Co-occurrence of sequential and practice effects in the Simon task: Evidence for two independent mechanisms affecting response selection


of 10
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 refers to the observation that responses to a relevant stimulus dimension are faster and more accurate when the stimulus and response spatially correspond than when they do not, even though stimulus position is irrelevant. Recent
  ©  2009 The Psychonomic Society, Inc.  358A human observer responds more efficiently to stimuli with a spatially corresponding response than with a spa-tially noncorresponding response, even if the location of the stimuli is irrelevant to the performance of the task and the response has to be emitted on the basis of a nonspa-tial stimulus feature (e.g., color or shape). For instance, if  participants are required to press a left key in response to a red square and a right key in response to a green square, responses will be faster when the red square appears on the left than when it appears on the right. The influence of the irrelevant spatial stimulus feature on performance is known as the Simon effect   (Simon & Rudell, 1967; for reviews, see Hommel & Prinz, 1997; Proctor & Vu, 2006), and it is believed to be the result of the interaction between two parallel and independent processing routes connect-ing perception to action (see, e.g., De Jong, Liang, & Lauber, 1994; Kornblum, 1994): an indirect route, called “conditional,” and a direct route, called “unconditional.” According to Barber and O’Leary (1997; see also Umiltà & Zorzi, 1997), the two processing routes rely on different memory associations connecting stimulus and response.When a stimulus appears, the conditional route acti-vates the required response on the basis of task-defined associations that connect a stimulus to a specific response. These associations are established by instructions and are supposed to be short-lived. The unconditional route acti-vates the response that spatially corresponds to the stimu-lus location through preexisting stimulus–response (S–R) associations, which are independent from instructions. These associations are supposed to be either determined genetically or overlearned as a result of extensive practice (Umiltà & Zorzi, 1997). When the two activated responses correspond, no competition arises; rather, the redundancy of the same response code might generate a facilitation effect (Umiltà, Rubichi, & Nicoletti, 1999). When they are different, however, the incorrect response needs to be aborted, thus causing a slowing of response time and an increased number of errors.The Simon effect has proven to be a robust phenom-enon whose magnitude can be reduced only partially through practice (see, e.g., Proctor & Lu, 1999; Simon, Craft, & Webster, 1973). Recent findings suggest that it can be strongly modulated by prior experience, however, at both a between- and a within-tasks level.When we consider the between-tasks level, we refer to the studies that have assessed the Simon effect as a function Co-occurrence of sequential and practice effects in the Simon task: Evidence for two independent mechanisms affecting response selection C RISTINA  I ANI   AND  S ANDRO  R  UBICHI University of Modena and Reggio Emilia, Reggio Emilia, Italy E LENA  G HERRI  Birkbeck College, University of London, London, England  AND R  OBERTO  N ICOLETTI University of Bologna, Bologna, Italy The Simon effect   refers to the observation that responses to a relevant stimulus dimension are faster and more accurate when the stimulus and response spatially correspond than when they do not, even though stimulus  position is irrelevant. Recent findings have suggested that the Simon effect can be strongly modulated by prior  practice with a spatially incompatible mapping and by correspondence sequence. Although practice is thought to influence conditional stimulus–response (S–R) processing, leaving response priming through the unconditional route unaffected, sequential effects are thought to represent trial-by-trial adaptations that selectively involve unconditional S–R processing. In the present study, we tested this assumption by assessing the effects of cor-respondence sequence both when the Simon task alone was performed and when it was preceded by a spatial compatibility task with either incompatible (Experiments 1–2) or compatible (Experiment 2) instructions. The observation that practice and correspondence sequence co-occur and exert additive effects strongly demonstrates that the two factors affect different processing routes.  