The perception of emotional expressions in social anxiety and autism

The ability to trust is pivotal to social life, yet many patients with mental
disorders have difficulties trusting others1,2. In contrast to healthy
individuals, they do not have the positive expectation that sharing emotions
can lead to help, assistance and support. This is detrimental to their
functioning as on a daily basis, quick evaluations of trustworthiness are
common, especially in interactions with strangers. Patients therefore miss out
on opportunities that interactions with people outside of their social network
can bring. Their lack of trust further prohibits them to reach out for the help
they need or to put confidence in a therapist, which blocks therapeutic
success3,4. Many patients, including those with ASD or SAD, show impaired
social skills including the avoidance of eye contact5,6, hyper-arousal7,8, less
emotional mimicry9,10, and deficits in emotion recognition11-14, and suffer
from social isolation15,16. Their low level of trust seems related to their
lesser ability to process signals of trust but in order to verify this
presumption, it is essential to disentangle the distinct components
constituting this putative relationship.
Humans* capacity to express, recognize and share emotions enables them to
navigate their social worlds and forms a core component of what it means to be
socially competent and healthy. In order to evaluate another*s trustworthiness,
they rely on various indicators of a safe interaction including emotional
expressions17-20. I recently proposed deficits in emotion processing as a
liability spectrum underlying a range of mental disorders, explaining
similarities in symptoms and high comorbidity rates. I concluded that research
in this area can advance significantly by directly comparing different clinical
groups with similar emotion deficits21. Comparing patients with ASD and SAD can
provide key insights into which deficits are disorder-specific or shared, which
is relevant for therapeutic interventions. The majority of research has focused
on the perception of explicit, isolated facial expressions, expressions rarely
seen in daily life22,23. My research on the perception of expressions beyond
the face, which I will detail later, shows that some standing assumptions
regarding social anxiety were incorrect24 and therefore, urgently calls for a
closer investigation of alternative expression modalities including expressions
of autonomic arousal, reflected in dilated pupils, tears and a blush22,24,25. I
will study these phenomena in a controlled setting, but also during real life
interactions.
The remainder of this section (a) provides a state-of-the art review of
literature on emotional expressions, different forms of mimicry, and
implications for the development of trust or distrust in patients with SAD and
ASD. I highlight where the literature falls short and where my research comes
in. This section is divided into three Work Packages (WP), each addressing
different sub-components of the linkage between emotion perception and
trustFig.1. WP1 focuses on emotional expressions, WP2 on mimicry and WP3 on
trust and distrust. The overarching aim of all WPs is to investigate the
linkage between mimicry, emotion recognition, and (dis)trust in patients with
SAD and ASD, compared to matched controls.

My research on the perception of expressions beyond the face shows that some
standing assumptions regarding social anxiety were incorrect24 and therefore,
urgently calls for a closer investigation of alternative expression modalities
including full body expressions and expressions of autonomic arousal, reflected
in dilated pupils, tears and a blush22,24,25. I will study these phenomena in a
controlled setting, but also during real life interactions. The overarching aim
of all WPs is to investigate the linkage between mimicry, emotion recognition,
and (dis)trust in patients with SAD and ASD, compared to matched controls.

Method and techniques
In all studies proposed below, 60 ASD and 60 SAD patients are being compared
with 120 carefully matched, healthy controls. Participants fill out
questionnaires measuring empathy, anxiety and autistic traits. Subjects wear
eye tracking glasses to measure gaze and pupil size, and electrodes to measure
facial muscle movements. Psychophysiological measures, e.g., heart-rate and
facial temperature give insight in arousal and blushing. This research consists
of three components.

Component 1 Emotion Perception
Study 1 focuses on the face. 1a is a passive viewing task during which subjects
view photos showing emotions expressed through different modalities
(constricted/dilated pupils, tears, blush). With an eye-tracker, attention and
avoidance patterns will be assessed. 1b investigates whether the groups
differentially perceive the expressions. Specifically, subjects will choose
amongst different emotional labels and rate the stimuli on valence and arousal
levels. 1c uses the dot probe paradigm, implemented on a touch-screen. In this
task, two photos are briefly presented. One of them is replaced by a dot, which
participants have to tap. As attention is drawn towards relevant stimuli, this
task typically shows faster reaction times when the dot replaces an emotional
compared to neutral image. This task robustly reveals biases in immediate
attention. Here, typical facial expressions will be used, in addition to
expressions from the eye region including dilated/constricted pupils, tears or
blushing faces.
*
Study 2 focuses on emotional body expressions. 2a In this eye-tracking task,
participants view expressions sent from the whole body including the face. In
condition A, facial and bodily expression are congruent, e.g., a happy face
above a happy body; in Condition B they express different emotions. I
previously showed that non-clinically anxious individuals avoided eye contact
but attended more to the hands, regardless of congruency. 2b investigates how
patients* compared to controls recognize body language. In addition to
photographs, short video clips will be shown. 2c uses the dot probe paradigm to
measure attentional biases towards body expressions.

