This article was downloaded by: [Michigan State University] On: 28 February 2015, At: 04:02 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

The Quarterly Journal of Experimental Psychology Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/pqje20

You can't always get what you want: The influence of unexpected task constraint on voluntary task switching ab

c

d

ab

Starla M. Weaver , John J. Foxe , Marina Shpaner & Glenn R. Wylie a

Kessler Foundation Research Center, West Orange, NJ, USA

b

Rutgers New Jersey Medical School, Newark, NJ, USA

c

Albert Einstein College of Medicine, Yeshiva University, New York, NY, USA

d

Click for updates

University of Vermont, Burlington, VT, USA Published online: 11 Jun 2014.

To cite this article: Starla M. Weaver, John J. Foxe, Marina Shpaner & Glenn R. Wylie (2014) You can't always get what you want: The influence of unexpected task constraint on voluntary task switching, The Quarterly Journal of Experimental Psychology, 67:11, 2247-2259, DOI: 10.1080/17470218.2014.917115 To link to this article: http://dx.doi.org/10.1080/17470218.2014.917115

PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or

Downloaded by [Michigan State University] at 04:02 28 February 2015

distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2014 Vol. 67, No. 11, 2247–2259, http://dx.doi.org/10.1080/17470218.2014.917115

You can’t always get what you want: The influence of unexpected task constraint on voluntary task switching Starla M. Weaver1,2, John J. Foxe3, Marina Shpaner4, and Glenn R. Wylie1,2 1

Kessler Foundation Research Center, West Orange, NJ, USA Rutgers New Jersey Medical School, Newark, NJ, USA 3 Albert Einstein College of Medicine, Yeshiva University, New York, NY, USA 4 University of Vermont, Burlington, VT, USA

Downloaded by [Michigan State University] at 04:02 28 February 2015

2

The current study assessed the effect that unexpected task constraint, following self-generated task choice, has on task switching performance. Participants performed a modified double-registration voluntary task switching procedure in which participants specified the task they wanted to perform, were presented with a cue that, on the majority of trials, confirmed the choice, and then performed the cued task. On a small portion of trials, participants were cued to perform a task that did not match their choice. Trials on which cues unexpectedly failed to match the chosen task were associated with costs. These costs were particularly large when participants chose to switch tasks but had to unexpectedly repeat the previous task. The results suggest that when participants choose to switch tasks, they prepare for that switch in anticipation of the stimulus, and the preparation is durable such that it cannot be readily undone. Keywords: Task switching; Voluntary task switching; Task preparation.

Cognitive flexibility, or the ability to adopt one action sequence out of several, is central to human cognition. While many animals are able to quickly learn to associate a given action with a particular stimulus (or context), our ability to flexibly switch from one action sequence to another “at will” is one of the things that, we like to think, sets humans apart. This has engendered a field of research aimed to better understand this important ability, which uses the task switching paradigm. In this paradigm, participants are given at least two tasks; on some trials, they are required to repeat the task performed previously, while on other trials they are asked to switch tasks. As many studies have now shown, there is a large performance cost associated with switching tasks (longer response

latencies, and higher rates of errors). Initially, this switch cost was interpreted as a relevantly direct measure of the time needed to perform the executive function of instantiating a new task (Jersild, 1927), but further inquiry showed that a large part of the switch cost derived from some sort of interference (Meiran, 1996; Wylie & Allport, 2000). While debate continues about exactly what the switch cost represents, consensus has begun to emerge that at least some portion of the switch cost represents the time necessary to overcome interference from previously relevant tasks (Vandierendonck, Liefooghe, & Verbruggen, 2010). While the task switching paradigm has furthered our understanding of how control of action is achieved by the brain, several investigators have

Correspondence should be addressed to Glenn R. Wylie, Kessler Foundation Research Center, Neuropsychology & Neuroscience Laboratory, 300 Executive Drive Suite 70, West Orange, NJ 07052, USA. E-mail: [email protected] ©2014 The Experimental Psychology Society

2247

Downloaded by [Michigan State University] at 04:02 28 February 2015

WEAVER ET AL.

pointed out that it might not be the best paradigm for studying willed action (Logan & Bundesen, 2003). Task switching experiments generally use cues (or a predictable task sequence in conjunction with cues) to inform participants of which task to perform on each trial (e.g., Rogers & Monsell, 1995). For example, in some experiments, participants repeat and switch tasks as directed by cues, which change randomly across trials. This approach has clear advantages from the experimenter’s perspective: The cues at once allow the participant to know which task to perform and the experimenter to determine whether the task was performed correctly. However, from the standpoint of investigating willed action, this paradigm is less attractive because participants are only doing what they are told to do on every trial. That is, participants are not themselves deciding which task to perform, and such decisions are central to what we mean by “willed action” or “free choice”. This has led researchers to develop a voluntary task switching paradigm in which participants themselves choose which task to perform on each trial (Arrington & Logan, 2004). As with nonvoluntary paradigms, voluntarily switching tasks results in a performance cost that can be interpreted as stemming from processes associated with reconfiguring or biasing the cognitive system for the forthcoming task and overcoming interference from previously relevant tasks. Nonvoluntary (e.g., cued) and voluntary task switching paradigms capture two situations that occur outside the laboratory: Sometimes you will perform actions that you have been assigned to do, and other times you will be free to choose which actions you perform (within the confines of the current context). However, a third situation frequently occurs that has received relatively little study: when you freely choose an action that you intend to perform, but find that your actual performance is unexpectedly constrained. For example, you may choose to drive down a particular route but find that the road is blocked, or intend to see a particular movie but find the show is sold out. The current study assessed the effect that unexpected task constraint, following self-generated task choice, has on performance.

