When using structural MRI information in MEG modeling and to constrain the inverse solutions, an accurate mapping between the coordinate systems of the two modalities is required. Co-registering a high-field MR image with MEG data involves many manual steps including a number of potential error sources. Moreover, geometrical distortions in high-field MR images, related for instance to magnetic susceptibility variations, worsen the local co-registration accuracy. With a hybrid MEG–MRI device using ULF-MRI-tailored sensors for both modalities, the co-registration problem can turned into an automatic calibration step. After this, the MEG and MRI coordinates will be the same, i.e., positions of the MRI voxels are known exactly in the laboratory frame with respect to the MEG sensor helmet. The key component in our calibration method are the sensitivity profiles of the superconducting pickup loops, which in ULF MRI are independent of the sample and therefore well-defined. In the most basic form, the spatial information of the profiles, captured in parallel ULF MR acquisitions, is made consistent with profiles computed in the sensor coordinate system by finding the exact coordinate transformation between the modalities. Furthermore, the transformation does not have to be affine as usual but, for example, second-order corrections can be included. A nonlinear mapping can be used to detect and correct for remaining minor image distortions due field imperfections such as concomitant gradient field components. Thus, after the calibration, distortion-free MRI and a high spatial accuracy for MEG source localization can be achieved. The spatial calibration method was assessed by simulations assuming known geometry for the sensor array. It was analyzed how the calibration method performs in different conditions, including nonlinear image distortions and high noise levels. With cubic voxels of size 4×4×4mm³, sub-millimeter calibration accuracy was achieved even in prominently high noise conditions.

Transfer entropy has been used to quantify the directed flow of information between source and target variables in many complex systems. While transfer entropy was originally formulated in discrete time, in this paper we provide a framework for considering transfer entropy in continuous time systems, based on Radon-Nikodym derivatives between measures of complete path realizations. To describe the information dynamics of individual path realizations, we introduce the pathwise transfer entropy, the expectation of which is the transfer entropy accumulated over a finite time interval. We demonstrate that this formalism permits an instantaneous transfer entropy rate. These properties are analogous to the behavior of physical quantities defined along paths such as work and heat. We use this approach to produce an explicit form for the transfer entropy for pure jump processes, and highlight the simplified form in the specific case of point processes (frequently used in neuroscience to model neural spike trains). Finally, we present two synthetic spiking neuron model examples to exhibit the pertinent features of our formalism, namely, that the information flow for point processes consists of discontinuous jump contributions (at spikes in the target) interrupting a continuously varying contribution (relating to waiting times between target spikes). Numerical schemes based on our formalism promise significant benefits over existing strategies based on discrete time formalisms. "Transfer entropy in continuous time, with applications to jump and neural spiking processes" Richard E. Spinney, Mikhail Prokopenko, and Joseph T. Lizier Phys. Rev. E 95, 032319 doi:10.1103/PhysRevE.95.032319

Brain decoding is a popular multivariate approach for hypothesis testing in neuroimaging. Linear classifiers are widely employed in the brain decoding paradigm to discriminate among experimental conditions. Then, the derived linear weights are visualized in the form of multivariate brain maps to further study spatio-temporal patterns of underlying neural activities. It is well known that the brain maps derived from weights of linear classifiers are hard to interpret because of high correlations between predictors, low signal-to-noise ratios, and the high dimensionality of neuroimaging data. Therefore, improving the interpretability of brain decoding approaches is of primary interest in many neuroimaging studies. Despite extensive studies of this type, at present, there is no formal definition for interpretability of multivariate brain maps. As a consequence, there is no quantitative measure for evaluating the interpretability of different brain decoding methods. In this study, first, we present a theoretical definition of interpretability in brain decoding; we show that the interpretability of multivariate brain maps can be decomposed into their reproducibility and representativeness. Second, as an application of the proposed definition, we exemplify a heuristic for approximating the interpretability in multivariate analysis of evoked magnetoencephalography (MEG) responses. Third, we propose to combine the approximated interpretability and the generalization performance of the brain decoding into a new multi-objective criterion for model selection. Our results, for the simulated and real MEG data, show that optimizing the hyper-parameters of the regularized linear classifier based on the proposed criterion results in more informative multivariate brain maps. More importantly, the presented definition provides the theoretical background for quantitative evaluation of interpretability, and hence, facilitates the development of more effective brain decoding algorithms in the future.

