ArticlePDF Available

Abstract and Figures

While we instantaneously recognize a face as attractive, it is much harder to explain what exactly defines personal attraction. This suggests that attraction depends on implicit processing of complex, culturally and individually defined features. Generative adversarial neural networks (GANs), which learn to mimic complex data distributions, can potentially model subjective preferences unconstrained by pre-defined model parameterization. Here, we present generative brain-computer interfaces (GBCI), coupling GANs with brain-computer interfaces. GBCI first presents a selection of images and captures personalized attractiveness reactions toward the images via electroencephalography. These reactions are then used to control a GAN model, finding a representation that matches the features constituting an attractive image for an individual. We conducted an experiment (N=30) to validate GBCI using a face-generating GAN and producing images that are hypothesized to be individually attractive. In double-blind evaluation of the GBCI-produced images against matched controls, we found GBCI yielded highly accurate results. Thus, the use of EEG responses to control a GAN presents a valid tool for interactive information-generation. Furthermore, the GBCI-derived images visually replicated known effects from social neuroscience, suggesting that the individually responsive, generative nature of GBCI provides a powerful, new tool in mapping individual differences and visualizing cognitive-affective processing.
Content may be subject to copyright.
This is the accepted version of the journal article, published in IEEE Transactions on Affective Computing. For the final
version, please see DOI: 10.1109/TAFFC.2021.3059043
© 2021 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in
any current or future media, including reprinting/republishing this material for advertising or promotional purposes,
creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of
this owkr in other works.
Brain-computer interface for generating
personally attractive images
Michiel Spap´
e, Keith M. Davis III, Lauri Kangassalo, Niklas Ravaja, Zania Sovij¨
arvi-Spap ´
and Tuukka Ruotsalo
Abstract—While we instantaneously recognize a face as attractive, it is much harder to explain what exactly defines personal
attraction. This suggests that attraction depends on implicit processing of complex, culturally and individually defined features.
Generative adversarial neural networks (GANs), which learn to mimic complex data distributions, can potentially model subjective
preferences unconstrained by pre-defined model parameterization. Here, we present generative brain-computer interfaces (GBCI),
coupling GANs with brain-computer interfaces. GBCI first presents a selection of images and captures personalized attractiveness
reactions toward the images via electroencephalography. These reactions are then used to control a GAN model, finding a
representation that matches the features constituting an attractive image for an individual. We conducted an experiment (N=30) to
validate GBCI using a face-generating GAN and producing images that are hypothesized to be individually attractive. In double-blind
evaluation of the GBCI-produced images against matched controls, we found GBCI yielded highly accurate results. Thus, the use of
EEG responses to control a GAN presents a valid tool for interactive information-generation. Furthermore, the GBCI-derived images
visually replicated known effects from social neuroscience, suggesting that the individually responsive, generative nature of GBCI
provides a powerful, new tool in mapping individual differences and visualizing cognitive-affective processing.
Index Terms—Brain-computer interfaces, Electroencephalography (EEG), Generative Adversarial Networks (GAN), Image generation,
Attraction, Personal preferences, Individual differences.
WHAT is beauty? Although in daily life we can instantly
judge whether a picture looks attractive, we com-
monly find it hard to explain the reasons behind such a deci-
sion and harder still to create beauty without extensive skill
and experience. The difficulty in describing and portraying
attractiveness stems from two interrelated problems. First,
ratings of attractiveness vary significantly between individ-
uals, for example in terms of age, culture, and gender [13],
[45], [68]. Second, describing what one finds aesthetically
pleasing requires awareness of what is thought to be an
implicit evaluation of a complex configuration of features
[34]. Therefore, attractiveness is rather a subjective, personal
characteristic than a objective, visual feature: Beauty is in the
eye of the beholder.
The creation of personally attractive images has thus far
been a challenge due to the complex nature of attractiveness.
Previous studies sought to determine the attractiveness of
a face by relying on simple, predefined features computed
from a picture [49], [56], [63], [65], but in so doing have
likely underestimated the complexity of attractiveness judg-
ments. Models relying on hand-crafted features through
simple methods (e.g. length of nose measurements, golden
K. M. Davis, L. Kangassalo, and Z. Sovij¨arvi-Spap´e are with the Depart-
ment of Computer Science, University of Helsinki, Finland. T. Ruotsalo
is with the Department of Computer Science, University of Helsinki,
Finland and Department of Computer Science, University of Copenhagen,
M. Spap´e and N. Ravaja are with the Department of Psychology and
Logopedics, University of Helsinki, Finland.
ratio, symmetry) or more higher-level algorithms (e.g. Ga-
bor wavelet transformations) do not reflect an individual’s
understanding of aesthetics [34]. Thus, such approaches fall
short in modeling psychologically relevant sources of attrac-
tiveness and cannot enable inverse inference by producing
novel, attractive images. This is because the automatically
captured features are limited to the observable distribution
of images within the models, which likely represents only
a small portion of the true distribution of features that
constitute variance in the visual image features.
Besides attractiveness being a complex, personal per-
ceptual decision, it may also be better judged implicitly
than explicitly. Humans typically respond emotionally to
attractive images [9] rather than purely on the basis of
rational, visually salient reasons. In order to generate attrac-
tive images, an ideal system should therefore rely on early,
implicit responses rather than explicit ratings. This could be
implemented as a physiologically adaptive system, such as a
brain-computer interface (BCI), which would utilize implicit
signals to generate personally attractive images. However,
the aforementioned complexity problem limits the use of
BCIs in this field. For example, if one face is implicitly
evaluated as more attractive than another, it will likely differ
in multiple ways. How do we infer which of the features is
important, and how can we generate another face that is
expected to be similarly attractive to the target face?
Here, we present a new paradigm, which utilizes implicit
reactions to perceived attractiveness to steer a generative
adversarial network (GAN) [30], thereby producing images
that are expected to be personally attractive. We refer to the
approach as generative brain-computer interfacing (GBCI).
By training a GAN with images, it learns to mimic the
underlying visual distribution, which enables us to draw
new, unobserved samples from that distribution [7], [38],
[47]. GBCI unites a GAN with BCI: in response to a series of
evoked brain responses to images, the GBCI iteratively pro-
duces novel (previously unseen), photorealistic images that
match a user’s individual aesthetic preferences. As further
explained in Figure 1, the GBCI works by classifying brain
activity evoked by faces being perceived as either attractive
or unattractive. Each face is represented as a coordinate
within the GAN space, so that with multiple attractive
faces being detected, we can triangulate GAN vectors that
are expected to be subjectively attractive. This expected-
attractive localization is iteratively updated whenever more
evidence is detected, each time producing a novel image
for the position. The final position is expected to match
the participant’s sense of personal attractiveness, which we
empirically test in the present study.
We report an experiment with 30 participants to validate
the GBCI in its ability to generate attractive faces. A GAN
model trained with celebrity faces was used to create a
sample of fictional faces. These were then presented to
each participant while their EEG was recorded. The GBCI
paradigm was then applied to iteratively generate the im-
age that was predicted to be the best match for a user’s
personal attraction. To empirically test the GBCI’s efficacy,
we requested users to blindly evaluate the best personal
match, expecting higher personal attractiveness ratings than
for control images.
In summary, we present generative brain-computer in-
terfacing for generating personally attractive images:
1) This is, to our knowledge, the first successful ap-
proach utilizing brain responses as an interactive
feedback to a generative neural network.
2) The approach was validated in a face generation
task and found to generate novel, personalized, and
highly attractive images.
In the following section, we will first review the current
state of research on the psychology of attraction before we
focus on brain processes involved in aesthetic judgements.
We then move our focus to research showing that general,
but not personal, attraction can be predicted from computer
vision algorithms. Likewise, the brain processes reviewed
earlier can be harnessed to enable automatic detection of
aesthetic relevance. In the subsequent section, we will pro-
pose GBCI as a new approach to unite the psychology of
aesthetics, the cognitive neuroscience of personal attraction,
and the computer science of generative adversarial models.
2.1 The psychology of personal attraction
The study of aesthetics, or the perception of beauty and
experience of attractiveness, has a long tradition within
psychology and related disciplines. Despite the common
idea that taste is intensely individual, psychological research
consistently shows a strong consensus on the visual features
that are considered attractive [46]. Symmetry in faces is
known to be seen as attractive, perhaps because symmetry
in general is an important evolutionary signifier. Visual
symmetry, for example, may point towards the nearby pres-
ence of fruit, flowers, and animals [70]. Indeed, it is even
thought that positive affect is a consequence of the com-
putational ease of processing due to symmetry necessarily
including a redundant set of visual features, streamlining
perception [55]. Another common theory in evolutionary
psychology is the sexual dimorphism account of facial at-
tractiveness, which holds that feminine displays in females
and masculine in males are attractive by signifying mate
quality [41]. It seems, however, that evolutionary factors
shaping visual feature perception do not entirely predict
attraction: Computer simulations that optimize averageness
or sexual dimorphism generate faces that are judged attrac-
tive but not maximally attractive faces for individuals [61].
The common consensus on what is beautiful notwith-
standing, individual differences in attractiveness judgments
do exist. Interestingly these differences have been found to
be larger for female ratings of male faces than the other
way around [36]. It is clear, however, that attractiveness
is not solely due to biological, genetic factors: Subjective
levels of attractiveness vary widely as a function of social
learning [49]. Typically, debates of biological and cultural
determinants of individual differences in faces are presented
in nature-vs-nurture terms, but cultural differences do not
negate evolutionary theory, or vice versa. Indeed, Darwin
himself already noted the tremendous differences in what
people find beautiful [14]. A lack of common consensus
in beauty may itself be adaptive, as variations within and
between cultures present an evolutionary advantage [50].
Yet, however much people differ in what they find
personally attractive, cognitive neuroscience suggests their
brains process attraction in very similar ways. Conse-
quently, if the cognitive and affective processes underlying
aesthetic relevance are comparable between individuals,
then it should allow us to detect whenever a stimulus is
deemed attractive, which can then be used as the GBCI’s
implicit measure. In the next section, we will discuss the
relevant literature on the neural response to detecting at-
tractive stimuli.
