You are on page 1of 14

Brain Research 1646 (2016) 160–173

Contents lists available at ScienceDirect

Brain Research
journal homepage: www.elsevier.com/locate/brainres

Research report

Electrophysiological correlates of the cognitive control processes
underpinning mixing and switching costs
Vincenza Tarantino a,n, Ilaria Mazzonetto a,b, Antonino Vallesi a,c
a

Department of Neuroscience, University of Padua, via Giustiniani 5, 35128 Padua, Italy
Department of Information Engineering, University of Padua, Via Gradenigo 6/b, 35131 Padua, Italy
c
Cognitive Neuroscience Center, University of Padua, via Venezia 8, 35131 Padua, Italy
b

art ic l e i nf o

a b s t r a c t

Article history:
Received 15 December 2015
Received in revised form
17 May 2016
Accepted 25 May 2016
Available online 26 May 2016

Typically, in task-switching contexts individuals are slower and less accurate when repeating a task in
mixed blocks compared to single-task blocks (mixing cost) and when switching to a new task compared
to repeating a previous one (switch cost). Previous research has shown that distinct electrophysiological
correlates underlie these two phenomena. However, this evidence is not a consistent result. The goal of
this study was to better characterize differences between the control processes involved in mixing and
switch costs. To this aim, we examined event-related potentials (ERPs) evoked during a cued taskswitching experiment. In order to minimize the confounding effects of cognitive demands unrelated to
task-switching, we asked participants to shift between two simple tasks (a letter identity task and a letter
position task). The mixing cost was defined, in terms of ERPs, by contrasting repeat and single-task trials,
whereas the ERP switch cost was obtained from the comparison of switch and repeat trials. Cue-locked
ERPs showed that the mixing cost was mediated by two sustained components, an early posterior positivity and a late anterior negativity. On the other hand, the switch cost was associated with two early
phasic positive components, one principally distributed over centro-parietal sites and the other located
over left posterior sites. In target-locked ERPs the mixing cost was expressed by a frontal positivity,
whereas the switch cost was expressed by a reduced parietal P3b. Overall, the results extend previous
findings by providing elucidating ERP evidence on distinct proactive and reactive control processes involved in mixing and switch costs.
& 2016 Elsevier B.V. All rights reserved.

Keywords:
Task switching
Cognitive control
ERP
Switch positivity
Task-set
Executive functions

1. Introduction
In everyday life, we frequently and rapidly shift from one task
to another. The cognitive ability underlying this operation is
known as task-switching. A successful example of this ability
within the verbal domain is represented by the capacity of bilinguals to switch language according to the interlocutor. A failure of
this ability is evident in perseverative behaviors exhibited by patients with brain injuries (typically frontal), who keep repeating an
action that has become inappropriate with respect to the context
or the intention (Shallice et al., 2007). In experimental settings,
researchers have investigated the mechanisms subtending taskswitching abilities using paradigms that require an individual to
promptly shift attention toward different features of stimuli and to
respond to them according to new stimulus-response (S-R) rules
(‘attentional-shift’ paradigms; Rushworth et al., 2002, 2005).
These studies, which simulate real life situations, have
n

Corresponding author.
E-mail address: vincenza.tarantino@unipd.it (V. Tarantino).

http://dx.doi.org/10.1016/j.brainres.2016.05.048
0006-8993/& 2016 Elsevier B.V. All rights reserved.

unequivocally shown that such cognitive flexibility implicates
costs in performance. Indeed, participants are slower and less
accurate in performing a task which is different from the previously executed one (i.e., in switch trials) compared to performing
the same task (i.e., in repeat trials). The required extra-time and
the reduction of accuracy reflect the so-called ‘switch cost’ (see
Kiesel et al., 2010 for a review; Meiran, 1996; Rogers and Monsell,
1995). Furthermore, participants are slower and less accurate on
repeat trials, that is, when repeating a task in blocks that also
include switch trials (mixed blocks), than on single-task blocks,
that is, when performing a single task throughout the whole block.
This effect is called the ‘mixing cost’ (Rubin and Meiran, 2005).1
To date, understanding the cognitive processes and neural
mechanisms that underlie the behavioral switch and mixing costs
remains an important issue (Karayanidis and Jamadar, 2014). One
1
Note that the cost referred to as the mixing cost in the comparison between
mixed blocks (switch þrepeat trials) and single-task blocks represents a more
'global' cost, whereas the cost referred as the switch cost in the comparison between switch trials and repeat trials represent a ‘local’ switch costs (Mayr, 2001).

which implies rule retrieval and goal activation processes (Jost et al. 2007. Nicholson et al.. Interestingly. Indeed. Tarantino et al..e. 2005. updating (or deleting) them in working memory. A less consistently observed ERP component linked to switch trials is a positivity emerging over frontal sites. multiple topographically and temporally distinct event-related potentials (ERPs) have been reported. This refers to an endogenous process that involves multiple sub-processes. the modulation of this component by trial type (i. 2012. Rushworth et al. and onset of the switch-positivity have been found to be modulated by the CTI duration and the amount of information about the upcoming task provided by the cue (Karayanidis et al. Finke et al.. Rogers and Monsell. duration. The task-set reconfiguration process plays a key role in task-switching operations by acting in anticipation of a stimulus.. 2005. In order to disentangle the contribution of specific processes related to the switch and mixing costs previous research has combined cued task-switching paradigms with the simultaneous recording of electroencephalography (EEG)... 2008b. 2015. we briefly overview the main cue-locked and target-locked ERP components associated with the proactive and reactive control processes. / Brain Research 1646 (2016) 160–173 of the most robust process that supports the switch cost is task-set reconfiguration (Meiran et al. 2009). Therefore. Some studies report larger negativities on switch trials (Astle et al. inhibitory processes have to be mentioned. A further cue-related component. the amplitude. Kray et al. 2009). shows a left-lateralized distribution (Elchlepp et al. 1995). Karayanidis et al. while others report larger frontal negativities on repeat trials (Goffaux et al.. Periáñez and Barceló. Rubin and Meiran. These findings support the view that this negativity may reflect inhibition of the competing S-R mapping. The topographic and temporal dynamics of the mixing-positivity as well as its functional role in relation to the switch-positivity have not yet been clearly characterized. Karayanidis et al. Rogers and Monsell. Rubinstein et al. it may represent the decoding of the cue (Karayanidis et al. 2002). this component was not been observed when covert responses were required (Astle et al. Interestingly.. 2005). 2001).. 2009. 2000). its precise functional role is still debated. Altmann..e. 2014). Nicholson et al. According to some studies... Indeed. . Interestingly.. the same applies for the mixing cost. Therefore.. 2003).. Jost et al. 1994) or ‘backward inhibition’ (Mayr and Keele. 2005. respectively. 2007. it has been theorized to reflect cue change detection. Wylie et al.. Lavric et al. 2008b. 2006. Mueller et al. The fact that a larger switch-positivity has been found to be associated with smaller behavioral switch costs suggested that it likely represents an index of proactive task-set reconfiguration processes (Capizzi et al. 2011. 2009. 2000). Moreover. 2007.... cue-to-target interval. Hsieh and Cheng. Karayanidis et al. sometimes labeled ‘switch-positivity’. 2012). 2008b) and it has been found to be reduced or absent in univalent conditions compared to bivalent conditions and when the tasks are mapped onto separate response keys compared to when they are mapped onto the same response keys (Astle et al. 2008. This evidence suggests that inhibitory processes intervene at both the preparation and implementation phases. retrieving task goals and rules. At least in part.e. Mueller et al. 2011a. 2014.. A posterior cue-related potential has been documented not only in switch trials as compared to repeat trials. 2006. which has been found to diminish with increasing preparation intervals (Kray. sometimes labeled ‘pre-target negativity’ (Astle et al.e. 2008a. 2008. Rushworth et al. Hsieh and Cheng. The excellent temporal resolution of EEG enables us to distinguish between cuerelated task preparation processes (i.. Poulsen et al. 1995... which induces endogenous preparatory processes.. has been consistently observed after cues signaling a switch as compared to cues indicating a task repetition (for reviews see De Baene and Brass.e. 2014). reactive control) (Braver. both switch and mixing costs have been shown to reflect the greater memory demands needed for retrieving the relevant task rules from long term memory and updating and maintaining more than one taskset active (Mayr and Keele.. such as shifting attention between stimulus attributes or features. Notably. Eppinger et al. This component has been predominantly associated with the contingent negative variation (CNV) and given a general anticipatory attention and/or task preparation function (Brunia and van Boxtel. occurring in the late portion of the CTI in both switch and repeat trials is a slow fronto-central negativity.. 2006. 2007.. This component has been referred to as the ‘mixing-positivity’ (Karayanidis and Jamadar. Karayanidis et al. i. 2002). even though long preparation intervals minimize the proactive interference of a previous task-set.. 2008). a phenomenon called ‘task-set inertia’ (Allport et al. With long CTI durations and with high levels of cue informativeness a ‘residual cost’ is still observed (De Jong. this component 161 reaches maximum amplitude over centro-parietal scalp sites and. 2008.V. Below. 2011a. However. Typically.. 2008. 2006a). 2000. 2009). 2008a... respectively).. Karayanidis et al. Capizzi et al. 2014. 2005. Manzi et al. the P2 amplitude was found to be more pronounced for informative switch cues compared to uninformative switch and repeat cues (Finke et al.. Importantly. Nicholson et al. Cue-locked ERPs A positive posterior ERP component. 1996. 2009). 2010). and enabling a different response set (Monsell. Karayanidis and Jamadar.. Mueller et al. Furthermore.. rapid context updating processes (De Baene and Brass. 2008).. 2005). Meiran.. 2016. at around 150– 200 ms after cue onset (P2.. Koch... the target stimulus triggers the activation of the competing S-R association.. These findings reveal that (i) the task-set reconfiguration process is likely not completed in the preparation phase but is still ongoing during the implementation phase and (ii) additional preparation processes are entailed during the CTI. switch or repeat) is not consistent across studies. Li et al. Thus far. 2011b. 2006). 2008a. 1996. 2012) and a greater amount of information carried by the cue (Czernochowski. Astle et al. 2006b) produce a decrement in the switch cost. this component could also reflect the ‘reloading’ of the new S-R mapping (Periáñez and Barceló. 2014. suggesting that this component is not a mere index of cue sensory changes but an index of goaldirected context updating. Lavric et al. in some cases. 2004.. These findings suggest that switch costs and mixing costs might in fact rely on common or partially-common preparatory mechanisms. 2012). 2001). but also in repeat trials as compared to single-task trials (Czernochowski.. which are thought to reflect different cognitive processes... 2000. 2006. 2011a. 2003.1. which aim at resolving the conflict and overcoming the interference produced by the previous task-set. 2012). 1. Poulsen et al. 2006. proactive control) and target-related task execution/implementation processes (i. Among these. Since this component has an early latency and reaches higher amplitude in switch compared to repeat trials. CTI. the benefits afforded by the duration of the preparation interval and by the information provided by the cue do not completely eliminate the switch cost. Nicholson et al. 2011. inhibitory processes seem to contribute to the mixing cost as well. the persistence of the competing S-R association causes conflict even in repeat trials (Los. it has been shown that in cued paradigms (where a cue signals the task that must be executed on the upcoming stimulus) longer preparation intervals (i.

