Dissociating Social Functioning in ASD and Schizophrenia
using Clinical Assessment and Neural Response to Gaze Cues
J Foss-Feig, A Naples, K Deckert, E Levy, K Stavropoulos, M Rolison, SA Chang, J Trapani, K Ellison, T McAllister, S Malak, N Santamauro, C Schleifer, A Anticevic, V Srihari, & J McPartland
McPartland Lab, Yale Child Study Center, New Haven, CT
-100
-50
N170
Direct
Age (SD)
21.89 (3.81)
28.81 (8.30)
28.80 (7.69)
N170 Amplitude (uV)
0
50
100
150
200
250
300
350
FSIQ* (SD)
102.56 (19.29)
95.79 (11.56)
105.42 (20.47)
Onset face
with averted
gaze
Gaze shift to
direct when
participant
looks to eyes
Clinical Measures:
To measure social and perceptual difficulties, participants completed:
• ASD diagnostic assessment: Autism Diagnostic Observation Schedule
• SZ diagnostic assessment: Structured Clinical Interview for DSM
Diagnosis; Positive and Negative Syndrome Scale
• ASD self-report measures: Social Responsiveness Scale; Broad Autism
Phenotype Questionnaire; Autism-Spectrum Quotient
• SZ self-report measure: Schizotypal Personality Questionnaire
• Behavioral assessments: Benton Facial Recognition Test
50
30
120
25
25
100
80
48
6
20
80
60
46
4
15
60
40
44
42
8
12
4
8
2
4
0
0
ADOS Total
2
PANSS
Negative
PANSS
Positive
0
5
20
5
20
0
0
0
300
TD
ASD
SZ
200
150
6
12
ADOS Social Communication Total
18
SZ
-2.5
-3
-3.5
250
200
150
-2.5
30
Direct
TD_direct
-3
TD_averted
SZ
ASD_direct
ASD
SZ_averted
150
170
190
N170 Latency (ms)
210
Figure 3. N170 response to direct and averted gaze in ASD, SZ, and TD. Main effect of
Condition, F(1,41)=7.79, p < 0.01, indicating enhanced amplitude to averted vs. direct gaze
in all groups. Particularly with gaze shift from averted to direct, individuals with ASD showed
greater N170 amplitude relative to both SZ (p = 0.019) and TD (p = 0.048), whereas
individuals with SZ and TD did not differ from each other (p = 0.83). There was a significant
interaction with N170 latency, F(1,41)=3.38, p = 0.04, wherein latency to direct vs. averted
gaze was equivalent in TD, delayed for direct in SZ, and delayed for averted in ASD.
4
2.5
TD
2
ASD
1.5
SZ
1
0.5
0
Direct
Averted
P300
TD_direct
3.5
3
TD_averted
3
ASD_direct
2.5
SZ_direct
SZ_averted
1.5
1
0.5
-100
-50
0
Time (ms)
0
50
100
150
200
250
300
350
-0.5
-1
SRS Total
40
Benton Score
40
45
Benton Total Score
50
55
-1 0
ADOS Social Communication Score
6
12
18
-2
-3
-4
-5
-6
-7
-8
Figure 6. Transdiagnostic relations between ERP
response to gaze cues and clinical variables.
Across groups, greater ASD symptomatology, as rated
by a clinician on the ADOS, associated with delayed
N170 latency (r = 0.36, p = 0.044), but enhanced
N170 amplitude (r = -0.41, p = 0.02) to gaze shift from
direct to averted. Greater face perception skill, as
measured with the Benton, associated with faster
N170 latency (r = -0.35, p = 0.025) when perceiving
gaze cues.
In line with a dimensional approach to understanding neurodevelopmental
disorders, preliminary results of this study suggest that neural response to gazecontingent shifts in eye gaze is a reliable marker of social dysfunction across
individuals with ASD and SZ.
• N170 latency is related to social function and dysfunction. Across groups, those
with better face perception skills during clinical assessment had faster N170
latencies, whereas those with greater ASD symptomatology had more delayed
N170 to gaze cues.
• Whereas previous literature has suggested attenuated N170 response in ASD,
across groups ASD symptomatology was also associated with a larger N170 to
gaze cues in the context of our novel, gaze-contingent face perception paradigm.
In contrast to ERP markers, clinical measures of ASD and SZ symptomatology are
variably effective in differentiating the two clinical populations. Specifically, though
self-report measures are reliable in differentiating clinical populations from TD, they
are ineffective in differentiating between diagnostic categories.
