Our desire and fascination with intelligent machines dates back to the antiquity's mythical automaton Talos, Aristotle's mode of mechanical thought (syllogism) and Heron of Alexandria's mechanical machines and automata. However, the quest for Artificial General Intelligence (AGI) is troubled with repeated failures of strategies and approaches throughout the history. This decade has seen a shift in interest towards bio-inspired software and hardware, with the assumption that such mimicry entails intelligence. Though these steps are fruitful in certain directions and have advanced automation, their singular design focus renders them highly inefficient in achieving AGI. Which set of requirements have to be met in the design of AGI? What are the limits in the design of the artificial? Here, a careful examination of computation in biological systems hints that evolutionary tinkering of contextual processing of information enabled by a hierarchical architecture is the key to build AGI.
Balance of excitation and inhibition is a fundamental feature of in vivonetwork activity and is important for its computations. However, its presence in the neocortex of higher mammals is not well established. We investigated the dynamics of excitation and inhibition using dense multielectrode recordings in humans and monkeys. We found that in all states of the wake-sleep cycle, excitatory and inhibitory ensembles are well balanced, and co-fluctuate with slight instantaneous deviations from perfect balance, mostly in slow-wave sleep. Remarkably, these correlated fluctuations are seen for many different temporal scales. The similarity of these computational features with a network model of self-generated balanced states suggests that such balanced activity is essentially generated by recurrent activity in the local network and is not due to external inputs. Finally, we find that this balance breaks down during seizures, where the temporal correlation of excitatory and inhibitory populations is disrupted. These results show that balanced activity is a feature of normal brain activity, and break down of the balance could be an important factor to define pathological states.
Beta (β)- and gamma (γ)-oscillations are present in different cortical areas and are thought to be inhibition-driven, but it is not known if these properties also apply to γ-oscillations in humans. Here, we analyze such oscillations in high-density microelectrode array recordings in human and monkey during the wake–sleep cycle. In these recordings, units were classified as excitatory and inhibitory cells. We find that γ-oscillations in human and β-oscillations in monkey are characterized by a strong implication of inhibitory neurons, both in terms of their firing rate and their phasic firing with the oscillation cycle. The β- and γ-waves systematically propagate across the array, with similar velocities, during both wake and sleep. However, only in slow-wave sleep (SWS) β- and γ-oscillations are associated with highly coherent and functional interactions across several millimeters of the neocortex. This interaction is specifically pronounced between inhibitory cells. These results suggest that inhibitory cells are dominantly involved in the genesis of β- and γ-oscillations, as well as in the organization of their large-scale coherence in the awake and sleeping brain. The highest oscillation coherence found during SWS suggests that fast oscillations implement a highly coherent reactivation of wake patterns that may support memory consolidation during SWS.
Self-organized critical states are found in many natural systems, from earthquakes to forest fires, they have also been observed in neural systems, particularly, in neuronal cultures. However, the presence of critical states in the awake brain remains controversial. Here, we compared avalanche analyses performed on different in vivo preparations during wakefulness, slow-wave sleep, and REM sleep, using high density electrode arrays in cat motor cortex (96 electrodes), monkey motor cortex and premotor cortex and human temporal cortex (96 electrodes) in epileptic patients. In neuronal avalanches defined from units (up to 160 single units), the size of avalanches never clearly scaled as power-law, but rather scaled exponentially or displayed intermediate scaling. We also analyzed the dynamics of local field potentials (LFPs) and in particular LFP negative peaks (nLFPs) among the different electrodes (up to 96 sites in temporal cortex or up to 128 sites in adjacent motor and premotor cortices). In this case, the avalanches defined from nLFPs displayed power-law scaling in double logarithmic representations, as reported previously in monkey. However, avalanche defined as positive LFP (pLFP) peaks, which are less directly related to neuronal firing, also displayed apparent power-law scaling. Closer examination of this scaling using the more reliable cumulative distribution function (CDF) and other rigorous statistical measures, did not confirm power-law scaling. The same pattern was seen for cats, monkey, and human, as well as for different brain states of wakefulness and sleep. We also tested other alternative distributions. Multiple exponential fitting yielded optimal fits of the avalanche dynamics with bi-exponential distributions. Collectively, these results show no clear evidence for power-law scaling or self-organized critical states in the awake and sleeping brain of mammals, from cat to man.