regardless of task switching requirements. 2011b. indeed. 2006.1. ERP mixing vs. in NoGo trials). 2011.. 1. Furthermore.. the use of a Go/NoGo task and of a short CTI duration (the longest CTI was 780 ms) did not allow the isolation of the switch-related proactive control components. Based on previous literature. 2010b. Barceló et al. 2. 2006.e. The 3  2 ANOVA on RTs showed a significant main effect of Trial type [F(1. 2003).e. 2005. However. timing and topography..21) ¼4. A cued task-switching experiment was designed with a long CTI (i. located over central and parietal scalp regions (Astle et al. (2006) found that the ERP mixing cost was represented by a larger cue-locked pre-target negativity. Response inhibition. Goffaux et al. 1200 ms). Wylie et al. we thought to extend previous studies by controlling some experimental features.69. Tarantino et al. they rely on similar neural mechanisms. Therefore.001. To our knowledge. 2000.. Gajewski et al. They showed distinct topographic ERP configurations for repeat and single-task trials. In addition. A limited number of target stimuli (two letters in one of two positions) promoted advanced preparatory processes.. the mixing.2. 2001) and/or the greater amount of attentional resources devoted to overcome the conflict induced by the competing S-R mapping (Hsieh and Liu. task-set reconfiguration processes cannot be fully completed in advance and stimulus categorization is needed in order to select and activate the appropriate response (Jamadar et al. More importantly. it is not possible to exclude that repeat and switch trials would differ in terms of reactive control processes. 2010b). although the mixing cost and the switch cost are separate phenomena. which allows. Gajewski and Falkenstein.4)¼ 52. respectively.001). however their findings still leave some open issues. no specific cuelocked ERP indices of switch cost emerged. the reduction of this component might represent a marker of both mixing and switch cost. Behavioral Results Mean RTs and accuracy for each Trial type (single. it is important to better characterize their electrophysiological correlates. Jost et al. A plausible explanation of these results could be that the use of semantic tasks increased the cognitive load and minimized the difference between task-switching conditions. Post-hoc tests revealed that participants were slower in responding to repeat compared to single-task trials and in switch compared to repeat trials (all pso.3. we expected to find an earlier cuelocked posterior positivity and a later sustained negativity as electrophysiological markers of the mixing cost. letter identity) are displayed in Table 1. switch) and Task type (letter position. ƞp2 ¼ . 2006. The present study The goal of the present study was to elucidate the electrophysiological correlates of the mixing and switch costs. These interpretations effectively explain the more pronounced reduction observed under bivalent compared to univalent task-switching conditions (i. 2008. Moreover. Interestingly. Hsieh and Liu.162 V. no specific electrophysiological patterns associated with the switch cost were found. the neural correlates of proactive and reactive control processes. To this aim. Karayanidis et al. Possibly. 2011a. On the other hand. In this study. 2008). 2009). Accordingly. and in repeat trials as compared to single-task trials. it has been inferred that this phenomenon in task switching contexts is associated with the reactivation of task rules (Rubinstein et al. Mixing and switch cost components would differ in amplitude. and a reduced target-locked P3b over centro-parietal sites. Furthermore. Goffaux et al. supporting the idea that they rely on separate mechanisms... Given that the P3b typically reduces in amplitude as a function of working memory and attention demands (Kok. Overall. switch trials differed from repeat trials only in terms of signal strength. p o.and switch-positivity components were not observed. the ERP findings described thus far suggest that. 2002. better isolation of the preparatory ERP components.4. Indeed.44] and a significant Trial type  Task type interaction [F(1.70]...17] emerged. they showed distinct electrophysiological correlates. from single-task to repeat to switch trials. This approach allowed us to apply multiple t-tests simultaneously to all electrodes and data points and to control for the family-wise error rate via permutation tests. 2011b). Kieffaber and Hetrick. in order to detect ERP markers that distinguish between the two types of costs. Rushworth et al. since the authors did not examine the target-evoked potentials. participants were required to carry out two simple tasks (i.. their putative brain sources were also estimated.. therefore it does not depend on the execution of the response per se but on processes related to response selection..014.62.. 2014). distinct S-R mapping.. Results 2.e. repeat. 2008. switch costs Taken together. at both the cue level (larger posterior positivity and anterior pre-target negativity) and the target level (an attenuated P3b). In order to better detect all possible differences between the two types of cost. broadly distributed over the scalp. overlapping vs.. such as task rule activation (Gajewski and Falkenstein. / Brain Research 1646 (2016) 160–173 1.. the attenuated target-related P3b in switch relative to repeat trials was also obtained when no overt response was required (i. A non-verbal cue was adopted in order to lower possible additional semantic processes. ƞp2 ¼.27. the ERP data were analyzed by means of a mass univariate approach (Groppe et al.2. By reducing the contribution of such factors. The simplicity of the tasks allowed us to limit the confounding effects linked to cognitive demands unrelated to task-switching requirements. This component would be accompanied by more frontal modulations in switch and repeat trials. (2009) focused on cue-related ERP components. principally indexed by an attenuation of the target-locked P3b amplitude. a letter identity task and a letter position task). p ¼ . it reflects the implementation of reactive control processes. furthermore. 1998).. we expected to find target-related markers of residual costs. we were able to better characterize the two types of costs in terms of ERP correlates.86.e.. Wylie et al. The ERPs evoked by the cue and by the target stimulus were examined separately in order to investigate both the preparation and implementation phases. expressed as global field power (see Dove et al..007. 2011). p¼ .22)¼8. that is. on the other hand. Post-hoc tests revealed that in single-task and repeat . In order to better clarify the anatomo-functional characteristics of the ERP components. a significant main effect of Task type [F(1. 2010. Jamadar et al. whereas the ERP switch cost was only represented by a decrement of the P3b. Target-locked ERPs A finding widely reported in cued task-switching studies is a reduced positive peak (P3b) elicited by the target stimulus in switch trials as compared to repeat trials. ƞp2 ¼. Whitson et al. affected a more frontal P3 (‘NoGo P3’). 1. 2000 for similar neuroimaging evidence).. where inhibitory processes are required. only two studies have purposely addressed the direct comparison of ERP components related to mixing and switch costs (Goffaux et al. Karayanidis et al. we expect to find an earlier frontal positivity and a later posterior positivity as markers of the switch cost. unlike the aforementioned studies.

001].032].25. Post-hoc Wilcoxon signed-rank tests revealed that the level of accuracy on repeat trials was lower compared to single-task trials in the letter position task (Z¼  2. p ¼. p ¼ .92. this difference did not reach statistical significance (p ¼. When the mixing and switch costs were computed at the individual level. Task type effects did not emerge on any type of trial (Zs o  1.080).028). p o.4 (1.20.5 (5.3 (2. The accuracy on switch trials was lower compared to repeat trials in both the letter position 2. Accordingly. .8) 487 (101) Switch ACC (%) RT (ms) ACC (%) 96.690). po . the presence of a Trial type Task type interaction was assessed.0) 2. Single Letter identity Letter position Repeat RT (ms) ACC (%) RT (ms) 454 (72) 98.001) and the letter identity (Z¼  2. namely the switch cost was slightly larger for the letter position task (97 771 ms) compared to the letter identity task (70764 ms) [t(22) ¼ 2. p ¼. In the letter identity task. trials for the two tasks were collapsed and the main effect of Trial type was analyzed. In order to investigate this result. The zero time point corresponds to the cue onset.2. The presence of a small asymmetric switch cost emerged. whereas in switch trials they showed similar RTs (p ¼. individual differences in amplitude between repeat and single-task trials (ERP mixing cost) and individual differences in amplitude between switch and repeat trials (ERP switch cost) were computed for the letter identity trials and for the letter position trials.7) 521 (91) 410 (64) 98. ps4. Table 1 Response times (RTs) and accuracy rates (ACC) on the single-task.29.75.1.024) task.2.2 (6. Individual averages were computed after collapsing the two types of tasks (letter identity and letter position). / Brain Research 1646 (2016) 160–173 (Z¼  3. a Task type effect was found in the switch cost (Z¼ 2.8) 97. 1. ERP results trials participants were slower in performing the letter identity compared to the letter position task (ps¼ . Accordingly.131). Cue-locked ERPs The effect of Trial type was initially examined separately for each task. In this way.4) 585 (154) 163 91.V. Standard deviation values are reported in parentheses. This finding disclosed the presence of an asymmetric switch cost also in terms of accuracy. The resulting differential ERPs were then compared between the two tasks by means of a cluster mass permutation procedure. we decided to examine the ERPs separately for each task type in order to detect potential ERP correlates of this asymmetry. Cue-locked ERPs.023) but not in the mixing cost (Z¼.1 (1. the individual switch cost was computed and contrasted between the two tasks. even at a more liberal test-wise alpha level (po . To this aim. repeat. and switch trials by task type (letter identity and letter position). This null result demonstrated that cue-locked ERPs did not differ between the two tasks either in terms of mixing cost or in terms of switch cost.001). The Friedman's ANOVA performed on mean accuracy in all six experimental conditions yielded a significant effect of Trial type [χ2(5) ¼42. Tarantino et al. Grand-average waveforms of cue-locked ERPs for each trial type (single.05). p ¼.3) 591 (131) 93. None of the t-test values reached the significance threshold in any electrode/time point.691. p ¼.490). Cue-locked ERPs µV 5 Fz F3 F4 0 -5 Single -10 Single 5 Repeat Repeat Switch C3 Cz C4 0 -5 -10 Pz P3 5 P4 0 -5 -10 -200 200 600 1000 1400 -200 200 600 1000 1400 -200 200 600 1000 1400 ms Fig. switch) in representative electrodes are depicted. repeat.5 (3.28.91.

