Deficits in Emotional Face Processing Indexed by N170 Modulation in Chronic and in First Hospitalized Schizophrenia

Abstract

Introduction:

Facial emotion recognition is impaired in schizophrenia and contributes to profound social impairments. Healthy adults exhibit larger N170 amplitudes to emotional compared to neutral faces. Preliminary evidence suggests an inability to modulate N170 amplitude by emotional expression during chronic stages of the illness. The present investigation examined N170 modulation by emotion among patients with chronic (ChSz) and first hospitalized (FHSz) schizophrenia.

Methods:

EEG was recorded from 26 FHSz and 28 ChSz participants as well as 19 young (YC) and 21 older (OC) matched controls. Participants were asked to detect neutral faces among happy, angry, disgusted, fearful, and sad faces. N170 amplitudes were measured from P9/P10 electrodes. Symptoms were assessed using the Positive and Negative Syndrome Scale (PANSS).

Results:

N170 amplitude modulation by facial emotion was observed across FHSz and YC (P < .001), though the typical right-hemisphere lateralization of this response observed in YC (P = .001) was absent in FHSz (P = .56). In contrast to OC (P = .009), ChSz did not exhibit N170 modulation by emotion (P = .32). Among ChSz, N170 modulation (mean N170 across emotional expressions minus N170 to neutral faces) at P9 were inversely correlated with PANSS negative scores (r = −.53).

Discussion:

Results suggests a progressive impairment of emotional facial expression processing as indexed by N170 modulation across illness stage. While losing the hemispheric specialization of face processing, FHSz exhibited preserved N170 amplitude modulation by facial emotion in contrast to ChSz. This deficit was also associated with negative symptoms, implicating progressive pathology of N170 generators in persistent and debilitating symptoms of the disorder.

Introduction

The ability to perceive the emotional content of faces is critical to successful communication, navigation of social situations, and perception of social threats. Individuals with schizophrenia have difficulties with social skills, and numerous studies have documented impairments in facial processing and emotion recognition in schizophrenia,^1–3 particularly with respect to negative or threatening emotions.⁴ Previous studies have indicated that patients with schizophrenia misattribute negative emotion to neutral faces more often and do not visually attend to negative stimuli to the same degree as healthy controls.^5–7 When examined longitudinally, deficits in emotional facial expression recognition are present during early-stage illness^8,9 yet are exacerbated by recurrent psychotic episodes^10,11 with deficits contributing to a progressive decline in social and occupational functioning.

According to models of face processing, recognition of emotional expressions can be broken into distinct component processes: initial sensory-perceptual visual processing, encoding of identifiable facial features, and the higher-order processes involved in assignment of an emotional state.¹² The P100 event-related potential (ERP), a visual evoked EEG component reflecting early perceptual processing, is diminished in schizophrenia.^13,14 Although this deficit appears to contribute to poorer face processing and recognition,^15–17 the P100 is elicited by various visual stimuli with no selectivity for faces. In contrast, the N170 reflects a relatively later stage of visual processing and is evoked by complex objects, exhibiting particular sensitivity to face stimuli.^18,19 Beyond a general impairment of the N170 elicited by complex objects,^13,20 the N170 enhancement to face stimuli is absent among individuals with chronic schizophrenia.^21,22 Structural neuroimaging has linked impaired N170 enhancement in schizophrenia to grey matter volume within its putative cortical generator, the fusiform face area (FFA).^23,24 Interestingly, impaired N170 enhancement may reflect longitudinal progression of illness debility with individuals following their first hospitalization for psychosis exhibiting intact N170 enhancement to faces compared to complex objects.²⁰

