spatial-orienting procedure that we used involvedthe presentation of an exogenous spatial cue (a brighten-ing) at one of two marked locations. The cues were notpredictive of the location of the subsequent target. In aseminal study in which this procedure was used, Posnerand Cohen (1984) demonstrated that less time was requiredto detect a target at cued locations than at uncued locationswhen the cue–target stimulus onset asynchrony (SOA) wasless than about 300msec. However, for longer cue–targetSOAs, the opposite pattern of results was observed; that is,response times (RTs) were longer for cued than for uncuedtargets. The latter effect is now commonly known as inhi-bition of return (IOR: see Klein, 2000, for a recent review). A common theoretical account of these cuing effects as-sumes that an abrupt onset cue captures attention auto-matically. Consequently, targets are responded to morequickly at the cued than at the uncued location when theyappear shortly after the cue. However, for longer intervalsbetween the cue and the target, attention is disengaged fromthe cued location prior to target onset. The resulting slowerresponses for cued than for uncued targets are thought toreflect an inhibition process that prevents attention fromreturning to where it has already been.Although the capture of attention by a peripheral cue isoften described as automatic, not all researchers agree onthis point. In particular, Folk, Remington, and Johnston(1992) have proposed that the capture of attention dependson the attentional set adopted by the observer (see Ruz &Lupiáñez, 2002, for a review). Furthermore, a growing lit-877Copyright 2003 Psychonomic Society, Inc. The research reported in this article was funded by an NSERC researchgrant awarded to B.M. and by Spanish Ministry of Education and Cul-ture Research Grant BSO2000-1503 awarded to J.L. We thank Bob Rafalfor informing us of a similar set of experiments conducted by Andrea Berger(see note 1) and Mari Reiss Jones for comments on earlier versions of the manuscript. Correspondence concerning this article should be ad-dressed to B. Milliken, Department of Psychology, McMaster Univer-sity, Hamilton, ON, L8S 4K1 Canada (e-mail: millike@mcmaster. ca).  Note—This article was accepted by the previous editorial team,while John T. Wixted was editor. Orienting in space and time: Joint contributions to exogenous spatial cuing effects BRUCE MILLIKEN  McMaster University, Hamilton, Ontario, Canada  JUAN LUPIÁÑEZ Universidad de Granada, Granada, Spain andMARTHA ROBERTS andBILJANA STEVANOVSKI University of Waterloo, Waterloo, Ontario, Canada We examined whether the time course of exogenous spatial-cuing effects is sensitive to the alloca-tion of attention in time. Expectation for a target within a particular time window following the cue wasmanipulated by varying the proportion of trials that appeared at each of three stimulus onset asyn-chronies in both a detection task and a two-alternative forced-choice discrimination task. The time courseof spatial-cuing effects was sensitive to the temporal expectation manipulation only in the discrimina-tion task. The results are discussed with reference to the role of attentional set in exogenous spatial-cuing paradigms. BRIEF REPORTS  878MILLIKEN, LUPIÁÑEZ, ROBERTS, AND STEVANOVSKIerature suggests that spatial-cuing effects measured fol-lowing attentional capture are subject to modulation byendogenous attentional processes. For example, severalrecent studies have demonstrated that the time course of exogenous spatial-cuing effects is sensitive to factors thatalter strategic aspects of processing (Danziger & King-stone, 1999; Lupiáñez & Milliken, 1999; Lupiáñez, Mil-liken, Solano, Weaver, & Tipper, 2001). Given these priorstudies, it seemed reasonable to ask whether a manipula-tion of endogenous attention in time would modulate theexogenous allocation of attention in space.The temporal-orienting procedure that we used varied theproportion of trials presented at each of three cue–targetSOAs: 100, 500, and 900msec. In principle, manipulatingthe proportions of trials that occur at each SOA ought toaffect how subjects allocate attention in time. Indeed, sev-eral recent studies have shown that attention can be en-dogenously oriented to a specific moment in time (Coulletal., 2000; Miniussi etal., 