2248

Previous research has explored the need to unexpectedly change plans using cued task switching paradigms in which an initial cue presented 1000 ms prior to the stimulus is sometimes made invalid by the presentation of an overriding cue at the time when the stimulus appears (Hübner, Kluwe, Luna-Rodriguez, & Peters, 2004; Ruthruff, Remington, & Johnston, 2001; Wendt, Luna-Rodriguez, Reisenauer, Jacobsen, & Dreisbach, 2012). Invalid trials are performed more slowly and are more likely to result in errors than validly cued trials. The cost appears to be separate from costs associated with stimulus-evoked competition (Hübner et al., 2004) or betweentask competition (Ruthruff et al., 2001) and instead appears to result from the need to unexpectedly override top-down processes that prepare for the invalidly cued task. The extent that unexpected task constraint influences performance following a voluntarily chosen task is also expected to be influenced by the need to override top-down preparation processes. Performance is likely to depend on the amount of preparation one has engaged in and the flexibility with which one can override that preparation once a task constraint is encountered. Voluntary task switching, like task switching, appears to be influenced by top-down preparation processes (Arrington & Logan, 2005). For example, increasing the interval between trials results in more quick and accurate performance (see Arrington, Reiman, & Weaver, in press, for review). However, in voluntary task switching participants are free to choose not only what task to perform, but when to make that choice. As a result it is not possible to differentiate between benefits at longer intertrial intervals that are due to active preparation processes and those that are due to passive processes, such as decay of previous trial activation. So while top-down processes are expected to play a role in task preparation following task choice (Arrington & Logan, 2005), the extent of this preparation (Vandamme, Szmalec, Liefooghe, & Vandierendonck, 2010), and in particular the flexibility with which one can perform an unexpected task after engaging in preparation processes that accompany task choice, is unclear.

THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2014, 67 (11)

UNEXPECTED TASK CONSTRAINT

Downloaded by [Michigan State University] at 04:02 28 February 2015

Figure 1. Depiction of timeline for a single trial for Experiment 1. Stimuli were either red or blue, and the task-tracking graphic at the bottom of the task prompt screen was green.

Here, we developed a paradigm that allowed us to investigate unexpected task constraint within a voluntary task switching framework. To do this we added a cue to the double registration variant of the voluntary task switching paradigm (see Figure 1). Participants were first presented with a prompt that allowed them to choose the task they wished to perform on the upcoming trial. Participants indicated their choice via key-press. Participants were then presented with a cue that indicated the task that should be performed on the upcoming trial. Finally, the target stimulus was presented, and participants performed the cued task. On the majority of trials (valid trials) the cue simply confirmed the participant’s task choice. That is, the task indicated by the cue matched the task that was chosen by the participant. However, on a small portion of trials the unchosen task was cued (invalid trials). Comparing performance on valid and invalid trials allowed for an assessment of the effect that unexpected task constraint, following self-generated task choice, has on performance.

EXPERIMENT 1 Method Participants Experiment 1 included 12 participants, but the data from one were lost due to equipment failure. All participants had normal or corrected-to-normal visual acuity and normal colour vision. Participants were paid $10 an hour for their participation. The procedures were approved by the

Institutional Review Board (IRB) of the Nathan Kline Institute, and all participants provided written informed consent. Apparatus A Dell Optiplex GX 260 microcomputer equipped with an Iiyama Vision Master Pro 514 CRT monitor was used to present stimuli and a USB response pad to record responses. The experiment was programmed using Presentation® software (Version 8.0, www.neurobs.com). Cues were either a square (80 mm2) or a diamond (the same 80 mm2 square, rotated 45°). Each cue was assigned to one of two tasks (e.g., the square cued the colour task and the diamond the letter task), and cue–task assignment was counterbalanced across participants. Task, stimuli, and responses Participants choose between performing the letter task and performing the colour task. The letter task required categorizing the stimulus as either an X or an O, and the colour task required categorizing the stimulus as either red or blue. To indicate their task choice participants used their right index and middle finger to press one of two keys on a USB response pad. Each key was associated with a single task (e.g., the left key was associated with the letter task and the right key with the colour task), and this association was counterbalanced across participants. Stimuli were the letters X (17 mm wide and 15 mm tall) or O (16 mm wide and 15 mm tall) presented in either red—red, green, blue (RGB) values in percentages = [100, 0, 0]—or blue—RGB = [0, 0, 100]. Participants responded with a two-alternative, forced-choice

THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2014, 67 (11)

2249

Downloaded by [Michigan State University] at 04:02 28 February 2015

WEAVER ET AL.

