E-mail: pharoram@umich.edu
Abstract
Objective. Typically, clinical measures of cognition require motor or speech responses. Thus,
a significant percentage of people with disabilities are not able to complete standardized
assessments. This situation could be resolved by employing a more accessible test
administration method, such as a brain–computer interface (BCI). A BCI can circumvent
motor and speech requirements by translating brain activity to identify a subject’s response.
By eliminating the need for motor or speech input, one could use a BCI to assess an
individual who previously did not have access to clinical tests. Approach. We developed
an asynchronous, event-related potential BCI-facilitated administration procedure for the
peabody picture vocabulary test (PPVT-IV). We then tested our system in typically developing
individuals (N = 11), as well as people with cerebral palsy (N = 19) to compare results to the
standardized PPVT-IV format and administration. Main results. Standard scores on the BCI-
facilitated PPVT-IV, and the standard PPVT-IV were highly correlated (r = 0.95, p < 0.001),
with a mean difference of 2.0 ± 6.4 points, which is within the standard error of the PPVT-IV.
Significance. Thus, our BCI-facilitated PPVT-IV provided comparable results to the standard
PPVT-IV, suggesting that populations for whom standardized cognitive tests are not accessible
could benefit from our BCI-facilitated approach.
2
J. Neural Eng. 14 (2017) 066001 R E Alcaide-Aguirre et al
assessments may have some form of cognitive impairment to provide results that are not difficult for the clinician to
that prevents them from responding at the same pace as a interpret.
non-impaired person. Also, the difficulty of the test questions 3. Brain-based cognitive assessment systems must be quick
will almost certainly vary between tests and within a test. If to set up (1 h or less).
a person must rush to answer a question, due to limitations 4. Brain-based cognitive assessment systems must have
of the BCI, then the results of the cognitive assessment may asynchronous control, thus allowing the subject to control
not accurately measure a person’s capacity. Thus, our fourth the pace of the assessment.
criterion: (4) a brain-based cognitive assessment system must 5. The BCI must be able to function in the target population.
have asynchronous control.
Using the criteria above, we developed an asynchronous
Perego’s study also allows us to glean information on the
ERP BCI, which retains the test and result format of the pea-
applicability of SSVEP BCIs. Perego’s BCI was only usable
body picture vocabulary test (PPVT-IV) [10]. All uses and
in 57% of his subjects, and six out of the seven subjects unable
adaptions to the PPVT-IV were approved by the publisher for
to use the BCI were people with cerebral palsy (CP). Other
research purposes only. We administered the BCI-facilitated
studies have also shown mixed results when using SSVEP
PPVT-IV to people without impairments and to people with
BCIs in populations with CP. Lower classification accuracy
CP. We chose the PPVT-IV because it has a strong test–retest
is usually attributed to the involuntary movements and muscle
reliability ranging from 0.91 to 0.94 across two different ver-
contractions in the neck, which are typical of CP. SSVEP BCIs
sions, form A and form B [10]. The strong retest reliability
rely heavily on occipital electrodes, which are the electrodes
allowed us to compare our BCI-facilitated PPVT-IV with the
closest to the subject’s neck [6, 15]. These electrodes are pro-
standard PPVT-IV.
foundly affected by muscle artifacts from the neck, which can
significantly alter signal quality. This unintentional interfer-
ence can ultimately lead to decreased BCI performance in 2. Materials and methods
people with CP. Another issue is that most SSVEP BCIs func-
tion like an eye-tracking system, requiring a person to focus The Institutional Review Board of the University of Michigan
and maintain their vision on the stimulus that corresponds to approved recruitment and data collection protocols. In total,
their selection [6, 15, 20]. For populations with conditions we recruited 11 people without impairments and 19 indi-
that include oculomotor impairments, maintaining such a gaze viduals with CP of whom 10 were able to complete the
may be too difficult. While some SSVEP systems can be oper- tests. Participants were ages 8–27, and were drawn from the
ated with closed eyes, or function using covert orienting of University of Michigan Health System and surrounding areas.
attention, these systems typically reduce the selection set to Subjects or their parents signed informed consent forms and
only two illustrations [21]. The reduced selection set means filled out demographic surveys.
