Differences in Prefrontal Cortex Activation Between Predominantly Positive and Negative Symptom Profiles in Schizophrenia: A Functional Near-Infrared Spectroscopy Study

Introduction

Schizophrenia is a heterogeneous mental disorder characterized by distinct positive and negative symptom dimensions. Each symptom type manifests with unique clinical characteristics. Schizophrenia with predominant positive symptoms (PSZ) is mainly characterized by delusions and hallucinations, while schizophrenia with predominant negative symptoms (NSZ) is marked by emotional apathy and reduced volitional activity. This dimensional perspective echoes the historical typological model proposed by Crow [1], who described Type I schizophrenia, characterized by positive symptoms and a putative neurochemical pathology, and Type II schizophrenia, characterized by negative symptoms, cognitive deficits, and a putative neurostructural pathology. While modern research has moved beyond this strict binary characterization, the conceptual distinction between positive and negative symptom domains remains a cornerstone for understanding the illness’s heterogeneity. According to Solmi et al [2], the global burden of schizophrenia has been increasing steadily since 1990, with a rise in prevalence, incidence, and disability-adjusted life years. Understanding the neurobiological distinctions between PSZ and NSZ is crucial for improving diagnosis and treatment.

Functional near-infrared spectroscopy (fNIRS) provides a non-invasive, cost-effective method for assessing cortical hemodynamics in schizophrenia, offering good temporal resolution. Diao et al [3] demonstrated the potential of fNIRS in differentiating schizophrenia from major depressive disorder by analyzing hemodynamic responses in the prefrontal cortex (PFC). Similarly, machine learning approaches employed by Chen et al [4] and Ji et al [5], which achieved high classification accuracy using fNIRS features, support our proposal that channel-specific activation patterns could enhance diagnostic precision. The verbal fluency task (VFT) is widely used in fNIRS studies to assess executive function and language processing, both of which are often impaired in patients with schizophrenia. For instance, studies using fNIRS have reported reduced activation in the PFC of patients with schizophrenia, compared with that of healthy controls during the VFT [6,7]. Yeung and Lin [8] conducted a systematic review and meta-analysis, finding that patients with schizophrenia generally exhibit reduced oxygenated hemoglobin changes in the PFC during VFT, suggesting that this task is a sensitive tool for detecting neural abnormalities in schizophrenia. Eken et al [9] used dynamic functional connectivity to differentiate schizophrenia and bipolar disorder, achieving high classification accuracy. Furthermore, Wang et al [10] investigated the neural mechanisms of expressed emotion in schizophrenia using fNIRS, finding significant differences in cortical activation between patients and healthy controls. Together, these studies highlight the importance of using fNIRS to explore the neural correlates of schizophrenia symptom subtypes during cognitive tasks.

Building on this literature, in the present study, we explored differences in PFC activation between patients with PSZ and patients with NSZ during a VFT. By analyzing hemodynamic responses in specific prefrontal regions, we sought to identify unique neurobiological features that can distinguish between these subtypes.

Previous studies have suggested that patients with PSZ can exhibit hyperactivation in certain brain regions, while patients with NSZ can show hypoactivation or distinct cortical engagement. For instance, Ma et al [11] documented a pronounced left dorsolateral prefrontal hyper-response during working-memory paradigms, especially in patients with negative symptoms. Zhu et al [12] further demonstrated accentuated medial prefrontal recruitment across a spectrum of cognitive challenges. Complementing these fMRI findings, Barconi et al [13] used simultaneous fNIRS and electroencephalography (EEG) to reveal altered prefrontal hemodynamics during emotion-regulation tasks. Extending the comparative perspective, Xiang et al [14] uncovered divergent prefrontal engagement profiles between schizophrenia and major depressive disorder during visual-form memory and Tower-of-London testing. Chou et al [15] also emphasized the potential role of fNIRS imaging as a clinical biomarker for schizophrenia.

In our study, we hypothesized that PFC activation patterns during visual-form memory tasks would differ significantly between patients with PSZ and NSZ presentations of schizophrenia. These findings may offer new insights for future research and improve diagnostic precision in clinical practice.

