The Simon Effect Occurs Relative to the Direction of an Attention Shift


of 12
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.
The Simon Effect Occurs Relative to the Direction of an Attention Shift
  Journal of Experimental Psychology: Copyright 1997 by the American Psychological Association, Inc. Human Pnteeplion and Pe~ 0096-1523/97/$3.~]0 1997, Vol. 23, No. 5, 1353-1364 The Simon Effect Occurs Relative to the Direction of an Attention Shift Sandro Rubichi Universit~ di Modena Roberto Nicoletti Universit~ di Urbino Cristina Iani Universi~ di Modena Carlo Umilt~ UniversitA di Padova We investigated whether the Simon effect depends on the orienting of attention. In Experiment 1, participants were required to execute left-right discriminative responses to 2 patterns that were presented to the left or right of fixation. The 2 patterns were similar, and the discrimination was difficult. A letter at fixation signaled whether the current trial was a catch trial. The results showed a reversal of the Simon effect. That is, spatially noncorresponding responses were faster than spatially corresponding responses. In Experiment 2, the discrimina- tion of the relevant stimulus attribute was easy. In Experiment 3, the discrimination of the relevant stimulus attribute was difficult, but the stimulus exposure time was long. In either experiment, the regular Simon effect was reinstated. In Experiment 4, the letter that signaled a catch trial appeared to the left or right of the imperative stimulus. The Simon effect occurred relative to the position of the letter. Stimulus location is coded and cannot be ignored in many situations for which it is irrelevant to the task to be performed (e.g., Tsal & Lavie, 1993, and the references therein). The effects of irrelevant location codes emerge clearly in the Simon and spatial Stroop tasks (for reviews, see Lu & Proctor, 1995; MacLeod, 1991; Umilt~t & Nieo- letti, 1990). In a Simon task, the stimulus features relevant for selecting the correct response are nonspatial (e.g., two colors or two shapes). These nonspatial features are assigned to spatially defined responses (typically keypresses; e.g., left and right or above and below), and the location (e.g., left and right or above and below) in which the stimulus is shown is task irrelevant. The Simon effect refers to the fact that responses are faster when the stimulus location corresponds to the location of the assigned response than when it does not correspond. For instance, participants are presented with two different patterns and are required to press the right-side key in response to one pattern and the left-side key in response to the other. Responses are faster when the location of the stimulus and the location of the response key are located on Sandro Rubichi and Cristina Iani, Dipartimento di Scienze Biomediche, Universi~ di Modena, Modena, Italy; Roberto Nico- letti, Institnto di Psicologia, Universit,,3 di Urbino, Urbino, Italy; Carlo Umilt~, Dipartimento di Psicologia Generale, Universith di Padova, Padua, Italy. This study was supported by the Consiglio Nazionale delle Rieerche (Grants 93.00752.PF41, 94.00667.PF41, and 95.00923.PF41) and ~ Ministero della Universi~ e della Ricerca Scientifica e Teenologiea. We thank Sandro ~lla for valuable help in planning the experiments. Correspondence concerning this article should be addressed to Carlo UmiltL Dipartimento di Psicologia Genemle, Universit~ di Padova, via Venezia, 8, 35131 Padua, Italy. Electronic mail may be sent via Internet to the same side (e.g., right-right or left-left; i.e., correspond- ing stimulus--response [S-R] pairings) with respect to the condition in which they are positioned differently (e.g., stimulus on the right side and response on the left side or vice versa; i.e., noncorresponding S-R pairings). Thus, in the Simon task, the irrelevant information concerning stimulus location is processed even though participants are clearly instructed to select the response exclusively on the basis of the relevant, nonspatial stimulus dimension. Most accounts of the Simon effect share two basic assumptions (for reviews and discussions of the various accounts, see Lu & Proctor, 1995; Umilt~ & Nicoletti, 1990; for dissenting views, see Hasbroucq & Guiard, 1991; Simon,  1990; Verfaellie, Bowers, & Heilman, 1988). One assump- tion is that a spatial code is generated for the irrelevant stimulus location attribute. In fact, multiple locational codes, all of which perhaps contribute to the Simon effect, are likely produced for a single stimulus (De Jong, Liang, & Lanber, 1994; Lamberts, Tavernier, & d'Ydewalle, 1992). In most situations, spatial coding is a function of the location of the target stimulus relative to the location of the alternative stimulus (e.g., Umilt/t & Liotti, 1987; Umilt~t & Nicoletti, 1985). Spatial coding is also thought to occur in terms of egocentric hemispace (e.g., Umilt~ & Liotti, 1987), relative to the position of the attentional focus (Nicoletti & Umilt/t, 1989), or as a function of configural cues in the display (Hommel, 1995). The other widely shared assumption is that the Simon effect occurs at the response-selection stage. The idea is that, provided there is enough similarity between the (irrelevant) spatial stimulus dimension and the (relevant) spatial re- sponse dimension (i.e., the two dimensions overlap; see Kornblum, 1994; Kornblum, Hasbroucq, & Osman, 1990; Kornblum & Lee, 1995), a stimulus automatically activates its spatially corresponding response code (also see Eimer, 1995; De Jong et al., 1994). For trials in which the 1353  1354 RUBICHI, NICOLETTI, ANI, AND UMILT.