The sustained negativity over frontal sites did not appear to represent a CNV. 2. In this time window. CP2. P5. Namely. from 310 ms to 520 ms. F4. / Brain Research 1646 (2016) 160–173 Fig. positive t-values). which reflected more negative ERPs in repeat trials compared to single-task trials over bilateral frontal and frontocentral electrode sites. CP3. The colors represent significant t-test values. CPz. is compatible with a contingent negative variation (CNV). Pz. but not frontal sites. F8. POz. Later. negative t-values). P7. P7. PO7. Electrodes are reported along the y-axis.e. in significant time windows. Hot colors stand for more positive t values. the ERP amplitude in repeat trials was significantly more positive relative to single-task trials for a longer duration (i. more positive potentials for switch compared to repeat trials were observed over a few midline and right-hemisphere sites as well (Pz. from 360 ms to 720 ms).. F5. and P6. the posterior positive cluster overlapped in time with the anterior negative cluster. . and F2. 2. and O1). CP5. nevertheless in this case t-values reached significance for shorter or more discontinuous time windows. it extended to CP1. F1. PO3. The ERP switch cost was mainly represented by two clusters. are illustrated. Tarantino et al. FC3. P3. FCz. the two types of tasks (letter identity and letter position) were collapsed. left panel). each colored electrode/time point represents a significant t-test (p o . FC6)... whereas O1.. Cue-locked ERP mixing and switch costs. over frontal sites its voltage was constant until the onset of the target stimulus both in repeat and switch trials. AF4. 1). 1 (here the two Task types are collapsed). which reflected more positive ERP amplitude on switch compared to repeat trials (see Fig. P2 and P4 electrodes were involved from 360 ms to 600 ms and from 670 ms to 700 ms. but was likely a component specific to mixed blocks. the topographical maps of differences between conditions. This cluster reflected the sustained differential negativity observed in the ERP waveforms over frontal electrodes. P1. P5. The first positive cluster developed from 270 to 400 ms after cue onset and was extensively distributed over the scalp. O2 electrodes were involved from 580 ms to 720 ms. an earlier positive cluster (i. Additionally.164 V. Oz. The test disclosed two main significant spatio-temporal clusters related to mixing cost. FT8. ERPs were significantly more negative in repeat trials compared to single-task trials. P4. In computing the mixing and switch cost. POz. In the diagram. The grand-average waveforms of the cue-locked ERPs on each Trial type in representative electrodes are depicted in Fig. The positive cluster involved at first only P1 and Pz electrodes. P2. FC5. F7.e. The results of the cluster mass permutation test are visualized in the raster diagram displayed in Fig. From 580 ms to 720 ms. and a later negative cluster (i. from 360 ms to 470 ms. In this case. The second cluster was found from 520 to 660 ms. which reflected more positive ERPs in repeat trials compared to single-task trials over bilateral parietal and occipital electrode sites. midline electrodes in the middle and electrodes positioned on the right hemisphere in the bottom. Over PO3. In the bottom panels. The topography and time course of this component over central. Oz.001). Fz. F3. CP4. The negative cluster emerged from 580 to 920 ms after cue onset and involved the following frontal electrodes: AF3. and PO4 sites.e. Fpz. which typically develops between a warning stimulus and an upcoming imperative stimulus and rises gradually until the onset of the imperative stimulus (Walter et al. In the top panels. whereas the time course is reported along the x-axis. P3. whereas over central sites it showed an increasing trend on all trial types. F6. FC1. 2. 1964). FT7. the raster diagrams representing the results of the cluster-based mass permutation tests performed on individual ERP differences between repeat and single task trials (left) and between switch and repeat trials (right) are depicted. It is worth noting that this long lasting negative potential showed a different morphology in frontal sites and central sites (see Fig. PO4. The electrodes positioned on the left hemisphere appear in the top. principally over left parietal and occipital sites (P1. This negativity was observed over other frontal electrodes as well (Fp1.

The zero time point corresponds to the target onset. The second peak. FC3. the latter cluster reflected the second. although in these electrodes not all time-points reached statistical significance. CP3. / Brain Research 1646 (2016) 160–173 165 that Trial type did not significantly interact with Task type. P2. F6. FC2. Cz. F7. which reached its highest amplitude over central and frontal sites for repeat trials and over frontal sites for switch trials. The and O2). P3. Pz. FC3. developing from 210 to 240 ms. P3. The first cluster (from 160 to 210 ms) was distributed over the following electrodes: F3. C3. None of the t-test values reached the significance threshold in any electrode/time point. namely Fp1. and C2. and P2). A small cluster. FC4. . and P5) showed significant differences from 710 to 800 ms..2. C1. more anterior. The comparison of the target-related ERPs between switch and repeat trials revealed significantly more negative amplitude in switch trials over centro-parietal electrodes. P3. FC2. AF3. C2. The first two clusters reflected the same component. FC5. involved additional frontal electrodes (Fp1. C3. The first cluster. meaning that the type of task did not significantly modulate cue-locked ERPs. we will not consider them further. and Fpz). P1. F8.V. switch – repeat) were then compared across the two tasks by means of a cluster mass permutation procedure. The mixing cost and switch cost were analyzed for each task separately. Within this time window three positive clusters could be distinguished. from 240 to 350 ms. F1. Pz. which comprised an early positive component. more sustained component. more phasic in time (i. AF8.05). namely the first peak of the bimodal positive fronto-central ERP (P3a). CPz. peak. CP2. C5. AF3. Target-locked ERPs. 1). FC4. C4. Individual averages were computed after collapsing the two types of tasks (letter identity and letter position). C5. this difference was sustained over almost the whole time window (from 220 to 400 ms). C6. which lasted from 220 to 290 ms.e. PO3. PO7). F5. over left posterior sites (P7. T7. 3. CP6. Cz. Fz. Significant more positive ERPs on repeat trials compared to single-task trials emerged from about 160 ms. The resulting ERP differences (repeat – single. Fpz. POz.2. F1. FC1.05). FCz. In some of these electrodes (P1. FT7. F2. lasting for a narrower time-window) and extensively distributed over the scalp. Specifically. Fz. The presence of a main effect of Task type (letter position vs. CPz. None of the t-tests reached the significance threshold. F2. FC1. switch) in representative electrodes are depicted. which was more pronounced over the left scalp regions. AF4. and a later. F4. C1. Target-locked ERPs Grand-average target-locked ERP waveforms in representative electrodes are displayed in Fig. The second cluster. C4. FCz. 3. two clusters were found between 220 and 400 ms (Fig. A further small positive cluster emerged from 740 to 810 ms. CPz. represents a P3b. P3. The third cluster was localized to more anterior sites. letter identity) was analyzed after collapsing data across trial types. Cz. F7. CP4. which reached maximum amplitude over parietal sites at about 300 ms. involved the following electrodes: P1. and P4. CP5. even at a more liberal test-wise alpha level (p o. Tarantino et al. Taken together these positive clusters reflected the bimodal morphology of the positive ERPs evoked in switch trials (see Fig. This means Target-locked ERPs µV 15 F3 Fz F4 Cz C4 Pz P4 12 9 6 3 0 -3 Single C3 15 12 Repeat Switch 9 6 3 0 -3 P3 15 12 9 6 3 0 -3 -50 150 350 550 750 -50 150 350 550 750 -50 150 350 550 750 ms Fig. FC2. 4). repeat. 2. Grand-average waveforms of target-locked ERPs for each trial type (single. Pz. trials were collapsed across the tasks in the subsequent analyses. Therefore. principally including parietal electrodes (CP3. F3. CP1. A positive bimodal deflection was evidenced. Fp2. even at a more liberal test-wise alpha level (p o. The first peak of this deflection might be associated to a P3a. Since these differences involve post-response processes (based on average RT). and P2.

in the 260–400 ms time window. the precuneus. the left supramarginal gyrus. are illustrated. CP5. PO3. in the 520–660 ms time window. and CP2) and lasted from 310 to 400 ms. The relative difference of this average between repeat and single-task trials.166 V. POz. PO3. P7. In the 720–920 ms time window. The results are depicted in Fig. the right angular gyrus. FC2. and the precentral gyrus. the topographical maps of differences between conditions. these two clusters reflect the reduction of the P3b over parietal sites in switch compared to repeat trials. each colored electrode/time point represents a significant t-test (p o . 7 and 8 of the Supplementary materials. Taken together.001). recruited an extensive group of fronto-parietal cortical areas. i. P1. P5. O1) and POz. the precuneus. midline electrodes in the middle and electrodes positioned on the right hemisphere in the bottom. PO7. CP1. in significant time windows. CP5. P1. The first one started at 170 ms and lasted up to 210 ms over left posterior scalp sites (CP3. and the following regions bilaterally: the superior frontal gyrus. In the bottom panels. P3. the right intraparietal sulcus. the activity was more spatially limited and included the anterior cingulate gyrus bilaterally. the anterior and middle cingulate gyrus. CP4. 2. P3. P5. and the precentral and postcentral gyri.2. whereas the time course is reported along the x-axis. CP3. repeat. In the 580–720 ms time window. the intraparietal sulcus. the central and postcentral gyri bilaterally. C5. the raster diagrams representing the results of the cluster-based mass permutation tests performed on individual ERP differences between repeat and single task trials (left) and between switch and repeat trials (right) are depicted. 4. Brain source analysis An estimation of the cortical generators of the cue-locked ERPs was performed in the time windows in which the previous cluster mass permutation test showed significant differences in amplitude between trial types. the superior occipital gyrus. and P2). was statistically tested at the whole-brain level by means of paired t-tests with a cluster- based non-parametric correction for multiple comparisons (see the Experimental procedure section for details). the superior parietal gyrus. and switch trials. 5. Tarantino et al. the left occipital lobe. The average of individual normalized sources (z-maps) within each of these time windows was computed. CPz. In the 360–580 ms time window. C1. at the beginning over left sites (C3. and the left superior temporal gyrus and sulcus. and the following areas bilaterally: the superior parietal gyrus. The cluster-based permutation analysis comparing the normalized sources between switch and repeat trials showed no statistically reliable neural generators for the first positive cluster. the right middle frontal gyrus. The second cluster emerged from 220 to 290 ms. The main effect of Task type on target-locked ERPs was investigated by collapsing single.e. manly reflected the brain activity localized in left parieto-occipital regions. In the diagram. the left postcentral gyrus. C4. / Brain Research 1646 (2016) 160–173 Fig. a greater activity principally located in left fronto-parietal cortical regions emerged in repeat as compared to single-task trials. compared to single-task trials. the two types of tasks (letter identity and letter position) were collapsed. C2. CP1. The relative grand-average waveforms and raster diagrams obtained from the mass univariate analysis are presented in Figs. Cz. Hot colors stand for more positive t values. The second cluster. The colors represent significant t-test values.3. Target-locked ERP mixing and switch costs. cue processing in repeat trials. Electrodes are reported along the y-axis. The mass permutation test revealed a more pronounced negativity in the letter identity task compared to the letter position task over left central and parietal sites in two clusters. second cluster involved these electrodes plus some central electrodes (C1. after 240 ms over midline and right sites (Cz. CP6.. In computing the mixing and switch cost. the angular gyrus. P7). In the top panels. FC4. CP2. in the posterior . These areas comprised the lateral portion of the left superior frontal gyrus. and between switch and repeat trials. The electrodes positioned on the left hemisphere appear in the top. CP3. namely. namely.

