SPECIFICITY OF ERP TO CHANGE OF EMOTIONAL FACIAL EXPRESSION. Michael Wright Centre for Cognition and Neuroimaging, Brunel University, Uxbridge, UB8 3PH, U.K.

1
SPECIFICITY OF ERP TO CHANGE OF EMOTIONAL FACIAL
EXPRESSION.Michael WrightCentre for Cognition
and Neuroimaging, Brunel University, Uxbridge,
UB8 3PH, U.K.
Background A change in emotional facial
expression is known to elicit a prominent N170
component (Miyoshi, et al., 2004 Neuroreport,
15, 911-914). There are many reports of rather
small differences in N170 to different facial
expressions (e.g. Batty Taylor, 2003 Cognitive
Brain Research, 17, 613-620). The present study
examines whether the ERP to expression change
gives greater sensitivity and emotion specificity
than does ERP to appearance of an emotional
face. Method Stimuli The stimuli were
grey-scale images of neutral and emotional faces
from a standardised database (Lyons, et al.
1998). Each trial consisted of presentation of
two facial expressions (of the same person)
successively for 0.5 s without an interval
(Fig.1). There were 10 individual faces in the
database, giving 10 unique stimuli per expression
change. These were presented in blocks,
Fig.2. Group averaged ERP to angry-neutral,
neutral-angry, zoom-neutral and neutral-zoom P4
(green) and FCz (blue). P1 is larger to first
(face onset) stimulus than second (face change)
stimulus. Face onset has a VPP, face change does
not. Face change has a large p2/p3 complex that
is larger to angry than to zoom. N170 peak
latency (at P4) to all onset stimuli is similar
(500-340160ms). Latency to neutral-angry (165ms)
is longer than to neutral-zoom (156ms) and
latency to angry-neutral (174ms) is longer than
to zoom-neutral (160ms).
Participants 20 participants Design In a given
experiment, a comparison was made between two
emotions, or between an emotion change and a
control change. The second variable in each
experiment was the order within each stimulus
pair of emotional and neutral expressions.
Recording and data analysis ERP was recorded
using a Neuroscan system (32 electrode, 10-20,
Synamps amplifier, 0.1 100Hz, Scan 3.2
software). EEG files were epoched, baseline
corrected to the pre-stimulus interval, artefact
rejected (-50µv all channels) and averaged. The
grand mean and SD of the average ERP was
computed. Difference waveforms were obtained for
pairs of experimental conditions in each
participant, enabling t-values to be computed for
each electrode. Grand average ERPs were baseline
corrected to the 50ms preceding the relevant
stimulus (first or second) prior to statistical
comparison of different experimental conditions.

• Results Emotional expression-change produced a
  characteristic ERP with a prominent N170, but
  with a reduced P1 in comparison with face-onset
  stimuli (Fig.1).
• Comparison of angry and non-emotional (control)
  changes. The control was a pixel-matched
  magnification change (zoom). Significant
  enhancement of N170 and later potentials was
  found for angry versus control faces, in both
  angry-neutral and neutral-angry pairs, though in
  the latter condition, differences were larger.
  For angry-neutral versus zoom-neutral,
  significant differences were seen in the N170 of
  the neutral stimulus (Fig 3ab).
• b) Comparison of happy and fearful expression
  changes.
• Significant differences were found for N170-VPP
  and late potentials for neutral-happy versus
  neutral-fear but not for happy-neutral versus
  fear-neutral. A difference was also noted in the
  N170-VPP to the neutral stimulus in happy-neutral
  and fear-neutral conditions (Fig 3cd, Fig4).

Conclusions The sensitivity of ERP to facial
expressions is greater for expression-change
stimuli than for face-onset stimuli. Latency,
waveform and amplitude differences around N170,
P2, P3 are evident between expression-change and
size-change images. Expression-change stimuli
show selectivity for different emotions
differences were found between happy and fearful
stimuli. There is some evidence that the ERP to a
neutral face is affected by the preceding
emotional face.
"Coding Facial Expressions with Gabor Wavelets"
Michael J. Lyons, Shigeru Akamatsu, Miyuki
Kamachi, Jiro Gyoba Proceedings, Third IEEE
International Conference on Automatic Face and
Gesture Recognition, April 14-16 1998, Nara
Japan, IEEE Computer Society, pp. 200-205.