There is reason to believe that there is an underlying structure to the way the lexicon in our semantic memory is organized. In order to uncover this structure, competing theories can be tested through a metric of semantic distance. For example, in a structure based on word associations, words that quickly come to mind given a cue word (DOG→BARK) are semantically closer than words that do not (DOG→FERN). To evaluate a theory, the semantic distances predicted by the proposed structure can be compared with for example the N400 component of the event-related potential (ERP), as measured through electroencephalography (EEG). Recently, advances in multivariate processing have boosted the signal-to-noise ratio (SNR) of EEG, diminishing the need for averaging and hence increasing the number of data points produced in a single study. This enables an exciting paradigm shift in how theories are evaluated with the help of electrophysiological data. A researcher may now approach the data analysis in an “unsupervised” manner, i.e. instead of labeling the data with experimental conditions, we can now generate enough data points so that underlying patterns may be deduced. In the present study, we demonstrate an unsupervised approach that can be used to aid our understanding of the structure of our mental lexicon. To this end, we performed unsupervised hierarchical clustering of the following 14 words: GIRAFFE CHAIR CLOSET ZEBRA LION COUCH RHINO HIPPO TIGER DESK TABLE ELEPHANT DOOR BED, using the amplitude of the N400 as the distance metric. All possible combinations of 2 written words were presented in sequential fashion to 16 participants, who indicated by button press whether they thought the words were related for any reason. The EEG recording during the presentation of the second word was analyzed using a variation of beamformer filtering to estimate the amplitude of the N400 with a sufficient SNR. The filter combined a template of the shape of the ERP component of interest with the covariance matrix of the current recording to estimate the amplitude of the component in each trial. To obtain an unbiased template of the N400, we used a previous recording on 10 participants that did not participate in the current experiment. Those participants were presented with 800 word-pairs with varying forward association strength, in a manner identical to the current experiment. The topmost two clusters of the dendrogram consisted of all animate versus all inanimate objects (within-cluster versus between-cluster N400 amplitudes: p = 0.021), as expected. This verifies the approach, which can now be used in situations where the semantic structure is disputed.

Brain decoding provides a possible tool for multivariate hypothesis testing on neuroimaging data that generally is higher in sensitivity and specificity than its univariate counterpart. Additionally, the learned parameters of brain decoding models can be visualized in the form of brain maps which can aid in understanding how brain activity in space and time underlies a cognitive function. However, the recovered brain maps suffer from lack of interpretability, especially in group analysis of multi-subject data. To overcome this limitation, we present an application of multi-task joint feature learning for group-level multivariate pattern recovery in single-trial magnetoencephalography (MEG) decoding. The proposed method allows for recovering sparse yet consistent patterns across different subjects, and therefore enhances the interpretability of the decoding model. We evaluated the performance of the multi-task joint feature learning in terms of generalization, reproducibility, and quality of pattern recovery against traditional single-subject and pooling approaches on both simulated and real MEG datasets. Our experimental results demonstrate that the mutli-task joint feature learning framework is capable of recovering meaningful patterns of varying spatio-temporally distributed brain activity across individuals while still maintaining excellent generalization performance. Our findings facilitate the application of brain decoding for characterizing the fine-level distinctive patterns of brain activities in group-level inference on neuroimaging data.

Previous behavioural studies have shown that humans act more altruistically towards kin. Whether and how knowledge of genetic relatedness translates into differential neurocognitive evaluation of observed social interactions has remained an open question. Here, we investigated how the human brain is engaged when viewing a moral dilemma between genetic vs. non-genetic sisters. During functional magnetic resonance imaging, a movie was shown, depicting refusal of organ donation between two sisters, with participants guided to believe the sisters were related either genetically or by adoption at young age. 90% of the participants self-reported that genetic relationship was not relevant to them, yet their brain activity told a different story. When the participants believed that the sisters were genetically related, inter-subject similarity of brain activity was significantly stronger in insula, cingulate, medial and lateral prefrontal, superior temporal, and superior parietal cortices. Notably, there were differences neither in eye-movements, nor in experienced emotions, nor in physiological arousal, which could have explained the differences in brain activity. Cognitive functions that these areas have been previously observed to support include moral and emotional conflict regulation, decision making, mentalizing, and perspective taking, suggesting more similar engagement of such functions when observing refusal of altruism from genetic sister. Further, these areas overlapped with those activated in a separate moral dilemma simulation where the subjects had to choose between saving their sister, best friend, or strangers, from danger. Our results show that mere knowledge of a genetic relationship between interacting persons robustly modulates social cognition of the perceiver.

Perineuronal nets (PNN) ensheath GABAergic and glutamatergic synapses on neuronal cell surface in the central nervous system (CNS), have neuroprotective effect in animal models of Alzheimer disease and regulate synaptic plasticity during development and regeneration. Crucial insights were obtained recently concerning molecular composition and physiological importance of PNN but the microstructure of the network remains largely unstudied. Here we used histochemistry, fluorescent microscopy and quantitative image analysis to study the PNN structure in adult mouse and rat neurons from layers IV and VI of the somatosensory cortex. Vast majority of meshes have quadrangle, pentagon or hexagon shape with mean mesh area of 1.29 mm2 in mouse and 1.44 mm2 in rat neurons. We demonstrate two distinct patterns of chondroitin sulfate distribution within a single mesh – with uniform (nonpolar) and nodeenriched (polar) distribution of the Wisteria floribunda agglutinin-positive signal. Vertices of the nodeenriched pattern match better with local maxima of chondroitin sulfate density as compared to the uniform pattern. PNN is organized into clusters of meshes with distinct morphologies on the neuronal cell surface. Our findings suggest the role for the PNN microstructure in the synaptic transduction and plasticity.

