Session IV: 30 June, 16.30-17.45 (CEST)

Neural oscillations play a crucial role in human brain functioning. Among these rhythms, beta oscillations (~13–30 Hz) have been extensively studied, yet their functional roles remain debated. Methodologically, traditional approaches treat beta activity as a sustained amplitude-modulated rhythm, yet recent studies show it is better described as transient bursts. These bursts vary in occurrence rate, duration, peak frequency, and amplitude, and such features have been shown to differ across behavioral contexts and clinical conditions, underscoring their functional relevance. Yet one critical aspect remains largely unexplored: the waveform shape. Oscillatory waveform shapes have been linked to underlying physiological mechanisms in several studies, suggesting that different waveform motifs may reflect distinct neural processes. In this study, we therefore aim to test whether specific waveform shapes are associated with different functional states. Here, we present a comprehensive analysis pipeline to systematically study beta burst waveforms. The pipeline was validated on simulated data, where it successfully recovered ground-truth waveform structure, and can now be applied to real data. On simulated data, an iterative burst detection algorithm identified bursts and provided their waveforms and features. For the waveform analysis, we adapted a topological data analysis (TDA) approach to assess beta burst waveforms. After that, each waveform was mapped into a high-dimensional space defined by its time samples and duration. This representation allowed us to analyze the geometry of beta bursts. We then applied TDA to capture the topology of this waveform space. To guide dimensionality reduction, intrinsic dimensionality was first estimated. Subsequently, the data were embedded into a low-dimensional space using Uniform Manifold Approximation and Projection (UMAP), which preserves local structure. Within this space, Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN) identified clusters of similar burst waveforms. We showed that our pipeline can robustly recover ground-truth waveform structures and lays the foundation for its application to real MEG data. Beyond methodological novelty, the pipeline enables us to directly address whether beta burst waveforms represent a continuum of shapes or can be meaningfully divided into discrete classes, and how these relate to their functional roles.

A key challenge in psychiatric neuroimaging is the limited replicability and predictive value of findings, with many reported neural correlates of depression failing to generalize across cohorts, scanners, or analysis pipelines. While large-scale studies show altered resting-state connectivity and structure in depression, heterogeneity in methodology and diagnoses has led to only modest individual-level prediction. An emerging approach is biotyping—identifying subgroups based on reproducible individual-circuit-level dysfunctions. Tozzi et al. (2024) recently introduced a standardized framework defining biotypes using three a priori networks: Default Mode (DMN), Salience (SN), and Attention (AN). Resting-state connectivity between predefined ROI pairs is aggregated into composite scores and expressed in SD units relative to healthy controls, enhancing reproducibility. Two resting-state biotypes are of particular interest: the AC- biotype (attention hypoconnectivity, linked to poor psychotherapy response) and the DC+SC+AC+ biotype (tri-network hyperconnectivity, linked to emotional slowing and better outcomes). This study aims to replicate these biotypes in patients with major depressive disorder (MDD) undergoing ECT and assess their predictive and mechanistic relevance. We hypothesize (1) that both biotypes can be identified using the original thresholds, (2) that baseline biotype membership predicts clinical response/remission, and (3) that biotype-specific connectivity patterns normalize post-ECT. We will analyze clinical, structural MRI, and resting-state fMRI data, preprocessed using fMRIPrep (with ICA-AROMA, CompCor, field maps, and MNI normalization). Composite connectivity scores will be computed from pre-defined ROIs and standardized against a healthy reference sample. Biotype membership will be determined using pre-registered thresholds. Clinical outcomes (HDRS, MADRS) will be modeled via linear mixed-effects models and logistic regression, adjusting for age, sex, site, baseline severity, and motion. This study seeks to replicate and validate the generalizability of functional connectivity biotypes based on personalized brain circuits to neuromodulatory treatment settings and may support the development of biotype-informed, precision-guided interventions in psychiatry.

Modern interactive media represents a double-edged sword for the developing adolescent brain, capable of both enhancing neuroplasticity and triggering maladaptive behavioral disruptions. This study investigates the implementation of a "closed-loop" neuro-rehabilitation framework using the Unity game engine and Virtual Reality (VR) to transform gaming from a passive stimulus into a precision tool for neural reinforcement. Central to this approach is the integration of single-channel EEG (Electroencephalography), a streamlined Brain-Computer Interface (BCI) used to monitor real-time cortical oscillations—specifically the ratio of Beta (focus) to Theta (distraction) waves. By mapping these electrical signatures to in-game mechanics within a VR environment, we create a system of "embodied cognition" where the digital world responds directly to the user’s internal cognitive state. Unlike traditional 2D gaming, which can lead to "attentional fragmentation" and reduced impulse control, Unity-based VR simulations require high-demand spatial navigation and executive processing, directly stimulating the Hippocampus and the Prefrontal Cortex (PFC). This engagement is critical for young people, as it builds "cognitive reserve" and strengthens the white matter integrity necessary to prevent early-onset cognitive decline. Furthermore, by utilizing single-channel EEG as a biofeedback mechanism, players are trained to regulate their own arousal levels, effectively mitigating the dopamine-driven irritability and behavioral disruptions often associated with unregulated gaming. The results suggest that the "Mind-Meld" between immersive VR and real-time EEG feedback provides a robust defense against neural atrophy, turning the gaming experience into a sophisticated "brain gym." By focusing on targeted circuit modulation rather than general entertainment, this research paves the way for a new era of "Neuro-Gaming," where digital environments are scientifically engineered to optimize the architectural health of the young human mind.

