Poster Sessions

In recent years, artificial neural networks have become very popular and capable of computational tools across a wide range of applications in interdisciplinary fields. Neural networks have been particularly useful in data analysis and computational models. Nonetheless, the principles underlying there are still not really understood. Thus, they are often likened to a black box, which researchers attempt to uncover using various conceptual, mathematical, and computational approaches. One particular important direction we want to take here is to understand the structural properties of representation learning in those networks. Here, we would like to understand how networks learn, represent, and transform the structure inherent in observed data to solve a given task. We want to particularly focus on the geometrical and topological properties of these representations in neural networks. Data we usually work with are in high dimensions, but we assume they form a structure called a manifold (manifold hypothesis). Our goal is to discover the latent structure of data or representational manifolds by applying generative models, specifically VAE, in order to reveal the geometrical structure of the data manifold. As part of our investigation, we are using differential geometry and topological data analysis to study its structure and how it relates to the tasks of the network, as well as whether these are predictive of the behaviour or performance of the networks.

Previous research has uncovered a large-scale organization of object categories in occipitotemporal cortex by the dimensions of animacy and real-world size (Konkle & Caramazza, 2013). The tripartite division of cortical zones with a preference for large objects, all animals, and small objects has been robustly replicated and appears to be reflected by the mid-level visual feature curvature, i.e. large objects tend to be boxier, and small objects and animals curvier (Long et al., 2017). However, given the factorial design in the original studies, it has remained open to what degree these findings generalize to larger stimulus sets. To address this question, we used THINGS-fMRI, a large-scale dataset comprising fMRI responses to 8,640 naturalistic images of 720 animate and inanimate object categories (Hebart et al., 2023). We then collected and applied a rich behavioral dataset of perceived animacy, real-world size, and image-wise curvature ratings (Stoinski et al., 2023). Our results replicate many facets of the characteristic animacy-size organization, such as alternating patterns of animate-to-inanimate and large-to-small gradients, respectively. However, beyond these previous findings, we found additional, bilateral clusters between FFA and PPA that responded preferentially to small objects. In addition, our results showed pronounced size gradients within animate selective regions of the right hemisphere. These results were replicated across all three subjects and could not be explained by the broader size range, category diversity, or display eccentricity. Instead, other factors, such as texture and other low to mid-level visual features or the higher sensitivity of our dataset, may contribute to the stronger alternation of large-small preferences. Finally, contrary to the view that high-level animacy and size distinctions are explained by curvature information, perceived curvature explained only limited portions of observed animacy-size selectivity while the overall pattern of results remained the same. Together, our results add important facets to understanding the large-scale functional organization of occipitotemporal cortex.

Interacting with the environment requires selection of actions based on our knowledge of the world. We have a remarkable ability to form and consider different action plans in order to reach our goals. However, the neuronal mechanisms underlying abstracted representation of actions are still unknown. Recent evidence suggests that humans organize spatial and non-spatial relational knowledge in form of a cognitive map in the hippocampal-entorhinal system, which facilitates learning and novel inferences. The question we wanted to investigate is whether and how cognitive maps could support representation and selection of available action plans. To answer this question, we designed a set of behavioral tasks using immersive virtual reality (VR) with head-mounted display and a hand controller. Participants (n=46) were trained to execute hand actions by moving two virtual joysticks, that trigger launching of a ball towards them. Participants learned to associate different actions with different probabilities to either catch a flying ball (outcome dimension 1), or for the ball to remain visible throughout its flying trajectory (outcome dimension 2). This mapping allowed for arranging actions along their outcome dimensions in a two-dimensional action-outcome space. In the subsequent testing phase, participants compared the action combinations based on the learned action-outcome contingencies. We performed multidimensional scaling on the behavioral data from the comparison task. Then, we fitted the subjective judgments to the actual outcome space, showing that participants’ mental map of differences between sampled action combinations correspond to the actual distribution of those combinations on the two-dimensional outcome space. To explore the neural mechanisms of this map-like representation, we tested participants’ knowledge of action combinations and action-outcome contingencies while recording brain activity with fMRI. We performed the representation similarity (RSA) and adaptation analyses of the neuroimaging data. Our results revealed that the relative positions of action combinations in the two-dimensional action-outcome space were represented with a grid-like code in the entorhinal cortex and with a distance code in the hippocampus. Furthermore, we could show that the cognitive coding of interactive actions was supported by the intraparietal sulcus. These results indicate a link between hippocampal cognitive maps and action representations.

Introduction Central post stroke pain (CPSP) is a common consequence of somatosensory stroke, typically occurring within 3-6 months after the stroke. CPSP is refractory to therapy and affect patients' quality of life. We conducted a prospective longitudinal study in which we recruited somatosensory stroke patients in the acute phase and performed detailed clinical, behavioral, and MRI assessments. Methods 77 stroke patients were recruited at Charité – Universitätsmedizin Berlin. Resting-state fMRI were performed before the pain (day2-10 days post-stroke) and after the pain (>60 days post-stroke). In total 61 patients were included in the analysis, 18 CPSP and 43 non-pain somatosensory stroke (NPSS). For the seed-based connectivity analysis, seeds were chosen from the pain network. For the connectivity gradients, we constructed affinity matrices and decomposed them using principal component analysis. The first gradient was included in the regression analysis. Results (1)In the cross-sectional analysis before the pain, CPSP group showed stronger functional connectivity between the contralesional pars triangularis and opercularis in the inferior frontal gyrus. When the pain had occurred, CPSP group showed lower functional connectivity between the ipsilesional superior parietal lobule and putamen. There was a significant group by time interaction showing that pain patients have lower functional connectivity between the ipsilesional superior parietal lobule and putamen. (2) In the cross-sectional analysis on gradient 1, 8 parcels before the pain and 14 parcels after the pain are close to somatosensory and pain. In the longitudinal analysis on gradient 1, 7 parcels which are close to somatosensory, show a group by time interaction. The post-hoc test showed that in the pain group 6 parcels are close to somatosensory. Conclusion This analysis shows for the first time that functional connectivity of multiple brain areas differs between CPSP and NPSS patients. The changes in connectivity before pain development could help predict who will develop pain after stroke.

Abstraction is the human ability that allows us to represent concepts beyond the available concrete (perceptual and/or motor), immediate, physical experience. Abstraction allows us to comprehend entities detached from grounded experience (e.g., intelligence), but also involves generalising concepts to other instances in which these occur (e.g., different intelligent situations). This ability is key to human cognition, behaviour and intellectual success, so the importance of understanding how it develops deems undeniable. While substantial research is available on how children learn specific concrete concepts (e.g., freezer) and generalise across exemplars and categories (e.g., from freezers to appliances), much less is known about abstract concept learning, where children learn concepts that refer to displaced entities, detached from physical experience, and their associated labels. Particularly, we lack knowledge on the development of abstraction abilities in its ecology, namely within social interaction with adults and peers. To understand how social interaction supports learning of abstract concepts in children, we propose a study that will follow an ecological, hyperscanning and multimodal approach. In this study, children aged 5-12 years old will be prompted to learn new abstract concepts and to generalise these to different situations, while socially interacting with a familiar adult. We will simultaneously record behavioural (verbal and non-verbal) and neural measures from both child and caregiver using fNIRS, allowing us to examine the moment-to-moment dynamics and to evaluate whether and how individual as well as dyadic coordination measures contribute to successful learning. This hyperscanning study will also allow us to move beyond standard approaches that focus on only one of the people in the interactive setting, either the learner or the teacher. Overall, this comprehensive approach will enable us to capture the full complexity of the learning mechanisms that take place in real-world settings. This project will have important scientific outcomes, as it will shed light on the neural substrate of conceptual learning in children in its ecological context, social interactions, as well as how behaviours, neural activity, and their coordination between child and adult support this learning mechanism. We seek feedback from the neuroscience community at the planning stage to enrich our research and ensure its robustness and success.

The fMRI and Eye-Tracking (ET) study is focused to investigate the cognitive architecture of action perception, specifically, which cues the brain processes during observation of action sequences and how the observer’s brain uses those cues to predict the actions of the actor. An artificial actor grammar describing legal transitions of a set of arbitrary action sequences using four study objects, two cups and cubes was developed. The grammar also consisted of action sequences with points of disambiguation to investigate anticipatory processes. Based on fixed transition probabilities, information theoretic parameters, namely, entropy and surprisal were quantified for each action step. The study comprised of four sessions on four consecutive days and fifty-eight participants were recruited for the study. Participants underwent an imitation training of legal sequences in third-person perspective on two consecutive days. The third day was fMRI-ET session, during which the subjects were presented with both legal (the action sequences shown during imitation training sessions) and, to ensure attention to sequences and introduce violation of expectation, a set of illegal (the legal action sequences with a specific step variation) were presented in randomized order. The last day was a behavioural memory task wherein the legal sequences were presented and paused after each action step, and the subjects must choose the correct next action step from the options displayed. The neural correlates and behavioural responses are being analysed for different levels of entropy and surprisal during action observation, perception, and prediction from an observer’s perspective. The findings will contribute to our understanding of anticipatory processes in action perception and its use for coordination in joint actions with other individuals or robots. Keywords: Joint Action, fMRI-Eye tracking, Action Perception, Action Prediction, Violation of Expectation.

Processing in auditory subcortical sensory pathway nuclei (i.e., the medial geniculate body, MGB and the inferior colliculus, IC) in typically reading individuals is based on the comparison between internally generated predictions and sensory input, a process known as predictive coding. Developmental dyslexia is associated with histological and functional alterations in the left MGB and structural changes in the connections between the left MGB and auditory association cortex. Our aim is to test the hypothesis that the predictive coding mechanism in the left MGB is altered in a group of individuals with dyslexia in comparison to matched controls. Based on previous work that showed an association of left MGB alterations in dyslexia with rapid automatised naming abilities, we also expect that the hypothesised left MGB alteration in predictive coding is associated with the ability to rapidly name letters and numbers (rapid automatized naming for letters and numbers, RANln). To address our hypothesis, we used high-resolution fMRI at 3 Tesla and an established oddball design. In the oddball design, the predictability of the stimuli is parametrically modulated and predictions and stimulus statistics are orthogonal to each other. We ran a pilot study in 7 participants with dyslexia and 7 typically-reading control participants (age, gender, education matched). The left MGB response in the dyslexia group had a significantly lower dependency on predictability than that of the participants in the control group. In addition, the alterations were specific to the left MGB in dyslexia, as the dependency of the responses on predictability in the inferior colliculus (IC) was comparable to that of the control group. We found that the speed of naming letters and numbers aloud (RANln) correlated negatively with the amount of dependency of the left MGB responses on predictability in the dyslexia group, but not in the controls. The pilot results give first indication about the functional relevance of the left MGB alterations in dyslexia and will be followed up with a preregistered study with a large sample size.

In humans, language processing is supported by a fronto-temporal network connecting Broca’s and Wernicke’s area through dorsal and ventral pathways. Despite the presence of a homologous brain system in some non-human primates, only humans can process language. By studying the brains of our closest ancestors, we can gain valuable insights on language evolution. Most of the current knowledge on primate brain connectivity originates from tracer studies in macaques. Researchers now use diffusion MRI for comparative neuroscience to retrace the evolution of the brain’s structure looking for shared features between primate species. The last decade of research has provided insights on the anatomy of language relevant tracts like the arcuate fascicle and its homologue in the non-human primates. The variety of techniques used so far led to some incoherence in the way the tracts are described, which in turns makes the interpretation of tractography results delicate. In this review we will first summarise the behavioural and anatomical context in which the tracts homologues have been defined through history. We will then provide an overview of the current state of the tract homologues’ definition used in the literature and will finish by bringing some perspective on said definitions.

