Elsevier

Brain Research

Volume 1157, 9 July 2007, Pages 56-65
Brain Research

Research Report
Behavioural and neurophysiological correlates of bivalent and univalent responses during task switching

https://doi.org/10.1016/j.brainres.2007.04.046Get rights and content

Abstract

A hallmark of human behaviour is its flexibility. In any given circumstance there is typically a range of possible responses that could be selected. In the current study participants were presented with stimulus displays that afforded two simple cognitive tasks and were required to switch predictably between them. The judgements for each task were either uniquely mapped onto separate effectors (univalent conditions) or else mapped onto shared effectors (bivalent condition). The results demonstrated that whilst behavioural switch costs were similar across the mapping conditions, these conditions differed in the patterns of brain activity observed during task preparation and early visual processing of the target. Specifically, a cue-locked switch-related late frontal negativity was present over frontal sensors for the bivalent condition only, and a target-locked N1 over occipital sensors was larger in the bivalent condition than the univalent conditions. In contrast, a switch-related target-locked P3b component was common to all mapping conditions. These findings are discussed with respect to differences in processing demands for switching between tasks with bivalent versus univalent responses.

Introduction

While human behaviour is noted both for its flexibility and coherence, adapting our actions to meet our current goals and circumstances may be cognitively demanding and require considerable neural resources. Patients with frontal lobe lesions, whose ability to self-regulate actions can be seriously impaired, clearly illustrate this dilemma. For example, in some of these patients actions are triggered by environmental cues rather than by intention (Lhermitte, 1983), whilst in other patients perseverative errors occur when a change of action is necessary (Milner, 1963). Even in healthy non-brain damaged individuals, slips of action happen in everyday life (Reason, 1984) and in the laboratory shifting between simple experimental tasks are associated with a decline in performance relative to repeating the same task consecutively (Monsell, 2003). While the decrease in performance associated with switching tasks can be reduced if sufficient time is available to prepare, switch costs are often not completely eliminated by preparation.

While the precise cause of residual switch costs (i.e., the slowing of mean response times that remains despite sufficient preparation time) is subject to much debate, a number of researchers have discussed the difficulty of switching between tasks that share the same set of responses (Meiran, 2000, Schuch and Koch, 2003). Although residual switch costs have been observed when tasks do not share the same set of responses (univalent responses), most task-switching paradigms have used overlapping response sets. For example, Rogers and Monsell (1995) introduced a task (which has since been used by a number of research groups) that involves switching between number and letter judgements. When performing the number classification task, participants indicate whether the number presented is even with their right index finger, or odd with their left index finger. When the task switches to letter classification, the index fingers are now used to specify different judgements. For example, a right key press is used if the letter is a consonant and a left key press if it is a vowel. When a response has values associated with two tasks it is referred to as being bivalent.

Most task-switching paradigms have used bivalent stimulus displays (i.e., items in the stimulus display are associated with more than one task). Items from two or more tasks can be presented together as in the Rogers and Monsell study (e.g., a digit and a letter) or else a single item affording several tasks can be presented. For example, Meiran and colleagues have used a simple spatial task that involves participants making judgements about the location of a single stimulus within a 2 × 2 grid. Depending on which task is cued, participants judge either whether the stimulus is at the top or bottom of the grid (one task) or whether it is at the left or right of the grid (second task).

Both the Rogers and Monsell and the Meiran stimulus displays are referred to as bivalent. However, filtering out the irrelevant stimulus dimension should be easier when the items associated with each task are spatially separate (a between-item switch), compared with when they are part of the same object (a within-item switch). Spatial attention could potentially be used in the former case to discriminate between the items and tasks (Duncan et al., 1997). In line with this proposal, Allport et al. (1994) have demonstrated that there are larger switch costs for within-item switching than for between-item task switching.

Very few switching studies have directly compared univalent and bivalent responses. An exception has been the work of Meiran and colleagues (Brass et al., 2003, Meiran, 2000, Rubin and Meiran, 2005). Meiran found that residual costs were absent when univalent responses were used for each of the four grid locations but present for bivalent responses. He concluded that residual costs arise because of a preference to respond to each task as if it had a unique set of responses. When two tasks are mapped to the same response the meaning associated with each response has to be recoded when the task shifts (e.g., the index finger moves from signifying top when the task is a top/bottom judgement, to right when the task shifts to a right/left judgement). According to Meiran and colleagues, response set recoding cannot be completed until the target is presented. When each task is associated with a univalent response no additional response recoding stage is needed and consequently no switch costs are incurred. Stimulus set recoding in contrast can occur prior to target presentation. Neuroimaging (fMRI) studies that have used the Meiran procedure have reported reduced switch costs in the univalent condition relative to the bivalent condition, rather than no costs. Using longer preparation intervals than the earlier behavioural studies, the difference between switch and repetition trials was small, even for the bivalent condition (Brass et al., 2003).

