Crafts

14 pages
6 views

Does IOR occur in discrimination tasks? Yes, it does, but later

of 14
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.
Share
Description
When a stimulus appears in a previously cued location several hundred milliseconds after the cue, the time required to detect that stimulus is greater than when it appears in an uncued location. This increase in detection time is known as inhibition
Transcript
  Attention is widely presumed to play an important rolein the rapid and efficient scanning of visual environments.In particular, when search is difficult, the movement of attention from one location to another may improve dis-crimination at each location, and thus also improve over-all search efficiency. However, the efficiency of searchalso depends on the ability to prevent attention from return-ing to previously examined locations. This issue has beenthe focus of considerable study by attention researchersover the past decade.Specifically, it has been shown that response to a tar-get is speeded if the location at which the target appearsis precued. In the cost–benefit paradigm (Posner, 1980),subjects are to respond to a target appearing in one of three boxes, one in the center of the screen and one to eachside of the center. Before the target appears, the subject’sattention is cued to one of the two peripheral locations.This attentional cuing is accomplished by making one of the two peripheral boxes flicker briefly. The target thenappears in either the cued location (cued trials) or the un-cued location (uncued trials). When the cue–target stim-ulus onset asynchrony (SOA) is less than about 300msec,responses are faster on cued than on uncued trials.However, Posner and Cohen (1984) demonstrated thatat SOAs of longer than 300 msec, the opposite pattern of results is observed; that is, response times (RTs) becomelonger on cued than on uncued trials. They used the term inhibition of return (IOR) to describe the effect, arguingthat “the inhibition effect evolved to maximize samplingof the visual environment” (p. 550).Following this initial study, other researchers extendedthe effect to different dependent variables and tasks.Thus, IOR has now been reported using manual keypress(Posner & Cohen, 1984) and eye movement latency (Ab-rams & Dobkin, 1994; Pratt, 1995) as dependent vari-ables, and with both target detection and localizationtasks (Maylor, 1985). The effect has also been reportedin both attentional cuing tasks (cue–target paradigm) andtasks requiring a response to the same location on con-secutive trials (target–target paradigm; Maylor, 1985).Within attentional cuing tasks, IOR is usually obtainedonly when a peripheral (exogenous) attentional cue isused. However, Rafal, Calabresi, Brennan, and Sciolto(1989) reported IOR using a central arrow (endogenous)cue, if subjects prepared an eye movement in the direc-tion indicated by the cue, even though they had “can-celed” that preparation before target onset.It has recently been argued that attention is inhibited to return not only to the location but also to the object inwhich the cue appears (Tipper, Weaver, Jerreat, & Burak,1994; but see Müller & Mühlenen, 1996). Tipper et al.called the former effect “location-based” and the latter ef-fect “object-based” IOR (Tipper, Driver, & Weaver, 1991).In summary, a good deal of experimental research sug-gests that IOR is a robust and general effect that demon-strates a general principle of information processing. How-ever, the generality of the effect has been disputed recently1241Copyright 1997 Psychonomic Society, Inc. We thank Steven P. Tipper, Bruce Milliken, Arthur F. Kramer, ananonymous reviewer, and especially Raymond Klein for their usefulcomments. Bruce Milliken also helped improve our English. Pilar Gon-zalvo helped us with data collecting. We are grateful to all of them. Thisresearch was financially supported by the Spanish Ministerio de Edu-cación y Ciencia (Grant PB931114 from the DGICYT to FranciscoMartos, and by MEC FPI Grant AP92-24228372 to J.L.). Correspon-dence should be addressed to J.Lupiáñez Castillo, Departamento dePsicología Experimental y Fisiología del Comportamiento, Facultad dePsicología, Campus Universitario de Cartuja, Universidad de Granada,18071-Granada, Spain (e-mail: jlupiane@platon.ugr.es). Does IOR occur in discrimination tasks?Yes, it does, but later   JUAN LUPIÁÑEZ, EMILIO G. MILÁN, FRANCISCO J. TORNAY,EDUARDO MADRID, and PÍO TUDELA  University of Granada, Granada, Spain When a stimulus appears in a previously cued location several hundred milliseconds after the cue,the time required to detect that stimulus is greater than when it appears in an uncued location. This in-crease in detection time is known as inhibition of return (IOR). It has been suggested that IOR reflectsthe action of a general attentional mechanism that prevents attention from returning to previously ex- plored loci. At the same time, the robustness of IOR has been recently disputed, given several failuresto obtain the effect in tasks requiring discrimination rather than detection. In a series of eight experi-ments, we evaluated the differences between detection and discrimination tasks with regard to IOR.We found that IOR was consistently obtained with both tasks, although the temporal parameters re-quired to observe IOR were different in detection and discrimination tasks. In our detection task, theeffect appeared after a 400-msec delay between cue and target, and was still present after 1,300 msec.In our discrimination task, the effect appeared later and disappeared sooner. The implications of thesedata for theoretical accounts of IOR are discussed.  Perception & Psychophysics1997, 59 (8), 1241-1254  1242LUPIÁÑEZ, MILÁN, TORNAY, MADRID, AND TUDELAon the grounds that it is limited to simple RT tasks (de-tection): “An IOR effect on choice RTs has been ob-served only in tasks requiring a saccadic or manual lo-calization response” (Müller & Mühlenen, 1996, p. 244).In fact, several studies have failed to obtain IOR withchoice RT tasks when target discrimination is necessaryto respond. Egly, Rafal, and Henik (1992) and Terry,Valdes, and Neill (1994) used shape discrimination: King-stone and Gazzaniga (1992, reported in Klein & Taylor,1994) and Tanaka and Shimojo (1996) used color dis-crimination; Pontefract and Klein (1988, reported in Klein& Taylor, 1994), and Tanaka and Shimojo used size dis-crimination; and Tanaka and Shimojo used orientation,vernier, and luminance discrimination.So far, only Pratt (1995) has reported IOR in a dis-crimination task. In Pratt’s experiment, subjects were re-quired to move their eyes toward and fixate a targetevent. Eye movement latency was used as the dependentvariable. Therefore, even though the target was to be dis-criminated—because two stimuli were displayed (a tar-get and a distractor)—a localization response was im- plied. Hence, it has not been established whether IOR occurs in discrimination tasks that do not require thesubject to localize the target. Another important issue inthis experiment is the fact that a longer-than-usualcue–target SOA was used (960 msec).In fact, we have obtained IOR with a choice RT task in our laboratory, but a 1,000-msec SOA was necessary(Lupiáñez, 1996; Lupiáñez, Milán, Tornay, & Tudela,1996; Lupiáñez, Tornay, & Tudela, 1996). For example,in one experiment, subjects were asked to discriminate thedirection indicated by an arrow that was displayed in oneof two boxes (to the left and right of fixation), one of which had previously been cued. Subjects responded byhitting one of two keys, depending on the direction of thearrow. Three different SOAs (100, 600, and 1,000msec)were used in this experiment, and their presentation wasmixed within a block. Importantly, IOR was obtainedonly at the longest SOA of 1,000 msec. In another ex- periment, only the two shorter SOAs were used, and fa-cilitation was observed for both. Thus, it appears thatIOR may be obtained in discrimination tasks, but that itis observed at longer SOAs than during detection tasks.This hypothesis was tested directly in the series of ex- periments reported in this article.A secondary issue addressed here concerns the factthat in discrimination tasks, a left-hand/right-hand re-sponse button assignment is customarily used to recorda response to the target, and the target itself can appear in either the left or the right visual field. Thus, the stim-ulus can be displayed in the visual field either ipsilateralor contralateral to the hand of response. Ipsilateral re-sponses are usually faster and more precise than con-tralateral responses. This effect is known as the Simoneffect  (Hommel, 1995; Simon, 1969; Simon & Rudell,1967) and has been discussed recently in relation to theorienting of attention (Umiltà & Nicoletti, 1992). Giventhat IOR has been previously obtained only in detectiontasks (a single response) and in localization tasks (onlyipsilateral trials), the relation between IOR and the Simoneffect has not been previously explored. Thus, the ex-amination of IOR in discrimination tasks that do not in-volve localization of the target enabled us to examine therelation between IOR and the Simon effect.In summary, previous research has challenged the ro- bustness and generality of IOR, given that it has not beenobtained reliably in discrimination tasks. In this work weexplored the robustness of IOR with discrimination anddetection tasks. Different SOAs were used across exper-iments to investigate the