2.2 The cognitive neuroscience of personal attraction
The event-related potential (ERP) technique within elec-
troencephalography research presents a useful method for
detecting when a user perceives a beautiful face. ERPs are
electrophysiological recordings of brain activity occurring
in response to known physical events. Since EEG provides
high temporal precision, it is possible to functionally disso-
ciate perceptual from cognitive operations. Thus, by detect-
ing whether an ERP is in response to an attended letter,
the classic brain-computer interface (BCI) allows people
to spell letters using EEG [25]. More controversially, this
approach was extended to situations in which participants
would rather hide their thoughts, but failed to disguise
their implicit guilt by having guilty knowledge [24] (but see
[53]). Due to this enticing possibility that hidden evaluations
might be detected from EEG, ERP research is more and more
conducted in the context of human-computer interaction
research [66].
In the present work, we focus on using properties of
the ERP that can inform us as to the aesthetic preference of
users. As preference entails a stimulus being found intrin-
sically relevant, we targeted the ERP component that has
been particularly associated with relevance detection, the
P300. The P300 in general is characterised by a late parietal
positivity occurring from ca. 300 ms after the onset of stimuli
in any modality if they are infrequent, attended, novel, and
relevant [12].
Later research, however, suggested the P300 can be
further divided into multiple subcomponents, of which at
least three seem related to aesthetic relevance detection.
The P3a predominantly affects frontal-central electrodes and
usually leads the other subcomponents in terms of latency,
and has functionally been related to exogenous relevance,
responding to stimuli that are novel or affectively evocative
[11]. The P3b, meanwhile, is the more commonly targeted
component in EEG research, and has generally been related
to top-down, endogenous, task-relevance related attention,
being modulated if stimuli of any modality [31] require
additional processing [42]. As such, it has theoretically
been seen as corresponding to a neurodynamic mechanism
related to working memory updating [19], or a process in
between attention and further memory processing [58]. It is
also the more reliable part of the P300, especially if stimuli
are improbable and require a mental or physical response
[57], [73]. Consequently, this component is often the one
targeted by BCIs, such as with the classic BCI-speller [25]
or applications utilizing relevance effects [22], [23], [35].
Finally, a further sub-component following the P3a and
P3b is sometimes referred to as the LPP, the late positive
potential, which has particular importance to attractiveness
research as it was shown enhanced on seeing beloved part-
ners [44], and aesthetically pleasing images [32].
To maximize the degree to which implicit attractiveness
evoked P300s, we employed three strategies based on the
literature. First, attractive images were made relatively im-
probable by presenting images of both the participants pre-
ferred and non-preferred gender. Thus, for the heterosexual
majority, fewer than 50 % of faces were both of preferred
gender and attractive. Second, we requested participants to
select unattractive images as reminders. These were shown
to the left and right of a screen to focus users on the
central, target image. Third, we asked participants to focus
particularly on attractive images by mentally counting their
occurrence. This showed that only about 1 in 5 images was
found attractive, matching relevant probability in traditional
BCIs [25].
2.3 Affective computing of personal attraction
Our work aims to detect personal attractiveness of images
based on implicit brain signals in order to optimize a pre-
diction of personal attraction. Our neuroadaptive interface
is designed to detect and predict personal attraction to vi-
sual images, which is related to the affective computation
of attractiveness. Traditionally, computer vision has been
used to extract complex features related to aesthetics from
images and using these to predict user evaluations. The
attractiveness of images in general has been done based on
automatic computation of low-level features, such as quality
[64], or more cognitively significant features, such as based
on Gestalt principles, rules of thirds, and visual weight [15],
[40], [52]. For images of faces, visual features such as nose-
to-forehead and nose-to-chin ratio were found to predict
how attractive a face is found with an accuracy of about
25% [20]. Recent years have enhanced the efficacy of this ap-
proach by extracting a combination of visual features from
image input and applying more advanced machine learning
algorithms to predict general attractiveness and affective
ratings [2], [3], [28], [62]. However, extracting invariable fea-
tures from a stimulus input necessarily optimizes prediction
of general attractiveness, dismissing interindividual variance
in what counts as attractive as merely noise.
Instead of solely relying on the objective visual features
of images, our work aims to infer what a particular indi-
vidual finds personally attractive and therefore targets the
subjective qualities evoked by an image. Here, we rely on
the growing literature on neuroadaptive computing [43],
which has previously focused on adapting a system via
neurofeedback in a very limited set of pre-defined states.
Examples of such neurofeedback systems include changing
the difficulty of a learning task to avoid overwhelming the
user’s cognitive capacities [75], or transforming a player ’s
character based on brain signals [72]. In contrast, our work
is based on the neuroadaptive framework by [37] towards
optimization of an inference of personal attraction within a
complex, multidimensional space represented using a gen-
erative adversarial network (GAN). Adapting the hypothet-
ical ”best guess” of what each individual finds attractive,
iteratively retrieving this point within the GAN generates a
visualization of the personal attraction.
To summarize, neuroadaptive systems have shown solid
promise in adapting an inference based on brain activ-
ity. The advent of GANs now enables extension of this
framework by allowing complex adaptations within ill-
constrained problem spaces [71]. Additionally, this provides
the possibility to generate photorealistic, yet artificial im-
ages of human faces, producing a novel window into mental
processes. In other words, a generative BCI is a neuroadap-
tive system with a deep-learning representation of image
data operating on individual users, which we believe to
represent the next natural step in predicting and visualizing
personal attraction.
Personal attraction as modeled via generative brain-
computer interfacing consists of four phases, reflecting the
example illustrated in Figure 1. In this section, we first
provide a general overview of how the GBCI functions. In
the following four subsections, we formally define each of
these phases using mathematical terms.
A GAN is first trained (A) using the CelebA-HQ dataset.
Next, a participant is asked to assess images randomly
sampled from the GAN while their EEG is recorded; a
classifier is then trained to associate their EEG with their
subjective assessments of the images (B). After calibration,
new images sampled from the latent space are shown to the
participants while their EEG is recorded. The EEG signals
are then classified to determine which images the partici-
pants found attractive (C). In the example, images 1, 3, and
4 are found attractive and image 2 is not found attractive.
The GBCI then tries to find optimal values for latent features
Fig. 1: The GBCI approach. A: A GAN model with generator Gand discriminator Dis trained using ca. 200k images of
celebrity faces, resulting in a 512-dimensional latent space from which sampled feature vectors used as Generator input
produce artificial images; B: Participants are shown images produced from sampled feature vectors while their EEG is
measured; Following, they are shown the same images and select based on personal attractiveness; These collected data
are then used to train an LDA classifier for each participant; C: Participants are shown new images produced using the
same generative procedure as in B; Now, their measured EEG responses are classified as attractive/unattractive using their
personal classifier; D: New images are generated from the latent representations (i.e. feature vectors) of images labeled by
the classifier as attractive. An image G(ˆz), estimated as personally attractive, is iteratively generated as more images are
classified as attractive and their combined feature vectors ziare used as inputs for the Generator.
within the GAN model that encode for attractive features.
Finally, a new image containing optimal values for attractive
features and non-attractive features is generated (D). The
key idea is that certain stimuli images have some features
that the participant finds attractive, but these features are
not necessarily all contained within a single stimulus image.
On the other hand, some images contain features which the
participant may find unattractive. In the example, image 2
is an image of a male that is not attractive for the participant
and thus has some features that are to be avoided. Now, the
GBCI is able to parameterize a vector that combines features
from the latent vectors of images 1, 3, 4, and features not
in the latent vector of image 2. The resulting latent vector
then corresponds to an image that is hypothesized to be
maximally attractive for the participant.
3.1 Phase A: GAN Training
First, a latent image space is created by training a progres-
sively growing generative adversarial network [38]. This
results in a 512-dimensional feature space Z. The generative
model provides a mapping G:ZX, where zZis
a point in the latent feature space Z, and xXis an
individual face in the set of faces. The goal is to produce
a feature space where it is possible to select a point ˆznZ,
for which Gzn) = ˆxn, and determine if it matches the
attractiveness criteria of the participant.
3.2 Phase B: GBCI Calibration
Next, the feature space Zis sampled to produce images of
artificial faces using the GAN architecture. These images
are presented to a participant. The brain activity Sn=
{s1, ..., sn}associated with the images is used to calibrate
a regularized Linear Discriminant Analysis (LDA) classifier
[5] with shrinkage chosen with the Ledoit-Wolf lemma [48].
The classifier learns a classification function f:SY,
where Yis a binary value discriminating target/non-target
stimuli. In this case, detecting that there were some features
in the presented face xithat was attractive for the partici-
3.3 Phase C: GBCI Classification
In Phase C, a set of nimages Xn={x1, . . . , xn}that
were not in the training set of the classifier are generated
from a set of latent representations Zn={z1, ..., zn}. This
information is displayed to the participant, whose brain
responses evoked by the presented information are mea-
sured. These responses are then classified using the trained
classifier, and their associated presented images xiand the
latent vectors ziused to generate these images are assigned
labels corresponding to the classifier outputs. The classifiers
are personalized and a separate classifier is trained per par-
ticipant. In the example case presented in Figure 1, images
x1,x3and x4are found to be attractive, while image x2
is not found attractive. This results in a set of images xi,
their associated latent vectors zi, and a set of binary labels
of whether an image is found to be attractive. Let us refer to
the set of latent vectors classified as attractive as ZP OS .
3.4 Phase D: GBCI Generation
Finally, in phase D, ˆznis updated by using a simple model
updating function h:Z, Y Zto generate a final image
from the latent model. Formally, this average vector was
computed with ˆzn=1
|ZP OS |PzjZP OS zj. This updating
procedure is a special case of the Rocchio algorithm [60].
