Prosser, Laura J and Swainson, Rachel and Yamaguchi, Motonori
(2023).
Predictable Task Switching: Manipulations of Preparation Interval and Preparation-interval Variability, 2018-2020.
[Data Collection]. Colchester, Essex:
UK Data Service.
10.5255/UKDA-SN-854762
GRANT ABSTRACT:
In this project we will investigate whether what we "know" and what we "do" have different effects on our subsequent behaviour. We will do this by looking at our ability to switch between different tasks. Specifically, we will compare how difficult it is to switch away from a task that we have either: a) only prepared to perform (we "knew" what the relevant task was but we didn't "do" it), or ii) actually performed (we both "knew" it and "did" it).
In our everyday lives we frequently need to switch between the different rules that guide our behaviour. For instance, when driving a car we might switch rapidly between the following “tasks”: visually assessing potential hazards at a junction; accelerating past a tractor; performing an emergency stop. From studies using laboratory tasks, we know that switching tasks usually leads to slowed responses, and that we occasionally even repeat the previous task in error. The existence of this “switch cost” reveals that some aspect of the previous task must persist in some way to affect the speed or accuracy of our subsequent behaviour, even though we know that it is no longer relevant. In this project, we wish to find out about what causes this cost of switching between tasks.
Our main question concerns whether just preparing a task ("knowing") will have different consequences from actually performing it ("doing"). There are various examples in psychology of situations where what we know has surprisingly little impact upon what we do. For instance, we can sometimes verbally repeat instructions given to us, and demonstrate that we understand and remember them, but then fail to implement them at all (a phenomenon known as “goal neglect”). A similar distinction has been drawn in task-switching research. It had been thought that doing a task would produce a subsequent switch cost, but that knowing which task should be performed without actually carrying it out would produce no subsequent cost. However, more recent evidence using a different method suggests that merely preparing a task can in fact produce a substantial switch cost, even if the prepared task was not performed.
We will conduct a series of psychological experiments in which people perform two different tasks. For instance, we will show them coloured shapes (like a blue circle) and ask them to press a button to indicate either what the colour is or what the shape is. By intermixing the two tasks randomly, we will be able to assess people's ability to switch between tasks, relative to repeating tasks - that is, we can measure each person's switch cost. On most trials, people will prepare a task and then perform it: for instance, they may see the word “colour” and then a blue circle, at which point they press the appropriate button to indicate that the colour is blue. Crucially, however, on some trials we will require a task to be prepared but not performed: e.g., we may show the word “colour” but then no coloured shape, instead moving straight on to the next trial. Therefore, we will be able to measure the switch cost that follows preparation separately from the switch cost that follows performance. Across a series of experiments we aim to find out what causes these types of switch cost to be established and abolished, and in what ways the switch cost driven by preparation may differ from that driven by performance.
The cost of switching tasks indicates a fundamental limitation in our cognitive system that is relevant to many situations (e.g., working in a busy office, driving in heavy traffic, preparing dinner while taking care of a baby). Understanding the mechanisms behind this limitation is of potential practical importance with respect to reducing risks associated with this cost. The present project will meet this challenge by illuminating the origin of this limitation.
Data description (abstract)
When more than one task has to be performed within a block of trials, a cost of switching relative to repeating tasks (a “switch cost”) is usually seen following completed trials (i.e., trials where the task was executed). In studies where completed trials involve making a response such as a button press, and where tasks are sequenced predictably (in alternating runs), that switch cost has tended to be overcome with just one instance of task-execution. In Experiment 1, we wished to see whether a switch cost that was overcome after just one trial of a new task-run would also be present following other types of trial: i) completed trials involving a nogo response (i.e., where not pressing the button is the appropriate response, such as where the task rule requires that the button not be pressed for even numbers); ii) cue-only trials, where a task cue but no target is presented, such that the prepared task cannot be performed. This experiment involved a single cue-type (colour cues), a fixed preparation (cue-target) interval and a fixed trial length. Because there was no way of knowing that participants had prepared the cued task before target onset in that first experiment, we explored in Experiment 2 whether we could observe pre-target task preparation by manipulating the preparation interval. This experiment included only completed trials, and it manipulated cue-type (between blocks; verbal or image), preparation interval (within participants) and whether preparation interval was varied between blocks or between trials (between participants). In Experiment 3, we used the between-trials method of varying preparation interval and image cues from Experiment 2, and we re-introduced cue-only trials.
