41) Normative Modeling of Resting-State MEG Data: Insights into Aging Brain Dynamics
Seyed Mostafa Kia1, Mohammad Zamanzadeh Nasrabadi1, Tomas Ros2, Thomas Wolfers3, Andre Marquand4, Marie Šafář1, Marijn van Wingerden1
1Tilburg University, Netherlands, 2University of Geneva, Switzerland, 3University of Tübingen, Germany, 4Radboud University Nijmegen Medical Centre, Netherlands
Normative modeling has recently emerged as a powerful framework for decoding the heterogeneity inherent in neuropsychiatric disorders. However, deriving lifespan normative ranges of functional imaging-derived phenotypes (fIDPs), such as magnetoencephalography (MEG) recordings, presents several technical challenges due to data and modeling complexities. In this study, we benchmark various configurations of the hierarchical Bayesian regression (HBR) algorithm for their potential in modeling non-linear, skewed, and heteroscedastic aging effects on several fIDPs in resting-state MEG data. We utilized the Cambridge Centre for Ageing Neuroscience (CamCAN) dataset, which includes more than 600 participants aged 18-88. The normative ranges are derived for periodic components (absolute and relative power spectrum in canonical and individualized frequency bands) and aperiodic components (offset and exponent) of the signal at the coarse sensor level. Our results show an overall decreasing trend in Alpha oscillations and an increasing trend in Beta oscillations with aging, while Theta and slow Gamma oscillations remain stable. Our observations further highlight the importance of i) adjusting the periodic power spectrum for the aperiodic component, especially for the Gamma band, and ii) using model selection to address the diverse effects of aging on different fIDPs. Overall, our findings confirm the feasibility of deriving reliable aging charts from MEG data. We plan to extend our model by i) incorporating additional datasets, ii) deriving normative ranges at finer spatial resolutions in the source space, and iii) benchmarking the model on clinical populations.
42) EEG correlates of multimodal feedback assisted movement learning
Gábor Péter Háden1
1Research Centre for Natural Sciences, Budapest, Hungary
Human motor learning is facilitated by action-perception loops (APLs) that develop from birth. Learning any new movement involves the reorganization and refinement of these loops. Perception can be limited to a single modality; however, the most effective learning is ensured by feedback from multiple modalities (visual, auditory, haptic perception). Knowledge about the neural background of APLs and particularly the interaction with multimodal perception is still limited.
To understand the relationship between motor learning and multimodal perception an experiment was conducted on adult participants while recording EEG. The participants had to navigate an "invisible maze" by moving their hands over a table. A neural network-based model identifies hand positions. The hand position is compared to a map and auditory and/or visual feedback is given if the participant hits the “wall” of the maze. Solving the maze requires learning a specific sequence of movements. Learning performance is measured in several procedurally generated mazes. Learning is operationalized as an increase in the accuracy and speed of repeated task performance, as well as the neural-motor synchronization observed during movement repetition and its changes.
Hypotheses were (1) learning can be detected from changes in functional brain networks during task performance and from changes in EEG-movement synchronization, (2) the effect of auditory and visual varies while there is a transfer between modalities. Preliminary results show that trial success can be predicted from EEG-movement synchronization and functional networks differ between feedback modalities.
43) Fluctuations in neural sensory encoding and prior expectation jointly account for biased auditory decision-making
Sarah Tune1, Iris Borschke1, Nikolai Dürrbeck1, Jonas Obleser1
1University of Lübeck, Germany
Perceptual decisions depend on integrating sensory evidence over time. Yet, human perceptual performance reveals different sources of suboptimality: Prior expectations introduce perceptual biases; neural sensory encoding fluctuates; and not least the sensory evidence get weighted unequally over time. Here we ask how these processes jointly impact auditory decision-making and its metacognitive corollaries.
In this EEG study (N=32, 18–33 yrs), participants listened to 1-sec trains of 20 clicks, each presented to either left or right ear (adapted from [1]). A Bernoulli process randomly presented clicks to either ear. Using a joystick, participants decided which side had more clicks. Prior expectations were manipulated via a preceding visual cue: For half of the trials, an informative (80% valid) cue indicated which ear was more likely to receive more clicks in a given trial.
