Abstract:

The neural correlates of consciousness (NCCs) are debatable due to the confounding effects of subjective reports. In addition, although previous studies have suggested that vMMN is relatively insensitive to the manipulation of visual attention, the relationship between vMMN and visual consciousness remains unclear. The inattentional blindness paradigm can not only effectively manipulate visual consciousness, but also explore the conscious processing without relying on subjective reports. Therefore, we used this paradigm to manipulate visual consciousness. Moreover, we introduced emotional (happy and fearful) faces, which are biologically and socially significant visual stimuli, to explore NCCs and the relationship between automatic detection of changes and visual consciousness.
Fifty-six Chinese participants took part in the present study. We recorded electroencephalography (EEG) in three phases. In phase A, the participants needed to detect changes of the red dots. However, because they were not informed of the existence of emotional faces, 26 participants were unconscious of the task-irrelevant emotional faces. In phase B and C, all participants were informed about the emotional faces. Thus, they were conscious of the emotional faces. Specifically, in phase B, the participants still needed to detect changes of the red dots, and the emotional faces are task-irrelevant. However, in phase C, the participants were asked to detect changes of emotional faces, and thus emotional faces were task-relevant. To check the conscious state of emotional faces, participants were required to fill out an awareness questionnaire after completing phases A and B. Then the participants were divided into unconscious group and conscious group according to their conscious state of emotional faces in phase A.
For NCCs, two analyses of ERP amplitude of standard stimuli in Phase A and B of the unconscious group were performed. First, for the ERPs at PO7 and PO8 electrodes, a three-factor repeated-measures analysis of variance (rANOVA) of Phase (A, B) × Emotion (Fear, Happy) × Hemisphere (PO7, PO8) was used. Second, for the ERPs at the three electrodes of FCZ, CZ and CPZ, a three-factor rANOVA of Phase (A, B) × Emotion (Fearful, Happy) × Electrodes (FCZ, CZ and CPZ) was adopted. Similarly, two three-factor rANOVA were used for the effect of task relevance. However, exploring the effect of task relevance requires an analysis of the ERPs of standard stimuli in phase B and C of all participants. For vMMN, we analyzed the ERPs at PO7 and PO8 electrodes during the 250~350 ms window of time.
Results can be summarized as following. (1) For NCCs, the ERPs at PO7 and PO8 during the 200~300ms window of time were analyzed, which yielded a significant main effect of Phase [F(1, 25) = 16.385, p < 0.001, η²_p = 0.396]. This showed that the emotional faces in phase B evoked stronger negativity than in phase A for the unconscious group, suggesting that the conscious processing of emotional faces evoked visual awareness negativity (VAN). The ERPs at PO7 and PO8 during the 400~600 ms window of time were analyzed, which yielded a significant main effect of Phase [F(1, 25) = 15.79, p = 0.001, η²_p = 0.39]. This showed that the emotional faces in phase B evoked stronger positivity than in phase A for the unconscious group, suggesting that the conscious processing of emotional faces evoked late occipital positivity (LOP). Moreover, the ERPs at FCZ, CZ and CPZ during the 300~400ms window of time were analyzed, which yielded a significant main effect of Phase [F(1, 25) = 11.481, p = 0.002, η²_p = 0.32]. This showed that the emotional faces in phase B evoked stronger positivity than in phase A for the unconscious group, suggesting that the conscious processing of emotional faces evoked late positivity (LP).
(2) For the effect of task relevance, the ERPs on PO7 and PO8 in the 180~250ms interval were analyzed, which yielded a significant main effect of Phase [F(1, 55) = 20.93, p < 0.001, η²_p = 0.28]. This showed that compared with the task-irrelevant condition (phase B), the emotional faces under the task-relevant condition (phase C) evoked stronger negativity, suggesting that task relevance evoked selection negativity (SN). The ERPs on PO7 and PO8 in the 400~500ms interval were analyzed and the main effect of Phase is significant [F(1, 55) = 6.12, p = 0.02, η²_p = 0.1]. This showed that compared with the task-irrelevant condition, the emotional faces under the task-relevant condition evoked stronger positivity, suggesting that task relevance evoked LOP. Moreover, the ERPs on FCZ, CZ and CPZ in the 300~400ms interval were analyzed and the main effect of Phase is significant [F(1, 55) = 29.77, p < 0.001, η²_p = 0.35]. The results showed that compared with the task-irrelevant condition, the emotional faces under the task-relevant condition evoked stronger positivity, suggesting that task relevance evoked LP that may reflect the post-perceptual processing. Therefore, this study provides evidence that LP and LOP are NCCs without the confounding effects of task relevance. In short, VAN may reflect the early perceptual process of emotional faces, LP and LOP may reflect the further process of classifying and recognizing the representations of emotional faces, such as assessing the emotional valence of faces.
(3) For the relationship between vMMN and consciousness, we analyzed the data of phase A with a four-factor rANOVA of Stimulus-type (standard, deviant) × Emotion (Fearful, Happy) × Hemisphere (PO7, PO8) × group (conscious, unconscious). The rANOVA revealed that the main effect of Stimulus-type was significant [F(1, 54) = 9.43, p = 0.003, η²_p = 0.149], and the interaction between Stimulus-type and other factors was not significant [ps > 0.05]. This showed that compared to standard emotional faces, deviant ones evoked stronger negativity in phase A. Importantly, the vMMN effect was observed for both the conscious and unconscious group in the phase A. Furthermore, no amplitude difference of vMMN was observed between the aware (phase B) and the unaware (phase A) conditions among unconscious group [t(25) = 0.14, p = 0.88], suggesting that the automatic processing of emotional faces is independent of visual consciousness. Compared with Chen (2020), this study provides evidence that the automatic processing of emotional faces is independent of visual consciousness under the condition that the unconsciousness level is manipulated more effectively.
(4) In addition, we analyzed the vMMN effect with a ANOVA of Phase (A, B and C), which yielded a significant main effect of Phase [F(2, 110) = 5.24, p = 0.007, η²_p = 0.087]. And compared with the task-irrelevant condition (phase B), the vMMN amplitude under the task-relevant condition (phase C) was larger (p = 0.003), suggesting that task relevance modulates the amplitude of vMMN and the attentional effect of task relevance promotes the automatic processing of emotional faces.
The conclusions of this study can be summarized as following. (1) VAN is a NCC under the condition of avoiding confounding effects of visual attention, and LP and LOP are NCCs without the confounding effects of task relevance. (2) The visual awareness of emotional faces has different ERP indicators at different time stages. Specifically, VAN reflects the early perceptual experience, LP and LOP reflect the late conscious experience of non-perceptual information. (3) The automatic processing of emotional faces is independent of visual consciousness but is modulated by visual attention.

Key words: inattentional blindness, emotional faces, visual mismatch negativity, visual awareness negativity, late positivity