The right frontal sulcus was more active in single-task trials from 240 to 350 ms as well. The switch cost emerged in terms of RTs and accuracy. Wylie and Allport.e. Kray et al. This means that the ERP switch cost was characterized by the same direction and the same effect size in both tasks. 1999). 2005. 2007. Wylie et al.V. In order to disentangle preparation/proactive control processes from implementation/reactive control processes. The behavioral findings confirmed the presence of mixing and switch costs. the mixing cost emerged in terms of RTs in both tasks.05). It was represented by significant slowing and lower accuracy when responding to switch trials compared to repeat trials. Behavioral and electrophysiological data were analyzed separately for each task. This finding is in line with previous literature. single comparison. 3. ERP mixing cost The mixing cue-locked ERPs were characterized by a sustained positivity. the angular gyrus. 1994.. the temporal superior gyrus. We cannot exclude that differences would involve response-related electrophysiological indices. Wylie et al. the target related switch negativity was associated with a greater activity in the anterior portions of the left superior and middle temporal gyri for repeat compared to switch trials. which has been observed in paradigms with a fixed long CTI. 5. the precuneus. see Supplementary materials). 5). 2011. 2010b). whereas the ERP correlate of the switch cost was represented by two early positive phasic components. 167 the results revealed different ERP patterns related to mixing and switch costs. which were not considered here. Specifically. Manzi et al. Based on the spatio-temporal features of such components. involving two simple tasks.. Mixing cost measures were obtained by contrasting repeat and single-task trials. both at the cue and at the target level. the superior occipital gyrus was additionally involved bilaterally.1. Discussion The present study aimed to clarify the differences between the electrophysiological correlates of mixing and switch costs. 3. 2014. A main effect of task was found on RTs which revealed that the letter identity task was overall more demanding.. Cue-locked ERPs The results of the cue-locked ERP analysis revealed the presence of distinct mixing. Brain sources. 2010a). Jamadar et al. Though a main effect of task was found in the target-locked ERPs (i. Namely. the ERPs evoked in a cued task-switching experiment. Overall. in repeat as well as in switch trials. after symbolic cues that signal the timing of the target and type of task to be implemented on the target relative to symbolic cues that signal the timing of the target but not the identity of the task (Karayanidis and Jamadar. Both components likely denote a sustained brain activity devoted to preparation processes. which emerged from 580 to 920 ms after cue onset and was mainly localized to bilateral centro-frontal sites. . which reflect the contribution of distinct brain processes... the switch cost was slightly larger for the letter position task compared to the letter identity task. and by a sustained negativity. / Brain Research 1646 (2016) 160–173 Fig. In the 220–290 ms and in the 310–400 ms time window.. top and right views. No statistically significant clusters of activity were detected when collapsing trial types in order to investigate differences between the two tasks. Karayanidis et al. the ERP mixing cost was represented by an early sustained posterior positivity and a late sustained anterior negativity. showing a greater activity in the right superior and middle frontal sulci in single-task compared to repeat trials. This subcomponent might be ascribed to the ‘informative-positivity’.. no significant interaction between trial type and task type emerged. 2011a. as resulted from the cluster-based permutation test on brain sources (po . The presence of differences between the estimated neural generators of the target-locked ERPs was examined in the following time windows: 160–240 ms and 250–350 ms for the repeat vs. ERPs locked to the presentation of the cue and to the presentation of the target were analyzed separately. and the middle occipital gyrus. 2000) or when shifting from the second to the first language (Meuter and Allport. It was represented by a significant slowing in responses to repeat trials compared to single-task trials. and in terms of accuracy in the letter position task. which showed that switch costs are larger when shifting from a more demanding task to an easier task (Allport et al. 2008a. The mixing-related negativity plausibly corresponds to the ‘pre-target negativity’ component (Astle et al. we may infer that the mixing-related sustained positivity corresponded to the ‘mixing-positivity’ reported in previous literature (Czernochowski. were examined. in both tasks.. 2008. 3. Asymmetric switch costs were observed in accuracy and RTs. Significant z-maps differences are shown in left. 2009).. This asymmetry did not emerge at the electrophysiological level. This interpretation is further corroborated by the fact that our cues in single-task blocks conveyed information only on the timing of the target (cf. the intraparietal sulcus. 2011b. repeat comparison (Fig.1. which emerged from 360 to 720 ms after the cue onset and was manly localized to bilateral centro-parietal sites. Furthermore. 220–290 ms and 310–400 ms for the switch vs. switch cost measures were obtained by contrasting switch and repeat trials. A minimum cluster of 30 contiguous vertices was considered in all z-maps portion of the superior parietal gyrus. Tarantino et al. Jost et al. Within this mixing-positivity an early centro-parietal positivity could be distinguished. Significant differences in brain sources emerged in the 160–240 ms time window. Eppinger et al. The task type did not interact with the switching requirements. at the implementation phase. 2009). Specifically.and switch-related preparatory components. The reconstructed cortical sources of significant ERP effects are reported. as discussed later.. 2006. and that the superior frontal gyrus was part of the putative source of such potential (Jamadar et al. To this aim. 2011.1.. in the latter time window the right precentral and postcentral giry and sulci were involved..

. 2008. Periáñez and Barceló. It could be argued that the cues in single-task blocks were not relevant to performance. 2008. 2010). ERP correlates differed in terms of amplitude. irrespective of its features. therefore it marks a cue change detection (Finke et al. for example.. therefore processes related to switch may actually derive from a cue-change and not a task-change (e. the time course and topography of the pre-target negativity observed in the current study disambiguated it from a CNV (see Section 2.. Unlike the ERP correlates of the mixing cost. 2007.e.. 3. Based on these findings. and with the involvement of the anterior cingulate gyrus.... they observed a more pronounced late sustained (pre-target) negativity in repeat compared to switch trials.. represents a shift of attention toward relevant features of the upcoming target stimulus (Hopfinger et al... 2009).2. We might infer that the greater involvement of the prefrontal cortex in repeat as compared to single-task trials likely reflected goal/task rule retrieval and activation into working memory (Braver et al. Likely. see also Lavric et al.. 2009).. observed from 720 to 920 ms after cue onset.. Goschke. These results extend the study by Wylie et al. Accordingly. 2013). Moreover. the early frontal positivity component). 2008a. The activity of the superior frontal lobe. which influence the presence of the informative-positivity. Karayanidis et al. Vallesi et al. Wager and Smith.2. In support of this. they exhibited clearly distinct time courses and topographies. left inferior parietal cortex. 2006a. Rushworth et al. where differences in neural correlates between repeat and switch trials emerged only in terms of signal strength (i. Namely.. Jost et al. This result is in line with previous findings (Jamadar et al. it should be noted that this difference is intrinsic to most cued task-switching paradigms and that the cue in single-task blocks is task-irrelevant. and that the contribution of bottom-up processes (i. whereas the late sustained negative component subtends the active maintenance of the relevant task set and the inhibition of the competing S-R mapping. 2012. has consistently been reported in previous neuroimaging literature on task-switching and has been confirmed to play a critical role in the rapid selection between competing task-sets (Crone et al. furthermore. timing. the results demonstrate that mixing and switch costs reflect the action of separate neural mechanisms. 2002).. However. from 580 to 720 ms). the longer CTI duration allowed us to better . The inferior parietal and superior temporal cortices might have subserved top-down attentional control processes evoked by a cue that directs attention in anticipation of a target (Hopfinger et al. Despite the fact that the brain source results must be taken cautiously. 2014). 2001. They did not report early significant modulations associated with repeat relative to single-task trials (i. we might speculate that the early sustained positive component underlies the updating of the task rule. The comparison between mixed and single-task blocks allows controlling for these generic preparation processes. whereas task-switching requirements affect later ERP components (the posterior positivity).. Mueller et al.. the intervention of the left temporal lobe plausibly indicated the retrieval of verbal labels associated with the S-R mappings (Bunge et al. Taken together... and topography. The first component likely represents a mismatch between the current cue and the previous one. 2015).. the manipulation of its degree of informativeness has been shown not to affect cue processing (Czernochowski. the following positivity. 2004). Finke et al. based on its temporal features it might be attributed to the early frontal positivity described in literature (Astle et al.168 V. 2000.. could partially reflect the detection of cue-change (i. were the putative generators of the mixing-related positivity in the first time window (360–580 ms).g. Ridderinkhof et al. 2008). this finding rules out the interpretation of this component as a marker of general preparation processes. They suggested that left fronto-parietal areas. they showed that cue changes affect cue-locked ERPs only when they convey information about the task to be performed on the upcoming stimulus (Finke et al. 2000). and left superior and middle temporal cortex.e. comprising the superior frontal gyrus and the superior parietal gyrus. Jamadar et al. Logan and Bundesen. It is worth noting that although the early part of the mixing-positivity slightly overlapped in time with the switch-positivity. Lavric et al. 2015. (2009). The anterior sustained negativity. 2012.. they provide useful insights into the understanding the functional role of the ERP components. Its neural generators were located in the left parietal cortex. global field power). The activity of the anterior cingulate gyrus likely subserved inhibitory processes linked to the suppression of the competing task-set (Botvinick et al. 2006.1. However.. 2011a. 2010a. Indeed.. with a widespread representation over the scalp. distributed over left parieto-occipital regions. we assumed the earlier portion of our switch evoked positivity. symbolic tones).e. Finke et al. Rubin and Meiran.e. 2003). 1997. Ullsperger et al. Whitson et al. Previous ERP findings clarified this issue and provided evidence that cue switch and task switch operate at different level of cue processing.. Nicholson et al. while the later ERP modulations likely denoted proper task-set reconfiguration processes. 2005. 2000. These results are partially in line with the study by Goffaux et al. 2005. Capizzi et al. ERP switch cost The ERP correlates of the switch cost were represented by two positive components. Nicholson et al. 2012). Cutini et al. This component might be properly attributed to the ‘switch-positivity’ (Karayanidis et al. The first switch-related ERP component was broadly distributed over the scalp and reached maximum amplitude values over central sites. Indeed.. whereas the second component was localized to left parietal sites. Tarantino et al. Classically.. 2008.. 2008b. Afterwards (i. Although its distribution was not limited to frontal areas. it was associated with the activity of more extended regions in the frontal and posterior parietal cortex. Once the cue change is detected. Its putative cortical sources. 2005. and supports the idea that it reflects more specific preparatory processes. and by the use of multiple (three) cues. and that this difference could have contributed to the cuelocked ERP effects reported here. 2006. given the limited spatial localization power of EEG. was associated with activity in the anterior cingulate gyri. Furthermore. Poulsen et al. The repetitiveness of the task and the consequent habituation evoke generic not specific preparatory processes in single-task blocks (see Jost et al. attentional resources are reallocated toward the alternative S-R mapping.. Hsieh and Cheng.. 2010b). this component has been ascribed to a CNV. Rushworth et al... 2014). 2012). the mixing positivity). 2002). Wylie et al. discussed below. switch trials differ from repeat trials in terms of attentional shift. A possible confounding factor is represented by the fact that a task switch always coincides with a cue switch.g.. whereas they were critical in mixed blocks. associated with distinct preparatory control processes. which critically comprises the supplementary motor area.. / Brain Research 1646 (2016) 160–173 2008b. the switch-related ERPs expressed more phasic processes with a different topographical pattern. of pure sensory cue changes) is little compared to the endogenous preparatory processes (e. Considering all cue-locked ERP results together. 2003. suggest that the pre-target negativity might underlie goal updating and inhibition processes.. these studies converged in showing that cue switch modulates ERPs at early stages (within 300 ms after cue onset). 2008.. (2006). which likely modulate the late preparatory component...e.. 2004. These discrepancies might be explained by the use of different cue types relative to the current study (words vs. whereas the greater involvement of the posterior parietal cortex underlies the active maintenance of the relevant task-set (Ruge et al. 2012).. 2010a) and suggests that when simple tasks are engaged..1.