We here report evidence for cue-related alterations at different levels: Psychophysical analysis showed a symmetric shift in response bias for informative vs. neutral cues. A psychophysical kernel (“reverse correlation”) analysis revealed more optimal (i.e., more uniform) temporal integration following informative cues. Cue-related changes in perceptual decisions were in part explained by fluctuations in neural sensory encoding, with lower neural encoding accuracy predicting less optimal decision-making.
In sum, our results show how suboptimal perceptual performance arises from biases situated along different stages of the processing hierarchy.
44) Topographical relocation of sleep spindles and their relation to internalizing problems: a longitudinal approach
Johanna Gfüllner1
1University of Salzburg, Austria
Introduction:
Research has established a strong link between affective disorders and sleep problems. Specifically, lower sleep spindle activity has been connected to internalizing Problems (IP) in children and adolescents. However, most studies have focused on examining this relationship at a single point in time. Here, we investigated the topographical developmental trajectories of sleep spindles and tested, whether deviations from these patterns predict IP.
Method:
We conducted polysomnographic recordings and measured IP via questionnaires across 15 years in a sample of subjects (N =28, 19 female) at three time points: (1) childhood (~9.5yrs), (2) adolescence (~16yrs), and (3) young adulthood (~22.5yrs). Two nights were recorded at each time point (adaptation and experimental night). To assess the topographical spindle pattern, we measured slow (11-13Hz) and fast (13-15Hz) spindle densities (N/min) during N2 sleep.
Results:
Frontal slow spindle density remained stable between childhood and adolescence (t(27) = 0.65, p = .677, MD = 0.19). A stronger individual decrease in frontal slow spindle density was associated with higher IP scores in adulthood (rsp = - .44, p = .018). Conversely, slow spindle density typically decreased from adolescence to adulthood (t(27) = 7.88, p < .001, MD = 1.57). During this period, a greater decline predicted lower levels of IP in adulthood (rsp = .44, p = .020).
Conclusion:
Our results study provide evidence that frontal slow spindle development is associated with IP during adulthood. We found this association to be dynamic, with the direction of the effect being dependent on the developmental period.
45) Grab it when off guard: State-level approach bias predicts eating under low top-down control
Hannah van Alebeek1
1University of Salzburg, Austria
The tendency to approach food faster than to avoid it (i.e., approach bias) is thought to
facilitate actual food intake, particularly when it comes to foods that conflict with one's
dietary goals. However, this relationship has been difficult to demonstrate, which
potentially relates to an ongoing debate about whether such cognitive-behavioral
biases represent stable traits or rather fluctuating states. Here, we thus investigated
the temporal fluctuations of food approach bias (1), its within-participant association
with food craving (2) and intake (3), as well as the role of top-down control in this biasintake association (4). A total of 76 participants, completed an impulsivity questionnaire
and were asked to perform a smartphone-based approach-avoidance task on nine
separate days. Additionally, they reported their daily craving, intake, and dietary
intention for 12 personalized foods they wanted to eat less or more often over the
whole study period. Multilevel models analyses showed that approach bias varied
considerably within individuals (1), correlated with craving (2) and with intake (3), but
the latter effect was only present for individuals with generally weak dietary intentions
or high impulsivity (4). Results emphasize the need to re-conceptualize approach bias
as comprising both state and trait components, and they indicate that top-down
processes gate the relationship of approach bias with intake. Findings explain why single-session approach bias measures often do not predict distal outcomes such as body weight. Furthermore, interventions targeting approach bias may be tailored to certain timepoints (high-risk situations) and individuals (those with weak intentions and high impulsivity).
46) Preliminary results: The impact of alterations in NREM sleep and progesterone on overnight memory consolidation during early pregnancy
Melanie Wöhler1, Johanna Gfüllner1, Kathrin Bothe1, Christopher Höhn1, Kerstin Hödlmoser1
1University of Salzburg, Austria
Aim: Investigating memory-related sleep parameters and their association with progesterone in early pregnancy.
