Abstract perceptual choice signals during action-linked decisions in the human brain.

Humans can make abstract choices independent of motor actions. However, in laboratory tasks, choices are typically reported with an associated action. Consequentially, knowledge about the neural representation of abstract choices is sparse, and choices are often thought to evolve as motor intentions. Here, we show that in the human brain, perceptual choices are represented in an abstract, motor-independent manner, even when they are directly linked to an action. We measured MEG signals while participants made choices with known or unknown motor response mapping. Using multivariate decoding, we quantified stimulus, perceptual choice, and motor response information with distinct cortical distributions. Choice representations were invariant to whether the response mapping was known during stimulus presentation, and they occupied a distinct representational space from motor signals. As expected from an internal decision variable, they were informed by the stimuli, and their strength predicted decision confidence and accuracy. Our results demonstrate abstract neural choice signals that generalize to action-linked decisions, suggesting a general role of an abstract choice stage in human decision-making.

Neural representations of the content and production of human vocalization.

Speech, as the spoken form of language, is fundamental for human communication. The phenomenon of covert inner speech implies functional independence of speech content and motor production. However, it remains unclear how a flexible mapping between speech content and production is achieved on the neural level. To address this, we recorded magnetoencephalography in humans performing a rule-based vocalization task. On each trial, vocalization content (one of two vowels) and production form (overt or covert) were instructed independently. Using multivariate pattern analysis, we found robust neural information about vocalization content and production, mostly originating from speech areas of the left hemisphere. Production signals dynamically transformed upon presentation of the content cue, whereas content signals remained largely stable throughout the trial. In sum, our results show dissociable neural representations of vocalization content and production in the human brain and provide insights into the neural dynamics underlying human vocalization.

Spectral Fingerprints of Cortical Neuromodulation.

Pupil size has been established as a versatile marker of noradrenergic and cholinergic neuromodulation, which has profound effects on neuronal processing, cognition, and behavior. However, little is known about the cortical control and effects of pupil-linked neuromodulation. Here, we show that pupil dynamics are tightly coupled to temporally, spectrally, and spatially specific modulations of local and large-scale cortical population activity in the human brain. We quantified the dynamics of band-limited cortical population activity in resting human subjects using magnetoencephalography and investigated how neural dynamics were linked to simultaneously recorded pupil dynamics. Our results show that pupil-linked neuromodulation does not merely affect cortical population activity in a stereotypical fashion. Instead, we identified three frontal, precentral, and occipitoparietal networks, in which local population activity with distinct spectral profiles in the theta, beta, and alpha bands temporally preceded and followed changes in pupil size. Furthermore, we found that amplitude coupling at ∼16 Hz in a large-scale frontoparietal network predicted pupil dynamics. Our results unravel network-specific spectral fingerprints of cortical neuromodulation in the human brain that likely reflect both the causes and effects of neuromodulation.SIGNIFICANCE STATEMENT Brain function is constantly affected by modulatory neurotransmitters. Pupil size has been established as a versatile marker of noradrenergic and cholinergic neuromodulation. However, because the cortical correlates of pupil dynamics are largely unknown, fundamental questions remain unresolved. Which cortical networks control pupil-linked neuromodulation? Does neuromodulation affect cortical activity in a stereotypical or region-specific fashion? To address this, we quantified the dynamics of cortical population activity in human subjects using magnetoencephalography. We found that pupil dynamics are coupled to highly specific modulations of local and large-scale cortical activity in the human brain. We identified four cortical networks with distinct spectral profiles that temporally predicted and followed pupil size dynamics. These effects likely reflect both the cortical control and effect of neuromodulation.

Testing the generalization of neural representations.

Multivariate analysis methods are widely used in neuroscience to investigate the presence and structure of neural representations. Representational similarities across time or contexts are often investigated using pattern generalization, e.g. by training and testing multivariate decoders in different contexts, or by comparable pattern-based encoding methods. It is however unclear what conclusions can be validly drawn on the underlying neural representations when significant pattern generalization is found in mass signals such as LFP, EEG, MEG, or fMRI. Using simulations, we show how signal mixing and dependencies between measurements can drive significant pattern generalization even though the true underlying representations are orthogonal. We suggest that, using an accurate estimate of the expected pattern generalization given identical representations, it is nonetheless possible to test meaningful hypotheses about the generalization of neural representations. We offer such an estimate of the expected magnitude of pattern generalization and demonstrate how this measure can be used to assess the similarity and differences of neural representations across time and contexts.