A key component to acquire in language development is the ability to process grammatically complex sentences. For this, the acquisition of the morpho-syntactic rule system is fundamental, which forms an intersection between lexical words and syntactic structures. The preschool period is characterized by major behavioral improvement in grammatical knowledge on both sentence and word level with milestones at around 4 years of age. In the adult brain, dorsal and ventral white matter fiber tracts connect the frontal and temporal cortex, and play different functional roles for language processing. Here, we ask which maturational changes in the child’s neural language network underlie the emergence of morpho-syntactic ability in the critical preschool years. For this, we reconstructed language-related fiber tracts from diffusion-weighted MRI data in two independent samples of 3- & 4-year-old (N = 40) and 3- & 5-year-old children (N = 90). In a preregistered procedure, we related indices of white matter maturation to children’s grammar performance extracted from a standardized test of general language development. For children’s ability in morpho-syntactic rule generation, we found in both samples no main effect but a significant age interaction for younger (3-y.o.s) versus older (4- and 5-y.o.s) preschoolers in the dorsal fiber tract that targets the premotor cortex (BA6). Additional analyses revealed that the effect was only present in the 4- and 5-year-olds but not the 3-year-olds yet. These results were consistent across several measures for white matter maturation. Our findings show that grammatical knowledge on the word level relies on the development of the dorsal fiber tract to BA6 in preschool children. While in adults the dorsal tract targeting BA44 in Broca’s area is involved in complex syntactic processes, the dorsal tract to BA6 supports auditory-to-motor mapping. As brain structures that are crucial for syntax in adults are not yet matured, preschool children might rely on structures supporting phonological regularities as the dorsal tract to BA6.

Introduction: Sensorimotor brain-computer interface (BCI) is a system that decodes imaginary movements from the recorded activity of the user's brain and provides commands to external devices such as prostheses. On average, around 20% of individuals are unable to learn to control the device, and the mechanisms of successful BCI training are not clear yet. Signal-to-noise ratio (SNR) of the mu rhythm and functional connectivity between sensorimotor areas were previously shown to predict BCI performance. In the current work, we aimed to validate the relationship between SNR, connectivity, and accuracy of BCI control and investigate their longitudinal changes in a dataset with multiple sessions of BCI training. Methods: We used publicly available EEG recordings of 62 participants that performed 7-11 sessions of a cursor control task based on the imaginary movements of their hands. We considered 24 combinations of methods for inverse modeling and extraction of time courses of activity in sensorimotor areas of both hemispheres. Then, we applied multiverse analysis in order to assess robustness of effects with respect to selection of the pipeline. For each pipeline, we used coherence, imaginary part of coherency, and lagged coherence to estimate connectivity between sensorimotor areas. Results: We observed a positive effect of SNR on accuracy and connectivity, which was robust to the selection of the pipeline. At the same time, effects of connectivity on accuracy and changes in connectivity over time were significant for less than 25% of pipelines after controlling for changes in SNR. Discussion: Our results suggest that SNR is a primary factor of the observed variability in performance, while effects of connectivity are less robust across different pipelines. This in turn may indicate that changes in neuronal connectivity within the sensorimotor system are not the main factor defining learning in BCI.

Driving requires multitasking which can lead to cognitive overload and therefore to mistakes and accidents. To investigate this phenomenon, we created a cognitive model in ACT-R. This model was supposed to simulate human thinking in a way that it uses a realistic strategy such that we can observe cognitive load in the model that is comparable to the participant data. The model and the participants were solving the same task: driving on a straight highway in a simulator while solving a serial recall task. In one setting they were supposed to recall information and in the other one they had to use the information immediately. We could observe that the relative change of the cognitive load between the different settings for the model and for the participants was comparable, but the absolute amount differed.

Decades of research have described a frontotemporal network critical for lexico-semantic processing. However, whether these models reliably account for individual variability can only be addressed via deep phenotyping: the within-subject collection of behavioral and neuroimaging responses to stimuli spanning a wide range of variables during tasks requiring different levels of semantic processing. We present the cneuromod-triplets dataset from the Courtois Project on Neural Modelling: a large-scale, freely available and multimodal dataset that covers a wide range of neuroimaging and behavioral tasks in a small sample of subjects. The dataset includes two tasks that enable 1) the isolation of neural activity associated to single word processing and its modulation by perceptual and conceptual features of words and 2) the investigation of semantic processing of those same words within a minimal context. Four right-handed participants (2 female, age: 44±4.3 years; education: 20±3.6 years) were scanned in a 3T fMRI scanner over multiple sessions while they completed a familiarity judgment for visually-presented single word. The stimuli, drawn from a large, publicly available and human-annotated dataset, comprised 1,588 unique words spanning a large range of lexico-semantic and psycholinguistic features. For each subject and session, we ran parametric modulation analyses to examine covariation of the fMRI-BOLD response with low-level visual, lexical and semantic aspects of the single words. Concerning single word processing during the familiarity task, features of the words modulated activity differentially across the hemispheres, with larger effects in the left hemisphere. Activity in bilateral occipital and left frontoparietal areas covaried with orthographic complexity, while word concreteness modulated activity around the left central sulcus and right occipital lobe. Finally, sensorimotor features modulated a broad network of regions covering a large portion of the left hemisphere and right occipitotemporal and parietal regions. Our results demonstrate that lexico-semantic features of single words differentially modulate neural activity across the cortex, with frequency and sensorimotor features of the words showing the greatest and most heterogeneous modulatory pattern. Our results highlight the quality of this soon-to-be-released dataset and enable further investigation of the organization of lexico-semantic processing at the individual level.

Movement vigor changes with reward magnitude: movements exhibit shorter reaction time and increased velocity when directed toward more rewarding stimuli. But what if the reward is actively turned down? In interactive economic games, people turn down “free money” to punish players who have treated them unfairly. How vigorously do they move in punishing an unfair offer? Is vigor proportional to the incurred self-cost? Or does it rather reflect the cost inflicted to the unfair other? Here, we combined experimental economics and kinematics to investigate how self-cost, other-cost, and the effectiveness of punishments (i.e., the factor by which the self-cost reduces the punished player’s income) influence movement vigor. In two studies, we used motion capture to track the arm kinematics of participants while they played a motor version of the Ultimatum Game and the Trust Game. Results showed that as the reward magnitude increased, vigor increased for accepted offers, but decreased for rejected offers (Study 1). However, what was crucial in modulating vigor was not the self-cost or the other-cost, but their relationship, that is, the effectiveness of the punishment (Study 2). These findings imply that in deviating from optimal game theoretic predictions, participants accurately compute the cost of punishments.

The COVID-19 pandemic has been associated with increased susceptibility to the adverse effects of SARS-CoV-2 infection and has led to changes in physical and mental health, often associated with declines in cognitive reserve in the older population. Because the long-term effects of the pandemic are not yet known, we conducted a longitudinal study of healthy adults aged 60 years or older who underwent neuropsychological testing before and during the pandemic. At follow-up, they were shown images reminiscent of the pandemic, while some physiological measures were recorded to assess emotional regulation. We hypothesized that global cognition might play a protective role against negative changes in mood and emotional regulation.

Empirical studies have found a processing asymmetry between Chinese subject-extracted relative clauses (SRCs) and object-extracted relative clauses (ORCs), whose processing difficulty is manipulated by merely changing word orders. Still, there is no consensus on how this SRC-ORC asymmetry occurs. Thus, we conducted an electroencephalography (EEG) study to examine the comprehension of Chinese SRCs and ORCs and aimed to elucidate how the neural activity, in the forms of both event-related potentials (ERPs) and brain oscillations (i.e., event-related synchronization/desynchronization, ERS/ERD) attuned to sentences with different levels of processing difficulty. The results showed an N400 and a P600 effect when comparing SRCs and ORCs. Simultaneously, alpha ERS was associated with an N400 effect and theta ERS with a P600 effect. By incorporating the ERP and ERS indexes, we propose that the dissociation between the integration and retrieval effort involved in sentence comprehension may account for the processing asymmetry between different relative clause sentences.

Hand somatosensory deficits, which are a frequent consequence of stroke, considerably reduce the patient’s dexterity and quality of life. Yet, research on touch rehabilitation remains scarce. One notable approach in this regard is Repetitive Somatosensory Stimulation (RSS), known to improve tactile acuity of the stimulated finger by inducing transient plastic changes in the corresponding primary somatosensory (S1) representation after passive mechanical stimulation. While RSS-induced tactile improvement has long been thought to be local (i.e., specific to the stimulated region), recent work from our group showed remote tactile improvement on the unstimulated hand. Specifically, RSS on the right index finger (rD2) induced tactile improvement at this finger as well as at the left thumb (lD1) and middle finger (lD3), while tactile acuity of the fingers adjacent to rD2 (right D1 and D3) remained stable. While this finding showcases the rehabilitative potential of RSS to reduce somatosensory deficits even remotely, the physiological mechanisms underlying these remote effects remain unknown. Given the specific pattern of improvement observed on each hand, we hypothesize that RSS produces remote tactile improvements by modulating inhibitory processes between S1 fingers’ representations both within (lateral inhibition) and between (interhemispheric inhibition) hemispheres. To test this hypothesis, we conducted a double-blind sham-controlled study in 40 healthy volunteers undergoing, before and after 3 hours of RSS on rD2, electroencephalography (EEG) recordings of somatosensory evoked potentials allowing to indirectly measure the level of inhibition between digits both within (rD2-rD3 and lD2-lD3) and between (rD2-lD2 and rD3-lD3) the two hands (i.e., double stimulation paradigm). Tactile acuity was additionally assessed on these fingers using the 2-point discrimination task. Preliminary results from 21 participants tend to replicate the transfer of improvement to the other hand following RSS on rD2, with tactile improvement on rD2 and lD3 and unchanged acuity on lD2 and rD3. EEG data showed a trend for an increased inhibition between rD2 and rD3 as well as between rD2 and lD2. These preliminary results suggest that RSS remote effects may be mediated by increased interhemispheric inhibition between homologous S1 regions. The results of this study will shed light on the plasticity mechanisms underlying interhemispheric transfers of tactile improvement.

When parsing sentences, comprehenders tend to interpret role-ambiguous arguments as the agent of the sentence. This agent preference is robust cross-linguistically and is possibly linked to a general cognitive bias for agents in cognition. Despite this ample literature, it is fairly unknown whether children can use semantic role information to guide sentence processing. In an auditory EEG experiment, we explored the interpretation of ambiguous noun phrases in 6-year-old Basque children (N=32). We exposed children to subject-verb intransitive sentences and we manipulated whether the initial noun phrase was ambiguously marked or not, and whether its role was agent or patient. We hypothesised that if 6-year-old children already use semantic role information, disambiguations to patients should lead to reanalysis effects, in line with the agent preference in adults. We additionally presented children with semantically congruent and incongruent sentences as control conditions for semantic reanalysis effects. We analyzed the ERP results as well as power decreases and increases in alpha and beta bands with generalized additive mixel models (GAMMs). As expected, ERP results showed an N400 effect for the incongruent sentences compared to congruent ones. Critically, disambiguations to patient also elicited a negativity over posterior and central electrodes compared to disambiguations to agents. We interpret this negativity as signaling a reanalysis effect, consistent with the agent preference in adults. We are currently working on the analysis and interpretation of the results in the alpha and beta bands and we expect to have it ready by the summer school in June. In any case, our preliminary results suggest that 6-year-old Basque children already use semantic role information to guide sentence processing in a top-down fashion. Our results also allow to compare the neural correlates of semantic role disambiguation in children and adults, by comparing the results in the current

Introduction: Interoception is the process by which the nervous system senses, interprets, and integrates signals originating from within the body, providing a moment-by-moment mapping of the body’s internal landscape across conscious and unconscious levels. This is essential for homeostasis and allostasis. The heartbeat-evoked potential (HEP) is an implicit, electrophysiological marker of interoceptive processing. Studies have shown that attention to the body and arousal can modulate the HEP. However, HEP studies have been limited by the lack of consensus and standardisation in the processing methods, and correction of artefacts such as the cardiac field artefact. The aim of this study is to develop a signal-processing pipeline for HEP derivation. Methods: A scoping review of the literature is being conducted focusing on previously published HEP processing methods. This is to extract current processing techniques in the literature and attempt to develop a signal processing pipeline for HEP. Different methods extracted from the literature will be implemented and tested on available scalp EEG data (Temple University EEG dataset) to derive improved methods of HEP processing using advanced computational approaches in Python. The HEP will be studied by analysing its variability according to electrodes, heart rate thresholds, time window used in relation to R wave of the ECG, filtering and artefact correction methods (e.g. comparing ICA based approaches to Hjorth source derivation or current source density estimates) amongst others. Results and outcomes: As a future plan, following the scoping review and surface EEG analysis, intracranial and scalp EEG recordings in epilepsy individuals will be used to further investigate HEP processing methods. Indeed, HEPs derived from intracranial EEG recordings are free of cardiac field artefact and consequently provide a ground truth for patients with surface EEG recordings. The outcome of this research will permit a greater understanding of the gaps and inconsistencies in HEP processing methods used to date in the field. The signal processing pipeline hoped to be achieved, will allow standardisation in the field and a better understanding and use of the HEP with a clinical application in the future. This first stage of this research including the scoping review and HEP processing methods studied in available scalp EEG data, will be completed by June 2023.