Semantic dementia (SD) is characterised by progressive deterioration of semantic knowledge and language, most commonly associated with frontotemporal dementia and anterior temporal lobe (ATL) dysfunction. Non-invasive brain stimulation offers a potential avenue to modulate ATL plasticity and improve semantic processing. This study investigated whether theta-burst focused ultrasound stimulation (tbFUS) applied to the ATL alters inhibitory (GABA) and excitatory (glutamate–glutamine; Glx) neurochemistry and enhances semantic memory in healthy adults. 23 native English speakers (aged 19–33 years) completed a three-session, within-subject, counterbalanced study. Structural MRI was acquired in the first session to localise stimulation targets. In subsequent sessions, participants received tbFUS to the ATL or ventricle (control), followed by behavioural testing and MR spectroscopy. Semantic processing was assessed using the Camel and Cactus Task, with a pattern-matching task as a control. Single-voxel MEGA-PRESS MRS quantified GABA and Glx concentrations in the ATL and occipital cortex. 2X2 Repeated measures ANOVA was performed to assess ATL stimulation effect on Semantic performance. ATL stimulation significantly improved semantic task performance, reflected by increased accuracy and reduced reaction times relative to control stimulation. Paired t test was conducted on neurotransmitters GABA and Glx and was found that tbFUS induced a significant increase in Glx concentration within the ATL (p = 0.024), while no significant changes were observed in GABA levels across conditions. These findings suggests that tbFUS preferentially enhances excitatory mechanisms underlying semantic processing. Clinically, tbFUS shows promise as a precise, non-invasive approach for strengthening semantic networks, with potential relevance for conditions such as semantic dementia and Alzheimer’s disease.

Major depressive disorder (MDD) is often accompanied by changes in appetite, body weight, and reward processing. Recent work by Thun et al. (2025) showed that altered food reward in depression may depend on the macronutrient composition of food, with reduced wanting and liking for fat- and protein-rich foods but relatively preserved responses to carbohydrate-rich foods. These findings raise an important mechanistic question: are such differences driven by the objective nutritional properties of food, or by participants’ conscious perception of those properties? We present a pre-registered replication and extension of this work in healthy controls (HC) and individuals with current MDD. Participants will complete a food cue reactivity task using a large set of AI-generated food images with validated nutrient estimates, rating wanting and liking for each item and then estimating its calorie, carbohydrate, fat, and protein content. To test whether the original findings generalize to this new image set, we will compare HC and MDD participants using linear mixed-effects models of food wanting and liking. We will examine whether objective macronutrient composition improves model fit and whether MDD is associated with relatively reduced wanting and liking for fat- and protein-rich foods compared with carbohydrate-rich foods, as reported by Thun et al. (2025). We will then repeat these analyses using participants’ subjective macronutrient estimates and compare model fit to determine whether objective or subjective nutrient information better explains food reward ratings. Planned analyses will additionally test moderation by depression severity, anhedonia, and trait anxiety. We hypothesise that the original pattern will replicate, such that MDD will be associated with reduced food wanting overall and with lower wanting and liking specifically for fat- and protein-rich foods relative to HC participants. Critically, if objective nutrient information outperforms subjective estimates in explaining food reward, this would support the idea that altered valuation is linked more strongly to implicit metabolic signalling than to explicit nutrient knowledge. By distinguishing conscious from putatively implicit contributors to food reward, this project aims to advance a brain-body account of reward dysfunction in depression.

INTRODUCTION: Stroke remains a leading cause of long-term disability, often resulting in persistent motor impairments that resist conventional treatment. While BCIs offer a promising way to foster neuroplasticity by bypassing damaged circuits, their adoption is limited by 'BCI inefficiency' in 15–30% of users. This lack of control is often due to the low amplitude or high variability of sensorimotor signals. In this study, we investigate healthy volunteers to establish the link between dopaminergic modulation, BCI learning, and EEG patterns. By pairing BCI training with Levodopa, we examine whether altering dopamine levels can enhance BCI learning, given dopamine’s essential role in neural plasticity and motor function. METHODS: In a double-blind, between-subjects study, 22 healthy individuals were randomly assigned to either a Levodopa or a placebo group. Participants completed daily one-hour motor imagery BCI training for six days in a row using the RecoveriX system (g.tec). Each session began 30 minutes after the administration of their assigned substance. EXPECTED RESULTS: We hypothesize that Levodopa will improve BCI learning by modulating key neurophysiological markers of motor imagery, particularly the event-related desynchronization (ERD) and the Signal-to-Noise Ratio (SNR) of sensorimotor rhythms. We expect these neural changes to positively correlate with higher BCI classification accuracy. Additionally, we will assess how dopaminergic modulation affects the stability of neural patterns used for classification over multiple sessions. Finally, we will examine changes in functional connectivity (FC); considering the mixed findings in longitudinal studies, we anticipate a complex interaction between FC and BCI learning. DISCUSSION: Boosting BCI learning speed and accuracy via dopaminergic modulation could greatly improve the clinical practicality of this technology. These developments are essential for creating more efficient, tailored rehabilitation programs for stroke survivors and individuals facing chronic motor impairments.

