IFIS and E-Prime can be a bit murky. This page clarifies some of the more opaque parameters in each and suggests helpful E-basic snippets and experimental structures.
IFIS
Default minimum continue time
Controversy has raged for years about the true meaning of the Default Minimum Continue Time parameter, specified in IFIS Defaults. Though its essential being [Da-sein] is no more accessible to us now than it was several years ago, we can use induction to draw conclusions about what Walt might have intended this to mean. Lab lore counsels that one should set this parameter to equal the length, in milliseconds, of one's longest trial. Whence this perfidy arose is not certain, but experimental observation on the part of Mike Schlund and Jeff Phillips suggests that, in a paradigm with varied trial length, setting the Default Minimum Continue Time so long can cause your paradigm to end before finishing all of the requested trials.
The purpose of this parameter seems to be to prevent IFIS from ending a run in the middle of a trial: it indicates how much clock time must be left in a run in order for IFIS to continue to the next trial. Imagine the following situation, and you can see why a large value for this parameter spells disaster:
- The Default Minimum Continue Time is set to 15000. Your paradigm has trials of varied length--say 7500, 9000, 12000, and 15000 msec--which are sampled randomly.
Each run of your paradigm has 30 trials, and you are currently approaching the end of the first run. The IFIS clock has 21 seconds left, and random sampling in this block has left one 9000 ms trial and one 12000 ms trial to complete.
The paradigm presents the 9000 ms trial, then checks the IFIS clock. Only 12000 ms remain; this value is less than the Default Minimum Continue Time, so IFIS exits the paradigm without completing it.
Note that the experimenter made no mistake in specifying the number of frames per run, in structuring the block and trial lists, or in setting the sampling method. The fault lies entirely with the Default Minimum Continue Time. While a safeguard to prevent IFIS from quitting in the middle of a trial seems desirable, this parameter does not seem to jibe with a paradigm having various, randomly sampled trial lengths. In such a case, try setting the variable to 1 ms--or to some value which is less than that of your shortest trial. Keep in mind that timing errors might sum up to be a few hundred milliseconds per run; for many experiments, this error is not critical, and it will not hurt terribly if the last trial in a run is cut short by this amount.
Don't be a freeloader! Contribute your own invective to this column.