In addition to a sensitivity for faces, the N170 also exhibits modulation to facial emotional expressions in healthy adults.²⁵ Though effects of specific facial emotions have been variable in the literature, emotional expressions in general appear to elicit larger N170 amplitudes relative to neutral faces.²⁶ A previous preliminary analysis revealed impaired modulation of the N170 by emotional expression in chronic schizophrenia beyond previously documented N170 amplitude reductions,²⁷ providing a potential mechanism for emotional processing deficits in this population. Unfortunately, a similar examination has not been conducted in earlier disease stages to evaluate its progression and contribution to longitudinal functional impairments in the illness. Examinations of the stages of facial processing suggest the more downstream, frontally generated N250 component may also capture emotional face processing deficits in schizophrenia.²⁸ However, given effects of illness chronicity on emotion recognition, N170 amplitudes, and structural pathology within the FFA,²⁹ examination of N170 modulation by emotional expression across illness stage would provide valuable insights into longitudinal social functioning deficits in schizophrenia.

To address this gap in the literature, the present study compared responses of individuals with schizophrenia either within 1 year of their first hospitalization (FHSz) or during a more chronic illness stage (ChSz) against age-matched controls during an emotional facial recognition task. Both N170 elicited by 6 different facial expressions as well as recognition performance were compared between groups. Clinical data were also collected from patients to examine associations between facial emotion processing deficits and symptoms across disease stage. In light of previous reports of preserved N170 enhancement to faces in FHSz, we hypothesize that N170 modulation by emotional expression will be intact among FHSz with impairments evident in ChSz.

Method

Participants and Study Design

Participants included individuals with a schizophrenia spectrum illness recruited from inpatient and outpatient settings and healthy controls between 18 and 55 years, IQ greater than 85 and at least 9 years of education. Patients were recruited from McLean Hospital and controls were recruited from newspaper advertisements in the greater Boston area. Prior to study entrance, all participants underwent a screening session. Exclusion criteria were a documented neurodevelopmental disorder or a learning disability, a neurological impairment, or a history of electroconvulsive therapy, seizures, a history of concussion or traumatic brain injury with sequelae, or substance dependence within the past 5 years. Controls were excluded for a personal history of or any first-degree family members with any Axis 1 disorders. Following screening, clinical assessments for all participants included the Structured Clinical Interview for DSM-IV (SCID-IV)³⁰ conducted by experienced clinicians. The MATRICS Consensus Cognitive Battery (MCCB) was used to assess cognitive deficits in these groups. Patient symptom ratings were obtained using the Positive and Negative Symptom Scale (PANSS).³¹ All subjects performed the Mini-Mental State Examination³² to rule out any dementia or delirium, the information subscale of the Wechsler Adult Intelligence Scale–Revised (WAIS-R)³³ as a gross estimate of fund of information, handedness assessed by the Edinburgh Inventory,³⁴ socioeconomic status of patients and their parents were assessed by the Hollingshead two-factor index,³⁵ and the digits forward and backward subscales of the WAIS-R to test immediate/short-term memory, attention, and concentration. Testing session procedures are described below. The study was approved by the McLean Hospital IRB and informed consent was obtained after a complete description of the study was provided. All subjects received payment for participation.

Twenty-six FHSz were matched on age, sex-distribution, handedness, WAIS-R Information scaled scores, and parental socioeconomic status to 19 young controls (YC). FHSz were within one year of initial hospitalization for schizophrenia-spectrum psychosis. The first-episode sample consisted of 16 paranoid, 2 undifferentiated, 1 disorganized, 1 schizoaffective-depressed, 3 schizoaffective-bipolar subtypes, 1 delusional disorder, and 2 schizophreniform disordered individuals. One FHSz patient was not taking medication at the time of testing, and medication information was not available for 4 additional FHSz patients.

Twenty-eight ChSz (illness duration: 16.92 ± 8.4 years) were matched to 21 older controls (OC). A subset of ChSz and OC participants were included in a previous report²⁷ examining N170 modulation in ChSz. The chronic schizophrenia group consisted of 17 paranoid, 1 undifferentiated, 6 disorganized, 2 schizoaffective-depressed subtype, 1 schizoaffective-bipolar subtype, 1 psychosis not otherwise specified individuals. One ChSz was not taking medication at the time of testing, and medication information was not available for 3 additional ChSz.