Multistage classification identifies altered cortical phase- and amplitude-coupling in Multiple Sclerosis.

Distinguishing groups of subjects or experimental conditions in a high-dimensional feature space is a common goal in modern neuroimaging studies. Successful classification depends on the selection of relevant features as not every neuronal signal component or parameter is informative about the research question at hand. Here, we developed a novel unsupervised multistage analysis approach that combines dimensionality reduction, bootstrap aggregating and multivariate classification to select relevant neuronal features. We tested the approach by identifying changes of brain-wide electrophysiological coupling in Multiple Sclerosis. Multiple Sclerosis is a demyelinating disease of the central nervous system that can result in cognitive decline and physical disability. However, related changes in large-scale brain interactions remain poorly understood and corresponding non-invasive biomarkers are sparse. We thus compared brain-wide phase- and amplitude-coupling of frequency specific neuronal activity in relapsing-remitting Multiple Sclerosis patients (n = 17) and healthy controls (n = 17) using magnetoencephalography. Changes in this dataset included both, increased and decreased phase- and amplitude-coupling in wide-spread, bilateral neuronal networks across a broad range of frequencies. These changes allowed to successfully classify patients and controls with an accuracy of 84%. Furthermore, classification confidence predicted behavioral scores of disease severity. In sum, our results unravel systematic changes of large-scale phase- and amplitude coupling in Multiple Sclerosis. Furthermore, our results establish a new analysis approach to efficiently contrast high-dimensional neuroimaging data between experimental groups or conditions.

Motion Coherence and Luminance Contrast Interact in Driving Visual Gamma-Band Activity.

Synchronized neuronal population activity in the gamma-frequency range (>30 Hz) correlates with the bottom-up drive of various visual features. It has been hypothesized that gamma-band synchronization enhances the gain of neuronal representations, yet evidence remains sparse. We tested a critical prediction of the gain hypothesis, which is that features that drive synchronized gamma-band activity interact super-linearly. To test this prediction, we employed whole-head magnetencephalography in human subjects and investigated if the strength of visual motion (motion coherence) and luminance contrast interact in driving gamma-band activity in visual cortex. We found that gamma-band activity (64-128 Hz) monotonically increased with coherence and contrast, while lower frequency activity (8-32 Hz) decreased with both features. Furthermore, as predicted for a gain mechanism, we found a multiplicative interaction between motion coherence and contrast in their joint drive of gamma-band activity. The lower frequency activity did not show such an interaction. Our findings provide evidence that gamma-band activity acts as a cortical gain mechanism that nonlinearly combines the bottom-up drive of different visual features.

Interobserver reliability is higher for assessments with 3D software-generated models than with conventional MRI images in the classification of trochlear dysplasia.

PURPOSE: Trochlear dysplasia is a significant risk factor for patellofemoral instability. The severity of trochlear dysplasia is commonly evaluated based on the Dejour classification in axial MRI slices. However, this often leads to heterogeneous assessments. A software to generate MRI-based 3D models of the knee was developed to ensure more standardized visualization of knee structures. The purpose of this study was to assess the intra- and interobserver agreements of 2D axial MRI slices and an MRI-based 3D software generated model in classification of trochlear dysplasia as described by Dejour. METHODS: Four investigators independently assessed 38 axial MRI scans for trochlear dysplasia. Analysis was made according to Dejour's 4 grade classification as well as differentiating between 2 grades: low-grade (types A + B) and high-grade trochlear dysplasia (types C + D). Assessments were repeated following a one-week interval. The inter- and intraobserver agreement was determined using Cohen's kappa (κ) and Fleiss kappa statistic (κ). In addition, the proportion of observed agreement (po) was calculated for assessment of intraobserver agreement. RESULTS: The assessment of the intraobserver reliability with regard to the Dejour-classification showed moderate agreement values both in the 2D (κ = 0.59 ± 0.08 SD) and in the 3D analysis (κ = 0.57 ± 0.08 SD). Considering the 2-grade classification, the 2D (κ = 0.62 ± 0.12 SD) and 3D analysis (κ = 0.61 ± 0.19 SD) each showed good intraobserver matches. The analysis of the interobserver reliability also showed moderate agreement values with differences in the subgroups (2D vs. 3D). The 2D evaluation showed correspondences of κ = 0.48 (Dejour) and κ = 0.46 (high / low). In the assessment based on the 3D models, correspondence values of κ = 0.53 (Dejour) and κ = 0.59 (high / low) were documented. CONCLUSION: Overall, moderate-to-good agreement values were found in all groups. The analysis of the intraobserver reliability showed no relevant differences between 2 and 3D representation, but better agreement values were found in the 2-degree classification. In the analysis of interobserver reliability, better agreement values were found in the 3D compared to the 2D representation. The clinical relevance of this study lies in the superiority of the 3D representation in the assessment of trochlear dysplasia, which is relevant for future analytical procedures as well as surgical planning. LEVEL OF EVIDENCE: Level II.