many cognitive assessment tests would have to be modified to Inclusion criteria for both groups were ages 8–29 and suf-
a two-choice format, creating psychometric incompatibilities ficient speech or movement and vision to participate in the
and violating our first design criterion. standardized version of the PPVT-IV with screening via the
An alternative to SSVEP BCIs is the visual event-related practice items for the test. Exclusion criteria included history
potential (ERP) BCI [22]. Like SSVEP BCIs, ERP BCIs use of moderate or severe acquired brain injury or other major
visual stimuli of flashing objects to elicit brain responses for neurological condition such as stroke, encephalitis, or refrac-
control. In an ERP BCI, each object (or group of objects) tory seizure disorder (for children with CP, this refers to
flashes one at a time. The flashing elicits an ERP brain response events subsequent to the onset and diagnosis of CP), major
only to flashes emitted by the object the subject is interested psychiatric disorder such as major depression, severe anxiety
in selecting. By determining which flashing object elicits the or psychosis that precluded participation, or for those under
ERP response, an ERP BCI can identify the subject’s desired the age of 18 the inability of the parent/guardian to complete
selection. Like SSVEP BCIs, ERP BCIs are easy to learn a child history. In the sample with CP, one participant was
and can incorporate asynchronous control [23]. The primary taking baclofen and one was taking sertraline. In the NCP
advantage of ERP BCIs over SSVEP BCIs is that they do not sample, one participant was taking sertraline.
rely as heavily on occipital electrodes for classification and do In the final group of 10 subjects with CP, primary tone in all
not require subjects to maintain visual fixation on the flashing participants was spasticity, with 60.0% exhibiting hemiplegia
object they want to select. For these reasons, ERP BCIs have and 40% diplegia. Functional mobility levels were assessed
a potential advantage over SSVEP BCIs in people with CP. using the Gross Motor Functional Classification System [24]
Thus, our final criterion is that: (5) the BCI must be able to criteria with participant level distribution as follows: level I
function in the target population. (5) 50.0%, level III (2) 20.0%, level IV (2) 20.0%, and level V
In summary, our criteria are as follows: (1) 10%. Manual Ability Classification System [25] levels
included level I (3) 30%, level II (4) 40%, and level III (3)
1. A cognitive assessment BCI should maintain the psy- 30%.
chometric properties of the standardized administration Subjects attempted the standard PPVT-IV and the BCI-
procedure. facilitated PPVT-IV. Subjects took the tests in a pseudo-
2. Brain-based cognitive assessment systems must automat- random order. We used two matched difficulty versions of the
ically abstract the complexity of brain activity analysis PPVT-IV, form A and form B, to minimize practice effects.
3
J. Neural Eng. 14 (2017) 066001 R E Alcaide-Aguirre et al
We used form A for the standard PPVT-IV and form B for the subject moves through the test questions until they have com-
BCI-facilitated PPVT-IV. To document perceived workload of pleted all the sets, or until they submit eight or more incorrect
our BCI-facilitated PPVT-IV and the standard PPVT-IV, sub- responses in one set. The final set is labeled the ceiling set,
jects filled out a NASA Task Load Index survey (NASA-TLX) and the number of incorrect responses is subtracted from the
after each test [26, 27]. highest question tested to determine the raw PPVT-IV score.
Using the PPVT-IV normative conversion score tables, the
raw PPVT-IV scores are converted into standardized scores
2.1. BCI setup
that are utilized in statistical analyses.
The BCI was set up and calibrated for each subject using a When the subject took the standard PPVT-IV, we used
32-electrode (EEG cap (Electro-Cap, Inc.)), with a sampling the standard PPVT-IV protocol outlined above [10]. The
rate of 600 Hz. Online classification only used 16 channels BCI-facilitated PPVT-IV used the same logic flow as the
(F3, Fz, F4, T7, C3, Cz, C4, T8, CP3, CP4, P3, Pz, P4, PO7, standard PPVT-IV. However, the subject viewed illustrations
Oz, and PO8), to match the classification montage of our pre- on a 28-inch monitor (running at 120 Hz refresh rate), and the
vious studies for future comparison [23, 28]. We reserved the subject heard each question spoken from computer speakers
other channels for future analysis [23, 28–29]. Before taking (figure 1).