Material and Methods

PATIENTS AND EXPERIMENTAL PROTOCOL:

A total of 36 patients with PSZ and 35 patients with NSZ from the Zigong Mental Health Center were included in this study. All participants were diagnosed according to the criteria of the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) [16], corresponding to ICD-10 codes F20.0–F20.9. The Chinese Classification of Mental Disorders, Third Edition was not used, as the DSM-5 is the international standard for research diagnostics; however, the criteria in both manuals are highly concordant for schizophrenia. The inclusion criteria were as follows: (1) age between 18 and 60 years (to minimize the confounding effects of age-related neurological and vascular pathology); (2) right-handed; (3) no history of other neurological disorders (eg, epilepsy, stroke, traumatic brain injury with loss of consciousness >30 min) or significant physical illness; and (4) ability to complete the experimental tasks. The exclusion criteria included a history of substance abuse (as confirmed by urine drug screening at admission for cannabis, cocaine, amphetamines, opioids, benzodiazepines, and barbiturates), significant head trauma, or other psychiatric comorbidities. All participants underwent a standard clinical workup, including medical history, physical and neurological examination, and routine blood tests to exclude hormonal, vitamin, infectious, and autoimmune causes. Brain magnetic resonance imaging (MRI) or EEG was performed based on clinical indication to rule out organic psychosis. The Positive and Negative Syndrome Scale (PANSS) [17] was administered, and a PANSS-derived Bipolar Index (PBI) was calculated as the difference between the positive and negative subscale totals (PBI=positive subscale total – negative subscale total; range:−42 to +42). Patients with PBI >0 were assigned to the PSZ subgroup (predominant positive symptoms, n=36), and those with PBI <0 to the NSZ subgroup (predominant negative symptoms, n=35), according to Kay [18]. The study was conducted from May 2024 to February 2025. All patients underwent a comprehensive clinical assessment, including the Brief Psychiatric Rating Scale (BPRS), to evaluate symptom severity. During the experiment, patients completed a Chinese version of the VFT [19,20] administered in Mandarin Chinese. The VFT included a 30-s pre-task baseline, a 60-s task period, and a subsequent 60-s post-task baseline. During the pre-task baseline, patients were instructed to count aloud from 1 to 30; during the post-task baseline, they were asked to count from 1 to 60. Throughout the task period, patients were required to generate as many 4-character Chinese phrases as possible using characters such as tian (“sky”), bai (“white”), shu (“tree”), and hua (“flower”). Automatic changes in stimulus sets occurred every 15 s. Before participation, all patients received a thorough explanation of the experimental procedures and provided written informed consent. The experimental protocol was approved by the Ethics Committee of the Zigong Mental Health Center (20240201). All procedures complied with the ethical standards of the Declaration of Helsinki [21]. Informed consent was obtained from all individual participants included in the study.

NIRS MEASUREMENT:

Hemodynamic changes in the PFC were measured using a 53-channel fNIRS system (BS-3000, Wuhan Znion Technology Co, Ltd, China). The system uses near-infrared light at wavelengths of 760 nm and 850 nm to detect changes in oxygenated hemoglobin and deoxyhemoglobin concentrations (16 sources and 16 detectors). The source-detector pairs were positioned 3 cm apart, and the probes were placed on the forehead of the scalp, according to the 10–20 system (source 9 is in the frontopolar midline position). The sampling rate was 20 Hz. The fNIRS system was calibrated before each measurement, and a pass rate of more than 90% was required before testing began. The channel distribution is shown in Figure 1.

DATA PRE-PROCESSING: The raw optical density data were first converted to relative changes in optical density using the Homer2 software toolkit [22]. tMotion was set to 0.3 s, tMask to 3 s, and the remaining parameters were left at their default settings for motion correction. Motion artifacts were identified and corrected using a moving standard deviation method and cubic spline interpolation. Physiological noise, including low-frequency drift and high-frequency fluctuations, was removed using a band-pass filter (0.01–0.1 Hz). Using the event mark as the starting point, a window of 2 s before and 60 s after each mark was extracted for overlay averaging. The 2-s pre-mark interval was used for baseline correction in each block. The optical density data were then converted to oxygenated hemoglobin and deoxyhemoglobin concentration changes using the modified Beer-Lambert Law [23,24].

