Script Author: Katja Borchert, Ph.D. (katjab@millisecond.com), Millisecond
Created: April 08, 2014
Last Modified: October 02, 2023 by K. Borchert (katjab@millisecond.com), Millisecond
Script Copyright © Millisecond Software, LLC
This script implements a Probabilistic Selection Task, a learning-transfer task that uses different "win" probabilities of symbols presented in pairs during training to investigate participant's selections of 'winning' symbols when confronted with new symbol pairings during a transfer task.
Frank, M.J., Seeberger, L.C. & O’Reilly. R.C. (2004). By Carrot or by Stick: Cognitive Reinforcement Learning in Parkinsonism. Science, 306, 1940-1943.
Doll, B.B., Hutchison, K.E. & Frank, M.J. (2011). Dopaminergic Genes Predict Individual Differences in Susceptibility to Confirmation Bias. The Journal of Neuroscience, April 20, 2011 • 31(16):6188–6198.
Millisecond thanks J. Bianchi for collaborating on this script!
Practice Block: ~3 minutes per practice block (repeated if learning criteria are not met) Test Block: ~5 minutes
Participants work through a learning and a test phase.
Learning Phase: participants get presented with 3 symbol pairs (AB, CD, EF) and have to pick the "winning symbol" of each pair. The win-probabilities of each symbol are controlled in such a way that A wins over B in 80% of the times; C wins over D in 70% of the times, and E wins over F in 60% of the times. The learning phase gets repeated if participants don't select the winning symbol with a pre-determined frequency.
Test Phase: by default, this scripts presents all possible pair combinations (15) during the test phase
This script provides the possibility to include an instruction manipulation by telling participants that one symbol (always the symbol being assigned to be symbol F) is either the symbol with the 'highest' or 'lowest' probability of winning (see Doll et al, 2011). To run a specific manipulation, go to section Editable Parameters and set parameters.instructionManipulationCondition accordingly. By default, this script runs the Probabilistic Selection Task without any instruction manipulation (see parameters.instructionManipulationCondition)
Letter-Symbol-Assignment: each of the 6 symbols gets randomly assigned to 'play the role' of letters A-F
Learning Phase:
• each learning block runs for 60 trials (20 per symbol pair -> can be edited via parameters.trialnumberLearning)
• for any given learning block the computer checks whether all learning criteria have been met
during the given learning block
-> learning criteria: symbol A is selected at least 65% of the times, symbol C is selected at least
60% of the times, symbol E is selected at least 50% of the times (unless the script is running within the
InstructionManipulation Framework in which case no restrictions are placed on pair EF)
If those criteria haven't been met in the block, a new learning block starts
Test Phase:
• the test block runs for 150 trials (10 trials for each of the 15 possible symbol pairs)
• by default, this script runs all possible symbols pairs (see list.trialselectionTest) in random order
Trial Set-up:
The left-right position of the symbol pairs is counterbalanced across each block
-> symbol A appears on the left side in 50% of the trials (same for all symbols)
The stimuli are the 6 Hiragana characters presented in Frank et al (2004).
Their size and location can be easily controlled via EDITABLE CODE -> Editable Parameters
To use a different set of images, simply replace the stimuli under
section Editable Stimuli -> item.symbols
Two additional Hiragana symbols were selected for example symbols by Millisecond.
Instructions are not original to the task. They are provided by Millisecond
as htm/html pages and can be edited by changing the provided htm/html files.
To edit htm/html-files: open the respective documents in simple Text Editors such as TextEdit (Mac)
or Notepad (Windows).
File Name: probabilisticselectiontask_summary*.iqdat
| Name | Description |
|---|---|
| inquisit.version | Inquisit version number |
| computer.platform | Device platform: win | mac |ios | android |
| startDate | Date the session was run |
| startTime | Time the session was run |
| subjectId | Participant ID |
| groupId | Group number |
| sessionId | Session number |
| elapsedTime | Session duration in ms |
| completed | 0 = Test was not completed 1 = Test was completed |
| minAAB | Proportion of A-selections in AB pairs that have to be met in a learning block (default: 0.65) |
| minCCD | Proportion of C-selections in CD pairs that have to be met in a learning block (default: 0.6) |
| minEEF | Proportion of E-selections in EF pairs that have to be met in a learning block (default: 0.5) !!!! if you use the task within the Instruction Manipulation Framework, (parameter) minE_EF is automatically adjusted to 0 a- |
| f | Contain the symbol itemnumbers assigned to letters A-F (randomly assigned during runtime) Letter A: highest win chance (during training wins with 80% probability) Letter B: lowest win chance (during training wins with 20% probability) |
| countTrainingBlocks | Counts the number of training blocks run |
| trainingPass | 0 = the training criteria were not met (performance should be flagged) 1 = the training criteria were met |
Training Performance |
|
| probAABTraining | Proportion A-selections in AB training pairs |
| probCCDTraining | Proportion C-selections in CD training pairs |
| probeEFTraining | Proportion E-selections in EF training pairs |
Test Performance |
|
| probAABTest | Proportion A-selections in ABTest pairs |
