Background

Visuospatial working memory (WM), comprised by the coordination of processing, maintaining, manipulating, and retrieving spatial and visual information within the brain [1], has been implicated in a variety of tasks necessary for the navigation of day-to-day life, including wayfinding [2] and visuomotor learning [3, 4]. Prominent WM deficits have been demonstrated across multiple sensory modalities in people with schizophrenia (PSZ), including in the visuospatial realm ([5 reviewed in 6]). Furthermore, better visuospatial WM function in PSZ has been shown to predict better functional and social outcomes [7, 8]. Spatial WM deficits are likewise observed in unaffected first-degree relatives of PSZ [9–11], suggesting they may represent genetic liability for schizophrenia and thus constitute a potential endophenotype for the disorder [12–15].

Electroencephalography (EEG) has been widely used to investigate the neural underpinnings of WM processes, traditionally in the form of event-related potential (ERP) studies (reviewed in [16]). PSZ have been found to demonstrate deficits in ERPs elicited from WM tasks [17–19] and tasks probing sustained attention [20–24], among others; similar deficits have been observed in first-degree relatives of people with the disorder [22, 25, 26]. More recently, EEG research has examined oscillatory activity in the brain associated with WM. Oscillatory activity in the brain, which arises from the rhythmic activation of large numbers of synapses [27], is crucial to coordinated activity in the normally functioning brain [28]. PSZ demonstrate many abnormalities in oscillatory activity, which interfere with efficient integration of activity within the brain and may contribute to the prominent cognitive and functional deficits observed in the disorder [27–29].

Generally speaking, high frequency oscillations are associated with synchronization in local cortical networks, and lower frequency oscillations with synchronization between brain structures/across greater distances [28]. Oscillations in the gamma-band (25–100 Hz) and the theta-band (4–8 Hz) in particular have been widely implicated in episodic memory and WM alike (reviewed in [30]). Furthermore, the synchronized superimposition of gamma cycles on oscillatory theta activity has been proposed as a mechanism for the representation of multiple items in WM, a mechanism termed the “theta/gamma code” [31]. Multi-item WM has been associated with phase-phase coupling (see Fig. 3) between theta and gamma frequencies in healthy people (e.g., [32]). In addition, theta-gamma phase-amplitude coupling (Fig. 3) has been observed in the hippocampus [33, 34] as well as from scalp EEG [35, 36] in healthy humans and, in both cases, has been predictive of WM performance [33–36]. Given these findings, abnormal theta-gamma coupling has been proposed as a mechanism for impaired WM function in PSZ [37, 38], though this hypothesis has yet to be formally investigated.

Much research has already shown abnormalities in the gamma- and theta- bands in PSZ, suggesting dysfunction in neural integration on local and global scales. The present review examines gamma and theta abnormalities during WM in PSZ in the context of the theta/gamma neural code. Though the theta/gamma code has not been formally investigated in schizophrenia to date, the mechanism and hypothesized utility of the superimposition of gamma cycles on theta cycles in relation to WM in healthy people will be reviewed. Subsequently, this review will recount observed abnormalities in PSZ for activity in the gamma-band, theta-band, and interactions between the two during WM, discussing how such abnormalities may interfere with efficient theta/gamma coding in schizophrenia. Through integration of current findings regarding abnormalities in gamma- and theta-band oscillations in schizophrenia during WM, the present review will attempt to elucidate the potential implications of these abnormalities with respect to the theta/gamma neural code.

The superimposition of gamma-band oscillations on theta oscillations: the theta/gamma neural code