Measuring standing hindfoot alignment: reliability of different approaches in conventional x-ray and cone-beam CT.

INTRODUCTION: Currently there is no consensus how hindfoot alignment (HA) should be assessed in CBCT scans. The aim of this study is to investigate how the reliability is affected by the anatomical structures chosen for the measurement. MATERIALS AND METHODS: Datasets consisting of a Saltzman View (SV) and a CBCT of the same foot were acquired prospectively and independently assessed by five raters regarding HA. In SVs the HA was estimated as follows: transversal shift between tibial shaft axis and heel contact point (1); angle between tibial shaft axis and a tangent at the medial (2) or lateral (3) calcaneal wall. In CBCT the HA was estimated as follows: transversal shift between the centre of the talus and the heel contact point (4); angle between a perpendicular line and a tangent at the medial (5) or lateral (6) calcaneal wall; angle between the distal tibial surface and a tangent at the medial calcaneal wall (7). Intraclass correlation coefficients (ICC) were calculated to assess inter-rater reliability. A linear regression was performed to compare the different measurement regarding their correlation. RESULTS: 32 patients were included in the study. The ICCs for the measurements 1-7 were as follows: (1) 0.924 [95% CI 0.876-0.959] (2) 0.533 [95% CI 0.377-0.692], (3) 0.553 [95% CI 0.399-0.708], (4) 0.930 [95% CI 0.866-0.962], (5) 0.00 [95% CI - 0.111 to 0.096], (6) 0.00 [95% CI - 0.103 to 0.111], (7) 0.152 [95% CI 0.027-0.330]. A linear regression between measurement 1 and 4 showed a correlation of 0.272 (p = 0.036). CONCLUSIONS: It could be shown that reliability of measuring HA depends on the investigated anatomical structure. Placing a tangent along the calcaneus (2, 3, 5, 6, 7) was shown to be unreliable, whereas determining the weight-bearing heel point (1, 4) appeared to be a reliable approach. The correlation of the measurement workflows is significant (p = 0.036), but too weak (0.272) to be used clinically.

Cortical correlation structure of aperiodic neuronal population activity.

Electrophysiological population signals contain oscillatory and non-oscillatory aperiodic (1/frequency-like) components. So far research has largely focused on oscillatory activity, and only recently, interest in aperiodic population activity has gained momentum. Accordingly, while the cortical correlation structure of oscillatory population activity has been characterized, little is known about the correlation of aperiodic neuronal activity. To address this, we investigated aperiodic neuronal population activity in the human brain using resting-state magnetoencephalography (MEG). We combined source-analysis, signal orthogonalization and irregular-resampling auto-spectral analysis (IRASA) to systematically characterize the cortical distribution and correlation of aperiodic neuronal activity. We found that aperiodic population activity is robustly correlated across the cortex and that this correlation is spatially well structured. Furthermore, we found that the cortical correlation structure of aperiodic activity is similar but distinct from the correlation structure of oscillatory neuronal activity. Anterior cortical regions showed the strongest differences between oscillatory and aperiodic correlation patterns. Our results suggest that correlations of aperiodic population activity serve as robust markers of cortical network interactions. Furthermore, our results show that aperiodic and oscillatory signal components provide non-redundant information about large-scale neuronal correlations. This may reflect at least partly distinct neuronal mechanisms underlying and reflected by oscillatory and aperiodic neuronal population activity.

