Visual Event-Related Potentials under External Emotional Stimuli in Bipolar I Disorder with and without Hypersexuality

Hypersexuality is related to functions of personality and emotion and is a salient symptom of bipolar I disorder especially during manic episode. However, it is uncertain whether bipolar I disorder with (BW) and without (BO) hypersexuality exhibits different cerebral activations under external emotion stimuli. In 54 healthy volunteers, 27 BW and 26 BO patients, we administered the visual oddball event-related potentials (ERPs) under external emotions of Disgust, Erotica, Fear, Happiness, Neutral, and Sadness. Participants’ concurrent states of mania, hypomania, and depression were also evaluated. The N1 latencies under Erotica and Happiness were prolonged, and the P3b amplitudes under Fear and Sadness were decreased in BW; the P3b amplitudes under Fear were increased in BO. The parietal, frontal, and occipital activations were found in BW, and the frontal and temporal activations in BO under different external emotional stimuli, respectively. Some ERP components were correlated with the concurrent affective states in three groups of participants. The primary perception under Erotica and Happiness, and voluntary attention under Fear and Sadness, were impaired in BW, while the voluntary attention under Fear was impaired in BO. Our study indicates different patterns of visual attentional deficits under different external emotions in BW and BO.


Introduction
Mood disorder is a psychiatric illness being predominated in emotional and cognitive disturbances [1,2], and its main type is bipolar disorder, a lifelong and recurrent disease being characterized by fluctuant mood and energy episodes of mania/hypomania and depression [3]. Bipolar disorder affects over 1% of the globe population, with high incidence rates of disability and suicide [4]. Its exact pathogenesis is not fully known, but nearly 70% of it is heritable [5]. Other contributing factors to its onset might be the monoaminergic dysregulation, inflammatory disturbance, and adverse environmental exposure [5][6][7]. In clinics, the precision diagnosis and management of bipolar disorder and its subtypes remain difficult after tremendous efforts these years, which might be due to the nonspecific symptoms or a depressive episode at the beginning and to the substantial psychiatric and somatic comorbidities [3,4].
For instance, the bipolar I disorder (BD I) has prominent impulsivity, irritability, and distractibility [8], and is characterized by the recurrent mania and depression episodes [9] which are subjected to severe dysfunctions of emotional processing and regulation [10,11]. Patients with BD I experienced great negative affectivity in their daily life [12], and showed impairment in recognizing facial happiness and disgust [13]. Interestingly, deficits in current study, we would like to use visual ERPs under externally emotional stimuli in BW and BO patients. We have hypothesized that: (1) compared to BO, BW would display more significant abnormalities in ERP components representing primary perception or involuntary attention under positive emotions, and in those representing voluntary attentional processing under negative emotions; (2) both BW and BO would have different patterns of cerebral activation under different external emotions; and (3) the ERP morphologies would correlate with the concurrently affective states of BW and BO.

Questionnaires
All participants were asked to complete the three self-assessment questionnaires below in a quiet room, using a paper-and-pencil style.
A. The Mood Disorder Questionnaire (MDQ; [47]) is an instrument with 13 dichotomous items (yes/no) assessing mania/hypomania symptoms and behaviors, and two items evaluating the frequency of those symptoms and the extent of functional impairment. The internal reliability of the 13 dichotomous items was 0.77 in the current study.
B. The Hypomania Checklist-32 (HCL-32; [48]) comprises 32 items for detecting hypomanic symptoms. There are 32 dichotomous items (yes/no) regarding emotions, thoughts, or behaviors of hypomania, and other items about the duration, impact on family, social as well as work life, or other people's reactions. Its internal reliability was 0.76 in the current study.
C. The Plutchik-van Praag Depression Inventory (PVP; [49]) consists of 34 items describing depression symptoms. Three scale points (0, 1, 2) of each item are corresponding to increasing tendencies of depression. If participants score between 20 and 25, they are considered to have "possible depression", or "depression" if they score above 25. The internal reliability of this inventory was 0.82 in the current study.

