CIBA: Continuous Interruption-free Brain Authentication

Florian Gondesen

1 a

and Dieter Gollmann

School of Computer Science and Engineering, Nanyang Technological University, Singapore

Hamburg University of Technology, Hamburg, Germany

Keywords:

Biometrics, Electroencephalography, Authentication, Visual.

Abstract:

The performance of contemporary biometrics systems based on electroencephalography (EEG) suffers from a

low signal to noise ratio due to the properties of the human EEG and the measurement on the scalp. There is a

trade-off between accuracy and the time required for data acquisition. Additional time is needed to mount an

EEG headset so that authentication requires several minutes, rendering it not very usable for most scenarios. In

a scenario where an EEG headset is already worn for a different purpose, the setup time can be neglected and

the time for data acquisition may be extended if it does not interfere with the subject’s actual task, allowing

continuous authentication. However, most proposed EEG-based authentication systems require the user to

perform a certain task during data acquisition, distracting the user from the actual task. We conceptualize an

EEG-based continuous authentication scheme that does not require the user to perform a task in addition to

working at a screen. We propose two approaches based on well known brain responses, SSVEP and ERP.

1 INTRODUCTION

With availability of consumer grade headsets, EEG-

based biometrics systems seem to become more ap-

plicable, increasing research interest in such systems.

Nevertheless, EEG-based biometrics have not estab-

lished themselves in real world applications yet. A

key impediment is the noisy nature of the electroen-

cephalogram (EEG). To extract discriminant infor-

mation, current EEG biometrics approaches capture

about ﬁve minutes of EEG data, whilst only achiev-

ing accuracies lower than desirable for practical ap-

plications (Gondesen et al., 2019a). Accuracy can

be increased by capturing longer time series of the

EEG, but extending the time for data acquisition

also reduces practical applicability. This is typically

even more inconvenient for the subject compared to

more commonly used biometrics, as most EEG-based

schemes require the user to perform a task during

data acquisition. The period of inconvenience for the

subject is extended by the time required for properly

mounting the EEG headset which may range from few

seconds to several minutes (Gondesen et al., 2019a;

Gondesen et al., 2019b). These properties render cur-

rent EEG biometrics approaches unsuitable for stan-

dard applications of biometrics as unlocking a phone

but may be suitable for a continuous authentication

https://orcid.org/0000-0001-6174-8854

scheme, especially in cases where an EEG is already

worn for other purposes. In recent studies, EEG was

used to determine mental parameters like awareness,

stress and workload in safety critical scenarios like air

trafﬁc control (Lim et al., 2018; Yeo et al., 2017; De-

bie et al., 2019). Safety critical work places are also

likely to be security critical, calling for continuous au-

thentication. But in such a scenario, subjects may not

be interrupted or distracted from their critical work,

which would be a requirement for most EEG-based

biometric schemes.

Therefore we present two possible concepts

for continuous interruption-free brain authentication

(CIBA), i.e., EEG-based continuous authentication

schemes aiming not to distract the users from their

work. Both concepts rely on delivering subliminal

stimuli to the subject by an off-the-shelf (gaming)

screen. The concepts are described and evaluated

based on related research efforts. Both approaches

utilize well known brain responses, namely steady

state visually evoked potential (SSVEP) and event-

related potential (ERP).

The paper is divided into the following sections:

In Sections 2 and 3, we explain the two approaches

and discuss their feasibility. In Section 4, we discuss

common issues of the proposed concepts. Section 5

provides a conclusion.

314

Gondesen, F. and Gollmann, D.

CIBA: Continuous Interruption-free Brain Authentication.