time course of IOR for bothsimple RT (detection) and choice RT (discrimination)tasks. Thus, the experiments differed only in the SOAsused and whether detection or discrimination responseswere required. GENERAL METHOD Subjects All subjects in these experiments were from the Faculty of Psy-chology of the University of Granada. Subjects were naive as to the purpose of the experiment and participated in exchange for coursecredit. A different group of 18 subjects participated in each of theeight main experiments, and all members within each group weretested simultaneously in a room equipped with 20 computers. A dif-ferent group of 6 subjects participated in each of the two additionalexperiments (eye movement monitoring), and subjects were testedindividually. Apparatus and Stimuli Stimuli were presented on a 14-in. color VGA monitor. An IBM-compatible 486/33 microcomputer running MEL software (Schnei-der,1988) controlled the presentation of stimuli, timing operations,and data collection. Responses were made by pressing a key on thecomputer keyboard. When only one response was required (detec-tion), subjects pressed the “B” key with any finger. In the discrim-ination task, subjects pressed either the “X” key (left response) withthe index finger of the left hand or the “M” key (right response)with the index finger of the right hand. Subjects sat approximately60 cm from the computer monitor. Procedure The target on each trial appeared in the center of one of two boxes, displayed to the left and right of fixation. The boxes re-mained on the screen throughout each trial and disappeared only between trials. The boxes subtended 17 mm in height  14 mm inwidth (1.62º and 1.33º of visual angle at a viewing distance of 60cm). The inner edge of each box was 77 mm (7.31º) from the fixa-tion point (the plus sign [+], displayed in white). The target to be de-tected or discriminated was a colored asterisk, which was either redor yellow with equal probability. The boxes were displayed in dark gray on a black background. On every trial, at varying temporal in-tervals before presentation of the target, one of the two boxes was presented in white for 50 msec before returning to its srcinal dark gray. This increase in luminance gave the impression of a brief flicker. This flicker is referred to hereafter as the “attentional cue.”The sequence of events on each trial is depicted in Figure 1. Afixation point was displayed together with the two boxes for 1,000msec. Then one of the two boxes flickered for 50 msec. Followingthe flicker, the fixation point and the boxes remained on the screenfor 50, 350, 650, 950, or 1,250 msec, depending on the SOA for thattrial. Following this interval, the target was displayed for 33 msec,and then the fixation point and boxes were again displayed aloneuntil the subject’s response, or for a maximum of 2,000 msec. If noresponse was made within 2,000 msec, the next trial began. The in-  IOR IN DISCRIMINATION TASKS1243 terval between trials was 1,000 msec in duration, and the screen re-mained black throughout this interval.The response required of subjects depended on the task. In thedetection task, subjects were given instructions to press the “B” keyif, and only if, an asterisk appeared, and regardless of the color of the asterisk. In the discrimination task, half of the subjects were to press the “X” key when the asterisk was yellow and the “M” keywhen it was red, and the other half were to do the opposite—pressthe “X” key for red and the “M” key for yellow. In both tasks, thetarget was absent on 20% of the trials (catch trials), in which casesubjects were simply required to wait for the beginning of the nexttrial. Auditory feedback (a 400-Hz computer-generated tone of 100msec) was provided on error trials.Trials were grouped in blocks of 100 and presented randomlywithin each block. The experiment was interrupted for 1 min every50 trials to allow the subject to rest. Subjects were instructed to press the space bar to continue the experimental session after eachrest period. Design Two independent variables were orthogonally manipulated ineach block of trials: cuing and SOA. Cuing took two values: Thetarget could appear either in the cued box (cued trial) or in the un-cued box (uncued trial). In each experiment, a 100-msec SOA was paired with another, longer, one—100 and 400 msec in Experi-ment1, 100 and 700 msec in Experiment 2, 100 and 1,000 msec inExperiment 3, and 100 and 1,300 msec in Experiment 4. In the dis-crimination task, because the target could be responded to with ei-ther the left or the right hand, and could also appear in the left or the right box, there were two kinds of trials—ipsilateral and contra-lateral. This variable was called “stimulus–response location