Habituation phases are generally set up as a trial (or set of trials) occurring within a loop. CRITERIONMET is a special loop terminating condition for habituation that is based on looking-time reduction from a basis window. How the basis window is chosen, and what reduction in looking time it takes to consider the child to be habituated, are defined at the beginning of a protocol as habituation criteria. The CRITERIONMET terminating condition, whenever it is evaluated, checks whether these habituation conditions have been satisfied.
Below is an example of the basic structure of a habituation protocol.
Using the habituation settings above, here is an example of how the basis window and criterion would be calculated by BITTSy across trials.
In this example, note that the target criterion window time (i.e. the maximum time a participant can look during a window and be considered to have met habituation) can change when a new basis window is identified, when BASISCHOSEN is set to LONGEST. The first window in which the child's total looking time is less than the current target criterion time is the window of trials 5+6. After trial 6, when this window is evaluated, CRITERIONMET is true, the loop terminating condition is met, and the loop will no longer repeat. The habituation phase will end, and execution will move on to the next phase of the protocol.
It is important to note that, like all loop terminating conditions, CRITERIONMET is only checked when the loop step is reached - that is, after each time the contents of the loop have been fully executed (see loop terminating conditions for an expanded explanation.) This means that if the loop contained two trial starts and trial ends, CRITERIONMET would only be evaluated after even-numbered trials. This does not prevent BITTSy from evaluating and identifying criterion windows based on the habituation settings in the protocol. But it does mean, in this case, if a criterion window was identified that ended with an odd-numbered trial, one additional habituation trial would be run before the loop would end. For this reason, it is generally recommended to only define one trial within a habituation loop, and if stimuli vary across trials, to define their cycling via selection from groups. (See the example protocols page for examples of setting up simple and more complex habituation phases.)
Many times, you want not only a terminating condition for when the participant habituates, but also a point at which the phase would end even if habituation is not reached. For example, you might want to end a phase either when the child habituates, or after 20 trials, whichever comes first.
In this case, you essentially set up two potential endings for the phase:
In a loop clause, if you have two different UNTIL statements, with a carriage return between them, they are treated as if they are linked by an OR condition (see this section for all available options). This would mean it would end either if the criterion was met or if it had already looped through 19 times (after the first one, so UNTIL 19 TIMES means there would be 20 times total).
Whichever terminating condition is met first, execution will progress to the next step: it is not possible to skip to a different step or phase based on which condition is met. Generally, this means that the post-habituation test phase of a typical habituation experiment would be shown to all participants, whether or not they habituated. When generating a habituation report from the event log of a test session, it is marked whether the participant habituated (i.e. met the CRITERIONMET condition rather than the alternate one) so that participants who did not habituate can be recorded as such, and excluded from analyses of looking time in the test phase.
All habituation settings are defined in the same section of the protocol as optional experimental settings such as key assignments. These settings are designed to allow experimenters to define and control habituation experiments in BITTSy similarly to established habituation procedures.
This defines the number of trials which constitute a window. The same size is used when evaluating potential basis windows and criterion windows.
There is no default value for window size. This setting must be specified in the protocol of any habituation procedure that uses basis and criterion windows.
This setting defines whether windows for evaluation as possible basis/criterion windows should overlap with each other. SLIDING allows overlapping windows, while FIXED does not. For example, with a WINDOWSIZE of 3 and the window type FIXED, trial sets 1-3, 4-6, 7-9, etc. will be evaluated; if SLIDING, trial sets 1-3, 2-4, 3-5, 4-6, 5-7, etc. will all be evaluated.
SLIDING is the default setting if the WINDOWTYPE is not defined within a protocol.
This setting defines whether the basis window and a criterion window are allowed to overlap. This setting is important whenever WINDOWTYPE is SLIDING; it is irrelevant for FIXED windows.
If this setting is YES to allow overlap, and the WINDOWSIZE was 4, you could have a basis window of trials 1-4 and trials 2-5, 3-6, 4-7, etc. would all be evaluated as possible criterion windows. But if WINDOWOVERLAP was set as NO, the first window to be evaluated as a potential criterion window would be trials 5-8. All prior windows would be disallowed because they would include trial 4, which was part of the basis window.
YES is the default if WINDOWOVERLAP is not defined in the protocol.
Defines which window is chosen as the basis window. This can either be the first window or the window with the longest total looking time.
LONGEST is the default if BASISCHOSEN is not specified.
This setting allows you to set a minimum total looking time for a basis window. This can help prevent a child who is fussy at the start of an experiment from reaching a habituation criterion without having met a minimal requirement for looking time.
Zero milliseconds is the default, i.e. no minimum.