The theory of a theta/gamma code as first posited by Lisman and Idiart [39] and elaborated elsewhere (e.g., [40]) offered three primary, interrelated proposals: 1) that systematic coupling of oscillatory activity of different frequencies within the brain is meaningful, 2) that this coupling serves to form a temporal code in which distinct pieces of information are associated with simultaneous activity across several cells at particular phases within a cycle of theta, and 3) that the theta/gamma code model can explain central features of WM. In an effort to support their first two hypotheses regarding the importance of cross-frequency coupling as a temporal code, the authors first addressed the plausibility of oscillatory activity as a sustained temporal organizer. The neuromodulators acetylcholine and serotonin have been shown to be released in the brain during periods of oscillatory activity [41, 42]. In the presence of these neuromodulators, firing of neuronal cells induces a period of membrane afterdepolarization (ADP) rather than the typical afterhyperpolarization, leading to a transient increase in cell excitability [43–45]. Building off these findings, the authors used computer simulations to show that the duration of ADP was on the time scale of oscillations in the alpha-theta range (5 to 12 Hz), and that such ADP could be propagated for many cycles. Thus, a single excitatory input could lead to sustained firing on subsequent oscillatory cycles, potentially serving a storage function (Fig. 1a; [39]). Furthermore, the authors noted that in Sternberg’s [46] classic work on serial scanning in WM, the addition of each stimulus to the string of stimuli to be recalled resulted in an increase in reaction time of roughly 38 ms—an increase corresponding to the cycle of a neural oscillation in the beta-gamma range. Jensen and Lisman [47] elaborated on this observation, showing that the theta-gamma code model could effectively account for reaction time data and serial position effects reported by Sternberg [46, 48].Footnote 1 Roughly seven cycles in the beta-gamma range could be superimposed on a lower frequency cycle in the alpha-theta range—such as those cycles induced by the ADP—which Lisman and Idiart [39] emphasized corresponded to the canonical average capacity of WM determined by Miller [49]. Thus, Lisman and Idiart [39] demonstrated the potential for theta-gamma cross-frequency coupling to serve an ongoing, organizing function.

Fig. 1
figure 1

a The afterdepolarization (ADP) allows information storage in a single cell. The neuron receives a suprathreshold informational input and a second, subthreshold input that induces the membrane potential to oscillate at theta frequency (negative phase due to inhibition). Simulations show membrane potential before and after informational input (arrowhead). b Network in which pyramidal cells make converging excitatory synapses onto an inhibitory interneuron that produces feedback inhibition of pyramidal cells. c The network can maintain the firing and correct phase of seven groups of cells that are active during different subcycles of the low-frequency oscillation. Each trace illustrates the synchronous firing of a group of cells whose spatial pattern encodes the memory of a letter. The dashed lines during the second and fourth theta cycles show the different subcycles. The limited memory capacity of the network is demonstrated by its failure to store eight memories. Input of the memory X is successful (arrowhead), but R is lost. d If feedback inhibition is removed (arrowhead), the “40-hz” oscillation and phase information is rapidly lost. The two traces represent two of the seven memories stored in the network. A small phase difference (too small to be shown) persists for one cycle after removal of inhibition. Reproduced from [39]

Full size image

In their discussion of the temporal properties of theta and gamma oscillations in relation to Sternberg’s classic studies [46, 48], Lisman and Idiart [39] transition to their final proposal regarding the role of the theta/gamma code in WM. The authors demonstrated through simulation how differing, non-overlapping stimuli or “memories” ([39], p. 1515) could be stored through systematic variation in the phases of the different high-frequency subcycles, with each stimulus being represented by different groups of cells that maintain the stimulus by firing simultaneously at a particular high-frequency subcycle within the nesting lower frequency cycle (Fig. 1c). Thus, particular stimuli or memories are represented by the particular spatial pattern of a group of cells, or neural ensemble [37], these memories are propagated or maintained through interactions between oscillatory activity in the theta and gamma frequencies, and sequential information regarding these memories are linked to the particular gamma subcycle within the nesting theta cycle a given memory’s neural ensemble produces (or perhaps more accurately, the phase offset between the neural ensemble’s gamma cycle and the nesting theta cycle; [37, 39]). The theta-gamma neural code as proposed by Lisman and Idiart [39] represents a theoretical framework which, at its inception, integrated a number of experimental findings to provide a plausible model for the storage of multiple items in WM. This framework, which has since been further developed [e.g., 31, 37], has served as the basis for a substantial body of experimentation over the past twenty years, the vast majority of which has produced results supporting the Lisman and Idiart [39] model. A brief review of this sizeable body of literature follows, focused on studies probing WM function.

