Perturbed sensory memory associated with schizotypy symptom load

Abstract

Impaired auditory event-related potential (ERP) components, including the N100 and MMN, are linked to sensory signatures of schizophrenia. Nonclinical individuals with schizophrenia-like traits (schizotypy) provide an attractive model for investigating these ERP biomarkers, as the individuals exhibit similar traits without clinical confounds (e.g., medications). The driving theoretical framework is that there is a continuum of traits that extends from the nonclinical population to patients diagnosed with schizophrenia spectrum disorders (SSD). Here, we tested the prediction that nonclinical participants high in schizotypy traits would also exhibit impaired N100 and MMN responses, reflecting impairment in early auditory processing, and/or in predicting sensory inputs. We used a simple odd-ball pitch-deviant paradigm and compared ERP amplitudes across a sample of healthy participants who completed the Schizophrenia Personality Questionnaire - Brief Revised (SPQ-BR). Participants with high total scores (high schizotypy; N=25) were compared to middle-scoring scorers (controls; N=32). We then investigated more nuanced relationships between ERP components and factors of the SPQ-BR (cognitive-perceptual, interpersonal, disorganized). High schizotypy exhibited prolonged MMN latencies but intact amplitudes—a pattern also observed in those with elevated negative symptoms. However, attenuated deviant N100 amplitudes were associated with higher disorganized factor scores, whereas standard and MMN amplitudes were not. Importantly, these data suggest that early deviance detection and predictive sensory memory mechanisms are differentially perturbed with schizotypy symptom load. Further research into schizotypy traits is required to validate the population for SSD risk. In conclusion, auditory impairments can be identified in nonclinical schizotypy and leveraged to understand disorder-related processes.

Introduction

Schizophrenia spectrum disorders (SSD) are characterized by the expression of delusions, hallucinations, disorganized speech/thoughts/behaviors, or negative symptoms (American Psychiatric Association, 2013). However, SSD traits are present non-clinically along a spectrum of severity. High schizotypy refers to non-clinical individuals who exhibit a high degree of SSD-like traits. Schizotypy trait load is valuable for SSD research, as it allows for the examination of SSD without clinical confounds (Ettinger et al., 2015; Haigh et al., 2023; Kelemen et al., 2013). Importantly, prospective SSD biomarkers can be measured as a function of increased schizotypy trait load, separately from disease process. High schizotypy may also capture a portion of individuals vulnerable to a future SSD diagnosis (Cohen et al., 2010; Raine, 2006). Thus, schizotypy may offer insights into the mechanisms underlying SSD during preclinical stages.

Schizotypy traits can be measured by SSD-like symptom-type (positive, negative, and disorganized; Ettinger et al., 2015). This allows investigation into the relationship between symptom type load and sensory mechanisms in SSD. We previously detected anomalous visual processing in participants with high scores on the disorganized subfactor, highlighting the value of investigating disorganized schizotypy traits to gain purchase on SSD (Torrens et al., 2023, 2024). This is clinically relevant because high disorganized scores predict worse clinical outcomes in those with SSD (Metsänen et al., 2006; Nestsiarovich et al., 2017; Reed et al., 2002; Shenton et al., 1992). Therefore, we assessed the effects of different schizotypy traits on two auditory biomarkers that have been well-characterized in the SSD population.

Proposed Auditory Biomarkers of SSD: The N100 and MMN

Two potential biomarkers of SSD are the auditory N100 and mismatch negativity (MMN). These event-related potentials (ERPs) are valuable because they are reliably and robustly impaired in schizophrenia (N100: Rosburg et al., 2008; MMN: Erickson et al., 2016; Umbricht & Krljes, 2005). Second, they arise from different brain regions (Korzyukov et al., 1999) and occur at different latencies, suggesting they represent different mechanisms. This allows us to identify if, when, and how neural activity becomes atypical in high schizotypy.