DOI: 10.5220/0010328003140319

In Proceedings of the 14th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2021) - Volume 4: BIOSIGNALS, pages 314-319

ISBN: 978-989-758-490-9

2 SSVEP

2.1 Prerequisites

When observing ﬂickering lights, the subject’s EEG

may exhibit activity at the ﬂicker frequency and its

(sub)harmonics, dominantly at the visual cortex. The

effect can be identiﬁed by peaks in the spectrum when

compared to the spectrum without stimulus. Brain-

computer interfaces (BCIs) based on SSVEP typically

use multiple simultaneous ﬂickering stimuli with dif-

ferent frequencies or phases, adding up to a complex

spectrum. To input data via such a BCI, the user

focuses on a single ﬂicker stimulus. The attention

to this single stimulus modulates the signals so that

the components elicited by the unattended stimuli are

decreased relative to the components elicited by the

stimulus focused on. The attention does not neces-

sarily need to be on the stimulus itself, it is only im-

portant to direct the attention to the space where the

stimulus is located, allowing the user for example to

read a text on a ﬂickering background (Morgan et al.,

1996).

The SSVEP paradigm works at frequencies of

about 3.5 Hz to 75 Hz (Herrmann, 2001; Amiri et al.,

2013). Similar to the pink noise characteristics of

the EEG baseline, the power decreases with increas-

ing frequency. BCIs often test a range of frequen-

cies to identify the frequencies with the best signal

to noise ratios for each user to select optimal stimu-

lus frequencies. Hence, frequency responses contain

discriminant information. The permanency, i.e., the

proportion of the individual frequency responses re-

maining constant over subsequent measurements, is

not extensively researched. In two studies the perma-

nency of an SSVEP biometrics identiﬁcation scheme

could be shown in multiple sessions with accuracies

over 90% (Piciucco et al., 2017; Falzon et al., 2017).

2.2 CIBA SSVEP

As the SSVEP paradigm does not require the sub-

ject to perform any task other than attending the lo-

cation of the stimulus, which can be the background

of the screen, it does not interrupt the user’s work-

ﬂow. Nevertheless, ﬂicker in the frequency range used

for SSVEP typically causes discomfort and strain and

may even cause epileptic seizures. For mitigation,

duty cycle and stimulus frequency can be optimized.

(Lee et al., 2011) showed that discomfort can be de-

creased by increasing the duty-cycle when using a

13.16 Hz stimulus. (Won et al., 2015) found no effect

of longer duty cycles on the comfort for frequencies

26 Hz to 34.7 Hz, but adverse effects on the perfor-

mance. This frequency range was reported to cause

less discomfort than 6 Hz to 14.9 Hz. If higher fre-

quencies generally decrease discomfort, one should

consider going to the limits of SSVEP. It has been

shown that the SSVEP paradigm works at frequencies

above the ﬂicker fusion threshold (Herbst et al., 2013;

Sakurada et al., 2015). A system that relies only on

frequencies above the ﬂicker fusion threshold can be

regarded as subliminal.

Implementation. In the CIBA SSVEP concept, a

PC screen with a high refresh rate would be used to

probe the subject for the responses to different stimu-

lus frequencies. The system can be tuned to test previ-

ously determined, most discriminant frequencies. The

stimulation could be realized by a software that cre-

ates a top layer which changes between a transparent

and a non-transparent black screen. This layer should

not capture inputs, as the subject must be able to per-

form regular work. A normal screen with a ﬁxed re-

fresh rate will most likely not sufﬁce, as properly re-

alizable stimulation frequencies are limited to even

divisors of the refresh rate. Scheduling of the oper-

ating system might also cause slight glitches in tim-

ing, creating visible artifacts that might be disruptive

to the user. Instead of a ﬁxed refresh rate screen, a

screen with an adaptive synchronization technology,

such as FreeSync, should be used. In a certain fre-

quency range, this allows arbitrary intervals between

two frames, enabling seamless delivery of a wide

range of stimulus frequencies. Gaming screens sup-

porting FreeSync often support refresh rates of 144 Hz

or higher, which allows to deliver stimuli in the range

of the ﬂicker fusion threshold. As the ﬂicker fusion

threshold is higher in the outer visual ﬁeld (Lee et al.,

2011), especially on large screens, ﬂicker might be

recognizable. By using an eye-tracker, the ﬂickering

area of the screen could be limited to the area ob-

served.