Empirical support for the theta-gamma neural code

The experimental support for Lisman and Idiart’s [39] model of the theta-gamma neural code is now widespread. In addition to other supporting observations, Lisman and Idiart [39] cited as some of the most significant early support for the theta-gamma neural code model the work of O’Keefe and Recce [50] (later clarified by [51]). These researchers examined pyramidal cells from the hippocampus of rats known as place cells, which fire bursts of spikes as the animal moves through a particular area in its environment termed the place field of that cell [50]. Concurrent to these bursts, theta band activity also occurs in the hippocampal EEG as a rat relocates itself in its environment. Though phase correlations between theta-band activity and bursts of hippocampal activity had then been well-established [52–55], the nature of the relationship was not well clarified. Using a combination of single unit and EEG recordings from hippocampal CA1 and CA3 place cells, O’Keefe and Recce [50] showed that for a single run through the field of a given place cell on a linear track, the cell fired bursts of spikes at progressively earlier parts of the of the concomitant theta cycle, a phenomenon termed “phase precession” ([51], p. 149; Fig. 2). This finding was taken to indicate that the hippocampus may use a neural code in which theta phase conveys important information [31]—namely, the theta-gamma neural code that Lisman and Idiart [39] proposed. Skaggs and colleagues [51] replicated and elaborated this finding to demonstrate that the phase precession effect was indeed robust throughout the rat hippocampus, and observed as animals moved through two-dimensional in addition to one-dimensional space. More significantly, the researchers emphasize the support their findings lend to the notion that the theta phase-offset at which a place cell fires carries important spatial information [51]. A considerable body of work has shown that this spiking in the hippocampus and entorhinal cortex is time locked to local gamma-activity, such that firing is effectively limited to discrete, gamma-locked time windows (see [56] and [57]). Bieri, Bobbitt, and Colgin [58] similarly observed theta-gamma phase precession in rat CA1 hippocampal cells and suggested differential relationships between coupling between high- and low-gamma signals. Phase precession has likewise been observed in entorhinal grid cells [59], supporting the organization of multi-item WM by way of theta- and gamma- cycles as by Lisman and Idiart’s [39] model and its elaborations. Phase-precession is further reviewed in [56].

Much subsequent experimentation has added to the support for Lisman and Idiart’s [39] model, demonstrating systematic relationships between theta- and gamma- oscillatory activity. These relationships are most frequently presented in one of two forms: phase-phase coupling (also called phase-synchrony, or simply phase-coupling), in which a given phase angle of a cycle of ongoing theta-activity tends to occur at a given phase angle of concurrent gamma-activity; and phase-amplitude coupling, in which the amplitude of gamma activity is modulated by the phase of concurrent theta-oscillations (Fig. 3). Though these types of coupling are most frequently reported separately, it has been suggested that they are likely two expressions of a single phenomenon [60, 61].

Fig. 2
figure 2

Extraction of the firing phase shift for each spike during a single run through the field of a CA3 place cell on the linear track. a Each firing of cell 3 as a single vertical line during the 1 s of data. b Phase of each spike relative to the theta cycle within which it falls as a horizontal line. c Hippocampal theta activity recorded at the same time as the single unit. d Result of the theta template matching algorithm. Note that the amplitude and the time between onsets of each template match varies to fit the variations in the theta. The small vertical ticks above the electroencephalogram (EEG) and below the template fits mark the beginning of each theta cycle. The cell clearly fires six bursts of spikes during this run through the place field. Comparison of each burst with the concomitant theta wave shows that each successive burst fires on an earlier part of the theta. This is shown clearly by the descending staircase of the phase correlates in B. EEG voltage in C is + 1 to - 1 mV. Total time between marks on the x axis is 1 s. Reproduced from [50]

