Eskimo3 Specs
Stimuli are Faces; Responses are Names; same materials as used in ForgSpace1.
Overview
In session 1, subject sees each face-name combination once (Study 1) and is tested on it twice (Test1 and Test2). There will be 300 trials in session 1. The design is 3 X 2. This is 3 possible gaps between Test1 and Test2 X feedback on test 2 vs. no feedback on test 2. Different face-name pairs are assigned randomly to one of the six resulting conditions. The only trick here is fitting all the pairs into a sequence like a jigsaw; the problem should be approached as in the ForgSpace series.
They won’t all fit in. If they did, there would be 100 pairs presented 3 times, and 100/6 or ~ 16 items in each condition. I figure we’ll be lucky if we get ten per condition per subject.
For that reason, we will end up with slightly different numbers of items in the different conditions simply because the program will be able to fit a few more of one condition into the sequence as compared to another condition.
In Session 2, the subject returns and gets a test without feedback on all of the items
Programming Strategy:
At Runtime but Prior to Presenting Stimuli
Basically what you want to do here is randomly assign all the face-name pairs to conditions individually for that subject, then fill in the jigsaw in advance of starting the presentations, and finally present the stimuli and collect and save the data.
First, take the face-name pairs and shuffle the list (every face can stay stuck to the same name across subjects, so that if subject 1 sees a particular face with the name “Joe”, subject 2 will also see that face paired with Joe) but shuffle the whole list at the beginning like a deck of cards. This will permit us to randomly assign face-name pairs to the six conditions.
Now (to put it in Pascal speak) create a 3-dimensional array
Faceindex[gap, feedback, n]
to assign face-name combinations to each of the 6 conditions.
gap takes
the values 2, 16, and 64
Feedback takes the value Yes and No.
n takes the values 1, …, 20
and x = index of the face in the shuffled list].
If we treat gap and feedback as numeric indices ranging from 1..3 and 1..2, respectively, in pascal you would write:
For gap=1to3;
For feedback=1to 2;
For x = 1 to 20;
FaceIndex[gap,feedback,x] =(gap-1)*60 + (feedback-1)*20 + x
End;
End;
End;
So now you have randomly assigned 20 face/name pairs to each of the six conditions.
Next you can start filling in the jigsaw puzzle. By that I mean you determine which 3 trials a given face-name pair will appear in. It is important that you do this in a way that doesn’t end up having items in one condition tending to appear earlier or later in the list than items from another condition. That means you can’t just fill in all the items in one condition, then all the items in the next condition, etc. The approach I recommend is to put them in six at a time (one for each of the six conditions, randomizing the order of the six you’re putting in). In addition, flip a coin to determine whether the filling-in starts from the beginning or the end of the list of trials.
Continue this until you can’t put any more items in any of the conditions. Set a flag when you find yourself unable to fit in an item from a particular condition, and quit trying for that condition, but don’t quit trying for the other conditions, because it may still be possible to put a few more in from one condition even after it’s impossible to fit any more in from another condition.
Once this is complete you will still have some blank (unassigned) trials where nothing has been scheduled. Just make these study trials: go to the face/name items that you haven’t used and cram those into the remaining empty spaces in the list.
Presenting Materials
and Collecting Data.
This can all work just like ForgSpace1. On Study Trial, present the face together with the name for 5 seconds. On Test Trials 1 and 2, present the face alone and provide a text box for the subject to type in the name (while still looking at the face). The subject should be able to hit enter when they’re done with that. If they don’t know, they can just leave it blank and hit enter. Feedback should consist of the windows job well done tone if they got it right, and the correct name for 4 seconds if they did not get it right.
The variable Feedback (yes or no) determines whether the subject gets feedback on Test 2. (On test 1 there is always feedback.)
Data Files
Write out two data files after Session 1 is over:
1. R-n. This contains one row per trial, and includes everything that
happened on that trial. What stimuli were presented, a presentation or test?; what the subject typed (if they don’t type anything and hit enter, write “999NoResponse”), feedback provided?
2. PerfRec-n. This contains a slightly more abstract representation of
what happened, focused on the faces rather than the trials (and logically speaking redundant to R-n). This should have one row per face presented or tested in the experiment. Columns in this file:
<face, correct response, trial of Study 1, trial of Test 1, trial of Test 2, were they correct in T1?; were they
correct in T2?>
Test Session
On the test session, the program should read PerfRec-n and construct a test for that subject. Every face that was presented 3 times on Session 1 should be tested in Session 2, and the subject should be forced to guess even if s/he doesn’t know the answer. Write test performance (correct/incorrect) and what the subject said back into PerfRec-n.
Then write a summary data file called S-n containing the six conditions, the number of words tested in that condition, the number of correct responses in that condition, and the percent correct in that condition. Six columns in this: <Subject, Gap, Feedback?, number of trials tested, number of trials correct, % correct>
So at the end there will be 3 files: R-n containing a record of Session 1, PerfRec-n containing a list of all the stimuli presented and describing what happened with these stimuli in both Session 1 and the test session, and S-n describing the overall performance on the test session as a function of condition.