Switch costs have been reported for univalent mappings, using other paradigms, even when preparation intervals of greater than 600 ms have been used. These paradigms have mainly involved more cognitively demanding tasks than were used by Meiran. For example, Arbuthnott (2005) found significant switch costs for univalent manual responses and tasks that involved classifying a singly presented digit according to its parity, magnitude, or prime status. While Arbuthnott did not compare univalent and trivalent conditions (three tasks mapped to two responses) within the same experiment (or even discuss this comparison), she did vary the number of mappings across experiments. Of more importance to the current study, switch costs in the univalent experiments (Experiment 1 and 3, verbal cues and six fingers) were larger, and reaction times slower, than in the trivalent experiment (Experiment 2, verbal cues and two fingers). Clearly, shared responses are not the only cause of switch costs, and for more complex cognitive tasks than were used by Meiran, it is currently unclear what contribution response bivalency might make to switch costs.

In the current study, we examined the behavioural and neural mechanisms associated with switching predictably between number and letter judgements, using a modified version of the Rogers and Monsell task. We utilised the excellent temporal resolution (millisecond accuracy) of the event related potentials (ERP) technique to examine the neurophysiological indicators of univalent and bivalent responses during task switching. ERPs allow us to isolate processes associated with preparation for a new task from those occurring after target onset (Nicholson et al., 2005, Swainson et al., 2006). They also enable us to look at processing when there is no behavioural outcome (e.g., performance on a covert task or on a No-Go trial) and to discover processing differences when behavioural outcomes appear equivalent. Unfortunately, given their low spatial resolution, ERPs are a relatively poor indicator of the neural generator(s) of the ERP signal.

While there have been many recent EEG studies investigating the ERP correlates of task switching (Astle et al., 2006, Karayanidis et al., 2003, Kieffaber and Hetrick, 2005, Nicholson et al., 2005, Poulsen et al., 2005, Rushworth et al., 2002, Swainson et al., 2003, Wylie and Allport, 2003), no published study has, to our knowledge, used EEG to examine the neurophysiological correlates of the ambiguity of response-task mappings in task switching. As with the behavioural studies, most EEG studies of task switching have used bivalent responses and bivalent stimuli requiring semantic judgements. Consequently, it is difficult to predict exactly which ERP components will be sensitive to response valency.

Anticipation of stimuli or responses is associated with an anterior negative shift in the ERP which is enhanced by increased difficulty (Cui et al., 2000) or more effortful preparation (Falkenstein et al., 2003). Slow wave activity of this type has been ascribed to the allocation of additional cognitive resources (Rosler et al., 1997). Poulsen et al. (2005), using a modified version of the Rogers and Monsell (1994), reported a late frontal negativity, which was enhanced in amplitude for task switch trials (see also Lorist et al., 2000). Although she attributed this negative shift to Meiran's stimulus set selection stage, she noted that this interpretation did not fit with Meiran and colleagues' (2001) proposal that stimulus set recoding does not require frontal lobe activity.

We have previously shown that frontal negativity during task preparation is sensitive to a range of response manipulations. For example, frontal negativity is observed during preparation to switch to a prosaccade task but not to an antisaccade task (Mueller, 2006). It is also absent during preparation for a task switch following a No-Go but not a Go response (Astle et al., 2006), and absent when bivalent covert responses are used (an internal count was increased or decrease according to the judgement required for example ‘+ 1’ for upper case and vowel ‘− 1’ for lower case and consonant-Astle et al., in press).

Another component frequently reported in the task-switching literature is a late parietal positivity that is delayed for switch trials (Wylie and Allport, 2003, Karayanidis et al., 2003, Nicholson et al., 2005). The timing of this component changes as the interval between a task cue and a target is altered. If the interval between the task cue and task is short this component can occur after the target (Nicholson et al., 2005). Although this component has been interpreted as reflecting anticipatory task set reconfiguration, as the timing of this component can occur after execution of the response on a task switch trial, it does not appear to be necessary for selection of the appropriate response (Swainson et al., 2006).

A post target component which has previously been reported in the task-switching literature is the P3b (Karayanidis et al., 2003, Lorist et al., 2000, Nicholson et al., 2005, Swainson et al., 2006, Wylie and Allport, 2003). The amplitude of this component, which is observed over centro-parietal electrodes, is enhanced for repeat task trials relative to switch trials. It is more reliably observed when a fixed order of task switching is used rather than an unpredictable task order (Lorist et al., 2000, Swainson et al., 2006, Wylie and Allport, 2003). Kok (2001) suggested that the P3b component indexes stimulus evaluation processes, specifically the matching of an external stimulus to an internal representation. When it is known that the same task will be repeated for a number of trials, it makes sense for task-related stimulus evaluation processes to be maintained and consolidated. If this interpretation is correct we would expect to see this repeat related component with the predictable task sequence design used here, however we should not expect it to be modulated by response bivalency.