GENERAL LINEAR MODEL ANALYSIS:

The general linear model (GLM) [7,25] approach was used to analyze task-evoked activation using a MATLAB-based toolbox, the Near-Infrared Spectroscopy-Knowledge Inference Toolbox [26]. A boxcar function corresponding to the VFT onsets was convolved with a canonical hemodynamic response function to generate the task regressor. The GLM model included a constant term and the task regressor. Task β values reflecting task-evoked activation were derived and used for group-level statistical analyses [26,27]. Because oxyhemoglobin has a typically superior signal-to-noise ratio and a stronger correlation with cerebral blood flow than does deoxyhemoglobin, oxyhemoglobin was selected as the primary measure for our subsequent analyses [28,29].

REGIONS-OF-INTEREST-LEVEL FEATURE EXTRACTION:

The integrated value of oxygenated hemoglobin concentration changes over the 60-s task period was computed for each channel. The average value of all channels within a given region of interest (ROI) was calculated as the ROI metric, and any coefficient of variation value exceeding 25% was considered a “bad channel” and would be omitted from the ROI calculation. Six ROIs were identified based on the Broadman probabilistic atlas: left and right Broca’s area, left and right dorsolateral prefrontal cortex, and left and right frontal pole area. The oxygenated hemoglobin concentration changes in these ROIs were analyzed to assess task-related activation.

STATISTICAL ANALYSIS:

Statistical analyses were performed using IBM SPSS Statistics (Version 26). Group differences in demographic variables and BPRS scores were assessed using independent-sample t tests, while sex differences were analyzed using chi-square tests. One-sample t tests were conducted to assess group-level task-evoked activation within each group. Group differences in activation were analyzed using independent-sample t tests. For ROI-level analyses, 2-sample t tests were performed for each ROI. Pearson correlation coefficients were calculated to explore the relationships between oxygenated hemoglobin changes and clinical scores. The significance level was set at P<0.05, and multiple comparisons were corrected using the Benjamini-Hochberg procedure to control the false discovery rate (FDR).

Results

DEMOGRAPHIC AND BPRS SCORES:

The baseline demographic and BPRS scores of patients with PSZ and NSZ are summarized in Table 1. The 2 groups did not differ significantly in sex distribution (P=0.400), age (P=0.230), marital status (P=0.320), education (P=0.420), body mass index (P=0.480), duration of untreated psychosis (P=0.080), or chlorpromazine equivalent dose (P=0.630). The BPRS total score was significantly higher in the PSZ group than in the NSZ group (P=0.002). On the BPRS subscales, the NSZ group scored higher on the Lack of Vitality dimension (mean [SD] 13.884, [2.957] vs 8.628 [1.68184]; P=0.000), while the PSZ group scored higher on the dimensions Activation (11.800 [3.02733] vs 6.961 [2.457]; P=0.000) and Hostility/Suspicion (13.057 [2.50814] vs 8.153 [2.257]; P=0.000). No significant differences were found between the groups on the dimensions Anxiety/Depression (P=0.920) and Thought Disturbance (P=0.140).

HEMODYNAMIC RESPONSES:

Figure 2 illustrates the concentration changes of oxygenated hemoglobin and deoxyhemoglobin in the different brain regions of patients with PSZ and NSZ: left and right inferior frontal gyrus (IFG-L, IFG-R), middle frontal gyrus (MFG-L, MFG-R), and superior frontal gyrus (SFG-L, SFG-R). Table 2 presents the activation of ROI regions in schizophrenia patients with different positive/negative symptom dimensions. These data revealed differences in the mean and standard deviation of oxygenated hemoglobin activation between the PSZ and NSZ groups in the brain regions IFG-L, MFG-L, SFG-L, SFG-R, MFG-R, and IFG-R. However, statistical analysis showed that these differences were not statistically significant (P>0.05).

