Bipolar disorder (BD) is a severe psychiatric condition associated with poor functional outcomes, including impaired familial relations and poorer social adjustment (Sanchez-Moreno et al., 2009; Tatay-Manteiga et al., 2018). Evidence suggests that social cognitive deficits contribute to this dysfunction in BD (Vlad et al., 2018). A critical determinant of social cognition is the ability to accurately discriminate the eye gaze direction of others, and its alteration may play a role in the functional impairment in BD. The present study sought to inform our understanding of altered social cognition in BD by investigating the neural correlates of altered gaze perception in BD.

As humans, the ability to accurately discriminate gaze direction allows us to make informed judgements about others’ intentions and inner states (Shepherd, 2010). Successful gaze discrimination relies on a range of basic- and higher-level cognitive processes (Itier & Batty, 2009). At the most basic level, we encode sensory information. Then, facial features including gaze direction (Berchio et al., 2017), head orientation (Itier et al., 2007), and facial emotion (Berchio et al., 2017, 2019; Carretié et al., 2001) are integrated to form a holistic representation (Bentin et al., 1996; Itier & Batty, 2009). At higher levels, we inhibit automatic judgements or emotional reactions to make informed judgements about others’ mental states (Sabbagh et al., 2004) and intentions (Carrick et al., 2007; Itier & Batty, 2009). Neuroimaging data show that earlier processes depend on temporal brain areas specialized for processing facial features (Allison et al., 2000), while higher-level processes rely on more frontal areas involved in higher level cognitive processes like self-referential processing and cognitive control (Cavanagh & Frank, 2014). Studies have shown that disrupted gaze perception observed behaviorally is associated with altered regional neural activity and connectivity between these regions (Tso, Burton, et al., 2021; Tso, Angstadt, et al., 2021). This suggests that effective gaze perception depends on the integrity of both local neural activity and inter-regional neural communications.

Behavioral studies show that BD is associated with altered face processing and theory of mind (ToM) (Samamé et al., 2012) and emerging evidence also indicates gaze processing abnormalities in BD. BD may over endorse eye contact relative to HC (Yao et al., 2017) and have difficulties inferring others’ intentions and emotional states from the eye region (Marotta et al., 2018). Difficulties with gaze perception may contribute to poorer social functioning (Tso et al., 2020) and may also play a role in the presentation of core symptoms associated with BD. For example, chronic over-perception of self-directed gaze in BD could interact with cognitive and affective states to exacerbate feelings of inflated self-esteem or grandiosity during hypo/mania (e.g., “Everyone is looking at me. I must be famous.”). Thus, altered gaze perception in BD has both clinical and broader functional importance. Understanding the neural mechanisms which perpetuate these difficulties is a crucial step towards translational research designing novel, targeted interventions (Chase et al., 2020; Reinhart et al., 2015; Yamada et al., 2021, 2022).

Thus far, only a few studies have examined the neural mechanisms of altered gaze processing in BD. One functional magnetic resonance imaging (fMRI) study found that, during gaze discrimination, individuals with BD exhibit reduced activation of medial prefrontal areas (implicated in higher level cognitive processing) and tempo-parietal areas (specialized for face processing) (Tso, Burton, et al., 2021). Along similar lines, one electroencephalography (EEG) study found general reductions in event-related potential (ERP) amplitudes over fronto-parietal areas during gaze perception in BD (Berchio et al., 2017). Together this suggests abnormal local recruitment of fronto-parietal regions involved in gaze processing. Recent perspectives on the neurobiology of BD emphasize the role of dysfunctional brain connectivity in the phenomenology of BD (Chase & Phillips, 2016; Dima et al., 2013; Perry et al., 2019; Yoon et al., 2020), leading some to frame BD as a disorder of functional dysconnectivity between cognitive/affective networks (Strakowski et al., 2012). Therefore, to fully understand the neural bases of aberrant gaze perception in BD, we must consider the role of impaired neural communications, in addition to local dysfunctions.