The resulting ˆznis then used as an input to the generator
of the latent GAN model Gto generate a new image,
Gzn) = ˆxn. The resulting image represents the point in
the latent space that is a novel, unseen face image, which is
expected to contain the attractive facial features. The process
starts from the first positively classified image, such that the
vector ˆzis initialized with the corresponding latent vector
ˆz0of that positively classified image
To evaluate the approach, we present an experiment that
tested whether GBCI could generate images that were
evaluated as personally attractive by 30 participants. The
experiment was run in two stages. In stage I, the pre-
trained GAN was used to produce 240 face images that were
used as stimuli. Participants viewed the images in a rapid
serial visual presentation (oddball) paradigm, particularly
concentrating on attractive images (relevant targets). Based
on their data, we trained a classifier to detect relevance from
their brain responses. This information was then used as a
positive model for generating individual, novel images that
were expected to be personally attractive to the participants.
In stage II, the participants explicitly evaluated the gener-
ated images randomly placed along with matched controls
in a blind test. We hypothesized that the positive-model
generated images would be evaluated as more personally
attractive than matched controls.
4.1 Participants
Thirty-one volunteers were recruited from the student and
staff population of the University of Helsinki. During the
recruitment procedure, volunteers were informed that par-
ticipation required they state their sexual gender preference.
They were fully informed as to the nature of the study, and
signed informed consent to acknowledge understanding
their rights as participants in accordance with the Decla-
ration of Helsinki, including the right to withdraw at any
time without fear of negative consequences. One volunteer
did withdraw due to lack of time and was removed from
data analysis. The full sample thereafter included 17 males
and 13 females, aged 28.23 (SD = 7.14, range = 18 to 45)
years on average. The study was approved by the University
of Helsinki’s Ethical Review Board in the Humanities and
Social and Behavioural Sciences. Participants received one
cinema voucher for participating in the acquisition phase
of the experiment, and two more for completion of the
validation phase.
4.2 Stimuli
A pre-trained Generative Adversarial Network (GAN) [38]1
was used to generate all stimuli used in this study. The
GAN was pre-trained with the CelebA-HQ dataset, which
consists of 30 000 1024 ×1024 images of celebrity faces.
The CelebA-HQ dataset is a resolution-enhanced version
of the CelebA-dataset [51]. The generator part of the GAN
provided a mapping from a 512-dimensional latent space to
a1024 ×1024 image. Only visual features (i.e. pixel data)
from the dataset were used to train the GAN. Other data
included in the Celeb-HQ dataset, including manually an-
notated labels describing various features contained within
an image, were NOT used in any way to train the GAN
Training images were initially generated via a random
process that sampled latent vectors from a 512-dimensional
multivariate normal distribution. These were then used to
produce corresponding images with the GAN generator.
Following, images were manually categorized by a human
(female) assessor into male and female-looking faces, with-
out regard for other visual attributes (e.g. age, ethnicity,
emotional expression), other than looking convincing and
being without significant artifacts. We then selected the first
120 male and 120 female faces for use in the present study.
To standardize the images with regards to face-unrelated
attributes such as the background, we removed the sur-
rounding area of the original 1024 x 1024 sized images using
a 746 x 980 pixels elliptic mask, replacing this with uniform
gray. To further improve presentation timing accuracy, we
then downsampled the images by a factor of 2. The images
were displayed on a 24” LCD monitor placed ca 60 cm from
the participant, running at 60 Hz with a resolution of 1920 x
1080, its timing and EEG synchronization optimized using
E-Prime [67].
For the evaluation procedure, three sets of stimuli were
generated: positive, negative and random images. Positives
and negatives were generated by averaging the latent vec-
tors representing images detected as attractive or unattrac-
tive, respectively. The matching controls were generated
similarly, but by averaging over randomly chosen latent vec-
tors. This procedure simulated a random-feedback classifier.
The formal averaging and generation process is provided in
the Image generation section (4.6).
1. Source code and pre-trained models:
progressive growing of gans
Fig. 2: Data acquisition procedure. During the RSVP, 8 x 60
images were presented at a rate of 2 stimuli / s. Following
the RSVP, users provided explicit feedback by clicking on
attractive images.
4.3 Stage I: Data acquisition procedure
In stage I of the experiment, participants viewed the images
in a rapid serial visual presentation paradigm. Following
EEG setup and signing of forms, the experiment was started
by the lab assistant. The participants were asked in private
to provide their sexual gender preference with options
given between male, female, or either. The experiment in-
cluded two parts: a presentation and a feedback. During
the former part, participants undertook 8 rapid serial visual
presentation (RSVP) trials. Each trial started by displaying 4
random images of a male or female face (order randomised),
asking participants to click on the face they found the least
attractive. This image was subsequently used as mismatch
”flanker”. Following this, an instruction screen appeared to
remind the participants of the task, which was to concen-
trate on attractive images by keeping a mental count. After
acknowledgment, the flankers were presented left and right
of the central, target location, for 1 s before the RSVP started,
which involved sequential presentation of 60 (30 male, 30
female) images being presented in the central location at a
rate of 2 per second without inter-stimulus interval. After
the last image, participants were requested to enter the
number of attractive images using the keyboard. One block
had 8 trials, such that after a block (480 images) all 240
images were presented twice, once flanked by males, and
once by females.
The feedback part was started after each of the three
blocks. Here, participants selected, or ”voted for”, images
they found attractive. Participants were shown all 240 im-
ages in random order, using four screens of 60 small buttons
laid out in 5 rows of 12 images, and asked to click on
the ones they had previously found attractive. To avoid
changes in decisional preference criterion, an estimate was
shown in the corner of the screen, indicating the number
of images they had provided previously, counting down
as participants selected images. After 3 blocks, taking on
average 37.15 (SD = 6.97) minutes in total, the experiment
was complete.
4.4 EEG acquisition and preprocessing
A BrainProducts QuickAmp USB was used to record EEG
from 32 Ag/AgCl electrodes positioned at equidistant lo-
cations of the 10-20 system by means of an elastic cap,
using a single AFz electrode as online reference. The time
series voltage amplitudes were then digitised with Brain
Vision Recorder running at a samplerate of 1000 Hz, with
a high-pass filter at 0.01 Hz, and a re-referencing to the
common average reference. Furthermore, two pairs of bipo-
lar electrodes – one pair placed lateral to the eyes and
the other above and below the right eye – were used to
capture EOG. Offline preprocessing included application of
a band-pass filter between 0.2 – 35 Hz to remove slow signal
fluctuations and line noise, after which the data were time-
locked to stimulus-onset and segmented into 900 ms epochs
of post- and 200 ms of pre-baseline activity. After removing
the average baseline activity from each channel, epochs
contaminated by artefacts such as eyeblinks were tagged
with an individually adjusted, threshold-based heuristic. As
a result, approximately 11% of each participants’ epochs
with the highest absolute maximum voltage was removed
from analysis. In order to speed up classifier training proce-
dures, the data were decimated by a factor of four. The final
dataset consisted of on average 1265 (SD = 109) epochs per
Standard feature engineering procedures were followed
to form a vectorized representation of the EEG data [5].
After preprocessing, the measured scalp voltages of each
participant were available as a Xn×m×ttensor, with n
epochs, mchannels and tsampled time points . Each epoch
was split into t0= 7 equidistant time windows on the 50
- 800 ms post-stimulus period, and the measurements in
each window were averaged. To generate spatio-temporal
feature vectors all available channels and the t0averaged
time points were concatenated, resulting in a data matrix
Xn×m·t0, where m·t0= 32 ·7 = 224.
4.5 Classifier
The attractiveness of faces was predicted in a single-trial
ERP classification scenario [5]. In detail, a regularized Linear
Discriminant Analysis (LDA) [27] classifier with shrinkage
chosen with the Ledoit-Wolf lemma [48] was trained for
each of the participants with the vectorized ERPs. LDA has
been shown to perform robustly for EEG classification [5].
Also included were binary labels indicating attractiveness
of the faces associated with the vectorized ERPs (attrac-
tive/unattractive). The label was assigned based on the
attractiveness votes (clicks) of each face, which were given
by the participant during the feedback phases: zero votes
for a given face labelled the face as unattractive, while two
or three votes labelled the face as attractive. Faces with
only one vote were deemed to be of unknown attractiveness
and removed from further analysis. This led to the average
participant having 1168 (SD = 123) data points prior to
splitting the datasets to training and test sets. The split was
done so that the first 80% of the data points were used for
training and the remaining 20% for testing. The test set thus
contained approximately 233 ERPs per participants.
After the training, the vectorized ERPs in the test set
were assigned to either the attractive or unattractive class
based on classifier confidence. For this purpose, two per-
participant thresholds were computed: one for the attrac-
tive class (positive prediction threshold) and one for the
unattractive class (negative prediction threshold); a face was
predicted to be attractive if the classifier confidence for the
attractive class exceeded the positive prediction threshold
and unattractive if it fell below the negative prediction
threshold. To ensure that the images generated from the
unattractive and attractive predictions were of equal quality,
the thresholds were computed so that the amount of positive
and negative (attractive/unattractive) predictions made by
the classifier for a participant were equal. This resulted in
a grand average of 41.13 (SD = 42.36) positive and negative
predictions by the classifiers.
The classifier performance was measured with an
Area Under the ROC Curve (AUC), and evaluated by
permutation-based p-values acquired by comparing the
AUC scores to those of classifiers trained with randomly
permutated class labels [54]. For each participant, n= 100
permutations were run, meaning that the smallest achiev-
able p-value was 0.01 [29].