ACTIVATION ANALYSIS OF PATIENTS WITH PSZ:

Figure 3 presents the activation analysis of patients with PSZ on the VFT. The results highlight significant activation of specific channels within the PFC, including Ch8 (triangular portion of Broca’s area), Ch35 (frontal pole), and Ch44 (triangular portion of Broca’s area and dorsolateral prefrontal cortex). The channels passed the correction for FDR. The 2-dimensional (2D) and 3-dimensional (3D) visualizations showed significant activation in these areas, which highlighted distinct neural activation patterns in the PFC of patients with PSZ during the VFT, particularly in Broca’s area and the dorsolateral PFC. Activation analysis in patients with NSZ did not pass FDR correction.

ACTIVATION DIFFERENCE ANALYSIS:

Figure 4 and Table 3 present the oxygenated hemoglobin activation differences between the PSZ and NSZ groups during the VFT. Figure 4 visually highlights the activation patterns in channels Ch21 and Ch51, which are associated with the frontal pole, orbitofrontal cortex, and triangular part of Broca’s area. These channels showed significant activation differences between the 2 groups after FDR correction. Table 3 provides the specific activation values, showing that the PSZ group had lower mean activation in Ch21 (−0.1337 vs 0.5068, P=0.047) and slightly lower mean activation in Ch51 (−0.0542 vs 0.0643, P=0.038), compared with the NSZ group.

CORRELATION ANALYSIS IN CHANNEL CH51:

Figure 5 shows a scatter plot of the correlation between β values and BPRS activators in channel Ch51. The results showed that the β value was negatively correlated with activation (r=−0.292, P=0.022), indicating that the β value of Ch51 channels tended to decrease with the increase of the BPRS activator. The correlation between activation of Ch51 and Ch21 and BPRS factor scores is shown in Table 4.

Discussion

In this study, fNIRS was used to investigate the differences in PFC activation between PSZ and NSZ in a cohort rigorously screened to exclude secondary organic psychoses. The main findings can be summarized as follows. There were no significant differences in hemodynamic responses in each region of the ROI between the PSZ and NSZ groups. However, significant activation differences were found in channels Ch21 and Ch51. The PSZ group exhibited lower mean activation in Ch21 and slightly lower mean activation in Ch51, compared with the NSZ group. Additionally, β values of Ch51 channels showed a significant negative correlation with activators of BPRS. Furthermore, groups were well-matched on key clinical variables, such as duration of untreated psychosis and chlorpromazine equivalent dose, suggesting that the observed neural differences were more likely attributable to symptom domain pathology rather than these potential confounders. These results suggest that specific channels can serve as potential biomarkers for distinguishing between PSZ and NSZ [15,30].

Our research results are consistent with those of previous studies in several aspects. Similar to Di et al [3], we observed the activation patterns of the PFC in different dimensions of schizophrenia symptoms, although these activation patterns occurred in specific regions. While Di et al focused on severe depression, our study extended these findings to schizophrenia with different symptom dimensions. Chen et al [4] also used fNIRS to classify patients with schizophrenia, using machine learning techniques to achieve high accuracy. Our study complements this by identifying specific channels, Ch21 and Ch51, with significant activation differences between PSZ and NSZ groups, offering new insights for future research and clinical applications. However, our results differ from those of some previous studies in the lack of broad regional differences in PFC activation. This discrepancy may stem from variations in sample characteristics, task designs, or data analysis methods. For instance, the study conducted by Yang and Lin [8] revealed that during the virtual graphics task test, the changes in oxygenated hemoglobin in the PFC of patients with schizophrenia were reduced. However, their research did not make specific comparisons among different symptom dimensions. Eken et al [9] used dynamic functional connectivity to differentiate schizophrenia and bipolar disorder, achieving high classification accuracy, but their focus was on different diagnostic categories. Our study expands upon the existing literature, highlighting the potential of specific fNIRS channels as biomarkers for differentiating PSZ and NSZ, which is of great significance for the development of more targeted therapeutic interventions.