The CIBA setup including an eye-tracker is shown

in Fig. 1. A possible sequence of SSVEP stimuli is

shown in Fig. 2. By default, the stimulation layer is

set to a transparency level that darkens the screen to a

gray level adjusted to the perceived brightness of the

SSVEP stimulation above the ﬂicker fusion threshold.

During a trial, a circular area around the point of gaze

is switched between full transparency and black. Tri-

als are separated by interstimulus intervals (ISIs) that

are used for baseline correction.

Discussion. The applicability of the CIBA SSVEP

concept mainly depends on having sufﬁcient discrim-

inant information in the responses to stimuli above

the ﬂicker fusion threshold for all users. If lower

CIBA: Continuous Interruption-free Brain Authentication

315

EEG

EYE

TRK

Stimulus

Controller

Feature

Extractor

Classiﬁer

User DB

Artifact

Rejection

Figure 1: The setup for CIBA consists of a stimulus con-

troller presenting stimuli on the screen at the point of gaze,

which is obtained from the eye-tacker (EYE TRK). Stim-

ulus information consisting of timing, position and type of

stimulus is added to the EEG. Appropriate stimuli are se-

lected from the user database. Artifacts are removed from

the EEG before feature extraction. For the authentication

decision the classiﬁer compares the extracted features to

templates stored in the user database.

Lorem ipsum dolor sit amet,

consectetur adipiscing elit, sed do

eiusmod tempor incididunt ut labore

et dolore magna aliqua. Ut enim ad

minim veniam, quis nostrud

exercitation ullamco laboris nisi ut

aliquip ex ea commodo consequat.

Duis aute irure dolor in reprehenderit

Lorem ipsum dolor sit amet,

consectetur adipiscing elit, sed do

eiusmod tempor incididunt ut labore

et dolore magna aliqua. Ut enim ad

minim veniam, quis nostrud

exercitation ullamco laboris nisi ut

aliquip ex ea commodo consequat.

Duis aute irure dolor in reprehenderit

Lorem ipsum dolor sit amet,

consectetur adipiscing elit, sed do

eiusmod tempor incididunt ut labore

et dolore magna aliqua. Ut enim ad

minim veniam, quis nostrud

exercitation ullamco laboris nisi ut

aliquip ex ea commodo consequat.

Duis aute irure dolor in reprehenderit

Lorem ipsum dolor sit amet,

consectetur adipiscing elit, sed do

eiusmod tempor incididunt ut labore

et dolore magna aliqua. Ut enim ad

minim veniam, quis nostrud

exercitation ullamco laboris nisi ut

aliquip ex ea commodo consequat.

Duis aute irure dolor in reprehenderit

Lorem ipsum dolor sit amet,

consectetur adipiscing elit, sed do

eiusmod tempor incididunt ut labore

et dolore magna aliqua. Ut enim ad

minim veniam, quis nostrud

exercitation ullamco laboris nisi ut

aliquip ex ea commodo consequat.

Duis aute irure dolor in reprehenderit

Lorem ipsum dolor sit amet,

consectetur adipiscing elit, sed do

eiusmod tempor incididunt ut labore

et dolore magna aliqua. Ut enim ad

minim veniam, quis nostrud

exercitation ullamco laboris nisi ut

aliquip ex ea commodo consequat.

Duis aute irure dolor in reprehenderit

T=-1000

T=0

T=15

T=30

T=2000

T=3000

[ms]

Figure 2: Sequence of SSVEP stimuli with a 66.67 Hz trial

of 2 s. The stimulation is limited to an area around the point

of gaze.

frequencies need to be used, the system is rendered

non-subliminal, uncomfortable to use and may even

pose risks to the users’ health. Using only high fre-

quencies might decrease accuracy, requiring an ex-

tended sliding window for continuous authentication.

It has to be made sure that this is within the security

requirements. Additionally, permanency needs to be

assessed. It is possible that the templates will change

over time, requiring an adaptive approach.

3 ERP

3.1 Prerequisites

An event-related potential (ERP) is the signal compo-

nent in the EEG that is caused by an event, for exam-

ple the presentation of a visual stimulus to the subject.

As the ERP is relatively weak, stimuli are typically

repeated to average out unrelated EEG components.