Post-stroke cognitive impairment and dementia (PSCI/PSD) affect up to 70% of stroke survivors and represent a major determinant of long-term functional recovery and quality of life. Previous studies have primarily investigated the neuroanatomical substrates of PSCI/PSD using lesion-based analyses. However, relatively few studies have examined neurotransmitter-specific alterations associated with PSCI/PSD by integrating neuroimaging data with neurochemical system. In this proposed study, we aim to investigate neurotransmitter-specific network disruptions associated with PSCI/PSD by cross-validating between two independent stroke cohorts: the randomized controlled INSPiRE-TMS trial and the longitudinal multicenter DEMDAS study. Patients without baseline dementia who underwent MRI within seven days after acute ischemic stroke will be included. Stroke severity will be assessed using the NIHSS, and cognitive outcomes will be evaluated one year after stroke using the Montreal Cognitive Assessment (MoCA). Normative neurotransmitter receptor and transporter density maps derived from PET imaging are integrated with a population-based structural connectome to generate neurotransmitter-informed connectivity maps for major neurotransmitter systems, including serotonin, dopamine, GABA, glutamate, and acetylcholine. These maps are constructed by weighting white-matter streamlines according to neurotransmitter density at their cortical and subcortical endpoints. Individual stroke lesion masks will then be projected onto this normative connectome to estimate lesion-induced network disconnections. For each neurotransmitter system, the weights of disconnected streamlines will be summed to quantify neurotransmitter-specific network disruption for each patient. PSCI/PSD status (MoCA < 26) will be analyzed as a binary outcome using multivariable logistic regression adjusted for global structural damage, age, pre-stroke cognitive status, lesion volume, and baseline NIHSS. Partial least squares (PLS) analysis will be applied to identify latent neurotransmitter-related predictor patterns associated with PSCI/PSD, and variable importance in projection (VIP) scores will be used to determine the contribution of individual predictors. Predictive models incorporating global PLS components, structural disconnection measures, or top-ranked neurotransmitter predictors will be compared with covariate-only models using Akaike Information Criterion (AIC).

The human brain exhibits intrinsic neural timescales (INTs) that vary systematically across cortical regions and reflect how these regions integrate incoming signals across different temporal windows. INTs follow a systematic hierarchy along the sensorimotor-association (S-A) axis, with unimodal sensory regions exhibiting fast timescales, while transmodal association areas demonstrate slow timescales. However, how INTs change across different cognitive states and between individuals is less well understood. Using high-resolution 7T fMRI data from the Human Connectome Project young adult dataset (HCP-YA, S1200 release) of healthy adults (n = 170, mean age: 29.3 ± 3.3 years, 102 females), INTs and their spatial covariation were compared between resting-state and movie-watching conditions. First, INTs were computed for each cortical parcel of the Schaefer-400 atlas. Group-level INT maps revealed a shift in the spatial distribution of timescales: during rest, the highest INTs were concentrated in default-mode, frontoparietal, and ventral attention networks, consistent with internally driven, integrative processing. In contrast, movie-watching enhanced INTs in visual and temporal default-mode networks, reflecting heightened engagement with external sensory input. Overall, during movie-watching, INTs were significantly increased in the visual and limbic network and decreased in the ventral attention network. Next, to examine how INTs covary across the cortex, inter-regional covariance matrices were computed for each condition, and diffusion map embedding was applied to extract low-dimensional gradients. In the resting-state, Gradient 1 (G1) captured the canonical S-A axis, while Gradient 2 (G2) reflected a visual-to-somatomotor axis. During movie-watching, this pattern reversed: G1 now aligned with the visual-somatomotor axis, while G2 recovered the S-A axis. These findings demonstrate that the covariance architecture of INTs is not static but dynamically reorganizes in response to the cognitive state. Altogether, these findings show that INTs follow a cortical organization, without explicitly relying on functional connectivity measures. A clear difference in INTs covariation between resting-state and movie-watching was found, highlighting the brain’s adaptive temporal processing mechanisms across cognitive states.

Memory consolidation during sleep is crucially dependent on precise coupling dynamics of neural activity between the cortex, thalamus, and hippocampus. Low-intensity non- invasive brain stimulation (NIBS) offers a unique opportunity to modulate this activity in humans as they sleep, potentially affecting the underlying consolidation mechanisms and subsequently memory performance. However, most commonly studied NIBS methods are limited in their potential to reach deep brain targets such as the hippocampus. This poster presents the methodological design choices for an experiment to assess the effects of NIBS on both episodic and procedural memory consolidation. We will use Temporal Interference (TI) to stimulate the left thalamus and left hippocampus during sleep. Our closed-loop design provides real-time detection of N2 and N3 slow-wave events, enabling targeted stimulation at the peak of cortical up-states. Personalisation of stimulation electrode position based on MRI anatomical images and neuronavigation tools further optimises the efficacy of TI stimulation by improving the focality of the electrical field generated at each target site. To measure changes in memory performance, subjects will complete a word-image association task and a motor sequence learning task before and after sleep. We discuss the task design, including measures taken to balance the stimuli sets and reduce inter-session variability, as well as the order in which subjects should perform them based on pilot data analysis. Finally, we show preliminary EEG processing, including slow wave-spindle coupling detection and spectral power analyses. Taken together, these methodological components provide a framework for evaluating closed- loop TIS during sleep, personalising stimulation of deep brain targets with MRI-informed TI and combining two behavioural tasks in a single experimental pipeline to delineate neural mechanisms underlying study sleep-dependent memory consolidation.