and the inhibition of the competing task-set. the target in repeat trials elicited a significantly more pronounced centro-frontal positivity as compared to single-task trials.. the P3b amplitude did not differ between repeat and single-task trial.e. Goffaux et al. the ERP switch cost was reflected in a reduced parietal P3b. Future research is needed to test these hypotheses. The switch cost was characterized by an early broadly distributed positivity and a late posterior positivity at the cue level. 2006). which was expected based on previous studies (e.79..2.. Importantly. Indeed.. 2011). such a high probability of switches has been shown to be associated with reduced switch costs (Dreisbach and Haider. Moreover. they confirm that common preparatory processes were involved in both types of trials (Karayanidis et al. and avoids carryover effects from the mixed to the single-task blocks (Meiran. indicating an increased taskset updating requirement. Target-locked ERPs were also significantly modulated by a main effect the task type. 2005).. based on these findings. / Brain Research 1646 (2016) 160–173 capture differences in neural correlates. their active maintenance. In some cases. The administration of single-task blocks at the beginning of the experiment encourages the learning of the single-tasks. 2001). the task rule activation in working memory. Unlike previous studies (Astle et al. the P3a originates from stimulus-driven attention mechanisms aimed at facilitating working memory processes and is followed by a parietal P3b (Polich. It should be taken into account that in the present study the occurrence of switch and repeat trials was equiprobable (. 2006. Gajewski and Falkenstein. we could exclude that confounding factors related to block order contributed to the quantification of the mixing cost. whereas the second part (i. the two single-task blocks started and ended the experiment whereas the mixed blocks were administered in the middle. We might speculate that the reactivation of the previously loaded relevant task rule and the resolution of response conflict generated by the alternative mapping were engaged in order to select the appropriate response at the target level. related to proactive as well as reactive processes.. The fact that the switchpositivity emerged earlier than the mixing-positivity might be attributed to the long CTI and to the low demand of the task requirements. The first part of this component (i. the P3b reduction in switch trials might be thought as the other side of a reduced negativity for repeat trials (and single-task ones).. Kieffaber and Hetrick. The source analysis did not detect reliable areas that selectively mediated the target-related mixing positivity. likely because it underpins the updating of response rule. none of these cue-locked components were selectively affected by the task to be performed. Based on the spatio-temporal characteristics these modulations might be assimilated to a P3a and NoGo-P3-like potential. using a ‘sandwich-like’ design. it could have an impact on the quantification of the mixing cost given that performance on mixed blocks could have been affected by either practice or fatigue. This evidence suggests that when comparing these two types of trial some control processes are cancelled out.. 2011b). Monsell and Mizon. Capizzi et al. 3 females). 2005. Tarantino et al. 3. 2015). 2012). carried over from the preceding trial (Jamadar et al. Specifically. Likely. 2000. but see Nessler et al.. therefore in our study its modulation did not constitute a correlate of the mixing cost.. the mixing cost was characterized by an early sustained posterior positivity and a late sustained anterior negativity at the cue level. this study demonstrates that switch and repeat trials share many control processes.2. On the other hand.313] and the letter position [F(1. we conducted an additional behavioral study on 12 new participants (mean age7 standard deviation: 2472.. the simplicity of the rules to be implemented on the upcoming target stimulus did not require specific cognitive resources to be recruited in advance. From a neural point of view. p ¼. This suggested that preparatory processes carried out during the CTI were similar for both tasks.. On the other hand. In order to control for the presence of such potential confounding factors. The NoGo-P3-like potential reflects inhibitory processes directed to override the interfering task-set (Falkenstein et al..2.V. Alternatively. . 2008b. 2003). Namely. Unlike the findings of Gouffaux et al. 2007). Miniussi et al. from 160 to 240 ms) was maximally expressed over an extensive area of central sites. namely. Therefore. which expresses the benefit of a positive priming effect.12.391] task (see Table 3 in the Supplementary materials). and a shorter early frontal positivity in switch compared to repeat trials. 1999.g.2. Typically.1. therefore the probability of switches was rather high. The results of this control experiment showed that performance on initial and final singletask blocks were similar. ERP switch cost At the target level..and post-central sulci might reflect the facilitation of task-set selection on repeat trials (Goffaux et al. p¼ . pointing out the importance to include single-task trials in task-switching studies. 2015. 3. from 240 to 350 ms) reached maximum amplitude over frontal sites. and by a more pronounced centro-frontal positivity at the target level. Therefore. 3. Namely the letter identity task evoked more negative potentials over left centro-temporo-parietal sites.5). it has been shown to generate a longer lasting parietal positivity in repeat compared to single-task trials (mixing positivity). Therefore.. with lower switching probability one could expect to find a less sustained mixing positivity and a longer lasting early switch positivity. and by an attenuated posterior P3 at the target level. (2006). Karayanidis et al.e. in both the letter identity [F(1. which may underpin the verbal processing of the stimulus.. which was not prepared in advance. 2012). On the other hand.7 years. suggesting less recruitment of attentional shifting resources (Nessler et al. this anterior positivity was present in switch trials as well. the results suggest that the mixing-related ERP components were present in both repeat and switch trials. Conclusions The present study clearly disambiguates the electrophysiological markers of mixing and switch costs.11) ¼ 1. These results support the idea that working memory and inhibition processes constitute the reactive control mechanisms that generate the residual ERP mixing cost. 2006. The greater activity that emerged in repeat compared to switch trials in the left temporal pole and the right pre. One potential limitation of the present experimental design could lie in the block order. In addition. while single-task blocks could have not. a main effect of task type did not emerge at the cue level. 2006. Meiran et al. Target-locked ERPs 3. 2008).11) ¼. This P3b decrement was considered a specific marker of the residual switch cost (Hsieh and Liu. 2008a. respectively.3. 169 the right superior frontal sulcus appeared to be more recruited in single-task trials compared to repeat trials. 2010b). The P3b decrement in switch trials has been interpreted as an index of the relevant task-set activation (Rubinstein et al. ERP mixing cost The target-locked ERP results add new evidence to previous literature by showing that the ERP correlate of mixing costs lies in an anterior positivity.

the next trial began. ‘E’ and ‘above’. In addition. two target stimuli (letters). however they were invited to carefully pay attention to it since it warned about the upcoming letter. ocular movements. the mixed blocks began with sixteen practice trials.2 (Delorme and Makeig. / Brain Research 1646 (2016) 160–173 4. at a viewing distance of about 70 cm from the computer screen.016–250 Hz and digitized at a sampling rate of 500 Hz.98°). Afterwards. adjacent to the two sides of the PC monitor. amplitudes exceeding a value of 7 100 μV. The target stimuli consisted of two black letters. Each trial started with the delivery of the tone cue (300 ms) together with the fixation asterisk. as indicated by the Edinburgh Handedness Inventory (Oldfield.170 V. During the practice trials participants received feedback on their accuracy.0. 6. and activity lower than 0. ‘Pay more attention’. participants were notified that the tone had no meaning. Woldorff. In these blocks. namely. The mixed blocks were subsequently presented. In the mixed blocks. two types of epochs were extracted: a) epochs from 200 to 1500 ms timelocked to the cue (cue-locked ERPs) and b) epochs from  50 to 800 ms relative to the target (target-locked ERPs). from cue onset to target onset).82°  0. An electrode placed under the right eye (EOG) allowed the monitoring of blinks and vertical eye movements. there were four taskkey combinations. against a white background. which appeared in 36 point Courier New bold font (1.e. 15 females) took part in the study. and from  50 to 50 ms for target-locked epochs (the 50-ms interval after the target onset was included in the baseline in order to minimize the distortion of the ERP averages due to the overlap with previous cue-related ERP modulations. or ‘No detected response’. namely the F key could be associated with ‘A’ and ‘above’. 2014). peak-to-peak deflections in a segment exceeding 7 100 μV within intervals of 200 ms. USA). respectively. the words ‘Correct’. one of the two letters appeared either above or below the location of the fixation stimulus. Experimental procedure 4.001 and a transition band of 20 Hz (Widmann et al. running in a Matlab environment (Version 8. The tone intensity was uniformly set at 60 dB. The trial procedure was identical for single-task and mixed blocks. Practice trials. and the two tasks were presented in separate blocks. the complementary mappings were used for the K key. a maximum pass-band deviation of . The assignment of categories to the response keys was counterbalanced across participants. tones) with the task was also counterbalanced across participants. the target stimulus appeared and was displayed for 1500 ms. Natick. The automatic detection criteria included an absolute difference between two sampling points exceeding 30 μV/ms.8). with the index fingers of the left and right hands. Namely.0. The continuous EEG trace was low-pass filtered by a windowed sinc FIR filter. mean laterality score7standard deviation: 82.23°. trials containing further artifacts were removed by means of a semi-automatic procedure. Within the mixed blocks. This was the case for three participants. Task and procedure The task was a cued task-switching paradigm and included two single-task blocks and four mixed blocks. MathWorks. which lasted 300 ms. the probability of a switch trial was 0. 4. they had to perform either a letter identity task or a letter position task. Apparatus and stimuli Stimulus presentation and response collection were controlled by Eprime 2 software running on a personal computer connected to a 19″ LCD monitor. Tarantino et al. The tone preceding each target presentation indicated with full certainty the upcoming task. labeled ‘1’ and ‘2’. from  200 to 0 ms for cue-locked epochs. 2004). A cue instructed participants about the specific task they had perform on each trial.5 72.14°  0. The cue stimuli comprised a high pitched tone (1500 Hz) and a low pitched tone (200 Hz). respectively. Germany) connected to 64 Ag/AgCl ring electrodes mounted on an elastic cap according to the extended 10–20 system. and ‘E’ and ‘below’. Participants repeated the practice phase of the mixed blocks if they made more than four errors out of sixteen trials (4 25%).5. irrespective of its spatial position. typically used in the ERP literature of . The resulting data were baseline-corrected using two different time windows. The ‘F’ and the ‘K’ keys on the computer keyboard. the first trial of each block. at a distance of 1. Participants Twenty-three university students (mean age7standard deviation: 22. A graphical illustration of an individual trial is depicted in Fig.82° visual angle). in 36 point Courier New font (0. the reverse assignments were used for the other half of participants. The off-line preprocessing of the EEG signal was performed in EEGLAB 12.1. The association of cues (i. who performed the practice phase twice.. ‘A’ and ‘below’. the uppercase vowels ‘A’ and ‘E’.2. EEG recording and data preprocessing The EEG was recorded using the BrainAmp equipment (Brain Products. MA. This computer was connected to a laptop computer that recorded continuous EEG. The on-line reference and ground electrodes were placed at FCz and AFz. In the single-task blocks. On each trial. TP10). In the letter position task participants had to respond to the position of the letter (above or below the fixation point). and a fixation stimulus (asterisk).1 years. Afterwards. with a cut-off frequency of 40 Hz. Participants were positioned in an electrically isolated and sound-shielded room.4 721. 1993). Each block included 60 trials. Following a variable inter-trial interval ranging from 500 to 1000 ms. and reported normal or corrected-to-normal visual acuity and normal hearing.65. They were presented one at a time in the central x-axis coordinate of the screen. which was displayed for 1500 ms. trials with errors and trials following errors were discarded. together with the asterisk. They signed a written informed consent prior to their participation.1 (Brain Products GmbH). All participants were right-handed. The order of the tasks was counterbalanced across participants. The study was approved by the Bioethical Committee of the Azienda Ospedaliera di Padova and was conducted according to the guidelines of the Declaration of Helsinki.2.3. Raw data were band-pass filtered between 0. for half of the participants the high pitch tone was associated with the letter identity task and the low pitch tone with the letter position task.4. participants were instructed that the tone indicated the task to be performed. The tones were presented by loudspeakers (Yamaha NX-50) placed in front of participants. Overall. The asterisk was displayed centrally on the monitor.1 μV within intervals of 200 ms. and muscle artifacts were detected and removed by means of an independent component analysis (ICA). The single-task blocks began with four practice trials. The stimulus material consisted of two auditory cue stimuli (tones). a Kaiser Window type with a beta of 5. irrespective of its identity. Then. The deadline for the response was 2000 ms. data were re-referenced to the average of the two mastoid channels (TP9. 4. 4. The impedance of each electrode was kept below 5 kΩ. All participants began the experimental session by performing the two single-task blocks. 1971). Munich. Blinks. followed by a fixed time interval of 900 ms (1200 ms SOA. In the letter identity task participants had to respond to the identity of a letter (either the letter ‘A’ or ‘E’).. and in BrainVision Analyzer 2. were used as response buttons.