The ERP’s waveform varies with the type of event and

individual characteristics of the subject, hence ERPs

are widely used in biometrics approaches. There are a

few paradigms that affect the components of the ERP

in certain ways. One of the most famous is the odd-

ball paradigm, where two types of stimuli are deliv-

ered randomly, with one type only occurring rarely.

The ERP of this rare event contains a peak after about

300 ms and is therefore called P300. The amplitude

of the P300 component is increased if the stimulus

is relevant to the task of the subject. This property

can be used to create an EEG-based authentication

system that is based on knowledge instead of biomet-

rics (Gondesen et al., 2019b). (Das et al., 2015; Das

et al., 2016) used an oddball paradigm for biometric

authentication achieving equal error rates about 14%.

Permanency was veriﬁed over a couple of weeks.

ERPs are susceptible to eye artifacts. Moving or

blinking eyes induce waves with relatively high am-

plitudes in the EEG. If an ERP is superimposed with

such an artifact, it will not be easily averaged out and

might lead to a misclassiﬁcation.

3.2 CIBA ERP

Common ERP-based biometrics approaches require

the users to pay attention to (most commonly visual)

stimuli, which are presented multiple times to in-

crease the signal to noise ratio. This contradicts the

CIBA concept, which does not allow disruptions of

the normal workﬂow. ERPs are elicited even with-

out a task related to the event, but this may affect

components that otherwise contain discriminant in-

formation. In addition to not including an authen-

tication task, stimulus presentation time needs to be

limited so that the subject cannot consciously per-

ceive it. (Bernat et al., 2001) used pictures of two

words as oddball stimuli presented for only 1 ms, en-

suring subliminal stimulation. The P300 component

was found to be increased for the rare case. (Frank

et al., 2017) conducted an ERP experiment aiming

to probe if a face is known to the subject. The sub-

jects were instructed to watch a movie, where occa-

sionally a known face or blurred face was inserted

for an interval of 13.3 ms, which was subliminal to

only four of 22 subjects. The other subjects per-

ceived the images to different degrees, seven being

able to name the depicted person. The known face

was shown rarely, creating an oddball paradigm. The

known face could be detected from the EEG in most

cases. As the experiment did not contain non-blurred

BIOSIGNALS 2021 - 14th International Conference on Bio-inspired Systems and Signal Processing

316

unknown faces, the effect might only be caused by a

general face detection, not by the known face. De-

tection whether a face is known to the subject could

be seen as a knowledge factor for an authentication

system. Non-subliminal ERP studies aiming to iden-

tify known faces in a concealed information test (CIT)

were conducted by (Meijer et al., 2007; Meijer et al.,

2009). Detection of known faces did not succeed un-

der one experiment variant where the depicted person

was of lesser importance for the subject and not a part

of the subject’s task. It was concluded that recogni-

tion of a familiar face may sufﬁce to elicit a P300,

but the P300 may increase with stronger familiarity

or an associated task. The experiments also indi-

cate that mere recognition of autobiographical infor-

mation elicits a P300. The properties of information

being autobiographical or faces being known depend

on the individual subject, i.e., they contain discrimi-

nant information, rendering autobiographical data and

known faces promising candidates for stimuli in the

CIBA paradigm.

Implementation. To deliver stimuli for a very short

duration, a screen with a high refresh rate is required.

With an off-the-shelf gaming screen with 144 Hz,

stimuli are at least displayed for 6.9 ms, which might

not be subliminal for all users. Reaching 1 ms is not

feasible with current off-the-shelf monitor technol-

ogy. As subliminality does not only depend on the

duration, it should be considered to design the stim-

ulus delivery system to mask the stimuli, making it

harder to perceive them. This could be done by de-

livering stimuli temporally and spatially aligned with

other events happening on the screen, as a stimulus

on a static background can be more easily noticed.