Memory & Cognition2009, 37 (3), 358-367  doi:10.3758/MC.37.3.358 C. Iani,  S EQUENTIAL   AND  P RACTICE  E FFECTS   IN   THE  S IMON  T ASK   359 term links, the same mechanism may explain the effects of correspondence frequency (cf. Stürmer et al., 2002).A different mechanism is considered to be responsible for sequential effects. According to some researchers, these effects indicate that activation through long-term links, considered to be automatic, is instead influenced by top-down control processes (Mordkoff, 1998; Ridderink-hof, 2002; Stoffels, 1996; Stürmer et al., 2002), with com- pensatory adjustments taking place after the detection of a conflict in the current trial to avoid conflict in the follow-ing trial (e.g., Botvinick, Braver, Barch, Carter, & Cohen, 2001). Such a view is supported by both electrophysi-ological (e.g., Gratton, Coles, & Donchin, 1992; Stürmer & Leuthold, 2003; Stürmer et al., 2002) and neuroimaging (e.g., Egner & Hirsch, 2005) data. What is still a matter of debate is whether these adjustments entail a modulation of the flow of activation through the unconditional route, as suggested by Mordkoff, or a suppression of the route, as  postulated by Stürmer et al. (2002).Some researchers have questioned the idea that sequen-tial modulations of the Simon effect reflect changes in cognitive control and instead have considered them to be the result of S–R priming (Mayr, Awh, & Laurey, 2003) or binding effects (e.g., Hommel et al., 2004; Notebaert, Gevers, Verbruggen, & Liefooghe, 2006; Notebaert, Soe-tens, & Melis, 2001). Crucially, in the typical Simon task, correspondence sequence is confounded with the presence of stimulus and response repetitions in consecutive trials. More precisely, whereas sequences of two corresponding trials (C–C) and sequences of two noncorresponding trials (NC–NC) are either complete repetitions of stimulus posi-tion and response or complete changes of both stimulus  position and response, mixed sequences (C–NC or NC–C) are always partial repetitions in which either stimulus posi-tion or response repeats. Since responses to both complete repetitions and complete alternations are always faster than those to partial repetitions (Hommel, 1998, 2004; Hommel & Colzato, 2004; Pashler & Baylis, 1991), the Simon effect turns out to be larger after a corresponding trial (C–C and C–NC trials), whereas it is smaller after a noncorresponding trial (NC–C and NC–NC). Hence, the advantage of correspondence- level repetition may be accounted for by the repetition of specific stimulus and response features, obviating the need to assume the inter-vention of a top-down control mechanism. If this account is true, sequential modulations of the Simon effect will disappear when stimulus and response repetitions are con-trolled. As some recent studies have suggested, however, this is not always the case (see, e.g., di Pellegrino et al., 2007; Stürmer et al., 2002; Ullsperger, Bylsma, & Botvin-ick, 2005; Wühr & Ansorge, 2005).In summary, both practice and sequential effects are evidence of an adaptation to specific experiences by fa-voring a specific S–R link, irrespective of task goals. On one hand, practice and asymmetries in the frequency of noncorresponding trials are thought to affect the strength of short-term links, leaving response priming through the unconditional route unaffected. On the other hand, sequen-tial effects are thought to represent trial-by-trial adaptations selectively involving unconditional S–R processing.of practice in a different task that was performed immedi-ately before the Simon task (  practice effects ). These studies have indicated that practicing with an incompatible spatial mapping (e.g., responding to the left stimulus with the right key and vice versa) before performing a Simon task can eliminate (Tagliabue, Zorzi, Umiltà, & Bassignani, 2000) or even reverse the Simon effect (Proctor & Lu, 1999). This modulation occurs not only when stimuli in the two tasks are visual, but also when stimuli in the spatial compatibility task are auditory and stimuli in the Simon task are visual (Tagliabue, Zorzi, & Umil tà, 2002) and, with extensive  practice, when stimulus and response in the two tasks vary orthogonally along the horizontal and vertical dimensions (Vu, 2007). In contrast, practicing with a spatially compat-ible mapping does not affect the Simon effect.At a within-tasks level, the Simon effect is reversed when location-relevant trials with an incompatible map- ping are intermixed with location-irrelevant trials in the same task (Marble & Proctor, 2000; Proctor, Marble, & Vu, 2000). It is also reversed when noncorresponding trials are 80% of the total trials (Hommel, 1994), but it increases when corresponding trials predominate (Toth