The Bonferroni correction for multiple comparisons was applied to p-values in post-hoc tests. a brain source analysis was applied in order to estimate the generators of significant scalp-level ERP differences. the first trial of each block. Differences between pairs of conditions. The single-subject ERP averages for each of the six Target Single Repeat Switch Single Repeat Switch 56 (2) 57 (2) 53 (5) 53 (4) 51 (6) 49 (6) 57 (2) 58 (1) 55 (5) 54 (4) 52 (6) 51 (6) experimental conditions were entered as input data. Tadel et al. between 50 and 800 ms (i. The mass univariate analysis was performed by means of an open-source toolbox implemented in Matlab (http://open wetware. 2005). Maris and Oostenveld.or a high-frequency tone.02). 2011a. wMNE.e. A cluster mass permutation test was applied in order to detect reliable differences (Bullmore et al. A schematic illustration of stimuli sequence in two successive trials.65 cm were considered spatial neighbors and adjacent time points were considered temporal neighbors. namely the Friedman's ANOVA and the Wilcoxon signed-rank tests as a post-hoc test. For each permutation. The most extreme cluster mass in each of the 2501 sets of tests is used to estimate the nullhypothesis distribution. In this analysis. 1999). 62  376 time points) were included in the analysis. the participants had to discriminate the letter identity or the letter position. 4. / Brain Research 1646 (2016) 160–173 171 Table 2 Mean number of trials (standard deviation) included in the analyses of Cue-locked and Target-locked ERPs. letter position) as within-subject factors. it elucidates the spatial (at the scalp level) and temporal distributions of the differences between experimental conditions. The two mastoid channels were excluded from analysis. 62  601 time points) and a total of 23. for each channel (n ¼62) and time point. based on minimum norm solutions (weighted minimum-norm current estimate. a total of 37. When necessary.262 comparisons for the cue-locked ERPs.6. as implemented in the Brainstorm software package (http://neuroimage. A 3  2 repeated measures ANOVA was performed. Cue Letter identity Letter position Fig. In order to reconstruct the spatial distribution of the cortical current sources. The inverse transformation was applied to the ICBM-152 standard anatomical template. Depending on the cue type.7. trials with errors and trials following errors were excluded. The z score values were then averaged across participants over the time windows which emerged as significant in the analysis of the ERP data. The z maps for each experimental condition were obtained and compared between conditions using whole brain two-tailed paired t-tests. The cue was represented by an auditory stimulus (tone)..001 (extreme t-scores) were included in clusters with neighboring extreme t-scores. 2010. 2007).. 4. Groppe et al. 2001).and target-locked ERP responses were estimated by modeling the conductive head volume of each participant according to the OpenMEEG Boundary Element Method (Gramfort et al. the absolute sources activity (in picoAmpere-meters) was transformed in z scores relative to such baselines. separately..usc. 2011a. The averaging procedure was performed on each epoch. Electrodes within a distance of approximately 3. Practice trials. all t-scores corresponding to uncorrected p-values less than 0. whereas the switch cost was assessed by contrasting repeat to switch trials. An alpha level of 0. for each trial type and task type. separately for letter identity and letter position trials. The mixing cost was examined by contrasting single-task to repeat trials. task-switching.org/wiki/Mass_Univariate_ERP_Toolbox. This data-driven approach contrasts pairs of experimental conditions across all electrodes and time points simultaneously and detects significant differences by means of non-parametric permutation tests (Groppe et al. Overall. Afterwards.05 was considered for statistical significance. The event sequence was the same in single-task and in mixed blocks. taking into account the several comparisons over space and time. which signaled a following letter. a Greenhouse-Geisser correction was applied and corrected p-values were reported. The cluster-based test exploits the fact that ERP effects are more likely than noise to extend across many adjacent electrodes and time points and is the most appropriate mass univariate procedure for detecting broadly distributed effects (Groppe et al. The number of included trials for each condition is reported in Table 2. between 0 and 1200 ms (i.. ERP data analysis A mass univariate analysis was performed on cue-locked and target-locked ERPs. Therefore. Behavioral data analysis Individual mean response times (RTs) and accuracy (percentage of correct responses) for each trial and task type were entered in statistical analyses. 2011). Kybic et al.e. it could be either a low. The sum of the t-scores in each cluster was the ‘mass’ of that cluster. For each participant. This relatively conservative significance threshold was chosen in order to capture only relevant clusters. nonparametric tests were applied. The sphericity assumption was checked by means of Mauchly test.V. were submitted to two-tailed repeated measures t-tests. 4. Since accuracy rates in four out of six conditions were not normally distributed (p o. 6. The permutation cluster mass percentiles ranking of each cluster from the observed data were used to derive the p-values. Brain source analysis The cortical generators of cue. repeat and switch) and Task type (letter identity. with fixed variance computed from 200 ms before cue onset to cue onset. All analyses were conducted with SPSS 22. including Trial type (single-task. Tarantino et al.5. 2011a. and from 50 ms before to 50 ms after target onset. 2007). The letter could be either an ‘A’ or an ‘E’ and could appear either below or above the fixation point (an asterisk).312 comparisons for the target-locked ERPs. vertices and time points were clustered on the basis ... 2011b). 2011b).edu/ brainstorm.002 fixed dipoles oriented normally to the cortical surface was used as a model. a three-dimensional grid of 15.. Effect size was calculated by means of partial eta squared (ƞp2).. Experimental procedure. A total of 2500 random withinsubject permutations of the data were generated by the algorithm to estimate the null distribution (Manly.. The covariance matrix was assumed to be independent across EEG sensors. The maximum possible number of trials is 60. The p-values were corrected for multiple comparisons with a cluster-based non-parametric test using 1000 permutations. Baillet et al.