Spatial learning in humans and other animals relies on specialized neural systems that support navigation in complex environments. Spatial information, such as objects, environmental boundaries, direction, and distance, is represented by populations of spatially tuned neurons. Although several spatial cell types have been identified and are known to play important roles in navigation, the neural mechanisms that support spatial representation and memory are still not fully understood. Over the past decades, a number of computational models have been proposed to explain how spatial navigation emerges and how environmental information is encoded along behavioral trajectories. Many of these models highlight the role of the hippocampal–entorhinal system and suggest that spatial environments are represented through coordinated activity across multiple neuronal populations and cortical regions. However, the precise network structure and dynamics underlying these processes remain debated. In addition, most studies have focused on rodents, and computational models of human spatial cognition remain relatively limited. This project aims to develop a human computational framework linking spatial navigation and episodic memory. we plan to start with egocentric bearing cells (EBCs), which were first identified by Kunz et al. in 2021. These neurons are thought to encode egocentric distance and direction during human spatial memory tasks. We propose a model in which multiple EBCs are involved, each showing specific tuning for both direction and distance relative to a reference field. As participants move through the environment, different EBCs may become active sequentially as the participant approaches particular reference points. By combining the BB model(Bicanski and Burgess, 2018), allocentric representations are assumed to be supported by neural populations in the medial temporal lobe, including grid cells, place cells, and allocentric object vector cells (OVCs). These cell types are thought to contribute to the construction of a stable spatial representation independent of the observer’s current viewpoint. At this stage, the model should be viewed as a preliminary idea. The model will likely need to be revised as more intracranial recording data become available and further analyses are conducted.

Functional Magnetic Resonance Imaging (fMRI) provides a non-invasive window into brain function, yet standard dimensionality-reduction techniques such as ICA or diffusion embedding mainly reveal group-level patterns and fail to capture individual variability. We present BRAINS, a deep latent modeling framework based on Convolutional Variational Autoencoders (CVAEs) that learns subject-specific representations of BOLD time series. The model reconstructs and denoises fMRI data, improving temporal signal-to-noise ratio by approximately 5 \% while preserving functional connectivity and spatial coherence. By aligning individual latent spaces with Procrustes analysis, we demonstrate shared and subject-specific functional components consistent with known cortical gradients. BRAINS offers a data-driven foundation for individualized fMRI analysis and supports more precise mapping of brain organization in research and clinical settings.

Structural brain maturation from childhood through early adulthood is characterized by progressive myelination and microstructural reorganization across cortical and subcortical regions. While cortical development has been extensively studied, coordinated cortico-subcortical maturation remains less well understood, despite its importance for cognitive and affective development. The T1-weighted/T2-weighted (T1w/T2w) ratio provides an in vivo proxy of regional microstructure, primarily reflecting myelin content, and offers a means to investigate synchronized developmental trajectories across brain systems. Using cross-sectional data from the Human Connectome Project Development (ages 5–21) and Young Adult (ages 22–26) cohorts (N = 959), we quantified age-related T1w/T2w trajectories in cortical and subcortical regions. Cortical parcels were defined using the Glasser atlas and subcortical regions using the Melbourne (Tian S4) atlas. Generalized additive models were used to estimate age-related changes, and first derivatives of T1w/T2w trajectories (ΔT1w/T2w) were computed to index regional maturation rates. Cortico-subcortical co-maturation was assessed by comparing maturation rates across regions, and principal component analysis (PCA) was applied to identify shared developmental patterns across four age bins spanning childhood to early adulthood. T1w/T2w ratios increased with age across most regions, with pronounced regional variation in maturation rates, such as prefrontal cortices, which exhibited slower and more protracted maturation. PCA revealed partial convergence of limbic and insular cortical regions with subcortical structures, indicating shared maturation profiles. Cortico-subcortical co-maturation was strongest during early adolescence (13–15 years) and diminished thereafter. Latent co-maturation axes were selectively associated with multimodal cortical organization, including myelin content, gene expression, functional gradients, neurosynth meta-analytic maps, and cortical thickness. These findings identify a temporally constrained window of coordinated cortico-subcortical maturation during adolescence, highlighting synchronized developmental processes that align with established structural, molecular, and functional brain organization and may support the emergence of mature cognitive systems.