The development and evolution of complex neocortical organisations is thought to result from the interaction of genetic and activity-dependent processes. Here we propose that a third type of process – mechanical morphogenesis – may also play an important role. We review recent theoretical and experimental results in non-linear physics showing how homogeneous growth can produce a rich variety of forms, in particular neocortical folding. The mechanical instabilities that produce these forms also induce heterogeneous patterns of stress at the scale of the organ. We review the evidence showing how these stresses can influence cell proliferation, migration and apoptosis, cell differentiation and shape, migration and axonal guidance, and could thus be able to influence regional neocortical identity and connectivity.

A key concept in system-level neuroscience is functional connectivity: any brain function builds on the activation and interaction of many specific and anatomically segregated brain areas. Due to their outstanding temporal resolution Magnetoencephalography (MEG) and Electroencephalography (EEG) seem to be particularly suitable to unravel such functional networks as well as their variations over time. However, this requires to design a proper analysis pipeline able to recognize the correct interactions even from short amount of data. In [1] we investigate the impact of the data length on Imaginary part of Coherency (IC) [2], generalized Partial Directed Coherence (gPDC) [3] and frequency-domain Granger Causality (fGC) [4] when their statistical significance is assessed by means of phase-randomized [5] and autoregressive [6] surrogate data test. In particular, we characterize the reliability of each combination of measures and statistical tests in the source space, i.e. when they are computed between the neural source time courses estimated from EEG time series. Data for such comparisons consist of four time series drawn from stable multivariate autoregressive (MVAR) models in which information flows only from the first to the second and the third source. We interpret the four signals as time courses of four dipolar sources and we compute the scalp potential generated by these sources. We add white measurement noise and different levels of biological noise, simulated as the scalp potential generated by 10 uncorrelated sources. Source reconstruction is performed by means of eLORETA [7] In order to study the impact of the data length on the considered measures, we compare results obtained by performing connectivity analysis from sub-samples of increasing length extracted from the reconstructed source time series. Exploiting knowledge of the MVAR models used to simulate data we compute the theoretical values of the connectivity measures under investigation. Then, for each pair of connectivity measures and statistical tests and for each sub-sample length we plot the histogram of the empirical values that pass the statistical test grouped on the theoretical values, for the pair of correlated sources, and the ratio of false positive, for the independent sources. Our results show that when the length of the data in input gets small, gPDC and fGC tend to provide a larger number of false positive, while the empirical values of IC pass the statistical test only with the longer sub-samples and in correspondence of the higher theoretical values. Moreover, IC seems to be less sensitive to the presence of biological noise. [1] Sommariva et al., Brain Topogr, submitted [2] Nolte et al, Clin Neurophysiol, 2004 [3] Baccalá et al, 15 th International Conference on Digital Signal Processing, 2007 [4] Geweke, J Am Stat Assoc, 1982 [5] Theiler et al., Physica D, 1992 [6] Schreiber and Schmitz, Physica D, 2000 [7] Pascual-Marqui et al, Philos Trans R Soc A, 2011

Experiments with real-time human-human and human-machine interactions started to clarify the neurobiological grounding of social cognition during interactive context, but also how relational dynamics could modulate brain dynamics. Specific signature of social roles can be tracked at neural, behavioral, and social scales. We will first see how spontaneous imitation can uncover both symmetric —joint leadership— and asymmetric —leader/follower— patterns. Then, we will present how both modes can be approach by principle based computational modelling and embedded in a virtual agent. Experiment with intentional forcing will illustrate how such real-time artificial leadership can support new sensorimotor and socio-cognitive therapeutics.

Neuroimaging can provide stable measures of human brain function non-invasively. However, the correspondence of the measured signals with molecular-level phenomena remains difficult to explore. Advances in genome-wide mapping of human traits provide tools to identify genetic loci involved in human cortical processes and, thus, a potential means of linking macroscopic cortical phenotypes with molecular-level processes controlled by specific genes. We have earlier shown in a combined magnetoencephalography (MEG) and genome-wide linkage study in >200 healthy siblings that auditory cortical activation strength is highly heritable and regulated oligogenically with linkages to chromosomes 2, 3, and 8 with several candidate genes with known functions in neuronal differentiation and axonal guidance (Renvall, Salmela et al. 2012). Similarly, we have observed the reactivity of the most prominent brain signal oscillation, the parieto-occipital 10-Hz alpha rhythm, to be linked to chromosome 10 with several functionally plausible genes that e.g. mediate synaptic transmission and have functions in brain development (Salmela, Renvall et al. 2016). We have now addressed the heritability of the rolandic somatomotor rhythm with distinct 10- and 20-Hz components that reacts to movement execution, observation, and motor imaginery (e.g. Salmelin and Hari, 1994). We measured spontaneous brain activity with a 306-channel Elekta Neuromag neuromagnetometer in 210 individuals (from 100 families) while the subjects were clenching their hands ~once per second or had their hands relaxed (3 min recordings with eyes open in both conditions). Power spectra at each MEG channel were calculated, and the peak frequencies and strengths of the 10-Hz and 20-Hz components were determined at the maximum channels over the bilateral somatomotor cortices in the hands-relaxed condition. The relevant peak was identified based on the spectral modulation elicited by the hand clenching. DNA was extracted from blood samples and genotyped. Peak frequencies of the 20-Hz but not of the 10-Hz component were highly heritable in both hemispheres (h2 > 0.69). In the right hemisphere, also the peak amplitude of the 20-Hz component showed very strong heritability of h2 > 0.76. Our present results suggest that the 20-Hz component of the rolandic somatomotor rhythm, originating mainly at the motor cortices (Salmelin and Hari, 1994) and with functions e.g. in imitation and motor learning, is under strong genetic control. We are currently analyzing the corresponding linkages at the genetic level. In general, we have demonstrated that identification of robust, heritable neurophysiological phenotypes and subsequently their genetic variants provides an interesting platform for interpreting neuroimaging data and for characterizing brain functions.

