Review of neurophysiological findings in patients with schizophrenia



Schizophrenia has been conceptualized as a failure of cognitive integration, and abnormalities in neural circuitry have been proposed as a basis for this disorder. In this article, we focus on electroencephalography and magnetoencephalography findings in patients with schizophrenia. Auditory-P50, -N100, and -P300 findings, visual-P100, -N170, and -N400 findings, and neural oscillations in patients with schizophrenia are overviewed. Published results suggest that patients with schizophrenia have neurophysiological deficits from the very early phase of sensory processing (i.e. P50, P100, N100) to the relatively late phase (i.e. P300, N400) in both auditory and visual perception. Exploring the associations between neural substrates, including neurotransmitter systems, and neurophysiological findings, will lead to a more comprehensive understanding of the pathophysiology of schizophrenia.

HUMAN PSYCHIC ACTIVITY is based on brain neural activities, and thus, neurophysiological approaches are important to understand the pathophysiology of neuropsychiatric disorders. Since the time of Bleuler, schizophrenia had been described as ‘splitting of the psychic functions’. In the past 3 decades, it has been proposed that the cognitive and affective impairments of schizophrenia may be related to a failure to integrate the activity of local and distributed neural circuits. Recent neurophysiological approaches to understanding the pathophysiology of schizophrenia include electroencephalography (EEG), electromyography, magnetoencephalography (MEG), positron emission tomography, single-photon emission computed tomography, functional magnetic resonance imaging, near-infrared spectroscopy and transcranial magnetic stimulation. These sophisticated neurophysiological techniques have revealed aspects of the pathophysiology of schizophrenia. In this review, we focus on EEG and MEG findings in patients with schizophrenia to present an overview of the results of recent studies measuring evoked or event-related potentials and neural oscillations in patients with schizophrenia.

Auditory-Evoked or Event-Related Responses


The human middle latency auditory-evoked potentials (MLAEP) are elicited by auditory stimuli in the 10–80 ms latency range after the stimulus onset. The typical MLAEP is composed of several different deflections, including the Na (or Nam in the MEG literature) at 18 ms, Pa or P30 (Pam or P30m) at 30 ms, Nb (Nbm) at 40 ms, and Pb or P50 (Pbm or P50m) at 50 ms. The P50 component has become of clinical interest in patients with various psychiatric disorders.[1-6] Suppression of P50 is considered to be a reflection of the brain's inhibitory mechanisms against information overload and is usually assessed by an auditory paired-click paradigm. In healthy subjects, the P50 wave shows reduced amplitude in response to the second click when presented 500 ms after the initial click. The degree of suppression is quantified as the ratio of the P50 amplitudes in response to the first and second clicks.[1] An elevated gating ratio in patients with schizophrenia has been repeatedly reported in many studies.[1, 2, 7] The P50m generators are estimated to be located in or near the primary auditory cortex.[8-11] Conversely, P50 may be a result of overlapping potentials. For example, Polyakov and Pratt[12] found that the primary auditory cortex generator and ascending subcortical generators both contribute to MLAEP.

P50 suppression is often absent or reduced in patients with schizophrenia[1, 2, 7] and this deficit is considered to be a well-validated and heritable neurophysiological biomarker of schizophrenia.[7, 13-17] Moreover, reduced P50m suppression in response to human voices was also reported in patients with schizophrenia.[18] P50 gating deficits have been detected in prodromal subjects[19, 20] and first-episode patients with schizophrenia.[20-22] Some schizophrenia family studies have reported abnormal P50 gating,[13, 14] although negative findings have also been reported.[23] In a study including a relatively large number of subjects, Clementz et al. concluded that family members of patients with schizophrenia show P50 gating deficits but that these are not severe compared with those in patients.[15] Some reports have suggested a possible recovery of P50 deficits, with equivocal results. Adler et al. reported that the P50 deficits in the relatives of patients with schizophrenia changed to normal levels after administration of nicotine gum.[24] P50 deficits in patients were also reported to be improved after treatment with risperidone[25] or quetiapine,[21] although Hong et al. did not find such an improvement with several antipsychotics.[22] It has been suggested that P50 gating dysfunctions may be a marker for the potential to develop positive symptoms (see the review by Potter et al.[26]).