~ automatically activated response code is the same as that signaled by the relevant stimulus feature (e.g., a red fight that is shown to the left of fixation and signals a left-side response), there is no competition at the response-selection stage and possibly even a benefit from the redundant response codes. For trials in which the automatically acti- vated response code is different from the one signaled by the relevant stimulus feature (e.g., a red light that is shown to the right of fixation and signals a left-side response), competi- tion between the two response codes occurs that must be resolved before the correct response can be executed. Thus, with the exception of the attention-based accounts of Simon (1990) and Verfaellie et al. (1988), it is widely held that, as srcinally proposed by Wallace (1971), a spatial code for stimulus location is formed even though it is not task relevant. This is the basis of the conflict that arises at the response-selection stage and perhaps also at the stimulus- identification stage, as suggested by Hasbroucq and Guiard (1991). Consequently, it becomes important to establish when and how the spatial code for the irrelevant stimulus location is formed. Two possible explanations have emerged (Lu & Proctor, 1995). One is that the stimulus spatial code is generated when there is a lateral shift of attention to the location occupied by the stimulus (Nicoletti & Umilt~, 1994; Stoffer, 1991; Umilttt & Nicoletti, 1992). This is the attention-shift account. An alternative explanation is that comparison of the target to a reference frame or object results in the spatial coding of stimulus location. This is the referential-coding account (Hommel, 1993a; Umilt~ & Nicoletti, 1985). In Stoffer's (1991) version of the attention-shift hypoth- esis (also see Stoffer & Yankin, 1994; but see Weeks, Chua, & Hamblin, 1996, for a failure to replicate), attention can be directed to different hierarchical levels of visual representa- tion, with the spatial codes that cause the Simon effect being produced when attention shifts laterally within a single representational level. In contrast, the Simon effect does not occur when attention zooms in from a higher order to a lower order level of representation. According to the version of the attention-shift hypothesis that was put forward by Umil~ and Nicoletti (1992; also see Nicoletti & Umil~, 1994), orienting attention to a stimulus produces the spatial code of its position, which in turn causes the Simon effect. This notion is based on the so-called premotor model of spatial attention (Rizzolatti, Riggio, Dascola, & Umil~, 1987; Rizzolatti, Riggio, & Sheliga, 1994; Umilt~, Riggio, Dascola, & Rizzolatti, 1991), in which covert attention orienting and overt attention orienting are strictly linked. In a Simon task, when the peripheral stimulus is pre- sented, attention covertly shifts toward its position. Accord- ing to the premotor model, covert orienting involves the preparatory stage of saccade generation (i.e., overt orient- ing), with the execution of the eye movement being with- held. In the motor program for the saccade, the directional feature is specified and becomes the spatial code of the stimulus. Evidence in favor of the attention-shift hypothesis srcinated from studies that showed that the Simon effect depended on the direction of the attention movement toward the stimulus (Nicoletti & Umilt~t, 1989, 1994) and that no Simon effect occurred when attention remained at fixation on stimulus presentation (Nicoletti & Umilt~, 1994). Hommel (1993a) countered the notion that attention shifting is the source of spatial coding and argued that a referential-coding account of the type proposed initially by Wallace (1971; also see Umilt~ & Nicoletti, 1985) may be sufficient to explain the Simon effect. In Hommel's view, spatial stimulus coding depends on the availability of frames or objects of reference, even though he admitted that the location of a reference object may well be the point of departure for an attention shift. Consider, for instance, Figure 1. When the stimulus on the right side is presented, the attention-shift account assumes that a right positional code for the stimulus is formed because attention