Chondroitin sulfate (CS) glycosaminoglycans inhibit regeneration in the adult central nervous system (CNS). We report here that HB-GAM (heparin-binding growth-associated molecule; also known as pleiotrophin), a CS-binding protein expressed at high levels in the developing CNS, reverses the role of the CS chains in neurite growth of CNS neurons in vitro from inhibition to activation. The CS-bound HB-GAM promotes neurite growth through binding to the cell surface proteoglycan glypican-2; furthermore, HB-GAM abrogates the CS ligand binding to the inhibitory receptor PTP (protein tyrosine phosphatase sigma). Our in vivo studies using two-photon imaging of CNS injuries support the in vitro studies and show that HB-GAM increases dendrite regeneration in the adult cerebral cortex and axonal regeneration in the adult spinal cord. Our findings may enable the development of novel therapies for CNS injuries. Published in Sci Rep. 2016 Sep 27;6:33916.

Recent data on pre- and postnatal auditory learning and competencies have shown that the brain’s capability to utilize sounds as material to focus auditory capacities starts already prior to birth and is extremely active during the first months of life. I will present neural evidence of prenatal learning (Partanen et al., 2014a,b) and discuss its consequences to our understanding of early auditory development. I will also review the competencies of typically developing infant’s auditory system (Teinonen et al., 2009 Winkler et al., 2003). In prematurely born infants, the acoustic environment during the first months in the hospital may be hostile for the development of the auditory competencies. Recent results of meta analysis studies show that the well-being of the infant and the mother can be helped with music therapy interventions (Bieleninik et al., 2016). Mothers working in noisy environments during pregnancy expose their fetuses to sound environments that are often repeated and non-vocal. We have followed the language development of these children to find potential decrements in learning speed. It is possible that learning repetitive, non-vocal patterns prior to birth might reduce the capacity to learn useful, language-related auditory rules and regularities after birth. I propose that music therapy interventions might be useful to help the language development of prematurely born infants and infants exposed to noise during pregnancy. I base this proposal on the knowledge of brain development in typically developing infants and on the data of effects of musical activities on the development of children with typical and reduced auditory system activity. I will present different types of subskills in language acquisition that would benefit from exposure to music and musical play. Finally, I will conclude by presenting a model of the usefulness of music exposure and musical play in infancy from the point of view of the developing auditory system.

Thanks to many data sharing initiatives, more and more brain imaging data is accessible online. This is great advance, but access to data is just the beginning. The key challenge remains the substantial amount of human curation, visual quality assessment and manual editing required by neuroimaging data. Currently, researchers need to download the data, and curate, edit and analyse it locally and redundantly in each research group. Many labs cannot embark on this time consuming process and have to discard a large proportion of the available data from their analyses, wasting time and funding. We aim at solving this challenge by proposing a co-editing platform for neuroimaging, similar to Wikipedia or Google Docs: BrainBox. It facilitates the creation of distributed teams of researchers collaborating in the analysis of open data – promoting community effort instead of competition. BrainBox uses various modern Web technologies to handle authentication, authorisation, and data harmonisation. User identification uses OAuth2 and GitHub login, and is the base for our access management. Server code uses Node and Express, and the database uses Mongo. This allowed us to develop an API which makes it easy for other applications to push and pull (access-controlled) information from BrainBox. Volume and text annotations use WebSockets and are updated in real-time. All code is available through GitHub, providing a common platform for developers to contribute. Our Web app makes it easy to work with shared brain imaging data directly online: Discover and visualise it, collaboratively curate and annotate it, connect research teams and manage access rights. BrainBox allows users to view data in a stereotaxic viewer or to compute and visualise a 3D model from segmentations. Users can collaboratively create or edit several multi-colour volume and text annotations for each MRI. Those are updated in real-time and let users see joint progress with all connected users. The link with MetaSearch allows users to find all MRI data relevant to them for use in BrainBox. Public user pages track all projects which a user has created or collaborates in and become a way to discover new data and projects. No data has to be downloaded or stored, no software has to be installed, and it will be possible to recruit a large, distributed group of collaborators online. Many online platforms exist for co-editing text (Wikipedia, Google docs etc). BrainBox is the first to bring co-edition to neuroimaging. Such platform can provide significant benefits: the amount of curated data available online can increase exponentially, there is greater trust in the data due to collaborative efforts, and it enables greater reproducibility by reducing redundancies in processing imaging data. As the amount of publicly available data grows, platforms such as BrainBox will allow increased collaboration, more effective hypothesis testing, and improved reproducibility. Supported by the Open Science Prize.

