Peter Zeidman’s research while affiliated with Wellcome Centre for Human Neuroimaging and other places

What is this page?


This page lists works of an author who doesn't have a ResearchGate profile or hasn't added the works to their profile yet. It is automatically generated from public (personal) data to further our legitimate goal of comprehensive and accurate scientific recordkeeping. If you are this author and want this page removed, please let us know.

Publications (154)


SPM 25: open source neuroimaging analysis software
  • Preprint
  • File available

January 2025

·

214 Reads

·

Nicholas A. Alexander

·

Nicole Labra Avila

·

[...]

·

Peter Zeidman

Statistical Parametric Mapping (SPM) is an integrated set of methods for testing hypotheses about the brain's structure and function, using data from imaging devices. These methods are implemented in an open source software package, SPM, which has been in continuous development for more than 30 years by an international community of developers. This paper reports the release of SPM 25.01, a major new version of the software that incorporates novel analysis methods, optimisations of existing methods, as well as improved practices for open science and software development.

Download

Dynamic Causal Models of Time-Varying Connectivity

November 2024

·

25 Reads

This paper introduces a novel approach for modelling time-varying connectivity in neuroimaging data, focusing on the slow fluctuations in synaptic efficacy that mediate neuronal dynamics. Building on the framework of Dynamic Causal Modelling (DCM), we propose a method that incorporates temporal basis functions into neural models, allowing for the explicit representation of slow parameter changes. This approach balances expressivity and computational efficiency by modelling these fluctuations as a Gaussian process, offering a middle ground between existing methods that either strongly constrain or excessively relax parameter fluctuations. We validate the ensuing model through simulations and real data from an auditory roving oddball paradigm, demonstrating its potential to explain key aspects of brain dynamics. This work aims to equip researchers with a robust tool for investigating time-varying connectivity, particularly in the context of synaptic modulation and its role in both healthy and pathological brain function.


Cognitive performance declines with higher AD pathological load
Cognitive performance is represented by the baseline PACC5 score, which was normalized to the unimpaired sample (cognitively normal individuals, subjective cognitive decliners, first-degree relatives of Alzheimer’s disease patients). A Quadratic model: PACC5 = b0 + b1 ⋅ PL² + c ⋅ COV. The black line depicts the predictions of a regression model (with 95% confidence intervals) with a quadratic effect of PL. B Same model as in panel (A), but with additional terms for years of education and its interaction with the quadratic PL score. Red and blue dots refer to individuals with high and low education, respectively, as obtained by a median split. Red and blue lines are the predictions of regression models for an individual with average covariate values and 17 (median of the high education group) or 12 years of education (median of the low education group), respectively. Shaded areas refer to the respective 95% confidence intervals. Source data are provided as a Source Data file. COV covariates (see “Methods”), PACC5 Preclinical Alzheimer’s Cognitive Composite 5, PL pathological load.
CR-related activity pattern that moderates effects of pathology