The auditory N100 peaks at ~100 ms after an auditory stimulus and is considered an index of simple, early auditory detection. With the repetition of frequent standard stimuli, the N100 amplitude diminishes, indicating sensory adaptation (Fruhstorfer et al., 1970; Ritter et al., 1968). In contrast, the N100 amplitude generated in response to infrequent deviant stimuli is typically larger, signaling the release from adaptation (May & Tiitinen, 2010). The N100 is a proposed marker of genetic risk for schizophrenia because it is reliably smaller than in neurotypical participants (Duncan et al., 2022). The N100 is also delayed (i.e., larger/slower latency) in schizophrenia (Boutros, 2004; Pultsina et al., 2022; Wang et al., 2024a; Williams et al., 2000). A delayed N100 could reflect a genetic link to schizophrenia as it has been observed in affected and unaffected schizophrenia relatives (Simons et al., 2011; Sumich et al., 2008a; Wang et al., 2024a), and the extent of delay worsens with duration of untreated psychosis in antipsychotic-naïve first-episode patients (Lu et al., 2025), pointing to a link to symptom severity.

The auditory MMN is a pre-attentive negative waveform that occurs later in processing than the N100, peaking at approximately 150–200ms after deviant auditory stimulation (Näätänen et al., 1978, 1978; Neustadter et al., 2016). The MMN is derived by subtracting the waveform elicited by deviant stimuli from the waveform elicited by standard stimuli, typically resulting in a negative potential (Näätänen, 1990). The MMN reflects the detection of a change in the auditory environment (Näätänen et al., 1978) and is considered an index of sensory memory; a violation of the expected stimulus that is stored in memory that then elicits a MMN (Ritter et al., 1995). The MMN is sometimes also referred to in computational terms as predictive coding. The MMN can detect statistical regularities and novelty in the environment to improve predictions (Garrido et al., 2008; Winkler, 2007; Winkler et al., 2009). However, there is debate in the literature as to whether the MMN serves as a predictive biomarker of schizophrenia or whether it is due to disease progress. Some studies report reduced MMN amplitudes in clinical high-risk individuals (Hamilton et al., 2022; Higuchi et al., 2013; Perez et al., 2014; Shaikh et al., 2012,), but the reliability of this finding has been questioned in different populations along the disease course (Brockhaus-Dumke et al., 2005; Erickson et al., 2016; Haigh et al., 2017; Salisbury et al., 2002).

MMN latency reports are less consistent. In schizophrenia, there is evidence for both shorter (i.e., faster) (frequency: Horton et al., 2011; Kärgel et al., 2014; duration: Domján et al., 2012; Kärgel et al., 2014) and larger latencies (frequency: Kathmann et al., 1995; Li et al., 2013; duration: Fisher et al., 2012; Nakajima et al., 2021). MMN latencies are also longer in non-remitters compared to remitters (Nakajima et al., 2021), suggesting a relationship between MMN latency and increased symptomology. Contrastingly, typical latencies have also been documented in clinical high-risk individuals who do and do not convert into psychosis (Duncan et al., 2022).

The N100 and MMN in Schizotypy

In non-clinical high schizotypy, there is some evidence for an atypical N100, but the effects depend on the symptom type. For example, higher scores on positive schizotypy traits (paranormal ideation and unusual experiences) were linked to reduced N100 amplitudes to target tones (Sumich et al., 2008a), whereas high anxiety-related schizotypy traits were linked with larger N100 amplitudes (Sumich et al., 2008b). However, there is also evidence for typical N100 amplitudes in high schizotypy (Klein et al., 1999). With respect to latencies, to our knowledge, the direct link between N100 latencies in schizotypy remains unexplored.

The auditory MMN in non-clinical schizotypy also has mixed findings that may be exacerbated due to differences in paradigm. Some studies report that MMN amplitudes to frequency, but not duration, deviants are smaller, but latencies in response to location deviants are longer, with higher schizotypy scores (Bissonnette et al., 2024). Others report larger MMN amplitudes in high schizotypy in response to duration deviants (Deng et al., 2024). Others report no relationship between MMN amplitudes to duration, frequency, or intensity deviants and high schizotypy (Bose et al., 2024; Broyd et al., 2016). Therefore, MMN and total schizotypy scores appear to correlate less reliably.

Focusing on the symptom-trait load approach has yielded some success. There is evidence for negative relationships between auditory MMN amplitude and positive (Bissonnette et al., 2024; Deng et al., 2024; Donaldson et al., 2021) and negative (Bissonnette et al., 2024; Donaldson et al., 2021) schizotypy traits. More specifically, higher scores on the cognitive-perceptual (CP) factor and the CP factor sub-factor ideas of reference have been linked with smaller MMN amplitudes to frequency (Bissonnette et al., 2024) and duration deviants (Deng et al., 2024). Whereas, higher scores on the CP subfactor suspiciousness scores have been weakly linked to larger MMN amplitudes to frequency deviants (Broyd et al., 2016).