Harmonicity plays a fundamental role in auditory perception because most natural sounds exhibit harmonic structure, making them more predictable and perceptually salient. Thus, harmonicity may support the formation of precise auditory predictions and robust neural responses to expectancy violations. Musical training may further modulate these processes, as prolonged experience with structured sound patterns refines internal predictive models and enhances sensitivity to statistical regularities in melodic sequences. Building on this framework, the present study investigates how harmonicity influences neural responses to melodic expectancy violations and whether musical training modulates this effect. We examine whether less probable tones in harmonic melodies elicit stronger brain potentials than comparable tones in inharmonic melodies, and whether musicians show enhanced neural responses relative to non-musicians. Understanding these mechanisms may clarify how the brain encodes auditory uncertainty and pitch structure in music perception. Hearing thresholds will be established using the Hughson–Westlake procedure, after which participants will undergo an EEG recording. During the experiment, participants will listen to monophonic melodies derived from instrumental works by Johann Sebastian Bach, divided into 150-second blocks. 50% of the melodies will be acoustically manipulated using the TANDEM-STRAIGHT algorithm to create inharmonic versions. Across 30 blocks (15 per condition), participants will rate the familiarity of each melody while neural responses to less probable tones, as estimated by the IDyOM model, are recorded and compared between harmonic and inharmonic conditions, as well as between musicians and non-musicians We expect that less probable tones in harmonic melodies will elicit stronger event-related potentials than those in inharmonic melodies, and that musically trained participants will demonstrate greater neural sensitivity to melodic expectancy violations. Data collection is currently in progress. The findings are expected to advance understanding of how the brain encodes auditory uncertainty on fine temporal scales relevant to music and speech perception. This research may also deepen knowledge of pitch processing mechanisms critical for language, music cognition, and broader auditory functions.

Synthetic voices are still perceived as sounding less human than recordings of human speakers (Wester & Larrouy-Maestri, 2025). This raises the question of whether text-to-speech (TTS) voices are processed differently from human voices. Lavan et al. (2024) revealed neural correlates of voice identity cues, of both physical traits (e.g., gender, age) and social traits (e.g., attractiveness, trustworthiness), with neural correlates of physical traits emerging early and social traits appearing later after stimulus onset. We hypothesize that the humanness of a voice might be another factor of voice identity, with a distinct neural correlate. To this end, we will record brain activity through electroencephalography (EEG) from participants listening to five German sentences spoken by eight human speakers and eight TTS-generated voices. In a second session, participants will rate each stimulus on a slider, for how human the voice sounds to them, as well as perceived arousal and valence of the stimuli. By examining the relation between EEG activity and behavioral ratings, using a representational similarity analysis (RSA), this study aims to identify a potential neural correlate of humanness and when in time neural activity significantly correlates with humanness perception. The precise timing of the neural correlate of humanness perception would clarify whether humanness is closer to a physical or a social trai

ABSTRACT: The adult auditory system adapts to changes in spectral cues for sound localization. This plasticity was demonstrated by modifying the shape of the pinnae with molds. Previous studies investigating this adaptation process have focused on the effects of learning one additional set of spectral cues. However, adaptation to multiple pinna shapes could reveal limitations in the auditory system’s ability to encode discrete spectral-to-spatial mappings without interference and thus help determine the mechanism underlying spectral cue relearning. In the present study, listeners learned to localize sounds with two different sets of earmolds within consecutive adaptation periods. To establish both representations in quick succession, participants underwent daily sessions of sensory-motor training. Both pinna modifications severely disrupted vertical sound localization, but participants recovered within each 5-day adaptation period. After the second adaptation, listeners were able to access three different sets of spectral cues for sound localization. Participants adapted to both sets of earmolds with equal success, and learning a second set of modified cues did not interfere with the previous adaptation. We found no indication of meta-adaptation as the rate of adaptation to the second molds was not increased.

Aims: In our daily lives, selective auditory attention allows us to focus on relevant information while filtering out distractions. Background stimuli can, however, shift the focus of attention if they are surprising or inherently relevant to the individual. While attention capture by deviant, surprising stimuli is well-examined, the role of background speech in this process remains less clear. Specifically, the extent to which unattended speech is semantically processed and to which degree its intelligibility and semantic content drive attentional shifts remain debated. To address these questions, we developed a novel experimental approach to isolate comprehension effects by using background speech in a constructed micro-language, both before and after participants acquired fluent listening comprehension. Method: We examined auditory attention mechanisms in healthy individuals by employing electroencephalography (EEG) and pupillometry during an auditory task with background speech in a micro-language. Participants performed this task both before and after training in this micro-language, allowing us to isolate the effect of speech comprehension on attentional processes. Additionally, we included deviant non-speech sounds in the background speech to compare attention capture mechanisms elicited by different auditory stimuli. Neural responses were analyzed using the multivariate temporal response function approach. Results: We show differences in neural tracking of target and distractor speech and of distractor speech before and after participants acquired comprehension of the language. Specifically, the distractor stream elicited a prolonged negative deflection beginning approximately 100 ms after stimulus onset. When participants understood the distractor speech, this negative deflection was significantly reduced, with a peak difference around 170 ms. Notably, the latency of this TRF modulation closely follows the positive encoding deflection elicited by the salient non-speech sounds, which were designed to robustly capture attention. Discussion: By isolating comprehension effects, this paradigm demonstrates sensitivity to detect comprehension-related changes in neural speech encoding. Reduced neural encoding of intelligible distractor speech may reflect changes in linguistic processing or attentional mechanisms. These findings provide evidence that speech comprehension modulates early neural responses to unattended auditory input.

