Rapid extraction of emotional expression: evidence from evoked potential fields during brief presentation of face stimuli

Abstract

Although the emotional expression of faces is believed to be accessed rapidly, previous ERP studies hardly found correlates of these processes. Here, we report findings from a study that investigated dichoptic binocular interaction using emotional face stimuli. Thirty-one subjects were briefly presented with schematic normal and scrambled faces (of neutral, positive, or negative expression) that occurred simultaneously in the left and right visual fields. Stimuli for both eyes could be congruent (control) or incongruent (dichoptic). Subjects decided which of the superimposed images in both hemi-fields appeared more “face-like” and during this task, the EEG was recorded from 30 channels. VEPs were analysed topographically according to the influence of the different experimental conditions (defined by presentation form, emotional expression, and location). Behavioural responses to the ambiguous dichoptic stimuli demonstrated a functional eye dominance not related to visual acuity and conventional eye preference. Electrophysiological data revealed three components with mean latencies of 85, 160, and 310 ms. Topography of the second component (equivalent to the face-related N170) differed in left–right and anterior–posterior direction compared with simple checkerboard stimuli. Dichoptic presentation caused reduced field strength of all three, and increased latency of the first component. Faces with negative expression yielded largest field strength of the second and third components. Besides that, emotional expression affected topography not only of late, but also the first component. This provides new evidence about the timing of perceptual processes related to facial expression, indicating that already VEP components occurring at 80–90 ms are sensitive to emotional content.

Introduction

Faces confront visual perception with exceptional processing demands, since they share a very similar overall configuration that, however, may change drastically by mimic movements that convey emotional state, or are induced by speech. Nevertheless, at a glance we can determine a persons identity, sex and age, and are able to read his or her mood or possible intentions.

These obviously specialised abilities, along with the consequences of their disturbance in e.g. prosopagnosia [6], made face perception a comprehensively studied topic of cognitive neuroscience. Viewing faces evokes differential neural activity in comparison to different other objects, as determined by studies using functional neuroimaging [16], human [1], [2], [15], and primate [7] electrophysiology. By now, we are beginning to understand the contribution of distinct brain structures to different aspects of face processing, as facial identity versus gesture or expression [13], which is also paralleled in neuropsychological findings demonstrating a potential dissociation of these two abilities [39].

Less clear so far is the time scale at which processes related to these different aspects of face perception operate. Emotional information, as concerned in this study, is widely acknowledged to be extracted in a rapid and automated way according to behavioural findings [12], [29]. This leads to the hypothesis that neural activity as a correlate of these processes should also differ already at early “sensory-specific” processing stages in the brain when faces of different emotional expression are perceived, even if this information may be irrelevant for the actual task at hand. When closer examining findings of previous human ERP studies as a method that should, in principle, provide sensitivity and temporal resolution to tackle this question, it is surprising that these studies mainly found effects of emotional content restricted to comparably late time periods [18], [20], [30], [35]. The lack of early influences in many studies, however, might also be due to methodological issues, since most of these recorded ERPs only in few channels. In consequence, analysis was restricted to waveforms at some locations that are reference-dependent and of limited sensitivity. In contrast, topographical ERP analysis considers potential fields changing over time that can be quantified by reference free measures [24]. Examples are evoked potential studies [31], [37] where the semantic content of word stimuli or the subjects’ attitude towards faces were found to affect brain electric activity comparably early after stimulus presentation.

Applying topographical analysis of brain evoked activity, we will present evidence here supporting the notion that emotional expression is indeed differentiated at a very early level in the brain. We used standardised schematic line drawings of faces, in order to prevent effects of physical stimulus parameters as luminance and contrast, and to exclude additional information related to personal identity. Our findings stem from an experiment that was designed with an additional purpose in mind, to address the influence of higher cognitive stimulus material such as faces on perception during dichoptic presentation. Dichoptic stimulation of both eyes is known to result under appropriate conditions in the phenomenon of binocular rivalry [4], [25], where perception alternates spontaneously between the two different stimuli that are permanently present. There is evidence that not only low-level sensory parameters but also stimulus configuration or meaning may exert an influence on the perceptual outcome during binocular rivalry: Yu and Blake [40], for example, illustrate how a face stimulus in one eye tended to dominate over a meaningless pattern of comparable physical stimulus properties in the other eye. Here, we presented faces laterally in a dichoptic arrangement and asked whether the perceptual outcome in this situation might also reflect such top-down mechanisms, and would be influenced by a higher right hemispheric importance in face processing [32]. Since this form of stimulation requires the short presentation times used in tachistoscopic studies to rule out the influence of eye movements, our experiment is different from the typical binocular rivalry paradigm where spontaneous perceptual alternations occur between stimuli that are presented for a comparably long time. In addition to addressing the neural processing of facial expression, we will present psychophysical and electrophysiological correlates of binocular interaction in our special paradigm, and also relate our findings to classical “face-responsive” potentials [2], [15].