4.6 Image generation
Using the same GAN architecture that produced the face
images used as stimuli (see 4.2), five different configu-
rations of the GBCI model were designed for generating
the evaluation images. The first configuration, POS, was a
positive feedback model that used only the latent vectors
of the face images classified as attractive AP OS . The second
configuration, NEG, used only negatively classified vectors
ANE G (i.e. the latent vectors of the face images classified as
unattractive). The third configuration, RND, used random
feedback ARND, where the labels for vectors used for the
POS and NEG models were shuffled. The fourth config-
urations, POS-NEG, subtracted the latent vectors used in
the NEG model from the latent vectors used in the POS
model. The fifth configuration, NEG-POS, subtracted the
latent vectors used in the POS model from those of the NEG
Operationalizing our hypothesis towards these models,
we expected that the POS model would generate faces that
were evaluated as more attractive than those generated by
the RND and NEG models. Furthermore, we expected that
the NEG model would produce faces that were evaluated
as less attractive than the baseline provided by the RND
model. As we had no clear a-priori hypothesis with regards
to the POS-NEG model, or the NEG-POS model, we left
these out of the confirmatory tests and analysis.
4.7 Stage II: Image evaluation procedure
In stage II, a blind evaluation procedure was used to test
the hypothesis that GBCI generated images were more
personally attractive than matched controls. Two months
after initial participation, we recalled the participants for
the follow-up validation procedure, in which they evaluated
their custom-generated images. Single generated images
from each of the models described in 4.6, along with 20
matched controls generated from the RND model, were em-
pirically tested for personal attraction using two tasks that
were presented in a set order. Following, an interview was
conduced with the participants in order to obtain qualitative
In the Free-selection task, the images were simultaneously
presented in 2 rows of 12 (similar to the feedback phase
described earlier) randomly arranged buttons, and partic-
ipants were requested to click on all which they found
attractive. We analyzed the percentage of times images
expected to be found personally attractive and unattractive
were selected.
In the explicit evaluation task, the 24 images were se-
quentially presented in random order, and participants
were requested to rate the attractiveness of each using a
1 (very unattractive) to 5 (very attractive) Likert-type scale.
To distinguish personal preference from general judgments
related to cultural norms or demand characteristics, we
asked participants to perform the task twice. During the
second run, we asked participants to estimate how attractive
the general population, given compatible orientation, would
rate the person. Thus, the explicit evaluation provided both
measurements of personal attractiveness, and estimations
of population attractiveness. We analyzed the average rat-
ings for the three types of images – expected to be attrac-
tive, unattractive, and neutral – using repeated measures
Finally, the personal predictions were revealed to the
participants and a semi-structured user interview was con-
ducted with the aim of determining whether participants
felt the generated images matched their personal attraction.
Guiding interviewees to reflect on the process of the study,
we explored the phenomenology of attraction in the context
of the experiment using thematic analysis [6]. In particular,
the participants reflected on the epistemology of attraction
in terms of how they defined and experienced attraction.
The interview was digitally recorded and answers were
transcribed for 8 randomly selected users. The complete
validation procedure, including free selection and explicit
evaluation tasks took ca. 20 minutes.
The generative brain-computer interface (GBCI) used event
related potentials (ERPs) to create an attractiveness classi-
fier, which was then used to generate images that were
empirically tested for matching personal attraction. In the
results, we first present the generalized effect of perceiving
attractive images on ERPs to confirm the expected pattern
was visible on the P3. However, this general analysis did not
affect individualized classifiers, the effectiveness of which
we describe in the subsequent section. The classifier was
then used to generate a set of novel images that were
hypothesized to match (positive generated images) and
not match (negative generated imagers) personal attraction.
The third section presents the results of the empirical test
of the GBCI to produce personally attractive images. The
final section summarizes how participants experienced the
decision-making process and the perceived efficacy of the
5.1 ERP results
ERPs to images voted as unattractive, attractive, and incon-
sistent were averaged per participant and analyzed for Fz
and Pz channels. As effects were predicted for both P3a
(commonly earlier and frontal), and P3b (usually parietal
and later) potentials, we first performed a confirmatory
Fig. 3: Grand average ERP from the Fz (top) and Pz (bottom)
showing evoked responses to faces consistently deemed
attractive (green) or unattractive (red) and inconsistently
rated faces (grey). Pink lines to the bottom of each graph
show significant differences between conditions (Bonferroni
corrected p <.05). Scalp topographies displaying the differ-
ence between unattractive (left, red), and attractive (right,
green) between 250-500 ms is shown below the panel.
analysis on the average amplitude between 250-350 for Fz
(P3a) and between 350-500 for Pz (P3b) using two repeated
measures ANOVAs with attractiveness (unattractive, incon-
sistent, attractive) as factor and component (P3a, P3b) as
measures to replicate the effect that observing attractive,
relevant images evokes a predictable pattern on average.
A significant effect of attractiveness was observed for the
P3a, F (2, 58) = 15.51, MSE = 0.33, p <.0001, η2= .12. Post-
hoc comparisons showed inconsistent and attractive images
evoked larger P3as than unattractive images, ps<.01,
and that attractive images evoked higher amplitudes than
inconsistent ones, p = .01. The effect was also significant
for the P3b, F (2, 58) = 51.24, MSE = 0.53, p <.0001, η2=
0.50. Again, attractive images significantly amplified the P3b
relative to unattractive, p <.001, and inconsistent images, p
<.001. Here, inconsistent showed P3bs roughly in between
unattractive and attractive images, as it was found to also
evoke amplified P3bs vs unattractive images, p = .005.
To provide a more comprehensive analysis, we further-
more explored the univariate effect of attractiveness on the
entire time-series of Fz and Pz activity using two windowed
repeated measures ANOVAs with the average amplitude of
Fz and Pz between 100-600 ms in bins of 20 ms as measure.
Figure 3 shows the result of these tests with short pink
lines under any interval in which a significant (Bonferroni-
corrected p* <.05 equals p <.00096) effect is observed.
This indicates a somewhat earlier effect of preference on
Fz (between ca 240-400 ms) than Pz (290-600 ms), likely
coinciding with a difference between P3a and P3b. For
both electrodes, attractive faces evoked more positivity, with
inconsistent faces roughly in between non-preferred and
preferred. Given that effects were observed mainly after 250
ms, we can infer that the GBCI likely did not benefit from
a flanker-induced N2 effect, and instead primarily relied on
patterns of activity within the P300 range.
5.2 Classification results
Classification results for all participants are shown in Figure
4. Within-subject permutation-based significance tests at p
<.05 showed that the classifiers performed significantly
better than random baselines for 28 out of 30 participants,
far better than chance level (less than 2 out of 30). Across
subjects, classifier performance had an average AUC of 0.76,
min = .61, max = .93. This suggests that the classifiers were
able to find significant structure in the data that discrimi-
nated brain responses for attractive and unattractive faces.
These classifications were then employed to generate the
preferred/non-preferred faces, as shown in Figure 5.
Fig. 4: AUC scores across participants, with scores for ran-
dom baselines computed using label permutation in gray.
Diamonds placed below boxplots indicate within-subject
significance of classifier performance against random.
5.3 Generated image evaluation results
The latent vectors of the faces expected to be found attrac-
tive were used to generate images such as those displayed
in Figure 5. To validate the GBCI in its ability to generate
personally attractive images, we performed two empirical
As can be seen in Figure 5 panel B, the results from the
free selection task showed that the positive generated image
was selected as attractive in 86.7% of cases from among
24 images (i.e. for 26/30 participants), while the negative
was selected in 20.0% (SE = 3.2%). In other words, there
were 86.7% true positives, 80.0% true negatives, 20.0 % false
negatives, and 13.3% false positives. Therefore, generative
Fig. 5: Individually generated faces and their evaluation. Panel A shows for eight female and eight male participants
(full overview available here) the individual faces expected to be evaluated positively (in green framing) and negatively
(in red). Panel B shows the evaluation results averaged across participants for both the free selection (upper-right) and
explicit evaluation (lower-right) tasks. In the free selection task, the images that were expected to be found attractive (POS)
and unattractive (NEG) were randomly inserted with 20 matched controls (RND = random expected attractiveness), and
participants made a free selection of attractive faces. In the explicit evaluation task, participants rated each generated (POS,
NEG, RND) image on a Likert-type scale of personal attractiveness.
performance (Accuracy = 83.33%) could be described as
Figure 5 panel B furthermore shows the results of the
second, explicit evaluation task. To analyse these results
we conducted two Bonferroni-corrected repeated measures
ANOVAs with image (unattractive, random, attractive) as
factor and average rating on personal attractiveness (mea-
sure 1) and population attractiveness (measure 2) as depen-
dents. This showed a significant effect of image on personal
attractiveness, F (2, 58) = 40.83, MSE = 1.00, p <.0001, η2
= .46. As shown in 5, positive images were rated higher
than either negative (p <.0001) or random (p <.0001)
images, while negative and random images did not show
a difference. For the second measure, of population attrac-
tiveness, the same analysis again showed a significant effect
of image, F (2, 58) = 43.88, MSE = 0.45, p <.0001, η2=
.52. In contrast with personal attractiveness, post-hoc com-
parisons of population attractiveness showed both positive
(p <.0001) and negative (p <.0001) to be significantly
higher evaluated than random images. Indeed, Bonferroni-
corrected t-tests showed no significant difference between
positive and negative images on popular attractiveness, p
= .07. Thus, negative generated images were evaluated as
highly attractive for other people, but not for the participant
Taken together, the results suggest that the GBCI was
highly accurate in generating personally attractive images
(83.33%). They also show that while both negative and pos-
itive generated images were evaluated as highly attractive
for the general population (respectively M = 4.43 and 4.90
on a scale of 1-5), only the positive generated images (M =
4.57) were evaluated as highly personally attractive.
5.4 Qualitative results
In semi-structured post-test interviews, participants were
shown the generated images that were expected to be found
attractive/ unattractive. Thematic analysis found predic-
tions of positive attractiveness were experienced as accurate:
There were no false positives (generated unattractive found
personally attractive). The participants also expressed being
pleased with results (eg.”Quite an ideal beauty for a male!”;
”I would be really attracted to this!”; ”Can I have a copy of
this? It looks just like my girlfriend!”).
However, there was some ambiguity regarding the accu-
racy of images that were expected to be seen as unattractive.