Full size image
Fig. 3
figure 3

Schematic overview of two cases in which oscillatory activity of a higher-frequency oscillation may be related to the phase of a lower-frequency oscillation. Dark and light boxes separate consecutive cycles of the lower-frequency oscillation. The lower-frequency oscillation of fluctuating amplitude (top) shows phase-amplitude coupling with a higher-frequency oscillation of fluctuating amplitude (middle). In this example, amplitudes of the higher-frequency oscillation are maximal during the up-phase of the lower-frequency oscillation. A higher-frequency oscillation of stable amplitude (bottom), shows phase-phase coupling with the low-frequency oscillation. Here, peaks of the higher-frequency oscillation always coincide with the same phase values of the lower-frequency oscillations. Reproduced from [79]

Full size image

Findings of both types of coupling are summarized in Tables 1 and 2. Review of the animal studies in Table 1 reveals well-documented increases of theta-gamma coupling using intracranial recordings from the entorhinal-hippocampal system of rodents during performance of memory-focused tasks (e.g., Table 1, references i-xi). Theta-gamma coupling is likewise observed using intracranial recordings from the entorhinal cortex and hippocampus during free exploration, sleep and anesthetization (e.g., xii-xxvi) as well as from in vitro isolated samples (e.g., xxvii-xxix) in rats and mice. Primate studies have also revealed theta-gamma coupling in auditory cortex (xxx) during a passive listening task as well as from anterior cingulate cortex during an attention task (xxxi). Thus, WM processes are associated with increased theta-gamma coupling in the animal literature, though coupling observed in less-memory-dependent paradigms indicates that theta-gamma coupling may represent a more widespread means of communication in the brains of rodents and primates.

Table 1 Animal findings regarding theta-gamma coupling (TGC)
Full size table

Table 2 reports findings of theta-gamma coupling in humans. Here, we see intracranial recordings utilized to demonstrate phase-phase coupling (e.g., Table 2, references i, vii) as well as phase-amplitude coupling (e.g., ii-vi) between theta- and gamma- activity in hippocampal areas as well as throughout the cortex. Furthermore, scalp-level recordings have shown phase-phase theta-gamma coupling in humans over primarily frontal, parietal and occipital areas (e.g., viii-xi, xv), as well as phase-amplitude coupling over much of the scalp, including prefrontal, occipital and parietal sites (e.g. xi, xiii, xiv). Notably, one recent experiment temporarily increased individuals’ short-term memory capacities by using transcranial alternating current stimulation to slow their theta cycles (xvi), while another used MEG to show progressively later peaks in gamma activity relative to concurrent theta activity for progressively later memory items in an encoding sequence (xvii). In addition, measures of theta-gamma coupling have similarly been related to WM performance in experimentation with humans (e.g., vii, xi, xiii, xv). Thus, theta-gamma coupling has been repeatedly associated with WM processes in both the animal and human literature.

Table 2 Human findings regarding theta-gamma coupling (TGC)
Full size table

The role of abnormal theta-/gamma- oscillations in schizophrenia

Given the well-documented importance of theta-gamma modulations to memory processes in animal and human studies alike, abnormal interactions between theta- and gamma-oscillations have been proposed to explain the prominent WM deficits observed in PSZ [37, 38]. In particular, predictive relationships demonstrated between indices of theta-gamma coupling and WM capacity [32, 34] as well as performance on WM tasks [36, 107] in healthy people suggest that impaired WM performance observed in PSZ may be attributed to deficient theta-gamma coupling. Formal investigation of this proposal has been scarce to date. As such, in addition to the few direct examinations of theta-gamma coupling in PSZ, demonstrated abnormalities in theta- and gamma- oscillations independently are likewise reported, and their potential relevance to interactions between the two frequency bands is discussed.

