The Krakencoder: a multi-modal connectome fusion tool for brain-behavior mapping

Prof. Amy Kuceyeski

Thursday 08 January 2026, 15:00-16:00

In this talk, Dr. Kuceyeski will introduce the Krakencoder – her lab’s new tool for multi-modal integration of brain network measurements that allows mapping between connectome processing pipelines as well as between structural and functional connectomes. This AI tool allows state‑of‑the‑art preservation of inter‑individual differences in brain networks through a shared latent space. In an extension from the original publication earlier this year, she will demonstrate the flexibility of the tool for jointly minimizing site differences (harmonization) as well as maximizing brain‑behavior mapping accuracy for various metrics. She will also demonstrate how the tool can be used to create interpretable feature importances when mapping to behavior or demographics. This tool represents the power of AI in better understanding the mapping between brain structure and function and behavior.

Complex Structure and Dynamics in Brain Networks

Prof. Olaf Sporns

Department of Psychological and Brain Sciences, Indiana University, Bloomington IN 47405, USA

Thursday, 04 December 2025, 15:00-16:00

How does the brain work as an integrated system? Understanding the organization of brain graphs (networks), their community structure and time‑varying topology is key to tackling this challenge. Network Neuroscience is an emerging field dedicated to understanding the structure and function of neural systems across scales, from neurons to circuits and ultimately to the whole brain. In this presentation I will review a few current themes and future directions. These include connectomics, structure/function relationships as disclosed in neuroimaging data, the growing use of computational models to map information flow and communication dynamics, and a novel edge‑centric approach to track dynamic functional connectivity. I will try to identify some key issues along three main dimensions: networks, dynamics, and information.

Changing Connectomes: Changes during development as well as changes after brain stimulation interventions to improve cognition

Prof. Marcus Kaiser

University of Nottingham, UK

Thursday, 20 November 2025, 15:00-16:00

The complete set of connections in the brain is called our connectome. Over the last 20 years we have found out more about how this network is organised and how it develops. I will outline some characteristic network features and how they could originate during brain network development. In particular, I will highlight the impact of the spatial organisation and embedding of brain networks and the role of delays on brain dynamics and cognition. Moreover, I will outline how interventions based on connectome information and computational models might be able to improve cognitive function for brain and mental health conditions. For this, I will also discuss how network metrics change after stimulation as well as our ongoing work on developing closed‑loop neuromodulation for brain and mental health (depression, schizophrenia, Tourette’s). Finally, I will give some information on the launch of our N3 Centre for Neurotechnology, Neuromodulation, and Neurotherapeutics at the University of Nottingham.

Causal Discovery and Network Inference for Biological and Biomedical Data

Prof. Hervé Isambert

Institut Curie

Thursday, 06 November 2025, 15:00-16:00

Discovering causal effects is at the core of scientific investigation but remains challenging when mostly observational data is available. In essence, causal discovery infers cause‑effect relations from specific correlation patterns involving at least three variables, which goes beyond the popular notion that pairwise correlation does not imply causation. Yet, in practice, causal representations have been difficult to learn and interpret, particularly for high‑dimensional data such as state‑of‑the‑art biological and biomedical data. In this talk, I will outline some network reconstruction methods and a broad range of applications. In particular, our group has developed novel causal discovery methods and tools (MIIC, CausalXtract, MIIC‑sdg, CausalCCC, MIIC search & score) to learn cause‑effect relationships in a variety of biological or biomedical data, from single‑cell transcriptomics and live‑cell imaging data to clinical data from medical records of patients. These machine‑learning methods combine multivariate information analysis with interpretable graphical models and outperform other methods on a broad range of benchmarks, particularly on complex non‑linear datasets, while allowing for unobserved latent variables that are ubiquitous in biomedical applications.

The Divided Mind

Prof. Ed Bullmore

King's College London

Thursday, 23 October 2025, 15:00-16:00

I will talk about this recently published book: why did I write it, what was I trying to say that it would have been difficult to communicate in a peer‑reviewed paper, and who was I hoping might read it? I will summarise the content and key themes of the book along the following lines, but I hope to leave a good amount of time for discussion. Briefly, the book tries to tell two occasionally interwoven histories. First, the world history of what we now call schizophrenia, especially the controversy between Freudian (brainless) and Kraepelinian (mindless) tribes, the dark crisis of the Kraepelinian concept of dementia praecox before and during World War II, and its long‑lasting imprint on how we continue to think about schizophrenia to this day. Second, my personal story as I became a psychiatrist and tried to get to grips with scientific questions about the origins of schizophrenia and the prospects for better treatments or preventions in future.