key-press. Responses for both tasks were mapped to the same two keys. Response keys were located immediately below the two keys used to indicate task choice on the USB response pad. The mapping of categories to responses was counterbalanced across participants. Participants were instructed to choose each task approximately equally and randomly. To guard against participants performing one task preferentially, a task-tracking graphic was presented at the bottom of the screen. It consisted of a horizontal green line (90 mm) with the word “Letter” at one end, the word “Color” at the other end, and a green hash mark (5 mm vertical line) in the centre. A short (5-mm) vertical red line located on this line served as a visual representation of the ratio of the number of times one task had been performed relative to the other. Thus, on the first trial, the red line bisected the green line (because each task had been performed an equal number of times: zero). After each trial, the location of the red line was updated, moving towards the task presented on that trial. Participants were instructed to use the task-tracking graphic to guard against performing one task preferentially. They were told to avoid keeping the red line in the middle at all times, but rather to use the graphic as a loose guide. The task-tracking graphic remained at the bottom of the screen throughout the experiment. Procedure The timeline of one trial is displayed in Figure 1. For the data under consideration, each trial began with a prompt where participants indicated the task they wanted to perform on the upcoming trial. Prompt time was unconstrained. The amount of time participants took to respond to the prompt was recorded in order to capture task choice response time (RT). Immediately following task selection a cue was presented. The majority of trials were valid, such that the task indicated by the cue matched that chosen by the participant. However, a randomly selected 15% of trials were invalid, such that the task that was cued was the task the participant had not chosen. On all trials, participants were required to perform the task

2250

that was cued, rather than the task that was chosen. The cue was presented for 700 ms and was followed by the presentation of the stimulus, which appeared at fixation for 200 ms surrounded by the cue. A response period then ensued that lasted until a response was made or 3 seconds elapsed. Trials occurred in sets of three, such that following a response to the first trial in the triad, two additional trials of the same task were presented. Each of the additional trials began as soon as the response period for the previous trial had ended. A stimulus surrounded by the cue (the cue was the same for all three trials of each triad) was presented for 200 ms and was followed by a response period that lasted until a response was made or 3 seconds had elapsed. The task choice prompt immediately followed the response period for the third trial. However, as first trial performance is of greatest interest here, performance of only the first trial of each triad is reported. Participants were seated a comfortable viewing distance from the monitor (approximately 76 cm) and received instructions. They then completed a practice block of 100 trials, all of which were valid. Prior to beginning the experimental blocks, participants were informed that 15% of trials would be invalid. Participants then performed three blocks of 140 trials. Due to the invalid and voluntary components of the procedure, the number of trials in each cell of the design varied. However, the mean trial number in each cell of the design was greater than 25.

Results For the task performance analyses, trials were sorted into repetitions and switches based on the cue presented on trial n and trial n − 1. For the task choice analysis, trials were sorted into repetitions and switches based on the task chosen on trial n and the task cued on trial n−1. The first trial of each block was excluded from all analyses as transitions could not be coded for these trials. Error trials and trials following an error were excluded from the all analyses except those assessing accuracy. Choice RT and task RT analyses were conducted using median RTs. Post hoc tests

THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2014, 67 (11)

UNEXPECTED TASK CONSTRAINT

Downloaded by [Michigan State University] at 04:02 28 February 2015

were conducted where Bonferroni’s corrections.

appropriate

using

Task performance Task RT was assessed via a 2 (validity: valid, invalid) × 2(transition: repetition, switch) repeated measure analysis of variance (ANOVA).1 A main effect of validity was found, such that valid trials (M = 608 ms) were faster than invalid trials (M = 815 ms), F(1, 11) = 5.50, p = .04, η2p = .36. The main effect of transition was not significant, F , 1. However, there was a significant Validity × Transition interaction, F(1, 10) = 7.03, p = .02, η2p = .41. As displayed in Figure 2, while the typical pattern of switch costs (repetitions performed more quickly than switches) was found in the valid condition, a reversal of that pattern was found in the invalid condition such that repetitions were performed more slowly than switches. Accuracy was assessed using the same analysis. The only result was a marginally significant effect of validity, F(1, 10) = 3.64, p = .09, η2p = .27. Participants tended to be less accurate on invalid trials (M = .81) than on valid trials (M = .88). Task choice The proportion of switches chosen was analysed separately for trials on which the previous trial had been valid and trials on which the previous trial had been invalid. As expected, based on previous voluntary task switching literature, when the previous trial was valid, a significant repetition bias was found (M = .35), t(10) = 2.31, p = .04. However, when the previous trial had been invalid, participants showed a bias for switching away from the task cued on the invalid trial (M = .65), t(10) = 2.46, p = .03. RTs for task choice also varied as a function of previous trial validity. A 2 (previous trial validity: valid, invalid) × 2 (transition: repetition, switch) repeated measure ANOVA on choice RT found a main effect of previous trial validity, F(1, 10) = 53.70, p , .001, η2p = .84, such that participants

were faster to select the current task when the previous trial was valid than when the previous trial was invalid (see Table 1). The Previous Trial Validity × Transition interaction was marginally significant, F(1, 10) = 4.47, p = .06, η2p = .31. The choice to repeat a task was made more quickly than the choice to switch tasks when the previous trial had been valid; however, this pattern was reversed when the previous trial was invalid.