The neurocognitive mechanisms of counterfactual false memories Maia J Armstrong1, Daniel Yon1, Silvia Seghezzi1 1 Department of Psychological Sciences, Birkbeck, University of London, Malet St, London, WC1E 7HX “I could have done otherwise” captures the essence of counterfactual thinking: the ability to imagine, and even plan, alternative actions that were possible but not actually executed. While this capacity supports planning, learning, and flexible behaviour, it may also render memory vulnerable to distortion. We investigated how counterfactual action plans influence memory using a novel maze-navigation task in which participants plan and execute the fastest route to a goal. Critically, on some trials a block appeared after planning but before execution, rendering a previously plausible path unavailable. Participants later completed a memory test in which they identified the path they believed they had taken among alternatives. To characterise the nature of this distortion, we modelled memory performance using a Signal Detection Theory framework. Across two experiments, participants were systematically more likely to falsely remember having taken paths that were planned but then blocked than unchosen paths that remained available. Together, these findings suggest that action plans for unexecuted actions can leave traces that ‘override’ those for executed actions during memory recall. To further test this theory, we explicitly asked participants to indicate the route they plan to take to reach the goal using the shortest path possible, and then manipulated whether or not their chosen route, or the unchosen route, was blocked. Across both in-person and online experiments, we found that participants are significantly more likely to have false memories for the untaken path, when that untaken path was the path they originally chose to take, but was subsequently blocked after planning. This finding confirms that false memories for counterfactual actions are most likely to be formed when an action plan is inhibited or suppressed before execution of an alternative action. In order to further investigate the mechanisms underlying how action plans can cause false memories, we are translating our maze paradigm to be tested with both EEG and fMRI to gain comprehensive insight into the neurocognitive mechanisms underlying false memory formation for counterfactual actions.

Temporal Interference Stimulation (TIS) is an emerging non-invasive brain stimulation (NIBS) method that enables the modulation of oscillatory dynamics in deep brain structures while minimizing exposure of superficial tissues. The insula is a central hub for nociceptive processing, and several stimulation studies have demonstrated its role in pain perception. However, conventional NIBS methods are limited by a depth-focality tradeoff. Therefore, we aim to validate TIS as a method to selectively modulate neural oscillations in the insula, with the goal of assessing its role in pain processing. We used Finite-Element Modeling (FEM) in 20 healthy participants to identify the optimal TIS montages that allow insula targeting. To assess the configurations, we used a multi-criterion ranking approach involving exposure, selectivity, and collateral activity. The optimized montage will be validated using a concurrent TIS-fMRI within-subject design. We will deliver active stimulation through two high-frequency currents at 2 mA, generating a 10 Hz envelope, while sham stimulation will use identical carrier frequencies without interference. Experimental sessions will include resting-state and tonic heat pain blocks, while primary outcomes will be insular BOLD responses and functional connectivity metrics. We identified montages that maximize insular exposure while minimizing cortical exposure through our simulation approach. In the validation experiment, we expect alterations in insular BOLD responses to tonic heat pain when applying TIS, as well as modulated resting-state connectivity within the insula and its related networks. Moreover, we hypothesize changes in local synchrony measures during TIS compared to sham. We will also assess interindividual variability by correlating neural effects with the simulated electric field strength and selectivity. Blinding will be checked using validated procedures which are expected to confirm equivalence between sessions. In conclusion, this study aims to establish TIS as a NIBS technique for selective targeting of the human insula. Successful validation of an optimized TIS configuration would support its capacity to modulate brain oscillations across different frequencies in the insula. Additionally, neural oscillations have been linked to pain across multiple frequency bands, and manipulating them using TIS provides a mechanistic approach to probe the role of insular oscillatory dynamics in pain perception.

Body image disturbance (BID) is a psychological phenomenon characterized by distorted body perception and negative self-evaluation. Recent advancements in neuroscience have identified distinct and overlapping neural mechanisms involved in BID, particularly in regions associated with self-referential thought, emotional regulation, and visual body processing. At the same time, social media engagement has become an integral part of daily life, with growing evidence suggesting its role in shaping self-perception, social comparison tendencies, and body-related attitudes. While prior research has explored the neural correlates of BID and social media use independently, few studies have systematically examined their shared neurobiological underpinnings. This systematic review addresses this gap by synthesizing neuroimaging findings on BID and social media engagement, aiming to identify common neural networks that may contribute to body image-related distress in digital environments. A comprehensive literature search following PRISMA guidelines identified 25 studies on BID and 33 on social media use, incorporating methodologies such as functional magnetic resonance imaging (fMRI) and electroencephalography (EEG). Findings reveal significant overlap in self-referential processing networks (medial prefrontal cortex, posterior cingulate cortex), emotional regulation circuits (amygdala, insula, anterior cingulate cortex), and reward-related areas (ventral striatum, nucleus accumbens), suggesting that habitual social media engagement may reinforce maladaptive self-evaluation patterns observed in BID. These insights not only advance our understanding of the neurocognitive mechanisms linking social media use and body image disturbance but may also inform future neuroimaging studies and interventions aimed at promoting healthier body perception in digital contexts.