| Data creators: |
|
| Contributors: |
| Name |
Affiliation |
ORCID (as URL) |
| Bonne Davide |
University of Aberdeen |
|
| Gimadieva Milena |
University of Aberdeen |
|
| Maranescu Emanuela |
University of Aberdeen |
|
| Nicolson Yasmin |
University of Aberdeen |
http://orcid.org/0000-0002-7716-7031
|
| Suonpää Pinja Pauliina |
University of Aberdeen |
|
| Ukkonen Jenna |
University of Aberdeen |
|
|
| Sponsors: |
ESRC
|
| Grant reference: |
ES/R005613/1
|
| Topic classification: |
Psychology
|
| Keywords: |
PSYCHOLOGY, COGNITIVE PROCESSES
|
| Project title: |
Preparation, performance and the subsequent cost of switching tasks
|
| Grant holders: |
Rachel Swainson, Motonori Yamaguchi
|
| Project dates: |
| From | To |
|---|
| 1 September 2018 | 31 August 2022 |
|
| Date published: |
08 Dec 2021 11:29
|
| Last modified: |
09 Jan 2023 12:49
|
| Collection period: |
| Date from: | Date to: |
|---|
| 1 October 2018 | 28 February 2020 |
|
| Country: |
United Kingdom |
| Data collection method: |
This collection includes three separate experiments. The participants were primarily university students. Participants were tested in a laboratory setting, with a PC used to display stimuli and collect button-press responses. In Experiment 1, data were collected over two 90-minute sessions; Experiments 2 and 3 took place over a single session each (Expt. 2, 45mins; Expt. 3, 1hr 15mins). In all three experiments, tasks were presented in alternating three-trial runs. Three different cues were used for each task (in a predictable sequence) in order to avoid cue-repetitions occurring within a run. The background to all trial displays was dark grey; target stimuli were shown in white (Expts. 1 & 2) or yellow (Expt. 3). Feedback was presented at the end of each trial if the participant made a response error (i.e., responding on a nogo trial or not responding on a go trial [Expt. 1], or making the wrong response [Expts. 1 & 2]), being shown in white (Expts. 1 & 2) or magenta (Expt. 3) for 500 ms and followed by a blank screen for 500ms. At the end of each block, the total number of errors and mean RT in that block were shown. Before the experimental trials began, participants practised each task in turn and then practised switching between them. EXPERIMENT 1: The key features of this experiment were: go/nogo response options; letter & number tasks; colour task cues; and fixed trial durations. This design was loosely based upon that used by Wylie, Javitt & Foxe (2003, European Journal of Neuroscience 17:667-672). The tasks involved letter-judgement and number-judgement. Target stimuli comprised a letter and a number displayed on-screen with a central fixation cross (e.g. A + 9). Participants responded either according to the letter (letter task: judge vowel or consonant) or the number (number task: judge odd or even). Response options for the letter and number tasks were go/nogo: for the letter task, vowels required a go response (press button) and consonants required a nogo response (do not press); for the number task, odd numbers required a go response and even numbers required a nogo response. Whether the relevant task was letter or number was cued in advance. Task cues were rows of five coloured fixation crosses (+++++) on a dark grey background. There were three cue colours for each task, which cycled through in a constant sequence (cues 1 to 3 for letter, then cues 1 to 3 for colour, etc.) There were six different mappings of cues to run-positions, with each participant being assigned one of these mappings. All trials were a fixed length of 2600 ms, with a preparation (cue-target) interval of 600 ms. 70% of trials were completed trials, on which the cue was followed by a target requiring the correct response; half of these trials required a go response and half a nogo response. On completed trials, the cue was displayed for 134 ms, then was removed (blank screen) for 466 ms, then the target was displayed for 117 ms, followed by a blank screen for the remaining 1883 ms. 30% of trials were cue-only trials (never occurring consecutively), on which the cue (displayed for 134 ms) was followed by a blank screen for the remainder of the trial 2466 ms. No response was required on any cue-only trial. The first task in the first block was always the letter task and after that, blocks alternated with respect to the starting task. Following the initial trial (always run-position 1), there were 60 trials (beginning with run-position 2 and ending with run-position 1) and finally a set of between zero and five “dummy” trials. Dummy trials were included so that the block could end on any trial within a run (so as to prevent participants altering their behaviour on a particular run position when they felt the end of a block was likely). N.B. Due to a programming error, the feedback was unreliable on the dummy trials in Expt. 1 (i.e., sometimes an incorrect response would not receive feedback, or a correct response would receive erroneous feedback that the wrong response had been made). Due to this error, we do not recommend analysing response data from these trials. Different participants completed different numbers of blocks in sessions 1 & 2 (respectively): 16 & 20, 16 & 23, and 19 & 23. EXPERIMENT 2: The key features of this experiment were: parity & magnitude tasks; left/right responses; and all trials being completed (no cue-only trials). Cue-type, preparation interval and variability of preparation interval were manipulated (although note that cue-type was confounded with session-half - see below). Target stimuli in this experiment were a single digit (1 to 9, excluding 5). The two tasks were to judge either its parity (odd/even, with left/right index-finger responses respectively) or its magnitude (lower/higher than 5, left/right responses respectively). Three task cues were used for each task; the exact cues used are shown in the DataCollectionReadMe.pdf document (and the image cue files are uploaded with the collection).