External Emotional Stimuli
The external emotional stimuli were composed of pictures selecting from the International Affective Picture System [50] and sounds from the International Affective Digital Sounds [51]

ERP Paradigm
After completing the above-mentioned questionnaires, participants were led to a dimly lit room and seated at 100 cm from a computer screen. Six successive sessions (Disgust, Erotica, Fear, Happiness, Neutral, or Sadness) with a two-minute interval between adjacent sessions were randomly presented for each participant. Within each session, a fixation cross in the middle of a black background were presented for 3000 ms, followed by 150 ERP trials (each trial lasting for 2400 ms), with an inter-trial interval of 1200~1500 ms. Within each trial, an external emotional stimulus of either Disgust, Erotica, Fear, Happiness, Neutral, or Sadness was shown for 2000 ms; then either a standard (a square of 40 mm × 40 mm, lasting for 400 ms) or target (a circle of 40 mm in diameter, 400 ms) stimulus appeared in the middle of the black background (Figure 1). In a randomized order, the standard stimuli were delivered 120 times (80%) and the target stimuli 30 times (20%). Participants were instructed to respond to the target stimuli actively by pressing a button with their right index finger as soon as possible, and no reaction was needed to the standard stimuli or emotional displays.

ERP Paradigm
After completing the above-mentioned questionnaires, participants were led to a dimly lit room and seated at 100 cm from a computer screen. Six successive sessions (Disgust, Erotica, Fear, Happiness, Neutral, or Sadness) with a two-minute interval between adjacent sessions were randomly presented for each participant. Within each session, a fixation cross in the middle of a black background were presented for 3000 ms, followed by 150 ERP trials (each trial lasting for 2400 ms), with an inter-trial interval of 1200~1500 ms. Within each trial, an external emotional stimulus of either Disgust, Erotica, Fear, Happiness, Neutral, or Sadness was shown for 2000 ms; then either a standard (a square of 40 mm × 40 mm, lasting for 400 ms) or target (a circle of 40 mm in diameter, 400 ms) stimulus appeared in the middle of the black background ( Figure 1). In a randomized order, the standard stimuli were delivered 120 times (80%) and the target stimuli 30 times (20%). Participants were instructed to respond to the target stimuli actively by pressing a button with their right index finger as soon as possible, and no reaction was needed to the standard stimuli or emotional displays. Figure 1. Timeline of events in the oddball paradigm. Participants were instructed to respond to a circle picture (target stimulus) as quickly as possible by pressing a button with their right index finger, and to do nothing to the square picture (standard stimulus) or emotional scenes (taking Neutral as an example). Note: +, fixation cross representing the beginning of the paradigm.

ERP Recording
EEG signals were recorded with 32 channel elastic electrocap (Electro-Cap International, Inc., Eaton, OH, USA) according to the 10-20 International System. The impedance of each electrode was maintained below 10 kΩ, and EEG signals were amplified by a DC amplifier (the ANT amplifier, Enschede, The Netherlands) with a sampling rate of 1024 Hz. Bipolar recordings of the electro-ocular activity were collected by electrodes placed at the outer canthus and supraorbitally to the right eye. Referred to the average activity of the two mastoid electrodes (M1 and M2), potentials were analyzed offline using a bandpass of 0.01~30 Hz in ASA software 4.7.3 (ANT Software B.V., Enschede, The Netherlands). The sampling epoch was 100 ms pre-stimulus and 600 ms post-stimulus. Any sweep in which the EEG exceeded ±70 μv or with electro-ocular activity was excluded from averaging.
Nine electrodes in frontal, central and parietal sites, i.e., F3, Fz, F4, C3, Cz, C4, P3, Pz, and P4 were selected, and ERP morphology determined by target stimuli were analyzed in terms of peak latency and baseline-to-peak amplitude. Latency ranges of potentials were 70~200 ms for N1, 150~300 ms for P2, 210~390 ms for N2, 300~500 ms for P3a, and 400~580 ms for P3b. Moreover, the reaction times and hit accuracies in response to target stimuli were recorded. Participants were instructed to respond to a circle picture (target stimulus) as quickly as possible by pressing a button with their right index finger, and to do nothing to the square picture (standard stimulus) or emotional scenes (taking Neutral as an example). Note: +, fixation cross representing the beginning of the paradigm.

ERP Recording
EEG signals were recorded with 32 channel elastic electrocap (Electro-Cap International, Inc., Eaton, OH, USA) according to the 10-20 International System. The impedance of each electrode was maintained below 10 kΩ, and EEG signals were amplified by a DC amplifier (the ANT amplifier, Enschede, The Netherlands) with a sampling rate of 1024 Hz. Bipolar recordings of the electro-ocular activity were collected by electrodes placed at the outer canthus and supraorbitally to the right eye. Referred to the average activity of the two mastoid electrodes (M1 and M2), potentials were analyzed offline using a band-pass of 0.01~30 Hz in ASA software 4.7.3 (ANT Software B.V., Enschede, The Netherlands). The sampling epoch was 100 ms pre-stimulus and 600 ms post-stimulus. Any sweep in which the EEG exceeded ±70 µv or with electro-ocular activity was excluded from averaging.
Nine electrodes in frontal, central and parietal sites, i.e., F3, Fz, F4, C3, Cz, C4, P3, Pz, and P4 were selected, and ERP morphology determined by target stimuli were analyzed in terms of peak latency and baseline-to-peak amplitude. Latency ranges of potentials were 70~200 ms for N1, 150~300 ms for P2, 210~390 ms for N2, 300~500 ms for P3a, and 400~580 ms for P3b. Moreover, the reaction times and hit accuracies in response to target stimuli were recorded.