This may have to do with inviting the social discourse of the
outside world into the lab. Despite being well aware that the
faces were not real, the subjects would not be rude to their
face. They would use conversational mitigation strategies,
roughly grouped into three categories:
First, when a participant would discuss an image that
was expected to be found unattractive, they volunteered
that others might: the ”it’s not you, it’s me!” strategy (”I
don’t find this person attractive... but I see people would
think so”). Second, they would couple the (fictional) person
in the image with negative personality traits (”I don’t like
his smile...too bossy”). This could be seen as ”shifting the
blame” away from themselves: The reason for not liking the
image is that the image is unlikable. Only as a last option,
might they deny the image personhood entirely, blaming
”weird artifacts” or ”bad source material” for finding an
image unattractive. Otherwise, they would assign the faces
personalities, even getting upset if they were predicted to be
found unattractive (”He’s quite good, if I were him I would
be a trifle annoyed!”). This may have the effect of confusing
just how negative a judgment they would admit to.
Various themes were discerned in the phenomenological
deconstruction of the preferences of the participants. They
took into account not only a variety of visual, physical
features, but also speculated, without being invited to do
so, on what these features signified. Some features were
very concrete, such as hair color, with a general preference
towards blonder generated images. Others were slightly
more complex: Male participants uniformly expressed a
preference for younger faces while female participants tied
perceived youth to specific attributes of the faces, such as the
presence or lack of hair (”he looks old...bald”). Participants,
however, moved beyond simple physical features to inferring
personality traits (”he looks too bossy”) and judging these
(”I cannot say anything against him. He is charming”).
Here again, they had a tendency to apply real-world social
discourse when disapproving. For example, one female
expressed dislike for a generated image due to assumption
she was required to like it, despite it not being presented as
her ”perfect match”: (”he is too Hollywood. I feel pressured
to like him”).
In sum, the qualitative results confirm quantitative tests
in that the predictions were experienced as accurately
matching personal preferences. Moreover, despite the ten-
dency to play down negative attraction, participants agreed
on what did not match their personal preferences. The qual-
itative analysis further enriched our understanding of how
participants determined attractiveness, and suggests GBCI
could present an effective tool for discussing preference.
We presented generative brain-computer interfacing for
visualizing personal attraction. Our approach used brain
responses related to aesthetic preference as feedback for a
generative adversarial network (GAN) that generated novel
images matching personal attraction. To empirically test
this, we recorded EEG responses to GAN produced faces,
training a model to classify ERPs. This model was applied
to a new sample of artificially produced faces, detecting la-
tent vectors expected to have personally attractive features.
Combining these vectors, we generated novel images that
were expected to be evaluated as personally attractive in a
blind test against matched controls. The results show that
GBCI produces highly attractive, personalized images with
high (83.33%) accuracy.
6.1 Summary of contributions
The generative brain-computer interface (GBCI) is able to
generate a priori non-existing images of faces that are seen
as personally attractive. Uniting BCI methods with a GAN
allowed us to generate photorealistic images based on brain
activity. Importantly, the generated images did not rely
on external assumptions of the underlying data (such as
what attributes make a face beautiful). Thus, the GBCI is
able to generate attractive images in a data-driven way
unaffected by current theories and opinions of beauty. The
contributions of our work to affective computing are both
methodological and empirical and are summarized as fol-
The GBCI shows personalized, affective decisions related
to visual aesthetics can be revealed in interaction between
a human user and a generative neural network. Intentions
are latent mental functions that can be hard to verbalize or
even unconscious. Aesthetic judgments are a prime example
of a common cognitive function we engage in without full
understanding: We make snap judgments that something
is beautiful, but are ill equipped to explain why. The GBCI
operates on implicit signals in response to complex configura-
tions of visual features to visualizing the aesthetic decision-
The GBCI’s generates images by operating on individual
processes of subjective features, resulting in higher perfor-
mance than computer vision approaches based on general
features. Unlike existing systems that are able to predict and
generate generally attractive-looking images, the neuroad-
aptive BCI allows personalized predictions. While images
generated by the GAN model are generically attractive,
GBCI was found to produce images that specifically match
individual user preferences.
6.2 Limitations
While the GBCI shows clear capability of generating per-
sonally attractive images, we would not go so far as to
suggest that the generated images correspond to a men-
tal representation of the participant’s ideal attraction. This
would naively assume that our participants engaged in the
task with a ”Platonic ideal” image in mind that was also
localizable within the GAN space, which they then repeat-
edly matched against the displayed stimuli so that the GBCI
would converge upon this point. Even if such a single image
existed either in the mind – a standpoint few in the literature
would endorse – or in the GAN, the final generated images
in the present study were simply weighted averages of
the 240 vectors representing the shown images within a
high-dimensional artificial neural network. Such a sample
is unlikely to provide representative coverage of the GAN,
let alone human face perception.
The approach may thus be limited by the structure
of mental representations of personal attraction, as well
as the GAN characteristics, such as coverage within this
space by the selected initial images and the algorithms
used to navigate this space. Further work could present
more images, provide an interactive path based on ongoing
neural feedback to correct and incorrect predictions, and
develop more sophisticated methods to iteratively identify
and explore dimensions of the GAN space so that attractive
images can be more reliably generated given a user’s unique
preferences [76].
On the other hand, the constrained sample space within
the GAN provides commensurability, enabling inference
across subjects [17] and allowing us to compare and contrast
them in terms of personal attraction [16]. For example, as
can be seen in Figure 5, the male positive matches show
generated images with a marked family resemblance, sug-
gesting high consensus in their personal attraction. This
presents an independent confirmation of a well-known
finding concerning attractiveness ratings in that males are
generally more in agreement than females as to what they
find attractive [74].
As the GBCI output is determined by the GAN archi-
tecture, it is critical to consider how the particular train-
ing set initially used to create the network, i.e. images of
thousands of celebrities [38], affected the results. Biased
data is a common problem throughout the machine learning
community, and even datasets widely used as benchmarks
for image classification tasks of everyday objects have been
shown to be biased in some manner [69]. Clearly, given
that images of famous people include people like fashion
models and movie stars, it is reasonable to assume that
the GBCI’s neuroadaptivity was biased towards producing
attractive-looking images. The results clearly indicate that
this was indeed the case: negative predictions were evalu-
ated and discussed as conventionally, even if not personally
attractive. Indeed, in blind tests, only positive predictions
were selected as matching the user’s individual aesthetics.
In other words, the use of a GAN trained on generically
attractive people increased the difficulty of demonstrating
validity, and yet, positive generated images were highly
preferred over matched controls. Thus, while a regular GAN
and computer vision based attraction detection algorithm
provides ample means for generating attractive images, the
GBCI performs better for personal attraction.
A similar limitation from the use of a celebrity images
GAN is the degree this over-emphasizes inequality and lack
of diversity in the popular media. For example, considering
the degree to which people of color are generally underrep-
resented in celebrity populations [21], [33], they are also less
likely to be generated by the GBCI, as can be seen in the
lack of ethnic diversity presented in Figure 5. On the other
hand, this may well correspond to the degree representation
influences social perception, given that our mainly white
sample demonstrated racial preferences favoring own-race
images over other-race images, replicating existing findings
from the social-psychological literature [8], [26], [59]. It is
possible that the GBCI’s high performance metrics are due
to the social inequality within the GAN model mirroring
cultural psychological biases. Thus, much like other ma-
chine learning applications, GBCI does not represent an
”objective”, ”fair” technology.
Another aspect related to the representation learned by
the GAN model is the face domain. Our experiments are
based on a latent structure learned in an unsupervised
manner, but are limited to a dataset of images of celebrity
faces. Therefore, the broader generalizability across different
domains, such as landscapes, animals, or even art, depends
on the ability of GAN architecture to capture the semantics
[4] and different affective dimensions that the learned rep-
resentations capture. Recent research has shown that GAN
models are, indeed, able to perform on a variety of domains
[39] and varying tasks, such as style transfer [1] and even
capture object-level semantics [10].
6.3 Implications and future work
As a novel paradigm, GBCI produces insights into human
cognitive and affective processing by visualizing individ-
ual differences and portraying implicit processing. In the
previous section, we already discussed how the uniform
procedure used to generate the individual images presents
interesting insights: The generated positive and negative
images of the male group of participants seems more similar
to one another than the female ones. While cognitive neuro-
science typically looks at the similarity between individuals
in terms of their averages, the reverse inference achieved
by GBCI allows us to speculate about interindividual vari-
ance: What causes groups of people to relate in terms of
GBCI generated faces, or to differ drastically between one
Furthermore, we are interested in exploring the degree
to which GBCI visualizes implicit bias. Noting the lack of
people of color in Figure 5 despite ethnicity being task
irrelevant, we speculated in the previous section how this
could either reflect a GAN limitation or an implicit bias
in participants. If the latter is true, then we would expect
even with more diversely oriented GANs, GBCI would
produce similar results. Furthermore, GBCI could provide
qualitatively different insights with other subjective catego-
rization tasks, such as recognizing images as trustworthy,
benevolent, or powerful. What kind of person would the
GBCI generate, but more importantly, what would this tell
us about the individual? While this remains speculation, the
results presented in the present study cause us to believe the
GBCI could be a significant step forward in social cognition.
Given the far-reaching importance the GBCI could have
if it were to present a tool for visualizing implicit bias, it is
critical for future work to address the degree to which the
results are affected by the GAN’s original source material
and/or by the diversity of the GBCI participants. To this
end, we envision two separate paths. Firstly, it is critical to
investigate whether GANs based on more diverse source
material may better represent diverse participant’s personal
attraction. Secondly, cross-cultural endeavors should be
made to estimate the efficacy of GBCI with groups that are
underrepresented within the GAN. For both, it may seem
obvious that disparity between GAN and sample diversity
limits the efficacy of the GAN. However, it is important
to keep in mind that the overrepresentation of certain de-
mographies in the popular media may itself affect mental
representations, thus causing implicit bias. As we see, the
present results suggest high performance of the GBCI across
genders and even though the sample pool hardly reflected
the original celebrity database.
