Background

Task-switching paradigms are routinely used to investigate aspects of cognitive control related to the updating, representation, and maintenance of frequently changing task rules. A common variant is cued-task switching, which in the relevant task for the upcoming trial is signaled via an advance cue. Following the presentation of the cue, a delay is provided, which allows for task preparation prior to the onset of the target stimulus.   When target stimuli are bivalent (contain features associated with multiple tasks), knowledge of the currently relevant task is necessary in order to know how to respond appropriately.  Cognitive control demands are thought to be higher on task-switching trials, relative to single-task trials (i.e., from a task block in which the task never switches), or even to trials in which the task is repeated  (i.e., from the previous trial).  These cognitive control demands are reflected in the switch cost (the increase in RT and error rates on task-switch vs. task-repeat trials) and mixing cost (the increase in RT and error rates on task-repeat vs. single-task trials).  

An additional indicator of cognitive control demand that we have focused on is the task-rule congruency effect (TRCE).  The TRCE is a Stroop-like effect that refers to the increase in RT and error rates on incongruent vs. congruent trials. An incongruent trial is one in which the two task sets activate competing responses whereas a congruent trial is one in which both task sets activate the same response.  The TRCE is thought to reflect cognitive control because on congruent trials the correct response can be made even if the currently relevant task is not actively represented, whereas on incongruent trials, conflict may arise if the irrelevant task-set is partially activated  (i.e., not fully inhibited) during processing of the target stimulus. 

Cued task-switching paradigms typically involve two tasks that randomly alternate.  A common task-pair that has been investigated is the letter-digit task.  In this paradigm, the target stimulus is a letter-digit pair (e.g., “D 3”,  “1 A”), and the task is to either categorize the letter as a consonant or vowel, or to categorize the digit as odd or even. The task cue indicates which of the two tasks is relevant for the current trial.  Often a highly transparent, semantically meaningful cue is used to minimize encoding demands (“Attend Letter” , “Attend Digit”).  A long cue-to-target interval (CTI) ensures that participants have sufficient time to engage in advance preparation for the target based upon cue information.

As Related to DMC Framework

In the DMC framework, it is assumed that manipulations that encourage participants to prepare the relevant task set at the time of the cue (and sustain preparation during the cue-to-stimulus interval) should result in an increase in proactive control. One such manipulation is use of reward motivation (monetary incentive). When the task cue appears in a certain color (e.g., green) participants have an opportunity to earn extra money if they are faster than in a baseline phase while maintaining accuracy. We have been using a list-wide mostly congruent condition to represent the baseline phase. This is because the TRCE effect is significantly larger in a list-wide mostly congruent condition compared to a list-wide mostly incongruent condition (Bugg & Braver, 2015). We interpret this finding as indicated a reduced level of proactive control when most trials do not require active task cue representation for correct responding. 


Since the TRCE is assumed to be close to maximal at baseline, it enables a test of whether reward motivation can reduce the TRCE in the proactive condition (which is also mostly congruent). Bugg and Braver (2015) found that a monetary incentive significantly reduced mixing cost but had no effect on the TRCE. Our most current task version (see [Distribution Tasks]) includes theoretically guided modifications to the proactive condition used in Bugg and Braver, including associating incentives with specific stimulus types.  

To increase reactive control, we have again elected to use a motivational manipulation in a mostly congruent list (see [Distribution Tasks] for details). However, it differs in two important ways from the proactive condition. One is that it is punishment-oriented. Participants begin the reactive condition with an extra pool of money and are told that on incentive trials, they will lose a substantial portion of this money for incorrect responses. Second, incentive trials are signaled not by the color of the cue but instead by the color of the target stimulus. If the stimulus is green, the punishment contingency is in effect.  Third, incentive trials are preferentially associated with incongruent target stimuli, increasing cognitive control demands in a somewhat item-specific manner.  Accordingly, any modulation of the TRCE may be attributed to a reactive process (adjustments made post-stimulus onset) and not a proactive process (i.e., preparation at the time of the cue).

By using a reward and punishment-oriented incentive in the proactive and reactive conditions, respectively, we anticipate that performance will be characterized by strategic shifts reflecting varying prioritization of speed and accuracy. Increased proactive control should be reflected in a sustained/global motivational context effect that affects all trials, speeding performance (but at the cost of accuracy) even on non-incentive trials. Increased proactive control should additionally reduce the RT TRCE on incentive trials.  Increased reactive control should lead to a trial-specific shift in performance. Specifically, incentivized trials should be selectively associated with a slowdown in performance and a reduction in errors, leading to a reduced error TRCE.

Description of Scripts Available for Distribution

Below is relevant information regarding the available task versions and conditions, in terms of the relevant procedures and parameters.