Discussion The current experiment assessed the effect of unexpected task constraint following voluntary task choice on performance. Previous research using the cued task switching paradigm has found that invalidly cued trials are performed less quickly than validly cued trials (Hübner et al., 2004; Ruthruff et al., 2001; Wendt et al., 2012). The current results replicate and extend this finding by demonstrating a similar invalid task impairment following voluntary choices. It seems that when participants choose to perform a task on the upcoming trial they engage in preparation for the chosen task, such that responding is impaired if performance of an alternative task is required. Evidence of task preparation in the current study, in which the task cue appeared immediately after task selection, suggests that participants were engaging in preparation for the upcoming task prior to indicating their task choice. In other words, choice RTs probably reflected not only the time required to select the upcoming task but also preparation for that task. The amount of preparation that participants engaged in prior to indicating their choice appeared to vary as a function of transition type. On invalid trials the typical pattern of switch costs, found in the valid condition and in previous voluntary task switching studies, was reversed due to a large increase in the time required to perform a task repetition once participants had chosen to perform a

1

Switch costs can be artificially inflated by trials on which both the stimulus and response repeat (Hübner et al., 2004). To protect against this possibility an additional analysis was conducted in which all stimulus–response repetitions were removed. Within this analysis, the main effect of validity remained significant. The significance of the interaction became marginal, p = .05. THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2014, 67 (11)

2251

Downloaded by [Michigan State University] at 04:02 28 February 2015

WEAVER ET AL.

Figure 2. Task response time (RT) for Experiment 1 as a function of expectancy and transition. Error bars represent standard error of the mean.

Table 1. Mean choice RT as a function of previous trial validity and transition for Experiments 1 and 2 Transition Experiment 1 Repetitions Switches Experiment 2 Repetitions Switches

Previously valid

Previously invalid

692 (29) 819 (33)

1083 (37) 918 (36)

616 (23) 922 (35)

795 (30) 868 (36)

Note: RT = response time, in ms. Standard errors in parentheses.

task switch. When participants chose to switch tasks, a great deal of preparation for that switch appears to have occurred prior to responding to the task prompt, such that this preparation could not be easily undone when participants were cued to perform a task repetition. So while preparation for the chosen task appeared to occur on both switch and repetition trials, that preparation was particularly difficult to overcome on invalid repetition trials. That participants were unable to undo early task preparation is somewhat surprising given the long interval allotted for preparation of the invalid task following the presentation of the cue. Previous task switching studies suggest that switch cost asymptotes at around 600 ms (e.g., Rogers &

2252

Monsell 1995; Wylie, Javitt, & Foxe, 2006). That is, as participants are provided more time to prepare for a task switch, the difference between performance on switch and repeat trials diminishes (but is almost never completely abolished), and the vast majority of incomplete task preparation can be carried out within the first 600 ms after the cue. In Experiment 1 the cue-to-stimulus interval (CSI) was 700 ms. In other words, on invalid trials, participants had 700 ms to prepare for the invalidly cued task following the presentation of the cue. Why was this interval not sufficient to allow for complete task preparation, or task preparation that matched that found in the valid condition? One possibility is that while 700 ms is sufficient time to switch to a new task (that is, to instantiate a new task representation), 700 ms is not enough time to switch away from a prepared for task and then to an alternative task (that is, undo preparation for the self-generated task and activate the representation for the cued task; Allport & Wylie, 2000; Wylie & Allport, 2000). Perhaps if participants are given even more time to prepare after encountering the cue on the invalid trials, then their performance would come to resemble the prepared state found on valid trials. Alternatively, preparation for a chosen task may be difficult to undo in the face of unexpected task constraint, such that additional increases in preparation time

THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2014, 67 (11)

UNEXPECTED TASK CONSTRAINT

would have limited effect on performance. Experiment 2 sought to differentiate between these two alternatives.

Downloaded by [Michigan State University] at 04:02 28 February 2015

EXPERIMENT 2 Experiment 2 explored the extent that the task preparation that occurs when one chooses to perform a new task can be undone when an alternative, unexpected task is encountered. More specifically, the experiment assessed whether preparation can be undone given sufficient time to prepare for the unexpected, cued task. The amount of preparation time that participants received after the cue varied between 150, 700, and 1100 ms. If more time is required to switch away from a prepared for task and then to an alternative task than to simply switch to a new task, then performance on invalid trials should become more like performance on valid trials as the CSI increases.

Method Participants Experiment 2 included 14 participants, but data from one participant who failed to complete the entire experiment were excluded from analysis. Participants’ vision, reimbursement, and IRB compliance were comparable to those of Experiment 1, and none had participated in Experiment 1. Stimuli, tasks, and procedure The stimuli, tasks, and procedure were the same as those in Experiment 1 except that the CSI randomly varied between 150, 700, and 1100 ms, and the experiment included 976 experimental trials. As in Experiment 1, the number of trials in each cell of the design varied. The mean number of invalid repetition at the 150-, 700-, and 1100ms CSIs was 9, 9, and 10, respectively. All other cells of the design had a mean trial number of 25 or greater. 2