Stimuli and Procedures

Participants viewed images of faces serially on a computer screen from a distance of 1.3 m. Stimulus presentation and button press responding were accomplished with SuperLab Pro 2 (Cedrus Corp.). Each trial began with a 200 ms fixation cross at the center of the screen followed by a 200 ms blank screen. Face stimuli were then presented in the center of the screen for 500 ms, subtending approximately 6° visual angle. Images were followed by an 800, 933, 1066, or 1200 ms blank screen, with variable durations balanced across expression, gender, and intensity.

Faces were taken from the Penn emotional face stimuli.³⁶ The stimuli comprised 16 examples each (8 male, 8 female) of neutral, sad, angry, happy, fearful, and disgusted facial expressions. Half of the stimuli from each emotional group were of mild intensity, and half were of extreme intensity. Each image was shown twice for a total of 192 face stimuli (32 trials of each facial emotion) with their order of presentation randomized within each block. Participants were instructed to button-press with their right index finger when they saw “a face with a neutral or blank expression.” All subjects passed a pretest in which they were asked to indicate which one of six simultaneously presented faces from a different stimulus set (Ekman faces) showed a neutral expression.

EEG Recording and Processing

EEG activity was recorded from 61 scalp sites and the nose tip using a pre-configured 64-channel cap (custom designed Electro-Cap International sintered Ag-AgCl caps using the 10–10 system). Activity was amplified and recorded continuously with embedded digital stimulus triggers using SynAmps and Scan Acquire (Neuroscan/Compumedics USA). During recording the right mastoid served as the reference. Two electrodes located medially to the right eye, one above and one below, were used to monitor vertical eye movements and blinks. Electrodes placed at the outer canthi of the eyes were used to monitor horizontal eye movements. The forehead served as ground. All electrode impedances were below 5 kOhms. The EEG amplifier bandpass was 0.10 (6 dB/octave roll-off) to 100 Hz (24 dB/octave rolloff). EEG was digitized at 500 Hz. Digital triggers were embedded using Superlab.

Data were processed using BrainVision Analyzer. Activity was digitally filtered off-line using a Butterworth zero-phase filter with a high cutoff of 20 Hz (48 db/oct). Ocular artifact correction was performed using the Gratton and Coles³⁷ method. Following interpolation of bad channels, recordings were re-referenced to the average reference as computed from all scalp electrodes. Epochs of 1100 ms (100 ms pre-stimulus baseline) were extracted based on stimulus presentation. Baseline correction was performed, and artifact rejection excluded segments with amplitudes beyond ±100 μV. Averages were calculated separately for each emotional expression. Only ERPs from trials in which participants correctly responded to neutral faces or withheld a response to emotional faces were included. Amplitude and latency of the N170 ERP were measured for each subject using automated peak detection between 150 and 300 ms post-stimulus at channels P9/P10 (sites with the largest N170 amplitudes) with subsequent manual confirmation, consistent with previous literature.^27,38,39 Channels P9/P10 were never marked as bad requiring interpolation in the current sample. Amplitude was measured over 10 msec window centered on each individual’s peak N170 for each facial expression.

Statistical Analyses

Analyses comparing FHSz versus YC and ChSz versus OC were conducted separately using SPSS. Basic demographic and task performance variables were compared between groups using t-tests or Chi-squared test where appropriate (Table 1). N170 event related potentials and their modulation by emotion were compared between groups using a 2 (group: controls/patients) by 6 (emotion: neutral/happy/sad/angry/fear/disgust) by 2 (hemisphere: LH (P9)/RH (P10)) repeated-measures analysis of variance (ANOVA). Simple effects and pairwise comparisons identified in the omnibus analyses were examined using repeated-measures ANOVA or t-tests. Pearson correlations between N170 amplitude and task performance and symptom measures were conducted within each patient group (FHSz and ChSz). Provided our primary interest in the illness-related impact of emotional modulation in face processing, correlations between N170 modulation averaged across emotional facial expression (mean N170 across emotional expression minus N170 to neutral faces) were also assessed among patient groups. All correlations were subjected to false discovery rate correction for multiple comparisons and considered significant at q < .1.⁴⁰

Table 1.