Statistical Analyses
One-way ANOVA was applied to the scale scores of MDQ, HCL-32, and PVP as well as reaction times in the three groups of participants. The latencies and amplitudes of ERP component in the three groups were analyzed by two-way ANOVA, i.e., group (3) × electrode (9). Whenever a significant main effect was detected, the Bonferroni test was employed as a post-hoc comparison. The p < 0.05 at no less than three coaxial electrodes (frontal, central, posterior in lateral axis; left, midline, right in sagittal axis) were considered to be significant and meaningful for group companions. Relationships between ERP components and questionnaire scores were examined using the Pearson correlation test, and only significant correlations with p < 0.01 at no less than three coaxial electrodes were considered as stable and meaningful.
The respective 3D sources were reconstructed based on data obtained at the 32 electrodes, to observe the involvement of cerebral areas corresponding to the significant differences of target stimuli under specific external emotional stimuli in the three groups. The source reconstruction applied the SPM12 software package, running in Matlab R2014b (Mathworks Inc., Natick, MA, USA). The figures of the averaged source map were generated by xjView 10.0 (http://www.alivelearn.net/xjview (accessed on 18 February 2022)).

ERP Components
The ERP components N1, P2, N2, P3a, and P3b under differently external emotions were collected in three groups of participants. For the sake of brevity, only data showing significant differences between groups were reported here, all other data are available upon request. There were significant differences on N1 latencies under Erotica (group effect, However, the post-hoc comparisons did not detect any meaningfully between-group differences. As an example, the grand averages of ERPs under external Erotica at nine electrodes in three groups were presented in Figure 2. The differences of N1 latencies under Erotica and Happiness, and P3b amplitudes under Fear and Sadness at Cz in three groups were illustrated in Figure 3.

Source Reconstructions
After significant differences on N1 latencies and P3b amplitudes under different external emotions were found, we located the possible neural sources for these components by performing 3D source reconstruction in 70-200 ms and 400-580 ms time-windows in three groups, respectively. In controls, the bilateral supramarginal gyri under Erotica and

Source Reconstructions
After significant differences on N1 latencies and P3b amplitudes under different external emotions were found, we located the possible neural sources for these components by performing 3D source reconstruction in 70-200 ms and 400-580 ms time-windows in three groups, respectively. In controls, the bilateral supramarginal gyri under Erotica and right inferior temporal gyrus under Happiness were mainly activated in N1 time windows. During P3b time window, the right inferior occipital gyrus under Fear and bilateral medial frontal gyri were activated. In BW, enhanced processing of left postcentral gyrus, right supramarginal gyrus, and bilateral inferior frontal gyri were found during Erotica and Happiness in N1 time window, and of bilateral lingual gyri as well as left medial frontal gyrus during Fear and Sadness in P3b time window. In BO, the bilateral medial frontal gyri and right inferior temporal gyrus were processed under Erotica in N1 time window and Fear in P3b time window, while the left superior/inferior frontal gyri were involved with Happiness during N1 time window and Sadness during P3b time window (Table 3). As an example, the 3D source reconstruction of N1 to Erotica in three groups was shown in

Relationships between ERPs and Concurrent Affective States
In controls, the N1 amplitudes under Sadness in frontal electrodes were positively correlated with MDQ (n = 54, r = 0.38~0.39, p = 0.003~0.005). In BW, the P3a amplitudes under Erotica in all nine electrodes except for C3 were positively correlated with PVP (n = 27, r = 0.50~0.60, p = 0.001~0.008). In BO, the N1 amplitudes under Fear in right electrodes

Relationships between ERPs and Concurrent Affective States
In controls, the N1 amplitudes under Sadness in frontal electrodes were positively correlated with MDQ (n = 54, r = 0.38~0.39, p = 0.003~0.005). In BW, the P3a amplitudes under Erotica in all nine electrodes except for C3 were positively correlated with PVP (n = 27, r = 0.50~0.60, p = 0.001~0.008). In BO, the N1 amplitudes under Fear in right electrodes were negatively correlated with HCL-32 (n = 25, r = −0.53~−0.58, p = 0.002~0.007). No other meaningful relationship between ERP components and affective states was found in any given group. Taking an example, the correlation between P3a amplitude (Cz) under Erotica and the PVP score in BW was displayed in Figure 5.