the BCI-facilitated PPVT-IV, subjects responded to 60 PPVT- The numbers in each selection box and the X in the cancel
like questions where the computer provided the correct answer box elicited ERP responses. Only one number or the X flashed
to the subject by highlighting the answer. Each question was at a time, prompting an ERP response only when the subject’s
presented on a monitor and showed four different illustrations. choice flashed (figure 1). Subjects responded to the BCI-
A spoken word was played through a pair of speakers that facilitated PPVT-IV by focusing their attention on the selec-
corresponded to the correct answer. The subject made his or tion box that corresponded to the illustration they wanted to
her selection by focusing their attention on the corresponding choose (figure 1).
selection box of each illustration. The selection boxes flash a
number inside them. However, only one selection box flashed
2.4. Classification
at a time. Only when the selection box the user wanted to select
flashed was an event related potential elicited. The subject did We used a three-stage classifier for ERP classification. During
two 30 question runs which took about 7 min per run. The data the first stage (stage 1), we applied the weights derived using
collected from these runs were used to calibrate the BCI. stepwise linear discriminant analyses (SWLDA) during the
calibration step to the subject’s EEG responses [30]. SWLDA
uses feature space reduction to find suitable features in a sub-
2.2. NASA-TLX
ject’s data to classify between two classes. In our case, the two
The NASA-TLX [26] is a survey instrument that is commonly classes were whether an EEG response contained an ERP or
used to assess the workload of a task. It consists of six ques- not. After establishing the features, the SWLDA classifier can
tions, and each question features a 21-point scale that the sub- then classify a subject’s EEG. EEG classification produces a
ject uses to convey the perceived difficulty of the task they value called the classification value. The classification values
did. The questions ask subjects to rate their perceived perfor- are either a negative or positive value, depending on whether
mance, mental demand, physical demand, temporal demand, a subject does or does not exhibit an ERP response. The larger
the degree of effort and level of frustration about the task they the positive or negative magnitude of the classifier value, the
performed. more likely the EEG response falls into either category. Thus,
a large positive classifier value more strongly suggests an ERP
occurred compared to a small positive classifier value. After
2.3. Peabody picture vocabulary testing
all selection boxes on the computer display had flashed at
We licensed the PPVT-IV from Pearson Education, Inc. for least once (called a flash sequence), our three-stage classifier
research purposes. The standard PPVT-IV contains 228 ques- entered its second stage called certainty.
tions separated into 12 sets of increasing difficulty. Each ques- We developed the certainty algorithm [31] (stage 2) to gen-
tion consists of a page with four illustrations in color. In the erate values corresponding to the probability that the subject
standard administration method, the examiner speaks a word is making a choice from the display. The certainty algorithm
when each question is presented. To respond, the subject must takes the SWLDA classifier values calculated for each flash
either point to or say the number of the illustration that best sequence in stage one and performs a t-test, then normalizes
matches the word spoken by the examiner [10]. the results. The outputs are the probabilities that a subject is
The test procedure involves identifying the subject’s basal making a selection, which we termed ‘certainty values’. To
and ceiling set. The basal set is identified as the first set the better estimate the certainty values of each selection box, we
subject completes with one or fewer incorrect responses. averaged the classifier values from different flash sequences
The starting set is based on age and is labeled the basal set for each selection. Averaging provides a better result than
if the subject meets the basal set criterion. Otherwise, the using only one ERP instance because it reduces the signal to
subject goes down one set at a time until they answer a set noise issues of EEG. In our application, we averaged up to five
with one or fewer errors. After determining the basal set, the of the most recent flash sequences. If certainty was reached
4
J. Neural Eng. 14 (2017) 066001 R E Alcaide-Aguirre et al
Figure 2. Example of hold-release confirmation step. On the left is the display during stage 1. After certainty is reached our system enters
the confirmation step. Then all illustrations are dimmed except the target selected by the certainty algorithm. Subjects can then continue to
select the target to confirm or select the cancel box (centrally located X label) to cancel their selection and try the question again.
before five sequences, we moved on to the next classification this case, the subject’s choice was assumed to be whichever
stage without waiting for more sequences. had the largest classifier value (figure 2).