Previous fMRI research in our lab demonstrated a network predominantly in the right hemisphere for processing linguistic prosody. It comprised an auditory ventral pathway connecting the posterior (pSTS) and anterior superior temporal sulcus (aSTS), and auditory-motor dorsal pathways connecting pSTS and the inferior frontal gyrus (IFG) and premotor cortex (PMC). However, it remains to be shown how these areas represent prosodic information over time. To this end, we collected magnetoencephalography data from 34 native German listeners while they listened to single words (“Bar” [bar], “Paar” [pair]) that gradually varied in prosody (statement – question) and word-initial phoneme (/b/ – /p/) along orthogonal continua generated by audio morphing. Participants categorized these words in terms of either prosody or phoneme in alternating blocks. We analyzed the data in two major steps. Firstly, we compared the source-localized neural activity evoked during the two tasks to replicate the prosody network identified in our previous study. A searchlight-based multivariate pattern analysis was also conducted using a linear task-classifier to complement the univariate comparison. These two analyses replicated our previous fMRI findings by identifying a distributed bilateral frontotemporal network that involved the left insula and the right PMC, STS, and IFG from 400 ms after the stimulus onset. These regions may reflect the decision-making processes for categorizing speech prosody. Next, a searchlight-based representational similarity analysis (RSA) was carried out to identify regions and time points that represent the acoustics of the prosodic contours and the abstract categories of statement or question. Therefore, the neural activity patterns across stimuli were compared with the modeled patterns based on either the acoustic or categorical dissimilarity of the stimuli. The searchlight RSA revealed that right temporal areas predominantly represented prosody acoustically and categorically, particularly during the prosody task. The prosodic representations emerged in early (100-200 ms) and late (around 450 ms) time windows. These recurrent representations could originate from feedback connections from e.g., the right PMC. This will be further investigated with a directed information transfer analysis probing the connectivity between the involved areas. Our study illuminates the dynamic processing cascade for the perception and evaluation of speech prosody in the brain.

The goal of the project was to explore the inter-subject analysis capabilities given by the M-CCA algorithm applied to an MEG data set. To achieve this, we selected an MEG data set that contained recordings of subjects listening to naturalistic speech stimuli, so we could track the sound envelope of this stimuli with the shared space created by the M-CCA procedure. Research has demonstrated that the sound envelope is closely related to the neural activity in the auditory cortex. Using the M-CCA procedure we were able to extract the more relevant components that were active in the data, and that were shared between participants. Correlation analysis shows that the relationship between the envelope and the canonical components is similar throughout participants. When fitting a linear model using the envelope as a dependent and the canonical components as independent it shows that the components have a different predictive power and explained variance. The sound envelope shows a relationship with our components, so the envelope could be used as a reference point of recording auditory stimuli. The M-CCA procedure shows relevance for inter-subject data analysis in MEG data.

According to ideomotor theory, actions are represented by their sensory outcomes, or action effects, and anticipating action effects is thought to enhance performance of motor tasks. In sequence learning, sequences of actions produce sequences of action effects, and the chaining of those action effects is thought to promote motor chunking. Behavioral studies suggest that the anticipation of action effects operates at the selection stage of action. This study aims to investigate whether action effects are integrated during the planning stage of a motor sequence, using magnetoencephalography (MEG). Previous research on the planning stage of sequences demonstrated Competitive Queueing (CQ), where representations of upcoming elements in a motor sequence, such as fingers or temporal intervals, exhibit a gradient of strength that is analogous to their order within the sequence: the first finger is represented the strongest, the second finger is represented the second strongest and so on. We hypothesize that action effects will exhibit the same CQ effect. Participants will learn four sequences over two days, associating each sequence with an abstract image cue. Each key press will produce a tone, such that performing a sequence will produce a melody. During MEG recording, participants will produce the sequences from memory when probed with the associated cue. In separate blocks, they will also be passively exposed to the learned melodies. We aim to decode representations of anticipated action-effects before movement onset by training a classifier on MEG data during passive exposure. The anticipated tones of the melodies are expected to follow the CQ gradient, indicating that chained action effects are indeed integrated into action planning of sequences.

Purpose: Attention deficit hyperactivity disorder (ADHD) is a common childhood neurodevelopmental disorder occurring in about 5% of children. About 4% of children and youths with ADHD will continue to have life impairing symptoms like inattention, impulsivity in adulthood. It is demonstrated that ADHD is a highly heritable disorder and, in addition to genetic component, multiple environmental risk factors have also been identified. Emerging evidence indicates that environmental risk factors of ADHD can in turn cause epigenetic modifications, which may play a role in the pathogenesis of the disorder. Studies have also shown that multiple drugs have the potential to cause epigenetic modifications and therefore mediate the treatment response. To our knowledge the epigenetic effect of the common drugs used to treat ADHD is not thoroughly examined. Aims: The aim of the study is to determine the effect of methylphenidate (MPH) and atomoxetine (ATX) on epigenetic modifiers conducting DNA methylation and demethylation in human peripheral blood mononuclear cells (PBMCs). We chose MPH and ATX for our study, because MPH is the first line pharmacological treatment option for children and adults with ADHD and ATX is used, when stimulant drugs like MPH prove to be ineffective or not tolerated by the patient. Changes in epigenetic modifiers in different brain regions are associated with changes in neuroplasticity and thereby may affect the treatment response of the drugs. Methods: We measured the effect of MPH and ATX on epigenetic modifiers called DNA methyltransferases (DNMTs) and Ten-eleven translocation (TET) enzymes in human PBMCs. In our experiment, cells were incubated with MPH, ATX or vehicle for 2 hours on four consecutive days. MPH and ATX were used in 3 different doses, which were either pharmacologically relevant or higher. PBMCs were collected 24 h after the last incubation. We used qPCR to measure mRNA levels of DNMTs and TETs. Results: Our preliminary qPCR results showed that ATX decreased the mRNA levels of DNMTs and TETs. qPCR results for MPH showed the opposite effect and increased mRNA levels of DNMTs and TETs. Conclusions: Our preliminary results showed that both drugs caused changes in the methylation pattern, which may contribute to their therapeutic effect. These results correlate with one previously published paper, which found that DNA methylation of DAT1 gene may be involved with the treatment response to methylphenidate.

Introduction Subthalamic deep brain stimulation (STN-DBS) is a neurosurgical intervention for patients suffering from motor symptoms of Parkinson’s disease (PD). It is well established that STN-DBS has positive effects on the motor symptoms of PD. However, how STN-DBS affects other non-motor domains of patients’ lives still remains unclear. It has been shown that anatomical and functional connectivity of the DBS site have an effect on treatment outcomes. Here, we present pilot data and a study plan to examine the relationship between different stimulation metrics (functional/structural connectivity and VAT-STN overlay) and instrumental activities of daily living (IADL). Methods Twenty patients were selected from a cohort of DBS-implanted PD patients at the General University Hospital in Prague. To measure IADL scores, we used the Functional Activities Questionnaire (FAQ). We collected self-ratings at baseline and 1 year after surgery to calculate the percentual change in score. Neuroimaging data were processed in the following way using the LEAD software: (1) CT & MRI co-registration, (2) normalization to MNI space + brain shift correction, (3) electrode localization, (4) volumes of activated tissue (VAT) extraction, (5) seed-based brain structural and functional connectivity from the VATs using a normative connectome, and (6) percentual overlay of VAT on STN. Results We found that patients in the ‘improvement’ group exhibited more functional connectivity between the VAT and a cluster of voxels in the anterior part of the right precuneus (p = 0.027). We used these results to define the ROI for the structural connectivity analysis. The structural analysis did not reveal any relationship between the ROI and change in FAQ. Lastly, the results from the VAT-STN overlay analysis did not show any significant correlations between percentual overlay and percentual change in FAQ after DBS. The results lay a foundation for the future directions of the study. Future Directions Given the heterogeneity of FAQ scores after DBS surgery, we will divide the study cohort into 3 groups according to these changes. Then, we will use the same process to compute functional and structural connectivity patterns. Additionally, VAT and STN overlap measures will be compared across groups. The overarching goal of this study is to map any differences in any of the 3 STN-DBS metric

Ever since Raymond and colleagues coined the term in 1992, the situations of attentional rivalry, where multiple stimuli compete for global access during the bottom-up processing, have been a major focus of attention for cognitive psychology and neurosciences. This phenomenon on visual cognition, called Attentional Blink, provides means for down-to-millisecond -behavioral and electrophysiological- assessment of temporal attention, it is sensitive to the sequentiality of the incoming information, and the processing of the stimuli can be modulated by the saliency of the provided information. One of the most salient properties in human cognition that can modulate the bottom-up dynamics is the structural processing. In this aspect, structures can be defined as mental functions bearing a causal relation across entities. Mental representations that inherit this relation can be considered as structured. Any effect that can modulate the processing of the structure can be considered salient; and any representation that have undergone this modulation, and that can be detected by a cognitive mechanism that is sensitive to it, can be considered as salient information. Our objective is to assess the structural processing on visual cognition and to examine to what degree the saliency of structural stimuli, namely T1 and T2 stimuli carrying salient structural natural language information can pass through the limitations of attentional resources. In an intuitive and straightforward manner, we base this saliency on the structural properties of functor-argument according to a Fregean account, in which the meaning of a complex expression is determined by not only the meanings of its constituents but also its structure. We will be testing this proposed structurality in a paradigm where functor (T1) is supposed to hinder the processing of the argument (T2) from the attentional limitation point of view. This study is important because it will provide us results that are important for understanding how the saliency of the structured representations holds under unconscious perception. We hypothesize that the structurally salient stimuli will be able to break through the AB limitations in a scenario where T1 is supposed to facilitate the processing of T2 from the compositional point of view, and that we will be able to observe different ERP dynamics across structurally salient and non-salient stimuli.

Proper name anomia is a common experience that can become amplified in patients with a diagnosis of dementia (PWD). The Gotcha! app aims to provide practice-based therapy for PWD to relearn the names of key people in their lives. It has been developed according to the principles of errorless learning, which have previously been shown to improve the remembering the familiar people’s names and benefit the relationship between the PWD and their loved ones. (Clare et al, 1999, 2000, 2003). Methods: Gotcha! is a digital confrontation naming therapy app which enables patients to train one face per day by using photos that the app represents. During the development phase we carried our qualitative research (thematic analysis) on why PWD get involved in research projects such as ours. Gotcha! therapy block lasts for six weeks and prior to the therapy patients complete a multiple baseline paradigm with eight weekly tests of free naming of the to-be trained faces. During the therapy, a novel speech verifier is used to provide real-time feedback (Barbera et al. 2020). Two analyses method is used to investigate the behavioural data: 1) within-subject non-parametric analysis using Tau-U metric (Parker et al. 2011); 2) a parametric group analysis using an ANOVA. Results: The thematic analysis revealed four themes that will be discussed in more detail on the talk.In terms of the quantitative data, our results from the first 16 subjects showed: 1) Tau-U. 73% showed a positive trend with better naming during the training phase with 5/10 reaching statistical significance. 2) ANOVA demonstrated a significant effect at the group level of training>baseline phase, F(1,9) = 6.68, p = .029. Conclusion: App-based proper name anomia retraining works for the majority of PWD in our trial thus far. Being able to freely recall and produce the name of a relative or loved one has a big impact on people’s lives.