1999). In these studies, a cuewas presented at fixation, indicating that a subsequent tar-get (also to be displayed at fixation) would most likely ap-pear after either a short (600msec) or a long (1,400msec)temporal interval. These cues were effective in inducingshifts of attention in time, since detection responses werefaster for targets appearing at expected intervals than fortargets appearing at unexpected intervals. The Interaction Between Spatialand Temporal Orienting How might spatial and temporal orienting interact? Astarting place for thinking about this issue is to assume thatthe orienting of attention in time is related to the prepara-tory state, or attentional set, of subjects in advance of theonset of a target stimulus. Characterizing temporal orient-ing in this manner establishes a bridge to conceptual issuesin the attention capture literature. In particular, Folk etal.(1992) proposed that attention capture by a cue dependson the task subjects are required to perform on a follow-ing target. In their study, when a target task required selec-tion of an abrupt onset singleton, an abrupt onset single-ton cue produced an attention capture effect, whereas a colorsingleton cue failed to do so. Similarly, when a target taskrequired selection of a color singleton, a color singletoncue produced an attention capture effect, whereas an abruptonset singleton cue failed to do so.To extend this framework to the study of temporal ori-enting, we assume that the nature of a task (e.g., color sin-gleton vs. onset singleton), as well as other factors relatedto preparatory state, can alter an attentional set and, con-sequently, modulate the influence of an exogenous spatialcue. In particular, we propose that the effect of an exoge-nous spatial cue may depend on whether that cue appearsduring a period of time in which a subject is optimally pre-pared for the onset of a target. For example, consider anexogenous spatial cue that appears just 100msec prior tothe point in time at which a subject expects a target to ap-pear. In this case, the preparatory set engaged by the sub- ject in anticipation of a target is likely to be in place whenthe cue appears. In contrast, if an exogenous spatial cue ap-pears a full 900msec before the subject expects a target toappear, the preparatory set engaged by the subject in an-ticipation of a target is less likely to be in place when thecue appears. The difference in the preparatory states upononset of the cue in these two situations could, in turn, mod-ulate the influence of the spatial cue on performance.Given this general framework, predictions concerningthe interaction between spatial and temporal orienting canbe made. For example, it seems plausible that attention cap-ture would be strongest when the cue occurs while the sub- ject is in an optimal state of preparation for the target. In thecontext of this study, this interaction between temporaland spatial orienting would reveal itself in larger spatial-cuing effects in the 100-msec SOA condition when sub- jects expect the target to appear 100msec after the cuethan when they expect the target to appear 900msec afterthe cue.Although predictions regarding cuing effects at longerSOAs are assumption dependent, it is worth specifying atleast one possibility. As a guide to this prediction, we referto Klein’s (2000) proposal that exogenous spatial-cuingeffects reflect two influences that produce opposite effectson performance. This proposal is depicted in Figure1. Thepositive influence is depicted by the dashed line and rep-resents the influence of the capture of attention at the cuedlocation. This influence speeds performance for cued rel-ative to uncued trials, is large at short cue–target SOAs,but decreases rapidly with increasing SOA. The negativeinfluence is depicted by the dotted line and represents theinfluence responsible for the IOR effect. Note that this in-fluence slows responses to cued, relative to uncued, trialsand is equivalent across the range of SOAs. The measuredcuing effect is depicted by the solid line and is simply thesum of the two influences described above.For the sake of simplicity, we assume that temporal ex-pectancy will affect only the positive influence on cuingeffects (i.e., the attention capture component), which wedepict by shifting the dashed line