In concert, the pattern of findings in the N100 and MMN point towards early sensory disruptions parallel to those in SSD, but in some cases, are contingent on symptom type. However, inconsistency across findings, and breadth of populations sampled hampers progress in understanding the underlying neural correlates of schizotypy traits.

Current study

We tested the prediction that nonclinical high schizotypy would exhibit altered N100 and/or MMN responses, reflecting impairment in early auditory processing, or in predicting sensory inputs, respectively. We contrasted the N100 and MMN amplitudes and latencies in a high schizotypy group with those of a middle-scoring schizotypy (controls) group collected during a classic odd-ball pitch-deviant paradigm. Pitch-deviant paradigms have yielded reliable effects in SSD (Hamilton et al., 2022; Perez et al., 2014; Umbricht & Krljes, 2005). Following our previous work (Torrens et al., 2023, 2024), we investigated relationships between the ERPs and individual factor scores. We predicted that the disorganized symptoms would elicit reduced amplitudes in the N100 and/or MMN, consistent with the SSD literature. Following a reviewer’s recommendation, latencies between groups and factors were also investigated. If the N100 is reduced or delayed, it would reflect impaired early auditory detection, whereas a reduced or delayed MMN would reflect sensory memory deficits, and if there were reductions in both, it would suggest early and later auditory deficits.

Methods

Participants

Following approval from the Institutional Review Board of the University of Nevada, Reno, we recruited undergraduate students. A total of 2,199 students completed an online screening survey through Qualtrics, including the Schizotypal Personality Questionnaire - Brief Revised (SPQ-BR; Cohen et al., 2010). Of the students who completed the screener, n=1,692 provided their email addresses indicating a willingness to be considered for the in-person EEG study. Inclusionary criteria included: no self-reported hearing loss or hearing aid; no vision problems that cannot be corrected with corrective lenses; no history of epilepsy; and no history of neurological disease or psychiatric problem. We contacted students who met inclusionary criteria and criteria based on their SPQ-BR scores (described below) for high schizotypy (n = 178) or controls (n=205).

Electroencephalography (EEG) data were recorded from 68 participants. Two participants were excluded due to EEG recording issues, and 9 were excluded for having too few epochs after artifact rejection (see data analysis section). ERP analyses were performed on 57 individuals (25 high schizotypy and 32 controls). Participants received credit through the Psychology Department’s online subject pool and were entered into a raffle for the opportunity to receive a $25 gift card.

Schizotypal Personality Questionnaire Brief-Revised (SPQ-BR)

The SPQ-BR (Cohen et al., 2010) is a 32-item self-report questionnaire that measures the number of SSD-like traits. Responses are on a five-point Likert scale (0 “strongly agree” - 4 “strongly disagree”). Example questions include, “I often hear a voice speaking my thoughts aloud.” The SPQ-BR scores range from zero to 128, with higher scores depicting more SSD-like traits (high schizotypy). There are three factor scores (Interpersonal [IP], Cognitive-Perceptual [CP], and Disorganized [DIS]).

Participants’ SPQ-BR scores from the online screener were not statistically normally distributed (W = .99, p<.001) but predominantly clustered around the center of the SPQ-BR score distribution (M = 60.44, SD = 18.21, N = 2,188; Figure 1). Participants at the center of the distribution served as our control group (score 50–70) because they were of higher incidence than lower-scoring individuals, suggesting greater typicality. Participants with scores 2 standard deviations above the mean total SPQ-BR score were included in the high schizotypy group (SPQ-BR>87). All schizotypy factors were normally distributed: CP (W = .97, p=.21), IP, (W = .98 p=.63), and DIS scores (W = .96, p=.96; see Figure 2).

Figure 1.

Distribution of the total SPQ-BR scores (N=2,188) of all adults who completed our screener survey. Scores >87 (black dashed line) are considered to be high schizotypy and scores 50 to 70 (grey dashed line) are considered controls.

Figure 2.

Distribution of schizotypy factor scores: Cognitive Perceptual (CP), Interpersonal (IP) and Disorganized (DIS).