Discussion
To the best of our knowledge, this is the first study examining visual ERPs under external emotions in BW and BO. Confirming our hypotheses, we found higher MDQ and HCL-32 scores in both BW and BO than those in controls, which is in accordance with former studies showing that BD I exhibit high levels of manic or hypomanic symptoms [41,52]. N1 latencies under Erotica and Happiness were prolonged in BW when compared to BO or healthy controls. P3b amplitudes were decreased under Fear and Sadness in BW compared to those in BO, and were increased in BO under Fear compared to controls. Abnormal activation of the cerebral areas especially frontal regions was detected corresponding to N1 and P3b components under different external emotions. Moreover, the concurrent affective states were correlated with ERPs in three groups.
In healthy controls, processing in Erotica and Happiness during N1 time window were mainly activated the bilateral supramarginal gyri and right inferior temporal gyrus, respectively. This might be supported by the connection of temporo-parietal lobe under positive emotion [53]. The right inferior occipital gyrus and bilateral medial frontal gyri were activated in Fear and Sadness during P3b time window, which was consistent with the previous studies showing that the neural activations of negative emotion were in right occipital gyrus, right amygdala, and bilateral frontal regions [54,55]. The correlation between mania and the elementary encoding of Sadness might be explained by the fact that the manic patients reported attenuated subjective sensation of facial sadness [56].
In BW, N1 latencies under Erotica and Happiness were prolonged. Since N1 reflects the primary processing of incoming information [44], the prolonged N1 latencies under Erotica and Happiness implied a delayed processing of positive emotions. Hypersexual individuals exhibited fewer positive emotions and displayed a paucity of happiness [57],