Animate and inanimate objects evoke distinct activity patterns in the human ventral temporal cortex (VTC). A core debate concerns the extent to which this organisation reflects selectivity to visual attributes that are characteristic of these categories (e.g., most mammals have four legs). By showing a robust animacy organisation for stimuli that were matched based on behavioural visual similarity metrics, a recent fMRI study argued that visual features do not fully explain the animacy organisation in VTC. In the present study, we followed up on this study by using convolutional neural networks (CNNs) to quantify visual similarity of the same stimuli at various feature depths. We found that intermediate layers of the CNN captured the shape and texture features of the images that drove behavioural visual similarity data in the previous study. Interestingly, values in the final layers of the CNN allowed accurate classification of the stimuli as animate or inanimate, also for the behaviourally-matched image pairs. Moreover, these classification scores correlated significantly with corresponding classification scores obtained from neural activity patterns in the bilateral object-selective cortex. We are now running a whole-brain searchlight analysis to search for any animacy clusters not completely driven by visual features. These results suggest that CNNs are sensitive to subtle image features that might capture categorical distinctions but may not be reflected in behavioural visual similarity measures. These findings reinvigorate the debate about the principles driving the animacy organisation in VTC.

Studies of apraxia (a disorder of complex movement) suggest that the inferior parietal lobule (IPL) plays a role in kinematic aspects of imitation, particulary for meaningless actions. Neuroimaging studies of healthy people often have less to say about subregions of the IPL, or how different types of action are processed for imitation. We used repetitive transcranial magnetic stimulation (rTMS) to bridge this gap and better understand the roles of the supramarginal and angular gyri in imitation. We also examined if these areas are differentially involved in meaningful and meaningless action imitation. We applied rTMS to the supramarginal or angular gyrus, then asked participants to imitate a confederate’s actions whilst the arm and hand movements of both individuals were motion-tracked. rTMS to area PF in the SMG significantly reduced velocity of participants' digits specifically during imitation of finger gestures, regardless of action meaning. Our results address discrepancies between studies of imitation in apraxia and healthy individuals, and explain why IPL activity is frequently reported in healthy individuals, regardless of the action type imitated. Our results support recent discussions in apraxia and confirm the role of the SMG, particularly the stimulated area PF, in asserting kinematic control during gesture imitation.

Introduction Virtually all organisms have a circadian rhythm, an endogenous clock with a rhythm of about 24h in humans and mice. In humans, circadian dysregulation has been linked with psychiatric disorders such as depression and schizophrenia. While the link is clear, the mechanism underlying these associations is unknown. We recently have found that a mutation in the gene Zfhx3 can lead to disrupted circadian behavioural rhythms in mice. Interestingly, in addition to strong expression in the hypothalamic circadian suprachiasmatic nucleus, ZFHX3 expression is also high in the central nucleus of amygdala (CeA), an important structure related to emotion, fear and anxiety. Could ZFHX3 inside the CeA play a role in the link between circadian rhythms and psychiatric diseases? Materials and Methods To investigate the role of ZFHX3 inside the central amygdala, we have used heterozygous (HET) female and male B6J/C3 mice with a mutation in Zfhx3 and compared them with wild-type littermates (WT). The use of the light-dark box and elevated-0 maze allowed one to assess aspects of anxiety in these mice. Moreover, at a cellular level, the expression of Zfhx3 and Spata13 in the CeA was evaluated by qPCR. Results In females, no differences were found in the time spent in the light compartment of the light-dark box or in the open-arm of the elevated-0 maze. Conversely, HET male behaviours in these tests indicated a reduced anxiety relative to WT males. Upon further investigation, both female and male HETs showed a reduction in “cautious” behaviour toward the light part of the light-dark box compared to WT mice. Interestingly, WT females also showed an increase in “inhibition behaviour” toward the light part of the light-dark box compared to male WTs but, surprisingly, female HETs showed a reduction of this behaviour equivalent to that of HET and WT males (no difference was observed between male genotypes). Additionally, Zfhx3 and Spata13 expression showed similar patterns to the “inhibition behaviour” with a reduced expression in HET females, HET males and WT males compared to female WTs. Conclusion ZFHX3 plays a role in 1) anxiety in males only; 2) “inhibition behaviour toward anxiogenic area” in females only; 3) “cautious behaviour” in both sexes. Ongoing studies Anxiety, risk-taking and inhibition behaviours are known to implicate the CeA. To investigate whether ZFHX3 in the CeA plays a role in the behaviours mentioned above we are using the Cre-Lox system to delete Zfhx3 in mouse CeA by viral vector injection. Subsequently, behaviours will be assessed in experimental and control groups. In parallel, we will investigate the cellular and molecular mechanisms and consequences of Zfhx3 deletion in the CeA. Importance of this study In this study we hope to uncover the hitherto unestablished role for Zfhx3 in the CeA. In doing so, we also expect to establish some insight into the link between circadian rhythms and neuropsychiatric disorders.

