stats: Greg Zarow
Basic Task: This is a Stroop-type task, but with location-word
compatibility instead of color compatibility. On each trial, subjects see
a word above or below a line [+++++]. Their task is to indicate the
location of the word, either above or below, using z or ? keys.
The stroop-like part is that the word ABOVE can appear below the line and that
might impair their properly responding below!
Stimuli (2)
Text: ABOVE , BELOW
uppercase sanserif font, white
text on black background.
~20pt font [we want to subtend ~1.7x0.7degrees visual angle]
Background: black, with a horizontal string of small
white plus signs (++++++++++) midscreen, dividing the screen into top and
bottom halves. The line should always remain present.
During each trial, a stimulus
appears, starting the RT clock. Stimuli should appear centered horizontally,
but a few pixels (~.25degrees visual angle) above (or below) the line of
++++++++++.
The subject’s task is to
indicate whether the text is above or below the midline by hollering above! or
below!
Make certain that responses
cannot occur before the stimulus comes on the screen.
There are four trialtypes:
two are compatible (ABOVE above the line; BELOW below the line) and two
are incompatible (ABOVE below the line; BELOW above the line).
We are interested in
distinguishing between compatible and incompatible trials.
We are interested in both
reaction time and errors.
There is another factor that
varies between blocks. This is the timing of the presentations.
Here is the design. There
are 2 blocktypes:
Blocktype
Interval from one trial to the next
1
Fixed 200 msec
2
Fixed 2000 msec
Before starting the experiment
proper, have them do ten practice trials. This should be done
while the experimenter is in the room, to check that the voice recognition is
working and to ensure that Ss understand the task.
There should be 8 blocks each composed of 100 trials per block.
The blocks should alternate between blocktype 1 and blocktype 2. Please
counterbalance across participants such that the subject s order condition (1
or 2) should determine which blocktype they start with in block 1.
At the end of each block, put up
a signal saying Rest Period click to resume and then continue when they click.
Instructions to include in
file:
Thank you for participating in
our experiment. In this study, you will be seeing words above or below
the line on the screen. Your task is to determine the placement of the
word, whether the word appears above the line or below the line regardless of
what the word is. You are to say above! as quickly as possible when the word is
above the line and you should say below! as quickly as possible when the
word is below the line. Sometimes it might be confusing. For
example, the word ABOVE might appear below the line, but the correct answer is below,
so you should quickly say Below! to this. Again, your task is to identify the
placement, not the word.
We are interested in both speed
and accuracy. Try to respond as quickly and as accurately as you
can. You will do 10 blocks of trials, each lasting a few minutes.
Your task is: as soon a word
appears, just identify the placement as quickly and accurately as you
can. Everyone makes some mistakes. Try to keep your eye on the line
and the words will appear above or below the line. There may be long pauses
between trials. The blocks of trials don t last more than a few minutes and
you can take a small break after every block of trials. If you have any
questions, please ask the experimenter now.
File Structure
In the R-file save the
Participant ID#, the block number, trial number, counterbalance order,
blocktype (Long or Short), trialtype (Compat or Incompat), the number of
seconds that passed from the previous trial, the word, the location, the
subject s response time, and their response identity (correct or incorrect).
Additionally, please make a
concatenated file at the end of the study, which includes all trials for all
participants and includes a column of participant ID#.
Note on Voice Data
This assumes the
voice-recognition is working. I suspect that subjects will often blurt
out one thing and correct themselves midstream. Sometimes that will
probably produce an unrecognizable combination ( Bel-above! ) and sometimes a
sequence of two ( Above, no, Below! ). If you can save Unrecog in the
first case, and Below Correction in the second, that would be perfect.
The trial should be counted as an error in any case. To be counted right
they must say the correct word and nothing else.