Finally, while future work must identify to what extent
GBCI output corresponds to mental processes, we envision
the GBCI as a practical, creative tool that generates images
optimized towards personal aesthetic preferences. By map-
ping brain activity to a representational space learned from
training data without predefined model parameterization,
we demonstrate the feasibility of visualizing attractiveness
without bias towards a-priori assumptions of the internal
structure underlying attractive images. Since the represen-
tational space is infinite, the model is theoretically capable
of reproducing mental images within the limits of the data
used to train the generative model [18]. The GBCI further-
more operates on brain signals alone and therefore requires
no artistic skills to create aesthetically pleasing portraits.
Thus, our approach may enable users to realize creative
customization of images using brain signals. Even though
future studies will be needed to determine the relationship
between GBCI-produced images, mental imagery, and men-
tal representation, we can already confirm that the GBCI
is capable in creating novel images that are evaluated as
personally attractive.
The research was supported by the Academy of Finland
(decision no: 313610, 322653, 328875).
[1] Rameen Abdal, Yipeng Qin, and Peter Wonka. Image2stylegan:
How to embed images into the stylegan latent space? In Proceed-
ings of the IEEE international conference on computer vision, pages
4432–4441, 2019.
[2] Oswald Barral, Manuel JA Eugster, Tuukka Ruotsalo, Michiel M
e, Ilkka Kosunen, Niklas Ravaja, Samuel Kaski, and Giulio
Jacucci. Exploring peripheral physiology as a predictor of per-
ceived relevance in information retrieval. In Proceedings of the 20th
international conference on intelligent user interfaces, pages 389–399,
[3] Oswald Barral, Ilkka Kosunen, Tuukka Ruotsalo, Michiel M Spap´
Manuel JA Eugster, Niklas Ravaja, Samuel Kaski, and Giulio
Jacucci. Extracting relevance and affect information from physio-
logical text annotation. User Modeling and User-Adapted Interaction,
26(5):493–520, 2016.
[4] David Bau, Jun-Yan Zhu, Hendrik Strobelt, Bolei Zhou, Joshua B
Tenenbaum, William T Freeman, and Antonio Torralba. Gan
dissection: Visualizing and understanding generative adversarial
networks. arXiv preprint arXiv:1811.10597, 2018.
[5] Benjamin Blankertz, Steven Lemm, Matthias Treder, Stefan Haufe,
and Klaus-Robert M ¨
uller. Single-trial analysis and classification of
erp components - a tutorial. NeuroImage, 56(2):814–825, 2011.
[6] Virginia Braun and Victoria Clarke. Using thematic analysis in
psychology. Qualitative research in psychology, 3(2):77–101, 2006.
[7] Andrew Brock, Jeff Donahue, and Karen Simonyan. Large scale
GAN training for high fidelity natural image synthesis. CoRR,
abs/1809.11096, 2018.
[8] David M Buss. Human mate selection: Opposites are sometimes
said to attract, but in fact we are likely to marry someone who is
similar to us in almost every variable. American scientist, 73(1):47–
51, 1985.
[9] Danilo Bzdok, Robert Langner, Svenja Caspers, F Kurth, U Habel,
Karl Zilles, A Laird, and Simon B Eickhoff. Ale meta-analysis
on facial judgments of trustworthiness and attractiveness. Brain
Structure and Function, 215(3-4):209–223, 2011.
[10] Xinyuan Chen, Chang Xu, Xiaokang Yang, and Dacheng Tao.
Attention-gan for object transfiguration in wild images. In Proceed-
ings of the European Conference on Computer Vision (ECCV), pages
164–180, 2018.
[11] Matthew A Conroy and John Polich. Affective valence and p300
when stimulus arousal level is controlled. Cognition and emotion,
21(4):891–901, 2007.
[12] Eric Courchesne, Steven A Hillyard, and Robert Galambos. Stim-
ulus novelty, task relevance and the visual evoked potential in
man. Electroencephalography and clinical neurophysiology, 39(2):131–
143, 1975.
[13] Michael R Cunningham, Alan R Roberts, Anita P Barbee, Perri B
Druen, and Cheng-Huan Wu. ” their ideas of beauty are, on the
whole, the same as ours”: Consistency and variability in the cross-
cultural perception of female physical attractiveness. Journal of
Personality and Social Psychology, 68(2):261, 1995.
[14] Charles Darwin. The Descent of Man, and Selection in Relation to Sex,
volume 2. D. Appleton, 1872.
[15] Ritendra Datta, Dhiraj Joshi, Jia Li, and James Z Wang. Study-
ing aesthetics in photographic images using a computational
approach. In European conference on computer vision, pages 288–301.
Springer, 2006.
[16] Keith M Davis, Michiel Spap ´
e, and Tuukka Ruotsalo. Collabo-
rative filtering with preferences inferred from brain signals. In
WWW ’21: Proceedings of The Web Conference 2021, 2021. To appear.
[17] Keith M Davis III, Lauri Kangassalo, Michiel Spap ´
e, and Tuukka
Ruotsalo. Brainsourcing: Crowdsourcing recognition tasks via
collaborative brain-computer interfacing. In Proceedings of the 2020
CHI Conference on Human Factors in Computing Systems, pages 1–14,
[18] Carlos de la Torre-Ortiz, Michiel M Spap´
e, Lauri Kangassalo, and
Tuukka Ruotsalo. Brain relevance feedback for interactive image
generation. In Proceedings of the 33rd Annual ACM Symposium on
User Interface Software and Technology, pages 1060–1070, 2020.
[19] Emanuel Donchin and Michael GH Coles. Is the p300 component
a manifestation of context updating? Behavioral and brain sciences,
11(3):357–374, 1988.
[20] Yael Eisenthal, Gideon Dror, and Eytan Ruppin. Facial attractive-
ness: Beauty and the machine. Neural Computation, 18(1):119–142,
[21] Morgan E Ellithorpe and Amy Bleakley. Wanting to see people like
me? racial and gender diversity in popular adolescent television.
Journal of youth and adolescence, 45(7):1426–1437, 2016.
[22] Manuel J. A. Eugster, Tuukka Ruotsalo, Michiel M. Spap´
e, Oswald
Barral, Niklas Ravaja, Giulio Jacucci, and Samuel Kaski. Natu-
ral brain-information interfaces: Recommending information by
relevance inferred from human brain signals. Scientific Reports,
6:38580, December 2016.
[23] Manuel J.A. Eugster, Tuukka Ruotsalo, Michiel M. Spap´
e, Ilkka
Kosunen, Oswald Barral, Niklas Ravaja, Giulio Jacucci, and
Samuel Kaski. Predicting term-relevance from brain signals.
In Proceedings of the 37th International ACM SIGIR Conference on
Research & Development in Information Retrieval, SIGIR ’14, page
425–434, New York, NY, USA, 2014. Association for Computing
[24] Lawrence A Farwell and Emanuel Donchin. The truth will out:
Interrogative polygraphy (”lie detection”) with event-related brain
potentials. Psychophysiology, 28(5):531–547, 1991.
[25] Lawrence Ashley Farwell and Emanuel Donchin. Talking off the
top of your head: toward a mental prosthesis utilizing event-
related brain potentials. Electroencephalography and clinical Neu-
rophysiology, 70(6):510–523, 1988.
[26] Raymond Fisman, Sheena S Iyengar, Emir Kamenica, and Itamar
Simonson. Racial preferences in dating. The Review of Economic
Studies, 75(1):117–132, 2008.
[27] Jerome H. Friedman. Regularized Discriminant Analysis. Journal
of the American Statistical Association, 84(405):165–175, March 1989.
[28] Junying Gan, Lichen Li, Yikui Zhai, and Yinhua Liu. Deep self-
taught learning for facial beauty prediction. Neurocomputing,
144:295 – 303, 2014.
[29] P.I. Good. Permutation Tests: A Practical Guide to Resampling Methods
for Testing Hypotheses. Springer, 2nd edition, 2000.
[30] Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu,
David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua
Bengio. Generative adversarial nets. In Advances in neural informa-
tion processing systems, pages 2672–2680, 2014.
[31] Ville J Harjunen, Imtiaj Ahmed, Giulio Jacucci, Niklas Ravaja, and
Michiel M Spap´
e. Manipulating bodily presence affects cross-
modal spatial attention: A virtual-reality-based erp study. Frontiers
in human neuroscience, 11:79, 2017.
[32] Lea H ¨
ofel and Thomas Jacobsen. Electrophysiological indices
of processing aesthetics: Spontaneous or intentional processes?
International Journal of Psychophysiology, 65(1):20–31, 2007.
[33] Darnell Hunt, Ana-Christina Ram ´
on, Michael Tran, Ambe-
ria Sargent, and Debanjan Roychoudhury. Hollywood di-
versity report 2018: Five years of progress and missed
opportunities. UCLA College of Social Sciences, February.
https://socialsciences. ucla. edu/wp-content/uploads/2018/02/UCLA-
Hollywood-Diversity-Report-2018-2-27-18. pdf, 2018.
[34] Miguel Ib ´
nez-Berganza, Ambra Amico, and Vittorio Loreto. Sub-
jectivity and complexity of facial attractiveness. Scientific reports,
9(1):1–12, 2019.
[35] Giulio Jacucci, Oswald Barral, Pedram Daee, Markus Wenzel,
Baris Serim, Tuukka Ruotsalo, Patrik Pluchino, Jonathan Freeman,
Luciano Gamberini, Samuel Kaski, et al. Integrating neurophys-
iologic relevance feedback in intent modeling for information
retrieval. Journal of the Association for Information Science and
Technology, 70(9):917–930, 2019.
[36] Victor S Johnston, Rebecca Hagel, Melissa Franklin, Bernhard
Fink, and Karl Grammer. Male facial attractiveness: Evidence for
hormone-mediated adaptive design. Evolution and human behavior,
22(4):251–267, 2001.