Autism spectrum disorder (ASD) is associated with difficulty in processing speech in a noisy background, but the neural mechanisms that underlie this deficit have not been mapped. To address this question, we used magnetoencephalography to compare the cortical responses between ASD and typically developing (TD) individuals to a passive mismatch paradigm. We repeated the paradigm twice, once in a quiet background, and once in the presence of background noise. We focused on both the evoked mismatch field (MMF) response in temporal and frontal cortical locations, and functional connectivity with spectral specificity between those locations. In the quiet condition, we found common neural sources of the MMF response in both groups, in the right temporal gyrus and inferior frontal gyrus (IFG). In the noise condition, the MMF response in the right IFG was preserved in the TD group, but reduced relative to the quiet condition in ASD group. The MMF response in the right IFG also correlated with severity of ASD. Moreover, in noise, we found significantly reduced normalized coherence (deviant normalized by standard) in ASD relative TD, in the beta band (14-25 Hz), between left temporal and left inferior frontal sub-regions. However, unnormalized coherence (coherence during deviant or standard) was significantly increased in ASD relative TD, in multiple frequency bands. Our findings suggest increased recruitment of neural resources that irrespective of task difficulty in ASD, alongside a reduction in top-down modulations, usually mediated by the beta band, that are needed to mitigate the impact of noise on auditory processing.

Background: Approximately one-third of patients with schizophrenia are considered treatment-resistant. For these patients, the atypical antipsychotic drug clozapine is recommended; however, there is still significant variability in treatment-response. Animal studies have demonstrated that clozapine induces epigenetic changes i.e. chemical alterations to the genetic code which influence subsequent gene expression. In the current study we explore whether clozapine induces DNA methylation changes (one form of epigenetic change) over 6 months in humans. Methods: We recruited 22 participants with a diagnosis of treatment-resistant schizophrenia, before they were prescribed clozapine. We then collected whole-blood samples at baseline and follow-up (6 weeks, 12 weeks and 6 months after clozapine start date), alongside clinical assessments to determine treatment response. We quantified DNA methylation at ~480,000 sites across the genome using the Illumina 450K HumanMethylation array. An epigenome-wide association study was performed comparing DNA methylation over the four time points. Results: Multiple CpG sites showed changes in DNA methylation associated with length of time exposed to clozapine, dependent on treatment response. The one epigenome-wide significant finding was a CpG site in the body of the DPP4 gene, which is involved in immune regulation. Conclusions: This is the first study to identify longitudinal epigenetic changes following clozapine exposure in human subjects, associated with treatment response. Recruitment is ongoing and further analysis will look at whether epigenetic changes are associated with different aspects of treatment response and adverse reactions. Ultimately, these data will help us understand the mechanisms involved in clozapine, potentially providing biomarkers to predict response to clozapine.

New techniques in brain imaging, such as the combination of magnetoencephalography (MEG) and magnetic resonance imaging (MRI), direct imaging of neuronal or impressed currents (NCI or CDI) in the brain require new verification and validation approaches. We propose a new concept for a physical head phantom to address verification and validation of these techniques. Compared to simulations, physical phantoms take into account the real world influences, such as environmental noise or 3D positioning errors. At the same time, physical phantoms allow for a representation of the ground truth, because of their well know structure and function. We present a phantom concept and a characterization study of phantom materials. The phantom consists of three conductivity compartments and interior components for signal generation and measurement purposes. We characterized the materials clay and polymer-gypsum with respect to their conductivity properties. Both materials provided conductivity values in the desired range for a skull modeling material. The phantom will be completed by an inner compartment filled with sodium-chloride solution and an outer compartment consisting of an agar gel mimicking the skin. Our proposed physical head phantom will be used also for verification and validation of transcranial electric stimulation and source reconstruction schemes in EEG and MEG.

In the BREAKBEN project (funded by Future and Emerging Technologies program of the European Union), our goal is to improve the accuracy and reliability of neuronal activity localization and characterization in the brain. Two methods involving ultra-low-field MRI (ULF MRI) using SQUIDs will be developed: 1) Combined MEG and MRI measurements, and 2) neuronal current imaging. In ULF MRI, the signals are recorded at about 100 microtesla instead of the several tesla in conventional MRI. BREAKBEN’s predecessor project MEGMRI (2008–2012) indicated that the power signal-to-noise ratio in the MRI measurement should still be improved by a factor of at least 1000 for the structural MRI to be clinically useful. This can be done by lowering sensor noise and by making the prepolarization magnet more powerful. In addition to improving the reliability of locating and characterizing brain activity, the new technology may also enhance the diagnosis of cancer patients thanks to improved MRI contrast at ultra-low fields. Cost reductions may be expected on the basis of improved workflow and more accurate diagnostics. Furthermore, there is hope that the new device can be used to measure the conductivity structure of the brain, which would improve the accuracy of locating brain activity with MEG as well as with EEG. Aalto University and Elekta Oy will build the hybrid MEG-MRI device using a new generation of SQUID sensors developed by VTT Technical Research Centre of Finland. The resulting technology will be used by the BioMag Laboratory of the Helsinki University Hospital in patient trials and by the University of Chieti–Pescara in studies of brain connectivity. Physikalisch- Technische Bundesanstalt (PTB) Berlin will attempt the detection of neuronal activity by ULF-MRI and NMR techniques. Sophisticated “phantoms” will be developed by the Technical University of Ilmenau to mimic the properties of the human head and brain to allow testing the sensitivity and accuracy of the new devices.