A Activation (hot colors) and deactivation (cool colors) during encoding of subsequently remembered compared to subsequently forgotten stimuli as identified by t-contrasts of the subsequent memory regressor in the whole fMRI sample. T values of voxels with pFWE < 0.05 are shown. B Group-level CR-related activity pattern that when expressed minimizes effects of AD pathology on cognitive performance as identified via a multivariate approach. The net contribution (moderation coefficient; positive/hot and negative/cool colors) of every voxel to the CR pattern is displayed (unthresholded). C Atrophy pattern in the whole baseline DELCODE sample as obtained by a VBM GM analysis of CN participants vs ADD patients. T values of voxels with pFWE < 0.05 are shown. Source data are provided as a Source Data file. ADD Alzheimer’s disease dementia, CN cognitively normal, CR cognitive reserve, GM gray matter, VBM voxel-based morphometry.
Significant regions in the CR-related activity pattern
Several clusters of voxels were determined via bootstrapping whose contribution to the CR pattern (wi) was found to be significant (p < 0.05, see “Methods”), displayed as A mosaic (multislice) view, B 3D view and C surface view. Displayed numbers refer to the clusters described in Table 2 with peaks in the following brain structures. 1: right inferior temporal cortex, 5: left precuneus. In panel (B), small clusters have been removed for illustrative purposes. It is important to note that the CR pattern is multivariate in nature, interpretable as a whole and cluster descriptives are reported for transparency of obtained non-negligible coefficients contributing to the pattern. Source data are provided as a Source Data file. CR cognitive reserve.
CR pattern and the subsequent memory contrast predominantly align
A The scatter plot displays the CR coefficients wi and subsequent memory contrast coefficients (beta) for every voxel with significant contribution to CR. They form three groups: 1. A concordant where both coefficients have the same sign (blue); 2. positive CR coefficient, but negative subsequent memory beta (CR+SM-; yellow); 3. negative CR coefficient, but positive subsequent memory beta (CR-SM+; green). The histograms display the frequency of the voxels in the corresponding groups. The gray dashed lines separate the four quadrants. B The CR-related activation pattern is shown color-coded corresponding to the colors in panel (A). It is important to note that the CR pattern is multivariate in nature, interpretable as a whole and cluster descriptives are reported for transparency of obtained non-negligible coefficients contributing to the pattern. Source data are provided as a Source Data file. CR cognitive reserve, SM subsequent memory.
Subsequent memory-related activity moderates the relationship between PL and PACC5
The relationship between the PL score and the PACC5 score (Box-Cox transformed and residualized for covariates) is moderated depending on the subsequent memory-related activity in two previously identified clusters (see Table 2 or Fig. 3). A Moderation effect of activation in cluster 1 located around the inferior temporal cortex including fusiform gyrus (positive moderation coefficients). B Moderation effect of deactivation in cluster 5 including bilateral cuneus and precuneus as well as posterior cingulate (negative moderation coefficients). The red lines in both panels depict the predicted PACC5 score for a hypothetical individual with an activation 1 SD above the mean, the blue lines for an activation 1 SD below the mean in the respective cluster. The shaded areas represent the 95% confidence intervals. Black dots represent the individual subjects' values for PL and (transformed + residualized) PACC5. Source data are provided as a Source Data file. PACC5 Preclinical Alzheimer’s Cognitive Composite 5, PL pathological load.