The auditory N100 component is a negative evoked potential peaking between 80 and 120 ms after the onset of a stimulus. EEG studies have suggested that the N100 consists of a component generated in the supratemporal auditory cortex and multiple other components consisting of complex potentials.[27] However, Hari et al.[28] recorded the N100 and the N100m simultaneously and reported that the neural source of the N100 was in the supratemporal auditory cortex, as was the neural source of the N100m, when the interstimulus intervals (ISI) were below 4 s. The auditory N100 component is elicited by any discernible auditory stimulus without task demands, and the amplitude is influenced by several factors, including ISI, stimulus intensity, arousal level, and subjects' attention. Reductions in N100 amplitude have been consistently reported in patients with schizophrenia.[29] Auditory N100 amplitude reduction has been proposed as a marker of functional brain abnormalities related to the genetic predisposition to schizophrenia.[30]

Mismatch negativity

Mismatch negativity (MMN) is the negative component of a waveform obtained by subtracting event-related responses (ERP) to a frequent stimulus (standard) from those to a rare stimulus (deviant), with ISI of approximately 500–1000 ms. The MMN can be elicited regardless of whether the subject is paying attention to the sequence, and it reflects the function of auditory sensory memory. The MMN is thought to reflect an automatic process that detects a difference between an incoming stimulus and the sensory memory trace of preceding stimuli.[31] The scalp MMN is generated from the auditory cortices bilaterally,[32, 33] but there may also be a contribution from the right frontal cortex.[34]

Attenuated MMN amplitude for changes in the duration and frequency of a repetitive sound indicates a pathophysiological mechanism of auditory automatic detection in patients with schizophrenia.[35] The overall effect size of the difference in MMN amplitude between patients and healthy controls was 0.99 (95% confidence interval: 0.79–1.29) for frequency or duration deviants. MMN to duration deviants produced a larger effect size than MMN to frequency deviants, suggesting that processing of duration deviants is more impaired than processing of frequency deviants;[36] however, this difference in effect size was not statistically significant. Michie et al.[37] reported reduced MMN amplitudes in unaffected first-degree relatives of patients with schizophrenia, suggesting that it is an endophenotype marker of vulnerability to schizophrenia. Of note, a progressive decrease in MMN was reported[38] and this may reflect functional deficits arising as the disease progresses in patients with schizophrenia.

According to a meta-analysis, no specific factor was significantly associated with MMN deficits, although MMN in response to duration deviants appeared more impaired in patients with schizophrenia than did MMN in response to frequency deviants. In addition, the effect sizes of frequency MMN were significantly correlated with the duration of illness.[39] Asymptomatic relatives of patients with schizophrenia also showed duration MMN abnormalities, but studies in larger samples have either not confirmed these findings or showed a trend only (see review[40]). Recent studies have confirmed the characteristics of duration MMN where first-episode schizophrenia subjects and even prodromal subjects showed duration MMN abnormalities.[41-43] Interestingly, in one study, only prodromal subjects who converted to schizophrenia later showed duration MMN abnormalities but those who did not convert showed no such abnormalities.[42] Hall et al. investigated duration MMN and P300 in psychotic bipolar disorder patients and found that these patients showed P300 amplitude reductions but normal duration MMN.[44] Taken together, these findings suggest that duration MMN could serve as a trait marker for schizophrenia.


The P300 component, or P3, is a positive event-related potential peaking at around 300–500 ms and two major subcomponents have been well investigated. The P300a is produced without a task when an infrequent distractor in a series of frequent stimuli is presented, whereas P300b is elicited by infrequent and task-relevant target stimuli during an oddball paradigm. P300a is related to automatic attention processing and located with more frontal distribution. P300b is thought to reflect a variety of cognitive processes elicited by the contextual updating of working memory and affected by conditions like the attention resource allocation to a deviant stimulus and target-to-target interval, and usually most prominent at parietal electrodes.[45] Intracranial recordings[46] have indicated that P300 is a component with multiple generators, including the prefrontal, anterior cingulate, superior temporal, and parietal cortices, and the hippocampus.

Although P300b is elicited by both auditory and visual modalities, reduced amplitudes and prolonged P300b latency in patients with schizophrenia have been consistently detected in the auditory modality.[7] Auditory P300b amplitude reductions and prolonged P300b latencies have been reported repeatedly in patients with schizophrenia. Moreover, previous studies have shown an increased rate of P300b latency prolongation with age in medicated chronic,[47] first-episode and drug-naïve patients with schizophrenia,[48] suggesting a pathological neurodegenerative process.

Mathalon et al. found that both visual and auditory P300b fluctuate, tracking changes in the clinical state, but only auditory P300b showed significant amplitude reductions, even when patients were least symptomatic.[49] From this point of view, auditory P300 amplitude could be viewed as a trait marker for schizophrenia, while visual P300 can be viewed as a state marker. Indeed, auditory P300 abnormalities have been reported in prodromal states[50, 51] and first-episode schizophrenia.[52, 53] Of note, Salisbury et al. reported a different distribution of P300 between patients with schizophrenia and bipolar disorder patients,[54] and this altered P300 distribution might have been due to a reduced superior temporal gyrus volume in patients with schizophrenia.[55]

Visual-Evoked or Event-Related Responses


The visual processing of objects in the brain can generally be divided into two processing pathways that carry qualitatively different information from the retina to the primary visual cortex: the magnocellular pathway (M-pathway), which is stimulated by low-contrast images, low spatial frequency informati on and motion; and the parvocellular pathway (P-pathway), which responds to high luminance contrast, high spatial frequencies and color. These two pathways project to the dorsal (the ‘where’ pathway) and ventral (the ‘what’ pathway) visual cortical streams, respectively.[56] Two neurophysiological paradigms have been used to investigate the early visual processing function. In the transient ERP approach, discrete stimuli are presented repetitively and elicit a sequence of visual-evoked potentials.

The P100 component is a positive potential at the occipital area, elicited at around 100 ms after the stimulus onset and generated within the extrastriate cortex.[57] Global processing is the initial stage of face categorization,[58] and the P100 potential is associated with the global processing of visual perception. P100 has dual sources in the dorsal and ventral visual streams with M-pathway dominance.[59] P100 deficits in patients with schizophrenia have been frequently reported.[60] For example, it was reported that patients showed reduced P100 amplitude in response to non-face stimuli,[61] and face stimuli, compared with healthy subjects.[62] A reduction in P100 amplitude was also reported in studies using M-pathway-biased stimuli, while studies using P-pathway-biased stimuli did not observe such reductions in response to non-face stimuli.[62] Obayashi et al.[63] reported that healthy subjects showed a significant P100 amplitude enhancement in response to low spatial frequency filtered face stimuli compared with responses to non-filtered stimuli, while this enhancement was not observed in patients. These results may support the existence of an M-pathway deficit in schizophrenia at the lower level (around extrastriate) of visual processing.

P100 reduction has also been observed in first-episode patients with schizophrenia[64] and unaffected schizophrenia relatives.[65] Furthermore, P100 abnormalities are correlated with social functions, but not with symptoms,[66] age, duration of illness, or medication.[60] These findings suggest that the visual P100 component might be a trait marker for schizophrenia.


The N170 component is a negative potential recorded over occipitotemporal areas at around 170 ms after the stimulus onset, and is considered to function as an index of the structural encoding of faces,[67] and the extraction of configural information.[68, 69] The N170 is also associated with local processing during the identification of individual faces.[58] Data from ERP studies have demonstrated that the negative potential recorded at occipitotemporal leads, the N170, is more negative (i.e. larger) for faces than for complex objects in healthy subjects.[70] However, several studies have found that schizophrenics show smaller differences in N170 amplitudes between faces and complex objects than do normal controls.[71, 72] Although both face and complex object perception use the ventral pathway, faces are perceived, at least in part, by a separate processing stream in the ventral pathway.[73] Based on converging evidence from EEG[74] and MEG,[75] the fusiform gyrus (FG) is considered to be one of the main neural sources of the N170 in response to faces. It has been repeatedly reported that patients with schizophrenia show reduced N170 amplitudes in response to neutral and emotional faces.[63, 76-80]

Even in healthy subjects, the effects of facial emotional expressions on the N170 still remain controversial. For example, Batty and Taylor found that the N170 amplitude in response to fearful faces was larger than the N170 in response to neutral faces in healthy subjects.[81] However, it was reported that the N170 was unaffected by emotional expressions in healthy subjects.[82, 83] The authors documented that the N170 reflects the process of identification and recognition of faces, but not facial emotional expressions. Lynn and Salisbury[79] demonstrated that healthy subjects showed bilateral differences in N170 amplitude among facial expressions, while schizophrenic patients failed to show this modulation. However, Obayashi et al.[63] reported no significant effects of facial expressions in normal controls or schizophrenics, regardless of the use of spatially filtered face images. They also suggested possible dysfunction of the magnocellular and parvocellular pathways, which may underlie the deficits associated with facial recognition in schizophrenic patients. Consistently, electrophysiological studies have indicated that the reduction in N170 amplitude in patients with schizophrenia during face and facial affect processing may reflect deficient processing of facial structures and facial structure encoding.

In terms of clinical correlations, more severe social dysfunctions were significantly correlated with reduced N170 amplitudes in response to faces in patients with schizophrenia.[63, 84] N170 amplitude reductions have also been found in prodromal subjects,[85] and unaffected schizophrenia family members failed to show N170 modulation according to the valence of facial stimuli.[86] The findings of face-N170 amplitude reductions suggest that basic perceptual deficits in patients with schizophrenia play a significant role in their inability to correctly interpret the state and intentions of others. A link between complex visual perception and social and occupational functions has been demonstrated. This link needs to be considered as a possible primary cause of the complex social deficits in schizophrenia, and presents an opportunity for the development of new interventions incorporating a remediation that addresses an ultimately perceptual deficit.[87]


The ERP component most widely used to probe semantic processes is the N400. The N400 is regarded as an index of an ease of connecting two semantic concepts. When two words are presented sequentially, the first word activates processing of semantically related words and inhibits processing of words that are not related to the first word within 400–500 ms after the stimulus onset. The process of fitting the word into the context occurs after this 400–500 ms period, and together these result in greater ease of cognizance of related words. Thus, designs using a short temporal distance (stimulus onset asynchrony [SOA]) between two words probe primarily the early processes of activation and inhibition, while designs using longer SOA over 500 ms probe the processes underlying context use and contextually driven inhibition.[88] The N400 amplitude has to be less negative when a word fits well into the previous word/context and more negative when the word does not fit well.

In a shorter SOA study, patients with schizophrenia showed smaller N400 amplitudes in a semantically unrelated condition but not in a related condition. In addition, healthy controls showed reductions of the N400 amplitude in the related condition compared with N400 amplitudes in the unrelated condition, but patients with schizophrenia did not show this effect, suggesting that the abnormality in schizophrenia was related to inefficient early inhibitory processes.[89] Conversely, in longer SOA studies, larger N400 amplitudes compared with those in healthy controls have been reported regardless of whether the presented sentence is nonsensical,[90, 91] which provides evidence that late processes of context use are inefficient in patients with schizophrenia.

Neural Oscillations

Neural oscillations and synchronization may reflect variable signals underlying flexible communication within and between cortical areas. In particular, neural oscillations in the gamma frequency band (30–80 Hz) are thought to play a crucial role in information processing in cortical networks.[92] Two types of neural oscillations exist: evoked neural oscillations, in which phases are locked to the stimulus onset; and induced neural oscillations, which are not strictly locked to the stimulus onset, but related to the stimulus. The auditory steady-state response (ASSR) is one of the evoked neural oscillations, and can be recorded by presenting click trains of 40-Hz frequency.[93] Although the ASSR itself may not reflect cognitive processes, the resonant frequencies of the ASSR suggest that basic neural circuits predominantly oscillate at 40 Hz.

Several previous studies have reported reduced 40 Hz-ASSR in patients with schizophrenia and their relatives. For example, Kwon et al.[94] reported that patients with schizophrenia showed diminished 40 Hz-ASSR-power. Hong et al.[95] reported that relatives with schizophrenia spectrum personality symptoms had reduced 40 Hz-ASSR-power. Light et al.[96] reported reductions in both evoked power and phase synchronization in response to 30 and 40 Hz-ASSR in patients with schizophrenia. Our group also reported a reduced 40 Hz-MEG-ASSR in patients with chronic schizophrenia.[97] Spencer et al.[98] reported that phase locking factor (PLF) deficits in the 40 Hz-ASSR were more pronounced over the left hemisphere in first-episode patients with schizophrenia.

For evoked neural oscillations to speech sounds, Palva et al.[99] reported different MEG patterns of evoked neural oscillations at 20–45 Hz between speech and non-speech sounds, suggesting the existence of a fast mechanism for identifying speech sounds in healthy subjects. Based on this finding, our group investigated evoked neural oscillations at 20–45 Hz in patients with schizophrenia.[100] The reported major findings are as follows: (i) patients showed a delayed neural oscillatory activity to speech sounds in the left hemisphere and to non-speech sounds in the right hemisphere; (ii) in the 0–50-ms period, patients showed significantly reduced evoked neural oscillation power to speech sounds in the left hemisphere; (iii) in the 100–150-ms period, patients showed significantly greater evoked neural oscillation power to speech sounds in the left hemisphere; and (iv) patients showed the opposite hemisphere patterns in the peak latency to both sounds and in the peak power to speech sounds. This study indicated that schizophrenia patents show different characteristics of evoked neural oscillations compared with healthy controls, and this is probably related to deficits in a fast mechanism for identifying speech sounds. Recently, neural oscillatory activities can be useful for translational research in psychiatry.[101]

Future Prospects

We have briefly overviewed neurophysiological findings in patients with schizophrenia. The findings suggest that patients with schizophrenia have neurophysiological deficits from the very early phase of sensory processing (i.e., P50, P100, N100) to the relatively late phase (i.e., P300, N400). It will be interesting to investigate how these neurophysiological components are related to each other. Friedman et al. investigated the correlation between auditory MMN and visual P100 in patients with schizophrenia, but they found no significant correlation across modalities.[102] Similarly, Atkinson et al. examined the correlation between auditory MMN and auditory P300a, but they also failed to find any significant correlations,[43] suggesting that these components represent independent deficits.

Newly emerging techniques, like frequency domain analysis, can shed light on novel aspects of the conventional components. Hall et al. recently investigated the evoked power of gamma- and beta-band responses using wavelet analyses to first (S1) and second (S2) stimuli in a paired-click paradigm. They found that patients with schizophrenia showed widely distributed reductions in gamma and beta oscillations to S1 stimuli and S2 stimuli, respectively, and impaired gating in both frequencies.[103] Other examples include wavelet transforms of P300 data. Ergen et al. performed a time frequency analysis, focusing on the delta frequency range, on both averaged ERP and single trials of visual odd-ball task data, which produced evoked (phase-locked) and total (phase-locked + non-phase-locked) delta responses. In patients with schizophrenia, evoked delta and P300 amplitudes in response to target stimuli were both reduced but total delta amplitudes were not, suggesting that the P300 amplitude reduction in schizophrenia results from a temporal jitter in the activation of neural circuits engaged in P300 generation.[104]

It will also be important to investigate the molecular functions underlying these components. For example, polymorphisms in the gene encoding the α7 nicotinic acetylcholine receptor may be related to the P50 sensory gating deficits in patients with schizophrenia.[105, 106] More recently, Szabolcs et al. reported that the reduction in the ratio of phosphorylated protein kinase B induced by neuregulin 1 was associated with impaired P50 sensory gating in patients with schizophrenia.[107] The glutamatergic system, especially NMDA receptors, is a longstanding candidate for modulation of MMN. It is relatively consistently found that administration of NMDA receptor antagonists reduces MMN amplitudes or delays MMN latencies;[108-111] however, other receptors like γ-amino butyric acid, serotonin, dopamine, muscarinic and nicotinic receptors might also be involved (see reviews[35, 112]). Moreover, P300a and P300b are thought to be associated with the dopaminergic and locus coeruleus norepinephrine systems, respectively (see reviews[45, 113]). Exploring the associations between neural processing, including neurotransmitter systems, and neurophysiological findings, will lead to a more comprehensive understanding of the pathophysiology of schizophrenia.