Stimuli could also be selected or modiﬁed to be more

similar to the background where they are placed. An

eye-tracker can help to place stimuli in the foveal area

to make sure it is observed. (Brunner et al., 2010)

showed increased accuracy in a P300 BCI study for

gaze directed to the target stimulus. This effect may

also apply to CIBA ERP. An eye-tracker can also help

identifying eye artifacts. Removing eye artifacts is

important as the user cannot be instructed to abstain

from blinking and ﬁxating a certain spot on the screen

in the CIBA scenario.

The CIBA setup including an eye-tracker is shown

in Fig. 1. A possible sequence of oddball stimuli is

shown in Fig. 3.

Discussion. The studies discussed indicate that a

passive, subliminal oddball paradigm can be used to

evoke a P300 component, but do not answer whether

the P300 contains sufﬁcient discriminant information.

Lorem ipsum dolor sit amet,

consectetur adipiscing elit, sed do

eiusmod tempor incididunt ut labore

et dolore magna aliqua. Ut enim ad

minim veniam, quis nostrud

exercitation ullamco laboris nisi ut

aliquip ex ea commodo consequat.

Duis aute irure dolor in reprehenderit

Lorem ipsum dolor sit amet,

consectetur adipiscing elit, sed do

eiusmod tempor incididunt ut labore

et dolore magna aliqua. Ut enim ad

minim veniam, quis nostrud

exercitation ullamco laboris nisi ut

aliquip ex ea commodo consequat.

Duis aute irure dolor in reprehenderit

Lorem ipsum dolor sit amet,

consectetur adipiscing elit, sed do

eiusmod tempor incididunt ut labore

et dolore magna aliqua. Ut enim ad

minim veniam, quis nostrud

exercitation ullamco laboris nisi ut

aliquip ex ea commodo consequat.

Duis aute irure dolor in reprehenderit

Lorem ipsum dolor sit amet,

consectetur adipiscing elit, sed do

eiusmod tempor incididunt ut labore

et dolore magna aliqua. Ut enim ad

minim veniam, quis nostrud

exercitation ullamco laboris nisi ut

aliquip ex ea commodo consequat.

Duis aute irure dolor in reprehenderit

Lorem ipsum dolor sit amet,

consectetur adipiscing elit, sed do

eiusmod tempor incididunt ut labore

et dolore magna aliqua. Ut enim ad

minim veniam, quis nostrud

exercitation ullamco laboris nisi ut

aliquip ex ea commodo consequat.

Duis aute irure dolor in reprehenderit

Lorem ipsum dolor sit amet,

consectetur adipiscing elit, sed do

eiusmod tempor incididunt ut labore

et dolore magna aliqua. Ut enim ad

minim veniam, quis nostrud

exercitation ullamco laboris nisi ut

aliquip ex ea commodo consequat.

Duis aute irure dolor in reprehenderit

T=-1000

T=0

T=5

T=1230

T=8385

T=8390

[ms]

target

non-target

target

Figure 3: Sequence of oddball stimuli using simple objects

(obtained from (Rossion and Pourtois, 2004)) ﬂashed at the

point of ganze for 5 ms.

It also remains open to which degree knowledge fac-

tors like known faces, or general familiarity with the

stimuli can contribute. If familiarity is an important

feature, it needs to be investigated whether sublimi-

nally shown pictures of unfamiliar persons or items

may render them familiar over time.

The P300 depends on the difﬁculty of the task,

showing decreased amplitudes for more difﬁcult

tasks (Kok, 1997). As more difﬁcult recognition tasks

also extend the P300 length (Twomey et al., 2015),

classiﬁcation accuracy might not degrade. It should

be also investigated to which extend difﬁculty affects

the passive, subliminal paradigm and if it is effected

by masking.

4 DISCUSSION

The approaches presented rely on well-established

paradigms used in BCIs, but these paradigms are

rather uncommon in EEG-based biometrics. Hence, it

is not clear if sufﬁcient discriminant information can

be identiﬁed. The quality of the information can usu-

ally be improved by capturing more EEG data. The

CIBA concept allows to adjust the sliding window of

EEG data used for continuous authentication. But in-

creasing the window size increases the response time

for deauthentication, rendering the system less se-

cure. Rather than mitigating the time-accuracy trade-

off, the CIBA concept impairs accuracy due to the

requirements of not using a task and using sublimi-

nal stimuli, eliminating the subjects attention. In both

paradigms, (spatial) attention improves the underly-

ing EEG signals. In addition, both paradigms need to

be used with unusual parameters like very high fre-

quencies or short stimulus intervals, contributing to

weaker signals.