Image cues were used for the first half of the session and text cues for the second half of the session. Image cues were designed to replicate a clock, with the filled segment appearing to rotate around from one trial to the next (see Monsell, Sumner & Waters, 2003, Memory & Cognition 31:327-342). Segments to the top-right of the diagonal line cued the magnitude task, those to the bottom-left the parity task. Text cues were designed to indicate the response-mapping for the current task by showing e.g. “odd” on the left of the screen (see Mayr & Kliegl, 2000, Journal of Experimental Psychology: Learning, Memory, and Cognition 26:1124-1140). Different styles and different missing letters were used for run positions 1 – 3 to minimise repetition of perceptual information between cues within a run. On each trial, the task cue was shown for 200 ms, then a central fixation dot for either 100 ms or 800 ms (preparation interval was therefore either 300 ms or 1000 ms in total), then the target digit was shown until the response was made, after which a blank screen was shown for 100 ms. Preparation interval was manipulated either between blocks (alternating) or between trials; each participant was assigned to either the between-blocks or the between-trials condition. EXPERIMENT 3: The key features of this experiment were: parity & magnitude tasks with image cues (those used in Expt. 2); left/right responses; cue-only trials were included; and preparation interval varied between trials. Trials could be one of the following three types: short-preparation completed (7/17 trials); long-preparation completed (7/17 trials); long-preparation cue-only (3/17 trials). These trials were intermixed such that cue-only trials never occurred consecutively. Within-trial timings corresponded to the group in Experiment 2 for whom preparation varied between trials, except that in Experiment 3, targets did not remain on-screen until response; instead, targets were displayed for 120 ms and then replaced with a central fixation dot until the response was detected. The target stimuli and image cues were as in Experiment 2, except that half the participants were now assigned the reversed cue-task mapping (upper right = parity; lower left = magnitude). Blocks comprised an initial first trial (run-position 1), then 60 experimental trials, plus a final 0-5 dummy trials as in Experiment 1 (but without the feedback error, so all are usable). At the end of a block, if the total number of errors was more than five, additional text reminded them to be “fast AND ACCURATE!”. There were 25 experimental blocks, across which the first task alternated between magnitude and parity. |
| Observation unit: |
Individual |
| Kind of data: |
Numeric, Text |
| Type of data: |
Experimental data
|
| Resource language: |
English |
|
| Data sourcing, processing and preparation: |
Data were collected using the experimental software E-Prime. E-Merge was used to compile a single file including all participants, and data were exported from E-DataAid into Excel. Demographic variables were removed and participant numbers were changed for anonymity. Columns of data that were collected automatically by the experimental software but that are not relevant for data analysis were removed. Misleading values were removed: in Expt. 1, on cue-only trials where no target was presented, target-relevant variables showed the value from the most recent completed trial instead (e.g. RT); these were removed in Excel. Columns were added to include variables required for data analysis; the values in these columns were computed from the information already present in other columns using Excel, and these included the following. In Expt.1: response accuracy for the current and preceding two trials; the preceding trial-completion status and the go/nogo/cue-only status of the current and preceding two trials. In Expts. 2 & 3: response accuracy for the preceding trial, and cue-target interval for the preceding trial. The file was then saved as a CSV file.
|
| Rights owners: |
|
| Contact: |
|
| Notes on access: |
The Data Collection is available to any user without the requirement for registration for download/access.
|
| Publisher: |
UK Data Service
|
| Last modified: |
09 Jan 2023 12:49
|
|
Available Files
Data and documentation bundle
Data
Documentation
Edit item (login required)
 |
Edit Item |