Theta-gamma coupling in people with schizophrenia

Few studies have directly examined interactions between theta- and gamma- activity in PSZ. Allen and colleagues [62] examined phase-amplitude coupling between low- and high- frequencies using independent component analysis (ICA) of EEG recorded from PSZ during performance of an auditory oddball task. In this ICA, data for comodulation between various frequencies were decomposed into separate components based on spectral and spatial composition. One such component showing widespread coupling between lower-frequency phases and higher-frequency amplitudes across the entire scalp was less prominent in PSZ data compared to that of healthy controls, and another component depicting modulation of high frequency amplitude by low frequency phase at frontotemporal sites was more prominent for PSZ than controls. Another component revealed strong theta-gamma coupling over occipital-parietal areas, but the loading parameters for said component did not differ between groups (e.g., the component was equivalently strong in PSZ and healthy controls). First-degree relatives of people with the disorder were included in this study and showed coupling intermediate to that of PSZ and healthy controls, though their data were not tested statistically due to power limitations. The authors interpret their results to indicate that abnormal cross-frequency coupling may represent an endophenotype for schizophrenia, particularly in light of genetic associations they report for genotypes for certain genes and loading parameters for various independent components in PSZ. However, though patterns of cross-frequency coupling were altered in PSZ, theta-gamma coupling is still prominent, especially over posterior sites. As such, theta-gamma coupling may not be significantly altered in the disorder, at least in the context of novelty detection. No correlations between measures of cross-frequency coupling and performance were observed [62]. Thus, this study provides evidence for preserved theta-gamma phase-amplitude coupling in PSZ; however, given the minimal role WM plays in the oddball paradigm, the theta-gamma interactions examined in this study may be distinct from those related to WM.

The only other published work examining theta-gamma cross-frequency interactions in PSZ comes from Kirihara and colleagues [63]. Using scalp EEG recorded during presentation of 40-hz steady state auditory stimuli, the authors found reduced intertrial phase coherence, increased theta amplitude, and undisturbed theta-gamma phase-amplitude coupling in PSZ relative to healthy controls. The authors interpret these findings to indicate a preserved hierarchical organization of theta and gamma activity in PSZ despite abnormalities in theta- and gamma- activity independently [63]. Notably, the simple auditory processing task used in this study does not tap memory processes, for which theta-gamma interactions have been shown to essential and greatly impaired in PSZ; thus, the reported intact theta-gamma coupling should not be interpreted as indicative of preserved coupling at large.

Two additional studies investigated theta-gamma coupling under influence of ketamine, an NMDAR antagonist which is used as a pharmacological model of schizophrenia in both animals and humans. Neymotin and colleagues [64] used a computer model of hippocampal CA3 to investigate the effect of the application of simulated ketamine. Ketamine administration disrupted theta-gamma coupling in the simulation, apparently due to its antagonistic effect on NMDA receptors on oriens-lacunosum moleculare cells (OLM); normal coupling was restored via injection of a continuous, depolarizing current into these OLM cells, suggesting a potential intervention for treating deficits in WM and other cognitive processes in PSZ. Similarly, Caixeta and colleagues [65] found that ketamine altered theta-gamma coupling in the hippocampus of rats, again suggesting the importance of theta-gamma interactions to cognitive deficits in PSZ. Still, theta-gamma coupling in the context of memory function in PSZ has not yet been directly examined.

Theta- and gamma- band abnormalities in people with schizophrenia

Though research on theta-gamma coupling in schizophrenia is sparse, findings from the individual bands support the notion of WM impairments in PSZ stemming from aberrant theta-gamma interactions. Abnormalities in theta- and gamma-activity during WM reported in PSZ are summarized in Table 3. These abnormalities include deficits in theta-synchrony (e.g., Table 3, reference iv) and aberrations in overall theta-amplitudes (e.g., i-iv) as well as modulations in response to task demands (e.g., i, iv), though results are variable regarding whether PSZ show enhanced or reduced amplitudes/modulation as compared with controls. Despite this variability, all studies show a failure of PSZ to modulate theta amplitudes with changes in task demands compared to controls, suggesting an inefficient allocation of cognitive resources (suggested by i, iii, iv) and/or deficiencies in neural responses associated with sequential presentation of stimuli (suggested by ii). Amplitudes of gamma-activity during WM have similarly been found to be abnormal in PSZ (e.g., v-ix), though findings are again mixed as to whether these abnormalities are enhancements or reductions compared to controls; as with theta, however, PSZ frequently fail to modulate (e.g., vi, viii-ix) or abnormally modulate (vii) gamma responses with changes in task demands. Again, these findings suggest that PSZ either lack or misallocate cognitive resources as compared to healthy controls. Discrepancies in whether oscillatory responses are increased or decreased in PSZ relative to controls have many potential sources, including heterogeneity in symptomatology, differences in stage of illness (e.g., early onset and/or first-episode patients), electrode site selection (e.g., single channel versus pooled sites) or means of collection (e.g., MEG vs. EEG). The literature nonetheless demonstrates consistent reports of abnormalities in theta- and gamma-band activity for PSZ as compared to healthy controls during WM processes, particularly in response to changes in task conditions related to WM function.

Table 3 Theta and gamma abnormalities in people with schizophrenia (PSZ) during working memory processes
Full size table

Though the present review focuses on the relation of theta- and gamma-oscillations to WM dysfunction in PSZ, the role of oscillatory activity in other frequencies and in relation to other cognitive functions has been reviewed elsewhere [28, 38, 66, 67]. Thus, abnormalities in oscillatory activity in PSZ do not appear to be limited to theta- and gamma-band activity, and may contribute to a wide variety of symptomatology observed in the disorder.

Relevance of abnormalities in theta- and gamma- activity in people with schizophrenia

PSZ demonstrate abnormalities in theta- and gamma- band activity during WM paradigms. Though no studies of WM in PSZ have examined coupling between theta- and gamma- activities, the ramifications of deficiencies within each of these bands independently on their interaction are easily ascertained. Inconsistencies in phase, as indexed by abnormal phase synchrony measures, would significantly interfere with any meaningful coordination between signals of differing frequencies. Similarly, amplitude deficiencies suggest abnormal generation of neural oscillations, be it through activation of fewer cells or impaired coordination of cellular assemblies that would function more cohesively in the healthy brain. Regardless, less reliable generators of oscillatory activity will limit the unitary functioning of cell assemblies that is necessary for coordination of signals that compose the theta-gamma neural code. Although coupling between theta- and gamma- activity in PSZ has been investigated, the few examinations of interactions between these bands in PSZ have not involved WM directly. Though theta-gamma coupling has been observed during processes other than WM in healthy people, the bulk of the literature ties it to memory processes; thus, further investigation is needed to clarify whether theta-gamma interactions are indeed abnormal during WM processes in PSZ, and whether those abnormalities explain behavioral performance deficits associated with the disorder.

Conclusions: the theta-gamma neural code and visuospatial WM in schizophrenia

Lisman and Idiart’s [39] model positing a theta/gamma neural code as a means for representing multiple items in WM has generated a significant body of literature, and has been implicated in an even greater collection of research. A considerable amount of evidence has been accumulated and overwhelmingly supports the notions advanced by the model, namely that the interaction between theta- and gamma- oscillations communicate meaningful information within the brain in terms of memory and potentially other cognitive processes. Recent experimentation with humans has supported the model more directly: gamma activity associated with particular sequentially presented stimuli has been found to peak at different phases of concurrent theta activity in accordance with stimulus position within the sequence, and electrical stimulation in the gamma frequency range has shown to improve WM performance when administered at particular phases of theta. Regardless of how generalized the theta/gamma code is within the brain, it is difficult to dispute the notion that theta-gamma interactions are central features of WM processes in humans and animals. Furthermore, measures of the strength of coupling between the two bands have repeatedly shown to predict WM performance, supporting the notion that theta-/gamma- interaction and memory are causally linked.

Despite this sizeable body of evidence supporting the importance of theta-gamma coupling to WM processes in humans and animals, few studies have examined interactions between the bands in PSZ [62, 63], and none in the context of WM. As such, much additional experimentation is required to determine whether deficient theta-gamma coupling may explain WM deficits in PSZ. Notably, though several studies have demonstrated predictive relationships between measures of coupling and WM ability in healthy people, few have compared the predictive ability of theta-gamma coupling to other indices of oscillatory activity, including measures of power and/or phase synchrony within theta-, gamma-, and other frequency bands. One notable exception is [32], who found an index of alpha-band power predicted WM capacity better than indices of theta-gamma coupling; the researchers interpreted this alpha measure as an index of the suppression of irrelevant information, a mechanism arguably distinct from that being presently examined. Regardless, the logical question is thus whether deficient theta-gamma coupling truly explains WM deficiencies, or if it is merely an important correlate of some other primary cause. Addressing such a question is an important next step in further asserting the causal role of theta-gamma coupling in WM processes as well as the veracity of Lisman and Idiart’s [39] theta/gamma neural code.Footnote 2

Despite the unanswered questions that remain, considerable evidence has been found to support the theory of a theta/gamma code. Much of this evidence is derived from examinations of spatial WM processes in both rodents and humans. Given the well documented deficits in visuospatial WM in PSZ [6] and their relatives [9–11], a clear and testable hypothesis is that abnormal interactions between theta- and gamma-activity explains these deficits in visuospatial WM and WM at large. The hypothesis is supported by the considerable body of literature demonstrating abnormalities in theta- and gamma-bands individually within PSZ, including in connection with WM. Examination of theta-gamma coupling during WM in PSZ may yield a better understanding of mechanisms underlying the prominent WM deficits observed in people with the disorder. Lisman and Idiart’s [39] theta-gamma neural code provides a framework for testing a specific source of WM dysfunction in PSZ. Using the Lisman and Idiart [39] model as a guide, we might expect to find generally reduced phase-phase coupling between theta- and gamma-activity during performance of tasks probing WM in PSZ. Theta-gamma coupling in healthy people has been observed prominently over prefrontal and posterior regions in scalp level recordings [32, 35, 36, 68–70]. Thus, deficits in synchrony between the two bands observed over anterior and posterior regions in scalp recordings, as well as in theta-gamma synchrony between the two areas, would particularly support the notion that deficits in the temporal alignment of theta- and gamma- activities are inextricably linked to WM function in the disorder. Furthermore, such synchrony deficits would mirror reports of inefficiencies in prefrontal areas [71] as well as deficient functional connectivity between prefrontal and posterior areas associated with WM processes observed in PSZ using functional MRI [72, 73], and may represent the same dysfunction. Theta-gamma interactions may likewise be disrupted in the hippocampus in PSZ, given the predominance of reported theta-gamma coupling during WM processes in animals and healthy people (e.g., Table 1) and observations that structural abnormalities in the hippocampus are prominent in schizophrenia [74] and observed even in cases of early-onset [75]. As activity from the hippocampus is difficult to observe using scalp level recordings (see [76]), intracranial EEG would be necessary to investigate hippocampal theta-gamma coupling deficits in PSZ; however, were such studies conducted, the Lisman and Idiart [39] model in combination with previous findings suggest abnormalities in theta-gamma coupling would be likely.

Furthermore, investigations of theta-gamma abnormalities during spatial WM processing in PSZ would be more informative if they included a task with a sequential presentation of stimuli. The Lisman and Idiart [39] model predicts different theta-phase preferences for different stimuli, represented by individual gamma cycles superimposed upon a theta cycle, in a memory sequence. The discussed phenomenon of phase-precession observed in rodents [50, 51, 56, 58, 59] represents one of the primary pieces of evidence supporting the theory and contributing to its continued consideration, and recent work has shown progressive shifts in gamma activity based on stimulus sequence [77]. However, paradigms that would allow for investigation of differential phase-preferences for individual stimuli in a spatial memory sequence are decidedly absent from human study. Thus, utilization of a paradigm featuring a sequential presentation of visuospatial stimuli would allow for a direct examination of whether phase-preferences are present in humans, and/or disrupted in PSZ.

In light of the dearth of literature testing the role of theta-gamma coupling in WM deficits in PSZ and their relatives, further study into these interactions between theta- and gamma-activity associated with WM in people genetically liable for schizophrenia is strongly warranted. Lisman and Idiart’s [39] model provides a concrete framework for the design of an ideal investigation of these processes in PSZ, and suggests that deficits in theta-gamma interactions are likely to be observed in such experimentation. Better understanding of how neural oscillations factor into visuospatial WM deficits in the disorder could help inform interventions for improving memory performance, particularly in light of continually developing techniques of neuromodulation that have been shown to improve cognitive function using oscillatory stimulation (e.g., [78]).