In science and engineering we use the concept of a black box to designate a device which we view in terms of its inputs and outputs without knowing anything in regards to its inner workings. For all practical purposes neuroscientists regard the brain as a black box whose mechanics are only loosely recognized.  Most of the information gathered about the brain comes from probing the same with electrodes; that is, plates of electrical conductors that are affixed to the gelatinous brain on one end and a voltmeter on the opposite side. Fluctuations of voltage over time are recorded by an instrument called the electroencephalogram or EEG.

Ever since my teenage years I have been fascinated by brain voltages.  My first crude attempt at building an EEG device came from a do-it-yourself electronic kit from Edmund Scientific.  Hacking the device gave me the opportunity to study different sub-bands of frequencies and perform some neurofeedback experiments on myself.  Back then, I was limited by all of the artifact I recorded.  Movement, eye blinks, sweating, and environmental stimulation all affected my recordings.

The cyclical repetitions of voltage waves represents its frequency (1Hz= 1 cycle per second).  Different bands of voltages are associated with different behavioral states.  The faster fequencies, for example, are associated with higher cognitive functions, among them executive functions (i.e., those cognitive functions that enable us to judge, plan and juggle multiple tasks).  In autism, the highest frequencies of the EEG (the gamma frequencies) are almost universally abnormal.  Gamma frequencies help to coordinate the actions of disparate brain regions.  When we see a flower or a face, the ability to recognize the whole or the gestalt, is due to the synchronization of gamma frequency across different brain areas.

Our group just published a study about gamma oscillations in autism which I find fascinating.  We used a new technique in analyzing these waveforms which suggests how gamma oscillations may be used as a biomarker for the condition, a possible outcome measure to gauge the effectiveness of interventions, and as an index of the excitatory/inhibitory imbalance of the cerebral cortex.

I am providing the first few paragraphs of the publication.  For those interested, the publication in its entirety can obtained through Researchgate (just click on the link provided).

Rhythmic patterns of neural activity, manifested in the electroencephalogram (EEG) asvoltage oscillations, have been linked to varied cognitive functions such as perception, attention,memory, and consciousness. The reciprocal interaction between excitation (pyramidal cells) and inhibition (interneurons) during cortical activation provides the genesis for brainwave oscillations [1].  Those brainwaves with the highest frequency, between 30 and 90 Hz, comprise the gamma bandwidth [1,2]. Fast-spiking interneurons that provide for the perisomatic inhibition of pyramidal cells, control the rhythm (clockwork) of these high frequency oscillations [1]. Immunocytochemical characterization of these cells reveals that they express the calcium-binding albumin protein parvalbumin (PV). The high metabolic activity of PV cells, which comprise the largest subgroup of cortical interneurons, makes them highly susceptible to oxidative injury. This pathoclisis helps explain their putative relationship to abnormalities of gamma aminobutyric acidergic (GABAergic) neurotransmission in many psychiatric disorders [3].

Reduced numbers of PV-expressing cells have been reported in human postmortem brain samples [4] and animal models of autism spectrum disorder (ASD) (e.g.,Fmr1, VPA ,Nlgn3,R451C,andCntnap2) [5]. More significantly, the reduced levels of PV expression correlate with ASD-like behavioral deficits (e.g., sociability, vocalization) and, curiously enough, with symptoms usually ascribed to ASD comorbidities (e.g., pain sensitivity, seizures) [6,7]. Long-lasting reversal of PV (GABAergic) deficits by pharmacologic or cell type-specific gene rescue, normalizes, or at least diminishes, cognitive dysfunction, and social deficits in these animal models [5,8,9]. It is therefore unsurprising that ASD researchers have proposed using gamma-band-based metrics, both a putative “electrophysiological endophenotype” [6] of PV pathology and a metric indicative of the cortical balance between inhibition and excitation [2], as an outcome measure for interventions aimed  at targeting the underlying pathology of ASD [10–12].

Gamma band activity is thought to reflect the mechanism for the integration of information in neural networks within and between brain regions (for reviews see [12,13]). Gamma rhythm is normally defined as EEG band in the frequency range between 30 to 90 Hz (or even higher), although there is an opinion [14] that different frequency sub-bands (e.g., 30–35 Hz, 40–48 Hz, etc.) may have distinct functional significance. Our study focuses on gamma sub-band within 35–45 Hz(so-called 40 Hz-centered gamma [15,16]). Oscillatory activity in the 40 Hz-centered gamma range has been related to Gestalt perception and to cognitive functions such as attention, learning,and memory [17,18]. Binding of widely distributed cell assemblies by synchronization of gamma frequency activity is thought to underlie cohesive stimulus representation in the brain [19,20]. It has been proposed that “weak central coherence” in autism could result from a reduction in the integration of specialized local networks in the brain caused by a deficit in temporal binding that depends on gamma synchronization [21–23]. It is important to emphasize that there are distinct functional differences between spontaneous gamma, evoked gamma power and coherence, and event-related induced gamma power and coherence [14]. Sensory evoked gamma coherences reflect the property of modality-specific networks activated by a sensory stimulation. Event-related (or cognitive) induced gamma and its coherences manifest coherent activity of sensory and cognitive networks triggered by and governed by requirements of a cognitive task. In autism, synchronization between these neural networks is abnormal and reflects an imbalance of the excitation/inhibition bias of the cerebral cortex (vide supra, [2]). Studies have shown that resting gamma power appears to be inversely correlated to ASD severity as measured by the Social Responsiveness Scale (SRS) [24].