Encoding new memories not only takes place simultaneously with occurring events but also against the backdrop of a rich library of information acquired through one’s life. Previous studies show that prior knowledge (e.g. schemas) strengthens the encoding and accelerates the recall of new memories that are both in agreement with (congruent), or in opposition (incongruent) to that previous knowledge. The contradictions between these two lines of research have not been resolved yet. A third suggestion of how prior knowledge influences memory comes from a recent framework, SLIMM (van Kesteren et al., 2012), which postulates that learning shows a non-linear, U-shaped function with degrees of congruency to prior information. However, the SLIMM model remains under scrutiny as not all of its hypotheses have been successfully tested yet. Furthermore, the neural underpinnings of such learning processes remain unknown. While the SLIMM model predicts a trade-off between the mPFC and MTL structures for congruent and incongruent effects respectively, other models predict an essential role of MTL structures in encoding congruent information. In this PhD project, we aim to use behavioural methods as well as neuroimaging techniques (fMRI) in order to understand whether and how prior knowledge structures enhance the encoding and retrieval of new events. Using a novel spatial schema paradigm, we will compare three conditions with varying degrees of congruency to previous knowledge in order to test the three seemingly contradictory patterns of findings in the literature. Additionally, we will use fMRI to directly compare learning systems in the brain that support learning under certain (congruent) and uncertain (incongruent) conditions. This will allow us to offer a refined neuroscientific model of how brain networks interact to successfully integrate new information with previous knowledge schemas. Importantly, to the best of our knowledge, we will be the first to use a machine learning classifier to decode schemas in the brain by investigating which brain networks are responsible for representing knowledge structures and integrating new learning into them. Understanding how learning is influenced by previous knowledge bear implications for improving clinical conditions, educational methods and machine learning techniques.

Statistical learning (SL) is the ability to extract statistical regularities from the environment. One of the domains in which SL plays a pivotal role is language acquisition. Many prior studies employed artificial languages to study how humans extract transitional probabilities (TPs) between neighboring syllables to infer and learn new artificial words. Recently, M/EEG has been employed to study whether brain rhythms align to these statistical patterns in speech. These studies used so-called neural frequency-tagging paradigms, where brain activity is recorded while a stream of syllables is presented at a fixed rate (e.g., 3 Hz). Sequences contain hidden, yet rhythmic patterns of low-TP events at a different rate (e.g., 1 Hz). The rationale is that the brain should track not only the 3-Hz acoustic rate (present in the stimulus), but also the abstract 1-Hz rate (computed internally by the brain). We here address a major confound of this prior work that limits interpretability: Most prior studies did not account for the possibility that the statistical regularities are confounded by regular acoustic events, in particular recurring phonetic features, at the same rate of TP patterns. With such confounds, it cannot be excluded that the brain tracks only phonetic regularities rather than statistical patterns. The present work presents a new algorithm that controls for both phonetic and statistical regularities. Our algorithm features a stationarity index to quantify the presence of phonetic patterns at the frequency of interest. The algorithm generates artificial language stimuli that minimize the stationarity of phonetic features in the acoustic stream of syllables. In addition, our algorithm employs a random-walk approach to stabilize the variance of TPs within streams of arbitrary length. We show that our algorithm yields a lower stationarity index of phonetic features and more stable TPs throughout the syllable stream, compared to less sophisticated randomization approaches used in previous studies. This method can be used to generate more controlled stimuli in future artificial language experiments in order to reduce the impact of potential lurking variables from the implications of linguistic SL findings. The algorithm will be made available publicly to the research community.

Finding locations in a familiar environment requires accurate representations of space (‘cognitive maps’). In many species, specialized neurons in the hippocampal formation provide the neural basis for cognitive maps. Previous research has found that stretching or squashing environmental dimensions directly controls the firing pattern of hippocampal place cells and entorhinal grid cells in freely moving rodents as well as human homing behaviour. This Indicates that boundaries defining the geometry of space play an important role in determining the nature of cognitive maps. Here, we examined how behavioural changes to environmental deformations relate to those on a neural level as measured with functional magnetic resonance imaging (fMRI) in humans, and how these effects can be explained by models of neuronal firing. In this two-day study, we first trained participants to learn the location of multiple objects inside a virtual arena. During subsequent scanning, we asked them to re-visit each location inside the arena as well as actively imagine them outside the arena. Critically, the arena deformed from square to rectangular shape across days. So far, we found systematic shifts in spatial memory related to the change in geometry that is best explained by a computational model of so-called boundary vector cells (BVCs) which respond to the presence of environmental boundaries. Furthermore, our fMRI data suggest that the brain uses spatial codes to represent locations and distances that scale with the current geometry of the environment. Our study expands our understanding of how behavioral and macroscopic neuronal adjustments to environmental deformations are linked in humans.

Caregiver-child relationships are crucial for developing children's emotion regulation capacities and subsequent psychosocial functioning. However, experiencing childhood maltreatment, including abuse and neglect, increases the risk of psychopathology. Dyadic synchrony has been proposed as a key mechanism impacting that development, and previous studies have shown that parent-child dyads with maltreatment experience display distinct patterns of cardiac co-regulation during stress (Lunkenheimer et al., 2018). Yet, little is known about how cardiac synchrony behaves during adolescence. To fill this research gap, we will investigate cardiac synchrony during an adapted Trust Game in adolescent-caregiver dyads with and without experiences of childhood maltreatment. The study will include ~150 dyads from a large-scale longitudinal study in Germany. We will assess maltreatment subtypes, chronicity, and severity through the Maltreatment Classification Interview and System (Barnett, Manly & Cicchetti, 1993). During the adapted Trust Game, adolescents will interact in alternating rounds with a peer and their respective caregiver, with the peer being a pre-programmed algorithm that cooperates little to induce stress. Cardiac synchrony will be calculated using Local Power (Bornemann et al., 2016), a short-term estimate of high-frequency Heart Rate Variability (hf-HRV) that allows for investigation of hf-HRV shifts within seconds. We hypothesise that dyads with maltreatment experience will display distinct cardiac synchrony signatures compared to dyads without such experiences. Additionally, we plan to explore maltreatment subtype-specific effects. As data collection is ongoing, we will present study design, set-up, hypotheses, and planned statistical analyses.

Humans effortlessly extract and re-apply the relational structure of our experiences to new problems. Prior research has attributed the underlying mechanisms of abstraction and generalisation to the hippocampal formation, which has a well-established role in mapping out both spatial and non-spatial stimuli. Nonetheless, existing experimental evidence does not clearly answer how these ‘cognitive maps’ guide complex behaviours. In models of the hippocampal formation, actions allow transitions between different states - but a clear neural substrate for action remains to be identified. We suggest that cognitive maps trace out action pathways which determine transitions through neural state space. These action pathways can then be mapped to new states, thereby both enabling and constraining how we use memory to solve problems. Inspired by work in ecological perception, we propose that at a macro level this can be seen as an encoding of (abstract) action affordances. To test this, we have designed an fMRI experiment in which participants learn a series of repeating mathematical rules. Using representational similarity analysis (RSA), we will then test for a representation of affordances. Specifically, we predict that states which afford the same actions will provoke a more similar neural representation compared to states with differing affordances. Crucially, our design allows us to tease apart the role of action (defined by addition and subtraction operations) from a positional code. In the present poster, we present results from pilot behavioural data which indicates that participants can learn and generalise a set of numerical operations. In sum, we hope to elucidate how structural knowledge can guide behaviour through action coding. This should extend our understanding of how the hippocampal formation uses relational memory to guide future action.

The hippocampal-entorhinal system is remarkably efficient in organizing relations between sensory stimuli, such as state transition probabilities, in a cognitive map. Such knowledge representation is assumed to enable fast learning of novel relations and the generalization of reward, likely facilitating goal-directed behavior. However, in addition to experienced transition probabilities, objects may simultaneously share other types of relational information, such as reward contingencies. This fMRI study investigates how the neural representation of relational knowledge is influenced by a subsequently learned latent reward structure, whether the structural relations between objects still are represented veridically. Participants first acquire knowledge about object relations based on object transitions that follow a hidden graph structure. In a subsequent decision-making task, each object gets associated with fluctuating reward values. Critically, two parts of the graph structure share the same reward contingencies (orthogonal to the initial graph representation). Behavioral data suggest that participants successfully acquire structural knowledge and can utilize it to find shortcuts when transitioning around the graph. Model-based analysis reveals that participants can extract the additional underlying latent reward structure and generalize over states sharing the same reward contingencies. Furthermore, they can apply their acquired structural knowledge to correctly infer current reward values of objects whose values they never directly experienced. Preliminary RSA results reveal hippocampal and vmPFC activation patterns reflecting structure and reward contingency representations. Suggesting participants represent both experienced transitions and shared reward contingencies between objects; moreover, they can combine both types of information for generalization and inference.

Short-range connections in the language network are largely unexplored. Broca's area in the left inferior frontal gyrus, which includes Brodmann areas (BA)44 and BA45, is a central node of the language network. These regions are tightly functionally connected and can inhibit each other during language processing. Furthermore, BA44 appears to be structurally and functionally heterogeneous: its anterior part is associated with language processing while the posterior part supports motor functions. Overall, the functional subdivision and interaction of BA44 and BA45 suggest the presence of short direct white matter connections between them, which however has never been directly detected. While long-range connections have been extensively studied, short-range association fibers (SAFs) in Broca's area remain unexplored - mainly due to the need for high resolution diffusion-weighted imaging (DWI) and specialized fiber models. We have recently developed and validated reliable SAF mapping methods using ultra-strong gradients from the Connectom scanner. We applied these methods to map SAF between subdivisions of Broca's area in a group of right-handed adults. Fiber tracts connecting BA44 and BA45 were detected bilaterally in all participants. At the group level, connectivity between BA44 and BA45 was left-lateralized, consistent with left-lateralization of language processing. We also found that anterior BA44 is stronger connected to BA45 than its posterior part, consistent with the described functional subdivision of BA44. For the first time, we mapped short-range connections between BA44 and BA45 with DWI in vivo at submillimeter resolution. The discovered SAF within the language network support previous work on receptor architecture and coactivation patterns in Broca's area and promise to fill the gaps in the current version of the language network.

Multiple sclerosis (MS) is a chronic autoimmune disease that causes progressive disability by affecting the central nervous system. Recent studies suggest that left-handedness may increase the risk of MS and lead to different disease outcomes compared to right-handed individuals. However, the underlying mechanisms of this association remain unclear. In this study, we aim to investigate the cellular and molecular mechanisms underlying the interaction between handedness and MS disease outcomes. We will recruit a cohort of MS patients and matched healthy controls. Using the lateralization quotient of the Edinburgh Handedness Inventory (EHI), varying handedness will be measured on an interval scale. We will use neuroimaging techniques such as functional magnetic resonance imaging (fMRI) and diffusion tensor imaging (DTI) to assess brain lateralization and connectivity. We will also perform functional assays to evaluate immune cell function and disease progression in the different patient groups. Finally, we will use the scale for assessment of somatic symptoms (SASS) and the World Health Organization Quality of Life questionnaire (WHOQOL) to assess patients' general level of physical and psychosocial functioning. The specific aims of our study are: [1] To investigate the relationship between handedness (as measured by EHI) and the risk of developing MS. [2] To examine the differences in brain lateralization and connectivity (as measured by fMRI and DTI) between MS patients and healthy controls, as well as between different subgroups of MS patients based on disease severity. [3] To explore the potential mediating or moderating effects of immune cell function and other variables in the relationship between brain lateralization and MS disease outcome. The findings of our study will provide new insights into the role of handedness in MS pathogenesis and disease outcomes, highlighting the importance of brain lateralization and connectivity in disease progression. This may lead to the development of new therapeutic interventions that target these mechanisms and improve the quality of life of MS patients.

