Information Technology Reference
In-Depth Information
8.8
Method
8.8.1 Participants
Twenty-two healthy volunteers participated in this experiment. All subjects were
BCI naive at the time of the experiment, and none of them reported neurological or
psychiatric disorders in their lifetime. Due to the presence of excessive artifacts in
the EEG data, three subjects were subsequently excluded from all analyses, leaving
19 participants, of which 9 females and 10 males, with age ranging from 20 to 30
with a mean and a standard deviation of 24 and 2.52, respectively.
8.8.2 Experimental Design
The experiment involved two sessions lasting altogether approximately half an
hour. Each session consisted of six blocks of six trials, for a total of 6
2=72
trials. Participants seated comfortably 80 cm in front of a 21-inch computer screen.
Nine square boxes were arranged in circle on the screen. Each trial consisted of the
same memory retrieval task: The trial started with the display of the current score
for 3,000 ms (initialized at zero), followed by a
×
6
×
fixation cross, also displayed for
3,000 ms (Fig.
8.2
a). Then, the memorization sequence started; each memorization
comprised a random sequence of two to nine digits appearing sequentially in
random positions, with each digit of the sequence randomly assigned to a different
box for each sequence (Fig.
8.2
b). Subjects were instructed to retain positions of all
digits. At the end of the sequence, the target digit (always contained in the previous
sequence) was displayed (Fig.
8.2
c) and subjects had to click with the aid of a
mouse on the box where it had appeared. Once the subject had answered, the
interface waited for 1,500 ms in order to avoid any contamination of ErrP by beta
rebound motor phenomena linked to mouse clicking (Pfurtscheller
1981
). Then, if
the answer was correct,
the chosen box background color turned into green
“
”
“
”
(
feedback). Subjects were
then asked to report if the feedback (error/correct) matched their expectation by a
mouse click (
correct
feedback); otherwise, it turned into red (
error
“
yes
”
/
“
no
”
) (Fig.
8.2
d). Following this answer, a random break of
1,000
1,500 ms preceded the beginning of the new trial.
In order to keep the subjects motivated throughout the experiment, the accumu-
lated score was computed at the beginning of each trial. When subjects localized
correctly the target digits, their score increased; otherwise, it remained unchanged.
The number of digits in the sequence was always between two and nine,
-
fixed within
blocks and updated, at the beginning of each block, according to the change in
performance from the block just
finished and the previous one, as assessed online by
means of statistical t tests. The
first block started always with four digits for all
subjects. The parameters of the adaptation were set, thanks to a pilot study and a
computer simulation, and were chosen to yield about 20 % of errors, regardless of the
working memory ability. Moreover, our learning approach is capable of adapting to