EEG is a promising means of measuring brain activity in order to investigate questions of this kind. Neural oscillations, electrical signals emanating from coordinated neuronal populations, can be extracted from EEG (Buzsáki & Draguhn, 2004). Neural oscillations are thought to reflect event-related functional de/synchronization (i.e., de/coupling) in the brain and, therefore, offer insight into functional network dynamics underlying cognitive processes (Bastiaansen et al., 2011) like gaze perception. Oscillations are traditionally categorized into different frequency bands simultaneously (including theta (4-8 Hz), alpha (8-12 Hz), beta (12-30 Hz), gamma (>30 Hz)) and activity at different frequencies are thought to reflect different physiological mechanisms underlying cognitive processing. For example, midline anterior theta is associated with higher-level cognitive processes like ToM (Seymour et al., 2018) and cognitive control (Cavanagh & Frank, 2014), while gamma over sensory cortices is linked to perceptual processing (Herrmann et al., 2010). Increasing evidence suggests that the synchrony or ‘coupling’ of oscillations from distal neural populations may underlie neural communications by way of cross-frequency phase-amplitude coupling (PAC), whereby the amplitude of high-frequency activity in one region is modulated by the phase of low-frequency activity at another (Canolty & Knight, 2010). Theta-gamma PAC is one of the most common forms of coupling in the human cortex and can shed light on inter-region communications supporting cognition. Studying neural oscillations using EEG, therefore, has the potential to provide insight into both regional activity and network dynamics (Bastiaansen et al., 2011) supporting gaze processing.

To date, no electrophysiology studies have examined neural oscillatory abnormalities associated with gaze processing in BD. However, studies examining the neural oscillatory correlates of face processing in BD offer some preliminary insights. Studies of affective and configural face processing using magnetoencephalography (MEG; measures magnetic fields produced by neural activity) show decreased midline anterior gamma in BD compared to healthy controls (HC) (Lee et al., 2010; Liu et al., 2012). In addition, there is evidence of alterations in the face-sensitive N170 ERP component over parietal regions during face processing in BD (Degabriele et al., 2011; Ibáñez et al., 2012; Sokhadze et al., 2011) (but between-study results are mixed (Feuerriegel et al., 2015)). Because N170 appears to reflect underlying theta activity (Tang et al., 2008; Torrence et al., 2021), we assume that altered N170 in BD could signal underlying theta-band abnormalities. Together, these findings indicate that examining midfrontal parietal oscillatory activity, particularly in theta and gamma bands, could provide insight into the neurobiology of abnormal gaze processing in BD.

The present study sought to address the existing knowledge gaps of neural oscillatory correlates of altered gaze perception in BD. To achieve this, we used time-frequency analysis of EEG data collected from BD and HC during a gaze discrimination task. We analyzed activity over midfrontal and parietal scalp areas that have previously been associated with higher-level cognitive processing and early face processing, respectively. Analyses captured both local oscillatory activity and inter-regional communications between these key locations during gaze processing. To characterize local oscillatory activity in BD and HC, we examined oscillatory power over separate midfrontal and parietal scalp locations in theta and gamma frequency bands.

To characterize information flow between these areas in BD and HC, we examined top-down and bottom-up theta-gamma PAC between midfrontal and parietal locations.

Considering that midfrontal theta is associated with higher-level cognition (Cavanagh & Frank, 2014; Seymour et al., 2018), and that higher-level cognition is impaired in BD (Latalova et al., 2011; Marotta et al., 2018; Samamé et al., 2012), we hypothesized that BD would show local reductions in theta power over midfrontal locations. Considering that there is at least some evidence of altered N170 over parietal locations in BD during face processing (Degabriele et al., 2011; Ibáñez et al., 2012; Sokhadze et al., 2011), and that theta activity appears to underlie N170 (Tang et al., 2008; Torrence et al., 2021), we hypothesized that BD would show local reductions in theta power over bilateral parietal locations. Considering evidence of reduced functional connectivity between higher-level cognitive regions and other brain areas in BD (Strakowski et al., 2012), altered prefrontal cortical functions in BD (Yoon et al., 2020), and difficulties with higher-level cognitive processing—but not low-level perceptual processing—in BD (Yao et al., 2017), we hypothesized that BD would have reduced top-down theta-gamma PAC between fronto-parietal areas during gaze processing. Additionally, little is known about the potential role of intermediate frequency bands in gaze processing in BD, so we also explored whether alpha and beta power differed between BD and HC.