The regular practice of mindfulness meditation (MM) and the repeated application of transcranial direct current stimulation (tDCS) have shown therapeutic benefits in the treatment of fibromyalgia syndrome (FMS). Anodal tDCS over the left primary motor cortex (M1) has been shown to enhance intracortical inhibition (ICI) in comparison to sham stimulation, correlated with improvement in FMS symptoms. On the other side, MM was shown to enhance GABA-mediated ICI. This project aimed to investigate the mechanistic effects of combining anodal tDCS over left M1 and MM in FMS patients trained in mindfulness, using transcranial magnetic stimulation (TMS) and recording motor evoked potentials (MEPs) of peripheral hand muscles. MEPs were measured at baseline, after a 4-week mindfulness training (MT) and after 2 weeks of 2 mA anodal tDCS over the left primary motor cortex (M1) paired with MM. Thirty FMS patients were assessed for corticospinal excitability (CSE) using single pulse TMS measures of resting motor threshold (RMT), recruitment curves (RECR) and stimulation intensity to give a mean MEP of 1 mV (SI1mV). Paired pulse TMS measures of short intracortical inhibition (SICI), long intracortical inhibition (LICI) and intracortical facilitation (ICF) were used to assess intracortical activity. Participants were randomly allocated to either receive anodal tDCS paired with MM or sham tDCS with MM. Based on previous findings, we hypothesized that MM will modulate CSE when combined with anodal tDCS of left M1, but not on its own. Moreover, the stimulation when delivered in combination with MM will further increase the ICI, in respect to MM only. The data are under analysis, however, this is the first study to assess the potential use of TMS measures of CSE and CI as a potential biomarker for MM and anodal tDCS therapies in FMS. (Registration ID: DRKS00029024)

Pain research in humans has mostly focused on the brain, even though the spinal cord is an initial processing site within the nociceptive pathway of the central nervous system and a key target of descending modulation. The application of functional magnetic resonance imaging (fMRI) to the human spinal cord is still a relatively young field of research and faces many challenges. Here we aimed to probe the limitations of task-based spinal fMRI by investigating the reliability of spinal cord BOLD responses to identical nociceptive stimulation across two consecutive days. In two fMRI sessions, 40 participants received painful heat stimuli on their left forearm. 3T fMRI data were acquired using a z-shimmed gradient-echo EPI sequence. Data were motion-corrected, high-pass filtered and denoised during first-level model estimation. Group-level analyses were carried out in a standardized space using multiple comparison correction via voxel-wise permutation testing. Test-retest reliability was quantified using the intraclass correlation coefficient (ICC). At the group level, we observed spatially specific BOLD responses at the expected location, but no spatial overlap in response patterns across days. While autonomic indicators of pain processing showed good-to-excellent reliability, both beta-estimates and z-scores of task-related BOLD responses showed poor reliability across days in an anatomical mask of the target region. However, using a slightly extended mask including the draining veins improved reliability. Heat pain stimuli as short as 1s are able to evoke a robust BOLD response in the ipsilateral dorsal horn in spinal cord segment C6, in line with previous studies that used more powerful stimulation parameters. However, on an individual level BOLD response patterns varied strongly within participants, resulting in poor reliability. Improvements in spatial normalization methods may help to improve reliability. Future research should investigate to what extent draining veins distort spatial specificity in spinal fMRI.

Theory-of-Mind (ToM) is the ability to reason about others’ beliefs and thoughts. Just as Premack and Woodruff (1978) originally identified ToM as ‘a system of inference’, the developmental relations between ToM and certain domains of abstract reasoning have been found (Perner et al., 2012). And both ToM and abstract reasoning show similar age-related changes in 3 to 5 years old (Frye et al., 1995). Therefore, we hypothesize that ToM will be related to other domains of abstract reasoning like relational reasoning and possibility reasoning. Interestingly, children from different cultural backgrounds have shown different developmental sequences in ToM tasks, for instance, Chinese children acquire knowledge attribution earlier while US children understand diverse beliefs earlier. Moreover, children from collectivistic societies may show a more relational-oriented tendency in relational reasoning tasks than children from individualistic societies. As we propose that ToM and abstract reasoning share a similar cognitive process – making a judgment based on one condition while ignoring the other conflicting one, inhibitory control will be necessary for success on both mental and non-mental reasoning tasks. Inhibitory control is more emphasized in collectivistic societies (e.g., China) than in individualistic societies (e.g., Germany). For instance, previous studies have shown that Chinese children have better performance on inhibitory control tasks than US children. Therefore, we aim to explore (1) the developmental sequence and trajectory in ToM and abstract reasoning across cultures; (2) the relationship between abstract reasoning and ToM across cultures; (3) the role of inhibitory control in the development of both abstract reasoning and ToM across cultures. In this study, we will recruit 3 to 5 years old children from China and Germany to participate in an online experiment. We will assess the different aspects of ToM, including diverse desires, diverse beliefs, knowledge, and explicit ToM. Abstract reasoning about relations and about possibilities will be measured, as well as the inhibitory control.
Keywords: theory-of-mind, abstract reasoning, inhibitory control, cross-cultural, preschooler

Understanding spoken words requires several processing steps, from the acoustic analysis of sound sequences to accessing lexical representations. While low-level processing of auditory stimuli, such as speech sound discrimination, is common among animals, evidence for the recognition of sequences of sounds and the understanding of their meaning is scarce in non-humans. Nevertheless, recent reports on dogs’ lexical processing capacities suggest that even non-primate mammals may process higher-level information in speech. However, it is debated whether dogs have similar abilities to humans in accessing the phonetic details of words during lexical processing. While a behavioural study has suggested that dogs may attribute a meaning-changing role to a single speech sound in a word, an electroencephalography (EEG) experiment has found no evidence for it. It is also an open question whether the previously identified brain areas involved in lexical processing are sensitive to the phonetic details of words. The aim of the present study was to investigate speech processing in the canine brain, with a particular focus on phonetic sensitivity during lexical processing. Twenty awake, unrestrained dogs were presented with (1) commands, (2) phonetically similar (differing in only one vowel) and (3) dissimilar nonsense words during the experiment, while the elicited brain activity was measured using functional magnetic resonance imaging (fMRI). In a whole-brain analysis, we found extensive bilateral auditory cortical activity when comparing all vocal stimuli to silence. Using region of interest analyses, we also found that within certain brain areas involved in lexical processing (mid and caudal ectosylvian gyrus, as identified in previous fMRI studies), the original commands and their phonetically similar forms were processed differently than dissimilar nonwords. However, the activity elicited by words and phonetically similar nonwords did not differ. These findings confirm the presence of lexical representations in dogs’ near-primary and secondary auditory cortex, but also shed light on the limited capacity of these brain areas to access the phonetic details of words.

A key adaptive function of anxiety is to increase vigilance and neurophysiological arousal towards potential threats in the environment. Previous studies suggest that heightened anxiety may be associated with alterations in face recognition, albeit with mixed results. In this study we therefore dissociated the encoding and retrieval stage of face recognition memory to determine how these are differentially affected by threat-of-shock induced anxiety. We tested participants from the general population (N=93) on an emotional face memory task, where images of emotional faces are presented during encoding and retrieval, either under threat-of-shock or safety. Using factorial analysis of variance (ANOVA) we investigated how being under threat-of-shock during encoding and retrieval affected participant’s face recognition accuracy at retrieval. We also combined our sample (N=93) with the data from previous studies which used the same task (N= 86 in Bolton & Robinson, 2017; N=32 in Garibbo et al., 2019) to conduct a meta-analysis of the standardized effect sizes (total N=211). In our sample (N=93) we found a significant main effect of encoding state (F=4.246, p=0.042, ηp2=0.044), with more accurate retrieval for faces encoded during safety (M=0.668, sd=0.172) compared to threat-of-shock (M=0.652, sd=0.157). We also found a significant main effect of retrieval state (F=10.495, p=0.002, ηp2=0.102), with higher face retrieval accuracy under threat-of-shock (M=0.675, sd=0.163) than safety (M=0.645, sd=0.166). Encoding and retrieval state did not interact. The random effects meta-analysis confirmed a significant pooled effect size for encoding (Cohen’s d=0.26, 95% CI=0.12-0.41, z=3.52, p<0.001) but not retrieval state (Cohen’s d =0.13, 95% CI=-0.02-0.31, z=1.73, p=0.084) across all studies. We conclude that threat-of-shock induced anxiety during the encoding stage appears to robustly impair subsequent face recognition. This may result from the attentional demands of a heightened anxious arousal state out competing the attentional resources required to encode faces, which are subsequently retrieved less accurately. Further analyses will investigate the neural activations and amygdala-cortical circuit connectivity underlying these behavioural effects.

Since the first report of transcutaneous spinal direct current stimulation (tsDCS) on humans in 20081, more than 50 experimental studies have been published. However, systematic reports of adverse effects (AEs) of tsDCS are scarce. In this pilot-study, we aimed to systematically record tsDCS AEs via a structured questionnaire. Twenty-two healthy participants (5 females, 20-40 years old) were recruited for this study. In study I, twelve of them underwent a single anodal tsDCS session with stimulation at 2.5 mA for 20 min and 15-second fade-in/fade-out periods. In study II, ten of them underwent three consecutive sessions (at least 1 week apart) with active (anodal/cathodal) or sham stimulation. A pair of rubber-electrodes was placed over the twelfth thoracic vertebra (anode) and the suprascapular region (cathode). A tsDCS adverse effects questionnaire (adapted from Brunoni et al., 20112) was filled in by participants directly after tsDCS termination. Several potential AEs and whether they were deemed to be associated with tsDCS were recorded. In study I, skin redness (91.67%), burning sensation (66.67%) and tingling sensation (58.33%) were most reported, with a lower proportion of people (25%) reporting a brief itching sensation after the single session tsDCS. We also observed a significant correlation between initial electrode impedance and the number of AEs reported. In study II, the electrode impedance was decreased by wiping the skin with alcohol, all the AEs were less observed: skin redness (53.33%), tinging (43.33%), burning (40%) and itching (10%). Most AEs occurred within the first minute of tsDCS onset, lasted for 1-2 minutes and solely occurred within the skin area beneath the electrodes. The participants could not distinguish between active and sham stimulation. Skin (sensory) problems at the site of electrode contact are the main reported AEs. Under the premise of ensuring safety, lowering the impedance of tsDCS stimulation electrodes reduce the occurrence of AEs. It is necessary to consider using sufficiently long stimulus durations (e.g., ≥ 1 min) for sham settings to achieve a better blinding effect. The observed onset and duration of AEs also serve as guidelines for designing sham conditions for better blinding. This pilot-study provides a systematic recording of tsDCS AEs in healthy volunteers. As the number of publications in this field increases, the need for safety-oriented explorations grow stronger.

Are cultures becoming more homogenous as a result of one-way global cultural flows? Or are we instead moving towards a cosmopolitan world with greater cultural diversity? What are the layers and substructures that transmit and sustain these cultural flows? Empirically addressing these questions has been challenging due to the lack of large-scale comparative data with high temporal and geographical resolution. Here, we quantify cultural flows through temporal patterns of music consumption spanning not only countries but also cities over extensive time periods. First, we quantify cultural flows and study the underlying structure of the global network by which songs spread over time and space. Second, we examine the impact of global and local structures on cultural diffusion processes. We measured daily music consumption via Shazam, a popular mobile application that allows users to identify music from short audio samples. For a period of two years, we collected top songs that were most searched among 1,423 cities in 53 countries. The resulting dataset includes 17.8 million music consumption events and 112,568 unique song cascades (temporal and spatial trajectories of songs from beginning to end). By combining methods from network science, music information retrieval, and computational sociology, we infer the underlying network of cultural diffusion and examine the role that cities play and the structures they form. We find that the global structure of the diffusion network can be characterised as a complex polycentric network consisting of 21 distinct cultural communities. They differ significantly in their patterns of music consumption and reflect cultural communities beyond national borders. Second, we show that cities play different roles in cultural diffusion, depending on how they are connected with other cities within and across these communities, forming four distinct roles: cultural bridges, gatekeepers, national hubs, and peripheries. Cultural bridges and gatekeepers are densely populated cities with more musical diversity, adopt musical trends earlier on, and act as a hub for cultural mixture and globalisation. By contrast, national and peripheral cities are smaller and less internationally connected, but nonetheless facilitate national transmission and maintenance of local culture.