com- patibility” and was also completely crossed in any block of trials.In the detection task (only one central response), this variable hadno meaning and was dummy coded. Given that all the independentvariables were completely crossed within each block and that allvariables took any value with the same probability (apart from catchvs. experimental trials), there was no predictive relation betweenthe attentional cue and the target’s location or color. Similarly, therewas temporal uncertainty.In each experiment, subjects performed one practice block andtwo blocks of experimental trials. The practice block consisted of four trials of each combination of compatibility (2)  cuing (2)  SOA (2), and eight catch trials (32 + 8  40 trials). Each block of experimental trials consisted of 10 trials of every combination and20 catch trials (80 + 20  100 trials). Thus, the experimental con-dition consisted of 200 trials, 160 of which represented the 20 repli-cations of each of eight different trial types, and 40 of which werecatch trials. EXPERIMENTS 1A–1B100- and 400-msec SOAs Two SOAs were used in these experiments: 100 and400 msec. The two experiments differed only in terms of the response given to the target. In Experiment 1A, sub- jects were required to detect the target, whereas in Ex- periment 1B, subjects were required to discriminate itscolor. However, because subjects used left- and right-handresponses to record their color discrimination, locationcompatibility was meaningful as an independent variablein Experiment 1B, but not in Experiment 1A. Thus, re-sults of these two experiments are reported separately. Figure 1. Experiment trial sequence, from top left to bottom right. Each trial began with a fixation point and two boxes dis-played in gray for 1,000 msec. Then one of the two boxes was displayed in white for 50 msec. After 50, 350, 650, 950, or 1,250msec (depending on the stimulus onset asynchrony), the target, a red or yellow asterisk, was displayed for 33 msec. Then thefixation point and the two boxes were again displayed in gray, either until subject’s response or for 2,000 msec if no responsewas made.  1244LUPIÁÑEZ, MILÁN, TORNAY, MADRID, AND TUDELA Results Rates of false alarms (responses to catch trials) 1 were0.061 and 0.050 (for short and long SOAs, respectively)in Experiment 1A and 0.047 and 0.036 in Experiment1B. Trials with correct responses faster than 100 msecor slower than 1,200 msec (1.04% of correct response tri-als),as well as incorrect responses, were excluded fromthe RT analysis. Mean RTs and percent errors 2 are shownin Tables 1 and 2. Experiment 1A: Detection task  . Mean RTs of cor-rect responses were introduced into a repeated measuresanalysis of variance (ANOVA) with SOA (2 levels) andcuing (2 levels) 3 as independent variables. This analysisrevealed a significant interaction between SOA andcuing [  F  (1,17)  21.26,  MS  e  456.38,  p < .001]. Fur-ther analysis of the interaction revealed a facilitationeffect at the short (21 msec) SOA; cued trials were re-spondedto faster than uncued ones [  F  (1,17)  8.57,  MS  e  449.14,  p < .01], and an IOR effect at the longSOA [  F  (1,17)  15.94,  MS  e  374.47,  p < .001]. Also,RTs in cued trials were significantly shorter at the short(381-msec) than at the long (402-msec) SOA [  F  (1,17)  16.97,  MS  e  445.01,  p < .001]. The opposite was truefor uncued trials: RTs were slower at the short SOA [410vs. 384 msec;  F  (1,17)  8.37,  MS  e  328.15,  p < .05].Thus, the usual IOR effect (i.e., slower responses to thecued than to the uncued location) was observed with thedetection procedure.Analysis of error percentages revealed no significanteffects. Experiment 1B: Discrimination task  . Mean RTs for correct responses in the discrimination task were sub-mitted to a repeated measures ANOVA, with compati- bility (2 levels), SOA (2 levels), and cuing (2 levels)treated as independent variables. This analysis revealeda significant main effect of cuing [  F  (1,17)  19.27,  MS  e  2,945.6,  p < .001], showing faster responsesincued than in uncued trials (40 msec). There were no other significant effects in the analysis (all  p s > .15).The analysis of error percentages revealed a main ef-fect of cuing [  F  (1,17)  10.43,  MS  e  131.79,  p < .01]:Responses on cued trials (6% errors) were more accuratethan responses on uncued trials (12% errors). Also, cuinginteracted with SOA [  F  (1,17)  7.85,  MS  e  124.36,  p < .05]. Further analysis revealed that the cuing effectwas significant only at the longer SOA [  F  (1,17)  9.64,  MS  e  242.07,  p < .01].Separate analyses of misses (no response) and mistakes(incorrect response) showed the same results: a facilitationeffect at the longer SOA for both misses [4.6%;  