upward. As was notedabove, changing this positive influence should affect themagnitude of positive cuing effects at short SOAs (notethe different lengths of the double-headed arrows in theupper and lower panels). However, note that magnifyingthe positive influence can produce two further effects: ashift in the point at which cuing effects change from fa-cilitation to IOR (note the different positions of the cir-cled area in the upper and lower panels) and smaller IOReffects at longer SOAs. These predictions were used as astarting point for interpreting the interaction between spatialand temporal orienting in the experiment described below. METHOD Subjects One hundred forty-four undergraduate students from McMasterUniversity received course credit or were paid for their participation.All had normal or corrected-to-normal vision.  ORIENTING IN SPACE AND TIME879 Design The experiment consisted of a 2 (task: detection/discrimination) 3 3 (bias: unbiased/short/long) 3 3 (SOA: 100/500/900msec) 3 2(cuing: cued/uncued) mixed factorial design, in which task and biaswere between-subjects variables and SOA and cuing were within-subjects (within-blocks) variables. Each task 2 was completed by 72 sub- jects, randomly assigned to one of the three levels of bias. In the un-biased condition, there were equal proportions of trials at each SOA.In the short- (long-) bias condition, 66% of the trials were presentedat the 100-msec (900-msec) SOA, 17% at the 500-msec SOA, and17% at the 900-msec (100-msec) SOA. Materials Stimuli were presented on a Sony SVGA monitor connected to anIBM personal computer, running MEL software (Schneider, 1988).Stimuli consisted of two black boxes (1.4º in width and 1.7º inheight) that were displayed 8.5º to the right and left of a central fix-ation cross (a plus sign, 1 ) on a pale gray background. The targetwas either a black X or a black O displayed in the center of one of the two boxes. Target letters were 0.4º in width and 0.8º in height andwere viewed at a distance of approximately 57cm. Procedure Prior to beginning the experimental session, the subjects read a setof instructions displayed on the computer monitor. In the discrimi-nation task, the subjects were asked to decide whether the letter wasan X or an O as quickly and accurately as possible and to respond bypressing the X or the M key on a standard keyboard (response keymappings were counterbalanced across subjects). In the detectiontask, the subjects were asked to make a response by pressing the B Figure1. Hypothetical cuing effects as a function of stimulus onset asyn-chrony (SOA) broken down into two components, as proposed by Klein(2000). In the top panel, the component that produces faster responsesfor cued than for uncued trials (the positive influence) is depicted as adashed line, whereas the component that produces slower responses forcued than for uncued trials (the negative influence) is depicted as a dot-ted line. The resulting cuing effect is the sum of these two componentsand is depicted as a solid line. The lower panel differs from the upperpanel only in that the positive influence, which is presumed to reflect thecapture of attention at the cued location, is larger in the lower panel.Note that this change in the positive influence can (1) magnify cuing ef-fects measured at short SOAs (see the different lengths of the gray double-headed arrows in the two panels), (2) delay the transition of cuing ef-fects from positive to negative (see the shift in position of this transitionpoint, marked by a circle, from the upper to the lower panel), and (3)produce smaller inhibition of return effects at long SOAs.  880MILLIKEN, LUPIÁÑEZ, ROBERTS, AND STEVANOVSKI key upon onset of any target and to withhold responses on the 20%of the trials in which a target was not presented (i.e., catch trials).Auditory feedback for incorrect responses allowed the subjects tomonitor their accuracy.A trial began with the display of the central fixation cross and thetwo boxes for 1,000msec. One of the boxes then changed to whitefor 50msec, which created the illusion of a flicker. At varying in-tervals following the offset of this cue, the target was then presentedfor 100msec, producing a cue–target SOA of 100, 500, or 900msec.The boxes and the fixation cross remained visible for 2,000msec afterthe offset