In previous studies, infants younger than 2 years have demonstrated correct expectations of how an agent with a false belief will act. However, the robustness of these findings and the underlying mechanism remain a matter of debate. The altercentric account proposes that infants’’ own representation of the world may be biased by the view of others. Specifically, infants might misremember objects based on where others saw them, allowing them to predict where the other will search for the object later on. Here, we tested whether infants' object representation and memory is spontaneously modulated by an agent's belief. In a preregistered eye-tracking study, we presented infants, aged 8-10 and 17-19 months, with videos, in which an agent observes an object moving into one of two locations. Subsequently, the agent then either observes (True Belief) or misses (False Belief) the object's change of location from location A to B. We predict that infants' expectation of the objects' location is altercentrically modulated, i.e., that infants expect the object where the agent believes the object to be. We test whether infants who correctly anticipate the agent’s action in the false belief condition also expect the object in the believed rather than the real object location. We collect data following a preregistered bayesian sequential testing scheme and have currently tested N=136 children, results are expected by May 2023.

Unfamiliar individuals may trigger strong emotions in us upon first contact. One mechanism to explain this is social-cognitive transference (Andersen & Chen, 2002), whereby a cue on the target triggers the mental representations of a significant other which in turn are projected back upon the target, leading one to perceive the new target as similar to the familiar significant other. Indeed, mental representations of significant others guiding one’s social world is the core idea of many clinical and developmental theories of psychology (Bowlby, 1973). However, the boundary conditions of this process are underspecified to date. Specifically, it is unclear how selectively individuals generalize and relatedly what conditions may drive one to generalize more readily. The latter is especially relevant for maltreated youth, as one learns to rely under extreme stress on negatively biased, automatic processing of the social environment (Luyten & Blatt, 2016). Therefore, this study aims to fill these gaps by selectively activating parental representations of maltreated adolescents during a virtual ball-tossing paradigm to examine how similarity vs non-similarity to a parent influences their physiological and mnemonic responses to new peers. To this end, maltreated and non-maltreated adolescents (N ≈ 140) described one of their parents using open-ended statements via an online survey a week before their testing appointment at the lab. During testing a week after, they engaged in Cyberball (Williams & Jarvis, 2006) with two co-players who were previously introduced to them via descriptive blurbs. These blurbs constituted the experimental manipulation, whereby one of the co-players (Target) was presented as character-wise similar to one of the parents of the participant. An electrocardiogram (ECG) simultaneously recorded participants’ heart rates during Cyberball. After the game, participants completed a recognition-memory test on previously seen blurbs of two co-players. We expect to find that maltreated and non-maltreated adolescents exhibit differential mnemonic and cardiac responses to two co-players as a function of parental resemblance and their maltreatment status. The main hypothesis is that maltreated adolescents will display less pronounced immediate heart-rate deceleration following not-receiving-the-ball from Target vs Control, on the premise that heart-rate deceleration is a correlate of heightened attention to salient negative stimuli (Bradley, 2009).

The hippocampus plays a central role in the study of behavioral and neurological disorders. However, the typical analysis often focuses solely on simple features such as volume, overlooking the significance of global shape. In this study, we propose an analysis pipeline that utilizes a joint clustering-dimensional reduction algorithm applied to the spherical harmonics of hippocampi. Additionally, we examine demographic data, including age and dementia diagnosis, associated with the identified clusters.

Chemical synapses play an essential role in the transmission of information between neurons [1]. Depending on the released neurotransmitters in the synaptic cleft, positive or negative ions can access the postsynaptic neuron, contributing to the complex balance between inhibition and excitation [2]. Adding more complexity to the information transmission process of neurons, the role of short-term plasticity (STP) can be understood as a modulatory property of synapses, whose mechanisms -facilitation and depression- enhance or diminish the synaptic efficacy [3]. Using mathematical models of STP can let integrate this modulatory property in artificial Spiking neural networks (SNN). Each model possesses a set of parameters that can be tuned, and, depending of its level of abstraction, it can simulate a biological synapses with some degree of accuracy. In this poster, we use two models of STP to simulate three examples of facilitation and depression synapses in rats [4], [5]: the parallel fiber to the Purkinje cell (PF-PC), the Calyx of Held (CH), and the cerebral mossy fiber to granule cell synapses (MF-GC). Additionally, the frequency response of the models is used to analyse its potential use to represent the modulatory property of synapses in a network level.

Perception is an active process shaped by the interplay of bottom-up information arriving at the sensory systems and top-down priors based on previous experience. While knowledge about the world is usually acquired via repeated encounters, the Mooney disambiguation effect is a commonly known example of one-shot perceptual learning. Mooney images are ambiguous two-tone images which are, without additional information, devoid of meaning for the perceiver. However, after a single exposure to the corresponding unambiguous original image, the content of the Mooney images becomes readily available. Usually, this effect is assessed by asking participants before and after disambiguation whether they subjectively recognized the figure of the Mooney image, but this subjective measure has several limitations. The present experiments aimed at identifying indirect measures of this special case of one-shot-learning by adapting a task in which participants had to identify whether red dots appeared on or off the main figures of different Mooney images. We found that, following disambiguation, error rates (ER) decreased while as discrimination rates (d’) increased. These objective measures of gradual nature, which do neither require conscious insight nor explicit subjective examination, could potentially allow follow-up studies to manipulate perceptual learning and assess its neural basis through neuroimaging methods.

Early-life emergence of Theory of Mind (ToM) corresponds to functional changes that span a range of frontal, parietal, and temporal regions in the cerebral cortex. Gathering evidence suggests that the cerebellum also plays a key role in ToM processing. However, the role of the cerebellum in the early-life development of ToM remains elusive. In the current study, we investigated the functional contribution of the cerebellum to ToM in a sample of children (N = 41, age range: 3-12 years) with varying ToM abilities, and compared it to that of adults (N = 82, age range: 19-48 years). We identified multiple clusters in the posterior cerebellum (crus I, lobule IX) that were activated in response to an in-scanner ToM movie-watching task across adult and developmental samples. Further, children who passed an out-of-scanner assessment of ToM abilities demonstrated activation clusters in cerebellar crura I-II, which were consistent with activations in adults. Contrarily, children who failed the ToM assessment lacked activations in crura I-II. Dynamic causal modelling (DCM) analyses revealed unidirectional endogenous connections from the posterior cerebellum of children with ToM abilities to the cerebral ToM network, which were scarce in children without ToM abilities. DCM in adults demonstrated inverse connectivity patterns as a function of ToM, in which regions of the cerebral ToM network unidirectionally modified activity in the posterior cerebellum. Together, our results reveal functional changes in the posterior cerebellum associated with the early-life emergence of ToM, and support a domain-specific contribution of the developing cerebellum to ToM. Additionally, our findings are in line with conceptualizations of a gradient of cerebro-cerebellar connectivity in social cognitive processing. In this account, forward connections from the cerebellum to the cerebral cortex, which support the initial construction of social schemas, are more prominent in early life, but are gradually replaced by inverse cerebro-cerebellar connections which support automatic use of these schemas later in life.

Pre-Stimulus Heartbeat Evoked Potentials predict P300 and reaction time in an auditory Oddball Task: real effect or artifact? Introduction: Heartbeat evoked potentials (HEPs) refer to the brain's response to one's own heartbeat and they are thought to reflect the brain's ability to monitor and regulate cardiovascular function. Previous research has shown that HEPs play an important role in cognitive and emotional processing. Since HEP amplitudes are significantly higher during interoceptive than exteroceptive attention, it has been hypothesized that changes in HEP amplitude represent an attentional shift from external task-relevant to internal interoceptive stimuli. In this framework, we aimed to investigate whether electrophysiological activity and task performance in an auditory Novelty-Oddball-Task relate to HEP amplitudes in the pre-stimulus time window. Methods. We analyze a large dataset of EEG recordings acquired in the LIFE-adult-study (N= 1725, 60 - 82 years). The subjects performed a 15min Auditory Oddball task. On the single-trial level, event-related responses were calculated and sorted according to high and low P300 amplitude. In addition, trials were sorted based on reaction time (RT). We identified HEPs in the pre-stimulus time window and calculated their amplitude with respect to high/low P300 amplitude and RT. Results. We were able to replicate a previously observed inverse relationship between task evoked ERP and HEP amplitude, with higher HEPs being followed by lower ERPs (and slower RT). However, control analyses point to the spurious nature of the observed association. Several strategies were applied to circumvent this problem and investigate the genuine HEP to ERP relationship. Discussion. While we cannot rule out the possibility that pre-stimulus HEPs are related to task-evoked ERPs in auditory oddball tasks, we present alternative explanations that account for the observed effects. Our results underscore the importance of appropriate control analyses in HEP and ERP research.

Psychosomatic symptoms are physical manifestations of mental or emotional distress, usually caused by stress or anxiety, and can be caused by a variety of factors. When it comes to tactile memory bias, the term refers to the tendency for individuals to value certain touch experiences more than others, and to remember those experiences more vividly. This study investigated the association between tactile memory bias and psychosomatic symptoms in an attempt to find out if there was an association between them.

In this project, a binary sparse coding (BSC) algorithm, which is commonly used in machine learning, was employed to effectively and succinctly extract auditory features from naturalistic speech stimuli. The resulting reconstruction was evaluated using various statistical tests, and the goal was to incorporate these features into an encoding model that could predict BOLD signals. This was motivated by the belief that the brain uses sparse coding to represent sensory information. To achieve this, the stimuli were first converted into a Mel-frequency spectrogram. Second, the BSC algorithm was utilized to generate 200 basis functions and identify the most probable active segments within 100ms intervals. The potential of this framework to enhance our understanding of the neural mechanisms of hearing and could offer an alternative diagnostic tool for hearing impairments was highlighted in a video (LINK TO YOUR VIDEO; https://drive.google.com/file/d/1F_cnJkCjV_uN1ktT19O7wgv_ryDLz-l5/view), which was rewarded by the Center of Excellence for Hearing Research (Hearing4all). However, although the BSC algorithm successfully restored some of the original input, the consistency of the reconstruction varied across different segments and frequencies, possibly due to scaling issues in the algorithm or applying it to the entire stimulus instead of just the speech segment.

Small variations in movement kinematics transmit specific intention-related information (Becchio et al, 2018). In previous studies, we found that human observers are able to read some, but not all of this information (Cavallo et al, 2016; Patri et al., 2020; Montobbio et al., 2022). In this study, we examined whether the ability to read intention information from movement kinematics can be improved with training. The study enrolled ten neurotypical adults who completed a one-interval two-alternative forced-choice intention discrimination task: they had to discriminate the intention of observed reach-to-grasp-to-pour and reach-to-grasp-to-drink movements. This task was administered both before (pre-test) and after (post-test) a training session without providing any feedback. During the training, participants performed the same task as in the pre- and post-test; however, visual feedback was provided after each trial in the form of a video displaying the full action sequence. Pilot data revealed a moderate improvement in intention discrimination abilities after training. Previous studies have shown that children with Autism Spectrum Disorders (ASD) have difficulties in extracting the correct intention information from movement kinematics (Montobbio et al, 2022). If replicated in a larger sample of participants, the current results may be useful to inform the design of novel training protocols that promote intention reading in ASD.

Goal-directed navigation in a complex environment requires not only a precise estimate of the target location but also a thorough plan of a navigational route avoiding obstacles along the journey. While a recent study showed that neurons in the prefrontal cortex encode navigational goals (Basu et al., 2021), previous studies reported that prefrontal neurons represent an animal’s movement directions or routes (Ito et al., 2015), leading to a question of whether these neurons encode a goal and a route independently or conjunctively. To address this question, we designed a new maze in which a goal location can be approached via different routes, allowing us to isolate representations of goal and route. We recorded the activity of OFC neurons while a rat performed this task and discovered that the OFC’s goal representation during navigation is consistent regardless of the animal’s route choices. Furthermore, in accordance with the task rule where the animal is required to target one of the two goals alternately, the representation of the next goal autonomously emerges in the OFC right after the completion of the ongoing journey, revealing neural dynamics supporting goal alternation. Notably, we also found that it is possible to decode the animal’s next movement directions or routes from the same population of OFC neurons. Detailed analysis of ensemble neural dynamics using a dimensionality reduction method revealed that the representation of movement directions is embedded in a subspace that is orthogonal to the goal-encoding subspace. This orthogonal representation may allow downstream circuits to retrieve the information of goal and route independently, likely supporting an animal’s flexible navigation.