et al., 1995). When there are an equal number of corresponding and noncorresponding trials, the Simon effect has been shown to reduce (see, e.g., Praamstra, Kleine, & Schnitzler, 1999; Ridderinkhof, 2002), disappear (see, e.g., Mordkoff, 1998; Stürmer, Leuthold, Soetens, Schröter, & Sommer, 2002), or even reverse (di Pellegrino, Ciaramelli, & Làdavas, 2007; Hommel, Proctor, & Vu, 2004; Wendt, Kluwe, & Peters, 2006) after a noncorresponding trial, whereas a regular effect is evident after a corresponding trial. The latter phenomena are referred to as  sequential effects .Taken together, these results suggest that the influence of irrelevant spatial information on performance is not as unavoidable as was once thought, but rather depends on the preceding experience and on the specific task context. Even though the majority of the accounts of between-tasks and within-tasks effects agree with this conclusion, there is disagreement regarding exactly how experience can influence performance.To explain the elimination (and reversal) of the Simon effect after practice with an incompatible mapping, Tagli-abue et al. (2000) introduced the notion of “long- lasting short-term links.” According to their hypothesis, the short-term (task-related) association between a stimulus location and the incompatible response that was created in order to perform the spatial compatibility task remains ac-tive and influences performance in the subsequent Simon task. This may occur because, as suggested by Logan’s (1988) automatization theory, if the same S–R mapping is repeatedly used, the S–R association is stored into the memory. As a consequence, when a stimulus with a given spatial code occurs, the response that has been associated with it for repeated instances is retrieved automatically, irrespective of a change in task instructions. According to this view, practice is not supposed to affect the long-term associations, which are considered unmodifiable. Note that, if one assumes that the presence of a higher number of noncorresponding trials establishes a prefer-ential noncorresponding S–R association through short-  360 I ANI , R  UBICHI , G HERRI , AND  N ICOLETTI Procedure . In the practice session, the stimulus was white, and  participants were asked to respond as quickly and accurately as pos-sible to the left stimulus with the right response and to the right stimulus with the left response. The practice session consisted of 384 trials that were divided into four blocks of 96 trials each.In the Simon task, participants were asked to respond as quickly and as accurately as possible to the color of the stimulus, ignoring its location. Half of the participants responded to the red square with the left hand and to the green square with the right hand, and the other half experienced the inverse mapping rule. The task consisted of 768 trials that were divided into eight blocks of 96 trials each, pre-ceded by 32 practice trials. Each block was composed of 32 three-trial sequences in which the first trial was named n  2, the second trial was named n  1, and the third trial was named n . Since each of the trials could be either corresponding or noncorresponding, eight different sequences were used (C–C–  C  , C–C–   NC  , C–NC–  C  , C–NC–   NC  , NC–C–  C  , NC–C–   NC  , NC–NC–  C  , NC–NC–   NC  ,   with italics denoting trial n ) and were repeated four times in each ex- perimental block. In this way, trial n  was preceded by either a cor-responding or a noncorresponding response with equal probability, as was trial n  1.In both tasks, the stimulus remained present for 1 sec, or until a response was made. The trial terminated if the participant did not respond within 1 sec. Visual feedback on speed and accuracy was  provided at the center of the screen for 500 msec. The intertrial in-terval was 1 sec. Results Only the data of the Simon task were analyzed. Late responses (reaction times [RTs]   1,000 msec) averaged 0.08%. Responses faster than 150 msec were excluded from the analysis (0.1%), as were incorrect responses and responses that were preceded by two incorrect responses (6.7%). Analysis of correspondence sequence . Mean RTs and arcsine-transformed error rates in trial n  were submit-ted to repeated measures ANOVAs with group (control vs. incompatible-practice group) as a between-subjects fac-tor, and with trial n  1 correspondence and trial n  corre-spondence as within-subjects factors. When necessary, comparisons were performed using Bonferroni’s test for multiple comparisons. The respective data are displayed in Figure 1. 