The ability to plan and understand complex actions considerably develops around the preschool years. Previous research suggests that structural action processing recruit the inferior frontal cortex (IFC), which is critically involved in hierarchical processing in language (Fadiga, et al., 2009). Recent meta-analyses on action and language processing in adults suggest that different subregions of the left IFC underlie structural processing of language and action (Zaccarella, et al., 2021). However, the underlying neural mechanisms of children’s action processing are largely unknown. Here, we compare structural action processing and its neural correlates in children and adults using a stacking game while measuring fNIRS. We instructed N=78 children and N=47 adults to place six blocks on to a small board to build a pathway for two toy figures. Task levels included two condition structures: Sequential action sequences, in which the order of block placement was flexible, and dependent action sequences, in which block placement was order-dependent. Participants were presented with twelve levels of increasing difficulty, six in each condition. We measured neural response using fNIRS in a 30 second planning phase, where the setup was revealed but out of reach, followed by a 60 second building phase, in which the task was solved. We hypothesized that structural complexity of action sequences modulates the activation of regions in IFC, with stronger activation for order-dependent than sequential sequences. If such activation indexes structural processing, we further expected correlations between task performance and neural response. The results suggest that the neural processing of structural action planning differs between children and adults. Our findings in children show activation in the left frontal, posterior region, located approximately over BA6, with stronger activation for order-dependent levels than sequential levels. The findings in adults are characterized by deactivation in the left and right frontal region, located approximately over BA44 and BA45, with less activation for sequential flexible levels than order-dependent levels. In sum, our results show that planning dependent action sequences is developing from an involvement of a more posterior frontal cortex in preschoolers to a more anterior frontal cortex in adults. This shift mirrors the functional allocation from an action-related posterior frontal cortex to a sequence-related anterior cortex.

Abstract: The social brain beyond the neocortex Social cognition, the ability to understand others’ thoughts and feelings, has predominantly been investigated at the level of the cerebral cortex. However, converging evidence indicates that non-neocortical regions (here termed “subcortex”), including the cerebellum, basal ganglia, and thalamus, contribute to social cognitive processing. The role of these subcortical regions relative to established cortical social brain networks remains incompletely understood. This study aims to characterize subcortical representations of social cognition and to test their role and specificity relative domain-general processing. The central research question is whether subcortical regions encode social cognitive information in a domain-specific manner and how these representations relate to neocortical social brain networks. We hypothesize that subcortical representations of social cognition are domain-specific, dissociable from non-social processing, and show systematic intercorrelations reflecting a shared representational structure. We will analyse task-based functional MRI data from the Spacetop dataset (n = 101; age: 24.7 ± 5.5 years; 69 males, 45 females, 2 other) across four social cognitive paradigms, focusing on false-beliefs, image-based mentalizing, as well as evaluations of naturalistic social scenes and vicarious pain. Task-specific contrasts will be used to derive activation maps across subcortical regions of interest (basal ganglia, cerebellum, thalamus) and cortical regions. To test the domain-specificity of subcortical systems and their relationship with the cortex, task-specific activation patterns will be extracted from predefined cortical and subcortical regions. Cortical regions will be defined using the Schaefer atlas with 400 parcels, while cerebellar and subcortical regions (thalamus, basal ganglia) will be defined using the King atlas for the cerebellum and the Tian atlas for the subcortex. Representational similarity analysis will quantify cross-task similarity of activity patterns, and multivariate pattern analysis will test whether subcortical regions encode social cognitive information distinct from non-social control conditions and whether these representations generalize across paradigms. Functional connectivity analyses will further characterize coupling between subcortical regions and cortical social brain networks.

Brain maturation is not uniform: structural and functional reorganization proceeds hierarchically across the cortex, with association areas exhibiting prolonged plasticity and heightened environmental sensitivity relative to sensory cortices [1]. Importantly, while brain maturation is driven by genetic factors, environmental factors also contribute at the individual level [2]. Therefore, a key question is how individual differences in cumulative environmental exposures – the exposome - translate into variation in maturational trajectories. Recent work has identified four major topological turning points in the lifespan organization of structural brain networks [3]. I aim to extend these observations by investigating how population-level turning points extrapolate to individuals. Deviations from population-level trajectories are expected to reflect exposome influences and are critical for interpreting the biological significance of the turning points. Using normative modelling of graph-theoretic measures of structural connectivity, I will quantify inter-individual variability across developmental epochs. I will test associations between these deviations and the individuals' exposome - capturing markers of social and physical environment - and explore relations to clinical and behavioral outcomes. To anchor these analyses in specific developmental contexts, I focus on two turning points: ages 9 and 66, periods linked to substantial endocrine, social, and health-related transitions. Using longitudinal data from the ABCD Study and UK Biobank, I will assess how individuals’ structural networks deviate from population motifs. These deviations will be related to environmental factors, including pubertal development and peer network changes in the second epoch, and variation in social roles, daily structure, and physical health in later adulthood. Taken together, this work aims to advance our understanding of how universal rules of brain organization are embedded within and shaped by unique developmental trajectories affected by individuals’ exposomes.

[1] Sydnor, V. J., et al. "Neurodevelopment of the association cortices: Patterns, mechanisms, and implications for psychopathology." Neuron (2021)

[2] Tooley, U. A., Danielle S. Bassett, and Allyson P. Mackey. "Environmental influences on the pace of brain development." Nature Reviews Neuroscience (2021)

[3] Mousley, A., et al. "Topological turning points across the human lifespan." Nature communications (2025)