Group Demographic, Clinical, and Task Performance Data (Mean ± SD).

	FHSz (n = 26)	YC (n = 19)	P-value	ChSz (n = 28)	OC (n = 21)	P-value
Age	25.9 ± 8.8	25.3 ± 7.2	.79	38.3 ± 9.9	31.0 ± 11.0	.67
Range	19–45	18–50		22–55	20–55
Sex (F/M)	10/16	6/13	.63	7/21	7/14	.52
Handednessa	0.62 ± .46	0.78 ± .18	.15	0.74 ± .21	0.78 ± .15	.46
Race			.70			.19
White	19	16		25	16
Black	5	2		3	3
Asian	1	1		0	2
Nonspecified	1	0		0	0
WAISb	12.78 ± 2.4	13.47 ± 2.1	.33	12.22 ± 2.3	13.19 ± 2.8	.22
PSESc	1.65 ± 0.9	1.52 ± 0.9	.66	2.04 ± 1.1	1.67 ± 0.9	.23
SESd	3.13 ± 1.2	2.11 ± 0.6	.003	3.77 ± 1.2	2.50 ± 1.2	.001
Education	13.91 ± 1.8	15.06 ± 1.2	.03	14.35 ± 2.2	15.78 ± 1.5	.02
CDe	26.00 ± 5.1	27.44 ± 4.1	.33	23.29 ± 7.2	27.67 ± 4.7	.02
FAf	15.17 ± 12.6	12.22 ± 8.2	.39	20.18 ± 16.9	13.24 ± 7.8	.09
PANSS-Posg	17.40 ± 5.1	-		18.00 ± 5.9	-
PANSS-Negh	18.52 ± 8.1	-		16.19 ± 7.2	-
PANSS-Geni	33.62 ± 9.8			28.65 ± 7.9
CPZ-Eqj	349.2 ± 332			383.5 ± 240

^aEdinburgh Inventory Score (1 = Right handed; −1 = Left handed);

^bWechsler Adult Intelligence Scale;

^cparental socioeconomic status;

^dsocioeconomic status;

^ecorrect detections;

^ffalse alarms;

^gpositive subscale of the Positive and Negative Syndrome Scale;

^hnegative subscale of the Positive and Negative Syndrome Scale;

ⁱgeneral psychopathology subscale of the Positive and Negative Syndrome Scale;

^jChlorpromazine equivalent dose

Results

Chronic Schizophrenia

Task Performance.

ChSz (23.29 ± 7.2) had significantly fewer correct detections (CD) of neutral faces relative to OC (27.67 ± 4.7) (t₄₇ = −2.42, P = .02). There was a trend towards increased false alarms (FA) among ChSz (20.18 ± 17.0) relative to OC (13.24 ± 7.8) (t₄₇ = 1.74, P = .09), inappropriately classifying emotional faces as neutral. There was also a trend towards longer response times (RT) among ChSz (761.3 ± 166) relative to OC (685.0 ± 121) (t₄₄ = 1.71, P = .09).

N170.

ERP waveforms elicited by all facial expressions, scatterplot depicting individual N170 values, and their group means are presented in Figures 1 and 2 and Table 2, respectively. Overall, N170 amplitude did not differ between groups (P = .22). However, a difference in modulation of the N170 by emotion was observed between groups (F_5,235 = 3.65, P = .003). Within-group RM-ANOVA revealed an effect of emotion in OC (F_5,100 = 3.24, P = .009), but not ChSz (P = .32). Among OC, N170 amplitudes were smaller for neutral relative to angry (t₂₇ = 2.18, P = .04) and disgusted (t₂₇ = 2.27, P = .03) facial emotions. Main effects of emotion (P = .32), hemisphere (P = .10), their interaction (P = .56), and the three-way interaction between emotion, hemisphere, and group (P = .54) were all non-significant.