Raven’s Standard Progressive Matrices (RSPM)

The Raven’s Standard Progressive Matrices (RSPM) Sets A-E (Raven, 1998) is a measure of fluid intelligence that includes 60 puzzles divided into sets of 12 (Raven et al., 2003; Raven, 1998). Possible scores range from zero (none correct) to 60 (all correct); see Table 1.

Table 1.

Demographic information for high schizotypy and control participants. Chi-square and t-tests were conducted to compare the high schizotypy and control groups on their demographic information.

	High Schizotypy	Controls	p-value
Gender F/M	18/7	26/6	.62
Age range (M/SD)	18 – 30 (20.24/2.68)	18 – 43 (19.97/4.50)	.77
Handedness (R/L)	21/4	30/2	.45
GPA (M/SD)	3.48/.54	3.56/.39	.55
RSPM Score range (M/SD)	41 – 58 (51.16/4.40)	37 – 60 (50.16/5.704)	.46

Note: Total number for gender (female/male). Age range in years (mean/standard deviation). Handedness totals listed for (right/left). Means (standard deviations) are listed for the remaining demographics. Raven’s Standard Progressive Matrices (RSPM) Scores and testing time listed in minutes. T-tests were used to compare most groups, except for gender and handedness for which we used chi-square tests.

Stimuli

Auditory stimuli standard pitch tones (1046.5Hz, 50ms duration, 5ms rise/fall, 80%) and pitch deviant tones (1108.73Hz, 50ms duration, 5ms rise/fall, 20%), with a pseudo-randomized interstimulus interval (ISIs) of 450 – 950ms. There were 600 tones divided evenly into 6 blocks, with self-paced breaks. Tones were presented via Etymotic insert earphones while EEG was recorded. Tones were generated and presented using MATLAB (Mathworks) and Psychtoolbox (Brainard & Vision, 1997; Kleiner et al., 2007; Pelli, 1997).

Data acquisition

EEG data were recorded using a BioSemi ActiveTwo system (Netherlands, Amsterdam). Electrodes were secured using a 32-channel nylon head cap following the international standardized 10–20 layout. A conductive gel (Signa, Parker) reduced impedance. EEG data were recorded relative to the BioSemi Driven Right Leg (DRL) and the Common Mode Sense (CMS) electrodes. Six flat-type active BioSemi electrodes on the head and body captured electrooculogram (EOG) and electrocardiogram (ECG) data. Two electrodes functioned as offline references, one positioned on each mastoid. To record EOG, a single electrode was placed on each outer canthi and another beneath the right eye. To record ECG, one electrode was positioned on the left collarbone. The data were digitized at 512 Hz.

Procedure

Participants completed an online survey through Qualtrics (10 minutes), containing the consent form, demographic questions, the SPQ-BR, and request for email address for those willing to participate in the EEG portion of the study. Participants who met the inclusion criteria with scores fitting the categories of high schizotypy or control groups were contacted via email and scheduled for EEG data collection.

Upon arrival at the testing room, participants signed informed consent documents. Participants then completed RSPM and provided their grade point average (GPA). Participants were then seated at a monitor (Display++ LCD Monitor from Cambridge Research Systems, using a Dell 3.5GHz Intel Xeon v3), and equipped with Etymotic insert earphones while EEG data were recorded. To ensure attention, participants were asked to fixate on a black central cross centered on a grey screen and instructed to press the spacebar when the cross flashed from black to white (5% of trials); see Figure 3.

Figure 3.

MMN task: participants were presented with a stream of pitch standard (black) and pitch deviant (grey) tones with a pseudo-random ISI (450–950ms), as they faced a monitor that displayed a gray screen with a and black fixation cross, that would randomly flash to white prompting the participant to hit the spacebar.

Data analysis

EEG data were analyzed using EEGLAB (Delorme & Makeig, 2004) and ERPLAB (Lopez-Calderon & Luck, 2014) toolboxes in MATLAB (Mathworks, R2023a). Data were preprocessed as follows: Electrodes were re-referenced to the average of the mastoids and signals were bandpass filtered between 0.1–50Hz. Data were visually inspected for EEG errors. An independent component analysis (ICA) was performed to visually identify ocular (blinks and horizontal eye movements) and ECG components to remove these sources of noise. Epochs were extracted from 100ms before stimulus onset to 330ms after stimulus onset. Epochs that contained artifacts exceeding ±100μV were automatically rejected. Participants with <75% of the epochs were excluded from further analyses (n=4 high schizotypy and n=5 controls). A t-test found no significant differences between the percentage of epochs kept between the remaining high schizotypy and control participants (t(44.96) = −1.32, p=.19).