shifts from fixation to the stimulus. In contrast, the referential-coding account assumes that the right positional code is formed because the fixation cross or the empty box (or both) on the left side acts as a reference object Accordingly, Hommel (1993a) proposed an alternative interpretation of the findings of Nicoletti and Umilt~ (1989). He argued that in that study, the stimuli were not coded relative to the position of attention but relative to a reference object, which also happened to be the object where attention was directed. Therefore, Nicoletti and Umilt~'s results could be explained within the framework of a pure referential- coding hypothesis, the object from which the attention shift was initiated acting as the reference object for the coding process. However, Hommel's interpretation cannot explain the results of the subsequent study by Nicoletti and Umilt~ (1994), in which, in the presence of a referential object at fixation, no Simon effect was found when attention did not move. Rationale of the Experiments We now consider how the spatial code of the stimulus might be formed. One can assume that spatial position is defined relative to a set of coordinates and that the srcin of the set of coordinates is determined by the position in space of a static reference object. This is what Hommel's (1993a) referential-coding hypothesis implicitly asserts. Alterna- tively, one can maintain that the srcin of the set of coordinates is determined by the position of the attentional focus, which can be moved in space. This is what Nicoletti and Umilt~ (1989) asserted. In contrast, Umilt~ and Nicolet- ti's (1992; Nicoletti & Umilt~, 1994) attention-shift hypoth- esis does not resort to the srcin of a set of coordinates and SQUARE i~ RECTANGLE Figure 1. Schematic drawing of the display used in Experiments 1-3. In Experiment 2, the two stimuli were more dissimilar.  SIMON EFFECT AND ATTENTION SHIFT 1355 asserts that what matters is not the position of the attentional focus but the direction of the attention shift. The saecade program that prodeces the attention shift contains a direc- tional feature. The directional feature is glued to the stimulus and becomes its spatial code. These three possibilities are difficult to evaluate indepen- dently, because all are based on the notion that the Simon effectis caused by the stimulus spatial code, that is, they all belong to the class of the coding hypotheses (Umilt~ & Liotti, 1987; Umiltb & Nicoletti, 1985; Wallace, 1971). In most experimental conditions, they make identical predic- tions. In fact, the referential-coding hypothesis and the first version of the attentional hypothesis are nearly indistinguish- able if one assumes, as Hommel (1993a) did, that the attentional focus is directed to the reference object. There are, however, experimental manipulations that allow one to pit the attention-shift hypothesis against the other two. Imagine that, in a display like the one depicted in Figure 1, attention is kept aligned with the fixation cross and does not move when the imperative stimulus is presented. The attention-shift hypothesis predicts that the Simon effect should not be observed because, in the absence of an attention movement, no directional feature is specified. Nicoletti and Umilt~ (I994) tested this prediction. On each trial of their Experiment 2, a letter was briefly presented at fixation, simultaneously with the imperative stimulus. They reasoned that attention did not have time to shift to the stimulus because it had to be kept at fixation, where a letter could signal a catch trial. Consistent with the prediction of the attention-shift hypothesis, no Simon effect was found. In contrast, ff what mattered was the position of the stimulus relative to a static reference object (i.e., the fixation cross, the empty box on the opposite side or both) or relative to the position of the attentional focus, which was aligned with the fixation cross, then the Simon effect should have been found. The attention-shift hypothesis can be distinguished from either the static or the dynamic version of the referential- coding hypothesis in another way. Imagine that the partici- pant is directing attention to an object at fixation and the imperative stimulus appears, for example, on the right side of fixation. If the participant shifts attention to the stimulus, the spatial stimulus code should be right according to every hypothesis. If the participant shifts attention to the opposite direction with respect to the stimulus, the spatial stimulus code should be left on the basis of the attention- shift hypothesis. This is because the motor program for the saccade contains a left-direction feature. In contrast, the spatial stimulus code should still be right on the basis of the referential-coding hypothesis because the stimulus re- mains on the right side relative to the reference object at fixation. The spatial stimulus code would still