Figure 1.

Event-Related Potentials Elicited at Left (P9) and Right (P10) Hemisphere Electrode Sites to Faces of various Emotions among YC (A), FHSz (B), OC (C), and ChSz (D).

Figure 2.

Scatter-plot of the quantification of N170 amplitude elicited by facial emotions, collapsed across hemisphere, for each participant by group.

Table 2.

Mean ± SD N170 Amplitude Values for Each Group and Emotion Condition.

	FHSz (n = 26)	YC (n = 19)	ChSz (n = 28)	OC (n = 21)
Neutral	−4.28 ± 2.6	−5.88 ± 3.12	−6.15 ± 3.2	−6.78 ± 4.0
Angry	−4.90 ± 3.0	−6.49 ± 3.3*	−6.20 ± 3.1	−7.57 ± 4.1*
Disgust	−5.53 ± 2.6*	−6.87 ± 3.9*	−6.13 ± 2.7	−7.52 ± 4.1
Fear	−5.65 ± 2.8*	−6.27 ± 3.6	−6.57 ± 2.8	−7.21 ± 3.4
Happy	−5.57 ± 3.1*	−6.52 ± 3.4*	−6.41 ± 3.0	−7.13 ± 3.8
Sad	−4.86 ± 2.7	−6.30 ± 3.7	−5.85 ± 3.5	−8.21 ± 4.2*

^*Significantly different than neutral

There was no significant effect of group or interaction involving it on N170 latency (P’s = .1). A main effect of emotion was present (F_5,235 = 4.86, P < .001) with neutral faces eliciting shorter latencies compared to all emotional expressions (Angry: t₄₈ = 2.55, P = .01; Disgust: t₄₈ = 4.87, P < .001; Fear: t₄₈ = 2.78, P = .008; Happy: t₄₈ = 2.82, P = .007; Sad: t₄₈ = 2.14, P = .04).

Correlations Between N170, Performance, and Clinical Measures.

Correlations between variables with uncorrected P-values are displayed in Table 3. Among ChSz, both task CD (r = −.53, q = .08) and N170 modulation at P9 (r = −.53, q = .08) were inversely correlated with PANSS negative scores. N170 amplitude at P10 was also associated with MCCB scores (r = −.55, q = .07). Exploratory correlations with N170 amplitude modulation for each emotional expression are presented in Supplemental Table 2.

Table 3.

Uncorrected Correlations Between Task Performance, N170, and Clinical Measures.

	CD	FA	PANSS Positive	PANSS Negative	MCCBc
FHSz
CDa	-	-	r = .01	r = −.06	r = −.26
			P = .98	P = .81	P = .27
FAb	-	-	r = −.12	r = .34	r = .36
			P = .64	P = .14	P = .08
P9 Modulation	r = .20	r = .24	r = −.43	r= −.47	r = .13
	P = .34	P = .26	P = .06	P= .03	P = .57
P10 Modulation	r = .04	r = −.03	r = −.15	r = .05	r = −.83
	P = .86	P = .89	P = .52	P = .83	P = .72
P9 Amplitude	r= −.59	r = −.12	r = .04	r = −.20	r = −.10
	P= .003	P = .57	P = .89	P = .38	P = .65
P10 Amplitude	r= −.63	r = .03	r = .23	r = .24	r = −.08
	P= .001	P = .89	P = .34	P = .30	P = .75
ChSz
CD	-	-	r = −.19	r= −.53	r= .42
			P = .37	P= .006	P= .03
FA	-	-	r = −.05	r = −.12	r = .09
			P = .79	P = .55	P = .67
P9 Modulation	r = .22	r = .09	r = −.16	r= −.53	r= .44
	P = .26	P = .64	P = .42	P= .006	P= .02
P10 Modulation	r = .05	r = .19	r = .12	r = .12	r = −.01
	P = .78	P = .33	P = .56	P = .60	P = .96
P9 Amplitude	r = −.07	r = −.03	r = −.22	r = .18	r= −.41
	P = .71	P = .87	P = .27	P = .38	P= .04
P10 Amplitude	r = −.13	r = −.11	r = −.27	r = .27	r= −.55
	P = .52	P = .60	P = .19	P = .18	P= .003