[37] Lauri Kangassalo, Michiel Spap ´
e, and Tuukka Ruotsalo. Neu-
roadaptive modelling for generating images matching perceptual
categories. Scientific Reports, 10(1):1–10, 2020.
[38] Tero Karras, Timo Aila, Samuli Laine, and Jaakko Lehtinen.
Progressive Growing of GANs for Improved Quality, Stability,
and Variation. arXiv:1710.10196 [cs, stat], October 2017. arXiv:
[39] Tero Karras, Samuli Laine, Miika Aittala, Janne Hellsten, Jaakko
Lehtinen, and Timo Aila. Analyzing and improving the image
quality of stylegan. In Proceedings of the IEEE/CVF Conference on
Computer Vision and Pattern Recognition, pages 8110–8119, 2020.
[40] Yan Ke, Xiaoou Tang, and Feng Jing. The design of high-level fea-
tures for photo quality assessment. In 2006 IEEE Computer Society
Conference on Computer Vision and Pattern Recognition (CVPR’06),
volume 1, pages 419–426. IEEE, 2006.
[41] Nicole Koehler, Leigh W Simmons, Gillian Rhodes, and Marianne
Peters. The relationship between sexual dimorphism in human
faces and fluctuating asymmetry. Proceedings of the Royal Society of
London. Series B: Biological Sciences, 271(suppl 4):S233–S236, 2004.
[42] Albert Kok. On the utility of p3 amplitude as a measure of
processing capacity. Psychophysiology, 38(3):557–577, 2001.
[43] Laurens R Krol and Thorsten O Zander. Passive bci-based neu-
roadaptive systems. In GBCIC, 2017.
[44] Sandra JE Langeslag, Bernadette M Jansma, Ingmar HA Franken,
and Jan W Van Strien. Event-related potential responses to love-
related facial stimuli. Biological psychology, 76(1-2):109–115, 2007.
[45] Judith H Langlois, Jean M Ritter, Lori A Roggman, and Lesley S
Vaughn. Facial diversity and infant preferences for attractive faces.
Developmental Psychology, 27(1):79, 1991.
[46] Judith H Langlois and Lori A Roggman. Attractive faces are only
average. Psychological science, 1(2):115–121, 1990.
[47] Christian Ledig, Lucas Theis, Ferenc Huszar, Jose Caballero, An-
drew Cunningham, Alejandro Acosta, Andrew Aitken, Alykhan
Tejani, Johannes Totz, Zehan Wang, and Wenzhe Shi. Photo-
realistic single image super-resolution using a generative adver-
sarial network. In The IEEE Conference on Computer Vision and
Pattern Recognition (CVPR), July 2017.
[48] Olivier Ledoit and Michael Wolf. A well-conditioned estimator
for large-dimensional covariance matrices. Journal of Multivariate
Analysis, 88(2):365–411, February 2004.
[49] Anthony C Little, Benedict C Jones, and Lisa M DeBruine. Facial
attractiveness: evolutionary based research. Philosophical Transac-
tions of the Royal Society B: Biological Sciences, 366(1571):1638–1659,
[50] Anthony C Little, Benedict C Jones, Lisa M DeBruine, and Chris-
tine A Caldwell. Social learning and human mate preferences:
a potential mechanism for generating and maintaining between-
population diversity in attraction. Philosophical Transactions of the
Royal Society B: Biological Sciences, 366(1563):366–375, 2011.
[51] Ziwei Liu, Ping Luo, Xiaogang Wang, and Xiaoou Tang. Deep
Learning Face Attributes in the Wild. arXiv:1411.7766 [cs], Novem-
ber 2014. arXiv: 1411.7766.
[52] Yiwen Luo and Xiaoou Tang. Photo and video quality evaluation:
Focusing on the subject. In European Conference on Computer Vision,
pages 386–399. Springer, 2008.
[53] Ewout H Meijer, Gershon Ben-Shakhar, Bruno Verschuere, and
Emanuel Donchin. A comment on farwell (2012): brain finger-
printing: a comprehensive tutorial review of detection of con-
cealed information with event-related brain potentials. Cognitive
neurodynamics, 7(2):155–158, 2013.
[54] Markus Ojala, Niko Vuokko, Aleksi Kallio, Niina Haiminen, and
Heikki Mannila. Randomization methods for assessing data anal-
ysis results on real-valued matrices. Statistical Analysis and Data
Mining: The ASA Data Science Journal, 2(4):209–230, 2009.
[55] Letizia Palumbo, Marco Bertamini, and Alexis Makin. Scaling
of the extrastriate neural response to symmetry. Vision Research,
117:1–8, 2015.
[56] David Perrett, KA May, and Sakiko Yoshikawa. Facial shape and
judgements of female attractiveness. Nature, 368:239–42, 04 1994.
[57] John Polich. Attention, probability, and task demands as determi-
nants of p300 latency from auditory stimuli. Electroencephalography
and clinical neurophysiology, 63(3):251–259, 1986.
[58] John Polich. Updating p300: an integrative theory of p3a and p3b.
Clinical neurophysiology, 118(10):2128–2148, 2007.
[59] Belinda Robnett and Cynthia Feliciano. Patterns of racial-ethnic
exclusion by internet daters. Social Forces, 89(3):807–828, 2011.
[60] J. J. Rocchio. Relevance feedback in information retrieval. In
Gerard Salton, editor, The SMART Retrieval System - Experiments
in Automatic Document Processing, pages 313–323. Prentice Hall,
Englewood, Cliffs, New Jersey, 1971.
[61] Christopher P Said and Alexander Todorov. A statistical model of
facial attractiveness. Psychological Science, 22(9):1183–1190, 2011.
[62] Jose San Pedro and Stefan Siersdorfer. Ranking and classifying
attractiveness of photos in folksonomies. In Proceedings of the 18th
international conference on World wide web, pages 771–780. ACM,
[63] Kendra Schmid, David Marx, and Ashok Samal. Computation of a
face attractiveness index based on neoclassical canons, symmetry,
and golden ratios. Pattern Recognition, 41(8):2710 – 2717, 2008.
[64] Hamid R Sheikh, Alan C Bovik, and Lawrence Cormack.
No-reference quality assessment using natural scene statistics:
Jpeg2000. IEEE Transactions on Image Processing, 14(11):1918–1927,
[65] Jiazheng Shi, Ashok Samal, and David Marx. How effective are
landmarks and their geometry for face recognition? Computer
vision and image understanding, 102(2):117–133, 2006.
[66] Michiel M Spap ´
e, Marco Filetti, Manuel JA Eugster, Giulio Jacucci,
and Niklas Ravaja. Human computer interaction meets psy-
chophysiology: a critical perspective. In International Workshop on
Symbiotic Interaction, pages 145–158. Springer, 2015.
[67] Michiel Spap ´
e, Rinus Verdonschot, and Henk van Steenbergen.
The E-Primer: An introduction to creating psychological experiments in
E-Prime. Leiden Univerisity Press, 2019.
[68] Randy Thornhill and Steven W Gangestad. Facial attractiveness.
Trends in cognitive sciences, 3(12):452–460, 1999.
[69] Antonio Torralba and Alexei A Efros. Unbiased look at dataset
bias. In CVPR 2011, pages 1521–1528. IEEE, 2011.
[70] Christopher W Tyler. Empirical aspects of symmetry perception.
Spatial Vision, 9(1):1–8, 1995.
[71] Antti Ukkonen, Pyry Joona, and Tuukka Ruotsalo. Generating
Images Instead of Retrieving Them: Relevance Feedback on Generative
Adversarial Networks, page 1329–1338. Association for Computing
Machinery, New York, NY, USA, 2020.
[72] Bram van de Laar, Hayrettin G¨
ok, Danny Plass-Oude Bos,
Mannes Poel, and Anton Nijholt. Experiencing bci control in
a popular computer game. IEEE Transactions on Computational
Intelligence and AI in Games, 5(2):176–184, 2013.
[73] Rolf Verleger. On the utility of p3 latency as an index of mental
chronometry. Psychophysiology, 34(2):131–156, 1997.
[74] Dustin Wood and Claudia Chloe Brumbaugh. Using revealed
mate preferences to evaluate market force and differential pref-
erence explanations for mate selection. Journal of personality and
social psychology, 96(6):1226, 2009.
[75] Beste F. Yuksel, Kurt B. Oleson, Lane Harrison, Evan M. Peck,
Daniel Afergan, Remco Chang, and Robert JK Jacob. Learn
piano with bach: An adaptive learning interface that adjusts task
difficulty based on brain state. In Proceedings of the 2016 CHI
Conference on Human Factors in Computing Systems, CHI ’16, pages
5372–5384, New York, NY, USA, 2016. ACM.
[76] Thorsten O Zander, Lauens R Krol, and Klaus Gramann. Towards
neuroadaptive technology: Implicitly controlling a cursor though a
passive brain–computer interface. In Neuroergonomics, pages 301–
302. Elsevier, 2018.
Michiel Spap´
ereceived his PhD in Psychology from Leiden University
in 2009. After working as a postdoc (Nottingham, Helsinki), and lecturer
(Liverpool), he is now docent in cognitive neuroscience at Helsinki
University, focusing on emotion, perception/action, and EEG.
Lauri Kangassalo received his MSc in Computer Science from the
University of Helsinki in 2019. As a specialist in machine learning,
he aided the present study in psychophysiology analysis. He currently
works at the Finnish Meteorological Institute.
Keith M. Davis III is a PhD student at the University of Helsinki. He spe-
cializes in machine learning and collaborative brain-computer interfaces.
Niklas Ravaja received a PhD degree in psychology from the University
of Helsinki, Finland, in 1996. He is professor of eHealth and well-being,
and expert on emotional and physiological processes during mediated
social interaction.