be right even if what mattered was the position of the attentional focus. This is because the stimulus remains on the right side relative to the current position of the attentional focus, which is moving to the left. Therefore, the attention-shift hypoth- esis is the only one that predicts a reversal of the Simon effect. Experiment 1 Experiment 1 was designed to discriminate the three hypotheses of the Simon effect on the basis of the foregoing reasoning. However, the experiment outlined earlier is difficult to perform in practice because attention is known to be automatically captured by abrupt increments of luminos- ity at the periphery of the visual field (e.g., Umilt~, 1988, for a review). It is possible to suppress the attention shift toward the peripheral stimulus or even to direct attention in a different direction (Yantis & Jonides, 1990). However, this operation is effortful and may succeed in just a few trials. Thus, there would be no guarantee that the reversing of the Simon effect, which is predicted by the attention-shift hypothesis, can be tested effectively. To overcome this difficulty, we had to use somewhat more complicated experimental conditions in this experiment and in those reported later. The rationale of this experiment, as well as that of the subsequent three experiments, is based on the assumption that attention moves, along the horizontal axis, from fixation to the imperative stimulus and then back to fixation. In Experiments 1 and 4, the attentional shift back to fixation would occur while the response is being selected. In contrast, in Experiments 2 and 3, the attentional shift back to fixation would not overlap in time with response selection. Method Participants. Fourteen students of the University of Modena were recruited for the experiment. All were fight-handed, had normal or corrected-to-normal visual acuity, and were naive about the purpose of the experiment. Apparatus and display. The participants were mated in front of a CRT screen driven by a Tulip dt 386SX computer. The head was positioned in an adjustable head-and-chin rest, so that the distance between the eyes and the screen was about 45 cm. The visual display (see Figure 1) included the following items: two empty boxes, 2 ° x 3.8 ° in size; one 1 ° x 1 cross, which was shown at the geometric center of the screen and posifinned'2.4 ° from the geometric center of the boxes; one 2.2 ° x 2 ° filled rectangle and one 2 ° x 2 ° filled square (i.e., the stimuli), which were shown, one at time, centered in one of the two boxes; and one letter (i.e., an H, N, Z, or W standard character from the computer's set), which was shown 1.9 ° below the fixation point (i.e., the cross). Responses were executed by pressing one of two keys on the computer keyboard. One key (the character z) was located to the left of the body midline and was pressed by the leftlndex finger, whereas the other (the character \) was located to the right and was pressed by the right index finger. The software was based on Micro Experimen- tal Laboratory (Schneider, 1988). Procedure. The fixation cross and the two boxes were visible on the screen throughout a trial. On each trial, a warning tone (25 ms in duration) was delivered, followed by a 300-ms interval, at the end of which one of the letters was shown for 100 ms below the fixation cross. After that, there was a 500-ms interval, at the end of which a stimulus (either the square or the rectangle) was shown for 100 ms inside one of the two boxes. Participants were instructed to maintain fixation and to press as fast as possible one of the two keys on stimulus presentation only if the letter was H, N, or Z. In the case of W, they had to refrain from responding (catch trials). Half the participants used the right-side key for the square and the left-side  1356 RUBICHI, NICOLETTI, ANL AND UMILTA 75 7OO 68 .i-o 45 4 O 71-Win ---n-- CAxr. /~ --o--- NOflCoff. e~l i424 , , , t bin1 bin2 bin3 bin4 I~n5 bin Figure 2. Means of individual reaction time (RT) bins for corresponding (Corr.) and noncorresponding (Noncorr.) trials in Experiment 1. key for the rectangle, whereas the others had the reverse assign- ment. At the end of each trial, feedback about reaction time (RT) and accuracy was provided below the fixation cross, followed by a l-s intertrial interval. Every participant was tested individually in one experimental session, which comprised 216 regular trials and 24 catch trials split into two equal blocks. The first experimental block was preceded by one block of practice trials. Stimulus presentation occurred according to a quasi-random sequence, with the constraints that both the square and the rectangle appeared half the time in one box and the other half in the other box, requiring half the time a right-hand response and the other half a left-hand response. Results Responses to catch trials were practically absent (0.9%), which confirmed that the participants were