^acorrect detections;

^bfalse alarms;

^cMATRICS Consensus Cognitive Battery

First-Episode Schizophrenia

Task Performance.

Groups did not differ in the number of CD (FHSz: 26.00 ± 5.1; YC: 27.44 ± 4.1; P = .33) or FA (FHSz: 15.17 ± 12.6; YC: 12.22 ± 8.2; P = .39) during the task. RT during CD also did not differ between groups (FHSz: 717.7 ± 142; YC: 678.5 ± 138; P = .39).

N170.

ERP waveforms elicited by all facial expressions, scatterplot depicting individual N170 values, and their group means are presented in Figures 1 and 2 and Table 2, respectively. Groups did not differ overall in N170 amplitude (P = .17). There was a significant effect of emotion across groups (F_5,215 = 5.53, P < .001). N170 amplitudes across groups were smaller for neutral relative to all other facial emotions (angry: t₄₄ = 2.91, P = .006; disgust: t₄₄ = 4.51, P < .001; fear: t₄₄ = 3.45, P = .001; happy: t₄₄ = 4.31, P < .001; sad: t₄₄ = 2.13, P = .04). N170 across groups was also larger for disgust (t₄₄ = 2.63, P = .01) and happy (t₄₄ = 2.28, P = .03) relative to sad faces. The effect of emotion did not differ between groups (P = .26). The LH also exhibited larger N170 amplitudes across groups and emotions relative to the RH (F_1,43 = 12.02, P = .001). This effect, however, differed between groups (F_1,43 = 7.53, P = .009) with FHSz exhibiting no hemispheric effect (P = .56) compared to YC that revealed larger LH amplitudes (t₁₈ = 3.77, P = .001). No interactions between emotion and hemisphere (P = .98) or three-way interaction (P = .23) were present.

No significant effects of group, emotion, or hemisphere or their interactions were observed for N170 latency (P’s = .1).

Correlations with Demographics and Clinical Measures.

Correlations between variables with uncorrected P-values are displayed in Table 3. Among FHSz, task CD was significantly correlated with N170 amplitude from both P9 (r = −.59, q = .07) and P10 (r = −.63, q = .07). Exploratory correlations with N170 amplitude modulation for each emotional expression are presented in Supplemental Table 1.

Discussion

Disruptions in the recognition and neural processing of facial emotions have been documented in schizophrenia and may contribute to many of the symptoms and functional impairments of the disorder. To assess the effects of illness chronicity on this deficit and its contribution to disease burden, the current investigation examined modulation of the N170 ERP by emotional facial expressions in chronic (ChSz) and first-hospitalized (FHSz) schizophrenia samples. Detection of facial emotion and modulation of N170 amplitude were both impaired in ChSz and associated with negative symptoms among these individuals. In contrast, despite diminished lateralization of the N170 response, task performance and N170 modulation by emotional expression remained intact in FHSz. This pattern of results suggests a progressive disruption of early perceptual face processing networks in schizophrenia and highlighting a potential target for preservation and source of resilience early in the disease course.

Consistent with previous reports on the effect of illness chronicity on emotional facial recognition,^10,11 ChSz exhibited impaired recognition and N170 modulation associated with severity of negative symptoms. These results, along with the preservation of these markers of emotional processing in FHSz, extend the finding of impaired N170 modulation by facial emotion in ChSz from a previous small-sample study.²⁷ The progressive pathology of the FFA in schizophrenia,²⁹ the proposed primary generator of the N170,^23,24,41 provides a potential mechanism for the impact of illness chronicity on its modulation by emotional expression. While N170 modulation was intact in the current FHSz cohort, previous reports have documented deficits in emotional facial expression recognition among individuals with first-episode schizophrenia.^8,9 Beyond potential differences in sample characteristics, this discrepancy in results suggests a more downstream source of emotional face processing deficits in early-stage disease involving a higher-order attribution of emotional state.