Childhood maltreatment and low socioeconomic status have been linked to reduced total and subfield-specific hippocampal volume. However, evidence of childhood adversity effects on hippocampal microstructure is scarce and limited to single-event trauma or adult samples. Here, we investigated the prospective developmental effects of adversity on hippocampal T1w/T2w ratio, an in vivo proxy for microstructure, while replicating prior findings on volume and examining hippocampal thickness as an intermediate phenotype. We analyzed behavioral and neuroimaging data of the Adolescent Brain Cognitive Development (ABCD) study at baseline (N = 5297) and assessed the hippocampus globally and along its anterior-posterior and proximal-distal axes using HippUnfold, an automated segmentation approach. To parse family-related adversity, Exploratory and Confirmatory Factor Analyses were performed on youth- and caregiver-reported questionnaires. Additionally, neighborhood disadvantage and income-to-needs ratio were included as socioeconomic indicators. Individual adversity domains were examined using Partial Least Squares Regression, while the effects of cumulative adversity exposure were tested using Linear Mixed Models based on a summed adversity score. Our results indicated that socioeconomic adversity was associated with alterations in hippocampal volume, thickness, and T1w/T2w ratios at global and axis-specific levels. Specifically, higher neighborhood disadvantage and income-to-needs ratio showed negative and positive associations, respectively. Within family-related adversity, reduced hippocampal volume was associated with an adverse family environment, whereas parental neglect predicted increases in hippocampal thickness. At the level of overall adversity load, cumulative adversity exposure was linked to reductions in hippocampal volume, thickness, and T1w/T2w ratio, particularly along the anterior-posterior and proximal-distal axes. Together, these findings extend prior work on volumetric alterations by demonstrating that adversity and cumulative adversity exposure relate to additional markers of hippocampal morphology at both global and axis-specific levels. The observed adversity-specific and cumulative associations suggest that different forms of adversity may affect distinct neurodevelopmental processes, reflected in alterations across hippocampal structural axes.

Social cognition encompasses several cognitive and affective processes and requires the complex interplay of various brain processes. Previous work has evidenced bidirectional associations between social cognitive deficits and deficient structural connectivity between relevant brain areas, suggesting that structural connectivity and white matter (WM) integrity might be an essential foundation for social cognitive abilities. Methods PubMed, Scopus, and Web of Science were systematically searched for studies reporting correlations between diffusion-based WM measures and behavioral SoC measures. Study characteristics including sample details, SoC measures, diffusion metrics, and statistical procedures of total of n=63 studies containing k=782 effect sizes in 4508 individuals were extracted. Whole-brain voxel-wise studies reporting correlations with fractional anisotropy (FA) were analyzed using Seed-based d Mapping (SDM). Studies reporting tract-wise diffusion measures were included in an effect size meta-analysis using (Robust) Bayesian model averaging. Subgroup and moderator analyses were conducted for SoC constructs, age groups, diagnoses, and white matter tracts. Results The voxel-wise meta-analysis (15 studies) showed no overall convergence across studies. However, construct-specific analysis revealed associations between empathy and FA in the corpus callosum, left striatum, and left superior longitudinal fasciculus. The tract-wise meta-analyses provided evidence for an overall association between SoC and WM integrity with substantial heterogeneity and risk of publication bias. Across models, the effect was most strongly moderated my age (older adults) and diagnosis (neurodegeneration, schizophrenia) while socio-cognitive constructs rarely differed. The tracts with strongest evidence for an effect were the uncinate fasciculus (UF), cingulum and corpus callosum, though only the UF model survived correction for publication bias. Conclusion The analyses provided evidence for an association between white matter integrity and social cognition, although results across studies are highly heterogeneous. Several major tracts — particularly the uncinate fasciculus — appear broadly relevant for SoC. Future work may benefit from network-based approaches and individualized tractography to better capture the structural connectivity underlying social cognition.

Magnetic resonance imaging (MRI) is sensitive to subject motion. Motion during acquisition can cause blurring and ghosting artefacts, leading to decreased image quality and loss of details in the image. With the advent of high-resolution imaging, quantitative MRI (qMRI) and in-vivo histology using MRI (hMRI), the need for more accurate and more robust motion correction methods for brain MRI has increased significantly over the last decade. This led to the development of many motion tracking methods, each having its own set of limitations. Some require a marker to be attached to the participant, some require the imaging sequence to be modified, some might be sensitive to certain types of motion and some might not be applicable for prospective motion correction. In this project we propose using three different radiofrequency field based methods for head tracking, as they do not suffer from the aforementioned limitations: field camera, pilot tone (PT) and beat pilot tone (BPT). A field camera consists of field probes that detect the magnetic field variations in the scanner. Typically a field camera is used to correct for system imperfections but it is also sensitive to the motion of the imaged subject, through the susceptibility change caused by the motion. For tracking the motion with pilot tone, a sinusoidal signal (which is the pilot tone itself) with frequency in the MR bandwidth but outside the image bandwidth is transmitted. This signal is superimposed on the k-space, and can be extracted as the transmitted frequency is known. The signal is modulated by the motion, and can be used to track it. Beat pilot tone takes pilot tone one step further. Instead of one signal near the Larmor frequency, two signals with a frequency difference near the Larmor frequency are transmitted. By exploiting the non-linearity of the receive elements, a signal corresponding to the frequency difference is received, which can be processed just like the pilot tone. This opens the way for using GHz range signals which are much more sensitive and can track smaller motion. By using a ground-truth optical motion tracking method, which is the Moiré Phase Tracker (MPT) in this case, we are going to acquire a dataset of ground-truth motion information and sensor data from the radiofrequency methods. We aim to use this information to calibrate the sensors so that radiofrequency tracking could be used for motion correction and thereby greatly improving image quality on its own.