The effects of neural activity on cerebral hemodynamics underlie human brain imaging with functional magnetic resonance imaging and positron emission tomography. However, the threshold and characteristics of the converse effects, wherein the cerebral hemodynamic and metabolic milieu influence neural activity, remain unclear. We tested whether mild hypercapnia (5% CO2 ) decreases the magnetoencephalogram response to auditory pattern recognition and visual semantic tasks. Hypercapnia induced statistically significant decreases in event-related fields without affecting behavioral performance. Decreases were observed in early sensory components in both auditory and visual modalities as well as later cognitive components related to memory and language. Effects were distributed across cortical regions. Decreases were comparable in evoked versus spontaneous spectral power. Hypercapnia is commonly used with hemodynamic models to calibrate the blood oxygenation level-dependent response. Modifying model assumptions to incorporate the current findings produce a modest but measurable decrease in the estimated cerebral metabolic rate for oxygen change with activation. Because under normal conditions, low cerebral pH would arise when bloodflow is unable to keep pace with neuronal activity, the cortical depression observed here may reflect a homeostatic mechanism by which neuronal activity is adjusted to a level that can be sustained by available bloodflow. Animal studies suggest that these effects may be mediated by pH-modulating presynaptic adenosine receptors. Although the data is not clear, comparable changes in cortical pH to those induced here may occur during sleep apnea, sleep, and exercise. If so, these results suggest that such activities may in turn have generalized depressive effects on cortical activity.
Intracranial recording is an important diagnostic method routinely used in a number of neurological monitoring scenarios. In recent years, advancements in such recordings have been extended to include unit activity of an ensemble of neurons. However, a detailed functional characterization of excitatory and inhibitory cells has not been attempted in human neocortex, particularly during the sleep state. Here, we report that such feature discrimination is possible from high-density recordings in the neocortex by using 2D multielectrode arrays. Successful separation of regular-spiking neurons (or bursting cells) from fast-spiking cells resulted in well-defined clusters that each showed unique intrinsic firing properties. The high density of the array, which allowed recording from a large number of cells (up to 90), helped us to identify apparent monosynaptic connections, confirming the excitatory and inhibitory nature of regular-spiking and fast-spiking cells, thus categorized as putative pyramidal cells and interneurons, respectively. Finally, we investigated the dynamics of correlations within each class. A marked exponential decay with distance was observed in the case of excitatory but not for inhibitory cells. Although the amplitude of that decline depended on the timescale at which the correlations were computed, the spatial constant did not. Furthermore, this spatial constant is compatible with the typical size of human columnar organization. These findings provide a detailed characterization of neuronal activity, functional connectivity at the microcircuit level, and the interplay of excitation and inhibition in the human neocortex.
Sleep spindles are bursts of rhythmic 10-15 Hz activity, lasting ∼0.5-2 s, that occur during Stage 2 sleep. They are coherent across multiple cortical and thalamic locations in animals, and across scalp EEG sites in humans, suggesting simultaneous generation across the cortical mantle. However, reports of MEG spindles occurring without EEG spindles, and vice versa, are inconsistent with synchronous distributed generation. We objectively determined the frequency of MEG-only, EEG-only, and combined MEG-EEG spindles in high density recordings of natural sleep in humans. About 50% of MEG spindles occur without EEG spindles, but the converse is rare (∼15%). Compared to spindles that occur in MEG only, those that occur in both MEG and EEG have ∼1% more MEG coherence and ∼15% more MEG power, insufficient to account for the ∼55% increase in EEG power. However, these combined spindles involve ∼66% more MEG channels, especially over frontocentral cortex. Furthermore, when both MEG and EEG are involved in a given spindle, the MEG spindle begins ∼150 ms before the EEG spindle and ends ∼250 ms after. Our findings suggest that spindles begin in focal cortical locations which are better recorded with MEG gradiometers than referential EEG due to the biophysics of their propagation. For some spindles, only these regions remain active. For other spindles, these locations may recruit other areas over the next 200 ms, until a critical mass is achieved, including especially frontal cortex, resulting in activation of a diffuse and/or multifocal generator that is best recorded by referential EEG derivations due to their larger leadfields.
OBJECTIVE: Spindles are rhythmic bursts of 10-16 Hz activity, lasting ∼1 s, occur during normal stage 2 sleep. Spindles are slower in frontal EEG and possibly MEG. The posterior-fast EEG pattern may predominate early in the spindle, and the anterior-slow pattern late. We aimed to determine the proportion of spindles showing this spatio-spectro-temporal interaction for EEG, and whether it occurs in MEG.
METHODS: We recorded high density EEG and MEG from seven healthy subjects during normal stage 2 sleep. High vs. low frequency (12 vs. 14 Hz) power was measured early vs. late (25th-45th vs. 55th-75th duration percentile) in 183 spindle discharges.
RESULTS: The predicted spatio-spectro-temporal interaction was shown by 48% of EEG and 34% of MEG spindles (chance=25%). Topographically, high frequency EEG power was greatest at midline central contacts, and low frequency power at midline frontal. This frequency-specific topography was fixed over the course of the spindle.