1 The RT analysis showed that responses were faster after a corresponding trial (389 msec) than after a noncor-responding (395 msec) trial, as indicated by the signifi-cant main effect of trial n  1 correspondence [  F  (1,30)   11.00,  MS  e     97,  p     .01,   p2     .27]. Corresponding tri-als (385 msec) were faster than noncorresponding trials (399 msec), as indicated by the significant main effect of correspondence on trial n  [  F  (1,30)   27.50,  MS  e     228,  p     .001,   p2     .48]. This correspondence effect was mod-ulated by prior practice, as indicated by the interaction be-tween trial n  correspondence and group [  F  (1,30)   44.99,  MS  e     228,  p     .001,   p2     .60], and by correspondence sequence, as indicated by the interaction between trial n  correspondence and trial n  1 correspondence [  F  (1,30)   162.67,  MS  e     218,  p     .001,   p2     .84]. Most importantly for the present purposes, there was a significant three-way interaction between group, trial n  1 correspondence, and trial n  correspondence [  F  (1,30)   4.69,  MS  e     218,  p     .04,   p2     .14].Following the logic of the additive factors method (Sternberg, 1969), if practice and correspondence sequence influence different mechanisms (that is, the conditional and unconditional routes of processing, respectively)—as can be concluded on the basis of the accounts described above—then their effects on performance will be additive. Alternatively, an interaction between these factors would suggest the possibility that they exert their influence on the same mechanism (that is, both influence the activity of the same processing route); as a consequence, either the accounts of practice effects or those of sequential effects would need to be revised.Two experiments were conducted to test these predic-tions. In both experiments, practice and correspondence sequence were concurrently manipulated. In Experi-ment 1, half of the participants performed only the Simon task, whereas the other half transferred to the Simon task after practicing a spatial-compatibility task with an incom- patible mapping. In Experiment 2, participants practiced with either a spatially incompatible mapping or a spatially compatible mapping before transferring to the Simon task. The orientation of the S–R set during the compatible prac-tice was either horizontal or vertical. This design allowed us to disentangle the specific, mapping-related effects of  practice from the nonspecific effects. EXPERIMENT 1 In Experiment 1, we assessed modulations of the Simon effect as induced by correspondence sequence when only the Simon task was performed or when it was performed after practice with a spatially incompatible S–R mapping. If correspondence sequence and prior practice affect the same mechanism, they will interact. If, as suggested by the accounts described above, they affect different mecha-nisms, they will exert additive effects.Furthermore, to explore whether correspondence se-quence effects could result from stimulus and response features repetition, we assessed sequential modulations  before and after the exclusion of both complete repeti-tions (that is, repetitions of both stimulus position and re-sponse) and response repetitions (e.g., di Pellegrino et al., 2007; Wühr & Ansorge, 2005). Method Participants . Thirty-two right-handed undergraduate stu-dents (20–32 years of age; 20 female and 12 male) with normal or corrected- to-normal vision participated in the experiment. Sixteen  performed only a Simon task, whereas 16 practiced with a spatially compatible task with an incompatible S–R mapping and then trans-ferred, after a 5-min interval, to the Simon task. Apparatus and Stimuli . Participants sat in front of a color moni-tor that was controlled by an IBM computer, in a dimly illuminated room, and at a viewing distance of approximately 57 cm. Stimulus  presentation and response collection were controlled by the E-Prime software system. The stimuli were white, red, or green solid squares (2º   2º), presented to the left or to the right of fixation with an ec-centricity of 5º.Responses were executed by pressing the “G” or the “L” key on the keyboard with the left or the right index finger, respectively. The keyboard was located centrally with respect to the body midline.  