This setting specifies the looking time reduction compared to the basis window that is desired in order to say that the participant has habituated (see meeting a criterion).
This value is a multiplier between 0 and 1, exclusive. It can be expressed as a decimal with or without a leading zero.
An error is thrown if CRITERIONMET is used as a terminating condition without a definition of CRITERIONREDUCTION being present in the protocol.
In any live-coded experiment, including habituation, it is sometimes desirable to specify a minimum length of a look to a stimulus or away from it that should "count." This can be controlled with the COMPLETELOOK and COMPLETELOOKAWAY settings. See the section on the optional experimental settings page for more information.
Habituation studies are ones in which an experiment or phase continues until the child no longer attends (or attends less) to a particular stimulus.
In BITTSy, habituation phases terminate based on your protocol's set termination criteria - a decrease in looking relative to baseline. This, in turn, is based on the following factors:
How many trials should be included in the baseline measure and in judging whether habituation has occurred (e.g., 3 trials, 4 trials, etc.). Note that the window SIZE (in terms of number of trials) is the same for both.
Which trials are included in the baseline (e.g,. the FIRST three trials, or the 3 trials with LONGEST looking overall...)
Percentage drop (e.g., the phase should end when looking has reduced by a certain percentage of baseline, such as 50% of baseline looking)
See the other pages in this section for specifics on how to , and proceed to another phase of the experiment, or that do not meet desired criteria.
You might decide that you want to exclude particular trials from calculation of habituation. For example, you might wish to exclude any trial that has no looking at all. Or you might want to allow the experimenter to exclude individual trials at the time (say, if they felt the child was distracted by something in the room, such that the trial was not an accurate measure).
This can be done, and is part of the STEP terminating conditions used within a trial. You have already learned how to denote end conditions that are "successful" and should let the trial be counted for habituation calculations - these are simply UNTIL statements. "Unsuccessful" end conditions are the same kinds - just with an UNSUCCESSFUL flag.
The above example would play the file "audiofile" until one of the following occurs:
You reach the end of the audio file
A look away is logged that exceeds 2 seconds (when the trial is started only when the experimenter judges the child is already oriented in the active direction, this requires that a look toward the stimulus has already been logged)
10 seconds have passed since the trial started and the child has not yet looked at all
The experimenter presses a designated key (X) to stop the trial
As with other combinations of step terminating conditions, these conditions are treated as "whichever comes first". So, if the 3rd or 4th condition were met first, the trial would be unsuccessful. The trial would end at that point, and would not be counted as a trial for the purposes of habituation calculations, which depend on successful trials. (More precisely, this means that any window containing this trial would not be evaluated as either a potential basis window or criterion window.) If condition 1 or 2 were met instead, the trial would then end, and it would be counted.
Because terminating conditions are checked sequentially, if there is one that is "held up" and cannot currently be evaluated, none of the others are evaluated yet either. In particular, SINGLELOOK and TOTALLOOK cannot be evaluated while a look to that stimulus is still in progress. They will be evaluated as soon as the experimenter records the look has ended, or the stimulus stops being presented, and then alternate terminating conditions that are listed after them can be checked.
Marking a trial as unsuccessful only excludes the trial from habituation calculations. It does not exclude it from being counted as one of a set number of trials, as in an alternate loop terminating condition. For example, if your phase is set up as
it repeats until either the criterion is met or the loop occurs 20 times - marking a trial as unsuccessful doesn't change the looping.
If you want to replace a no-look trial altogether, you could have another loop that loops the trial until a condition was met. Then the inner loop would do 2 trials if its condition wasn't met the first time, and the outer loop would still go the same number of times, getting you an extra trial for each trial that was not successful.
It is also worth noting that in the syntax above, not only does the trial not count if the child hadn't looked in 10 seconds, but it also ends at that point - that is, it does not continue to play the rest of the sound file. You could have it finish the trial anyway with the statement:
but that would mean that any looks that occurred after 10 second mark would still be ignored from the count of habituation, and yet would presumably influence the child nonetheless. We instead recommend that if you are ignoring a trial for non-looking, that the trial end at whatever point you make that decision.
What about if you want it to play the whole file, and only mark the trial as unsuccessful if they never look at all? For that, recall the COMPLETELOOK setting defines the minimum look length that counts for looking time calculations. This is the smallest amount that can be logged, so "never looking at all" is anything less than the COMPLETELOOK value.
Here, it ends and is marked as successful if the sound file ends and there were any looks during that time, or if the child looked and then looks away for 2 seconds. It is unsuccessful if the experimenter pressed the X key, or if there weren't any looks at all by the time the audio file ends.