Averaged ERPs were computed for each participant and latencies were selected using established windows for each component (N100: Annanmaki et al., 2017; Lijffijt et al., 2009; Rosburg et al., 2008; MMN: Fruhstorfer et al., 1970; Korpilahti et al., 2001; Näätänen et al., 2004, 2007; Rentzsch et al., 2015). To assess the effects of schizotypy on the N100 amplitude, mean peak amplitudes to standard and deviant tones were extracted for each participant between 80–170ms. For MMN, mean peak amplitudes were extracted between 150–250ms. In accordance with our previously published analyses (e.g., Haigh et al., 2022), data were also analyzed using 20ms latencies determined by grand-averaged peak window (these data can be found in OSF). Notably, the findings remained the same. Following a reviewer’s recommendation, peak latency values were also extracted to assess latency differences across groups. For each participant, peaks were extracted using a custom algorithm that detected the most negative (minimum y-value) value within time windows for each component separately (code available on OSF).

The N100 and MMN are maximal over frontocentral electrode sites (Näätänen et al., 2007; Rosburg et al., 2008); see supplemental materials for heatmaps. The amplitudes for all components were maximal at both Fz and Cz and signals were highly correlated across electrode sites (see Supplemental Materials examining their linear relationships). Therefore, statistical analyses will focus on the average of both Fz and Cz for all the ERPs. A 20Hz low-pass filter was applied to plot ERP waveforms, not in statistical analyses or scatterplots.

Initially, we created ERPs per participant and detrended to correct for slow linear drift in the waveforms (See Figure 4). The slow linear drift in the waveforms was most pronounced in the standard and deviant N100 amplitudes but less evident in the MMN amplitudes, as it was mitigated by the subtraction. However, the detrending method moved the baselines away from zero, complicating interpretation (see Supplemental Materials for Figures of detrended ERPs).

Figure 4.

Grand averaged standard N100, deviant N100, and MMN amplitudes for high schizotypy (black) and controls (grey) from the average of electrodes Fz and Cz. The shaded areas indicate the windows used to calculate the N100: (80–170 ms) and the MMN (150–250 ms). Asterisk represents a significant difference in latency (p<.05).

Statistical Analyses

Behavioral Data

The behavioral responses confirmed that participants were awake and attentive during the task. Reaction times over two trials after the presentation of the flash (~ 5 secs long) were assumed to be missed targets, and reaction times that were +/− 2 deviations from the median (0.37 ms) were excluded from analyses. T-tests measured differences in reaction times between the groups.

Total SPQ-BR Score

A multivariate analysis of variance (MANOVA) test measured statistical differences in component (N100 to standard and deviant tones and the MMN) amplitudes and peak latencies between the two groups (high schizotypy and controls) using the average signals from electrodes FZ and Cz.

Factor Scores: Amplitudes

Multiple regression models assessed relationships between each schizotypy factor (CP, IP, DIS) and their standard N100 amplitudes, deviant N100 amplitudes, and MMN amplitudes, respectively. For ERP components, the average signals from electrodes FZ and Cz were used in analyses, as the signal did not differ by electrode location. Bonferroni correction was applied per regression model (α = 0.05/3 =.017). A variance inflation factor multicollinearity test was performed to ensure the predictors in our models were not closely related. The variance inflation scores for CP (1.23), IP (1.15), and DIS (1.10) were close to the value of 1, indicating the absence of multicollinearity.

Factor Scores: Latencies

Multiple regression models assessed relationships between each schizotypy factor (CP, IP, DIS) and their standard N100 latencies, deviant N100 latencies, and MMN latencies, respectively. For ERP components, the average signals from electrodes FZ and Cz were used in analyses, as the signal did not differ by electrode location. Bonferroni correction was applied per regression model (α = 0.05/3 =.017).

Additional EEG Analyses

Participants were recruited based on their total scores (high schizotypy vs controls), see Participants section. To ensure that factors with statistically significant ERP amplitudes had sound score distributions and to visualize whether individuals’ factor scores were linked to their total SPQ-BR scores, we plotted IP and DIS cores as a function of total scores (see supplemental Figure S4).