One of the first methods to study human brain functioning was to observe what happened when a part of the brain was lost. Galen, among other Roman physicians to the gladiators, noted that the brain was the seat of mental faculties. The lesion deficit method studies the behavior of brain-damaged patients and has helped to make many associations between brain structure and function. It follows the logic that if a patient cannot do X, then the execution of X must depend on the lesioned area. This method is still potentially the most powerful method for establishing the necessity of brain structures for given functions. With the advent of numerous functional imaging methodologies, some question the continued contribution of the lesion method. New functional imaging techniques including functional magnetic resonance imaging (fMRI), electroencephalography (EEG) and functional connectivity (fCN) has allowed researchers to observe the working brain. While both the lesion method and newer functional methods suffer from limitations, they are complementary weaknesses. Through the use of converging methodologies, multiple methods provide complementary information about human brain function. Patients with brain lesions can provide a unique view of human brain functioning and when paired with functional methods can provide information about treatment and plasticity. As more advanced technologies are developed, it will be important to continue to combine these technologies with the lesion method.

This project aims to address the risk of information being lost through coercion/torture by training participants a high-entropy underlying target sequence in a visuomotor typing task until the desired skill is sufficiently and implicitly acquired outside of conscious awareness. In the study, participants are asked to type a seemingly random sequence of keys. However, the structure of the sequence is divided into two underlying sequences (Target and Foil), which is blind to the participants. The Target sequence is repeated more frequently than the Foil sequence to create an implicit Target- Foil motor learning disparity. Prior to a RT test, Group 1 is given 45 mins of training, unlike Group 2. The RT test is then taken again a week later to test for skill retention. On average, the RT during the Target sequence was found to be significantly lower (p<.05) in the trained group compared to the untrained group. Participants failed to recall sequence structure, establishing tacit skill expression outside of explicit awareness. This ensures that an authentication protocol is built around what users implicitly learns through prior sensorimotor training. Thus guaranteeing that information is not susceptible to coercion. The findings hope to lay a framework for future crypto primitives in large-scale protocols.

The coordinated activity of subsets of neurons across multiple circuits is thought to support complex behaviours. These functionally coupled subsets are often referred to as cell assemblies. The detection of cell assembly patterns from single-unit recordings usually relies on finding significant co-firing within a particular time bin. Choosing a bin length based on synaptic integration times, e.g. 20 ms, makes these methods well-suited to detecting Hebbian-like cell assemblies within a single structure such as the hippocampus. However for assemblies that span multiple circuits it may be that the assembly-forming neurons interact at longer latencies or over successive temporal windows. Here we apply independent component analysis to the cross-correlation between each neuron pair at multiple lags in order to incorporate these interactions. We show that this method is able to capture cross-structural assemblies, and contrast its performance to other methods, using both spike-train simulations and in vivo recordings from the rodent hippocampus and ventral tegmental area. Importantly we found that different assemblies detected in this manner show distinct neurophysiological correlates such as their coupling to different phases of hippocampal theta oscillations, responses during sharp-wave ripples, and speed modulation.

A common method of investigation in cognitive neuroscience is to present the brain with stimuli of interest, and record the electrical potentials that arise at the surface of the scalp. When we specifically time-lock these potentials to the time point at which stimuli occurred, we call it the event-related potential (ERP) method. This approach has been used in thousands of experiments in order to investigate how and when the brain processes information, with over 400 000 ERP papers currently listed on PubMed. The popularity of this method, and the scale of the literature, make it difficult to keep up with new developments, and to efficiently summarize what is known from these kinds of investigations. In this project we present an automated method to summarize the ERP literature, using a basic text-mining approach. We use the PubMed E-Utilities to scrape the relevant literature, and perform automated meta-analyses to examine the relationships between ERP terms, cognitive domains, and disease states. We curated dictionaries of terms, including over 80 previously described ERP components, and determined co-occurrence probabilities in published papers between ERP components and cognitive and disease terms to investigate what different ERP components are associated with. We also extracted words from the abstracts of all articles found using the same ERP dictionary, allowing us to build a data-driven profile for each ERP, including the terms with which they are most affiliated. The results serve not only as a tool to summarize individual ERP components, but also to investigate patterns and consistencies across components with different characteristic timings. For example, by ordering ERPs by their typical delay, and investigating the cognitive terms with which they are associated, we can build a profile of how the brain typically responds to stimuli, through time. By doing this we observe that the earliest components, up to about 150 ms post-stimulus, are typically associated with sensory processing, where as later processing, especially from about 400-600 ms post-stimulus is much more associated with cognitive terms such as conflict, language, and semantics. This data has been combined into an easily searchable database, allowing for efficient look-up of the ERP profiles, efficiently summarizing a large body of research. This database can be used both as a learning and teaching tool, and as a method of inquiry into the previously hidden structure of the existing literature.

Different measures of cortical morphology have been shown to index distinct aspects of inter-individual differences (e.g., volume, thickness, surface area, gyrification). Here I consider the additional measure of structural complexity, as quantified by fractal dimensionality. Using open-access MRI data across the adult lifespan, I examined the relationship between each measure and age-related differences in brain morphology. Here I demonstrate that fractal dimensionality, which incorporates shape-related properties, to be a more sensitive measure of age-related differences in cortical and subcortical structures.