F  (1,17)  10.04,  MS  e  37.68,  p < .01] and mistakes [6.8%;  F  (1,17)  7.77,  MS  e  107.21,  p < .02]. Discussion Experiments 1A–B differed only in the response thatsubjects made to the target. When subjects were requiredto detect the target, a facilitation effect was observed atthe 100-msec SOA, and the opposite was observed at the400-msec SOA, thus demonstrating the usual IOR pattern(Posner & Cohen, 1984). However, when subjects were re-quired to discriminate the target’s color, a facilitation ef-fect was observed at both the 100- and 400-msec SOAs. Infact, a subsequent ANOVA that treated task (detection/dis-crimination) as a between-subjects variablerevealed a sig-nificant three-way interaction between task,cuing, andSOA [  F  (1,34)  7.4,  MS  e  430,  p < .02]. Thus, the cuing  SOA interaction was indeed modulatedby differencesinherent in the detection and discrimination tasks.The data from these experiments support previous re- ports of failure to demonstrate IOR in a discriminationtask (Pontefract & Klein, 1988, reported in Klein & Tay-lor, 1994; Terry et al., 1994). Pontefract and Klein usedtwo similar values of SOA (100 and 500 msec) in separate blocks of trials and obtained similar results: When achoice RT task was used, facilitation was observed at bothSOAs. However, when a simple RT task was used, facil-itation was observed at the 100-msec SOA and IOR wasobserved at the 500-msec SOA.Thus, the effect of cuing a location appears to have different effects depending on the nature of the task. Atshort SOAs, cuing appears to benefit both detection anddiscrimination of the target. However, at longer SOAs(400msec here, 500 msec in Pontefract and Klein,1988), cuing appears to help discrimination but hinder detection.Given the lack of IOR in such discrimination tasks,many researchers have argued against the notion thatIOR is a robust and general effect (Terry et al., 1994) andagainst its attentional nature (Klein & Taylor, 1994;Schmidt, 1996). Terry and colleagues argued that if IOR reflects an innate principle of attentional processing,namely preventing attention from returning to recentlyattended locations, then the effect should be obtainedwith any task. However, given that the effect is sensitiveto changes in task, IOR may not reflect a general princi- ple of information processing, but rather a specific mech-anism that is relevant for some types of information pro-cessing but not for others. Others have argued that if IOR were the opposite of the facilitative effect observed atshort SOAs (i.e. attentional inhibition), then IOR oughtto be observed in all tasks that demonstrate such facilita-tion effects. However, given that IOR has been observedreliably only in tasks requiring target detection or local-ization, it has been argued that IOR may be associatedwith responding (Klein & Taylor, 1994) or with the visual– motor action system, rather than with attention (but seeReuter-Lorenz, Jha, & Rosenquist, 1996, for an attentionalconception of IOR).As we noted in the introduction, the IOR effect mightoccur in both detection and discrimination tasks, yet withdifferent temporal parameters. The following experimentswere conducted to explore whether the same dissociationregarding IOR between detection and discriminationtasks would be obtained with longer SOAs.  I   ORI   NDI   S  C RI  MI   NAT I   O NT A S K S 1 2 4  5  Table 1Mean Correct Response Time (RT) and Percentage of Errors (PE) in Detection Task Results for Cuing and Stimulus Onset Aynchrony (SOA) (Across Experiments) SOA100 msec400 msec700 msec1,000 msec1,300 msecCuedUncuedEffectCuedUncuedEffectCuedUncuedEffectCuedUncuedEffectCuedUncuedEffectExperimentRTPERTPERTPERTPERTPERTPE1A3812.644023.06210.424102.923845.1   262.22A3741.393891.2515   0.13A4025.004225.4200.424A3794.04012.822   1.3Total3843.264033.119   0.14102.923845.1   262.2RTPERTPERTPERTPERTPERTPERTPERTPERTPE3913.23411.94   50   1.34446.674032.36   41   4.34086.93773.5   32   3.53913.23411.94   50   1.34446674032.36   41   4.34086.93773.5   32   3.5 Table 2Mean Correct Response Time (RT) and Percentage of Errors (PE) in Discrimination Task Results for Cuing and Stimulus Onset Aynchrony (SOA) (Across Experiments) SOA100 msec400 msec700 msec1,000 msec1,300 msecCuedUncuedEffectCuedUncuedEffectCuedUncuedEffectCuedUncuedEffectCuedUncuedEffectExperimentRTPERTPERTPERTPERTPERTPE1B5578.896019.86440.975663.6160115.03511.42B5636.946019.44372.503B5719.5860111.5301.944B61914.264316.8242.64Total5779.9061111.9342.015663.6160115.03511.4RTPERTPERTPERTPERTPERTPERTPERTPERTPE59310.85637.78   31   3.16089.035748.19   34   0.865718.964212.4   15   6.559310.85637.78   31   3.16089.035748.19   34   0.865718.964212.4   15   6.5
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