In our study, we used a variety of methods to investigate the activation of the PFC in schizophrenia across different symptom dimensions. Notably, we used the VFT to induce PFC activation and analyzed hemodynamic changes using GLM analysis and ROI-level feature extraction. This combination enabled us to precisely identify the specific channels that showed significant activation differences between patients with PSZ and those with NSZ. Previous studies have used similar fNIRS tasks and analysis methods to explore schizophrenia, but few studies have specifically compared different symptom dimensions. For example, Chou et al [15] highlighted the potential role of fNIRS as a clinical biomarker in schizophrenia but did not delve into subtype distinctions. Our study builds on this foundation by providing evidence for distinct activation patterns in specific channels. Further correlation analysis showed that β values of Ch51 channels were significantly correlated with BPRS activators, a finding that echoed the study by Wei et al [31]. The significance of specific neurobiological markers in understanding the different symptom dimensions of schizophrenia has been emphasized. The use of fNIRS in our study offers a non-invasive and cost-effective method for probing these markers, which could be beneficial for future research and clinical applications. Other studies have also suggested the potential of fNIRS in revealing neurobiological differences between schizophrenia and other psychiatric disorders. For instance, Diao et al [3] successfully used fNIRS to differentiate schizophrenia from major depressive disorder. Our research further enriches this field of work, indicating that functional fNIRS can also distinguish different symptom dimensions of schizophrenia. The identification of specific channels with significant activation differences may pave the way for more targeted interventions and personalized treatment strategies in schizophrenia.

The clinical implications of our findings align with growing interest in fNIRS-guided interventions. The channel-specific activation patterns we identified could inform targeted neuromodulation approaches, such as the theta burst stimulation protocols described by Gao et al [32] and Li et al [33], which demonstrated prefrontal activation normalization accompanying symptom improvement. Furthermore, our results support the potential integration of fNIRS biomarkers with digital phenotyping frameworks, as proposed in recent multimodal studies [34].

This study has several limitations. First, all patients were recruited from a single site, which can introduce selection bias; multi-center studies are needed to validate these findings more broadly. Second, the absence of a healthy control group limits our ability to compare activation patterns in schizophrenia subtypes with typical brain activity [35,36]; future investigations should include controls to clarify schizophrenia-specific neurobiology. Third, we employed only the VFT; incorporating a broader battery of cognitive paradigms would provide a more comprehensive assessment of PFC function [14]. Fourth, and perhaps most importantly, our cross-sectional design captured symptom profiles at a single time-point. Symptom domains in schizophrenia are dynamic states, not fixed traits: an individual classified as PSZ at one assessment can shift toward a NSZ profile later [37]. Consequently, our PBI-based group assignments offer a valid dimensional snapshot but cannot account for intra-individual symptom evolution. Longitudinal studies that repeatedly assess clinical presentation alongside fNIRS-derived activation are required to uncover truly dimension-specific neurobiological markers. Finally, integrating fNIRS with complementary neuroimaging modalities such as fMRI could yield a more complete understanding of the mechanisms underlying schizophrenia [38].

Conclusions

This study highlights distinct PFC activation patterns between patients with PSZ and those with NSZ during the VFT. Specific channels – Ch21 and Ch51 – showed statistically significant activation differences between the 2 groups. These findings, derived from a cohort rigorously screened for organic psychosis, indicate that fNIRS can effectively detect neurobiological differences between symptom dimensions within schizophrenia.

However, as a cross-sectional snapshot, this study cannot address the dynamic longitudinal interplay between positive and negative symptoms. Future studies should prioritize longitudinal designs to track neurofunctional changes alongside clinical evolution. Larger sample sizes and multi-center recruitment will improve the generalizability of the findings. Introducing a healthy control group and integrating fNIRS with other neuroimaging techniques, such as fMRI or EEG, could provide a more comprehensive understanding of the neurobiological mechanisms underlying different symptom profiles.

Ultimately, this integrated approach can facilitate the development of more targeted therapeutic interventions and personalized treatment strategies for patients with schizophrenia.