ASD_averted
2
AQ Total
TD
Averted
SZ_direct
-4
35
N170 Amplitude (uV)
250
-1
ASD
BAP-Q Total
0
300
-2
SPQ Total
1
-0.5
TD
10
Figure 5. Clinical Assessments of Symptomatology. Regarding clinician administered and rated
measures, ADOS scores and the PANSS positive symptoms scale differentiated ASD from SZ;
however ASD and SZ patients were not well discriminated by the PANSS negative symptoms scale
or the Global Social Functioning Rating. When administered self-report measures of ASD and SZ
symptomatology, individuals with SZ and ASD did not differ from each other and both clinical
groups endorsed elevated symptom levels relative to controls across measures.
Averted
-1.5
15
40
0
Global Social
Functioning
20
10
100
3.5
Fixation
100
-4
ASD_averted
P300 Amplitude (uV)
Figure 1. Trial Structure. After
participants fixated on a crosshair
for ~300ms (panel 1), a face with
either direct or averted gaze was
presented (panel 2). After the
participant looked to the face for
≥500ms, a second face with shifted
gaze (panel 3) was presented for
600ms. Inter-trial interval ranged
from 200-1200ms.
-2
-3.5
*groups matched on sex and FSIQ
Experimental Paradigm:
• Participants were presented with 80 distinct photorealistic, animated faces
matched for low-level visual features.
• Contingent upon participants’ fixating on the face, stimuli responded by
shifting eye gaze (either from direct to averted or averted to direct).
-1.5
P300 Amplitude (uV)
ASD
SZ
TD
Sex*
13M
13M
9M
30
16
0
Time (ms)
-1
N
17
14
13
6
140
10
0
-0.5
Participant Demographics
8
35
20
100
0.5
0
10
N170 Latency (ms)
Objective: To evaluate relations among social function and dysfunction and
neural markers of gaze processing during simulated face-to-face interactions in
individuals with ASD, SZ, and TD controls.
12
N170 Latency (ms)
Specifically, we evaluated N170 and P300 response to direct and averted gaze
in adults with ASD, SZ, and typical development (TD) to determine betweengroup differences as well as transdiagnostic patterns in neural processes
associated with face decoding. In parallel, we evaluated social functioning with
a battery of diagnostic, clinical, and neurocognitive assessments. In this way,
we evaluated whether specific abnormalities in gaze processing differed by
diagnostic category or were general indicators of social dysfunction across
neurodevelopmental disorders.
EEG and ET Data Acquisition and Collection:
• EEG recorded at 1000 Hz with a 128-channel Hydrocel Geodesic Sensor net
• ET data collected using an Eyelink-1000 remote camera system
ERP Preprocessing and Analysis:
• Data were filtered from 0.1-100 Hz. Data were then
rereferenced to average reference, segmented from 100ms to 400ms relative to shift in stimulus gaze,
baseline corrected, and artifact detected.
• N170 was extracted from right occipitotemporal
electrodes 130-250ms. P300 was extracted over fronto- Figure 2. Recording
central scalp from 250-400ms. Mean amplitude (N170, sites. Fronto-central
P300) and latency (N170) were included as within- P3 and right occipitotemporal N170 were
subject variables in repeated measures ANOVAs (with extracted from
group as the between-subject factor), with follow-up electrode clusters
one-way ANOVAs and t-tests to clarify effects.
depicted.
N170 Amplitude (uV)
Social difficulties, including deficits in maintaining and interpreting social gaze
and in recognizing faces and emotions, are hallmark features of autism
spectrum disorder (ASD). Atypical social functioning and gaze processing are
not, however, unique to ASD. Both are also affected in schizophrenia (SZ), a
disorder with genetic, neurobiological, and phenotypic commonalities with ASD.
This study utilized novel methods, integrating eye-tracking and
electrophysiology (EEG), to study social behavior and brain function during
simulated face-to-face interactions in individuals with ASD and SZ.
Figure 4. P300 response to direct and averted gaze in ASD, SZ, and TD. Main effect of
Condition, F(1,41)=12.27, p < 0.01, wherein all groups showed enhanced amplitude to
averted vs. direct gaze. For shift from averted to direct gaze, individuals with ASD showed
increased P300 amplitude relative to SZ (p = 0.043), but no differences between ASD and
TD or between SZ and TD were observed.
These preliminary findings suggest that, across multiple neurodevelopmental
disorders, neural indices of social processing can reveal differences in gaze
processing related to clinically-relevant social difficulties that behavioral measures
of overt symptomatology do not capture.
Funding Sources: NIMH R01107426 (McPartland, Srihari); NIMH R01 MH100173 (McPartland); NIMH R01
MH100173-02S1 (McPartland); Autism Speaks Translational Postdoctoral Fellowship (Naples); NARSAD Young
Investigator Award (Foss-Feig); Autism Science Foundation Research Mini Enhancement Grant (Foss-Feig); NIH
DP50D012109-02 (Anticevic); Yale Center for Clinical Investigation (Anticevic); NIH R01 MH103831 (Srihari)