Probabilistic Reversal learning

From PEBL WIKI
Jump to: navigation, search
A screenshot of PEBL's Probabilistic Reversal Learning task.

Contents

[edit] About

Task involves learning a 'rule' which determine which of two abstract stimuli are selected. The rule changes in the middle of the block, and the goal is to see how quickly the participant can adapt to the rule change. Feedback is probabilistic.

[edit] Notes

Located in probrev/probrev.pbl

[edit] Options

One can control the blocksize, stage size, distribution of probabilistic errors, and substitute new images for the stimuli. Variables that can be changed, and their default values:

numblocks <- 4    # Each block is around 150 trials.
numStages <- 10   #How many learning stages per block?
stageMin  <- 10   # Size of smallest stage
stageMax  <- 15   # Size of largest stage
pErrorMin <- 0    # Smallest number of errors in a stage
pErrorMax <- 5    # Largest number of errors in a stage

[edit] Data Output

Data is saved to probrev-<gsubnum>.txt. Columns are:


sub,block,stage,totaltrial,trial,order,valid,corrfeedback,acc,corr,rt
111,1,1,1,1,2,1,0,0,0,876
111,1,1,2,2,2,1,0,0,0,1004
111,1,1,3,3,2,1,1,1,1,665
111,1,1,4,4,2,1,1,2,1,617
111,1,1,5,5,1,1,1,3,1,1008
111,1,1,6,6,1,1,1,4,1,423
111,1,1,7,7,2,1,1,5,1,408
111,1,1,8,8,1,1,1,6,1,453
111,1,1,9,9,1,1,1,7,1,406
111,1,1,10,10,2,1,0,7,0,519
111,1,1,11,11,1,1,1,8,1,581
111,1,1,12,12,2,1,1,9,1,648
111,1,1,13,13,1,1,1,10,1,496
111,1,1,14,14,1,1,1,11,1,439
111,1,2,15,1,1,1,0,0,0,623
111,1,2,16,2,2,1,0,0,0,503
111,1,2,17,3,2,1,1,1,1,398
111,1,2,18,4,1,1,1,2,1,520

Here,

  • totaltrial and trial are just trial counters--trial starts over on each stage.
  • order indicates which side the stimulus appears on
  • valid indicates whether valid feedback was given. The trials in which invalid feedback is given are selected at the beginning of each stage. But *invalid feedback is only given after correct responses.
  • corrfeedback is the actual feedback given (smiley vs frowney) indicating whether they were told they were correct or incorrect.
  • acc is a counter indicating the running total of correct responses--resetting on each stage
  • corr is whether a trial was correct
  • rt is how long (in ms) the response took.

In addition, after data collection is complete a report file is generated. This basic report is saved to probrev-report-<gsubnum>.txt:


---------------------------------------------------
 Report for PEBL Probalistic Reversal Learning Task
 Version 0.3
 PEBL Version 0.14
 Mon Jan 11 14:13:17 2016
 Participant Code: test1
 http://pebl.sf.net
---------------------------------------------------
Statistic                               Value 
---------------------------------------------------
Number of Trials:       		172
Correct Trials:         		133
Blocks:                 		1
Reversal stages/block   		10
Number of reversals     		10
Mean Correct Trials/stage		13.3
Mean Total Trials/stage		        17.2
Mean Errors per reversal		3.9
---------------------------------------------------

[edit] References

Cools, R, Clark, L., Owen, A. M., Robbins, T. W. (2002) Defining the Neural Mechanisms of Probabilistic Reversal Learning Using Event-Related Functional Magnetic Resonance Imaging.

[edit] See Also

PEBL Test Battery

Personal tools
Namespaces

Variants
Actions
Navigation
Toolbox