Section snippets

RT

The results showed significant main effects of trial type and mapping (F(1,21) = 107, P < 0.001 and F(2,42) = 5.6, P < 0.01, respectively). The interaction, however, was not significant (F(2,42) = 1.1, P = 0.32). Fig. 1 illustrates the main effect of trial type showing that switch trials were slower than repetition trials. It also suggests that the bivalent condition was faster than both univalent conditions. As the main effect of mapping was significant, comparisons between each mapping condition were

Discussion

This study examined the behavioural and electrophysiological correlates of bivalent versus univalent response-task mappings in a task-switching paradigm. Switch costs were equivalent for all three mapping conditions, although bivalent responses were faster than univalent responses. The ERPs showed a cue-locked switch-related frontal negativity, accompanied by a coincident parietal positivity, for the bivalent but not univalent response conditions. By contrast, a stimulus-locked P3b over

Subjects

Twenty-two neurologically normal adults participated in the study (12 female, mean age 23.6 years ± 7.4 years) and received an incentive of £10 for participation. Local ethical approval was obtained and written consent was obtained from all individuals prior to participation in the study. ERP data from one participant were excluded due to articulation and movement artefacts but all behavioural data were kept for analysis.

Task design

The experiment consisted of two tasks, the letter task and the number task.

Acknowledgment

This study was supported by a BBSRC grant to GMJ.

References (48)

  • R. Swainson et al.

    Using advance information in dynamic cognitive control: an ERP study of task-switching

    Brain Res.

    (2006)
  • D.M. Tucker

    Spatial sampling of head electrical fields: the geodesic sensor net

    Electroencephalogr. Clin. Neurophysiol.

    (1993)
  • M.G. Woldorff et al.

    Modulation of early auditory processing during selective listening to rapidly presented tones

    Electroencephalogr. Clin. Neurophysiol.

    (1991)
  • G. Wylie et al.

    Task switching: a high-density electrical mapping study

    NeuroImage

    (2003)
  • A. Allport et al.

    Shifting intentional set: exploring the dynamic control of tasks

  • K.D. Arbuthnott

    The influence of cue type on backward inhibition

    J. Exper. Psychol.: Learn. Mem. Cogn.

    (2005)
  • Astle, D.E., Jackson, G.M., Swainson R., in press. Fractionating the cognitive control required to bring about a change...
  • R. DeJong et al.

    Use of partial stimulus information in response processing

    J. Exp. Psychol. Hum. Percept. Perform.

    (1988)
  • M. Eimer

    ERP modulations indicate the selective processing of visual stimuli as a result of transient and sustained spatial attention

    Psychophysiology

    (1996)
  • M. Falkenstein et al.

    Short-term mobilization of processing resources is revealed in the event-related potential

    Psychophysiology

    (2003)
  • T.C. Handy et al.

    Attention and spatial selection: electrophysiological evidence for modulation by perceptual load

    Percept. Psychophys.

    (2000)
  • W.E. Hick

    On the rate of gain of information

    Quart. J. Exp. Psychol.

    (1952)
  • M. Hübner et al.

    Response selection difficulty and asymmetrical costs of switching between tasks and stimuli: No evidence for an exogenous component of task-set reconfiguration.

    J. Exp. Psychol.: Hum. Percept. Perform.

    (2004)
  • R. Hyman

    Stimulus information as a determinant of reaction time

    J. Exp. Psychol.

    (1953)
  • Cited by (40)

    • Proactive and reactive control differ between task switching and response rule switching: Event-related potential evidence

      2022, Neuropsychologia
      Citation Excerpt :

      Kieffaber et al. (2013) demonstrated that, compared to switching stimulus-set alone, switching both stimulus- and response-set resulted in a larger switch cost and enhanced switch-positivity over the central and parietal regions. In addition, the stimulus-set and response-set interact with each other due to their hierarchical organization (Kleinsorge and Heuer, 1999; Mueller et al., 2007). The switch cost at the higher level (i.e., task goal) is greater than that of lower level (i.e., response-set) (Han et al., 2018, 2019; Kleinsorge and Heuer, 1999), possibly because the change in task goal will automatically prime the change in response-set (Kleinsorge and Heuer, 1999).

    • Adaptiveness in proactive control engagement in children and adults

      2020, Developmental Cognitive Neuroscience
      Citation Excerpt :

      Amplitude was averaged across these channels for the following windows: 600−1000 ms at age 6, 500−900 ms at age 9, and 300−700 ms in adults. A lateralised CNV was observed in adults, which is consistent with prior work (Mueller et al., 2007) but not in children. In adults, it was left-lateralised and maximal over frontocentral channels Fz, F1, F3, F5, FCz, FC1, and FC3.

    • Electrophysiological correlates of the cognitive control processes underpinning mixing and switching costs

      2016, Brain Research
      Citation Excerpt :

      At least in part, it may represent the decoding of the cue (Karayanidis et al., 2009), which implies rule retrieval and goal activation processes (Jost et al., 2008). A further cue-related component, occurring in the late portion of the CTI in both switch and repeat trials is a slow fronto-central negativity, sometimes labeled ‘pre-target negativity’ (Astle et al., 2006, 2008; Hsieh and Cheng, 2006; Karayanidis et al., 2011a; Mueller et al., 2007; Nicholson et al., 2005; Poulsen et al., 2005; Rushworth et al., 2002). 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, 2001).

    View all citing articles on Scopus
    View full text