In our application, we used the certainty algorithm as a To further increase accuracy, the hold-release algorithm
gatekeeper that prevented the BCI from making any decisions can be adjusted to prevent classification until the subject
until one of the selection boxes reached a certainty value of selects the target selection box a predetermined number of
90% [31]. In literature, this form of BCI is called an asyn- times (called number of times to verify). In the original hold-
chronous BCI since it prevents the BCI from making a choice release paper, two times to verify were used. In our study, four
until the subject is ready to respond. These steps minimize times to verify were used to increase BCI accuracy. In contrast
false positives and allow subjects to take their time to think to the original paper, we increased the times to verify because,
about which illustration they want to choose. Once a subject in real-time, classification accuracy dropped compared to the
has made his/her choice, he/she can then focus on the respec- original hold-release paper. We hypothesized that the decrease
tive selection box allowing the BCI to reach the 90% certainty in accuracy was because of the decreased number of items
threshold. Once the threshold was met, we labeled the selec- displayed to the user (11 versus 4).
tion box the subject choose the ‘target’, and our classification Two other variations (3 total variations) of the original
system entered stage 3; hold-release [23]. hold-release algorithm were used to test potential optimization
During the hold-release stage, we dimmed all illustrations methods. In the first variation, the third hold-release condition
except the target. At this point, the cancel box in the middle was ignored. Thus, the classification was not altered, even
of the screen began to flash in addition to the other selection when both illustrations had positive classifier values but were
boxes (figure 1). We asked subjects to continue focusing their still below the positive classification threshold.
attention on the selection box they chose (i.e. the target) if no In the second variation, the third hold-release condition
color change occurred on the illustration they were selecting. was applied when the target had a classifier value larger
If their illustration dimmed, they were instructed to focus their than the cancel box. Otherwise, the times to verify were not
attention on the cancel box. altered. This modification biases the BCI into choosing the
The hold-release algorithm produces a decision when any target, thus, increasing the speed of confirmation if the target
one of three conditions is met. The first condition uses as a was selected correctly initially.
threshold (called the positive threshold) the smallest classi-
fier value that separated ERPs from non-ERP. In the original
hold-release paper, this threshold was set to 99% accuracy 3. Analysis
difference between ERP and non-ERPs, determined from the
subject’s training data. In our study, the positive hold-release Across all subjects and both CP versus typically developing
threshold was set to the mean plus the standard deviation of (TD) groups, we calculated the mean and standard deviation for
the classifier values for the attended labels in the calibration the following measures: time/set, time/question, time/attempt
data. This represented a threshold that separated ERPs from at a question, time in classification stages 2 and 3; the number
non-ERP with 85% accuracy. We changed the method of set- of cancellations/question and the number of attempts/question.
ting the positive threshold to explore how a lower threshold The mean and standard deviation of the difference in the
would impact hold-release performance. If the classifier PPVT-IV scores for the two administration methods (standard
value of either the target or cancel box was above the positive and BCI-Facilitated) were calculated. The Pearson correlation
threshold, that was considered the choice of the subject. The between the scores was determined. NASA-TLX scores and
second condition was whether the target was a negative clas- the time required for test administration were evaluated using
sifier value. In this case, the cancel selection was classified paired t-tests.
as the choice of the subject. The final condition was invoked An ANOVA was used to test hold-release accuracy based
when both the target and cancel box had positive classifier on changes to the third hold-release rule, and a t-test was used
values, but those values were below the positive threshold. In to compare the accuracy of our 3-stage classifier (SWLDA
5
J. Neural Eng. 14 (2017) 066001 R E Alcaide-Aguirre et al
6
J. Neural Eng. 14 (2017) 066001 R E Alcaide-Aguirre et al
Note: summary of NASA TLX results of CP and TD means for perceived: mental demand, physical demand, temporal demand, perceived performance,
effort and frustration.
a,b,c,d,e,f
Correspond to statistical significance between each respective group. Entries with no symbols had no statistically significant differences.
believe the increase in mental demand and effort was because verification can become quite slow as the number of responses
the BCI-facilitated test required people to focus their atten- in a cognitive test increase. For example, in a two-choice test,
tion on making selections, compared to verbalizing a selection 2–3 actions are required to select, but if presented with six
as in the standard PPVT-IV. For populations without impair- choices (as is in Perego’s study), it may take the subject 2–7
ment or those that can take the standard test easily, such as actions or more to confirm a choice. These additional steps
those in our study, we expected the BCI-facilitated test to be break the flow of the assessment and may become frustrating
more challenging than simply replying verbally. The results of to a subject, leading to changes in assessment results. Using
our study support this as our BCI-facilitated assessment was hold-release allows for a more natural confirmation step com-
perceived as more physically challenging (but not mentally pared to using a secondary prompt to confirm a subject’s
challenging). However, we believe that for populations with choice. In our implementation, the subject only needs to pro-
severe movement and speech impairments for whom actual vide an additional response if their choice is being classified
physical movement is a great burden, the BCI-facilitated test incorrectly. Otherwise, the subject continues focusing on their
will be less challenging than the standard PPVT-IV, and per- choice until the BCI progresses to the next question.