Dissociating harmonic and non-harmonic phase-amplitude coupling in the human brain.

Phase-amplitude coupling (PAC) has been hypothesized to coordinate cross-frequency interactions of neuronal activity in the brain. However, little is known about the distribution of PAC across the human brain and the frequencies involved. Furthermore, it remains unclear to what extent PAC may reflect spurious cross-frequency coupling induced by physiological artifacts or rhythmic non-sinusoidal signals with higher harmonics. Here, we combined MEG, source-reconstruction and different measures of cross-frequency coupling to systematically characterize local PAC across the resting human brain. We show that cross-frequency measures of phase-amplitude, phase-phase, and amplitude-amplitude coupling are all sensitive to signals with higher harmonics. In conjunction, these measures allow to distinguish harmonic and non-harmonic PAC. Based on these insights, we found no evidence for non-harmonic local PAC in resting-state MEG. Instead, we found cortically and spectrally wide-spread PAC driven by harmonic signals. Furthermore, we show how physiological artifacts and spectral leakage cause spurious PAC across wide frequency ranges. Our results clarify how different measures of cross-frequency interactions can be combined to characterize PAC, and cast doubt on the presence of prominent non-harmonic phase-amplitude coupling in human resting-state MEG.

Task-evoked activity quenches neural correlations and variability across cortical areas.

Many large-scale functional connectivity studies have emphasized the importance of communication through increased inter-region correlations during task states. In contrast, local circuit studies have demonstrated that task states primarily reduce correlations among pairs of neurons, likely enhancing their information coding by suppressing shared spontaneous activity. Here we sought to adjudicate between these conflicting perspectives, assessing whether co-active brain regions during task states tend to increase or decrease their correlations. We found that variability and correlations primarily decrease across a variety of cortical regions in two highly distinct data sets: non-human primate spiking data and human functional magnetic resonance imaging data. Moreover, this observed variability and correlation reduction was accompanied by an overall increase in dimensionality (reflecting less information redundancy) during task states, suggesting that decreased correlations increased information coding capacity. We further found in both spiking and neural mass computational models that task-evoked activity increased the stability around a stable attractor, globally quenching neural variability and correlations. Together, our results provide an integrative mechanistic account that encompasses measures of large-scale neural activity, variability, and correlations during resting and task states.

Gradual progression from sensory to task-related processing in cerebral cortex.

Somewhere along the cortical hierarchy, behaviorally relevant information is distilled from raw sensory inputs. We examined how this transformation progresses along multiple levels of the hierarchy by comparing neural representations in visual, temporal, parietal, and frontal cortices in monkeys categorizing across three visual domains (shape, motion direction, and color). Representations in visual areas middle temporal (MT) and V4 were tightly linked to external sensory inputs. In contrast, lateral prefrontal cortex (PFC) largely represented the abstracted behavioral relevance of stimuli (task rule, motion category, and color category). Intermediate-level areas, including posterior inferotemporal (PIT), lateral intraparietal (LIP), and frontal eye fields (FEF), exhibited mixed representations. While the distribution of sensory information across areas aligned well with classical functional divisions (MT carried stronger motion information, and V4 and PIT carried stronger color and shape information), categorical abstraction did not, suggesting these areas may participate in different networks for stimulus-driven and cognitive functions. Paralleling these representational differences, the dimensionality of neural population activity decreased progressively from sensory to intermediate to frontal cortex. This shows how raw sensory representations are transformed into behaviorally relevant abstractions and suggests that the dimensionality of neural activity in higher cortical regions may be specific to their current task.

Dissociated neuronal phase- and amplitude-coupling patterns in the human brain.

Coupling of neuronal oscillations may reflect and facilitate the communication between neuronal populations. Two primary neuronal coupling modes have been described: phase-coupling and amplitude-coupling. Theoretically, both coupling modes are independent, but so far, their neuronal relationship remains unclear. Here, we combined MEG, source-reconstruction and simulations to systematically compare cortical amplitude-coupling and phase-coupling patterns in the human brain. Importantly, we took into account a critical bias of amplitude-coupling measures due to phase-coupling. We found differences between both coupling modes across a broad frequency range and most of the cortex. Furthermore, by combining empirical measurements and simulations we ruled out that these results were caused by methodological biases, but instead reflected genuine neuronal amplitude coupling. Our results show that cortical phase- and amplitude-coupling patterns are non-redundant, which may reflect at least partly distinct neuronal mechanisms. Furthermore, our findings highlight and clarify the compound nature of amplitude coupling measures.