Results from the current investigation also support a relationship between emotional facial processing and disease symptomatology identified by previous work.^2,42 Among ChSz, both impaired emotion recognition task performance and reduced N170 modulation were associated with more severe negative symptoms, highlighting the role of impaired emotional processing in some of the most treatment resistant symptoms of the disorder. That the emotion recognition impairments in FHSz might result from later processing impairments might also explain the weaker, yet still significant, association between N170 modulation and negative symptoms observed in FHSz relative to ChSz.

Although N170 amplitude and its modulation by emotion were intact among FHSz, these individuals did not exhibit the typical right-hemisphere lateralized response observed among YC indicating disruptions of facial encoding even at this early illness stage. Previous reports have documented a loss of N170 lateralization with aging,^43,44 a finding supported by the lack of hemispheric effects in our older cohort. This pattern of results provides some support for an advanced aging hypothesis of schizophrenia,^45,46 though aberrant functional lateralization has been observed in schizophrenia across domains not typically susceptible to aging effects as well.^47,48 Within the aging literature, authors have argued that diminished lateralization reflects either a compensatory mechanism⁴⁹ or an inability to recruit specialized neural systems.⁵⁰ While inverse correlations between left-hemisphere N170 modulation and negative symptoms in our patient cohorts lends support to a potential neural compensation, this theory remains highly speculative at this time.

Correlations between N170 amplitude, its modulation by emotional expression, and neurocognition assessed by MCCB in the older patient cohort also suggest a sensitivity of emotional face processing deficits to age-related cognitive deterioration in schizophrenia. The N170 has previously been associated with cognitive performance and is sensitive to task-related cognitive load,^51,52 likely reflecting a shared reliance of emotional face decoding and various facets of cognition more broadly on basic perceptual processing. It is also important to note that social cognition is a component of the MCCB which may be driving this relationship as well.

Limitations

There are certain limitations of this work to consider when evaluating its results. While the PANSS is a well-validated instrument for assessing symptoms in schizophrenia, its reliance on evaluators’ behavioral observations undervalues the experiential aspects of patient’s negative symptoms relative to newer measures such as the brief negative symptom scale (BNSS) and clinical assessment interview for negative symptoms (CAINS).⁵³ In addition, our ability to assess medication effects on N170 modulation was limited by the high antipsychotic treatment adherence in our sample and lack of data available to determine lifetime antipsychotic exposure. Furthermore, the lack of certain detailed clinical information such as the stability of medication doses, number of trials, and number of hospitalizations in ChSz limited characterization of the sample. Finally, the cross-sectional nature of the data makes it difficult to completely isolate effects of illness chronicity given additional variables such as long-term medication side effects and medical comorbidities that might differentiate individuals in FHSz and ChSz groups.

Conclusion

Taken together, findings from the current study indicate an impaired tuning of emotional facial perceptual processing in schizophrenia that appears to develop over the course of the illness. While aberrant lateralization of the specialized facial processing network is present proximal to disease onset, diminished sensitivity of this system to emotional attributes of face stimuli do not manifest until later stages of the illness. Importantly, this deficit appears to be related to negative symptom severity, providing a potential target to treat, in the case of ChSz, and potentially prevent, in the case of FHSz, some of the most refractory symptoms contributing to social dysfunction in schizophrenia. Recent evidence also suggest that the N170 may provide a useful marker of response to treatments targeting prosocial behaviors in autism spectrum disorder,⁵⁴ an approach that may prove useful in schizophrenia as well.

Supplementary Material

Supplemental material for this article is available online.