Results Tasks were sorted, and trials were excluded using the same criteria as those in Experiment 1. Task performance The effect of preparation time on RT performance was assessed via a 2(validity: valid, invalid) × 2 (transition: repetition, switch) × 3(CSI: 150 ms, 700 ms, 1100 ms) repeated measures ANOVA.2 As the three-way interaction was significant, F (2, 24) = 12.01, p , .001, η2p = .5, separate follow-up analyses were conducted for valid and invalid trials. Valid trials A 2(transition: repetition, switch) × 3(CSI: 150 ms, 700 ms, 1100 ms) repeated measures ANOVA was used to assess RT on valid trials. The pattern of performance is presented in Figure 3A. As expected, repetitions were performed more quickly than switches, resulting in a significant main effect of transition, F(1, 12) = 12.11, p = .005, η2p = .50. Increases in CSI were associated with decreases in RT, as indicated by a main effect of CSI, F(2, 24) = 49.59, p , .001, η2p = .80. Post hoc tests indicated that responses were faster at the 700-ms CSI than at the 150ms CSI, p , .001, and faster at the 1100-ms CSI than at the 700-ms CSI, p = .04. The Transition × CSI Interaction was also significant, F(2, 24) = 21.21, p , .001, η2p =.64, such that switch costs became smaller as CSI increased. Invalid trials The same 2(transition: repetition, switch) × 3(CSI: 500, 700, and 1100 ms) repeated measures ANOVA was used to assess RT on invalid trials (see Figure 3B). The main effect of CSI was significant, F(2, 24) = 36.55, p , .001, η2p = .7. Post hoc tests indicated that RTs decreased from the 150-ms CSI to the 700-ms CSI, p , .001, but did not decrease any further at the 1100-ms CSI, p = .18. In other words, trials in the invalid condition reached an asymptote following 700 ms and did

Removing stimulus-repetition trials did not change the significance levels of any of the reported effects. THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2014, 67 (11)

2253

Downloaded by [Michigan State University] at 04:02 28 February 2015

WEAVER ET AL.

Figure 3. Task response time (RT) for Experiment 2 as a function of validity, transition, and cue-to-stimulus interval (CSI).

not benefit from further preparation time. Neither the main effect of transition nor the interaction reached significance. Accuracy was assessed using the same analysis. A main effect of validity was found such that accuracy was higher on valid trials (M = .918) than on invalid trials (M = .859), F(1, 12) = 14.13, p = .003, η2p = .54. The main effect of transition was also significant, F(2, 24) = 2.04, p = .02, η2p = .39. Accuracy was higher on repetitions (M = .90) than on switches (M = .88). No other effects reached significance.

Task choice The proportion of switches chosen was analysed separately for trials on which the previous trial had been valid and trials on which the previous trial had been invalid. As in Experiment 1, when the previous trial was valid, a significant repetition bias was found (M = .26), t(12) = 8.04, p , .001. However, when the previous trial was invalid a bias toward switching was found (M = .70), t(12) = 4.57, p = .001. RTs for task choice as function of transition and previous trial validity are displayed in Table 1. A 2 (previous trial validity: valid, invalid) × 2 (transition: repetition, switch) repeated measure ANOVA on choice RT found a main effect of transition, F(1, 12) = 6.10, p , .03, η2p = .18, such that participants were faster to select a task repetition than to select a task switch. This switch selection cost was larger when the previous trial had been

2254

valid than when the previous trial had been invalid, as indicated by a significant Previous Trial Validity × Transition interaction, F(1, 12) = 7.87, p = .02, η2p = .40.

Discussion Of critical interest in the current experiment was whether invalid task preparation, following task choice, can be undone given sufficient preparation time. In other words, would invalid trials, in which one chooses and prepares to perform a given task but is required to perform an alternative task instead, come to resemble the pattern of prepared task performance found on the valid trials if participants are given sufficient time to prepare for the unexpected task? To answer this question we compared task RTs found on valid and invalid trials. The pattern of performance seen on valid trials was typical to that seen in previous task switching and voluntary task switching studies (Kiesel et al., 2010; Vandierendonck et al., 2010) and serves as an excellent model of prepared task performance to which RTs on invalid trials can be compared. Responses for repetition trials were fast, even at the shortest CSI and did not decrease with increased CSI, suggesting that repetition trials were at floor, or that participants were performing at their most prepared rate. Responses for switch trials were somewhat slow at the shortest CSI, but became faster as preparation time increased

THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2014, 67 (11)