IQ and Ravens Scores with ERPs

To identify potential relationships between fluid intelligence (Raven’s), GPA, and ERP amplitudes and latencies, exploratory multiple regressions were performed (see Supplemental Materials). No effects were found between high schizotypy and controls’ fluid intelligence metrics and ERP amplitudes or latencies.

Results

Behavioral Data

The behavioral task data confirmed that there were no between-groups differences in reaction times to the flashing fixation cross (t(55) = 1.36, p = .18). Thus, both groups were awake and attentive to the task.

Total SPQ-BR Score

We then assessed whether there were any differences between high schizotypy (total score) and controls’ (medium scoring schizotypy) N100 (standard, deviant) and MMN ERPs for the average of electrodes Fz and Cz. Although the full model was not statistically significant, [F(6,55)=1.22, p=.310, η²=.13], latency analyses revealed a statistically significant difference between the MMN latencies of high schizotypy and controls [F(1,55)=6.94, p=.011]. High schizotypy (M = 205.31, SD = 24.52) elicited a slower MMN than controls (M = 188.63, SD = 23.08), see Figure 4. N100 latencies to standard [F(1,55)=.39, p=.533] and deviant [F(1,55)=1.57, p=.215] tones were not different between the groups. No statistical differences were found between the N100 amplitudes to standard [F(1,55)=.00, p=.955], deviant tones [F(1,55)=.01, p=.923], or MMN [F(1,55)=.10, p=.749] of high schizotypy and controls.

Factor Scores: Amplitudes

Our primary interest was in the relationship between DIS factor score and N100 (standard, deviant) and MMN ERPs. First, we investigated the potential relationships between schizotypy factors (CP, IP, DIS) and deviant N100 amplitudes; see Figure 5. Although the model did not reach statistical significance [F(3,53)=2.60, p=.062, Cohen’s f²=.15], DIS was significantly related to smaller (i.e., more positive) N100 amplitudes to deviant tones [β = .37, t(52) = 2.69, p =.009], as predicted. No significant relationships were observed for CP (t = −.56, p =.576) or IP (t = −.81, p =.420). However, the schizotypy factors were not related to standard N100 amplitudes [F(3,53)=1.70, p=.178, Cohen’s f²=.10] or MMN amplitude [F(3,53)=.93, p=.433, Cohen’s f²=.05]; see Figures 6.

Figure 5.

The relationship between the schizotypy factor scores (cognitive-perceptual, interpersonal, and disorganized) and the amplitude and latency of their standard N100 waveform, deviant N100 waveform, and MMN obtained from the average of Fz and Cz. Cognitive Perceptual (CP) is in medium grey, disorganized (DIS) is in black, and interpersonal (IP) is in light grey.

Factor Scores: Latencies

An investigation into the relationship of schizotypy factors (CP, IP, DIS) and their component peak latencies yielded a significant relationship with MMN latencies [F(3,53)=3.92, p=.013, Cohen’s f²=.22]. IP was significantly related to greater latencies [β = .35, t(52) = 2.62, p =.011], see Figure 5 for scatterplots. No significant relationships were observed for CP (t = −.29, p =.771) or DIS (t = 1.71, p =.093). The schizotypy factors were not related to standard N100 latencies [F(3,53)=.50, p=.686, Cohen’s f²=.03], or deviant N100 latencies [F(3,53)=1.54, p=.215, Cohen’s f²=.09].

Discussion

In SSD, the auditory N100 and MMN are reliably perturbed in amplitude (N100: Rosburg et al., 2008; MMN: Erickson et al., 2016; Umbricht & Krljes, 2005). Our aim was to test if the N100 and/or MMN were also affected in participants exhibiting more SSD-like traits. We examined the relationship between schizotypy symptoms and responses to an oddball pitch-deviant paradigm. We found that the N100 to deviant tones was significantly lower in amplitude in those with more disorganized symptoms, but there was no reliable effect on the standard N100 amplitude or the MMN amplitude. There was no difference in N100 or MMN amplitudes with total schizotypy scores or with IP or CP scores. We also found that MMN latencies were delayed in high schizotypy compared to controls, and that greater latencies were related to more interpersonal symptoms. However, N100 latencies were not different between groups or related to the CP or DIS factor. Collectively, these findings provide insight regarding detectable ERP differences in a non-clinical high schizotypy trait load population.