haps the only accessible option. Other research groups have also developed asynchronous
There was no significant difference between the PPVT-IV BCIs. Typically, probabilistic models of ERPs, ERP ampl
scores of subjects with or without CP. However, on the NASA
itude, classifier values, SSVEP, or EEG power bands are used
TLX, subjects with CP reported significantly lower per-
to determine when a subject is making a choice [34]. Some
ceived performance for both the standard and BCI-facilitated
groups have also combined two methods to increase the reli-
PPVT-IV, suggesting that the CP subjects had lower confi-
ability of their asynchronous BCI. These hybrid systems typi-
dence than the TD subjects.
cally combine an ERP based method (probabilistic models of
Our 3-stage classifier significantly increased the accuracy
ERPs, ERP amplitude or classifier values) along with a fre-
compared to other classification methods we used (SWLDA
quency-based method (EEG power bands, spectral analysis or
and Certainty). Along with accuracy gains, our 3-stage clas-
SSVEP responses). Frequency-based methods rely heavily on
sifier also allowed the BCI to function asynchronously.
occipital electrodes to determine whether a subject is selecting
Asynchronous functionality allows subjects the time to think
a response with the BCI, making SSVEP BCIs less suitable
as much as needed to provide their best answer, while a con-
for people with CP [34–38]. Our P300 only method has the
firmation step reduces incorrect selections.
advantage of not requiring frequency-based analysis, reducing
Two other variations of the original hold-release algorithm
the likelihood of incorrect classification due to neck muscle
were used offline to test potential accuracy differences. In the
artifacts.
first variation, the third hold-release condition was ignored.
In our approach, we used our certainty algorithm for
In the second variation, the third hold-release condition was
asynchronous BCI functionality. Based on the classification
applied when the target had a classifier value larger than the
methods described above, we will now consider how our BCI
cancel box. When these changes were applied to both vari-
met the criteria we outline previously.
ations, there was a decreased accuracy for the hold-release
system compared to the original paper [23]. This is most 1. A cognitive assessment BCI should maintain the psy-
likely due to our comparatively lenient positive hold-release chometric properties of the standardized administration
threshold of 85% versus 99% compared to the original hold- procedure. Results from the difference analysis sug-
release paper. gest that our BCI-facilitated PPVT-IV yields adequate
Confirmation steps usually require a subject to respond to measurement agreement with the standard version of the
a secondary prompt or make another choice to confirm. To PPVT-IV, though more extensive analyses with larger
illustrate, Perego’s cognitive BCI used an indirect selection samples would be important in this regard.
method and a secondary response [6]. Subjects would first 2. Brain-based cognitive assessment systems must automat-
indirectly scroll through the possible choices and then provide ically abstract the complexity of brain activity analysis
a second command to confirm their final choice. This form of to provide results that are not difficult for the clinician
7
J. Neural Eng. 14 (2017) 066001
Flashes to
Time/ Time in stage Time in stage Sequences confirm or Number of Number of
Time/set Time/question attempt 2/attempt 3/attempt to reach certainty cancel cancelations/set cancelations/question Attempts/question
All Mean 338.21 ± 97.05 s 24.57 ± 17.41 s 18.58 ± 6.39 s 3.85 ± 4.28 s 6.26 ± 3.44 s 6.29 ± 6.76 12.13 ± 9.60 3.46 ± 3.28 0.29 ± 0.67 1.29 ± 0.67
sequences flashes
8
CP BCI mean 363.68 ± 108.41 s 26.70 ± 19.91 s 19.22 ± 6.76 s 4.03 ± 4.41 s 5.21 ± 3.36 s 6.63 ± 6.92 13.31 ± 10.63 4.24 ± 3.77 0.35 ± 0.76 1.35 ± 0.76
N = 15 sequences flashes
TD BCI mean 313.14 ± 77.35 s 22.37 ± 14.07 s 17.86 ± 5.86 s 3.64 ± 4.13 s 5.77 ± 4.01 s 5.91 ± 6.55 10.79 ± 8.07 2.69 ± 2.51 0.23 ± 0.55 1.23 ± 0.55
N = 11 sequences flashes
R E Alcaide-Aguirre et al
J. Neural Eng. 14 (2017) 066001 R E Alcaide-Aguirre et al
to interpret. Our adapted BCI provided an output that [3] Wagner B T and Jackson H M 2006 Developmental memory
matched the format of the standard PPVT-IV. Therefore, capacity resources of typical children retrieving picture
communication symbols using direct selection and visual
our approach meets the second criterion.
linear scanning with fixed communication displays J.