Downloaded by [Michigan State University] at 04:02 28 February 2015

UNEXPECTED TASK CONSTRAINT

until at the longest CSI they reached the speed of performance found on repetition trials, suggesting that preparation for the upcoming task had been maximized. The pattern of performance found for invalid trials was quite different from that found on valid trials. Both switch and repetition trials in the invalid condition were quite slow at the shortest CSI (compared to switch trials in the valid condition). Responses for invalid trials did become faster at the 700-ms CSI, suggesting that participants were able to use this increased preparation time to prepare for the upcoming task. However, RTs reached an asymptote following 700 ms of preparation, such that the further preparation time provided by the 1100-ms CSI did not result in any additional reductions in response speed. The results suggest that once one chooses and prepares for a voluntarily chosen task, that preparation cannot be easily undone, regardless of the time that one has to prepare for the newly assigned task. Unlike in Experiment 1, in which a reversal of switch costs was found in the invalid condition, in Experiment 2 RTs in the invalid condition did not vary as a function of transition, even at the 700-ms CSI replication of Experiment 1. This result is surprising given that the only difference in methodology between the two experiments was the addition of a CSI manipulation in Experiment 2. Nevertheless, past task switching research has found that the presence of a withinsubject CSI manipulation can lead to variations in performance. Specifically, Altmann (2004) found that switch costs are reduced at longer CSIs when CSI is manipulated within subjects but not when the same manipulation occurs between subjects, suggesting that within-subject CSI manipulations may encourage increased utilization of the CSI interval for task preparation. Within the current study, the introduction of a within-subject CSI manipulation in Experiment 2 may have encouraged participants to better utilize the CSI interval, allowing for the dramatic increase in RT found on invalid repetition trials in Experiment 1 to be somewhat ameliorated in Experiment 2. Regardless, the most critical effects of interest from Experiment 1 were replicated in Experiment

2. Invalid trials were still performed more slowly than valid trials, and, as in Experiment 1, the effect of validity was larger for task repetitions than for task switches. The results demonstrate the difficulty associated with performing an unexpected task, particularly after choosing to switch tasks. It seems that after participants choose and begin preparation to switch tasks, the requirement to perform a repetition trial instead, or to switch away from the newly prepared “switch” task and back to the “repetition” task, results in particularly slow performance.

GENERAL DISCUSSION The current study assessed the effect that needing to perform an unexpected task following self-generated task choice has on performance. On the majority of trials, participants chose and performed tasks according to their choice (valid trials). However, on a small proportion of trials, participants were cued to perform the unchosen task (invalid trials). Experiment 1 demonstrated that invalid trials were associated with large performance costs, particularly when participants were required to perform a repetition after choosing to switch tasks. The results of Experiment 2 suggest that this preparation is not easily undone, despite an extended preparation interval. The results suggest that once one has chosen to perform a task, the ability to perform a different, unexpected task is greatly impaired.

Task preparation Across both experiments, invalid trials were performed significantly more slowly than valid trials. This finding implies that participants were doing some important work towards preparing themselves for the forthcoming task prior to the presentation of the task cue. In the context of the current paradigm, this result suggests that participants were engaging in both task choice and preparation for that chosen task prior to making their task selection. Previous research on the extent that task preparation occurs prior to responding to the prompt (or

THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2014, 67 (11)

2255

Downloaded by [Michigan State University] at 04:02 28 February 2015

WEAVER ET AL.

indicating their task selection) in the double registration variant of the voluntary task switching procedure has been mixed. Arrington and Logan (2005), who first used this procedure, failed to find evidence of preparation prior to the prompt. In their study, choice RTs were very fast and did not vary as a function of transition type. However, in a more recent use of the paradigm, Poljac, Poljac, and Yeung (2012) found longer choice times prior to selecting a task switch than selecting a task repetition, indicating that participants may have used the increased choice time to begin preparation for a task switch. A similar pattern of results was found (following valid trials) in the current study. Choice RTs were longer for switches than for repetitions. However, this pattern of results alone is not necessarily indicative of advanced task preparation on switch trials, as simple motor priming, which occurs as a result of pressing the same choice key as that on the previous trial, could also produce this difference. However, the large performance cost associated with invalid trials in the current study indicates that task preparation was occurring. Further, since the response to the prompt, in which participants indicated their task choice, was immediately followed by the presentation of the cue, this preparation must have occurred prior to point when participants indicated their task choice. Researchers using the double registration procedure should be aware of the possibility that participants may sometimes engage in task preparation prior to indicating their task choice. Participants appear to have engaged in task preparation prior to indicating their task choice. One open question is how extensive was that preparation? Since the amount of time that participants could take to indicate their task choice was unconstrained in the current study, it is possible that participants could delay indicating their choice in order to completely prepare for the task. The CSI manipulation used in Experiment 2 allowed for an assessment of this possibility. If participants engaged in complete preparation prior to indicating the selected task, then additional preparation, provided by longer CSIs, should have no additional effect on switch costs in the valid condition.

2256

Alternatively, reductions in switch costs as a function of CSI would indicate only partial task preparation. The results in the valid condition of Experiment 2 are straightforward. Switch costs reduced as CSI increased. In other words, the increased preparation time available at the longer CSIs allowed for more complete preparation for the task switch and reduced switch costs. Thus, while participants clearly engaged in some preparation for the upcoming trial prior to responding to the task prompt, this preparation was not sufficient to leave them completely prepared for the upcoming task.