Resting-state blood oxygen level-dependent (BOLD) functional MRI (fMRI) has progressively been used to investigate the complex functional architecture of the brain. The sensitivity and reproducibility of these functional networks is associated with several confounding factors, including motion artefacts, cardiac and respiratory fluctuations. The changes in hemodynamic response function (HRF) shape between regions and individuals also make a significant contribution to intra- and inter-subject variability, and subsequently affect the connectivity estimation. The HRF is always used to measure the latency, duration and amplitude of the underlying neural activity from BOLD fMRI signal, acts as an effective indicator of aging and diseases. However, most of algorithms for HRF estimation are based on the task-related fMRI data, few of them could be directly applied to resting-state protocols. Here, we present a procedure to explore voxelwise HRFs of spontaneous neural activity from resting-state BOLD fMRI data. The validity of proposed algorithm is firstly tested on neurophysiological simulations. Then a 7T BOLD fMRI dataset revealed that resting-state HRF is associated with retrospective indicators of self-generated thought in default mode network (DMN) and non-DMN regions, which are considered as the central areas for spontaneous thought in the existing literature. A resting-state ASL and BOLD fMRI dataset further demonstrated that there is highly correlation between resting CBF and HRF, which indicates the underlying neurovascular coupling mechanism of resting-state HRF. Taken together, we provide the simulated and empirical evidences supporting that resting-state HRF originates from neurovascular coupling changes, and reflect the resting-state brain dynamics. Autonomic nervous system fluctuations are a key component in the variance of BOLD signal, and definitely it biases the estimation of the HRF at rest. We find that the resting-state HRF estimation is significantly modulated in the brainstem and surrounding cortical areas. From the analysis of two high-quality datasets with different temporal and spatial resolution, and through the investigation of intersubject correlation, we suggest that spontaneous point process response durations are associated with the mean interbeat interval and low-frequency power of heart rate variability in the brainstem. Most datasets used and all the code are publicly available.

The amount of data that can be generated in a typical neuroscience lab grows at an amazing rate as measurement technology becomes widely available. For example, researchers can use neuroimaging (e.g. confocal microscope) to visualize cells, electrophysiology to measure electrical signalling and molecular techniques to see the changes in gene and protein expression. This wide range of spatial and temporal scales of investigation is challenging for neuroscience. Data is collected at resolutions ranging from the subcellular, cellular, regional and whole brain levels representing processes that occur over picoseconds to many years, and in addition, data is often collected in various formats at each of these sublevels. Neuroinformatics tackles the challenge of integrating information across all levels and scales of neuroscience. Tools and standards provide solutions for fitting together and analyzing data that comes from vastly different timescales, techniques and animals across all levels, from single cells to the whole brain. The field also provides guidelines for recording information about data and how it is analyzed so that experiments can be reproduced and the data can be reused in new studies. For neuroinformatics to be useful, it requires the coordinated efforts of many scientists coming together as a community to resolve issues and deliver solutions, such as global standards, best practices, tools and workflows. INCF (incf.org) supports, coordinates, and grows this community by enabling groups with common interests to unite in addressing challenges of modern neuroscience. We do this by: - providing seed funding for collaborative projects - building https://knowledge-space.org, community encyclopedia that links brain research concepts with data, models, and literature from around the world - development and provision of training and educational resources in neuroinformatics - supporting Special Interest Groups, suggested and led by the global neuroscience community

Transcranial Magnetic Stimulation (TMS) is a tool for explore the excitability of central tracts on the human sensory systems. The use of TMS with electrical peripheral stimulation can be used for the study and the understanding of the descending control over postural and voluntary tasks. Functional implications of these findings indicated the processing information at spinal propiospinal interneuronal level and how can be affected by posture related activities. Specific time intervention and the integration of excitatory and inhibitory inputs over the soleus alpha-motoneurons are the most likely responsible cause for the different effects of modulation of posture related mechanisms.

Brain function can be studied from a variety of perspectives in both time and space. A vast amount of data has been acquired at each of these scales, contributing to a growing knowledge of brain function at the level of the individual neuron, neural ensembles, or large-scale networks. A fuller understanding of brain function remains elusive, however, because no single technology currently exists to simultaneously acquire data across all spatial and temporal scales. As a result, little is known about the dependencies of one scale on the next and tasks such as directly linking cellular mechanisms with specific cognitive functions or dysfunctions remain a great challenge. TheVirtualBrain (TVB, thevirtualbrain.org) neuroinformatics platform begins to address this gap by building a unifying theoretical framework to quantify the relations between scales. TVB makes use of empirical data as the foundation for large-scale computer simulations of brain dynamics. Models based on individual subjects EEG and fMRI data have identified common local biophysical and global network properties that fuse modalities. The direct link to empirical data enables a personalization of clinical models, opening a potential use for diagnosis and prognosis. We have used TVB to identify biophysical parameters estimating local excitation/inhibition balance that predict therapeutic outcome in stroke patient. TVB models for epileptic patients can help confirm identification of seizure focus and accurately simulate seizure propagation. TVB thus acts as a much needed “computational microscope” that allows the direct inference of neurobiological mechanisms underlying human brain function in both health and disease.