3. Brain-based cognitive assessment systems must be quick Speech Lang. Hear. Res. 49 113–26
to set up (1 h or less). While our current system does fall [4] Meyers A R and Andresen E M 2000 Enabling our
within an hour of setup, there were still subjects who instruments: accommodation, universal design, and access
could not complete the test due to the lengthy setup time. to participation in research Arch. Phys. Med. Rehabil.
81 S5–9
Most of the setup time was spent applying gel to each
[5] Byrne J M et al 1995 An innovative method to assess the
electrode. New dry electrode technology developed by receptive vocabulary of children with cerebral palsy using
companies such as Wearable Sensing have the potential event-related brain potentials J. Clin. Exp. Neuropsychol.
of removing this barrier and reducing setup time to less 17 9–19
than 10 min [32]. [6] Perego P et al 2011 Cognitive ability assessment by
brain–computer interface validation of a new assessment
4. Brain-based cognitive assessment system must have
method for cognitive abilities J. Neurosci. Methods
asynchronous control, thus allowing the subject to control 201 239–50
the pace of the assessment. Due to our certainty and hold- [7] Kaufman J et al 2013 A comparison of visual inspection time
release algorithms, we satisfied our fourth criterion. measures in children with cerebral palsy Rehabil. Psychol.
5. The BCI must be able to function in the population it is 59 147–54
[8] Warschausky S et al 2012 Modified test administration using
targeting. We tested our technology with people who have
assistive technology: preliminary psychometric findings
cerebral palsy and selected a BCI modality that appears to Assessment 19 472–90
function well in this relatively high-functioning popula- [9] Spaulding T J et al 2013 Investigating the interchangeability
tion. Before we can fully say we met our fifth criterion, and diagnostic utility of the PPVT-III and PPVT-IV for
testing should be done with people with a higher severity children with and without SLI Int. J. Speech Lang. Pathol.
15 453–62
of cerebral palsy.
[10] Dunn L M and Dunn D M 2007 Peabody Picture Vocabulary
Test (Minneapolis, MN: Pearson Assessments) 4th edn
[11] Birbaumer N et al 1999 A spelling device for the paralysed
6. Conclusion Nature 398 297–8
[12] Birbaumer N and Cohen L G 2007 Brain–computer interfaces:
Here, we presented a BCI that can administer the PPVT-IV, communication and restoration of movement in paralysis
J. Physiol. 579 621–36
a test of receptive vocabulary. Our BCI provided equivalent [13] Buch E et al 2008 Think to move: a neuromagnetic brain–
results to the standard PPVT-IV, suggesting that our BCI- computer interface (BCI) system for chronic stroke Stroke
facilitated PPVT-IV could be used for cognitive assessment in 39 910–7
populations for whom standardized tests are not accessible [37, [14] Cincotti F et al 2008 Non-invasive brain–computer interface
38]. There were some limitations with our system and method- system: towards its application as assistive technology
Brain Res. Bull. 75 796–803
ology. First, all subjects that went fully through our study were [15] Daly I et al 2013 On the control of brain–computer
able to take both standard and BCI adapted PPVT-IV. While interfaces by users with cerebral palsy Clin. Neurophysiol.
this allowed us to validate the measurement agreement of the 124 1787–97
system, future studies should focus on subjects with more [16] Allison B Z et al 2007 Brain–computer interface systems:
significant motor and speech impairments. Furthermore, our progress and prospects Expert Rev. Med. Devices
4 463–74
sample size was small, thus precluding more extensive psy- [17] Iversen I H et al 2008 A brain–computer interface tool to
chometric analyses of reliability and validity. Lastly, we only assess cognitive functions in completely paralyzed patients
tested our BCI on the PPVT-IV, which is an untimed multiple with amyotrophic lateral sclerosis Clin. Neurophysiol.