Discussion
To the best of our knowledge, this is the first study examining visual ERPs under external emotions in BW and BO. Confirming our hypotheses, we found higher MDQ and HCL-32 scores in both BW and BO than those in controls, which is in accordance with former studies showing that BD I exhibit high levels of manic or hypomanic symptoms [41,52]. N1 latencies under Erotica and Happiness were prolonged in BW when compared to BO or healthy controls. P3b amplitudes were decreased under Fear and Sadness in BW compared to those in BO, and were increased in BO under Fear compared to controls. Abnormal activation of the cerebral areas especially frontal regions was detected corresponding to N1 and P3b components under different external emotions. Moreover, the concurrent affective states were correlated with ERPs in three groups.
In healthy controls, processing in Erotica and Happiness during N1 time window were mainly activated the bilateral supramarginal gyri and right inferior temporal gyrus, respectively. This might be supported by the connection of temporo-parietal lobe under positive emotion [53]. The right inferior occipital gyrus and bilateral medial frontal gyri were activated in Fear and Sadness during P3b time window, which was consistent with the previous studies showing that the neural activations of negative emotion were in right occipital gyrus, right amygdala, and bilateral frontal regions [54,55]. The correlation between mania and the elementary encoding of Sadness might be explained by the fact that the manic patients reported attenuated subjective sensation of facial sadness [56].
In BW, N1 latencies under Erotica and Happiness were prolonged. Since N1 reflects the primary processing of incoming information [44], the prolonged N1 latencies under Erotica and Happiness implied a delayed processing of positive emotions. Hypersexual individuals exhibited fewer positive emotions and displayed a paucity of happiness [57], which might influence the recognition of happiness stimuli. The differences of N1 latencies under Erotica between BW and BO might be due to the compulsive use of sexual materials in BW [31], thus leading to a decrease of emotional sensitivity to erotica. On the other hand, the hypersexual behaviors were associated with negative emotions [57], and the impaired recognition of unpleasant emotions in BD I [13,37] might result in the delayed perception of Erotica in BW in the current study. Through source analyses, we also found different parts of the brain activated under external erotica and happiness in BW. The literature shows that the postcentral gyrus plays a critical role in processing sensory information and regulation of emotion [58], and the supramarginal gyrus is a part of the ventral attention network [59] associated with overcoming emotional egocentric biases [60]. The inferior frontal gyrus is activated during attentional control and emotional perception [61,62]. In addition to encoding the primary perception of stimuli, the activations of these regions might imply that Erotica rather than Happiness induces the activation of somatosensory areas in BW. The finding that negative and distressing emotions triggered excessive sexual behaviors [63] helps explain the positive correlation between PVP and P3a amplitudes under Erotica in our BW patients.
In BO, P3b amplitudes under Fear were increased, suggesting the heightened voluntary attention to the negative emotions in these patients, which might be due to difficulties of ignoring the threating stimuli during attentional engagement [64]. The decreased P3b amplitudes in BW rather than in BO under Fear and Sadness might be explained by the following documentation. Hypersexuality is a typical symptom of mania episode [24], patients with mania displayed an impaired recognition of facial fear and sadness [56,65], and displayed the emotional dysfunction manifested in hypersexuality [26]. These abnormalities might lead to insufficient voluntary attention and evaluation to Fear and Sadness in BW. On the other hand, the neural regions activated during Fear were diverse in BW and BO during P3b time window. Indeed, bilateral lingual gyri associated with visual memory and stimuli perception [66] were activated in BW. Interestingly, bilateral medial frontal gyri which play a critical role in executive function, decision making, and cognitive control of emotion [67,68], and right inferior temporal gyrus, which correlates with object recognition [69], were mainly activated in BO. Moreover, the N1 amplitudes under Fear in right electrodes were negatively correlated with HCL-32 in BO. In clinics, individuals with hypomania show elevated mood, increased energy or activity, flights of ideas, and racing thought [9], which might facilitate the primary recognition of fearful or threating scenes in BO.
Similar to our present findings, Sagar et al. [10] showed that BD I had similar activated brain regions under fearful and happy stimuli. More specifically, we discovered that even during different time windows, activations of prefrontal cortex were altered in BW and BO under both Happiness and Sadness. Indeed, the activation of prefrontal cortex is critical in the emotional regulation [23]. In our BO, there were more activations of bilateral medial frontal gyri and right inferior temporal gyrus under both Erotica and Fear, which might imply that the primary perception under Erotica and the voluntary attention under Fear were more contrasted to distinguish BW and BO than those under Happiness and Sadness.
Considering the disruption of cognitive and cerebral processing under external emotions in BW and BO, the non-invasive brain stimulation might be considered as an effective approach to enhance neurocognition and regulate behaviors of the two BD I subtypes through applying to the specific brain areas, such as the prefrontal cortex which ties closely with cognition and emotional regulation [70]. Indeed, this stimulation method has been adopted in the management of psychiatric and neurological diseases [70,71], in improvement of emotion recognition [72], and in modulation of memory, particularly fear-related ones [73,74]. For the current challenges of precision diagnosis and management, this stimulation method within brain functional regions is potentially applicable in the individualized treatment of psychiatric problems including bipolar disorder.

Limitations and Future Directions
Nonetheless our study suffers from several design limitations. Firstly, our participants were all young adults, whether the results can be generalized to other age groups remains to be seen. Secondly, we only enrolled BW and BO patients, recruiting other groups such as major depression or personality disorder might add more confirmation to our current findings. Thirdly, we did not include other external emotions such as surprise, anger, or contempt, which might also display their emotional effects on the attentional process. Fourthly, we failed to measure personality traits in our participants, since BD I and hypersexuality might be related to either normal or disordered personality traits. Fifthly, we did not follow up to the eventual therapies of our patients, which might affect their sexuality in a far-reaching way. Nevertheless, we have demonstrated that BW had delayed primary perception under Erotica and Happiness and decreased voluntary attention under Fear and Sadness, and demonstrated differences between BW and BO on attentional function under Erotica and Fear. Thus, our study might provide hints of different emotional processes of and the clinical intervention for the two subtypes of bipolar I disorder.

Conclusions
We have found that BW and BO demonstrated different cerebral processing and activations under external emotions, especially Erotica and Fear. Our study thus indicates different patterns of visual attentional deficits and emotional processes and provides a basis for developing emotional intervention therapy and applying the non-invasive brain stimulation to brain areas including prefrontal cortex in the two bipolar I disorder subtypes.

Institutional Review Board Statement:
The study was conducted in accordance with the Declaration of Helsinki, and approved by the Ethics Committee of Zhejiang University School of Public Health (ZGL201606-1-6).
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. Data Availability Statement: Data are available from the corresponding author (W.W.) upon reasonable request.

Conflicts of Interest:
The authors declare no conflict of interest.