J.. G. Jackson. Mecklinger. M.1016/j. 43. Tarantino et al.doi.neuropsychologia.3389/fpsyg. Pollmann.doi. EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis.doi. Attentional set shifting modulates the target P3b response in the Wisconsin card sorting test.. Front.00910. C. P. http://dx.1006/nimg.doi. Falkenstein. Brain Res. S. 1203. L. T. G. Penolazzi.T.org/10.org/10. Wallis...Y.1016/S0028-3932(00) 00046-4. M. Capizzi.. Fulham.S. Aubert. 2014. Jackson.. Barceló. F..brainres. Diversity of the P3 in the task-switching paradigm.2011. Electrophysiological correlates of residual switch costs. E. 2011a. 2011b. Styles. 1137–1154. 148–158. Murphy... Pozo. Hsieh..2008. 1084.1152/jn. Bullmore.).x.J. http://dx. 230–236. 15. P. An intention-activation account of residual switch costs.. Braver. M.org/10.1016/j. D.doi. Process..org/10. 2000.014. Swainson. 32–42. Hoormann.M.doi. http://dx. Psychon. Electrophysiological correlates of anticipatory and poststimulus components of task switching. tics.E.. http://dx. 2000. Brass.P. Lavric.172 V. G. http://dx. S.05. 2006. Baillet. Taylor..D...01..A. 132–145. J. D. Control Cogn. 2001.1016/j. J. G. Cutini. M.biopsycho.1016/j. Hohnsbein.06439. N. http://dx.1016/j. Michie.D.doi. Cohen. Moscovitch (Eds..org/10. Online 9.org/10. Neurosci. 329–348. Hughes. T. Vallesi. Cortex 16. http://dx. Acta Psychol. 1–14. Mangin. 2003.org/10. M. A special thank goes to the Città della Speranza in Padova for its logistic support. Trends Cogn. Jamadar.2011...doi. Falkenstein. Whitson.org/10. S...org/10. Dreisbach.. I.. Kleinsorge. 2004. M. Biol. Brain Res.. Swainson. G. Czernochowski. http://dx. Brunia. M.org/10. Neurosci.org/10.org/10. Electrophysiological evidence for domain-general processes in task-switching. Buonocore.org/10. http://dx.2014. Neurosci. Biol. Methods 134. Preparatory adjustment of cognitive control in the task switching paradigm.21277. 28. brainres.. Supplementary material Supplementary data associated with this article can be found in the online version at http://dx. Scatturin. H. J. Dissociating strategy-dependent and independent components in task preparation. Cereb. Sequence effects in cued task switching modulate response preparedness and repetition priming processes..doi. XV.doi.3758/BF03193853..1016/j. Muñoz-Céspedes. 2006. J. Urbach.06.jneumeth.2008. S.8.. 2010b.2016. Neuropsychologia. T.. D. http://dx. D. Acknowledgments This work was funded by the 7FP/2007-2013 European Research Council Starting grant LEX-MEA (GA no. E.2011. Kutas. doi. Dissociating neural indices of dynamic cognitive control in advance task-set preparation: an ERP study of task switching. von Cramon. 278–290. A brain-potential study of preparation for and execution of a task-switch with stimuli that afford only the relevant task. pp. but not long preparatory intervals.. Variability in proactive and reactive cognitive control processes across the adult lifespan.. J. 1999.doi.2005. S.P. J. L. P.1016/j. 1999. http://dx.. Front.1093/cercor/bhi127.doi. Kahn.2002.. http://dx.. Brain Res. 3389. 101. S.1469-8986. 75. Hsieh.... Hopfinger. van Boxtel. 2011.1016/j.. 1–10. T.. Med. 1342–1355. Coltheart. 421-452.3758/CABN. 1–19. 2008b. C. G. Urbach.D. J. S.. Michie.. 90. Biomed. J.009.E.doi.org/ 10.E. NeuroImage 5. Crone. A. J.. 357–376.org/10..1016/j. Smith. D.2006. Psychophysiology 40. P. http://dx. Evaluation of inverse methods and head models for EEG source localization using a human skull phantom. 3.. a.2016.1002/ hbm. D.. Phillips.1111/j. D. 94–103. Barceló. R.R. Psychophysiology 48.J. 1138–1148. Conflict monitoring and cognitive control.doi. http://dx.org/10. 46. Laura Babcock for her careful English proofreading. http://dx. S.doi. K.. 2011. T. S..12.3. Rev. A. R. http://dx...S. Gajewski. Escera. 2000. 2008. voxel. Fractionating the cognitive control required to bring about a change in task: a dense-sensor event-related potential study.010..org/10. F.2006.1016/j. Hsieh. L. 2001. Electrophysiological correlates of task conflicts in taskswitching. 2000.E. vol. M. Jamadar. Hum. 2008a.. E. The preparation effect in task switching: carryover of SOA.1126/science. I.. Cogn.. . 2003. S. E. biopsycho.1038/72999. 110.. Neurophysiol... Garnero.T.S. NeuroImage 42. Delorme. 117–131. 106–113.pone.001.doi. Bisiacchi. Neural evidence for dissociable components of task-switching.org/10. 52–67.brainres. T. Jonides. Bunge. H. S..07.org/10. De Baene.. De Jong. R. http://dx. T. 59–75.. Imaging 18.doi... Mangun. 2008....org/ 10. Capizzi. B..010. Sinai. http://dx..M.org/10.092.. J.. http://dx.org/10.1134404〉. 9.00932.doi.H. G.2006.. Clerc.10. Olivi. Mizon.. Psychol. Int. 2012. The effects of foreknowledge and task-set shifting as mirrored in cue. Control Cogn.1111/j. Goffaux. Michie. http://dx. Task reconfiguration and carryover in task switching: an event-related potential study. org/10. M.11..00221. Med Biol. J. E. Overmeyer. 475–486. Wiggins.doi. http://dx.org/10. 624–652.07.624.D. M.doi.K. Cheng. http://dx...2011.B. A. 〈http://dx. 945–955. Psychol.org/10. In: C. Astle. Global.org/ 10.. Wendelken.0049486.3758/s13415– 014-0302-y.. W. E.M..20015. http://dx. Karayanidis. F.2010. 2006.x. http://dx. Michie. 13.013.060.1469-8986. Suckling.D..2337/ db11-0571.3389/fnhum. P.J..doi.02.. 1994. Brammer..T.S. S.doi. Jost. / Brain Research 1646 (2016) 160–173 of spatial and temporal adjacency (Maris and Oostenveld. Braver. A. 77–96.1155/2011/545023. Psychol. 87–97.09. Neuropsychologia 38. Neurosci. Liu. Miller.1016/j. G. Finke.2008.. pp.).. F. The variable nature of cognitive control: a dual mechanisms framework. Swainson. Cohen. 2006.org/10. U. Vallesi. Significant areas were identified according to the Destrieux atlas parcellation.org/10. M.doi. (Eds.. http://dx. A.C.. D.. M. F. P. 432–449.org/10. Brain Res. http://dx. 16. Karayanidis.1111/j. E. 2010..F.. K.. 2010a. Selective activation of the superior frontal gyrus in task-switching: an event-related fNIRS study. 32. S.D. http://dx. Botvinick. ERPs dissociate proactive and reactive control: evidence from a task-switching paradigm with informative and uninformative cues. Hum.and target-locked event-related potentials.. 153–163. IEEE Trans. Neurosci. 2004.2003.. Donohue..015. Papadopoulo. O...doi. P.00037. C. Bull. E...M. 2007.doi. Behav. Riera. Altmann. http: //dx. Gamberini... Rabe-Hesketh.1162/ jocn.. F. F. 2001. and Ettore Ambrosini for his support in data analysis. A parametric study of prefrontal cortex involvement in human working memory.doi.00318.doi. J.01272.. J. Czernochowski.doi. S. Karayanidis..1186/1475-925X-9-45. 284–291.. Process. Cogn.1016/ j.. R. 2. B.. C. P. http://dx. The neural mechanisms of top-down attentional control.1371/journal. Intentional reconfiguration and involuntary persistence in task set switching. A. Falkenstein.M...1111/ j. for a difference between two groups of structural MR images of the brain.108.doi. M.S. Schubert. 1411.org/10. J. D..biopsycho.R.00124. http: //dx.1469-8986. Cogn. Umilta & M. A. F.. J.750253.org/ 10. Monsell. 2012. Nat. Neurosci. http://dx. Prefrontal cortex activation in task switching: an event-related fMRI study. E..3389/fpsyg. 33. Behav. ERP components in Go/Nogo tasks and their relation to inhibition. 267–291. Behavioural and electrophysiological measures of task switching during single and mixed-task conditions.. L. Mayr. P. http://dx. D. 8.doi.1037/0033-295X.. 2. Sci.01. Karayanidis. a.1996. 49–62. 10. 2008. Houghton.2009. PLoS One 7. Mass univariate analysis of event-related brain potentials/fields II: simulation studies. Affect. Neural circuits subserving the retrieval and maintenance of abstract rules. J. Astle. Kray..E. Noll.. Cortex 46.org/10.doi..x.doi. brainres.. Arbula... Nystrom.. Task-switching preparation across semantic and spatial domains: an event-related potential study. E.doi. cortex. 116–125.org/10.2009. R.org/10. Ambrosini.. W.090. J..2011.doi.1109/42.x. Groppe. Carter. M. 331–355.. Attention and Performance. J. 2016. Front.org/ 10..1016/S0001–6918(99)00008–6. Event – related potentials reveal multiple components of proactive and reactive control in task – switching. Psychol. References Allport. A. Astle. Eur.048. Mass univariate analysis of event-related brain potentials/fields I: a critical tutorial review. Task Switching and Cognitive Control. M. 2010. Affect. Gajewski.2011.1016/j... The role of spatial information in advance task-set control: an event-related potential study..A.05. Marin. Neurosci. Age differences in task switching and response monitoring: evidence from ERPs..org/10. Brain Mapp.org/10. 365–386. 2015. Brain Res. Makeig. http: //dx. Wait and see.org/10. T. Jamadar. M.. a.neuroimage. Cogn.doi. The authors wish to thank Mariagrazia Capizzi for her precious comments on earlier versions of the manuscript. S. Zorzi. In: Grange. Pushkar.M. 1997. Fehér.1. 9–21.T. Groppe.M..74. Gramfort. Dove.1016/j.2008. Elchlepp. 313692) to AV. Cogn. Kutas. J. Dell’Acqua. F.. Brain Res. 1–10. Psychol. Rösler. Karayanidis.1460-9568. The spatial and temporal dynamics of anticipatory preparation and response inhibition in task-switching.org/ 10. Haider. Rev. Goschke.doi. Mazzonetto. http://dx. Heathcote. neuroimage. T. C.org/10. S. D. 2006. K.009. Eng. 2007).doi. Shifting intentional set: exploring the dynamic control of tasks. Menon.3758/BF03195828. 74–84. http://dx.doi. http://dx. Psychol.M. a.. 45. http://dx. ERP evidence for scarce rule representation in older adults following short. Jackson. 3419–3428.. brainres. 2014. Barch.01273.. Appendix A. S.04. Braver.T... Psychophysiol..1111/1469– 8986.. P. 2014. M.. http://dx. M..011.E. Rubia.doi. Phys. A.A. Psychophysiology 47. 72.. NeuroImage 51.D. http://dx. Eppinger. Psychophysiology . Mem. OpenMEEG: opensource software for quasistatic bioelectromagnetics.org/10. 103–109. 1404–1418. 2012.doi. by theory and permutation. 334–338..2015.12.0247. Biol.M. 108. H.092. 20. and cluster tests. A. 255–267.. G.R. Is task switching nothing but cue priming? Evidence from ERPs. John. Wagner. 2000. 2011a.. 1–15. 1125.. Bunge.