(Plan of a proposed study) Action perception, as a foundation for social interactions, enables us to decode and comprehend the actions performed by others. The neural basis of action perception has been linked to the action observation network (AON), which comprises frontal, parietal, and occipitotemporal regions (Caspers et al., 2010). Several studies have investigated the neural representational space of observed actions in young adults. Tucciarelli et al. (2019) found that semantic features explained the most variance in fMRI data. Other studies discovered that sociality (i.e., presence of social interaction) was the primary organizing dimension of action representations in young adults' brains (Dima et al., 2022; Tarhan & Konkle, 2020). However, it is unclear how older adults' representation of daily actions may differ from that of young adults. In our recent behavioral pilot study, we asked participants to view naturalistic static photos of daily actions presented for varied durations (17-133ms) and make perceptual decisions by pressing keys to examine age differences in perceiving the communicative features of observed daily actions. We found that older adults, compared to younger individuals, required more time to perceive the communicative nature of a picture. Additionally, by fitting participants' response data with the diffusion model, we found that older adults' decision-making process was noisier compared to that of the young. In our proposed study, we plan to adopt a multiple arrangement task in combination with Magnetoencephalography (MEG) measurement and representational similarity analyses to investigate how older adults' neural representation of observed actions differs from that of younger individuals, leading to the slower perception of observed actions. Specifically, we will manipulate sensory uncertainty (operationalized by image blurring) as a potential contributing factor to age differences in action perception. Furthermore, we will explore the age differences in how action perception unfolds over time (i.e., disentangle the contributions of different features of our naturalistic stimuli and the relative timing in MEG data).

Transcranial magnetic stimulation (TMS) has been widely used to modulate brain activity in healthy and diseased brains, but the underlying mechanisms remain unclear. Previous research leveraged biophysical modeling of the induced electric field (E-field) to map causal structure-function relationships in the primary motor cortex. This study aims at transferring this localization approach to spatial attention, which helps to understand the TMS effects on a higher cognitive function, and may ultimately result in an optimized stimulation scheme. Thirty right-handed healthy participants participated in a functional magnetic imaging (fMRI) experiment and seventeen of them were re-invited for a TMS experiment. The fMRI activation peak within the right inferior parietal lobule (rIPL) during a Posner-like attention task defined the center target for TMS. Thereafter, participants underwent 500 Posner-task trials with 5 TMS pulses at 10 Hz each to modulate attentional processing. The TMS-induced electrical fields (E-fields) for every cortical target were correlated with the behavioral modulation to identify the relevant cortical regions for attentional orientation and reorientation. We found that TMS selectively inhibited attentional reorienting, resulting in task-specific behavioral impairments, and this effect was especially pronounced for stimulating the supramarginal gyrus (SMG). The BOLD-measured neuronal activity and TMS-evoked neuronal effects showed different patterns, which highlights the principal distinction between neural activity being correlated with (or maybe even caused by) particular paradigms, and activity of neural populations exercising a causal influence on the behavioral outcome.

In functional magnetic imaging (fMRI) in Parkinson’s disease (PD), a paradigm consisting of blocks of finger tapping and rest along with a corresponding general linear model (GLM) is often used to assess motor activity. However, this method has three limitations: (i) Due to the strong magnetic field and the confined environment of the cylindrical bore, it is troublesome to accurately monitor motor output and, therefore, variability in the performed movement is typically ignored. (ii) Given the loss of dopaminergic neurons and ongoing compensa- tory brain mechanisms, motor control is abnormal in PD. Therefore, modeling of patients’ tapping with a con- stant amplitude (using a boxcar function) and the expected Parkinsonian motor output are prone to mismatch. (iii) The motor loop involves structures with distinct hemodynamic responses, for which only one type of modeling (e.g., modeling the whole block of finger tapping) may not suffice to capture these structure’s temporal activation. The first two limitations call for considering results from online recordings of the real motor output that may lead to significant sensitivity improvements. This was shown in previous work using a non-magnetic glove to capture details of the patients’ finger movements in a so-called kinematic approach. For the third limi- tation, modeling motion initiation instead of the whole tapping block has been suggested to account for different temporal activation signatures of the motor loop’s structures. In the present study we propose improvements to the GLM as a tool to study motor disorders. For this, we test the robustness of the kinematic approach in an expanded cohort (n = 31), apply more conservative statistics than in previous work, and evaluate the benefits of an event-related model function. Our findings suggest that the integration of the kinematic approach offers a general improvement in detecting activations in subcortical structures, such as the basal ganglia. Additionally, modeling motion initiation using an event-related design yielded superior performance in capturing medication- related effects in the putamen. Our results may guide adaptations in analysis strategies for functional motor studies related to PD and also in more general applications.

Learning strategies are a key aspect of the solvability of complex tasks. In reinforcement learning (RL), agents interact with their environment to learn a task, a process in appearance understandable as similar to the one of humans. However, the underlying computational mechanisms unfolding during the learning process are still largely unknown. Here we study the development of task representations in RL agents during learning using a multi-scale approach inspired by cognitive neuroscience. We show that neurons’ activity changes during learning as neurons get higher temporal resolution and develop different selectivity profiles. Based on the neural activity we propose and develop an abstraction framework to identify the successive levels of the task abstraction process. Finally, we show that the structural changes of the neural population display different phases during learning and are invariant to the RL types and games studied. These findings allow for more understanding and transparency of RL agents' learning mechanisms.

In social contexts, humans frequently infer what other people think and believe. This ability to reason about other minds is called Theory of Mind (ToM). In preschool-aged children, a recent study identified distinct and independent cortical brain regions associated with the early development of a non-verbal form of this ability, termed implicit ToM, in contrast to the classic, verbal form, termed explicit ToM (Grosse Wiesmann et al., 2020). We here ask whether different fiber connections support these abilities in development. Based on previous findings, we hypothesized a dual pathway model of ToM: a dorsal tract, the arcuate fascicle (ARC) connecting the medial prefrontal cortex (mPFC) and the temporoparietal junction (TPJ), supporting explicit ToM and a ventral tract connecting the supramarginal gyrus (SMG) and the anterior insula (AI) supporting implicit ToM development. To test these hypotheses, we used micro- and macrostructural measures of structural connectivity in an automated fiber quantification (Yeatman et al. 2012) in an existing dataset of behavioral and diffusion MRI data collected from a sample of 3-to-4-year-old children (Grosse Wiesmann et al., 2017). Specifically, anatomical tracts of interest are reconstructed from diffusion data and segmented into slices. Each slice is then correlated with behavioral measures. As predicted, explicit ToM scores correlated with the right ARC connecting mPFC and TPJ and the bilateral cingulum (CGC) connecting mPFC and PC. Further, implicit ToM scores correlated with the white matter connection SMG-AI. The effects in the CGC and SMG-AI were independent of age, language, and executive functions, while the ARC effect was dependent on these co-developing abilities. Importantly, the effects found for explicit ToM are independent of implicit ToM and vice versa. The ARC and CGC are part of the default mode network (DMN) which is known to support ToM capacity. The SMG and AI are both part of the ventral attention network (VAN) which is responsible for saliency detection and bottom-up attentional processing. The effects we find here for implicit and explicit ToM make use of distinct structural brain networks and are independent of one another. Our findings thus support the theory that these abilities make use of two different cognitive systems: Explicit ToM relies on the well-known ToM network in the DMN while implicit ToM depends on lower-level attentional processes supported by connections in the VAN.

Introduction: Visual perception of food cues can trigger both reward-related and cognitive control networks of the brain. This plays a crucial role in food decision-making and eating behavior. However, when considering the effect of body weight on food decision-making, fMRI studies have produced contradicting results, leaving the evidence inconclusive. Moreover, neuroimaging studies often fail to consider the implications of methodological issues in both original studies and reviews of the field. To overcome these limitations, we provide preregistration of a systematic review and meta-analysis of the current fMRI literature on food cue reactivity, taking diverse measures of effect as well as reliability into account. Method: First, we conducted a preliminary search on published papers and pre-registered projects on PubMed and PROSPERO, respectively. Next, we developed the optimal main search strategy syntax (apt combination of key/MeSh terms) for PubMed and Web of Science databases by considering recall, precision, and number needed to read as measures of performance. The search covered the time-window from the 1st of January 2000 to the 31st of October 2022. Respective study populations and neuroimaging features are defined through tailored inclusion and exclusion criteria. Two independent reviewers are assigned to do the screening, quality check, and data extraction of the papers. For the meta-analysis, we will use “Brainmap GingerALE software” and leave-one-out jackknife for sensitivity analysis. Outcome: For details, see PROSPERO: CRD42022365310. Briefly, the peak coordinates for the task-based fMRI activations will be the main outcome measures. In addition, the following specifications of fMRI BOLD response will be reported and systematically evaluated: Direction of ß and contrasts between food/non-food or high/low-caloric, and, if applicable, higher-order contrasts (e.g., obese vs. lean); fMRI acquisition parameters; task designs, etc. Secondary outcomes will include publication, study, and participant characteristics. Discussion: In this preregistration, we propose a framework on how to systematically extract information showing which brain (de-)activations towards food cues are reliably associated with food-decision making, especially when taking body weight variation into account. In sum, implementing rigorous and unbiased preregistrations in open science platforms can affect the quality and reliability of human fMRI studies in neurosciences.

Mindfulness-based interventions have become a popular means to reduce stress. However, the mechanisms driving observed stress reduction remain understudied. Prior research aiming to pinpoint these specific mechanisms has benefitted from dissecting mental training interventions into distinct training modules containing different types of mental practices. This had led to the development of the Monitor and Acceptance Theory. This asserts that during mindfulness training, attentional and interoceptive capacities are first developed which may lead to emotional agitation and symptom exacerbation as monitoring is learned. However, through the secondary cultivation of acceptance and acceptance related capacities individuals learn how to manage these affective states, and thus stress reduction occurs. In the context of a longitudinal mindfulness training study, the ReSource project, healthy participants underwent three 3-month mental training modules targeting either attentional (Presence module), socio-affective (Affect module) or socio-cognitive skills (Perspective module). We tracked the development in a range of mental capacities relating to either monitoring or acceptance and related changes in these mental capacities to participants’ stress-reactive cortisol levels following standardized psychosocial stress induction. We found that monitoring and acceptance capacities modulated stress sensitivity. However, this was dependent on the type of training undergone. In those who trained presence, cultivation of attention based capacities was associated with higher stress sensitivity, however the reverse effect was found for those who trained in the Affect of Perspective modules. Trend level effects were observed for acceptance related capacities. Discussion: Our results corroborate the requirement of attention/monitoring to fully access the supporting emotional capacities to build stress resilience. Moreover, cultivating acceptance is a benefactor in lowering cortisol stress reactivity. Overall, mental training relies on the co-development of several interacting processes to successfully attenuate stress.

High frequency oscillations (HFOs) in response to somatosensory stimulation were first observed in the 1970s as small notches on the earliest cortical EEG response, known as the N20. More recently, concurrent surface EEG and single unit recordings have suggested that HFOs may be a non-invasive marker for neural population spikes. Despite these cortical findings, thus far, only very limited data point to the existence of HFOs in the human spinal cord, the first station of somatosensory processing. In this project, we aimed to leverage spatial filtering techniques to non-invasively uncover HFOs not only in cortex, but also in the spinal cord, potentially providing a novel window into neuronal activity over the entire central nervous system. To achieve this aim, we used a dataset previously collected in our lab where 36 participants each received 2000 non-painful electrical stimuli. The stimulation was applied to either the median nerve at the wrist, or the tibial nerve at the ankle and data was recorded simultaneously from both the cortex and spinal cord using surface electrodes. To extract the HFOs from the surface recordings, we relied on a spatial filtering approach known as canonical correlation analysis (or CCA), which we used to find spatial filters that maximise the correlation between single-trial data and the trial-averaged signal. Having replicated previous findings to establish our data is of sufficient quality to investigate HFOs at the cortical level, we were able to establish the existence of HFOs in both the cervical and lumbar spinal cord, as well as the spatial specificity of such signals. We now hope to leverage these findings to investigate the degree to which HFOs provide complementary information to more typical low-frequency analyses when it comes to neural processing at the level of the spinal cord.

