Information Technology Reference
In-Depth Information
“
follow incorrect motor action (e.g., wrong key press). It is a
negative potential (error negativity ERN) in the EEG 80
response ErrP
”
100 ms after the
incorrect movement followed by a positive peak between 200 and 500 ms.
“
-
ects incorrect
performance (e.g., system presents a feedback that tells the user he has done an
error). It is a negative de
feedback ErrP
”
follow the presentation of a stimulus that re
ection in the EEG 250 ms after the stimulus.
“
follow the observation of errors made by an operator during
choice reaction task where the operator needs to respond to stimuli. It is a
negative de
observation ErrP
”
ection 250 ms after an incorrect response of the operator was
observed.
“
”
appear when the user interacts with an application that does
not respond in the expected way. It is composed of a positive peak 200 ms after
the feedback, a negative peak after 250 ms, a second larger peak after 320 ms, a
larger negative peak after 450 ms, and a late peak after 600 ms.
interaction ErrP
Interaction ErrP can, for example, be used to detect error in a P300 speller (see
Sect.
13.4.2
).
13.3.4 Mental Workload
Real-time measurement of mental workload could present bene
ts in different
contexts as suggested in Blankertz et al. (
2010
) and Coffey et al. (
2010
). For
instance, it could be used for safety purpose (e.g., raising alert), for improving
human effectiveness and reducing errors (e.g., modifying task demands or acti-
vating assistance in times of cognitive overload), or as an objective measure in
usability evaluation of new products.
Two relevant markers are used to estimate mental workload: ERP and EEG
oscillatory activity (Van Erp et al.
2010
). ERP have been shown to be affected by
the user
is mental workload, whereas EEG rhythmic activity is correlated with
mental workload levels.
There are no obvious best markers to estimate user
'
s mental workload. As
reported by Berka et al. (
2007
), the requirement of an electing stimulus into real-
world tasks to elicit the potentials is a limitation of the ERP-based approach.
Oscillatory rhythmic activity-based estimators show the advantage of being able to
be used without disrupting the primary mode of interaction of the user. Indeed,
unlike ERP-based estimators, they do not require any external stimulus. They could
be used in a completely passive BCI context. The combination of the two tech-
niques has shown improvement of
'
the mental workload estimator accuracy
(Brouwer et al.
2012
).