The finding of similar N100 amplitudes to standard and deviant tones, and MMN amplitudes, in high schizotypy (total scores) compared to controls replicates previous findings (Francis et al., 2023; Klein et al., 1999). Similarly, N100 amplitudes were equivalent across controls and individuals diagnosed with schizotypal personality disorder (Mannan et al., 2001; Trestman, 1996). We note that we used a random ISI to control for any effects of stimulus timing predictability, thereby isolating neural responses to changes in frequency. This contrasts with other studies using a consistent ISI, who also found no significant effects of total schizotypy scores (Bissonnette et al., 2024; Donaldson et al., 2021). Therefore, our findings suggest that the frequency-detecting sensory predictive mechanisms (Garrido et al., 2008; Winkler, 2007; Winkler et al., 2009) or frequency-adaptive neuronal mechanisms (May & Tiitinen, 2010) that yield MMN amplitude (strength of response) is typical in high schizotypy.

Conversely, high schizotypy exhibited greater peak latencies than controls in the MMN, suggesting that the mechanisms underlying the MMN are taking more time to process, particularly in those with more IP symptoms. Prolonged auditory MMN has been documented in non-clinical high schizotypy in response to location deviants (Bissonnette et al., 2024). In schizophrenia, there is evidence of shorter MMN latencies to frequency and duration deviants (Horton et al., 2011; Kärgel et al., 2014), but prolonged latencies to frequency deviants with enhanced verbal memory performance (Kärgel et al., 2014).

Schizotypy Factors

Investigating schizotypy factor scores and their ERP amplitudes revealed that, in line with findings in schizophrenia (Boutros et al., 1997; See Rosburg et al., 2008 for review), DIS traits were linked to smaller N100 amplitudes to deviant tones. No relationship existed for CP (positive schizotypy) or IP (negative schizotypy). Although prior research has found relationships between MMN amplitudes and positive (Bissonnette et al., 2024; Broyd et al., 2016; Deng et al., 2024; Donaldson et al., 2021) and negative (Bissonnette et al., 2024; Donaldson et al., 2021) schizotypy traits, MMN amplitudes here did not change with trait load across the schizotypy factors. Importantly, impaired N100 amplitude but normal MMN amplitude suggests individuals who report disorganized symptoms experience a dysfunction of early auditory deviance detection, accompanied by intact sensory memory.

Conversely, greater IP (negative schizotypy) traits were linked to longer MMN latencies, which is consistent with previous work (Ford et al., 2020). Other factors did not relate to ERP peak timings. Intact N100 latencies but longer MMN suggest increased negative symptoms are linked to slower or impaired mechanisms involved in sensory memory, but intact auditory detection. Prolonged latencies have been documented in schizophrenia patients with increased negative scores on the PANSS or primarily displaying negative symptoms (Fisher et al., 2012; Pultsina et al., 2022). Pinpointing the effects of perturbed sensory processing on disorganized and interpersonal behaviors should be further explored.

Symptoms of disorganization

Disorganized symptom severity during the prodromal stage is strongly linked with transitioning into psychosis (Demjaha et al., 2012), earlier disease onset (age<20; Nestsiarovich et al., 2017), and the risk for more severe disease progression (Metsänen et al., 2006; Nestsiarovich et al., 2017; Reed et al., 2002; Shenton et al., 1992).

In non-clinical schizotypy, DIS SSD-like traits are associated with several impairments observed in schizophrenia. First, there is evidence for hypo-excitable auditory responses, such as longer P3 latencies to auditory frequency targets (Deng et al., 2023), and reduced P50 amplitude to paired clicks (Evans et al., 2007). Second, there is additional evidence for impaired emotional and communication processing, and speech production (Kerns, 2006; Kerns & Becker, 2008; Marggraf et al., 2019), pointing to auditory impairments across auditory regions. Together, this suggests a progressive worsening of auditory functioning that is linked to the severity of DIS symptoms.

Interestingly, we found no association between DIS symptoms and MMN amplitude. There is currently a relevant debate regarding the effectiveness of MMN as a biomarker for disease presence. MMN is reliably lower in chronic schizophrenia (Umbricht & Krljes, 2005), the reduction is less prevalent at first-episode (Erickson et al., 2016; Haigh et al., 2017), and it is negligible for at-risk individuals (Erickson et al., 2016). This suggests that MMN reflects disease progression more than a genetic risk for schizophrenia (Erickson et al., 2016).