+1

Cognitive reserve against Alzheimer’s pathology is linked to brain activity during memory formation

November 2024

·

161 Reads

The cognitive reserve (CR) hypothesis posits that individuals can differ in how their brain function is disrupted by pathology associated with aging and neurodegeneration. Here, we test this hypothesis in the continuum from cognitively normal to at-risk stages for Alzheimer’s Disease (AD) to AD dementia using longitudinal data from 490 participants of the DELCODE multicentric observational study. Brain function is measured using task fMRI of visual memory encoding. Using a multivariate moderation analysis, we identify a CR-related activity pattern underlying successful memory encoding that moderates the detrimental effect of AD pathological load on cognitive performance. CR is mainly represented by a more pronounced expression of the task-active network encompassing deactivation of the default mode network (DMN) and activation of inferior temporal regions including the fusiform gyrus. We devise personalized fMRI-based CR scores that moderate the impact of AD pathology on cognitive performance and are positively associated with years of education. Furthermore, higher CR scores attenuate the effect of AD pathology on cognitive decline over time. Our findings primarily provide evidence for the maintenance of core cognitive circuits including the DMN as the neural basis of CR. Individual brain activity levels of these areas during memory encoding have prognostic value for future cognitive decline.


Figure 4: a) Design matrix used for second-level analysis, with an intercept as first column and the continuous regressor as second column. b) Recovered spectral density, showing both an effect of the continuous regressor and an effect of the task condition. c) Effect of the continuous regressor on the power (left) and exponent (right) of the aperiodic component. The dashed black line indicates the value that was used to generate the data. The blue line indicates regressed mean of the effect. The shaded area surrounding the mean indicates the 95% confidence interval. d) Effect of the continuous regressor on the power (left), frequency (middle), and width (right) of the alpha mode. e) Effect of the continuous regressor on the power (left), frequency (middle), and width (right) of the beta mode.
Figure 6: Bayesian model comparison results for the LEMON dataset. a) The upper most plot indicates the absence (black) or presence (white) of a particular peak in each model. For instance, a delta peak was present in models 9 to 16. The lower most plot indicates the model evidence, as estimated by the free-energy, for each of the 16 models. b) Log Bayes factors for the presence of each peak. These are obtained by averaging the free-energy over models that feature a particular peak, and subtracting the average free-energy over models without that peak. The dotted horizontal lines indicate a Bayes factor of 3, commonly accepted a threshold for strong evidence. Both theta and alpha peak have decisive evidence, while beta peak only has strong evidence.
Figure 8: Effect of age on model parameters of the aperiodic component (a) and theta (b), alpha (c), and beta modes (d). For each plot, the blue line indicates regressed mean of the effect, whilst the shaded area surrounding the mean indicates the 90% confidence interval.
Parameters of interest, their link function, and default priors. The function σ appearing in the link function for the peak frequency is the hyperbolic arc-tangent.
BSD: a Bayesian framework for parametric models of neural spectra

October 2024

·

18 Reads

The analysis of neural power spectra plays a crucial role in understanding brain function and dysfunction. While recent efforts have led to the development of methods for decomposing spectral data, challenges remain in performing statistical analysis and group-level comparisons. Here, we introduce Bayesian Spectral Decomposition (BSD), a Bayesian framework for analysing neural spectral power. BSD allows for the specification, inversion, comparison, and analysis of parametric models of neural spectra, addressing limitations of existing methods. We first establish the face validity of BSD on simulated data and show how it outperforms an established method (\fooof{}) for peak detection on artificial spectral data. We then demonstrate the efficacy of BSD on a group-level study of EEG spectra in 204 healthy subjects from the LEMON dataset. Our results not only highlight the effectiveness of BSD in model selection and parameter estimation, but also illustrate how BSD enables straightforward group-level regression of the effect of continuous covariates such as age. By using Bayesian inference techniques, BSD provides a robust framework for studying neural spectral data and their relationship to brain function and dysfunction.


Evaluating Models of the Ageing BOLD Response

October 2024

·

46 Reads

·

1 Citation

Neural activity cannot be directly observed using fMRI; rather it must be inferred from the hemodynamic responses that neural activity causes. Solving this inverse problem is made possible through the use of forward models, which generate predicted hemodynamic responses given hypothesised underlying neural activity. Commonly‐used hemodynamic models were developed to explain data from healthy young participants; however, studies of ageing and dementia are increasingly shifting the focus toward elderly populations. We evaluated the validity of a range of hemodynamic models across the healthy adult lifespan: from basis sets for the linear convolution models commonly used to analyse fMRI studies, to more advanced models including nonlinear fitting of a parameterised hemodynamic response function (HRF) and nonlinear fitting of a biophysical generative model (hemodynamic modelling, HDM). Using an exceptionally large sample of participants, and a sensorimotor task optimized for detecting the shape of the BOLD response to brief stimulation, we first characterised the effects of age on descriptive features of the response (e.g., peak amplitude and latency). We then compared these to features from more complex nonlinear models, fit to four regions of interest engaged by the task, namely left auditory cortex, bilateral visual cortex, left (contralateral) motor cortex and right (ipsilateral) motor cortex. Finally, we validated the extent to which parameter estimates from these models have predictive validity, in terms of how well they predict age in cross‐validated multiple regression. We conclude that age‐related differences in the BOLD response can be captured effectively by models with three free parameters. Furthermore, we show that biophysical models like the HDM have predictive validity comparable to more common models, while additionally providing insights into underlying mechanisms, which go beyond descriptive features like peak amplitude or latency, and include estimation of nonlinear effects. Here, the HDM revealed that most of the effects of age on the BOLD response could be explained by an increased rate of vasoactive signal decay and decreased transit rate of blood, rather than changes in neural activity per se. However, in the absence of other types of neural/hemodynamic data, unique interpretation of HDM parameters is difficult from fMRI data alone, and some brain regions in some tasks (e.g., ipsilateral motor cortex) can show responses that are more difficult to capture using current models.