of the target or until the subject made a response. The en-tire display then disappeared, leaving only a blank screen. The boxesand the fixation point reappeared 500msec later, marking the startof a new trial. Half of the trials were cued (the target appeared at the cued loca-tion), and the other half were uncued, so that the cue provided no pre-dictive information about the location of the target. Short-SOA tri-als were most probable in the short-bias group, long-SOA trials weremost probable in the long-bias group, and the three SOAs wereequally probable in the unbiased group. Given that the duration of the fixation point was constant across trials, the subjects in the twobiased groups could use the fixation point to anchor an expectationfor when the target would appear: 1,100msec after fixation onset (and100msec after cue onset) in the short-bias group or 1,900msec afterfixation onset (and 900msec after cue onset) in the long-bias group.After the instructions had been read and understood, the subjectspressed the space bar to begin a set of 48 practice trials, followed byfour blocks of test trials. Each block of test trials proceeded until 96correct responses were recorded in the discrimination task or until120 correct responses were recorded (96 correct responses to targetsand 24 correct response omissions to catch trials) in the detection task.In each bias condition, the proportion of trials in each conditionwithin a block mirrored that for the experimental session as a whole.After every 16 trials in both the practice and the test sessions, a mes-sage appeared on the screen, reminding the subjects to keep theireyes on the fixation cross and to be fast and accurate. The experi-mental session lasted approximately 30 min. RESULTS AND DISCUSSION RTs for correct responses in the discrimination task andfor hits in the detection task were first submitted to an out-lier elimination procedure (Van Selst & Jolicœur, 1994)that excluded 2.3% of the RTs from further analyses. 3 Mean RTs were computed using the remaining observa-tions and then were submitted to a 2 3 3 3 2 3 3 (task 3 bias 3 cuing 3 SOA) mixed factorial analysis of variance(ANOVA) with task and bias as between-subjects variables.Mean RTs and error rates for each condition are displayedin Table1. Spatial Orienting The usual effects of exogenous spatial cues on perfor-mance were reflected in a significant interaction betweencuing and SOA [ F  (2,276) 5 59.18,  MS  e 5 379.63,  p , .001]. Simple main effects tests revealed significantlyfaster responses for cued than for uncued trials for the100-msec SOA (458 vs. 467msec) and significantly slowerresponses for cued than for uncued trials for both the 500-msec (466 vs. 444msec) and the 900-msec (465 vs.441msec) SOAs. These results correspond to those inmany prior studies of exogenous spatial cuing. 4 Temporal Orienting As is illustrated in Figure2, which presents RT collapsedover cue status, there were robust temporal-orienting ef-fects in this experiment. The effect of temporal orientingwas revealed by a significant interaction between bias andSOA [ F  (4,276) 5 76.97,  MS  e 5 475.13,  p , .001]. Al-though this interaction was modulated by task [ F  (4,276) 5 2.79,  MS  e 5 475.13,  p , .03], separate analyses of the twotasks revealed significant bias 3 SOA interactions that werequalitatively similar [see Figure2; F  (4,138) 5 47.09,  MS  e 5 493.62,  p , .001, and [ F  (4,138) 5 32.08,  MS  e 5 456.64,  p , .001, for the detection and the discriminationtasks, respectively]. In the short-bias condition of both thedetection and the discrimination tasks, RTs were lowestfor the 100-msec SOA and increased monotonically withincreases in SOA. This pattern of data produced signifi-cant linear trends across SOA for both tasks [ F  (1,23) 5 127.00,  MS  e 5 268.62,  p , .001, and F  (1,23) 5 12.45,  MS  e 5 590.59,  p , .005, respectively]. In neither case wasthe residual quadratic trend significant. In the long-bias Table1Mean Response Times (RTs, in Milliseconds) and Percentages of Errors (ERs) for Each Task, Stimulus Onset Asynchrony (SOA) Bias, and Level of Cuing SOA (msec) in Detection TaskSOA (msec) in Discrimination Task100500900100500900RTERRTERRTERRTERRTERRTERCondition  MSEMSEMSEMSEMSEMSEMSEMSEMSEMSEMSEMSE  Short BiasCued3349.01.50.637410.32.71.038810.84.81.651819.4 