CONCLUSIONS: An evolution from posterior-fast to anterior-slow generators commonly occurs during spindles, and this is visible with EEG and to a lesser extent, MEG.
SIGNIFICANCE: The spatio-spectral-temporal evolution of spindles may reflect their possible involvement in coordinating cortical activity during consolidation.
The resistive or non-resistive nature of the extracellular space in the brain is still debated, and is an important issue for correctly modeling extracellular potentials. Here, we first show theoretically that if the medium is resistive, the frequency scaling should be the same for electroencephalogram (EEG) and magnetoencephalogram (MEG) signals at low frequencies (<10 Hz). To test this prediction, we analyzed the spectrum of simultaneous EEG and MEG measurements in four human subjects. The frequency scaling of EEG displays coherent variations across the brain, in general between 1/f and 1/f(2), and tends to be smaller in parietal/temporal regions. In a given region, although the variability of the frequency scaling exponent was higher for MEG compared to EEG, both signals consistently scale with a different exponent. In some cases, the scaling was similar, but only when the signal-to-noise ratio of the MEG was low. Several methods of noise correction for environmental and instrumental noise were tested, and they all increased the difference between EEG and MEG scaling. In conclusion, there is a significant difference in frequency scaling between EEG and MEG, which can be explained if the extracellular medium (including other layers such as dura matter and skull) is globally non-resistive.
Forward solutions with different levels of complexity are employed for localization of current generators, which are responsible for the electric and magnetic fields measured from the human brain. The influence of brain anisotropy on the forward solution is poorly understood. The goal of this study is to validate an anisotropic model for the intracranial electric forward solution by comparing with the directly measured 'gold standard'. Dipolar sources are created at known locations in the brain and intracranial electroencephalogram (EEG) is recorded simultaneously. Isotropic models with increasing level of complexity are generated along with anisotropic models based on Diffusion tensor imaging (DTI). A Finite Element Method based forward solution is calculated and validated using the measured data. Major findings are (1) An anisotropic model with a linear scaling between the eigenvalues of the electrical conductivity tensor and water self-diffusion tensor in brain tissue is validated. The greatest improvement was obtained when the stimulation site is close to a region of high anisotropy. The model with a global anisotropic ratio of 10:1 between the eigenvalues (parallel: tangential to the fiber direction) has the worst performance of all the anisotropic models. (2) Inclusion of cerebrospinal fluid as well as brain anisotropy in the forward model is necessary for an accurate description of the electric field inside the skull. The results indicate that an anisotropic model based on the DTI can be constructed non-invasively and shows an improved performance when compared to the isotropic models for the calculation of the intracranial EEG forward solution.
Sleep spindles are approximately 1 s bursts of 10-16 Hz activity that occur during stage 2 sleep. Spindles are highly synchronous across the cortex and thalamus in animals, and across the scalp in humans, implying correspondingly widespread and synchronized cortical generators. However, prior studies have noted occasional dissociations of the magnetoencephalogram (MEG) from the EEG during spindles, although detailed studies of this phenomenon have been lacking. We systematically compared high-density MEG and EEG recordings during naturally occurring spindles in healthy humans. As expected, EEG was highly coherent across the scalp, with consistent topography across spindles. In contrast, the simultaneously recorded MEG was not synchronous, but varied strongly in amplitude and phase across locations and spindles. Overall, average coherence between pairs of EEG sensors was approximately 0.7, whereas MEG coherence was approximately 0.3 during spindles. Whereas 2 principle components explained approximately 50% of EEG spindle variance, >15 were required for MEG. Each PCA component for MEG typically involved several widely distributed locations, which were relatively coherent with each other. These results show that, in contrast to current models based on animal experiments, multiple asynchronous neural generators are active during normal human sleep spindles and are visible to MEG. It is possible that these multiple sources may overlap sufficiently in different EEG sensors to appear synchronous. Alternatively, EEG recordings may reflect diffusely distributed synchronous generators that are less visible to MEG. An intriguing possibility is that MEG preferentially records from the focal core thalamocortical system during spindles, and EEG from the distributed matrix system.
The electroencephalogram (EEG) is a mainstay of clinical neurology and is tightly correlated with brain function, but the specific currents generating human EEG elements remain poorly specified because of a lack of microphysiological recordings. The largest event in healthy human EEGs is the K-complex (KC), which occurs in slow-wave sleep. Here, we show that KCs are generated in widespread cortical areas by outward dendritic currents in the middle and upper cortical layers, accompanied by decreased broadband EEG power and decreased neuronal firing, which demonstrate a steep decline in network activity. Thus, KCs are isolated "down-states," a fundamental cortico-thalamic processing mode already characterized in animals. This correspondence is compatible with proposed contributions of the KC to sleep preservation and memory consolidation.