Figure 2. Modulatory connection strengths in the winning models relative to the OBS and EXE tasks surviving at BMA thresholding at >95% posterior probability (strong evidence). Red dots represent positive modulation effects during the VF observation (A1), control observation (A2), VF execution (B1) and control execution (B2). Yellow arrows indicate driving inputs where the visual and motor information of action start.
Two distinct networks for encoding goals and forms of action: An effective connectivity study

June 2024

·

236 Reads

·

2 Citations

Proceedings of the National Academy of Sciences

Goal-directed actions are characterized by two main features: the content (i.e., the action goal) and the form, called vitality forms (VF) (i.e., how actions are executed). It is well established that both the action content and the capacity to understand the content of another’s action are mediated by a network formed by a set of parietal and frontal brain areas. In contrast, the neural bases of action forms (e.g., gentle or rude actions) have not been characterized. However, there are now studies showing that the observation and execution of actions endowed with VF activate, in addition to the parieto-frontal network, the dorso-central insula (DCI). In the present study, we established—using dynamic causal modeling (DCM)—the direction of information flow during observation and execution of actions endowed with gentle and rude VF in the human brain. Based on previous fMRI studies, the selected nodes for the DCM comprised the posterior superior temporal sulcus (pSTS), the inferior parietal lobule (IPL), the premotor cortex (PM), and the DCI. Bayesian model comparison showed that, during action observation, two streams arose from pSTS: one toward IPL, concerning the action goal, and one toward DCI, concerning the action vitality forms. During action execution, two streams arose from PM: one toward IPL, concerning the action goal and one toward DCI concerning action vitality forms. This last finding opens an interesting question concerning the possibility to elicit VF in two distinct ways: cognitively (from PM to DCI) and affectively (from DCI to PM).


Important concepts of the theory of dynamical systems. (A) An example of flow in state-space (grey arrows), governing the evolution of trajectories (coloured curves) from different initial states (coloured circles). (B) The corresponding trajectories in the time domain for both x1 and x2 axes. (C) An example of bifurcation of the flow: trajectories converge towards a fixed point of state space when the bifurcation parameter α is below a critical value αc, and towards a limit cycle when the bifurcation parameter is above the critical value (Andronov-Hopf bifurcation). (D) An example of a multistable system: the attractor to which the trajectory evolves depends on the initial state, as indicated by the colours of the trajectories.
Taxonomy of the different modelling frameworks discussed here. The key factors guiding the selection of a particular framework are the nature of dynamics (discrete or continuous) and the nature of state space (stochastic or deterministic). In addition, the nature of the inputs (stochastic or deterministic) has relevance for continuous deterministic systems. In the particular case of continuous deterministic systems with stochastic inputs, one can use a linear response function, which is the first-order term of the Volterra kernel representation of the system, to directly approximate the outputs from the inputs without reference to the states. Effectively, this implies that the dynamics do not need to be integrated over time, which greatly simplifies model inversion. In all other cases, model inversion entails tracking the states or their distribution through time.
Multiscale dynamics of brain signals: mapping slow and fast variables. (A) Slow quantities in the brain, such as synaptic efficacy between regions, exhibit large time constants and evolve slowly over time. The evolution of two slow variables are illustrated here, as yellow and purple lines. (B) The evolution of slow variables can also be represented by dynamics in a slow state space. (C) Importantly, for every location in the slow state space (horizontal axes), there is a corresponding mode of fast dynamics (vertical axis). Three modes are depicted here, numbered 1–3. These fast dynamics give rise to rapid brain signals, such as field potentials in pyramidal neurons (D). The mathematical relationship between the slow and the fast timescale is given by dt = εdT (ε ≪ 1); in other words, the dynamics at faster scale t unfolds over a fraction (ε) of the slower scale T. In summary, the brain is understood to navigate slowly (A) through a repertoire of fast stable dynamics (D). Crucially, the slow variables are directly linked to the dynamics of the fast variables (C). Similarly, changes in the fast variables’ dynamics can be attributed to changes in the slow variables. Therefore, modelling the complex dynamics of multiscale dynamical systems can be simplified by focusing on the dynamics of the slow variables and the mapping from slow to fast variables.
Illustration of the centre manifold theorem with a three-dimensional dynamical system. (A) The three-dimensional state-space of the system. The blue surface is the centre manifold, and gives a height x = h(θ1, θ2) to each point of the (θ1, θ2) plane. Trajectories initialized away from the manifold converge rapidly towards the surface (black curves). This is due to the presence of a strong flow orthogonal to the centre manifold (B) for a section of state space. The strong flow (green arrows) converges towards the centre manifold (blue curve). The flow parallel to the manifold (blue arrows) is weaker by orders of magnitude. Hence, trajectories quickly collapse to the centre manifold before evolving alongside it. This is reflected in the exponential decay of the distance to the manifold (C). Hence, the x-component of the trajectory is well approximated by a static function from the location on the (θ1, θ2) plane (D). This can motivate an adiabatic approximation: as the rapidly changing x component of the trajectory can be approximated by the mapping h(θ1, θ2), we may consider x as a spurious dimension of the system and restrict our description to the evolution on the (θ1, θ2) plane; in other words, we can approximate the fast vanishing states by a fixed mapping from the slow states.
Hierarchical modelling approach used to link slow effects to fast observations. First, the authors extracted sliding window data from their LFP time series (A). Then, they estimated the power spectral density for each window, to produce a time-frequency representation of the data (B). Then they fitted a state-space model, called a canonical microcircuit (CMC) dynamic causal model (DCM), for each window of power spectral densities (C). This resulted in a time course of posterior densities for the parameters of the DCM models. Finally, the authors added a second level to the model to test for between-window effects, enabling them to evaluate hypotheses of interest, that is, the interaction between interventions (PTZ concentration, presence or absence of NMDAr-Ab) and the parameters of the CMC model (D). Adapted with permission from R. E. Rosch et al. (2018).
Linking fast and slow: The case for generative models