Neurotransmitter receptor gradients enable flexible control of cortex‑wide network dynamics during cognition

Dr. Seán Froudist-Walsh

Thursday, 12 June 2025, 15:00-16:00

Recent advances in neuroanatomy and physiology make it possible to probe whole‑brain mechanisms of cognition and behavior. However, as yet, few models in computational neuroscience have tackled the mechanisms underlying highly distributed neural activity during cognitive tasks. In this talk I will describe our anatomy‑led approach to developing multiscale connectome‑based models of neural dynamics during cognitive tasks and our recent anatomical findings of gradients of neurotransmitter receptors in the cortex. I will highlight our investigations into how inputs from neuromodulatory systems (e.g., dopamine, serotonin) may dynamically alter the connectome, shifting the cortical dynamical landscape, and how this can confer flexibility on distributed cognitive functions such as conscious perception and working memory.

Comprehensive profiling of anaesthetised brain dynamics across phylogeny

Dr. Andrea Luppi

University of Oxford

Thursday, 15 May 2025, 15:00-16:00

Abstract: Intrinsic dynamics of neuronal circuits shape information processing. Combining neuroimaging with causal perturbation offers the opportunity to understand how local dynamics mediate the link between neurobiology and functional repertoire. We compile a unique dataset of multi‑scale neural activity during wakefulness and anaesthetic‑induced suppression of information processing encompassing human, macaque, marmoset, mouse and nematode. Applying massive feature extraction, we comprehensively characterise local neural dynamics across over 6,000 time‑series features. Using dynamics as a common space for cross‑species comparison reveals a phylogenetically conserved dynamical profile of anaesthesia, characterised by reduced intrinsic timescales and dampened inter‑regional synchrony, confirmed by computational modelling. Spatially, this profile covaries with conserved transcriptional profiles of neurotransmission across human, macaque and mouse cortex. This dynamical regime is experimentally reversed in vivo by deep‑brain stimulation of the macaque centromedian thalamus, restoring behavioural responsiveness. Altogether, comprehensive dynamical phenotyping reveals that across species and scales, anaesthetics induce spatiotemporal isolation of local neural activity.

Multiscale Organization of Neuronal Activity Unifies Scale-Dependent Theories of Brain Function

Dr. Mac Shine

University of Sydney

Thursday, 01 May 2025, 10:00-11:00

Abstract: Analysis of cellular recordings from five phylogenetically diverse species uncovers a conserved, multiscale organization of neuronal activity that resolves disparate theories of brain function. This hierarchical structure allows the brain to operate across multiple timescales, enhances information flow, flexibly reconfigures activity during behavior, and balances functional resiliency with efficiency across scales.

Reproducible Brain Charts: Accelerating Studies of Brain Development with Open Science

Dr. Ted Satterthwaite

University of Pennsylvania

Thursday, 20 March 2025, 15:00-16:00

Abstract: Major mental illnesses are increasingly understood as disorders of brain development. Neuroimaging studies of brain development can help track healthy brain maturation and may identify deviations linked to mental illness, but large samples are required. The talk discusses the rationale, challenges and lessons from the Reproducible Brain Charts (RBC) initiative—an open data resource aimed at aggregating, curating and harmonizing large-scale developmental neuroimaging data to accelerate psychiatric research.

Revolutionizing Connectomics using Generative, Federated and Holistic Learning

Dr Islem Rekik

Imperial College London

Thursday 06 March 2025, 15:00-16:00

The brain is a highly complex network that can be represented as a matrix using multimodal MRI (including T1‑weighted and resting-state fMRI). This talk provides an overview of prime graph neural network models designed for generating brain connectivity across various dimensions (scale, time, domain), aiming to revolutionize connectomics by learning to generate brain matrices without acquiring additional MRI data. It also covers emerging trends such as federated connectome generation.