Zania Sovij¨
estudied at the Universities of Amsterdam and
Helsinki. She is a lab assistant at the Interaction Lab and the Cognitive
Computing Lab and provided expertise in qualitative data analysis.
Tuukka Ruotsalo is an Academy Research Fellow at the University of
Helsinki, and an Associate Professor at the University of Copenhagen,
He is an expert on machine-learning and cognitive computing.
... If we can transform neural activities into language effectively, that would be revolutionary for human-computer interaction as well as for the functional restoration of aphasia. At present, human-computer interaction is commonly developed using EEG or ECoG (Ieracitano, Mammone, Hussain, & Morabito, 2021;Spape et al., 2021), but both EEG and ECoG have limitations on spatial resolution or invasiveness. In contrast, fMRI is non-invasive and has good spatial resolution, which also shows potential application in helping the brain to directly communicate with an external device by decoding subject's perceptions or intentions from fMRI signals and converting them into a set of suitable commands or language. ...
... In order to promote the substantive development of decoding study, we need to deeply understand the neural mechanisms by which the brain integrates vision and language. Our language decoding model has potential applications in the field of the brain-machine interfaces (BCIs) (Anumanchipalli, Chartier, & Chang, 2019;Brumberg, Pitt, Mantie-Kozlowski, & Burnison, 2018;Ieracitano et al., 2021;Pandarinath et al., 2017;Spape et al., 2021) where the decoded texts can be directly input to the machine as instructions to achieve intelligent control. Hopefully, our model plus a speech synthesizer (Wang, Skerry-Ryan et al., 2017) can convert visual neural activities into speech, which will improve clinical feasibility of using speech neuroprosthetic technology to help aphasics to communicate more efficiently through what they are seeing. ...
Transforming neural activities into language is revolutionary for human–computer interaction as well as functional restoration of aphasia. Present rapid development of artificial intelligence makes it feasible to decode the neural signals of human visual activities. In this paper, a novel Progressive Transfer Language Decoding Model (PT-LDM) is proposed to decode visual fMRI signals into phrases or sentences when natural images are being watched. The PT-LDM consists an image-encoder, a fMRI encoder and a language-decoder. The results showed that phrases and sentences were successfully generated from visual activities. Similarity analysis showed that three often-used evaluation indexes BLEU, ROUGE and CIDEr reached 0.182, 0.197 and 0.680 averagely between the generated texts and the corresponding annotated texts in the testing set respectively, significantly higher than the baseline. Moreover, we found that higher visual areas usually had better performance than lower visual areas and the contribution curve of visual response patterns in language decoding varied at successively different time points. Our findings demonstrate that the neural representations elicited in visual cortices when scenes are being viewed have already contained semantic information that can be utilized to generate human language. Our study shows potential application of language-based brain-machine interfaces in the future, especially for assisting aphasics in communicating more efficiently with fMRI signals.
... We conclude the chapter by mentioning that, in addition to the topics covered here, there are several other aspects related to aesthetics that could be further considered. In particular, in addition to numerous applications of image aesthetics to enhancement, recommendation, etc., mentioned throughout the chapter, we need to mention video aesthetics (Yeh et al. 2013;Bhattacharya et al. 2013) and related applications (e.g., thumbnailing (Song et al. 2016)), and finally recent studies linking brain-computer interfaces to the generation of aesthetically pleasing pictures (Spape et al. 2021), which appear to be a promising avenue to understand and predict aesthetic judgment mechanisms. ...
Humans search for, identify, and interact with objects efficiently, utilizing not only the visual characteristics of the object itself but also contextual information to generate optimal predictions about objects in scenes. Over the course of our lives, we have acquired knowledge regarding co-occurring local objects as well as the global scene contexts in which they are usually encountered, creating strong predictions regarding what objects are typically found where in our environment. A number of studies from the last decades have characterized how such knowledge may guide attention in scene viewing and modulate object perception, using diverse methodologies like psychophysics, eye tracking, and neurophysiology, with various degrees of realism ranging from on-screen experiments via virtual reality to real-world studies. Some recent work has focused on investigating what “ingredients” of scenes actually influence object search and perception. Scenes tend to be hierarchically organized with some objects—so-called “anchor objects”—holding stronger predictions than others. Apart from meaningful objects, global scene properties (e.g., spatial layout or texture) have been shown to predict object identity. In order to tease apart the influence of such ingredients, large-scale databases and machine learning techniques have become increasingly popular. Here, we review recent advances in the field that help to better capture human efficiency in real-world scene and object perception, particularly focusing on which contextual information we take advantage of most and when. Further, we explore how these findings could be useful in pushing computer vision further ahead and how computer vision could mutually further our understanding of human visual perception.
In the last two decades, advancements in artificial intelligence and data science have attracted researchers' attention to machine learning. Growing interests in applying machine learning algorithms can be observed in different scientific areas, including behavioral sciences. However, most of the research conducted in this area applied machine learning algorithms to imagining and physiological data such as EEG and fMRI and there are relatively limited non-imaging and non-physiological behavioral studies which have used machine learning to analyze their data. Therefore, in this perspective article, we aim to (1) provide a general understanding of models built for inference, models built for prediction (i.e., machine learning), methods used in these models, and their strengths and limitations; (2) investigate the applications of machine learning to categorical data in behavioral sciences; and (3) highlight the usefulness of applying machine learning algorithms to non-imaging and non-physiological data (e.g., clinical and categorical) data and provide evidence to encourage researchers to conduct further machine learning studies in behavioral and clinical sciences.
Computational image aesthetics aims at designing algorithmic approaches to perform aesthetic decisions, in a similar fashion as humans. In the past fifteen years, computational aesthetics has undergone unprecedented development, thanks to the availability of large annotated datasets and deep learning approaches, impacting many applications in multimedia from image enhancement to recommendation and retrieval. In this chapter, we first overview the several interpretations that aesthetics has received over the centuries and propose a set of suitable dimensions for a taxonomy of computational aesthetic approaches. Then, we present the advances of computational aesthetics in the past decade by providing a critical analysis of the most popular datasets, early methods based on hand-crafted features, and modern approaches using deep neural networks. In the last part of the chapter, we discuss some open challenges in computational aesthetic quality assessment: dealing with the intrinsic subjectivity of the scores, and providing explainable aesthetic predictions. In particular, throughout the chapter, we stress the fundamental importance of data collection in computational aesthetics.
Full-text available
Brain-computer interfaces enable active communication and execution of a pre-defined set of commands, such as typing a letter or moving a cursor. However, they have thus far not been able to infer more complex intentions or adapt more complex output based on brain signals. Here, we present neuroadaptive generative modelling, which uses a participant's brain signals as feedback to adapt a boundless generative model and generate new information matching the participant's intentions. We report an experiment validating the paradigm in generating images of human faces. In the experiment, participants were asked to specifically focus on perceptual categories, such as old or young people, while being presented with computer-generated, photorealistic faces with varying visual features. Their EEG signals associated with the images were then used as a feedback signal to update a model of the user's intentions, from which new images were generated using a generative adversarial network. A double-blind follow-up with the participant evaluating the output shows that neuroadaptive modelling can be utilised to produce images matching the perceptual category features. The approach demonstrates brain-based creative augmentation between computers and humans for producing new information matching the human operator's perceptual categories.
Full-text available
The origin and meaning of facial beauty represent a longstanding puzzle. Despite the profuse literature devoted to facial attractiveness, its very nature, its determinants and the nature of inter-person differences remain controversial issues. Here we tackle such questions proposing a novel experimental approach in which human subjects, instead of rating natural faces, are allowed to efficiently explore the face-space and “sculpt” their favorite variation of a reference facial image. The results reveal that different subjects prefer distinguishable regions of the face-space, highlighting the essential subjectivity of the phenomenon. The different sculpted facial vectors exhibit strong correlations among pairs of facial distances, characterising the underlying universality and complexity of the cognitive processes, and the relative relevance and robustness of the different facial distances.
Full-text available
The use of implicit relevance feedback from neurophysiology could deliver effortless information retrieval. However, both computing neurophysiological responses and retrieving documents are characterized by uncertainty due to noisy signals and incomplete or inconsistent representations of the data. We present the first-of-its-kind, fully integrated information retrieval system that makes use of online implicit relevance feedback generated from brain activity as measured through electroencephalography (EEG), and eye movements. The findings of the evaluation experiment (N = 16) show that we are able to compute online neurophysiology-based relevance feedback with performance significantly better than chance in complex data domains and realistic search tasks. We contribute by demonstrating how to integrate in interactive intent modeling this inherently noisy implicit relevance feedback combined with scarce explicit feedback. While experimental measures of task performance did not allow us to demonstrate how the classification outcomes translated into search task performance, the experiment proved that our approach is able to generate relevance feedback from brain signals and eye movements in a realistic scenario, thus providing promising implications for future work in neuroadaptive information retrieval (IR).
Brain-computer interfaces (BCIs) are not only being developed to aid disabled individuals with motor substitution, motor recovery, and novel communication possibilities, but also as a modality for healthy users in entertainment and gaming. This study investigates whether the incorporation of a BCI in the popular game World of Warcraft (WoW) has effects on the user experience. A BCI control channel based on parietal alpha band power is used to control the shape and function of the avatar in the game. In the experiment, participants , a mix of experienced and inexperienced WoW players, played with and without the use of BCI in a within-subjects design. Participants themselves could indicate when they wanted to stop playing. Actual and estimated duration was recorded and questionnaires on presence and control were administered. Afterwards, oral interviews were taken. No difference in actual duration was found between conditions. Results indicate that the difference between estimated and actual duration was not related to user experience but was person specific. When using a BCI, control and involvement were rated lower. But BCI control did not significantly decrease fun. During interviews, experienced players stated that they saw potential in the application of BCIs in games with complex interfaces such as WoW. This study suggests that BCI as an additional control can be as much fun and natural to use as keyboard/mouse control, even if the amount of control is limited. Index Terms-Brain-computer interface (BCI), games, human factors, presence, user experience.