[solved] repeat cycle

Comments

• eduard

  April 2015 edited 8:59PM

  Hi Andrea,

  Is your experiment random on trial level or only on condition level? I mean, are the 3 trials of one conditions always presented after each other, or are all 21 trials presented in a mixed fashion in the initial round?

  If they are blocked, you do something like this:

  1) In a loop item, select the condition you're going to run
  2) Within this loop, you put an inline_script running each trial of a condition in a while loop. So, while accuracy != 100: run_trial, or something similar

  Was this helpful?

  Let us know if you need more detail.

  Eduard

  
• andrea

  April 2015 edited 8:59PM

  Hi Eduard,

  My experiment is random on a trial level, so that my 21 trials (3 for each condition) are presented totally random. I have only one loop (and the same sequence) for all the trials.

  For example, I need to pick up only three out of the 21 trials of the original loop.

  Could your suggestion work also in this case?
  Anyway, I'll try and I'll let you know!

  Thank you so much!
• andrea

  April 2015 edited 8:59PM

  Hi Eduard,

  I've tried your suggestion but it doesn't work: using the while loop the experiment presents only the first trial of the first condition to infinity.

  How can I do?

  Thanks,

  Andrea
• eduard

  April 2015 edited April 2015

  Hi Andrea,

  My experiment is random on a trial level, so that my 21 trials (3 for each condition) are presented totally random.

  This makes things a bit more complicated, but it still is doable. Here a (very general) way of doing it (sorry for the lack of detail).

  1) First you ran your initial loop and sequence in a rather basic way (all 21 one trials randomly). It is super important that you keep track of the accuracy of each condition. For example you can use a dictionary, with the conditions as keys and the number of total trials, accurate trials and average accuracy as values (Don't forget that you would have to initialize the dictionary somewhat earlier in your script.

  # intialization a dict with a list of [correct trial, total trials, accuracy]
  accuracies = dict(cond1 = [0,0,0],cond2= [0,0,0],...,cond7=[0,0,0]) 
  
  # later in your loop
  # assuming 'cond' has a value that equals one of the keys in a dict, e.g. cond1
  
  accuracies[cond][1] += 1 # updating trial count
  if correct:    accuracies[cond][0] += 1 # updating correct count (NOT PROPER CODE!)
  # average accuracy. Watch out to do a proper division (don't divide integers!)
  accuracies[cond][2] = float(accuracies[cond][0]) / accuracies[cond][1] 
  

  It might also be handy to keep track of other trial information, like which stimuli were drawn, etc.

  2) Now the tricky part begins. You have to decide which of the conditions need a repetition and create a new loop on the fly. I recommend doing it in an inline_script, because you have more freedom there. The idea is to select every condition that needs a redo and put its trials into the new pool. This procedure should then be repeated until the pool is empty, so every condition has 100% accuracy

  import random
  
  while True:
      # trials will contain all trials that need to be run again
      trials = []
      # conditions is a list with all your conditions
      for cond in conditions:
          if accuracies[cond][2] != 100:
              trials.append([list_with_all_trial_information])
  
      # now trials contain all information about trials to be repeated
      # shuffle it
      random.shuffle(trials)
  
      # and present the stimuli
      # this depends on your design. Check out the canvas function documentation
      # if you're unsure about how to do it
  
      for trial in trials:
          run(trials) # or whatever
          # while running to update the accuracy dictionary accordingly. Otherwise
          # you won't change anything.
  
      # finally you have to include a possibility to break the loop
  
       if all_accuracies == 100:
           break
  
  

  Does this help?

  Good luck,

  Eduard

  
• andrea

  April 2015 edited 8:59PM

  Hi Eduard,

  thank you so much for your help. I'm working on it but I have some troubles in appending the trial: which information do I need to put in this line?

  trials.append([list_with_all_trial_information]) 
  

  Thanks again,

  Andrea
• eduard

  April 2015 edited 8:59PM

  Hi Andrea,

  Every information that you need to show your stimuli. This can be anything, but boils down to What, How, When and Where. What your goal is in the second loop (the one in which you repeat the conditions that haven't 100% correct, is to show the same trials again, right? So, you have to make sure that you know what to present.

  Does this make sense?

  Eduard

  
• andrea

  April 2015 edited 8:59PM

  Hi Eduard,

  thanks to your suggestion I have figured out that problem, so that I have created a new list that the include all the trials I need to repeat.

  However I can't display the new trials; I've tried with

  for trial in trials
       exp.items['sequence'].prepare()
       exp.items['sequence'].run()
  

  but it doesn't work. The sequence is exactly the same of the previous loop (basically I have created a second loop with the same sequence and at the beginning of this new sequence I have inserted the inline_script for selecting only the trials I need)....is it right? Or I need to specify the entire sequence in the inline_script?

  Moreover, I'm not sure of the right position of the inline_script....

  Thanks again,

  Andrea
• andrea

  April 2015 edited 8:59PM

  Hi Eduard,

  sorry for bother you again (I'm quite new of Opensesame and Phyton code in particular)!
  I've tried in a second way to do it: instead of randomize the 21 trials, I've randomized only the condition level.
  I've created one big loop that contains 7 small loops (one for each condition) of 3 trials each. I've used an inline code to keep track of the accuracies as you suggested.
  At the end of the big loop I've inserted a repeat_cycle plug-in that re-runs the big loop only if one of the accuracies is not 100% (and re-runs only the small loop in which the accuracy is less than 100%).

  Everything works fine but there is only one problem: when all the accuracies are 100% (and so the repeat cycle should stop) the experiment remains freeze on the last screen of the last trial, and does not continue to the next phase. The only way I can stop the experiment is to press the esc button.

  How can I solve this? Is it an error in the repeat_cycle plug-in?

  Thank you so much again,
  Andrea
• eduard

  May 2015 edited 8:59PM

  Hi Andrea,

  Sorry for not having responded earlier.

  Your problem sounds, that you haven't defined a break criterion. Would you mind sharing your experiment with us, or at least your settings for the repeat_cycle plugin? Without more information it's hard to offer a solution.

  Thanks,

  Eduard

  
• andrea

  May 2015 edited 8:59PM

  Hi Eduard,

  No problem!!! I figured it out! It was a problem in the break criterion.
  I've replaced the repeat_cycle plug-in with an inline_script in which for each condition I've used a while:

  while self.get('accuracy condition 1') != 3:
      exp.items['condition1'].prepare()
      exp.items['condition1'].run()
  

  Thank you so much!!

  Andrea
• eduard

  May 2015 edited 8:59PM

  Good to hear.

  Eduard