http://dx...doi. 136. M. 1995. Psychol.doi.. 4–26. http://dx.00041. 2005.. On the origin of mixing costs: exploring information processing in pure and mixed blocks of trials. J. http: //dx.T. I. Kieffaber. http://dx. C. Distinct neurophysiological mechanisms mediate mixing costs and switch costs.M. Arbula.1016/j.org/10.1007/s004269900003. F. 849–874.1016/S1364-6613(03)00028–7. 380–384.00298. Sapir. 2012. T.D. Mem. D.org/10...04. 40. 2007.doi. Modulation of brain activity by selective task sets observed using event-related potentials.255.. Picton... Fogelson. N. Components of attentional set-switching. J. Med. 11. N.E. A. R. T. P. http://dx. N. 540–554. http://dx. 2015.org/ 10. http: //dx. Woldorff.org/10.2011. U.00025. 7.doi. F. C. 1514–1528. Pantazis. Changing internal constraints on action: the role of backward inhibition.. 2012.. 1996. Rev. Los.101. Rev..1037//0096-1523. Steinhauser. Cogn.. 133–154. A. 173–183. A. 45–73. D. J. Nicholson. Heathcote. 134–140. Gen. Tarantino et al.. Components of switching intentional set. C.. 13.. 1999. Leahy. Mizon.2000. 1258–1275.doi. R.. Fulham. R.biopsycho.3389/fpsyg. http://dx.. Q..1037/0096-1523..doi. 2009. / Brain Research 1646 (2016) 160–173 Karayanidis... S.014.1469-8986. T..F. 52.21420. Philipp. 2004. R.1016/j. Psychol. 2128–2148... 255–274. A. http://dx. Event-related potential correlates of task switching and switch costs. Neurosci.neuropsychologia. org/10. R. http://dx. D. Vallesi. Exp. 83–98.doi.. 2006. 2011a. Heathcote.1469–8986.R. Learn.04. J... In: Grange. R. R. Front. brain activation. A.07.06372.F.1016/j.1460–9568.2005. a.1126/science. J.doi. Domain-independent neural underpinning of task-switching: an fMRI investigation. Cogn. Bull. Brown.. Houghton. http://dx. Winter.1016/S0167-8760(98)90168-4〉. D. Cogn. Components of task-set reconfiguration: differential effects of “switch-to” and “switch-away” cues. P. Neurosci.A. 383.W. 2010. Nessler. 56–65.2006...2. A. V.org/10.1111/j.. Percept. Marzi. Neurosci.M.1155/2011/879716.doi..doi.W. Neural mechanisms underlying the cost of task switching: an ERP study.016. 1423–1442.org/10.org/10.. G.cogbrainres. Meiran.. Shallice. G. . Jost. F. L. 2014.doi. Very clever homunculus: compound stimulus strategies for the explicit task-cuing procedure. Provost.org/ 10. http://dx. Polich. Psychol.124.01. S.. H.2005.x. 2015. Michie. Ullsperger.J. Nessler. Walter.1993..2010. Michie.org/10.I.. Updating sensory versus task representations during task-switching: Insights from cognitive brain potentials in humans.. Meyer..doi.doi. http://dx.. W. 2004.3758/BF03196709.00262. http://dx.1423. Acta Psychol.doi. Brain Res. Cogn. 5.org/ 10.doi. S. Dynamics of task sets: evidence from dense-array event-related potentials. B. Evans.04.. N. 21.x. Smith.. Behav. C. . Kiesel. Res. R. 2000.org/10.doi. Exp. 202–215. 2013.9.1037/0278-7393.doi. I.83.M.. L.. 2009.763.D. http://dx.brainres. 2001.E. J. A.005.doi. Exp. S.2005. J.. F.x.doi.. Kybic..2007. Task practice differentially modulates task-switching performance across the adult lifespan. Allport. On the origins of the task mixing cost in the cuing taskswitching paradigm.. 31.1. 1095.08.doi. Wendt. Neurosci. R. F.. 2011.1038/203380a0.M. T.doi.G. 2005.207 Rubin. adolescents. 9.1016/0001–6918(95)00050-X.. http://dx. Nieuwenhuis. E. 177–190. (Ed. Davey.org/ 10.org/10. 24. F. Clin.. G.019.clinph.E. Kok. Lavric.03. A.00383. H.1016/j. Wylie.01115.org/10.R. Contingent negative variation: an electrical sign of sensorimotor association and expectancy in the human brain. Nicholson.brainres. Lang..09. http://dx. A.22. 1993. 25–40. M. Passingham. 1121.00350.. Task switching. Mansfield.2. Kray.doi. R. 145–188. Friedman. C. E.doi. jneumeth. 94. R.L.. Baillet.doi. http://dx. org/10. Hum. Exp. Task switching and the measurement of “switch costs”. Acta Psychol. Eur. J. http://dx. E.. Science 306.org/10. Psychophysiology 30.3. A common formalism for the integral formulations of the forward EEG problem. W. Oldfield.. A. 1996. a. E.. 2006. F.org/10. 245–262.S.. Tucker. Meiran. J.U.org/10.org/10. Brain Res.. 28.002. 2005. Psychophysiology 42. 2007.2.1100301 (80-. http://dx.... G. A new account of the effect of 173 probability on task switching: ERP evidence following the manipulation of switch probability.. 94. Karayanidis. 107–123.2004.1162/ 089892902760807159. A. Exp. 12–28. Karayanidis..org/10.52. R. http://dx. Bull.3758/CABN.S... 14. Mecklinger.. Psychol.x. 2002a. Age differences in the selection of mental sets: the role of inhibition. Oostenveld. Jocham. Davies.tb03209. 493–516. Mueller. 2006b. Percept. 2010. 67. J.doi. Behavioural and neurophysiological correlates of bivalent and univalent responses during task switching.0042233.. http://dx. 118.01192.jneumeth.1111/j. Mayr. http://dx.1469-8986.. Psychol.4..doi. Rushworth.. 2007. 96–109.doi.. 1. Multiple effects of prefrontal lesions on task-switching. A.. 1–13. a C.x.2005. neuropsychologia.3. Neurosci.doi. D. N.. D. Psychol. and young adults.1016/j.T..2007.x.01. 2000.org/ 10.T.. P. Affect.1111/ j. O.2005. Gen.brainres.doi.1162/jocn. Hsieh. A.1016/j. 2006a.. 12–35.A. Meiran. Exp. Psychophysiology 42. F. Neuropsychologia 47. Age differences in attentional control: an event-related potential approach. Brain Res.cortex. Wylie.. 124-136.S. Foxe. W. 2014. stimulus ambiguity.-J... J. Advance preparation in task-switching: Converging evidence from behavioral. Methods .. Neurosci. Provost. Manly. http://dx... http://dx. Michie. M. Chorev.2307/1271419 Manzi.doi. Psychophysiology 48.. Psychon.1111/j. Cogn..E.doi. Anticipatory reconfiguration elicited by fully and partially informative cues that validly predict a switch in task. G. A.. http://dx. Czernochowski.doi.. 832–840..T. B. Perform. Maris.org/10. Brain Res. A.J.L. Javitt.. Brain Res.org/10.. Monsell. Koch. Rubinstein. McCallum.. Keriven. Imaging 24. Karayanidis. Allport.C..org/10. Rev. Hum. Nobre. 2003. Switch-specific and general preparation map onto different ERP components in a taskswitching paradigm. 10.. 2011. http://dx. e42233.1006/cogp. Randomization...16.A. U. and model-based approaches. 1998. 488–492. 1157. Digital filter design for electrophysiological data – a practical approach.C.1016/j. Swainson.. Control and interference in task switching – a review.1037/ 0096-3445. Paton.org/10.F..L. 2000. The development of anticipatory cognitive control processes in task-switching: an ERP study in children. Psychol..1. 41. Meiran chapter.org/10.2006. 91.. Friedman. Neurosci.doi.actpsy. Phillips. Rushworth. Neurosci. http: //dx. Neuropsychologia 9. Biol.. 2012. http://dx.. 1.1111/ j.. J. M.3758/ BF03196511.pone. http://dx. org/10. Galloway.6. 2001. Aldridge. http://dx. A..1371/journal.org/10. Poulsen. Reactive control processes contributing to residual switch cost and mixing cost across the adult lifespan. J. Executive control of cognitive processes in task switching. Hetrick.doi. Updating P300: an integrative theory of P3a and P3b.1111/j. G. Task-set switching under cue-based versus memory-based switching conditions in younger and older adults. Karayanidis.doi.. Falkenstein.. http://dx.. Psychol.org/10.008. Bull.doi.12. 1105.org. F. Monsell. Can the task-cuing paradigm measure an endogenous task-set reconfiguration process.org/10. K.. 2014. PLoS One 7..3389/neuro. 30. B..doi.org/ 10. Murray. 2003. M.837363.2004. Costs of a predictable switch between simple cognitive tasks. Faugeras..doi. 2003. 83–92.2009. http://dx.. P. The role of external cues for endogenous advance reconfiguration in task switching.doi. 1160–1172. Nobre. S. J.R. Eppinger. org/10. Front. 1016–1029. Bundesen. http://dx.x.C. S.1111/j. Tadel. S..P. Meuter. 1139–1150.1469-8986.C.4.).6.org/10. 763–797..2015. http://dx.493. Poboka. S. 34. 2005. On the utility of P3 amplitude as a measure of processing capacity. Koch.1037/0096– 3445. J. Intell. Heathcote. Provost. Sci.. J. U. Jamadar.1152/physrev. E. S.org/ 10.. R.10.R. R. cue informativeness and predictability. 2005. Alexander. 407–416. (Amsterdam) 139. D’Avella.. 2005. Passingham. 160–176. P. Task Switching and Cognitive Control. Psychol.org/10. 22. Periáñez. org/10. The assessment and analysis of handedness: the Edinburgh inventory. Mem. and response-set overlap. J.4.. Gillingham.2602.2012.32. 212–233.31. ). The many faces of preparatory control in task switching: reviewing a decade of fMRI research. Papadopoulo. A. Li. 97–113. Psychol.2011. http://dx.C. 2007. Barceló.. M. doi. 27. http://dx.T. Karayanidis. http://dx.doi.1037/a0019842. Mayr. L.046. 2005... Capizzi.014.org/10. J.E.doi.. D. . Wang... N. http://dx. A.1109/ TMI.doi.1477. bootstrap and Monte Carlo methods in biology. Affect. Stuss. Physiol.F. ERPs dissociate the effects of switching task sets and task cues. http: //dx. Johnson. 2008.10. Psychophysiology 42. D. J. 443–447.1016/j.1469-8986. T.G.202.2014. Front. Abboud. 211–253. Zhao. M. Electrophysiological correlates of anticipatory task-switching processes. Karayanidis. F.016. Bilingual language switching in naming : asymmetrical costs of language selection.. Hum. Monsell.016.. Karayanidis.. S. Brain Mapp. Clerc.. Bumak.org/10. Cogn. Jackson. M. G. Causin. Poboka. Danielmeier.. a. 2007. D. S. J. S.2014. Perform. Rogers. Whitson.1016/j. K... J..D. Psychon.pdf. 〈http://dx.doi. Michie.org/10. Methods 164.11. Cogn. Mosher.1016/j.org/10. Comput.. M. 1477–1491. Michie.doi. Mem. Logan. http://dx. Neurophysiol. 2011. Psychophysiology 48.. F.brainres.002.C. B..org/10.1998. K. G. Cogn. Aging 16.1037/ 0278-7393. J.. 〈http://dx. Zimmermann. http://dx....org/10. Z.org/10. http: //dx.2007. 3. Psychol.2010...T. Nonparametric statistical testing of EEG. Nicholson. Exp. 129. Karayanidis. C. Keele. 2005.1016/j.org/10. http://dx. A.org/10. Neuropsychologia 43. Heathcote. W. Sci.D.. Forstmann.. 559–568.0736.doi. Distortion of ERP averages due to overlap from temporally adjacent ERPs: analysis and correction. Reconfiguration of processing mode prior to task performance.org/10. Ruge. Texts Stat. a. Mizon.. Neurophysiological signature of effective anticipatory task-set control: a task-switching investigation.2012. 1971. http://dx. Behav. Heathcote.. Smith. Brainstorm: a userfriendly application for MEG/EEG analysis Franc.96.org.1002/hbm. Meiran. Kray. S. 56–71. Nobre. M. 32. Ullsperger.024.. Maess. A. Cortex 65. Psychol. http://dx.M. Wager.. D. A.org/10.2009.. D.1016/j.doi. pp. Learn. Component processes in task switching.1016/j.. IEEE Trans. S. http://dx. 1964. 35–79. http://dx.G. http://dx.P.J.. Luu. Psychol.27. Jamadar..21009.1027/1618– 3169. S. M. Monsell.. Widmann.. F..1469-8986.1037/0882-7974.09. Schröger.1006/jmla.C. E. Nature 203.doi.and MEGdata. 98–119. M. P.doi. O.doi.. P.. Int.. Psychophysiol. R. D. Crone. J..2007.org/10.. Ridderinkhof. 63. Ruge. The role of the medial frontal cortex in cognitive control. 105–118. http://dx.3389/fpsyg. M. Hum.3758/CABN.doi. Trends Cogn. Psychol.129.V.08.. Cooper. D. Neuroimaging studies of working memory. J.01. 2..doi. 2009..org/10. Miniussi. Neurophysiology of performance monitoring and adaptive behavior. Neurosci.004〉.2008. Whitson. B.