In contrast, reduced N100 amplitudes in DIS non-clinical schizotypy could reflect a heightened risk for SSD that either did not manifest thanks to other protective factors or that could occur in the future. Most of our participants are within the typical age range of first episode of schizophrenia (Miettunen et al., 2019; Tanskanen et al., 2021), so conversion to SSD is possible. The stronger relationship between DIS symptoms and the deviant (compared to the standard) N100 suggests that low-level adaptation in primary auditory cortex may be impaired and provides a simple marker to explore for perturbed mechanisms of SSD that are impaired before psychosis.

Negative symptoms

The negative schizotypy factor (i.e., IP) measures traits associated with social anxiety and social anhedonia (subfactors: No close friends/constricted effect and social anxiety) (Asan & Pincus, 2023; Cohen et al., 2010). Interpersonal schizotypy has been linked to grey matter alteration and gyrification in critical cognitive and social-cognitive regions (Nenadić et al., 2025; Quidé et al., 2021; Wang et al., 2015), and is inversely related to P300 amplitude (Deng et al., 2023). There is also evidence for impaired auditory N100 with worse negative symptoms in schizophrenia (Giordano et al., 2021), Although here we found support for this in disorganized, not negative (IP) schizotypy. Certain impairments are linked to both negative and disorganized symptoms. Negative and disorganized schizotypy are linked to P300 impairments (Deng et al., 2023), exhibit functional activation during mentalizing in neural regions consistent with those activated in individuals with schizophrenia during mentalizing (Acosta et al., 2019), and both are associated with the environmental schizophrenia risk factor, social adjustment (Morton et al., 2016).

Our findings should be interpreted considering a few limitations. The fact that we found differences in auditory ERP amplitudes is striking because our sample was young, educated, and healthy. This also likely limited our score distribution. In future work, it will be useful to collect participants with SPQ-BR scores across a broader range of values. This may be possible by loosening the inclusionary criteria to include individuals with mood disorders (e.g., depression and anxiety), as these are highly comorbid with SSD (Buckley et al., 2009; Zhao et al., 2024). Methodologically, there is the prospective risk of obscuring individual differences in amplitude peak latency when selecting the ERP latency windows from the grand average collapsed across groups. However, amplitude measures are widely used in clinical samples (See Meta-analyses by Erickson et al., 2016; Haigh et al., 2017, Umbricht & Krljes, 2005), and there is little reason to assume that a sample of otherwise neurotypical subjects would be more heterogeneous in their ERP timings than clinical subjects (see Figure 4). Another limitation of our study is that although we asked participants to report whether they had normal hearing, we did not verify normal hearing with a hearing test (e.g., Rinne and Weber tests for air-to-bone conduction). Participants could be unaware of auditory impairments, so in future iterations of this study, an auditory examination should be included. Importantly, we sampled SPQ-BR scores from a total of 2,188 individuals, revealing the manifestation of total SPQ-BR traits within the undergraduate population at our university. We selected middle-scoring individuals as controls due to the increased prevalence and, therefore, greater likelihood that these amounts of traits (scores 50–70) represent the most “typical” manifestation. This approach could be a potential limitation, as it reduces the distribution of this group. There is also a need for longitudinal studies to track possible conversion into SSD, particularly with symptom-type load. Finally, the SPQ-BR is a self-report measure and thus, subject to reporting bias.

Conclusions

Individuals with high schizotypy exhibited prolonged MMN latencies but intact amplitudes—a pattern also observed in those with elevated negative symptoms. This points to delayed processing of auditory sensory memory with intact strength of activity and intact sensory detection. High schizotypy and those with more negative symptoms are exhibiting slower, less efficient auditory mechanisms. In contrast, smaller N100 responses to deviant tones were observed with more DIS symptoms. Interestingly, MMN amplitude was not related, suggesting that early auditory deviance detection worsens with more DIS SSD-like traits, but predictive sensory memory mechanisms are unaffected. Our findings highlight the need to focus on subfactors of schizotypy in addition to total scores, as they can reveal more about the differences in underlying physiology in a non-clinical population. Further research into DIS and negative (IP) schizotypy traits is required to validate schizotypy as a population at risk for SSD.