In the CIBA concept the continuous authentica-

tion process must not distract the subject from reg-

ular work, but the EEG data must contain discrimi-

CIBA: Continuous Interruption-free Brain Authentication

317

nant information. Evoking EEG signals by sublimi-

nal stimuli means that the stimuli are processed sub-

consciously by the brain. This might create a con-

ﬂict in allocation of brain resources, limiting regular

work or the accuracy of the authentication. (Allison

and Polich, 2008) showed that the P300 of an auditory

counting task decreased with the difﬁculty of a game

played simultaneously.

Besides the performance of the subject, it also

needs to be studied how subliminal continuous au-

thentication affects the subject’s well-being. Both

might not only be affected by the subliminal stimu-

lation itself, but also by the subject knowing to be

probed continuously or a lack of accuracy leading

to occasional false negatives, deauthenticating a valid

user.

5 CONCLUSION

Both CIBA approaches are based on well known

paradigms that are widely used in BCI research. In

both cases the parameters are pushed to the limits

where it is not clear how well the paradigms work

and if they can deliver sufﬁcient discriminant infor-

mation without requiring a sliding window that is too

large to meet the security requirements. It can be ex-

pected that real world applications will not only need

research in the feasibility of the approaches but also

in the optimization of data acquisition and feature ex-

traction algorithms.

REFERENCES

Allison, B. Z. and Polich, J. (2008). Workload assessment

of computer gaming using a single-stimulus event-

related potential paradigm. Biological psychology,

77(3):277–283.

Amiri, S., Fazel-Rezai, R., and Asadpour, V. (2013). A

Review of Hybrid Brain-Computer Interface Systems.

Advances in Human-Computer Interaction.

Bernat, E., Shevrin, H., and Snodgrass, M. (2001). Sublim-

inal visual oddball stimuli evoke a P300 component.

Clinical neurophysiology, 112(1):159–171.

Brunner, P., Joshi, S., Briskin, S., Wolpaw, J. R., Bischof,

H., and Schalk, G. (2010). Does the ‘P300’ speller

depend on eye gaze? Journal of neural engineering,

7(5):056013.

Das, R., Maiorana, E., and Campisi, P. (2016). EEG Bio-

metrics Using Visual Stimuli: A Longitudinal Study.

IEEE Signal Processing Letters, 23(3):341–345.

Das, R., Maiorana, E., Rocca, D. L., and Campisi, P. (2015).

EEG Biometrics for User Recognition Using Visually

Evoked Potentials. In International Conference of the

Biometrics Special Interest Group, pages 1–8.

Debie, E., Rojas, R. F., Fidock, J., Barlow, M., Kasmarik,

K., Anavatti, S., Garratt, M., and Abbass, H. A.

(2019). Multimodal fusion for objective assessment

of cognitive workload: a review. IEEE transactions

on cybernetics.

Falzon, O., Zerafa, R., Camilleri, T., and Camilleri, K. P.

(2017). EEG-based biometry using steady state visual

evoked potentials. In 2017 39th Annual International

Conference of the IEEE Engineering in Medicine and

Biology Society (EMBC), pages 4159–4162. IEEE.

Frank, M., Hwu, T., Jain, S., Knight, R. T., Martinovic, I.,

Mittal, P., Perito, D., Sluganovic, I., and Song, D.

(2017). Using EEG-based BCI devices to sublimi-

nally probe for private information. In Proceedings

of the 2017 on Workshop on Privacy in the Electronic

Society, pages 133–136.

Gondesen, F., Marx, M., and Gollmann, D. (2019a). EEG-

Based Biometrics. In Biometric-Based Physical and

Cybersecurity Systems, pages 287–318. Springer.