Design: Five women (MAge = 28.60 +/- 1.67) in their first trimester of pregnancy and nine non-pregnant women (MAge = 25.70 +/- 3.00) in their luteal phase participated in a seven-day experimental block. Polysomnography was recorded during two nights. Progesterone levels were assessed by saliva samples. A declarative word-pair learning task was performed before the experimental night. Sleep architecture, as well as the density of slow (11-13 Hz) and fast (13-15 Hz) sleep spindles, slow oscillations, and their coupling were analyzed.
Results: Pregnant women showed less N3 sleep in the second and third quarters of the night than non-pregnant women. They also tended to have reduced slow sleep spindle density and frontal slow oscillation positive peak amplitudes. Fast sleep spindle density and the precise coupling of spindles with slow oscillations were not altered. Frontal sleep spindle coupling with slow oscillations was positively correlated with overnight memory consolidation in pregnant and non-pregnant women. Progesterone was negatively correlated with fast and slow sleep spindle density in non-pregnant women. Furthermore, the association between progesterone and positive peak amplitudes of slow oscillations at frontal and central sites tended to follow a U-shape. Pregnant women showed comparable overnight memory consolidation to non-pregnant women.
Conclusion: Pregnant women already showed deviations in sleep architecture and memory-related sleep parameters during the first trimester. Some of these parameters were related to progesterone. However, these changes did not significantly impact overnight memory consolidation in this sample.
47) Intact auditory processing of low-level stimulus features but disrupted pre-stimulus prediction-related and feature-specific processing during human sleep
Pavlos Topalidis1
1University of Salzburg, Austria
As attentional resources are naturally minimised in sleep, recent studies have sought to determine the extent to which the brain engages in the predictive processing of auditory inputs. Here, for the first time, we examine the brain's ability to predict or pre-activate the low-level stimulus features of an expected stimulus prior to its actual presentation. In a passive listening paradigm, 34 participants listened to tone sequences comprising of four simple tones (i.e., low to high-pitch), while recording simultaneous EEG and MEG brain activity during wakefullness (20 mins) and a 2.5 hour nap. We presented the tones continuously at a fixed presentation rate (3 Hz), to establish strong temporal predictions, and manipulated the tone transition probabilities to create predictable and unpredictable/random sequences. Using multi-level pattern analysis (MVPA), we show that the low-level stimulus properties of the four tones remain decodable in light non-REM N1 and non-REM N2 sleep. However, compared to wakefulness decoding accuracies dropped significantly and were less sustained over time. In addition, we find that in wakefulness the feature-specific neural activations of an expected tone are even decodable before its actual presentation. Going beyond previous findings, we show that these neuronal prediction or pre-activation patterns are still evident in light N1 sleep, but cease during N2. Altogether, the data suggest that stimulus-specific auditory processing is retained despite the fading of consciousness, while stimulus-specific anticipatory processing is dependent upon minimal levels of conscious processing such as in transitory N1 sleep.
48) The impact of vibrotactile stimulation on the neural tracking of speech
Ioana-Sabina Rautu1, Julie Bertels1, Mathieu Bourguignon1, Veikko Jousmäki2, Xavier De Tiège1
1Université Libre de Bruxelles, Brussels, Belgium, 2Aalto University, Espoo, Finland
In the current study, we investigated whether vibrotactile stimulation derived from the speech temporal envelope can facilitate the neural tracking of speech in normal-hearing listeners. Magnetoencephalographic (MEG) activity from 30 French-speaking participants was recorded while they attended to videos of native French speakers reciting stories with neutral content. The heard audio was presented either in silence or embedded in multi-talker noise, with (Audiotactile [AT] condition) or without (Audio-only [A] condition) vibrotactile stimulation. During the AT conditions, vibrotactile envelope-based stimulation was delivered to the participants’ left palm either congruently (i.e., in synchrony) or incongruently with the speech signal of the attended speaker. Neural tracking was then assessed per condition using speech-brain coherence at the phrasal and syllabic levels (0.2 – 1.5 Hz and 2 – 8 Hz, respectively). Results indicate a robust enhancing effect of the congruent vibrotactile stimulation on the neural tracking of speech. This vibrotactile benefit was present at the syllabic-level tracking and was significantly more pronounced in the right hemisphere. Moreover, the effect of vibrotactile input was particularly enhanced under noise conditions. Contrastingly, no effect of the vibrotactile stimulation was found on the neural tracking of speech at the phrasal level. Vibrotactile input also impacted behavioral performance, with comprehension in noise being significantly improved by congruent vibrotactile input compared with audio-only or incongruent AT conditions.
