[open] random trial order but distributed catch trials w/out repetition

Comments

• HannesBln

  June 2014 edited June 2014

  To clarify what my experiment looks like, this is the general structure:

  - experiment (sequence)
     *startscreen (scatchpad)
     *run_loop (loop)
        -block_sequence (sequence)
           *trials (loop)
              -trial_sequence (sequence)
                 *check_catch_break (inline script)
                 *question (scatchpad)
                 *jitter (inline script)
                 *reply (scatchpad)
                 *mouse response
                 *logger
                 *catch_log (inline script)
           *thankyou (feedback)
           *question_strategy (text_input)
           *logger
     *end (scatchpad)`
  

  I now tried the following:
  I added a variable "catch_log (0)" to the run_loop in order to keep track of whether the trial before was a catch trial and a variable "catch_twice (0)" which indicates if two catch trials are following each other.
  The check_catch_break inline script is set to run if "always" and contains the following in the prepare phase:

  if exp.get("catch_log") and exp.get("catch"):   #if before and now are catch trials
      exp.set("catch_twice", 1)                               
      exp.items['trial_sequence'].count = exp.items['trial_sequence'].count - 1   
  else:
      exp.set("catch_twice", 0)
      exp.set("catch_log", exp.get("catch"))
      #check if its time for a break [...]`
  

  I was hoping that this will have the effect that if a catch trial is followed by an other one opensesame would go back one step and randomly choose again which trial is coming up, until it is not a catch trial again. Therefore I set all other items in the trial_sequence to run if "not [catch_twice]".

  It seems to work so far that now at least it does not happen that two catch trials are right after each other anymore. It is not the perfect solution, since they are not evenly distributed. But I do not want to do the pseudo randomisation by hand with an inline script since I already added all the stimulus images and other variables in the loop items variable cells which total up to 3828 variables.
  But there is one problem left: It looks like that the catch trials that are skipped because they follow a previous catch trial are lost and do not come up during the entire experiment.

  Any idea how to prevent that? I hope my solution of decrementing the internal item count does not have any other bad consequences, since I am not yet understanding opensesame's underlying internal structure of the items and phases.

  I would appreciate any ideas and comments on this,

  cheers
  Hannes
• sebastiaan

  June 2014 edited 6:49PM

  Hi Hannes,

  One way to do this, a bit easier than what you're doing right now, is to add a catch trial (let's call it catch_sequence) to your trial_sequence, and use a run-if statement so that catch_sequence is only executed once every 24 trials.

  So let's say that you have a structure like this:

  

  Then you could use the following Python-style run-if statement so that it's only executed after every 24 trials:

  =(self.get('count_trial_sequence')+1) % 24 == 0
  

  For the logic behind this statement (although in a slightly different context), see:

  • http://osdoc.cogsci.nl/rovereto2014/#extra-medium-add-breaks-and-per-block-feedback

  Of course, you can have catch_sequence do whatever you want, including running five catch trials in a row.

  Does this help?

  Cheers,
  Sebastiaan

  
  
  Check out SigmundAI.eu for our OpenSesame AI assistant!