Illusory contour or illusory figure (e.g., Kanizsa figure [25]) perception is a very useful model to study the integration of local image features into a coherent precept, and tests based on several illusory figures were productively used to investigate the impairment of such integration in children with ASD.Brown et al. [22] tested adolescents with autism in an experiment that presented Kanizsa shapes with visual illusions and reported excessive evoked gamma at 80 and 120 ms post-stimulus, in addition to enhanced induced gamma (200–400 ms). Inability to reduce gamma activity would lead to the inability to decide which event requires attention when there are multiple choices. In autism, uninhibited gamma activity suggests that none of the circuits in the brain can come to dominance because too many are active simultaneously [21–23]. Abnormalities of gamma synchrony can result in significant cognitive deficits, such as reduced attentional control, and other dysfunctions present in ASD. In addition, EEG recordings during a Kanizsa figure task have shown an overall increase in gamma oscillatory activity in ASD as compared to neurotypicals [22]. These findings are thought to reflect a reduction in the“signal to noise” level due to diminished inhibitory processing [22]. These observations are of clinical significance as several studies have now reported in ASD that abnormalities in gamma oscillations are normalized by low frequency repetitive transcranial stimulation (TMS). This neuromodulatory therapy also provides improvements in both repetitive behaviors and executive functions [10,26–29].

In a recent study comparing ASD and neurotypical controls, spectral analysis of the outer envelope joining the upper peaks of gamma oscillations allowed researchers to characterize the settling time after peak voltage amplitude [30]. At baseline, with no active treatment instituted, the latency of the ringing decay assessed using frequency analysis was significantly diminished in ASD as compared to control subjects. A short ringing time indicates a system whose efficiency of operation or sensitivity is diminished [31]. The oscillations induced by tasks involving the integration of features, as for example in a reaction time tasks using Kanizsa illusory features. Our group has used oddball task paradigms of target classification and discrimination which required a response to target Kanizsa squares among non-target Kanizsa triangles and other non-Kanizsa distractor figures in order to examine event-related potentials (ERP) and amplitude of gamma-band EEG activity [10,12,29]. We reported differences between neurotypical children and children with ASD diagnosis in reaction time and ERP measures as well as amplitude of gamma responses. Furthermore, we reported normalization of ERP responses and improved behavioral symptoms in children with ASD following repetitive transcranial magnetic stimulation (rTMS) treatment [28]. The current study was focused on more advanced analysis of evoked and induced gamma oscillations using the same illusory figure task. The gamma waveforms elicited by this task exhibit a characteristic dampening after peak amplitude in which the outer envelope of successive peaks traces a decay curve that persists until baseline.

The study used demodulation of gamma oscillations allowing to examine both the envelope ofa signal as well as the periodic waveform that carries the same suggesting that resonance behavior,exemplified in the carrier wave may tie neural populations operating at the same frequency. Analysis of the envelope of gamma oscillations can be used to investigate the impedance of involved circuits and the excitatory inhibitory balance of the cerebral cortex [30]. We propose that the metrics of gamma oscillations, ingrained in both its carrier and its envelope, may provide important information contributing to better understanding of functional significance of EEG gamma waveforms.

Despite all of the evidence, the utility of gamma-band related variables as diagnostic biomarkers is currently unexplored, suggesting an urgent need for using gamma oscillation measures as functional markers of response to interventions such as rTMS or other types of neuromodulation. This sensitivity is what allows a system to respond selectively to a given frequency while eliminating others.Brainwave oscillations are not a finely tuned process; amplitude, frequency and phase all vary across individual gamma cycles [32]. In autism, the low sensitivity makes the synchronization between neuronal networks imprecise.The end result is a distortion in the ability to form cohesive perceptual experiences and a reduction in the brain’s ability to provide for nuanced responses to both environmental and social exigencies.

The findings described in the previous paragraphs led the authors to study and compare, in ASD and a neurotypical control population, the metrics for gamma oscillations that describe its envelope.This study expands on previous findings by analyzing the evoked and induced components of gamma oscillations using wavelet transformation. Given the many reports in the literature which translate gamma oscillations to possible behavioral states we also analyzed for possible correlates to aberrant and/or repetitive behaviors in our ASD population post-TMS treatment.