5.70.955122.76.61.2 56126.05.51.2Uncued33510.73.11.134212.22.50.93568.24.31.254721.56.41.053419.46.31.053919.65.11.1UnbiasedCued38212.11.30.4 39110.80.80.3 40310.41.20.554118.35.21.053518.75.41.053620.14.10.8Uncued38311.91.00.334710.70.80.336410.50.90.355618.75.71.153718.86.11.2 53117.44.31.1Long BiasCued413  ORIENTING IN SPACE AND TIME881condition, RTs were highest for the 100-msec SOA anddecreased monotonically with increases in SOA, againproducing significant linear trends for both tasks [ F  (1,23) 5 54.63,  MS  e 5 971.17,  p , .001, and F  (1,23) 5 122.12,  MS  e 5 508.01,  p , .001, respectively]. The residual qua-dratic trends were also significant [ F  (1,23) 5 8.56,  MS  e 5 365.71,  p , .01, and F  (1,23) 5 9.16,  MS  e 5 201.47,  p , .01] but accounted for relatively small proportions of vari-ance (.06 and .03, respectively). These results illustrate a robust effect of temporal ex-pectancy in both the detection and the discrimination tasks.These effects were observed both when the most frequentSOA was short and when the most frequent SOA waslong, demonstrating that they were not due simply to in-creases in readiness with increases in SOA (Niemi & Näätä-nen, 1981). Instead, the results suggest that the subjectsgenerated an expectancy for a stimulus within a particulartime window (see also Coull etal., 2000; Miniussi etal.,1999). Whereas prior studies of temporal orienting used adetection task and just two SOAs, the use of both detectionand discrimination tasks and three SOAs allowed us tomeasure a robust linear trend in performance across SOAthat generalized across tasks. Interactions Between Spatial and Temporal Orienting The interaction between spatial and temporal orientingis reflected in statistical interactions that involve the biasand cuing variables. The highest order significant interac-tion involving these variables was the four-way interactionbetween task, bias, cuing, and SOA [ F  (4,276) 5 2.62,  MS  e 5 379.63,  p , .05]. We examined this interaction fur-ther by conducting separate analyses for each task. Theseanalyses treated cuing effects (uncued RT 2 cued RT) asthe dependent variable and bias and SOA as within-subjectsindependent variables. Note that when cuing effects serveas a dependent variable, an interaction between spatial andtemporal orienting will produce a main effect of bias,rather than an interaction between cuing and bias. Fur-thermore, if an interaction between spatial and temporalorienting itself depends on SOA, the bias 3 SOA interac-tion ought to be significant. Cuing effects are shown in Figure3. Values greater thanzero depict facilitation effects, whereas those less than zerodepict IOR effects. Note that in all the task and bias condi-tions, cuing effects shifted in a negative direction with in-creases in SOA. This trend was reflected in significantmain effects of SOA for both detection and discriminationtasks [ F  (2,138) 5 62.51,  MS  e 5 467.30,  p , .001, and F  (2,138) 5 15.93,  MS  e 5 1,051.22,  p , .001, respectively].Of central interest were effects that involved the biasfactor. In the detection task, neither the main effect of biasnor the interaction between bias and SOA approached sta-tistical significance (both  p s . .30), suggesting that spa-tial and temporal orienting did not interact in this task. Incontrast, in the discrimination task, although the main ef-fect of bias was not significant (  p . .30), the interactionbetween bias and SOA was significant [ F  (4,138) 5 3.04,  MS  e 5 1,051.22,  p , .02]. This result indicates that spa-tial and temporal orienting did interact in the discrimina-tion task and, in particular, that this interaction dependedon cue–target SOA.We further examined this interaction in the discrimina-tion task by evaluating whether it conformed to predictionsset forth in the introduction. One prediction was that spatial-cuing effects at the 100-msec SOA ought to be larger for Figure2. Mean response times (RTs, in milliseconds) f or the short-bias andlong-bias groups as a function of stimulus onset asynchrony (SOA; 100, 500,and 900msec) and task (detection vs. discrimination). Note that responses werefastest for short SOAs in the short-bias group and for long SOAs in the long-bias group for both tasks.
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks