Same clock, different time read-out: Spontaneous brain oscillations and their relationship to deficient coding of cognitive content

Abstract

Neuronal oscillations provide an efficient means of communication, fostering functional neural states supporting action and reaction. High in the action hierarchy, cognitive abilities are severely compromised in neuropsychiatric disease such as schizophrenia. Current thinking highlights a clocking mechanism provided by the phase of an ongoing slow oscillation that offers a temporal frame for coding of perceptual and computational elements. Yet unclear is whether and how a dysregulated clocking mechanism accounts for diminished cognitive performance. Neuromagnetic oscillatory activity was related to cognitive performance assessed by the MATRICS Consensus Cognitive Battery in 58 healthy individuals (HC) and 46 schizophrenia patients (SZ). HC showed a correlation between gamma-band oscillations (> 40 Hz) and working memory performance. This relationship was disrupted in several ways in SZ. First, patients evidenced lower gamma power, poorer working memory performance, and no relationship between these measures. Second, the power spectra were dominated by ~ 10 Hz alpha oscillations with no group differences in amplitude. However, analysis of phase-to-amplitude coupling (PAC) revealed exaggerated clocking of gamma activity by alpha phase in SZ, associated with poor working memory performance. Third, despite entrainment by the same 10 Hz clock, gamma amplitude was abnormally distributed across the duty cycle in SZ, a potential consequence of compromised interneuron inhibition. Fourth, SZ showed over-engagement of a fronto-parietal network measured by gamma phase coherence, suggesting a brain state hindering cognitive output.

Such an endogenous temporal organization may be a core dysfunction in SZ: a segregation/integration input imbalance fostering reduced cognitive performance and compromised behavioral output.

Introduction

Timing is of the essence in brain function (Buzsaki, 2006). Accounts in the literature propose a clocking mechanism in which neuronal activity is coordinated by ongoing brain oscillations. The oscillations play an important modulatory role in information integration by providing spatio-temporal “opportunity” windows of reduced or enhanced excitability patterns (Haegens et al., 2011, Weisz et al., 2014). These patterns in turn interact with input to shape the coding, integration, and distribution of information throughout the brain (Buzsaki and Watson, 2012, He et al., 2010, Jensen et al., 2014). Ongoing brain oscillatory activity is no longer considered mere noise but a temporally coordinated structure that is preserved across species (Buzsaki et al., 2013) and substantially determines cognitive ability (Fries, 2005).

Coordination of neural assemblies is reflected in gamma (> 40 Hz) amplitude clocked by the rhythm (phase) of slower oscillations (Buzsaki and Wang, 2012, Canolty and Knight, 2010). This phenomenon is commonly termed cross-frequency phase-to-amplitude coupling (PAC). It is a mechanism creating “gamma packets” assembled in unison by a lower-frequency clock proposed to support efficient cognitive coding and information transmission in the mammalian brain (Buzsaki and Moser, 2013, Canolty et al., 2006, Jensen and Colgin, 2007). Conceivably, deficits in gamma activity may also manifest in dysfunctional clocking by slow oscillatory activity (Allen et al., 2011). Thus, coordination deficits in controlling coherent gamma activity can hamper network communication and result in deficient cognitive performance.

Studies convincingly demonstrate that brain oscillatory activity is critical for working memory performance (Jensen et al., 2002, Palva et al., 2010, Raghavachari et al., 2001) and is dysregulated in neuropsychiatric diseases such as schizophrenia (Haenschel et al., 2009, Kirihara et al., 2012, Uhlhaas et al., 2008). In addition to task related, spontaneous brain activity may contribute to cognitive performance. Brain signals recorded over several minutes to several hours show a characteristic 1/f function in the power spectrum (Buzsaki, 2006, He et al., 2010), i.e., decreasing power as a function of increasing frequency. This 1/f statistic is an indicator for the presence of a particular temporal dependence within the sampled signal, or nested oscillations (He et al., 2010). Slow once determine faster once and faster once regulate even faster once (Buzsaki, 2006). It is this fundamental and intrinsic temporal coordination that is hypothesized to be behind the aberrant coding of cognitive content in schizophrenia (Buzsaki and Watson, 2012, Roux and Uhlhaas, 2014, Uhlhaas and Singer, 2012). It is altered during both rest and various levels of working memory load suggesting a stable characteristic present across cognitive states (Repovs and Barch, 2012).

The present study tested the hypothesis that altered temporal organization is associated with a dysregulated clocking mechanism as a factor in compromised cognitive performance in schizophrenia. A sample of healthy comparison subjects (HC) served as a normative reference for behavioral and neurophysiological data. Analyses focused on ongoing oscillatory activity during an awake, eyes-open resting-state (RS) condition. First, frequency-domain analysis of magnetoencephalographic (MEG) recordings served to characterize the relationship of different frequencies to cognitive performance as measured by the MATRICS Consortium Cognitive Battery (MCCB) (Nuechterlein et al., 2008). Second, PAC served to determine temporal organization on a local scale and its relationship to cognitive test performance was examined. Finally, the magnitude of large-scale integration into a coherent brain network was evaluated in order to establish a more global description of the network state.