Schizophrenia is conceptualized as a failure of cognitive integration, and altered oscillatory properties of neurocircuits are associated with its symptoms. We hypothesized that abnormal characteristics of neural networks may alter functional connectivity and distort propagation of activation in schizophrenic brains. Thus, electroencephalography (EEG) responses to transcranial magnetic stimulation (TMS) of motor cortex were compared between schizophrenia and healthy subjects. There was no difference in the initial response. However, TMS-induced waves of recurrent excitation spreading across the cortex were observed in schizophrenia, while in healthy subjects the activation faded away soon after stimulation. This widespread activation in schizophrenia was associated with increased oscillatory activities in the proximal central leads and in fronto-temporo-parietal leads bilaterally. A positive correlation was found between increased TMS-induced cortical activation in gamma frequency and positive symptoms of schizophrenia, while negative symptoms were correlated with activation in theta and delta bands. We suggest that excessive activation in response to stimulation in schizophrenia brains may lead to abnormal propagation of the signal that could potentially result in aberrant activity in areas remote from the activation origin. This mechanism may account for the positive symptoms of schizophrenia and could worsen signal to noise deficits, jeopardizing adequate information processing with ensuing cognitive deficits.
Schizophrenia is a heterogeneous disorder characterized by a complex constellation of seemingly unrelated symptoms, including hallucinations, delusions, thought disorder, affective flattening, and pronounced cognitive deficits. Recent evidence supports the notion that a wide range of deficits in schizophrenia may result from a failure to integrate the activity of local and distributed neural circuits. This includes abnormal power and synchronization patterns of induced or evoked electroencephalography (EEG) rhythmic activity in both medicated and medication-naive patients (Spencer et al. 2004; Cho et al. 2006; Uhlhaas et al. 2006; Minzenberg et al. 2010) as well as decreased entrainment of oscillatory activity, primarily in high-frequency bands, in response to a steady-state stimulation (Kwon et al. 1999).
Impaired ability of distributed neuronal networks to integrate information in schizophrenia has been attributed to abnormalities in the rhythm-generating neuronal networks, such as inhibitory interneurons (Lewis et al. 2005). The crucial role of inhibitory deficits in the pathophysiology of schizophrenia has been suggested by abnormalities in the functional integrity, morphology, and distribution of inhibitory interneurons in schizophrenic patients (Lewis et al. 2008). Further support for impaired inhibitory neurocircuits in schizophrenia is provided by evidence of significant deficits in intracortical inhibition in response to transcranial magnetic stimulation (TMS) in patients with schizophrenia (Daskalakis et al. 2002). These deficits have also been documented in first-episode schizophrenia patients (Wobrock et al. 2008) and are more pronounced in unmedicated patients (Daskalakis et al. 2002).
Despite ample evidence linking affective and cognitive symptoms of schizophrenia with abnormalities of neural oscillatory activity (Uhlhaas et al. 2006), to date, we have a little understanding of how impaired synchrony between distributed neuronal networks translates into specific symptoms of schizophrenia. Current theories of cognitive functioning associate perceptual awareness with long-range synchronous rhythmic oscillatory activity. Intricate loops of feedback and feed-forward inhibition are known to segregate the activated neuronal assemblies into fine spatial and temporal domains specific to the incoming stimulus, which thus generate geometrically discrete rhythmic oscillations in distributed neuronal networks (Llinas et al. 2005). Binding of the multisensory inputs into a coherent cognitive experience is reliant on this inhibitory rhythm-generating “clustering,” as diminution of γ-aminobutyric acid release (GABAergic) inhibition has been shown to not only distort synchronized brain activity (Fingelkurts et al. 2004), but also alter perceptual selectivity (Wang et al. 2000, 2002). There is substantial theoretical and empirical evidence indicating that inhibition determines the spread of cortical activation by sculpting oscillatory patterns in time and space. Blocking inhibition has been shown to alter spatial and temporal patterns of neuronal activity, resulting in lateral spread of stimulation-induced neuronal activation in in vitro model systems (Contreras and Llinas 2001). Spatio-temporal patterns of spreading synaptic activity in response to stimulation were also reported in rat brain slices in the presence of both, dopamine and the GABAergic antagonist, bicuculline, 2 neurotransmitter mechanisms implicated in the pathophysiology of schizophrenia (Bandyopadhyay et al. 2005). It is possible, therefore, that deficient inhibitory neurocircuits lead to disrupted signal propagation of neuronal excitation in schizophrenia brains.
Hence, we hypothesized that deficits in inhibitory neuronal networks, as has reliably demonstrated in patients with schizophrenia, may ultimately result in excessive spread of neuronal excitation in response to an incoming stimulus in the brains of patients with schizophrenia. However, addressing this question in live humans is not a trivial task. A recent study by Massimini et al. (2005) has provided elegant methods through which to approach such questions. Specifically, the group assessed effective cortical connectivity in humans using TMS and EEG recordings during sleep and wakefulness. Electrical oscillations evoked by TMS during wakefulness have been shown to propagate to connected cortical areas several centimeters away, while same stimulus delivered during the non-rapid eye movement (NREM) sleep was rapidly dampened and did not propagate beyond the site of stimulation. The study suggests that diminished cortical connectivity reflects the failure of brain's ability to integrate information, which may underlie loss of conscious awareness during sleep (Massimini et al. 2005). Of note, TMS-induced electrical field stimulates primarily horizontally situated cortical interneurons and perpendicularly aligned pyramidal and non-pyramidal cells located within the sulci, while no direct stimulation is transmitted to deep subcortical structures. Therefore, it can be inferred that changing patterns of brain activity from high frequency (wakefulness) to slow-wave rhythmic activity (NREM sleep) are sufficient to alter functional (e.g., physical) properties of the cortex, or its functional conductivity. Of note, we purposely avoid attributing altered patterns of neuronal activity to changes in functional (or effective) connectivity, as it is well-known that in some pathological conditions increased excitation may result from decreased anatomical, functional, or effective connectivity (e.g., loss of inhibitory or some excitatory neuronal populations enhance local excitability while decreasing long-range connectivity; Moeller et al. 2009), so this term may in fact become misleading. Instead, we propose to use a more impartial characteristic, functional cortical conductivity, depicting a purely physical property of the brain, which is however a product of functional interactions between its neuronal elements. A large-scale mathematical modeling of thalamocortical systems demonstrated that a shift in the excitation/inhibition balance toward inhibition may account for this diminished intercortical signal propagation (Esser et al. 2009). This, along with evidence of diminished cortical inhibition in schizophrenia, provides further support for a possibility of altered cortical conductivity in schizophrenia brains.
To test this hypothesis, we used a combination of TMS–EEG recordings in order to compare cortical activity resulting from direct cortical stimulation in schizophrenia subjects to their healthy counterparts. If our hypothesis is confirmed, we could expect to see a disrupted pattern of cortical conductivity in patients with schizophrenia.
Materials and Methods
We studied 16 patients (mean age 36.7 ± 10.4 years; 12 males, 4 females) with a diagnostic and statistical manual of mental disorders (DSM) diagnosis of schizophrenia or schizoaffective disorder confirmed by the structured clinical interview for DSM IV and 16 healthy subjects (mean age 36.1 ± 7.9; 11 males, 5 females). Fourteen of 16 patients with schizophrenia were treated with antipsychotic medications (clozapine, n = 6, mean dose 400 ± 54.8 mg/day; risperidone, n = 3, mean dose 3.2 ± 2.5 mg; haloperidol, n = 2, mean dose 2 ± 1.4 mg; quetiapine n = 1, 100 mg; perphenazine n = 1, 16 mg; olanzapine, n = 1, 7.5 mg) and with no other psychotropic medications. All participants were right-handed as assessed by the Oldfield Handedness Inventory. Healthy volunteers were screened for psychopathology and excluded if they had psychiatric, neurological, or major medical illness or were suffering from substance abuse disorder. The ethics committee at the Centre for Addiction and Mental Health approved the study and written informed consent was obtained for each participant.
Our experimental design consisted of measuring global cortical activation and cortical excitability in response to a single monophasic TMS pulse. Cortical excitability was determined using the motor-evoked potential (MEP) size, which was defined as the intensity of TMS stimulus sufficient to produce mean MEP amplitude of 1-mV peak-to-peak response at the baseline (stimulus intensity of 1 mV or SI1 mV). This was consistent with previous studies using both TMS–electromyography (EMG) and TMS–EEG recordings (Farzan et al. 2010, Frantseva et al. 2008).
To determine SI1 mV, the average MEP of 20 stimuli at rest was calculated. TMS was administered over the left motor cortex of all subjects. One hundred TMS stimuli were delivered per trial for every 5 s. Moreover, to control for the effect of TMS click-induced auditory activation on the cortical-evoked potentials, single-pulse Sham-stimulation was administered in all healthy and schizophrenia subjects at the same intensity as used for active stimulation but with the coil angled at 90° from the scalp resting on one wing of the coil. During the experiment, subjects were sitting in a comfortable armchair with their eyes open, their elbow flexed, and their hands rested on a pillow placed on their laps.
Transcranial Magnetic Stimulation
TMS of the left motor cortex was performed with a 7-cm figure-of-eight coil and a Magstim 200 stimulator (The Magstim Company, Whitland, United Kingdom). The coil was placed at the optimal position for eliciting EMG-recorded MEPs from the right abductor pollicis brevis (APB) muscle, which typically corresponded to a region between FC3 and C3 electrodes on the 10–20 EEG system. The optimal position was marked on the scalp with a felt pen to ensure identical placement of the coil throughout the experiment. The handle of the coil pointed backward and was perpendicular to the presumed direction of the central sulcus, about 45° to the midsagittal line. The direction of the induced current was from posterior to anterior and was optimal to activate the motor cortex transsynaptically.
Surface EMG was recorded from the right APB muscle with disposable disc electrodes placed in a tendon-belly arrangement over the bulk of the APB muscle and the first metacarpal–phalangeal joint. The subject maintained relaxation throughout the experiment, and the EMG was monitored on a computer screen and via speakers at high gain. The signal was amplified (Intronix Technologies Corporation Model 2024F, Bolton, Ontario, Canada), filtered (band pass 2 Hz to 2.5 kHz), digitized at 5 kHz (Micro 1401, Cambridge Electronics Design, Cambridge, United Kingdom), and stored in a laboratory computer for off-line analysis.
To evaluate TMS-induced and Sham-induced cortical-evoked potentials, EEG was recorded concurrently with the EMG. EEG was acquired through a 64-channel EEG cap, and 4 electrodes were placed on the outer side of each eye, and above and below the left eye to monitor the eye movement artifact. All electrodes were referenced to an electrode placed posterior to the central electrode. EEG signals were recorded direct current and a low-pass filter of 100 Hz at the 20-kHz sampling rate, which was shown to avoid saturation of amplifiers and to minimize the TMS-related artifact (Daskalakis et al. 2008).
All TMS–EEG signals were downsampled from 20 to 1 kHz. The \ 60-\Hz powerline (f0) artifact was removed from the signals by using the Thomson F-test based on multitaper spectral estimate techniques (Percival and Walden 1993). The signals from all 60 channels were subsequently average referenced (Nunez and Srinivasan 2006). The preprocessing of Sham-EEG signals follows that of TMS–EEG preprocessing explained above in this section.
An average signal was obtained from each electrode for TMS–EEG and Sham-EEG. To further smooth the data, a sliding window with 100-ms width stepping at 1 ms was employed. The mean voltage within this window at each time instance was calculated. We thus obtained 60 smoothed each for TMS–EEG and Sham-EEG signals from 60 electrodes for the healthy and the schizophrenia (SCZ) subjects, respectively. Moreover, the spatial voltage distributions over scalp surface were derived from the signals with the topographic toolbox of EEGLAB.
The power spectra of the signals were calculated for a window of 400–800 ms post-test stimulus (TS). Optimized spectral smoothing/concentration was achieved using the multitaper method (Jarvis and Mitra 2001). In this analysis, we chose the number of tapers (k = 2), and the time–frequency product time -frequency window (TW) = 1.5, resulting in the frequency resolution of 7.5 Hz.
The data epochs were extracted from the sliding windows with 200-ms width that were moved over the TMS–EEG signals in steps of 10 ms. For each epoch extracted, the corresponding power spectrum was obtained.
To obtain an error bar of the estimate at each time–frequency location on the spectrogram for Figure 3A,B, we applied the method of Jackknife, that is, we leave one trial out and calculate the spectrum at each time instant from the rest trials. By repeating this procedure, we obtain an ensemble of estimates at each time–frequency location. The error bar indicates the standard deviation of the estimates, which are considered significantly different if the estimates are separated by more than 2 standard deviations.
To obtain a relation between measure of SCZ symptoms and conductivity (Fig. 4), time–frequency signal power in delta (1–3.5 Hz), theta (4–7 Hz), alpha (8–12 Hz), beta (12–28 Hz), and gamma (30–50 Hz) was evaluated from the average TMS–EEG signal across 60 electrodes for each subject. A time-varying maximum power was computed for each frequency band providing a measure of conductivity (maximum spectral power) for the time period of interest. Positive and negative symptoms based upon the positive and negative syndrome scale (PANSS) score were used as the measure of SCZ symptoms. A linear correlation test between the symptom score (positive and negative) and maximum spectral power (within each frequency band at 200 ms and between 400 and 700 ms) was performed. Regression plots were generated for the statistically significant correlation results.
Brain responses evoked by TMS were recorded using a 64-channel EEG amplifier in a group of 16 healthy volunteers and 16 schizophrenia patients, 14 of whom were treated with antipsychotics, and 2 were unmedicated. Monophasic single TMS pulses were delivered to the motor cortex, as described previously (Daskalakis et al. 2008). While some previous studies have shown significant difference in resting motor thresholds between schizophrenia and normal control subjects, no statistically significant difference in the resting motor threshold was found in our experiments (P = 0.2, Table 1).
|Schizophrenia (n = 16)||Healthy (n = 16)||P-value|
|Age (years)||36.7 ± 10.4||36.1 ± 7.9||0.85|
|Illness duration (years)||9.7 ± 7.3||NA||NA|
|Resting motor threshold||56.6 ± 13.9||51.3 ± 7.5||0.20|
|1 mV Intensity||66.9 ± 16.5||64.1 ± 10.0||0.56|
|Total||65.3 ± 17.6||NA||NA|
|Positive||16.3 ± 4.3||NA||NA|
|Negative||18.3 ± 6.1||NA||NA|
|Global||30.4 ± 8.6||NA||NA|
|Schizophrenia (n = 16)||Healthy (n = 16)||P-value|
|Age (years)||36.7 ± 10.4||36.1 ± 7.9||0.85|
|Illness duration (years)||9.7 ± 7.3||NA||NA|
|Resting motor threshold||56.6 ± 13.9||51.3 ± 7.5||0.20|
|1 mV Intensity||66.9 ± 16.5||64.1 ± 10.0||0.56|
|Total||65.3 ± 17.6||NA||NA|
|Positive||16.3 ± 4.3||NA||NA|
|Negative||18.3 ± 6.1||NA||NA|
|Global||30.4 ± 8.6||NA||NA|
Note: Data are given as mean ± standard deviation.
PANSS, positive and negative syndrome scale; NA, not applicable.
Characteristics of TMS-Induced Cortical Activations
Figure 1 illustrates a comparison of TMS-induced EEG activity between schizophrenia patients and healthy control subjects. In healthy subjects, initial voltage deflections elicited by a TMS pulse were primarily limited to 300-ms poststimulus interval (Fig. 1A), whereas in schizophrenia patients, who displayed similar characteristics of the initial (75–150 ms) response, we consistently observed delayed EEG activity around 200 ms and between 400 and 750 ms after stimulation (Fig. 1B,C). To quantify total brain activation in response to TMS, we calculated the mean average global voltage, estimated as a surface area under the rectified EEG traces across all electrodes for each subject. While we observed no statistically significant difference between averaged global voltages in the 75–150-ms poststimulus interval (Fig. 1C), the average global voltage 400–750 ms poststimulus was significantly higher in schizophrenia when compared with healthy control subjects. Subtraction of Sham-EEG signal from active TMS-induced EEG responses did not diminish the observed difference statistically in delayed EEG responses as illustrated in Figure 1D. Of note, Sham-stimulation, which is delivered with a coil placed at 90° from the scalp, may provide less of a sensory stimulation than active TMS, given that tactile stimulation component of it is diminished. Figure 2 illustrates the topography of average voltage distribution in both healthy control and schizophrenia groups. This data suggests that schizophrenia subjects experience prolonged and wide spread cortical activation in response to TMS.
Given a wealth of evidence for spectral abnormalities in schizophrenia, we wanted to explore time and frequency domains of evoked TMS responses in both study groups. We were particularly interested to characterize the spatio-temporal properties of aberrant oscillatory activity evoked by TMS in schizophrenia subjects. For this we compared time–frequency signal power in delta (1–3.5 Hz), theta (4–7 Hz), alpha (8–12 Hz), beta (12–28 Hz), and gamma (30–50 Hz) ranges following TMS between 2 study groups. As shown in Figure 3a,b, significantly enhanced delta frequency oscillations were observed in the ipsilateral frontal and temporo-parietal leads and in bilateral occipital and parietal electrodes. High-frequency (beta–gamma) oscillations were also pronounced in ipsilateral leads proximal to stimulation and in homologous contralateral areas (motor cortex) of schizophrenia patients. Finally, high-frequency oscillations were recorded in the bilateral fronto-temporo-parietal leads of schizophrenia subjects.
A correlation between measure of SCZ symptoms and conductivity based upon PNASS (positive and negative scores) and maximum spectral power within each frequency bands at 200 ms and between 400 and 700 ms post-TMS were performed. Positive symptoms were positively correlated with gamma band (P < 0.0425) between 400 and 700 ms. Negative symptoms were positively correlated with theta (P < 0.0218) and delta bands (P < 0.0237) at 200-ms post-TMS. Regression plots between the correlation results are illustrated in Figure 4.
To our knowledge, this is the first demonstration of excessive spread of neuronal excitation induced by TMS in schizophrenia subjects. In this study, we provide evidence that, although TMS elicits the initial response of similar magnitude in both study groups, this response remains limited to the same area proximal to the stimulation site in normal control subjects, while in schizophrenia subjects it propagates both in time and in space. Enhanced TMS–EEG activity in schizophrenia group was observed post-stimulus, and this was associated with increased oscillatory activity primarily in gamma, beta, and delta frequency ranges in ipsilateral central leads proximal to stimulation, and in fronto-temporo-parietal leads bilaterally. These oscillatory abnormalities are suggestive of increased interhemispheric functional connectivity, which seems to be most pronounced between homologous (motor cortex) areas, but also extends toward temporo-parietal regions in schizophrenia brains, areas most implicated in schizophrenia pathology. Increased cortical conductivity in gamma-band frequency between 400 and 700 ms was positively correlated with positive schizophrenia symptoms (e.g., delusions and hallucinations), while negative symptoms were positively correlated with functional cortical conductivity in theta and delta bands at 200-ms post-TMS, suggesting that these indices may be functionally related.
Of interest, prior studies of TMS-evoked activity in schizophrenia brains analyzed EEG responses within 100-ms poststimulus window, reporting reduced evoked gamma oscillations (Ferrarelli et al. 2008) and significantly reduced field power (Ferrarelli et al. 2008; Levit-Binnun et al. 2010), particularly at time points corresponding with the peaks of 2 TMS-evoked gamma oscillations (Ferrarelli et al. 2008). This, along with our data showing increased global voltage and high-frequency oscillatory activity after 200-ms poststimulus, suggests that altered resonant properties of neuronal networks likely result in delayed, but prolonged activation (especially in high-frequency bands) in schizophrenia.
These observations suggest that pathophysiological changes characteristic for schizophrenia lead to fundamental difference in functional properties of cortical surface, which reflects the altered functional interactions between groups of neurons. Given that lateral cortical inhibition determines the extent of spread of activity along the cortical surface (Bandyopadhyay et al. 2005; Llinas et al. 2005), increased functional cortical conductivity may be a biophysical manifestation of inhibitory (and/or any other) deficits underlying impaired filtering in schizophrenia. The abnormal sensory filtering, or deficient gating, hypothesis is based on significantly reduced behavioral and electrophysiological indices of sensory gating observed in schizophrenic patients and in animal models of schizophrenia (Thaker 2008). An impaired sensory filtering in schizophrenia is thought to underlie an inability to selectively process relevant information, supposedly due to being flooded with unconstrained and irrelevant “perceptual noise” (Braff et al. 1992). Human and animal data indicate that similar information gating deficits may be a corollary to various molecular abnormalities, which include glutamatergic (Ma et al. 2009), cholinergic (Bickford and Wear 1995), dopaminergic (Bickford-Wimer et al. 1990; Hajos et al. 2005), and GABAergic (Hershman et al. 1995) mechanisms. Interestingly, amphetamine-induced thalamic auditory gating deficits in rats were linked to impairment of bursting activity of the inhibitory thalamic reticular neurons, deficits that were restored either by dopamine D2 antagonist haloperidol (Krause et al. 2003) or by acetylcholine receptor agonists (an agonist of α7 nicotinic acetylcholine receptor, genetically linked to schizophrenia), supposedly via restoration of GABAergic neurotransmission (Hajos et al. 2005). Likewise, GABAergic, dopaminergic, and cholinergic mechanisms were implicated in glutamatergic (ketamine-induced) impairment of auditory gating in rat hippocampus (Ma et al. 2009). These studies suggest that multiple molecular underpinnings of schizophrenia may manifest via a somewhat uniform phenomenon, such as impaired filtering. We speculate that increased functional cortical conductivity is an electrophysiological correlate of impaired ability to filter information in schizophrenic brains.
Could aberrant cortical conductivity with ensuing excessive spread of neuronal excitation provide any insight with regards to the mechanisms of schizophrenia symptoms? Or, in other words, what are the possible functional consequences to this “excitation leak”? Indirect answer to this question may be derived from in vivo studies utilizing intracortical injections of a GABAa antagonist bicuculline (Alloway and Burton 1991; Kyriazi et al. 1996; Wang et al. 2000, 2002), shown to enhance lateral spread of cortical activation (Contreras and Llinas 2001; Bandyopadhyay et al. 2005). Local microiontophoretic application of bicuculline has been shown to enlarge the size of receptive fields of neurons in somatosensory rat and visual primate cortices (Alloway and Burton 1991; Wang et al. 2000), and to change their spatial selectivity so that affected neurons become responsive to a broader range of stimuli (Kyriazi et al. 1996; Wang et al. 2002). Moreover, a subset of these neurons became responsive to stimuli that did not resemble the previously preferred inputs (Rao et al. 2000; Wang et al. 2002). These results parallel data obtained with EMG recordings from the rat motor cortex, demonstrating that following a local bicuculline injection stimulation of the motor cortex responsible for vibrissal movement activated a forelimb (Jacobs and Donoghue 1991). It has been concluded that intracortical blockade of GABAergic neurotransmission unmasks normally silent excitatory connections, resulting not only in loss of neuronal selectivity, but also making neurons respond to non-specific stimuli. In other words, some neurons disinhibited by bicuculline “hallucinate” by identifying a stimulus that has never been presented. We propose that similar mechanisms may be relevant to generating aberrant rhythmic activity responsible for perceptual abnormalities. In other words, the signal that “leaks” to the neighboring cortical area due to unfaithful filtering may excite “untuned” neurons, whose activation may be misinterpreted by the brain as, for example, a voice when no one speaks. Of interest, in our experiments, the TMS pulse delivered to the left motor cortex seems to induce fast oscillatory activity not only in homologous contralateral cortical area, but also in contralateral temporo-parietal regions, most implicated in schizophrenia pathology and, specifically, linked to auditory hallucinations. For example, several studies using functional neuroimaging techniques provide evidence for increased brain activity during auditory hallucinations in both, primary and secondary, associative auditory cortices (Dierks et al. 1999; Lennox et al. 2000; Bentaleb et al. 2002) of schizophrenia patients. In addition, increased excitatory activity in the right temporo-parietal cortical areas and increased interhemispheric coherence between auditory cortices have been implicated in the genesis of auditory hallucinations by EEG techniques (Line et al. 1998; Sritharan et al. 2005), strengthening a potential link between abnormal neuronal activation in speech-related areas and auditory hallucinations. Positive correlation between increased functional cortical conductivity in gamma-band frequency and positive symptoms, as well as between cortical conductivity in theta and delta bands and negative symptoms, demonstrated in our study seems to further suggest a link between schizophrenia symptoms and delayed cortical activation. Bilateral oscillatory activity in temporo-parietal areas in response to the left motor cortex, TMS shown in our study seems to be in line with evidence of increased functional correlation between these regions, thus providing a further support to pathologically increased synchrony between left and right temporo-parietal regions in schizophrenia. There is recent evidence relating locally increased white matter connectivity in temporo-parietal and other speech-related areas to auditory hallucinations (Hubl et al. 2004). The same study has also demonstrated that inner speech in patients with frequent hallucinations leads to abnormal coactivation in regions related to the acoustical processing of external stimuli. An interesting interpretation of these findings was proposed by Uhlhaas and Singer (2006), who hypothesize that “hyperconnectivity between higher- and lower-order cortical areas favors backpropagation to the respective primary auditory cortices of oscillatory activity generated in higher sensory areas during visual and auditory imagery, thus generating activation patterns that resemble those induced by sensory stimulation.” The “leaky cortex” may be an additional (or, in some cases, primary?) mechanism, underlying spontaneous ectopic activity in areas responsible for hallucinatory experiences. Given that “backpropagation” of cortical activation due to abnormal cortical conductivity is bidirectional, ectopic activation of secondary auditory cortices resulting from activation of lower-order auditory cortex is just as possible. A well-known clinical phenomenon of auditory hallucinations triggered by the humming sounds of appliances that typically start as indistinct noises and gradually progress through whisper-like sounds to distinctive voices (Hoffman et al. 2003; Perez Velazquez and Frantseva 2011) seems to be in line with this hypothesis. Another well-known clinical phenomenon is that some schizophrenia patients are able to keep their voices under control by listening to loud music (Sadock et al. 2009), an intervention that may hypothetically sharpen their auditory receptive fields via thalamic depolarization (Llinas et al. 2005), potentially limiting lateral spread of cortical excitation.
Our results seem to be in concordance with the interpretation of an EEG study (Breakspear et al. 2003) that analyzed the topographic organization of nonlinear interdependencies in patients with schizophrenia. According to this study, the rate of occurrence of dynamical interdependencies did not differ at any of the sites between subjects; however, the topography across the scalp was significantly different between schizophrenia and normal control groups: Nonlinear interdependences tended to occur in larger concurrent clusters across the scalp in schizophrenia than in healthy subjects. These findings were interpreted as they “do not support a simple ‘disconnection’ of cortical interactions, as implied by disconnection hypothesis of schizophrenia.” Disconnection hypothesis of schizophrenia postulates that multiple deficits in schizophrenia are brought about by the lack of structural and functional connectivity across multiple brain regions that result in abnormal functional integration of brain processes (Braff 1999). “Instead, they suggest a loss of fine-grained organization of cortical interactions and hence can be cautiously interpreted as evidence of impoverishment of flexibility across hierarchical brain regions in schizophrenia” (Breakspear 2006).
One of the major limitations of this study is the fact that 14 of the 16 schizophrenia patients were treated with antipsychotic medications. Although oscillatory abnormalities in medication-naive and first-episode patients with schizophrenia have been widely reported (Gallinat et al. 2004; Symond et al. 2005), there are also studies suggesting EEG alterations (e.g., general slowing) induced by antipsychotic medications (Koshino et al. 1993). In contrast, others report normalizing effects of neuroleptic medications in schizophrenia (Saletu et al. 1994; Canive et al. 1996). Importantly, deficits in cortical inhibition, a putative neurophysiologic mechanism of enhanced cortical conductivity, have been found to be more pronounced in unmedicated schizophrenia patients (Daskalakis et al. 2002). Nevertheless, future research confirming our preliminary findings in medication-free patients with schizophrenia is required.
Another important potential limitation of our study is suggested by recent publications linking EEG-recorded gamma-band rhythms with muscular activity (Whitham et al. 2007). However, increase in gamma frequency in schizophrenia subjects, reported in our study was calculated as EEG signal difference between 2 groups. To our knowledge, there is no evidence that schizophrenia patients have higher resting EEG activity. Therefore, it is likely that subtraction of EEG signal between 2 groups would eliminate the noise induced by muscle movement. Accordingly, increased gamma frequency observed in our experiments gravitates toward central rather than circumferential leads: No increase in high-frequency activity was detected in most lateral and posterior leads, as would be expected if they were a product of a muscle artifact (Whitham et al. 2007). As well, most recent study by Pope et al. (2009) that performed comparative analysis between paralyzed and nonparalyzed subjects within the group (e.g., before and after paralysis) demonstrates that evoked responses to cognitive tasks were qualitatively unaffected by muscle paralysis, although some gamma responses were still obscured by EMG. Further experiments comparing high-frequency TMS-induced activity in paralyzed subjects would be helpful to address this issue.
Another important avenue that needs to be further addressed is the specificity of abnormal cortical conductivity to schizophrenia (or psychotic illness). Of interest, TMS–EEG recordings were recently utilized in order to assess cortical excitability in epilepsy, another disorder characterized by altered excitation–inhibition balance. Delayed (>100 ms and <1 s after TMS stimulus) responses to TMS were observed in 3 of 15 epileptic patients. These responses frequently resembled the patient's epileptiform discharges and were elicited only when stimulating over epileptogenic, but not other cortical regions (Valentin et al. 2008). In addition, no evidence of increased excitability (measured as average global voltage after TMS) was obtained in a group of bipolar patients, and in patients suffering from major depressive or obsessive–compulsive disorders (manuscript in preparation). These findings suggest that abnormal cortical conductivity may be a phenomenonspecific for schizophrenia.
In summary, our study provides preliminary evidence for increased functional cortical conductivity in schizophrenia. We propose that increased cortical conductivity is a novel electrophysiological representation of a core information processing deficit in schizophrenia. We suggest that abnormal cortical conductivity in schizophrenic brains may favor propagation of oscillatory activity to the adjacent and remote cortical areas, thus generating areas of aberrant cortical activation. It is possible that these abnormalities may be the underlying responsible mechanisms mediating perceptual abnormalities characteristic for schizophrenia. In addition, they could potentially worsen (or even underlie) signal to noise deficits, jeopardizing cognitive functioning of affected individuals. Future studies linking impaired cortical conductivity to specific clinical signs and symptoms of schizophrenia, and to specific neurophysiologic deficits may open new possibilities for understanding the neural substrates of altered information processing in schizophrenia patients.
Z.J.D received external funding through Neuronetics and Brainsway Inc., Aspect Medical, and a travel allowance through Pfizer and Merck. Z.J.D has also received speaker funding through Sepracor Inc., and served on the advisory board for Hoffmann-La Roche Limited. This work was supported by the Ontario Mental Health Foundation (OMHF), the Canadian Institutes of Health Research (CIHR), the Brain and Behaviour Research Foundation, and the Grant Family and Temerty Family through the Centre for Addiction and Mental Health (CAMH) Foundation. M.F. was supported by internal funding at CAMH.
Conflict of Interest: None declared.