April 2024

·

110 Reads

·

7 Citations

A pervasive challenge in neuroscience is testing whether neuronal connectivity changes over time due to specific causes, such as stimuli, events, or clinical interventions. Recent hardware innovations and falling data storage costs enable longer, more naturalistic neuronal recordings. The implicit opportunity for understanding the self-organised brain calls for new analysis methods that link temporal scales: from the order of milliseconds over which neuronal dynamics evolve, to the order of minutes, days, or even years over which experimental observations unfold. This review article demonstrates how hierarchical generative models and Bayesian inference help to characterise neuronal activity across different time scales. Crucially, these methods go beyond describing statistical associations among observations and enable inference about underlying mechanisms. We offer an overview of fundamental concepts in state-space modeling and suggest a taxonomy for these methods. Additionally, we introduce key mathematical principles that underscore a separation of temporal scales, such as the slaving principle, and review Bayesian methods that are being used to test hypotheses about the brain with multiscale data. We hope that this review will serve as a useful primer for experimental and computational neuroscientists on the state of the art and current directions of travel in the complex systems modelling literature.


Active Data Selection and Information Seeking

March 2024

·

260 Reads

·

2 Citations

Algorithms

Bayesian inference typically focuses upon two issues. The first is estimating the parameters of some model from data, and the second is quantifying the evidence for alternative hypotheses—formulated as alternative models. This paper focuses upon a third issue. Our interest is in the selection of data—either through sampling subsets of data from a large dataset or through optimising experimental design—based upon the models we have of how those data are generated. Optimising data-selection ensures we can achieve good inference with fewer data, saving on computational and experimental costs. This paper aims to unpack the principles of active sampling of data by drawing from neurobiological research on animal exploration and from the theory of optimal experimental design. We offer an overview of the salient points from these fields and illustrate their application in simple toy examples, ranging from function approximation with basis sets to inference about processes that evolve over time. Finally, we consider how this approach to data selection could be applied to the design of (Bayes-adaptive) clinical trials.