Algorithmic Information Dynamics

Dr Hector Zenil

King's College London

Thursday 20 February 2025, 15:00-16:00

Abstract: The talk introduces Algorithmic Information Dynamics (AID), a framework based on information theory, compressibility, algorithmic probability and perturbation analysis to study objects—particularly evolving networks—from the perspective of their generating mechanisms in software space. It explores how AID can characterize aspects of living systems and recent research on biosignatures.

Challenges and opportunities for biological network inference

Dr Kimberly Glass

Thursday 06 February 2025, 15:00-16:00

Abstract not available.

The Genetics of Person‑specific fMRI Networks

Dr Aaron Alexander‑Bloch

Thursday 23 January 2025, 15:00-16:00

Variability in the spatial layout of human brain functional networks on the cortex is an understudied phenotype, particularly in imaging‑genetic studies aimed at understanding the genetic basis of fMRI networks. Early results show that individual-specific topography of Person‑specific Functional Networks (PFNs) is strongly associated with domains of psychopathology and cognition during adolescence. PFNs capture individualized aspects of brain function with unique associations to clinical and developmental outcomes compared to standard fMRI approaches. Targeting PFNs may accelerate discovery of genetic contributions to brain organization, leading to mechanistic insights into genetic risks for behavioral health conditions. (Additional biographical details about Dr Alexander‑Bloch were provided in the talk description.)

Robust brain functional connectivity network estimation

Dr Hanâ Lbath

Université Grenoble Alpes

Thursday 07 December 2023, 15:00-16:00

Functional connectivity estimation between brain regions is challenging because noise and regional heterogeneity can lead to under‑ or over‑estimation of inter‑regional correlation. Existing methods handle either issue but not both. The speaker presented a novel clustering‑based non‑parametric estimator that offsets the impact of noise and arbitrary intra‑regional correlation through intermediate aggregation. Empirical validation on synthetic data and real rat and human rs‑fMRI datasets demonstrated its effectiveness. The method also provides an empirical distribution of the inter-correlation. An ongoing extension introduces distribution‑weighted connectivity networks.

Network effects in the neurobiology of psychiatric disorders

Prof. Alex Fornito

Monash University, Australia

Thursday 16 November 2023, 10:00-11:00

Abstract not available.

Idiosynchrony: Using naturalistic stimuli to draw out individual differences in brain and behavior

Dr. Emily Finn

Dartmouth, USA

Thursday, 09 November 2023, 15:00-16:00

Neuroimaging studies often collapse data across individuals, but understanding how brain function varies across people is critical for basic science and translational applications. Dr Finn’s work shows that whole‑brain functional connectivity patterns serve as a “fingerprint” that can identify individuals and predict trait‑level behaviors. While fingerprints can be detected during rest and task conditions, manipulating brain state using naturalistic paradigms (e.g., movie watching, story listening) enhances aspects of these patterns most relevant to behavior. The talk also discusses extensions to the inter‑subject correlation framework that model both shared responses and individual variability to naturalistic stimuli.

Exploring the human connectome

Prof. Martijn van den Heuvel

CNCR, VU Amsterdam, Netherlands

Thursday, 02 November 2023, 15:00-16:00

The human brain network is an efficient communication network. This talk highlighted advances in micro‑ and macroscale connectomics and discussed theories about the “principles of wiring” that may govern connectome organization and shape cognition, functional specialization and diversity. Principles such as cost-effective wiring, short communication relays and network hubs were considered across species, along with the idea that modifications in brain connectivity may have played a role in recent human cognitive evolution and are relevant to neurological and neuropsychiatric conditions.

Broad Transcriptomic Dysregulation Occurs Across the Cortex in ASD

Prof. Michael Gandal

University of Pennsylvania (UPenn)

Thursday, 18 May 2023, 15:00-16:00

The talk addressed molecular dysregulation in autism spectrum disorder (ASD). RNA‑sequencing of 725 samples across 11 cortical areas from 112 post‑mortem brains showed widespread transcriptomic changes across the cortex in ASD, with an anterior‑to‑posterior gradient and pronounced differences in primary visual cortex. Single-nucleus RNA‑seq and methylation profiling indicated that changes reflect cell-type‑specific gene expression affecting excitatory neurons and glia. Rare and common ASD-associated genetic variations converge on downregulated synaptic signalling modules, while common variation alone is enriched in an upregulated protein-chaperone module. The results highlight pervasive molecular changes across the cerebral cortex in ASD.