Predictions
SAD and ASD patients are expected to show heightened physiological arousal,
avoid eye contact and, as a consequence, miss out on eye signals compared to
controls. Since many SAD patients fear blushing, it is expected that they
better notice this cue than ASD and controls and I expect this to be reflected
in an attentional bias towards blushing others. Further, patients are expected
to pay more attention to body language than controls. However, whereas for SAD
patients this can work as an adaptive coping-strategy, it is expected to be
less of use to ASD patients as previous research has shown they have deficits
in the recognition of body expressions.

Component 2 Emotional Mimicry
Study 3 compares patients* and controls* mimicry of expressions. 3a is a facial
EMG mimicry study measuring congruent reactions to facial displays of emotions.
This straightforward comparison between ASD and SAD has not been made
previously. 3b focuses on pupil mimicry. Stimuli are images of faces in which
the pupil sizes are manipulated to constrict or dilate. An eye-tracker measures
subjects* pupil size. 3c This study investigates whether seeing someone cry is
as contagious for patients as it is for healthy individuals. Stimulus materials
are video fragments of crying people, made by collaborator Prof. A. Fischer.
Participants are videotaped to detect tears and facial EMG is applied to
measure expressions of sadness. Furthermore, a thermal camera measures nasal
temperature, indicative of negative arousal. 3d studies contagious blushing.
Subjects observe faces that have been subtly retouched to make the cheeks
slightly redder or less red. The temperature of subjects* cheeks is indicative
for contagion.


Predictions
SAD patients will show enhanced mimicry of negative compared to positive
expressions and ASD patients are expected to mimic less. As there is no
research on the mimicry of emotional cues or body language in either of these
patients groups, this study is the first to explore whether the known facial
mimicry deviances extend to other modalities.

Component 3 Trust and Distrust
WP3 studies the implications of mimicry for social decisions in VR and during
real life interactions. Two economic games are used that measure trust and
distrust. They are identical in their outcome spaces, with the only difference
being in how outcomes are achieved. In the trust game, the investor allocates a
certain amount to the trustee. Subjects are told that this will be tripled and
sent to the trustee, who decides how much to return to the investor. The amount
the investor invests is taken as a measure of trust. In the distrust game,
Subject A starts with an empty pocket and Subject B with an endowment. Subject
A decides how much of Subject B*s endowment he or she wants to take. The
experimenter divides this amount by 3 and gives that to Subject A. Next,
Subject B is asked to decide to send any possible amount to Subject A, and this
amount is not altered. Subject A*s taking money is closely aligned with
previous operationalisations of distrust as an action to mitigate one*s
vulnerability to a counterpart. In particular, the desire to buffer oneself
from the effects of others* actions is reflected in taking money in
anticipation of Subject B not intending to share it. In this study we will have
an additional group of control participants.
Study 5 investigates mimicry with help of virtual reality where in 5a an
embodied virtual agent real-time mimics the facial expression of the subject,
or refrains from mimicry. In 5b, the virtual agent*s pupils either align with
subject*s pupils, or not. 5c aligns or misaligns the blush on the cheeks of the
avatar with subjects* facial temperature.
Study 6 is similar to Study 5 but participants play distrust rather than trust
games.

Prediction
Whereas typically, mimicry boosts trust, a study in SAD suggests this might not
work for patients. However, the head movements in that study were quite obvious
and it remains to be investigated what influence another*s autonomic mimicry
has on patients. As autonomic mimicry is an ancient bonding mechanism, this is
expected to be intact in patients and foster trust. The mimicry of negative
expressions, however, is expected to link with distrust decisions.
Study 7 Patients and controls sit opposite each other. In Condition A, they
cannot see each other due to a screen separating them. In Condition B, they see
each other directly. In Condition C they see each other live on a computer
screen (Skype) and in Condition D, there is a slight visual delay. In 7a
subjects play trust games and in 7b distrust games.

Prediction
Healthy subjects are expected to show more emotional signals such as smiles,
when they can see (A) as compared to when they cannot see their partner (B).
For ASD patients this difference is expected to be smaller. Patients will show
hyper-arousal in the seeing versus no-seeing condition and avoid eye gaze.
Levels of arousal, as well as emotional expressions are expected to be
synchronized and mimicked less in ASD compared to SAD and controls. The visual
delay will not hamper mimicry in the ASD group compared to SAD and controls.
Both patients groups are expected to show less trust, and more distrust, but
for different reasons. SAD patients are expected to mimic negative expressions
more, and rely on their bodily feedback when making decisions. As these
patients are often perceived more negatively by others, not trusting partners
in sharing gains, might sadly be a socially wise decision. Patients with ASD
are expected to mimic less overall and be somewhat more calculative.

There are no risks for the patients or control participants. Because this is a
thorough investigation into emotion processing, the only disadvantage is the
time cost.