Undoing invalid task preparation The results of Experiment 1 suggest that participants were engaging in preparation for the chosen task prior to making their task selection. Experiment 2 assessed the extent that preparation for a chosen task could be undone given sufficient time to prepare for the unexpected task. The results suggest that preparation for a chosen task is not easily undone, despite an extended preparation interval. Indeed, even after an 1100-ms CSI, RTs for invalid trials never reached the RTs of the prepared state found on valid trials. Invalid trials appeared to reach asymptote following 700 ms of preparation; however, this asymptote (approximately 670 ms) was significantly slower than the prepared state achieved in the valid condition (approximately 530 ms). The results suggest that once a participant has chosen and prepared for a task, the preparation associated with that choice is not easily undone, even if one is given over a second to prepare for the unexpected task. Why is preparation for a chosen task so resilient? The strength of this initial preparation could be due to the nature of the processes required in order to perform a voluntary task switch. As many studies have now shown, there is a large performance cost associated with switching tasks (longer response latencies, and higher rates of errors), and a large part of the switch cost is derived from interference from the previously performed task (Meiran, 1996; Wylie & Allport, 2000). Tasks performed during voluntary task switching may be subject to even

THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2014, 67 (11)

Downloaded by [Michigan State University] at 04:02 28 February 2015

UNEXPECTED TASK CONSTRAINT

greater between-task interference than those performed in a cued task switching paradigm (Yeung, 2010). During voluntary task switching, participants cannot rely on task cues to help activate the task to be performed on the current trial, as is the case during cued task switching (Goschke, 2000; Koch & Allport, 2006). Instead, once the decision to switch tasks has been made, increased exertion of top-down control processes needs to be employed to overcome the activation of the previously performed task and make a switch (Arrington & Logan, 2005). Therefore, task preparation efforts may be particularly strong and difficult to overcome in voluntary task switching. Indeed, Liefooghe, Demanet, and Vandierendonck (2009) found that participants are more likely to make use of preparation intervals in voluntary task switching than in cued task switching. Within the current study, initial task preparation appears to have been very difficult to undo, particularly on invalid repetition trials. The difficulty associated with undoing the initial preparation for the chosen task may also be a product of the mechanism used to prepare for a task switch. When one switches away from a task, that task is inhibited. Switching back to the recently abandoned task requires overcoming this inhibition. In their seminal study, Mayr and Keele (2000) provided evidence for the presence of backward inhibition by demonstrating that switching back to a task that has recently been performed (ABA) takes more time than switching to a task that has been performed less recently (CBA). Backward inhibition has been demonstrated repeatedly within the task switching literature (Koch, Gade, Schuch, & Philipp, 2010) and within the voluntary task switching paradigm (Lien & Ruthruff, 2008). Backward inhibition appears to be proactively applied at the point when a participant knows that a task switch will be required on the upcoming trial (Hübner, Dreisbach, Haider, & Kluwe, 2003) and persists across long intervals and even across the performance of several intervening tasks (e.g., Altmann,

3

2007; Mayr & Keele, 2000; Phillipp, Jolicoeur, Falkenstein, & Koch, 2007). Applied to the current study, the backward inhibition account predicts that the effect of invalid trials should be greater for task repetitions than for task switches. On invalid switch trials, participants have chosen and prepared to repeat the previously performed task and therefore should not have imposed backward inhibition, whereas on invalid repetition trials participants have chosen to switch tasks and would therefore have imposed backward inhibition prior to the presentation of the cue. The need to overcome this recently applied inhibition is expected to make invalid repetition trials particularly slow and error prone. The results of the current study are consistent with the pattern of RTs predicted by backward inhibition. Across two experiments, the effect of validity was greater for repetitions than for switches. The results suggest that backward inhibition may play a critical role in voluntary task set preparation. Backward inhibition is able to neatly account for the present results; however, alternative explanations should be considered. Rather than applying inhibition when preparing to switch tasks, participants may instead eliminate previous preparation.3 This idea stems from theories of task switching that assume that actually performing a task prepares one for a task repetition in a way that cannot be matched by other forms of task preparation (Rubinstein, Meyer, & Evans, 2001). When one expects to repeat a task, this preparation is maintained and provides a performance benefit. However, it may be that when participants expect to switch tasks they eliminate or fail to maintain high preparation, such that the performance benefit is lost. Alternatively, the slowing that occurs on invalid trials may be due to a state of confusion caused by the need to abandon a chosen task set and adopt a new one. The current paradigm cannot differentiate between these explanations. What is clear is that regardless of the mechanism, the preparation that allows one to perform a

We thank Eric Ruthruff for this suggestion. THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2014, 67 (11)

2257

WEAVER ET AL.

chosen task is difficult to undo, and as a result, unexpected task constraint following self-generated task choice results in impaired task performance. Original manuscript received 21 October 2013 Accepted revision received 19 February 2014 First published online 11 June 2014