This study provides novel insights into the mechanisms of proven haptic enhancement of speech comprehension in silent and noisy auditory conditions, with potential implications in the design of haptic auditory devices.
49) Foraging in conceptual spaces: neurophysiological mechanisms of mental search in semantic memory
Giuliano Giari1, Simone Viganò1, Roberto Mai2, Christian F. Doeller3, Roberto Bottini1
1University of Trento, Italy, 2Ospedale Niguarda Ca' Granda, Milan, Italy, 3Max Planck Institute for Human Cognitive and Brain Sciences, Max Planck Society, Leipzig, Germany
The medial temporal lobe (MTL) represents both spatial and abstract relational information. However, how do we search and access stored knowledge? During navigation the hippocampal formation displays rhythmic oscillatory activity in the theta band (3-10 Hz), conveying information about traveled distance and velocity. Here we asked if this physiological signature of physical exploration also extends to mental exploration of abstract spaces, such as when recalling concepts from memory.
We used stereo-EEG to record local field potentials from the MTL of epileptic patients performing a categorical verbal fluency task, randomly “foraging” for concepts from different categories (animals, professions, or famous cities).
Preliminary results from 11 patients indicate that, in the period preceding the utterance of a word, theta power in MTL was significantly higher than during or after word pronunciation. This effect was independent of the semantic category that was mentally explored, and was more pronounced in the MTL than lateral temporal cortices. Furthermore, we used linear mixed models to test whether theta power was modulated by semantic distances between words, modeled using distributional semantics. We observed significant modulation from -1000 to -600 ms relative to word onset when considering high-dimensional semantic distances. In contrast, low-dimensional semantic distances more strongly modulated theta power in the time interval from -300 to -100 ms.
Although preliminary these results suggest that physiological signatures of hippocampal activity during physical exploration might also extend to mental exploration of abstract spaces, and potentially reveal novel mechanisms underlying the access of conceptual information from memory.
50) CAUSAL NEURAL SYNCHRONY IN ATTENTIONAL SIGNAL TRANSMISSION: TMS-EEG EXPERIMENT
María Melcón1
1University of Glasgow, United Kingdom
Neural synchronization has been suggested as a mechanism for information flow, with slower alpha/beta frequencies associated with feedback signals and gamma frequencies linked to feedforward propagation. Combining Electroencephalography (EEG) and online Transcranial Magnetic Stimulation (TMS) can provide insights into these dynamics by examining their causal role. To this purpose, EEG activity was measured in 32 participants performing a near-threshold visuospatial cueing task while single-pulse TMS was applied to key nodes of the attentional network (FEF, V5, and a control site, foot area) during the orienting period. Addiontally, sine gratings with concentric motion were presented in the half of the trials to enhance feedforward propagation. Time-frequency amplitude and intertrial phase coherence (ITPC) were calculated relative to the TMS pulse. The control condition, a no-attentional cue, was used to isolate the TMS effect on the neural fingerprints of attention, being subtracted from each spatial cue (left and right). Alpha/beta and gamma bands' amplitude and ITPC were analyzed in pre- and post-TMS pulse windows using a four-way repeated measures ANOVA. Preliminary results indicate an interaction between the time window and stimulus presence, with alpha amplitude reduced after the TMS pulse only in trials with stimulus presentation. This suggests that FEF TMS disrupts the attentional feedback process when feedforward propagation is previously enhanced. Further analysis of control site data will determine if the effect of stimulation depends on the brain's attentional state and the role of brain oscillations in information flow.