| probAACTest | Proportion A-selections in ACTest pairs |
| probAADTest | Proportion A-selections in ADTest pairs |
| probAAETest | Proportion A-selections in AETest pairs |
| probAAFTest | Proportion A-selections in AFTest pairs |
| probBBCTest | Proportion B-selections in BCTest pairs |
| probBBDTest | Proportion B-selections in BDTest pairs |
| probBBETest | Proportion B-selections in BETest pairs |
| probBBFTest | Proportion B-selections in BFTest pairs |
| probCCDTest | Proportion C-selections in CDTest pairs |
| probCCETest | Proportion C-selections in CETest pairs |
| probCCFTest | Proportion C-selections in CFTest pairs |
| probDDETest | Proportion D-selections in DETest pairs |
| probDDFTest | Proportion D-selections in DFTest pairs |
| probeEFTest | Proportion E-selections in EFTest pairs |
File Name: probabilisticselectiontask_raw*.iqdat
| Name | Description |
|---|---|
| build | Inquisit version number |
| computer.platform | Device platform: win | mac |ios | android |
| date | Date the session was run |
| time | Time the session was run |
| subject | Participant ID |
| group | Group number |
| session | Session number |
| minAAB | Proportion of A-selections in AB pairs that have to be met in a learning block (default: 0.65) |
| minCCD | Proportion of C-selections in CD pairs that have to be met in a learning block (default: 0.6) |
| minEEF | Proportion of E-selections in EF pairs that have to be met in a learning block (default: 0.5) !!!! if you use the task within the Instruction Manipulation Framework, (parameter) minE_EF is automatically adjusted to 0 |
| blockCode | Name of the current block |
| blockNum | Number of the current block |
| trialCode | Name of the current trial |
| trialNum | Number of the current trial |
| stimulusitem.1/stimulusitem.2 | The presented stimuli in order of trial presentation does not need to correspond to spatial presentation of stimuli |
| winLetter | Contains the winning letter during learning |
| response | The scancode of the participant's response key 30 = A = left 38 = L = right 57 = spacebar |
| selectedLetter | Contains the letter selected by participant during training and test phase |
| correct | The correctness of the response (1 = correct; 0 = otherwise) |
| latency | The response latency in ms; measured from onset of stimulus probAABTraining- |
| probCCDTraining | Calculates the probentage A/C/E selections for the training pairs (current training block) a- |
| f | Contain the symbol itemnumbers assigned to letters A-F (randomly assigned during runtime) Letter A: highest win chance (during training wins with 80% probability) Letter B: lowest win chance (during training wins with 20% probability) |
| countTrainingBlocks | Counts the number of training blocks run |
| trainingPass | 0 = the training criteria were not met (performance should be flagged) 1 = the training criteria were met |
The procedure can be adjusted by setting the following parameters.
| Name | Description | Default |
|---|---|---|
| instructionManipulationCondition | 0 = no manipulation; parameters.minE_EF runs as set Manipulation Conditions, parameters.minE_EF will automatically be set to 0 1 = participants are told that one symbol (symbol F) is the highest winning symbol 2 = participants are told that one symbol (symbol F) is the lowest winning symbol 3 = control group (no manipulation presented but parameters.minE_EF will automatically be set to 0) (see Doll et al, 2011, for manipulation explanations) | |
| learningCriteria | To remove a criterion, set the corresponding value to 0. | |
| minAAB | Proportion of A-selections in AB pairs that have to be met in a learning block | 0.65 |
| minCCD | Proportion of C-selections in CD pairs that have to be met in a learning block | 0.6 |
| minEEF | Proportion of E-selections in EF pairs that have to be met in a learning block !!!! if you use the task within the Instruction Manipulation Framework, parameters.minEEF is automatically adjusted to 0 | 0.5 |
| maxNumberTrainingBlocks | The maximum number training blocks are run before participants proceed with the test. Performance of participants who fail the last training block will be flagged in the data file. | |
| responseKeyLeft | Response key for symbol on the left | "A" |
| responseKeyRight | Response key for symbol on the right | "L" |
| readyDuration | The duration (in ms) of the get-ready trial | 3000ms |
| xRight | Horizontal coordinate of the right symbol in screen probentages | 65% |
| xLeft | Horizontal coordinate of the left symbol in screen probentages | 35% |
| symbolHeight | The height of the symbol image in screen height probentages | 40% |
| trialNumberLearning | The number of trials in a learning session the minimum number for the default win probabilities and the balancing of left-right symbol presentation is 10 trials per symbol pair the learning block gets repeated until specific learning criteria are met (see above) within a block | 60 -> 20 per symbol pair: AB, CD, EF |
| trialNumberTest | The number of trials in the test session !!! by default, all 15 possible combinations of trials are run by this script for the test phase. To remove test combinations that are of little interest, go to list.trialselectionTest under section EDITABLE LISTS | 150 -> 10 per symbol pair |
| feedbackDuration | The duration of the feedback stimuli | 1000ms |
| winAAB | The winning proportion of A in AB pairs | 0.8 |
| winCCD | The winning proportion of C in CD pairs | 0.7 |
| winEEF | The winning proportion of F in EF pairs | 0.6 |