Behavioral Versions

  1. Baseline & Reward-Incentive (Bugg & Braver, 2015)
    Experiment 3 of Bugg & Braver (2015) implements a mostly congruent baseline and reward incentive proactive version of the letter-digit cued task-switching paradigm, but does not include a reactive control condition.   If you are interested in testing this variant, please the paper for details on all methods information.
    Eprime tasks scripts and raw data
    Access this script
  2. Baseline, Reward, and Punishment Incentives (Web-based)
    This web-based version uses both reward and punishment incentive conditions during letter-digit cued task-switching to investigate baseline, proactive and reactive control.
    Request access to the task scripts

In each condition, participants are presented with the same list of stimuli.  The list is 67% congruent (i.e., mostly congruent) and consists of sixteen different stimuli (1A, A1, 2A, A2, 1B, B1, 2B, B2, 3E, E3, 4D, D4, 5H, H5, 6I and I6).  There are three main trial-type categories (note that these categories are meaningful only for the proactive and reactive condition—in the baseline condition, there are no incentive trials) and each category is represented by only a subset of the possible stimuli: A) Incongruent trials that are always incentivized (comprising 25% of the total list), for example, the stimuli 5H, H5, 6I and I6; B) Congruent trials that are incentivized 33% of the time (comprising 25% of the total list), for example, 3E, E3, 4D, and D4); C) A set of congruent and incongruent trials (83% are congruent) that is never incentivized (comprise 50% of the total list), for example, 1A, A1, 2A, A2, 1B, B1, 2B, and B2.  There are a total of 192 trials.

Each trial begins with a 300 ms fixation cue (“+”) in the center of the screen.  Following the fixation is a 500 ms task cue (“attend number”/”attend letter”) presented in red (except for incentive trials in the proactive condition).  A 3500 ms CTI follows.  The stimulus is then presented in black (except for incentive trials in the reactive condition) until a response is made. Feedback following the response is displayed for 1250ms, followed by a 1700ms blank screen before the next trial begins.

Baseline

There are no incentive trials in this condition. This condition presents all task cues in red and stimuli in black. Participants are asked to respond as quickly as possible while maintaining accuracy. 

Proactive

In the proactive condition, 33% of trials are incentivized.  The incentivized trials are denoted by a green task cue.  Participants are told that they will earn a reward (up to $2 in total across all incentivized trials) for each incentivized trial on which they answer both correctly and more quickly than they did on the same subset of trials in the baseline condition. The criterion is set from the baseline phase, based on the median correct RT.  For example, on an incentive trial from Category A above (i.e., Incongruent trials that are always incentivized—5H, H5, 6I, I6), the participant’s performance would be compared to their performance on trials 5H, H5, 6I and I6 during the baseline phase.   If performance is accurate and faster than the baseline median correct RT, they would receive positive reward feedback for that trial. 

Reactive

In the reactive condition, 33% of trials are incentivized. These trials are denoted by a green target stimulus.  Participants begin this version with a pool of potential reward money ($2) and are told that they will lose a portion of this reward ($0.25) for each incorrect response on incentivized trials.  Participants are told that speed is not as important, but that they should still try to respond as quickly and accurately as possible on all trials.

fMRI Version

The scanner variants are modeled after the behavioral variants with a few notable changes.  All conditions are implemented in a mixed block/event-related design. Within each scanning run, 3 task blocks alternate with 4 resting fixation blocks (30 sec duration); in each task block, the inter-trial interval randomly varies (across 3 step sizes).

Request access to these task scripts

To reduce the length of the task for scanning purposes, total number of trials was cut from 192 to 108, while closely preserving the proportion congruency and proportion of each trial type represented in the behavioral version (33% incentivized trials, 67% congruent trials). Likewise, the trial timing was changed to more closely match other tasks in the battery and the scanner TR.  Each trial starts with a flickering fixation cue (300ms), followed by the task cue (500ms) and CTI (4000ms), The target is presented for a fixed duration (3200ms), followed by visual feedback (800ms).

Relevant Behavioral Measures to Compute

In addition to global behavioral measures of RT and accuracy, there are a number of derived measures that are useful for this task (for any of the above versions), which are described below.   Additionally, it can be useful to calculate the internal consistency (as split-half reliability, Cronbach’s alpha, or ICC) for various measures, particularly if you plan to use these to investigate individual differences.

Incentive TRCE

This computes the magnitude of behavioral interference measured as incongruent – congruent performance on the subset of stimuli that are incentivized (in the proactive and reactive conditions; this  stimulus subset is also used for computation in the baseline condition even though the trials are not incentivized).  It can be computed on RT and accuracy.  The DMC framework predicts a reduction in TRCE for both the proactive and reactive conditions, but in the proactive the reduction should be found in RT, whereas in reactive it should be found in accuracy. 

Nonincentive Trials

This computes behavioral performance, both RT and accuracy, selectively on the 100% non-incentive trials, since these are identical across all three conditions.  The DMC framework predicts a sustained reward motivational context in the proactive condition will produce faster RTs compared to both baseline and reactive conditions, along with increased error rates.  However, the non-incentive TRCE is predicted to be unchanged from baseline, reflecting a more generalized effect.