able to focus attention below the fixation point at the beginning of each trial. Errors were 12.6%. Only correct RTs in the 200- to 900-ms range were considered for the subsequent analysis. This yielded 3.0% outliers, of which 2.5% were RTs longer than 900 ms and 0.5% were RTs shorter than 200 ms. Using the Vincentization procedure introduced by Ratcliff (1979; also see De Jong et al., 1994), we calculated the mean RT for the first through the fifth bins of the rank-ordered raw data separately for corresponding and noncorresponding trials (Figure 2). Corresponding trials were those in which right-side and left-side keypresses were executed in re- sponse to right-side stimuli and left-side stimuli, respec- tively. Noncorresponding trials were those in which right- side and left-side keypresses were executed in response to left-side and right-side stimuli, respectively. 1 Correct RTs were entered into an analysis of variance (ANOVA) with two within-subjects variables: type of trial (corresponding vs. nonconesponding trials) and bin (first through fifth). All sources of variance were significant: F(I, 13) = 9.58, MSE = 1,805.74, p = .0085, for the type of trial main effect; F(4, 52) = 629.81, MSE = 491.62, p < .0001, for the bin wain effect; and F(4, 52) = 5.46, MSE = 123.50, p = .001, for the interaction. There was a 22-ms reverse Simon effect, with noncorre- sponding trials producing faster RTs than corresponding trials (542 vs. 564 ms; see Table 1 for a summary of regular and reverse Simon effects obtained in every experiment). Not surprisingly, RT lengthened from the first to the fifth bin (427, 495, 544, 597, and 705 ms, respectively). Palrwise comparisons (a = .05) showed that the reverse Simon effect was small and nonsignificant at the first bin (6 ms), whereas it was greater and significant a all other bins (19, 27, 31, and 27 ms, respectively; see Figure 2). An ANOVA on error data, with type of trial as a within-subjects variable, showed that fewer errors were made in noncorresponding than in corre- sponding trials, 5.35 versus 7.28, F(1, 13) = 4.95, MSE = 24.26, p = .044. Discussion The results of Experiment 1 demonstrate that, beginning from the second bin, there was a clear and reliable reverse Simon effect. We interpret this reverse Simon effect by assuming that the participants first shifted attention to the position of the imperative stimulus and then shifted attention back to fixation. Because the discrimination was difficult, the stimulus code relevant for selecting the response became available late. Therefore, the response was selected when t More precisely, we calculated the RT distributions for corre- sponding and noncorresponding trials for each participant. We then divided each of these distributions into five proportional bins such that each bin contained the same proportion (one fifth) of trials. The difference between these mean RTs is a bin-by-bin measure of the Simon effect. When plotted as a function of the average RT, it provides a measure of the changes over time in the magnitude of the Simon effect. Table 1 Mean Reaction Times and Standard Deviations, and Regular (+) or Reverse (-) Simon Effects, for Corresponding and Noncorresponding Trials in Experiments 1-4 and for Central Letter Trials in Experiment 4 Experiment 1 Experiment 2 Experiment 3 Experiment 4 Experiment 4 Variable M SD M SD M SD M SD (central letter trials) Corresponding trials 564 51.4 400 31.6 522 60.6 409 45.6 415 65.8 Noncorresponding trials 542 53.6 418 28.8 536 63.2 425 56.0 386 41.8 Simon effect -22 + 18 + 14 + 16 -29  SIMON F_~ ND TTENTION SHIFT 1357 attention was in the process of being shifted back from stimulus position to fixation. Some features of our experimental conditions were prob- ably instnnnental in inducing the observer to immediately shift attention back to fixation. The stimulus remained on the screen for only I00 ms.2 After it had disappeared, it was useless to maintain attention on its position. In addition, separating attention from the line of gaze requires some effort. If such effort is not useful for performing the task, the observer might have a tendency to realign attention with the fixation point. In fact, reorienting attention to fixation was an implicit task demand, considering that feedback about the response would soon appear at fixation followed by a letter that might signal a catch trial. These considerations give rise to two predictions. The regular Simon effect should be reinstated if the stimuli become easier to discriminate, even though they are shown briefly. This is because the relevant response code should become available in a shorter time, and thus the response should be selected before the attention shift back to fixation begins. The regular Simon effect should also be reinstated by presenting the stimulus for a longer exposure time, even though the discrimination is difficult. This is because maintaining attention at the location