Downloaded by [Michigan State University] at 04:02 28 February 2015

REFERENCES Allport, A., & Wylie, G. (2000). Task-switching, stimulus-response bindings, and negative priming. In S. Monsell & J. S. Driver (Eds.), Control of cognitive processes: Attention and performance XVIII (pp. 35–70). Cambridge, MA: MIT press. Altmann, E. M. (2004). The preparation effect in task switching: Carryover of SOA. Memory & Cognition, 32, 153–163. Altmann, E. M. (2007). Comparing switch costs: Alternating runs and explicit cuing. Journal of Experimental Psychology: Learning, Memory & Cognition, 33, 475–483. Arrington, C. M., & Logan, G. D. (2004). The cost of a voluntary task switch. Psychological Science, 15, 610–615. Arrington, C. M., & Logan, G. D. (2005). Voluntary task switching: Chasing the elusive homunculus. Journal of Experimental Psychology: Learning, Memory, and Cognition, 31, 683–702. Arrington, C. M., Reiman, K. M., & Weaver, S. M. (in press). Voluntary task switching. In J. Grange & G. Houghton (Eds.), Task switching and cognitive control. Oxford University Press. Goschke, T. (2000). Intentional reconfiguration and involuntary persistence in task set switching. In S. Monsell & J. Driver (Eds.), Control of cognitive processes: Attention and performance XVIII (pp. 331– 355). Cambridge, MA: MIT Press. Hübner, M., Dreisbach, G., Haider, H., & Kluwe, R. H. (2003). Backward inhibition as a means of sequential task-set control: Evidence for reduction of task competition. Journal of Experimental Psychology: Learning, Memory & Cognition, 29, 289–297. Hübner, M., Kluwe, R. H., Luna-Rodriguez, A., & Peters, A. (2004). Task preparation and stimulusevoked competition. Acta Psychologica, 115, 211–234. Jersild, A. T. (1927). Mental set and shift. Archives of Psychology, 14, 5–81. Kiesel, A., Steinhauser, M., Wendt, M., Falkenstein, M., Jost, K., Philipp, A. M., & Koch, I. (2010).

2258

Control and interference in in task switching: A review. Psychological Bulletin, 136, 849–874. Koch, I., & Allport, A. (2006). Cue-based preparation and stimulus-based priming of tasks in task switching. Memory & Cognition, 34, 433–444. Koch, I., Gade, M., Schuch, S., & Philipp, A. M. (2010). The role of inhibition in task switching: A review. Psychonomic Bulletin & Review, 17, 1–14. Liefooghe, B., Demanet, J., & Vandierendonck, A. (2009). Is advanced reconfiguration in voluntary task switching affected by the design employed?. The Quarterly Journal of Experimental Psychology, 62, 850–857. Lien, M. C., & Ruthruff, E. (2008). Inhibition of task set: Converging evidence from task choice in voluntary task switching paradigm. Psychonomic Bulletin & Review, 15, 1111–1116. Logan, G. D., & Bundesen, C. (2003). Clever homunculus: Is there an endogenous act of control in the explicit task-cuing procedure? Journal of Experimental Psychology: Human Perception and Performance, 29, 575–599. Mayr, U., & Keele, S. W. (2000). Changing internal constraints on action: The role of backward inhibition. Journal of Experimental Psychology: General, 129, 4–26. Meiran, N. (1996). Reconfiguration of processing mode prior to task performance. Journal of Experimental Psychology: Learning, Memory, & Cognition, 22, 1423–1442. Phillipp, A. M., Jolicoeur, P., Falkenstein, M., & Koch, I. (2007). Response selection and response execution in task switching: Evidence from a go-signal paradigm. Journal of Experimental Psychology: Learning, Memory, & Cognition, 33, 1062–1075. Poljac, E., Poljac, E., & Yeung, N. (2012). Cognitive control of intentions for voluntary actions in individuals with a high level of autistic traits. Journal of Autism Developmental Disorder, 42, 2523–2533. Rogers, R. D., & Monsell, S. (1995). Costs of a predictable switch between simple cognitive tasks. Journal of Experimental Psychology: General, 124, 207–231. Rubinstein, J. S., Meyer, D. E., & Evans, J. E. (2001). Executive control of cognitive processes in task switching. Journal of Experimental Psychology: Human Perception and Performance, 27, 763–797. Ruthruff, E., Remington, R. W., & Johnston, J. C. (2001). Switching between simple cognitive tasks: The interaction of top-down and bottom-up factors. Journal of Experimental Psychology: Human Perception and Performance, 27, 1404–1419.

THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2014, 67 (11)

UNEXPECTED TASK CONSTRAINT

Wylie, G., & Allport, A. (2000). Task switching and the measurement of “switch costs”. Psychological Research, 63, 212–233. Wylie, G. R., Javitt, D. C., & Foxe, J. J. (2006). Jumping the gun: Is effective preparation contingent upon anticipatory activation in task-relevant neural circuitry? Cerebral Cortex, 16, 394–404. Yeung, N. (2010). Bottom-up influences on voluntary task switching: The elusive homunculus escapes. Journal of Experimental Psychology: Learning, Memory and Cognition, 36, 348–362.

Downloaded by [Michigan State University] at 04:02 28 February 2015

Vandamme, K., Szmalec, A., Liefooghe, B., & Vandierendonck, A. (2010). Are voluntary switches corrected repetitions? Psychophysiology, 47, 1176–1181. Vandierendonck, A., Liefooghe, B., & Verbruggen, F. (2010). Task switching: Interplay of reconfiguration and interference control. Psychological Bulletin, 136, 601–626. Wendt, M., Luna-Rodriguez, A., Reisenauer, R., Jacobsen, T., & Dreisbach, G. (2012). Sequential modulation of cue use in task switching paradigm. Frontiers in Psychology, 3, 1–6.

THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2014, 67 (11)

2259

You can't always get what you want: the influence of unexpected task constraint on voluntary task switching.

The current study assessed the effect that unexpected task constraint, following self-generated task choice, has on task switching performance. Partic...
232KB Sizes 0 Downloads 4 Views