S EQUENTIAL   AND  P RACTICE  E FFECTS   IN   THE  S IMON  T ASK   361 105,  p     .001,   p2     .79]. The latter difference was larger for the incompatible-practice group (34 msec) than for the control group (21 msec), as indicated by the interaction  between group and trial n  1 correspondence [  F  (1,30)   6.98,  MS  e     105,  p     .02,   p2     .19].With regard to errors, the interaction between group and trial n  correspondence reached significance [  F  (1,30)   36.53,  MS  e     0.009,  p     .01,   p2     .55]. This was because the incompatible practice group showed a reverse Simon effect, with fewer errors in noncorresponding trials (3.5% vs. 7.6% in the control group and 8.7% vs. 4.0% in the incompatible-practice group for corresponding and noncor-responding trials, respectively). There was also a significant interaction between trial n  correspondence and trial n  1 correspondence [  F  (1,30)   81.01,  MS  e     0.005,  p     .001,   p2     .73]. After a corresponding trial, there was a signifi-cant Simon effect (5.3%), whereas a reverse effect was evi-dent after a noncorresponding trial (  4.7%). The interac-tion between group, trial n  correspondence, and trial n  1 correspondence did not reach significance (  p     .19).After a corresponding trial n  1, the Simon effects were 59 and 35 msec for the control and incompatible-  practice groups, respectively. Pairwise comparisons showed that this difference was significant. After a noncorrespond-ing trial n  1, no Simon effect was evident for the control group (4 msec, n.s.), but a 43-msec reverse effect was evi-dent for the incompatible-practice group. Pairwise com- parisons showed that this difference was significant.To better understand the three-way interaction, we ran separate ANOVAs on corresponding and noncorre-sponding trial n s. Corresponding responses did not differ  between the two groups and were faster when trial n  1 was corresponding (366 msec) than when it was noncor-responding (405 msec) [  F  (1,30)   116.32,  MS  e     210,  p     .001,   p2     .80]. Noncorresponding responses were faster for the incompatible-practice group (381 msec) than for the control group (418 msec) [  F  (1,30)   4.95,  MS  e     4,462,  p     .04,   p2     .14], and they were faster when trial n  1 was noncorresponding (385 msec) than when it was corresponding (413 msec) [  F  (1,30)   115.13,  MS  e      200250300350400450500C NC Control C NC n  1 Incompatible Practice C NC Control C NC Incompatible Practice     R    T    (   m   s   e   c    ) CNC05101520     E   r   r   o   r   s    (    %    ) 59 * 4 –43 * 35 * n  1 n  1  n  1 Figure 1. Mean reaction times (RTs, upper panel) and percentages of error (lower panel) for the current event in Experiment 1, as a function of practice group (control vs. incompat-ible practice) and preceding event (trial  n  1) correspondence. Error bars indicate standard errors of the mean. The magnitude of the Simon effect for separate conditions is reported on top. Asterisks denote significant values (  p     .05). C, corresponding; NC, noncorresponding.  362 I ANI , R  UBICHI , G HERRI , AND  N ICOLETTI Discussion The results of Experiment 1 showed that the Simon ef-fect was modulated by practice and by correspondence sequence. Interestingly, 384 practice trials with an incom- patible mapping reversed the Simon effect in errors but not in RTs. Since the Simon task in the present study was lon-ger than those in previous studies, we analyzed the Simon effect as a function of experimental block to exclude the  possibility that transfer effects dissipated throughout the 768-trial transfer session. The interaction between block and correspondence did not reach significance (  F   1), however, suggesting that only a very long practice session can reverse the overall effect on RTs.At a within-tasks level, the Simon effect on RTs was modulated by the correspondence of the preceding trial in different ways for the two experimental groups. When the Simon task alone was performed, a large Simon effect (59 msec) was evident following a corresponding trial, whereas it was eliminated following a noncorresponding trial. These results replicate those of previous studies (e.g., Stürmer et al., 2002) and support the idea that the detec-tion of a conflict in trial n  1 triggers adaptations that are aimed at eliminating the impact of spatial S–R correspon-dence on response selection in the following trial. The ob-servation that this pattern of results persisted even when exact stimulus and response repetitions were excluded from the analysis supports the idea that sequential mod-ulations are not due simply to feature repetition and/or response-priming effects. Rather, these results, along with those of studies that used other conflict tasks—such as the Stroop (Kerns et al., 2004) and the flanker tasks (Botvin-ick, Cohen, & Carter, 2004; Botvinick, Nystrom, Fissel, Carter, & Cohen, 1999)—provide further support for the view that the automatic activation of the response cor-responding to stimulus location is permeable to context- dependent influences.The finding that the Simon effect