choice test. Different BCI adaptions would be required for 119 2214–23
time-sensitive assessments or assessments with different pre- [18] Aloise F et al 2013 Asynchronous gaze-independent event-
related potential-based brain–computer interface Artif.
sentation formats and response demands. While our method Intell. Med. 59 61–9
was only applied to the PPVT-IV, a multiple-choice format test [19] Leeb R et al 2007 Self-paced (asynchronous) BCI control of
with a quadrant stimulus array, our system can be extended a wheelchair in virtual environments: a case study with a
to other visual multiple-choice tests. Also, we demonstrated tetraplegic Comput. Intell. Neurosci. 2007 79642
a novel, natural confirmation step that significantly increases [20] Hwang H J et al 2012 Development of an SSVEP-based BCI
spelling system adopting a QWERTY-style LED keyboard
BCI accuracy without the need for a secondary prompt. J. Neurosci. Methods 208 59–65
[21] Lesenfants D et al 2011 Design of a novel covert SSVEP-
based BCI Proc. of the 5th Int. Brain–Computer Interface
References Conf. 2011 (Graz, Austria, 22 September–24 September
2011)
[22] Farwell L A and Donchin E 1988 Talking off the top of your
[1] Yin Foo R et al 2013 Intelligence assessments for children head: toward a mental prosthesis utilizing event-related
with cerebral palsy: a systematic review Dev. Med. Child brain potentials Electroencephalogr. Clin. Neurophysiol.
Neurol. 55 911–18 70 510–23
[2] Bagnato S and Neisworth J 1994 A national study of the social [23] Alcaide-Aguirre R E and Huggins J E 2014 Novel hold-release
and treatment ‘invalidity’ of intelligence testing for early functionality in a P300 brain–computer interface J. Neural
intervention Sch. Psychol. Quart. 9 81 Eng. 11 066010
9
J. Neural Eng. 14 (2017) 066001 R E Alcaide-Aguirre et al
[24] Palisano R et al 1997 Development and reliability of a system [31] Aref A and Huggins J 2012 The P300-certainty algorithm:
to classify gross motor function in children with cerebral improving accuracy by withholding erroneous selections
palsy Dev. Med. Child Neurol. 39 214–23 EEG Clin. Neurosci. Soc.
[25] Eliasson A C et al 2006 The manual ability classification [32] Slater J D et al 2012 Quality assessment of
system (MACS) for children with cerebral palsy: scale electroencephalography obtained from a ‘dry electrode’
development and evidence of validity and reliability Dev. system J. Neurosci. Methods 208 134–7
Med. Child Neurol. 48 549–54 [33] Donchin E et al 2000 The mental prosthesis: assessing the
[26] Hart S G and Staveland L E 1988 Development of NASA- speed of a P300-based brain–computer interface IEEE
TLX (task load index): results of empirical and theoretical Trans. Rehabil. Eng. 8 174–9
research Adv. Psychol. 52 139–83 [34] Nicolas-Alonso L F and Gomez-Gil J 2012 Brain–computer
[27] Carswell C M et al 2010 Hands-free administration of interfaces, a review Sensors 12 1211–79
subjective workload scales: acceptability in a surgical [35] Yin E et al 2014 A dynamically optimized SSVEP brain–
training environment Appl. Ergon. 42 138–45 computer interface (BCI) speller IEEE Trans. Biomed. Eng.
[28] Thompson D E and Huggins J E 2011 A multi-purpose brain– 62 1447–56
computer interface output device Clin. EEG Neurosci. [36] Li Y et al 2013 A hybrid BCI system combining P300 and
42 230–5 SSVEP and its application to wheelchair control IEEE
[29] Thompson D E et al 2013 A plug-and-play brain–computer Trans. Biomed. Eng. 60 3156–66
interface to operate commercial assistive technology [37] Yin E et al 2013 A speedy hybrid BCI spelling approach
Disabil. Rehabil. Assist. Technol. 9 144–50 combining P300 and SSVEP IEEE Trans. Biomed. Eng.
[30] Krusienski D J et al 2006 A comparison of 61 473–83
classification techniques for the P300 speller J. Neural Eng. [38] Wang Y et al 2006 A practical VEP-based brain–computer
3 299–305 interface IEEE Trans. Neural Syst. Rehabil. Eng. 14 234–40
10