My question is similar to the one that Jack Grinband asked a few days ago
but with an added condition.
I have a block design with three conditions and rest periods between trials.
This is a visual memory task in which subjects memorize a pattern and,
after a short delay, have to decide whether a second pattern is either the
same or different as the first. The first pattern is preceeded by an
instruction to study and the second pattern is preceeded by an instruction
to recall. For this analysis the entire sequence of study, pattern, recall,
pattern is considered as a single trial. The conditions are as follows:
The first condition called "easy memory" follows the sequence decribed above
with an easily memorized pattern.
The second condition called "hard memory" follows the same sequence but with
a more complex pattern.
The third condition called "no memory" follows the same sequence but the
instructions have changed. Instead of studying the pattern subjects are
instructed to "look" at the pattern and instead of the word "recall"
subjects are given the correct answer to the following pattern. Subjects
are familiarized with the task prior to scanning and reaction times support
the assumption that they are, in fact, not making the effort to remember
during the "no memory" trials.
If we consider the order of the conditions as no memory, easy memory, and
hard memory then, following Christian Beckman's recommendation, there should
be one regressor equal to 1 whenever a pattern or instruction is present
(1,1,1). Another regressor whenever subjects have to memorize (0,1,1) and
finally a contrast to differentiate between the 2 memory conditions (0,-1,1).
Any feedback would be greatly appreciated.