While the brain constantly regulates our bodily functions, it also receives interoceptive feedback about the internal state of the body. One specific body-brain interaction is the cardiac axis, where the neurophysiological indicator of the cardiac signal is the heartbeatevoked potential (HEP). Analyses time-locked to the R-peak reveal major cardiac cycle effects that influence perception and attenuate pathological findings. These may indicate that the HEP represents a neuronal prediction for each heartbeat. Although many studies have focused on the regular heart rhythm, few have contextualised the HEP with arrhythmia. A common type of arrhythmia are extrasystoles, which are mostly benign, premature beats that occur outside the physiological heart rhythm. The proposed project aims to analyse the neuronal response to extrasystoles as an interruption of the physiological heart rhythm using a novel approach in the field of heart-brain EEG analysis. Further, it will be investigated whether brain activity before the irregular beat can be used to predict an impending extrasystole. Machine and deep learning methods have been widely used for classification tasks in brain-computer interfaces. However, the full potential of these methods for exploratory research questions in the field of heart-brain interactions has not yet been explored. By combining state-of-the-art methods with novel research questions, this project aims to identify useful approaches for further exploratory studies in this field.

In this study we use a multimodal approach to assess the pathophysiology of Tourette Syndrome (GTS). A 7T MAGNETOM TERRA (Siemens) is employed for the acquisition of T1-w, QSM, T2* and MRS (aMCC and striatum) while a 3T Biograph mMR PET-MRI (Siemens) is used for acquisition of [11C]-SCH23390 D1-Dopamine receptors availability maps as well as T1-w, rs-fMRI and DWI data. The aim of the study is -for first time in GTS- to identify Quantitative Susceptibility Mapping (as a surrogate for local iron distribution) and D1 receptors binding potential alterations in the basal ganglia and brain stem nuclei between the GTS cohort and the controls, indicating potential implications of brain iron accumulation in the roots the pathology’s characteristic clinical manifestation. Additionally, statistical analysis of the T1 and DWI data might indicate structural alterations associated with the pathology, in regions of interest, like basal ganglia and cerebellum. Our results until now have indicated reduced iron accumulation (through QSM) in subcortical nuclei, in patients with Tourette Syndrome as well as widespread reduced D1 dopamine receptors availability in orbito-frontal, striatal, temporal and thalamic areas. Strong correlations are found between local iron distribution and D1-receptor binding potential in basal ganglia. Additionally, correlations are also observed between QSM and tic severity in Thalamus, Caudate and Substantia Nigra -a nucleus known for its involvement in motor function. These links between susceptibility reductions obtained with QSM and PET-derived abnormalities in dopaminergic transmission and clinical manifestation, suggest that disruptions in iron regulatory mechanisms may be involved in GTS pathophysiology, and that neurotransmitter abnormalities may be related to mechanisms regulated by iron-containing enzymes. While susceptibility measures lack the unique specificity of PET to target individual aspects of dopaminergic signaling (i.e., dopamine vesicles, individual receptors, and transporters), the non-invasiveness of QSM opens new routes to indirect investigations of the dopaminergic system in GTS, including maturational effects during adolescence, where PET scanning is not an option.

Autism is a neurodevelopmental condition involving atypical sensory-perceptual functions together with language and socio-cognitive deficits. Previous work has reported subtle alterations in the asymmetry of brain structure and reduced laterality of functional activation in individuals with autism relative to non-autistic individuals (NAI). However, whether functional asymmetries show altered intrinsic systematic organization in autism remains unclear. Here, we computed inter- and intra-hemispheric asymmetry of intrinsic functional gradients capturing connectome organization along three axes, stretching between sensory-default, somatomotor-visual, and default-multiple demand networks, to study system-level hemispheric imbalances in autism. We observed decreased leftward functional asymmetry of language network organization in individuals with autism, relative to NAI. Whereas language network asymmetry varied across age groups in NAI, this was not the case in autism, suggesting atypical functional laterality in autism may result from altered developmental trajectories. Finally, we observed that intra- but not inter-hemispheric features were predictive of the severity of autistic traits. In sum, our findings illustrate how regional and patterned functional lateralization is altered in autism at the system level. Such differences may be rooted in altered developmental trajectories of functional organization asymmetry in autism.

Recent research has shown that learning to suppress the likely location of a salient distractor can boost visual search efficiency, known as statistical learning of distractor suppression. However, the course of statistical learning and the influence of sequential volatility are still not well understood. In the present study, we investigated how the volatility of distractor presence affects statistical learning of distractor suppression. We adopted a distractor probability cueing search paradigm and used Markov chain transitional matrix to construct volatility sequences to compare the statistical learning of distractor suppression in high- and low-volatile distractor environments while maintaining a consistent global spatial distribution of distractor - one location with the same high occurrence of the distractor. We replicated the distractor probability cueing effect - faster responses when the distractor appeared at the frequent location relative to the rare location. Although response times were generally slower in highly volatile environments, the distractor-location probability cueing and the inter-trial distractor repetition suppression were similar between both environments. Interestingly, the target location effect shown in distractor-absent trials, an indicator of proactive suppression at the priority map, which was prominent in the highly volatile environment, disappeared in the low volatile environment. These findings suggest that statistical learning of distractor suppression is largely unaffected by the volatility of the distractor occurrence, but location-based proactive suppression depends on the volatility of the distractor occurrence.

Stress-related disorders, such as anxiety and depression, have been increasingly linked to the brain-gut axis, which highlights the intricate communication between the central nervous system and the gastrointestinal tract. The study of this relationship demands the integration of various types of data, including neuroimaging, behavioral, and physiological measures. In this paper, we propose a novel approach to model stress-related disorders based on multimodal brain-gut data. We collected a comprehensive dataset consisting of gut microbiota features such as abundance, metabolite profiles, and gene expression patterns, and brain data, such as electrophysiological data, neurotransmitter profiling data. The data was collected from a diverse group of mice, including individuals diagnosed with stress-related disorders and healthy controls. We employed machine learning techniques, such as support vector machines, decision trees, and deep learning, to analyze the multimodal data and identify patterns that could distinguish individuals with stress-related disorders from healthy controls. Furthermore, we used feature selection and ranking methods to determine the relative importance of each data modality in predicting stress-related disorders. Our results demonstrate that the integration of brain-gut multimodal data significantly improves the prediction of stress-related disorders compared to using single modality data. Additionally, we identified specific neural and gut microbiome features that were strongly associated with stress-related disorders, providing insights into potential therapeutic targets. In conclusion, this study highlights the potential of utilizing multimodal brain-gut data in understanding and predicting stress-related disorders, paving the way for personalized interventions and treatments based on an individual's unique brain-gut profile. Future research should focus on longitudinal studies to explore the causal relationship between brain-gut interactions and stress-related disorders, as well as on the development of more sophisticated analytical methods to further refine our understanding of this complex relationship.

Smartphones are playing an increasingly important role in the lives of individuals. At the same time, many countries around the world are facing an epidemic of obesity due to an increased intake of energy-dense foods. I hypothesize that instant gratification in regards to technological engagement results in unhealthier eating decisions. There is evidence indicating that technological engagement has an influence on intertemporal preferences (Wilmer H & Cein J, 2016). Shuval et al. (2016) suggest that higher future time preferences are related to a lower frequency of unhealthy (fast-)food consumption. My research proposal aims to investigate the potential correlation between a decrease in instant gratification resulting from reduced technological engagement, specifically focused on smartphone usage, and healthier eating decisions. Moreover, the study seeks to explore the impact of this decrease on individuals' tendency to delay gratification, ultimately enhancing their ability to resist unhealthy eating choices. To conduct this research, a sample of 1000 individuals aged 18-45, who spend at least 4 hours daily on screens, with half of their screen time dedicated to social media, will be recruited. These participants will be divided into two groups: a detox-group, instructed to limit their phone usage to a maximum of 2 hours per day, and an unrestricted group. The study will involve the assessment of 1) the independent variable technological engagement which will be measured using the technological engagement scale developed by Wilmer and Chein (2016), along with the utilization of an installed app to track smartphone usage, 2) eating behavior, the dependent variable, will be assessed through the use of a food diary and a device capable of measuring caloric intake, 3) glucocorticoid levels will be tested and incorporated as a mediator in the model, considering the assumption that high smartphone use-induced stress affects glucocorticoid levels and subsequently influences food intake (Adam T & Epel E, 2007). Additionally, 4) fMRI scans are used to assess neural activation patterns associated with cognitive processes during experimental tasks that engage participants in smartphone-related and food-related decision-making scenarios while undergoing fMRI scanning. The study helps understand the relationship between technology, cognition, and behavior and provides insight into how a decrease of instant gratification an individual is exposed to could affect their life, since healthier eating decisions are linked to higher levels of well-being with physical and mental health benefits. On a macro-sphere their decision-making in eating and their ability to delay gratification play a significant role in general public health.

Unlike the positive blood oxygenation level-dependent (BOLD) response (PBR), commonly taken as an indication of an ‘activated’ brain region, the physiological origin of negative BOLD signal changes (i.e. a negative BOLD response, NBR), also referred to as ‘deactivation’ is still being debated. We attempted to gain a better understanding of the underlying mechanism by obtaining a comprehensive measure of the contributing cerebral blood flow (CBF) and its relationship to the NBR in the human visual cortex, in comparison to a simultaneously induced PBR in surrounding visual regions. A newly developed multi-echo version of a center-out echo planar-imaging (EPI) readout was employed with pseudo-continuous arterial spin labeling (pCASL) facilitated a simultaneous detection of functional CBF and BOLD changes at 3T with improved sensitivity. Evaluations of CBF changes and the effective transverse relaxation rate (R2*), the coupling ratios, and their dependence on CBF at rest, indicated differences between activated and deactivated regions. Analysis of the shape of the respective functional responses also revealed faster negative responses with more pronounced post-stimulus transients. The application of a simplistic model of neuronal control of changes in CBF and oxygen metabolism to our data in regions of PBR and NBR, further suggested differences in neuronal contributions and inhibitory control of changes in CBF between the two regions.

Neuromelanin-sensitive MRI receives interest as potential biomarker in neurodegenerative diseases, such as Parkinson’s and Alzheimer’s disease. It is known that human neuromelanin pigments bind large quantities of toxic metal ions, especially iron but also other transition metals including copper. These melanin-iron complexes are a potential source of paramagnetic relaxation enhancement of water proton. In relaxometry investigations, we found deviations from a simple linear concentrationdependent T1 shortening in synthetic neuromelanins containing different amounts of iron and copper. Knowledge of the occupation of distinct metal binding sites seems crucial for contrast optimization or attempts to quantify metal content by MRI.

Utilizing simulated autonomous agents as virtual model organisms for natural intelligence mirrors the use of rats as biological model organisms for human learning. While traditional model organisms exhibit noise and unpredictable behavior, computer programs provide the advantage of being studied within precise environmental conditions, allowing direct investigation of their mechanisms. Reinforcement learning is a versatile framework for developing artificial agents that learn to solve tasks through experience. Although this framework offers a generalized approach to model decision-making and trial-and-error learning in the brain, the agent's autonomy is not absolute, as the algorithm designer must specify an extrinsic reward for each task. However, recent advancements in self-supervised reinforcement learning present promising opportunities to model intrinsic motivation in intelligent agents. How can autonomous agents learn to control their environment without external hints, thereby determining their own desired outcomes? How can agents acquire the ability to identify affordances and construct neural representations of them? In this study, we propose strategies to address these questions by employing simulations of intrinsically motivated intelligent agents based on self-supervised reinforcement learning in complex environments.