Sensory processing and categorization in cortical and deep neural networks.

Many recent advances in artificial intelligence (AI) are rooted in visual neuroscience. However, ideas from more complicated paradigms like decision-making are less used. Although automated decision-making systems are ubiquitous (driverless cars, pilot support systems, medical diagnosis algorithms etc.), achieving human-level performance in decision making tasks is still a challenge. At the same time, these tasks that are hard for AI are easy for humans. Thus, understanding human brain dynamics during these decision-making tasks and modeling them using deep neural networks could improve AI performance. Here we modelled some of the complex neural interactions during a sensorimotor decision making task. We investigated how brain dynamics flexibly represented and distinguished between sensory processing and categorization in two sensory domains: motion direction and color. We used two different approaches for understanding neural representations. We compared brain responses to 1) the geometry of a sensory or category domain (domain selectivity) and 2) predictions from deep neural networks (computation selectivity). Both approaches gave us similar results. This confirmed the validity of our analyses. Using the first approach, we found that neural representations changed depending on context. We then trained deep recurrent neural networks to perform the same tasks as the animals. Using the second approach, we found that computations in different brain areas also changed flexibly depending on context. Color computations appeared to rely more on sensory processing, while motion computations more on abstract categories. Overall, our results shed light to the biological basis of categorization and differences in selectivity and computations in different brain areas. They also suggest a way for studying sensory and categorical representations in the brain: compare brain responses to both a behavioral model and a deep neural network and test if they give similar results.

Analyzing EEG and MEG signals recorded during tES, a reply.

Transcranial Electric Stimulation (tES) is a widely used non-invasive brain stimulation technique. However, strong stimulation artifacts complicate the investigation of neural activity with EEG or MEG during tES. Thus, studying brain signals during tES requires detailed knowledge about the properties of these artifacts. Recently, we characterized the phase- and amplitude-relationship between tES stimulation currents and tES artifacts in EEG and MEG and provided a mathematical model of these artifacts (Noury and Siegel, 2017, and Noury et al., 2016, respectively). Among several other features, we showed that, independent of the stimulation current, the amplitude of tES artifacts is modulated time locked to heartbeat and respiration. In response to our work, a recent paper (Neuling et al., 2017) raised several points concerning the employed stimulation device and methodology. Here, we discuss these points, explain potential misunderstandings, and show that none of the raised concerns are applicable to our results. Furthermore, we explain in detail the physics underlying tES artifacts, and discuss several approaches how to study brain function during tES in the presence of residual artifacts.

Phase properties of transcranial electrical stimulation artifacts in electrophysiological recordings.

Monitoring brain activity during transcranial electric stimulation (tES) is an attractive approach for causally studying healthy and diseased brain activity. Yet, stimulation artifacts complicate electrophysiological recordings during tES. Design and evaluation of artifact removal methods require a through characterization of artifact features, i.e. characterization of the transfer function that defines the relationship between the tES stimulation current and tES artifacts. Here we characterize the phase relationship between stimulation current and tES artifacts in EEG and MEG. We show that stimulation artifacts are not pure in-phase or anti-phase signals, but that non-linear mechanisms induce steady phase deflections relative to the stimulation current. Furthermore, phase deflections of stimulation artifacts are slightly modulated by each heartbeat and respiration. For commonly used stimulation amplitudes, artifact phase deflections correspond to signals several times bigger than normal brain signal. Moreover, the strength of phase deflections varies with stimulation frequency. These phase effects should be accounted for during artifact removal and when comparing recordings with different stimulation frequencies. We summarize our findings in a mathematical model of tES artifacts and discuss how this model can be used in simulations to design and evaluate artifact rejection techniques. To facilitate this research, all raw data of this study is made freely available.

Gamma-band activity reflects attentional guidance by facial expression.