Gondesen, F., Marx, M., and Kycler, A.-C. (2019b). A

Shoulder-Surﬁng Resistant Image-Based Authentica-

tion Scheme with a Brain-Computer Interface. In

2019 International Conference on Cyberworlds (CW),

pages 336–343. IEEE.

Herbst, S. K., Javadi, A. H., van der Meer, E., and

Busch, N. A. (2013). How long depends on how

fast—perceived ﬂicker dilates subjective duration.

PloS one, 8(10).

Herrmann, C. S. (2001). Human EEG responses to 1–

100 Hz ﬂicker: resonance phenomena in visual cortex

and their potential correlation to cognitive phenom-

ena. Experimental brain research, 137(3-4):346–353.

Kok, A. (1997). Event-related-potential (ERP) reﬂections

of mental resources: a review and synthesis. Biologi-

cal psychology, 45(1-3):19–56.

Lee, P.-L., Yeh, C.-L., Cheng, J. Y.-S., Yang, C.-Y., and Lan,

G.-Y. (2011). An SSVEP-based BCI using high duty-

cycle visual ﬂicker. IEEE Transactions on Biomedical

Engineering, 58(12):3350–3359.

Lim, W. L., Liu, Y., Subramaniam, S. C. H., Liew, S. H. P.,

Krishnan, G., Sourina, O., Konovessis, D., Ang, H. E.,

and Wang, L. (2018). EEG-based mental workload

and stress monitoring of crew members in maritime

virtual simulator. In Transactions on Computational

Science XXXII, pages 15–28. Springer.

Meijer, E. H., Smulders, F. T., Merckelbach, H. L., and

Wolf, A. G. (2007). The P300 is sensitive to con-

cealed face recognition. International Journal of Psy-

chophysiology, 66(3):231–237.

Meijer, E. H., Smulders, F. T., and Wolf, A. (2009). The

contribution of mere recognition to the P300 effect in

a concealed information test. Applied psychophysiol-

ogy and biofeedback, 34(3):221–226.

Morgan, S., Hansen, J., and Hillyard, S. (1996). Selective

attention to stimulus location modulates the steady-

state visual evoked potential. Proceedings of the Na-

tional Academy of Sciences, 93(10):4770–4774.

Piciucco, E., Maiorana, E., Falzon, O., Camilleri, K. P., and

Campisi, P. (2017). Steady-state visual evoked poten-

tials for EEG-based biometric identiﬁcation. In 2017

BIOSIGNALS 2021 - 14th International Conference on Bio-inspired Systems and Signal Processing

318

International Conference of the Biometrics Special In-

terest Group (BIOSIG), pages 1–5. IEEE.

Rossion, B. and Pourtois, G. (2004). Revisiting snodgrass

and vanderwart’s object pictorial set: The role of sur-

face detail in basic-level object recognition. Percep-

tion, 33(2):217–236.

Sakurada, T., Kawase, T., Komatsu, T., and Kansaku, K.

(2015). Use of high-frequency visual stimuli above

the critical ﬂicker frequency in a SSVEP-based BMI.

Clinical Neurophysiology, 126(10):1972–1978.

Twomey, D. M., Murphy, P. R., Kelly, S. P., and O’Connell,

R. G. (2015). The classic P300 encodes a build-to-

threshold decision variable. European journal of neu-

roscience, 42(1):1636–1643.

Won, D.-O., Hwang, H.-J., D

ahne, S., M

uller, K.-R., and

Lee, S.-W. (2015). Effect of higher frequency on the

classiﬁcation of steady-state visual evoked potentials.

Journal of neural engineering, 13(1):016014.

Yeo, L. G., Sun, H., Liu, Y., Trapsilawati, F., Sourina,

O., Chen, C.-H., Mueller-Wittig, W., and Ang, W. T.

(2017). Mobile EEG-based situation awareness recog-

nition for air trafﬁc controllers. In 2017 IEEE Interna-

tional Conference on Systems, Man, and Cybernetics

(SMC), pages 3030–3035. IEEE.

CIBA: Continuous Interruption-free Brain Authentication

319