of a still visible stimulus would be beneficial. Therefore, the response should be selected before the attention shift back to fixation begins. In Experiments 2 and 3 we tested these predictions. Before proceeding, however, an alternative possibility needs to be taken into consideration. Hommel (1993b, 1994) reasoned that irrelevant spatial codes, not being actively maintained, are likely to decay over time and their impact on the process of response selection should decrease when response selection is delayed. As a consequence, the Simon effect should also decrease when response selection is delayed. Because the time point of response selection is determined by the time required to process the relevant stimulus feature, the lengthening of this process should result in a smaller Simon effect. In accordance with this prediction, Hommel found that the Simon effect diminished, or even disappeared, when stimulus discrimination was made more difficult. In Experiment 1, discrimination was difficult and RT was relatively slow; therefore, Hommel's (1993b, 1994) decay hypothesis might apply. However, a spontaneous, passive decay of the irrelevant spatial code can explain a small or null Simon effect, but it cannot explain its reversal. For clarity, it must be emphasized that Hommel's decay hypoth- esis is unrelated to his referential-coding hypothesis. The fact that irrelevant spatial codes may decay concerns the temporal dynamics of the Simon effect. It does not concern whether spatial codes are formed through attention shifting or referential coding. Therefore, the failure to explain the reversal of the Simon effect .in the slow bin region is damaging to the hypothesis of a passive decay, but it does not affect the tenability of the referential-coding hypothesis. Considering that the spatial code is irrelevant, one might suggest that it is actively inhibited (see, e.g., the review in Neill, Valdes, & Terry, 1995). For brevity, we assume that the irrelevant spatial code of the stimulus might be inhibited. However, inhibition might instead affect the response code that is associated with the location of the stimulus. Here and in what follows, it is not necessary to distinguish between the two possibilities because the reasoning applies to either the stimulus or the response cede. If it is assumed that inlfibition builds up over time, and thus its effects emerge only when discriaiination is difficult and RT is slow, the inhibition hypothesis is no doubt congruent with the results of Fatperiment 1. Note that De Jong et al. (1994), using a bin analysis, found that the regular Simon effect was present in the fast RT region and then reversed in the slow RT region. They interpreted the reversal of the Simon effect in terms of inhibition of the spatial code. 3 Experiment 2 This experiment was identical to Experiment 1, except that the discrimination between the square and the rectangle, which were much more dissimilar, was easier. We reasoned that overall RT should become faster because the relevant, nonspatial code of the stimulus should take less time to be formed. Therefore, the response should be selected before attention reoriented to fixation. As a consequence, a regular Simon effect should be obtained. Method Participants. Fourteen students of the University of Modena participated. They were selected as before, were naive about the purpose of the experiment, and had not taken part in the previous experiment. Apparatus, display, and procedure. Tbe appmatus and display were the same as those described earlier, except that the rectangle was 1 x 3.8 °. The procedure Was identical to that used in Experiment 1. Results and Discussion Responses on catch trials were rare (3.9%). Errors were 4.9%. With the same cutoff points as in the previous experiment (i.e., 200 and 900 ms), there were 0.8% outliers, of which 0.3% exceeded 900 ms and 0.5% were faster than 200 ms. Correct RTs and errors were entered in two ANOVAs identical to those already described for Experiment 1. The error data did not yield any significant effects. In the RT analysis, all sources of variance were significant: F(1, 13) = 2 Because of visual persistence, and in the absence of a mask, the duration of stimulus visibility is unlikely to coincide with stimulus exposure time. However, differences in stimulus exposure time should be related to differences in the duration of stimulus visibility. 3 To our knowledge, Hedge and Marsh (1975) were the first to report that when stimuli and responses were both iateralized and colored, the Simon effect could be made to reverse. They inter- preted this finding in terms of an identical logical Iransformation that was applied to all stimulus features. However, on the basis of a confounding that was present in Hedge and Marsh's experiment, Simon, Sly, and Vilapakkam (198i) offered a more parsimonious interpretation.
Related Search
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