for errors reversed after a noncorresponding trial suggests that these modu-lations may not act simply by suppressing the long-term corresponding associations but rather by changing the weight of both corresponding and noncorresponding S–R associations, favoring either one or the other depending on the corresponding level of the preceding trial. Indeed, the suppression of the automatic activation of a correspond-ing response can explain the absence of the Simon effect  but not its reversal: Since the unconditional route is sup- pressed after a noncorresponding trial, a response would  be selected solely on the basis of the relevant (nonspatial) stimulus dimension; hence, no advantage for correspond-ing responses would be evident. 2 The spatially incompatible practice had no effect on corresponding RTs, which did not differ between the two groups and were faster than noncorresponding RTs when preceded by a corresponding event. It speeded up noncorresponding responses, however, thus leading to a reduced but still present 35-msec Simon effect follow-ing a corresponding event and resulting in a reverse effect (  43 msec) following a noncorresponding event.Interestingly, the interaction between practice and cor-respondence sequence, a result that contradicts the idea Analysis of stimulus and response feature sequence . Following Hommel et al. (2004), data were considered in terms of repetition or alternation of the response and the stimulus location in the current and preceding trial, and they were entered into a repeated measures ANOVA with practice group as a between-subjects factor and with corre spondence of trial n , stimulus-location sequence (repetition vs. alternation), and response sequence (repeti-tion vs. alternation) as within-subjects factors. The inter-action between stimulus-location sequence and response sequence was significant [  F  (1,30)   155.57,  MS  e     457.82,  p     .001,   p2     .84], and it was modulated by group [  F  (1,30)   4.33,  MS  e     457.82,  p     .05,   p2     .13]. Separate analyses by group showed that the interaction be-tween stimulus-location sequence and response sequence was significant for both groups. For the control group, response repetitions were 28 msec faster when the stimu-lus position was repeated (383 vs. 411 msec for total rep-etitions and response repetitions, respectively), whereas alternations of the response were 27 msec faster when the stimulus location was alternated (393 vs. 420 msec for total alternations and response alternations, respectively) [  F  (1,15)   54.34,  MS  e     454.97,  p     .001,   p2     .78]. For the incompatible-  practice group, response repetitions were 38 msec faster when the stimulus position was re- peated (363 vs. 401 msec for total repetitions and response repetitions, respectively), whereas alternations of the re-sponse were 39 msec faster when the stimulus location was alternated (364 vs. 403 msec for total alternations and response alternations, respectively) [  F  (1,15)   105.23,  MS  e     460.68,  p     .001,   p2     .87].Furthermore, the interaction among stimulus-location sequence, response sequence, and trial n  correspondence was significant [  F  (1,30)   8.61,  MS  e     204.99,  p     .01,     p2     .22]. Pairwise comparisons showed that the dif-ference between corresponding and noncorresponding responses was significant after complete alternations (26 msec) but not after complete repetitions (12 msec) or after partial repetitions of either the response (13 msec) or the stimulus position (4 msec).To exclude the possibility that the results from the correspondence- sequence analysis were due to repeti-tions, all sequences that involved complete repetitions (i.e., sequences in which both response and stimulus posi-tion were repeated) or partial repetitions (i.e., sequences in which the response was repeated but stimulus position changed) were excluded from the analysis. In this way, 50% of C–C and NC–NC sequences and 50% of NC–C and C–NC sequences were excluded. The remaining data were analyzed as a function of practice group and corre-spondence level in trial n  1 and trial n .In line with the correspondence-sequence analysis, the interaction between trial n  1 and trial n  correspondence was significant [  F  (1,30)   87.65,  MS  e     409.7,  p     .001,   p2     .75], indicating that a positive Simon effect (49 msec) was present following a corresponding trial, whereas a reverse effect (  18 msec) was evident follow-ing a noncorresponding trial. In contrast with the previ-ous analysis, this interaction was not modulated by group (  p     .11).
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