Emotional face is a classic social stimulus conveying information about others’ emotional states. As an important social signal, it plays a crucial role in daily interpersonal interactions (Fredrickson, 2001). Deficits in the processing of emotional faces are often associated with poor social cognition and social functioning (Gao et al., 2021). A growing body of research has found that people with mental illness have deficits in emotional facial recognition (Cotter et al., 2018; Gao et al., 2021; Hwang et al., 2021; Krause et al., 2021; Wang et al., 2023), especially in the recognition of negative emotional faces such as fear and sadness (Pena-Garijo et al., 2023; Shenoy et al., 2019; Suárez-Salazar et al., 2020; Wang et al., 2023).

Previous studies have found that the brain's processing of emotional faces is rapid, unconscious, automatic, and volume-free (Anderson et al., 2003; Brooks et al., 2012; Palermo and Rhodes, 2007). Emotional information is prioritized for processing, even when attention is not focused on the current task (Carretié, 2014; Hinojosa et al., 2015; Kovarski et al., 2017), thus indicating an effect of emotion facilitation. The emotional significance of stimuli is initially evaluated under pre-attentive conditions, and subsequently, stimuli that have been tagged with emotional significance are prioritized for access to selective attention mechanisms that operate within a limited-capacity system (Compton, 2003). This hierarchical processing framework, particularly its pre-attentive initiation phase, carries critical implications for understanding atypical emotional perception.

To precisely capture these millisecond-level neural dynamics, event-related potentials (ERPs) have emerged as a pivotal neurophysiological tool, offering unparalleled temporal resolution for investigating the early perceptual stages of facial emotion processing. Current research on pre-attentive processing of emotional faces has adopted a “multistage” cognitive mode, including the detection of basic visual processes (Kim et al., 2015), the encoding of facial structure (Liang et al., 2019; Zhang et al., 2023), and the decoding of emotion information (Priyesh et al., 2021; Sandhya et al., 2019).

The mismatch negativity (MMN), originally identified in auditory research (Näätänen and Winkler, 1999; Näätänen et al., 2004), manifests as a neurophysiological signature of prediction-error encoding during pre-attentive processing, with its visual counterpart (vMMN) showing pronounced activity in occipitotemporal regions. Building upon this neural signature, researchers have developed the expression-related visual mismatch negativity (EMMN) - an electrophysiological index that sensitively tracks automatic neural responses to dynamic changes in facial emotional information (Stefanics et al., 2014; Zhao and Li, 2006). To operationalize this measure, experimental paradigms typically employ multi-deviant oddball designs that systematically manipulate facial emotion deviations while controlling for low-level visual confounds (Csukly et al., 2013). The EMMN amplitudes were calculated by subtracting the amplitudes elicited by high-probability standard stimuli from the amplitudes elicited by low-probability deviant stimuli (Astikainen and Hietanen, 2009; Czigler, 2014). Using the oddball paradigm, several studies have found decreased EMMNs in patients with schizophrenia and major depressive disorder, indicating deficits in the automatic processing of emotional faces under pre-attentive conditions (Chang et al., 2010; Csukly et al., 2013; Priyesh et al., 2021; Wu et al., 2017; Yin et al., 2018).

These early-stage perceptual deficits may mechanistically underlie the transdiagnostic phenomenon of social anhedonia (Barkus and Badcock, 2019) - a core symptom characterized by reduced pleasurable experiences and motivation for social interactions (Gooding and Pflum, 2022) that predicts psychiatric vulnerability across schizophrenia spectrum disorders (Meehl, 1962), major depression (Atherton et al., 2015), and autism spectrum conditions (Gadow and Garman, 2020). Longitudinal studies further establish social anhedonia as a robust prodromal marker preceding clinical symptom onset (Gooding et al., 2005).

Some previous studies have preliminarily found the recognition defects of emotional faces in individuals with high social anhedonia (HSA) and its relationship with pre-attentive processing. In one study, using a task that requires participants to classify different schematic faces by valences, researchers found that individuals with HSA performed a blunted categorization advantage for happy over sad faces with comparable 170 latencies in response to happy and sad faces (Nie et al., 2020). The categorization advantage of happy faces exists in healthy people (Nummenmaa and Calvo, 2015), with positive (happy) faces typically being recognized faster than negative ones. This phenomenon is associated with earlier N170 latencies for happy faces than for negative faces (Batty and Taylor, 2003). The disappearance of N170 specificity in the HSA group may account for the role of pre-attentive processing deficits in this group's poor performance on face classification tasks (Astikainen and Hietanen, 2009). Neuroimaging evidence from an fMRI study using an implicit masked face processing task also indicated that individuals with HSA exhibited abnormal automatic activation increases to sad faces in the bilateral thalamus and left red nucleus compared to those with low social anhedonia (LSA) (Günther et al., 2017). These studies may imply facial emotion pre-attentive processing deficits in HSA groups, but this hypothesis has not been tested.

In addition, the traditional oddball paradigm used in previous studies involved presenting standard stimuli and deviant stimuli with different emotional valences in a sequence. It may introduce the influence of low-level physical differences in stimuli. Moreover, the excessive presentation of standard stimuli in the traditional paradigm may result in neural refractoriness (Kreegipuu et al., 2013), which further inhibits the neural response to subsequent stimuli presentations. To avoid this problem, an equal-probability sequence (Jacobsen and Schröger, 2001) was added to the oddball paradigm. In this modified oddball-control paradigm, EMMN is derived by subtracting the ERP elicited by emotional faces in the control sequence from the ERP elicited by faces of the same emotional valence in the oddball sequence. This approach effectively controls for neural refractoriness and eliminates the influence of low-level physical differences between stimuli (Ding et al., 2020), thus resulting in a genuine EMMN based on the probability of stimulus occurrence (Stefanics et al., 2014).

In this study, we adopted the modified oddball-control paradigm to investigate the pre-attentive processing of emotional faces in individuals with HSA. It was hypothesized that the HSA group would exhibit deficits in the pre-attentive processing of emotional faces, with EMMN amplitudes elicited by emotional faces being significantly lower than those of the LSA group. Moreover, the EMMN amplitudes would be significantly correlated with self-reported anhedonia.