Facial expressions attract attention due to their motivational significance. Previous work focused on attentional biases towards threat-related, fearful faces, although healthy participants tend to avoid mild threat. Growing evidence suggests that neuronal gamma (>30Hz) and alpha-band activity (8-12Hz) play an important role in attentional selection, but it is unknown if such oscillatory activity is involved in the guidance of attention through facial expressions. Thus, in this magnetoencephalography (MEG) study we investigated whether attention is shifted towards or away from fearful faces and characterized the underlying neuronal activity in these frequency ranges in forty-four healthy volunteers. We employed a covert spatial attention task using neutral and fearful faces as task-irrelevant distractors and emotionally neutral Gabor patches as targets. Participants had to indicate the tilt direction of the target. Analysis of the neuronal data was restricted to the responses to target Gabor patches. We performed statistical analysis at the sensor level and used subsequent source reconstruction to localize the observed effects. Spatially selective attention effects in the alpha and gamma band were revealed in parieto-occipital regions. We observed an attentional cost of processing the face distractors, as reflected in lower task performance on targets with short stimulus onset asynchrony (SOA <150ms) between faces and targets. On the neuronal level, attentional orienting to face distractors led to enhanced gamma band activity in bilateral occipital and parietal regions, when fearful faces were presented in the same hemifield as targets, but only in short SOA trials. Our findings provide evidence that both top-down and bottom-up attentional biases are reflected in parieto-occipital gamma-band activity.

Spectral fingerprints of large-scale neuronal interactions.

Cognition results from interactions among functionally specialized but widely distributed brain regions; however, neuroscience has so far largely focused on characterizing the function of individual brain regions and neurons therein. Here we discuss recent studies that have instead investigated the interactions between brain regions during cognitive processes by assessing correlations between neuronal oscillations in different regions of the primate cerebral cortex. These studies have opened a new window onto the large-scale circuit mechanisms underlying sensorimotor decision-making and top-down attention. We propose that frequency-specific neuronal correlations in large-scale cortical networks may be 'fingerprints' of canonical neuronal computations underlying cognitive processes.

Physiological processes non-linearly affect electrophysiological recordings during transcranial electric stimulation.

Transcranial electric stimulation (tES) is a promising tool to non-invasively manipulate neuronal activity in the human brain. Several studies have shown behavioral effects of tES, but stimulation artifacts complicate the simultaneous investigation of neural activity with EEG or MEG. Here, we first show for EEG and MEG, that contrary to previous assumptions, artifacts do not simply reflect stimulation currents, but that heartbeat and respiration non-linearly modulate stimulation artifacts. These modulations occur irrespective of the stimulation frequency, i.e. during both transcranial alternating and direct current stimulations (tACS and tDCS). Second, we show that, although at first sight previously employed artifact rejection methods may seem to remove artifacts, data are still contaminated by non-linear stimulation artifacts. Because of their complex nature and dependence on the subjects' physiological state, these artifacts are prone to be mistaken as neural entrainment. In sum, our results uncover non-linear tES artifacts, show that current techniques fail to fully remove them, and pave the way for new artifact rejection methods.

Measuring the cortical correlation structure of spontaneous oscillatory activity with EEG and MEG.

Power correlations of orthogonalized signals have recently been introduced for MEG as a powerful tool to non-invasively investigate functional connectivity in the human brain. Little is known about the applicability of this approach to EEG, and how compatible the results are between EEG and MEG. To address this, we systematically compared power correlations of simultaneously recorded and source co-registered 64-channel EEG and 275-channel MEG in resting human subjects. For both modalities, connectivity peaked at around 16 Hz. For this frequency range, seed-based correlation maps showed comparable patterns across modalities, with generally more distinct patterns for MEG. A brain-wide pattern correlation analysis also revealed maximum similarity around 16 Hz. Correcting for different signal-to-noise ratio (SNR) across frequencies and modalities revealed pattern correlation between modalities close to one across a broad frequency range from 1 to 32 Hz and only slightly smaller for higher frequencies. The decrease above 32 Hz likely reflected higher susceptibility to muscle artifacts for EEG than for MEG. Our results show that power correlation of orthogonalized signals is feasible for studying functional connectivity with 64-channel EEG. Furthermore, besides differences in SNR, for frequencies from about 8 to 32 Hz, EEG and MEG measure the same correlation patterns across the entire brain.