Efficacy of a gamified digital therapy for speech production in people with chronic aphasia (iTalkBetter): behavioural and imaging outcomes of a phase II item-randomised clinical trial

February 2024

·

43 Reads

·

2 Citations

EClinicalMedicine

Background Aphasia is among the most debilitating of symptoms affecting stroke survivors. Speech and language therapy (SLT) is effective, but many hours of practice are required to make clinically meaningful gains. One solution to this ‘dosage’ problem is to automate therapeutic approaches via self-supporting apps so people with aphasia (PWA) can amass practice as it suits them. However, response to therapy is variable and no clinical trial has yet identified the key brain regions required to engage with word-retrieval therapy. Methods Between Sep 7, 2020 and Mar 1, 2022 at University College London in the UK, we carried out a phase II, item-randomised clinical trial in 27 PWA using a novel, self-led app, ‘iTalkBetter’, which utilises confrontation naming therapy. Unlike previously reported apps, it has a real-time utterance verification system that drives its adaptive therapy algorithm. Therapy items were individually randomised to provide balanced lists of ‘trained’ and ‘untrained’ items matched on key psycholinguistic variables and baseline performance. PWA practised with iTalkBetter over a 6-week therapy block. Structural and functional MRI data were collected to identify therapy-related changes in brain states. A repeated-measures design was employed. The trial was registered at ClinicalTrials.gov (NCT04566081). Findings iTalkBetter significantly improved naming ability by 13% for trained items compared with no change for untrained items, an average increase of 29 words (SD = 26) per person; beneficial effects persisted at three months. PWA’s propositional speech also significantly improved. iTalkBetter use was associated with brain volume increases in right auditory and left anterior prefrontal cortices. Task-based fMRI identified dose-related activity in the right temporoparietal junction. Interpretation Our findings suggested that iTalkBetter significantly improves PWAs’ naming ability on trained items. The effect size is similar to a previous RCT of computerised therapy, but this is the first study to show transfer to a naturalistic speaking task. iTalkBetter usage and dose caused observable changes in brain structure and function to key parts of the surviving language perception, production and control networks. iTalkBetter is being rolled-out as an app for all PWA and anomia: https://www.ucl.ac.uk/icn/research/research-groups/neurotherapeutics/projects/digital-interventions-neuro-rehabilitation-0 so that they can increase their dosage of practice-based SLT. Funding 10.13039/501100000272National Institute for Health and Care Research, Wellcome Centre for Human Neuroimaging.



Citations (57)


... Reinforcement learning is a prime candidate framework, but it tends to be costly in terms of computational resources and training time [3]. A more appropriate framework for resource-constrained mechatronic systems is active inference, which characterizes itself by including optimal information gain in its data acquisition protocol [22,21]. Here we present scalar active inference agents that are coupled together to jointly control a mechatronic system with multiple inputs and multiple outputs [20]. ...

Reference:

Coupled autoregressive active inference agents for control of multi-joint dynamical systems
Active Data Selection and Information Seeking

Algorithms

... Logopenic PPA causes relatively different problems, as phonological processing and working memory are the more prominent impairments. Researchers in their study [86] the option of using tDCS to activate the posterior parietal region responsible for phonological retrieval to improve sentence repetition and word retrieval in patients. After some time, the routine use of tDCS resulted in consistent improvement in language performance due to the increased synaptic plasticity at the temporoparietal junction. ...

Efficacy of a gamified digital therapy for speech production in people with chronic aphasia (iTalkBetter): behavioural and imaging outcomes of a phase II item-randomised clinical trial
  • Citing Article
  • February 2024

EClinicalMedicine

... Regardless of the recording modality, slow fluctuations changes in neural activity and confounds neuronal coupling will are expected to impact the measured signals in a nonlinear fashion and need to be acknowledged assessed to ensure the soundness validity of any statistical analysis. This can be addressed by constructing statistical models of how the these data are generated, i.e., building generative models that explicitly accomodate these slow fluctuations (Medrano et al., 2024b). Far from the ambition of perfectly tackling all the complexities underlying brain dynamics, generative models can provide an idealised summary of the key mechanisms generating observed brain responses, recapitulating the activity in a sparse set of regions using a few parameters with a clear biophysical meaning, such as average number of synaptic projections or average membrane potentials of a population. ...

Linking fast and slow: The case for generative models

... Such differences during the course of AD are potentially reflective of disease-related functional reorganization and adaptation. In the whole sample (n = 493), successful memory encoding was related to activation of a large network including lingual gyri, occipital, and prefrontal regions, while widespread deactivations were observed in the precuneus, posterior cingulate cortex, inferior parietal lobule, and frontotemporal regions ( Fig. 2A), replicating previous observations in the same dataset 39,40,42 and in independent cohorts 35,37,38,43 . We observed that more advanced disease scores were related to hyperactivation in the precuneus, inferior parietal lobule, and posterior cingulate cortices bilaterally, as well as anterior cingulate cortex and superior frontal gyrus (Fig. 2B, for non-linearities, we did not find indications for inverted U-shaped associations between disease stage and successful memory encoding. ...

Cognitive Reserve Against Alzheimer's Pathology Is Linked to Brain Activity During Memory Formation

... This variational free energy is a lower bound on the (log) marginal likelihood (the term in the denominator of Bayes' rule; Box 3), and in this context, its maximization is equivalent to finding the parameters of an approximate posterior distribution that optimally balances model fit and model complexity (captured by the likelihood and prior, respectively, in the numerator of Bayes' rule; Box 3). Conveniently, the marginal likelihood (or model evidence) facilitates straightforward model comparison, where the Bayes factor represents the ratio of model evidence assigned to a given set of observations under competing models and permits comparison of alternative (inverted) dynamic causal models (differing, for example, in terms of which effective connections between regions are set to zero) 43 . Thus, DCM not only facilitates inference of effective connectivity via model inversion but also guides the selection of the simplest network structure that explains the observed data well. ...

A primer on Variational Laplace (VL)
  • Citing Article
  • August 2023

NeuroImage

... CEN), temporal networks, and subgenual ACC on ECT response was suggested by Leaver et al. [85]. A role of the connectivity between the CEN and the salience network in predicting ECT response has also been reported [86]. Moreover, important contributions from the fronto-limbic network connectivity in ECT response prediction have been demonstrated by two studies [84,87]. ...

Effective resting-state connectivity in severe unipolar depression before and after electroconvulsive therapy
  • Citing Article
  • July 2023

Brain Stimulation

... This deficiency may result in misperceptions, sensory overload, distractibility and cognitive fragmentation. 33 We have previously reported a link between working memory deficits in schizophrenia and aberrant functional connections in brain regions contributing to the integration of sensory and motor resources. 34 The findings of this study may contribute new insights into the role of sensorimotor regions in the working memory deficits observed in schizophrenia. ...

Dysconnection and cognition in schizophrenia: A spectral dynamic causal modeling study

... We observed a significant association between mild SARS-CoV-2 infection and functional metrics in one subcortical brain area, the left amygdala. Previous literature linked COVID-19 to structural and functional brain changes and cognitive deficits [6,[42][43][44]. Our results suggest that decreased functional connectivity in the amygdala competitively mediated the COVID-19 associated disruption in cognitive function, specifically spatial working memory. ...

Amygdala connectivity related to subsequent stress responses during the COVID-19 outbreak

... This evidence has led to a divestment in tb-fMRI research and a transition to research using more reliable indices, including brain structure and resting state fMRI (Supplemental Figure 1), as well as an increased focus on multivariate and machine learning approaches that combine measures to capture more variance 4 . However, by inducing specific cognitive states, tb-fMRI captures unique information that is not present in structural or resting state data 5,6 , and the majority of machine learning approaches require exceptionally large samples when effects are small 7 . There thus remains an urgent need to improve task-fMRI-derived indices of brain function, particularly to increase the utility of already acquired data and enable the design of informative future studies. ...

An information-theoretic analysis of resting-state versus task fMRI

... The recognition of this input information by the frontal lobe, including the orbitofrontal cortex, is referred to as top-down processing 25 . In patients with PD, VH is related to the altered integration of sensory input (bottom-up) and prior knowledge (top-down) within the visual system 26 . Several studies have previously reported functional connectivity reduction between bottom-up processing and the right insula, right hippocampus, left putamen, and bilateral caudate in PD-VH compared to healthy controls 27,28 . ...

Changes in both top-down and bottom-up effective connectivity drive visual hallucinations in Parkinson’s disease

Brain Communications