From Connectome to Computation Through the Genomic Bottleneck

Prof. Tony Zador

Cold Spring Harbor Laboratory

Thursday, 20 April 2023, 15:00-16:00

Artificial neural networks (ANNs) have revolutionized artificial intelligence, but they still fall far short of animals in terms of learning speed and sample efficiency. One reason is the so‑called “genomic bottleneck”: genes do not have enough capacity to specify the entire wiring diagram of the brain. This talk explored how animals overcome this bottleneck—leveraging genetically encoded wiring rules—and how similar principles might inform the design of more efficient ANNs.

Network approaches to psychopathology: An overview

Prof. Denny Borsboom

University of Amsterdam

Thursday, 19 January 2023, 15:00-16:00

In the network approach to psychopathology, mental disorders are conceptualized not as latent disease entities but as sets of causally connected symptoms. This perspective offers new insights into why symptoms co‑occur, how disorders develop and persist, and why comorbidity is common. The talk provided an overview of theoretical foundations, methodological advances, and empirical findings in the symptom‑network literature, highlighting opportunities and challenges for future research and clinical practice.

Relations and Predictions in Brains and Machines

Dr. Kim Stachenfeld

DeepMind

Thursday, 02 March 2023, 15:00-16:00

Humans and animals learn and plan with flexibility and efficiency far beyond that of modern machine‑learning methods, owing in part to their ability to build structured representations of their environments and rapidly adapt these representations to new situations. This talk discussed theoretical work connecting hippocampal and entorhinal representations to algorithms for relational reasoning and planning. It also explored how predictive models that capture long‑range correlations can enable agents to generalize to novel tasks and environments.

Linking cortical microstructure to in vivo measures of cortical structure in health and disease

Dr. Konrad Wagstyl

University College London

Thursday, 02 February 2023, 15:00-16:00

Advances in neuroimaging now allow us to infer aspects of cortical microstructure—such as layer thickness and neuronal density—from non‑invasive data. This talk presented efforts to create atlases linking post‑mortem histology with in vivo MRI, enabling researchers to interpret structural MRI findings in terms of underlying cellular architecture. It also showcased how these tools can improve detection of subtle cortical lesions in conditions like epilepsy and help explain inter‑individual variability in cortical thickness.

Neural network models as mechanistic explanations of brain information processing

Prof. Nikolaus Kriegeskorte

Columbia University

Thursday, 16 February 2023, 15:00-16:00

Abstract not available.

The brain in the social world: Integrating approaches from social psychology, neuroscience, and social network analysis

Dr. Carolyn Parkinson

UCLA

Thursday 08 December 2022, 16:00-17:00

This talk will cover work integrating theory and methods from psychology, neuroscience, and social network analysis to examine how people track, encode, and are influenced by the social networks that they inhabit. One set of studies tests if, when, and how people retrieve knowledge of familiar others’ positions in their real‑world social networks when encountering them. Related research tests how this knowledge, once retrieved, shapes downstream processing and behavior. A second set of studies tests if human social networks exhibit assortativity in how their members perceive, interpret, and respond to their environment. Consistent with this possibility, we find that proximity between people within their social networks is linked to similar neural responses to naturalistic stimuli, similar subjective construals of such stimuli, and similar patterns of brain connectivity. A final set of studies examines how shared understanding relates to overall levels of social connectedness within communities. We find that people who process the world in a manner that is more reflective of community norms have greater overall levels of subjective and objective social connection. All human cognition is embedded within social networks, but research on information processing within individuals has progressed largely separately from research on the social networks in which individuals are embedded. The set of findings to be reviewed in this talk suggests that integrating approaches from social psychology, neuroscience, and social network analysis can provide new insights into how individuals perceive, shape, and are shaped by the structure of their social world.

Changing Connectomes: How Brain Network Changes are Linked to Cognition in Health and Disease

Prof. Marcus Kaiser

University of Nottingham, UK

Thursday 24 November 2022, 15:00-16:00

The complete set of connections in the brain is called our connectome. Over the last 20 years we have found out more about how this network is organised and how this organisation is linked to brain function. I will outline how characteristic network features arise during evolution, how they are linked to brain function, and how they originate during individual brain development. For example, small‑world features enable the brain to rapidly integrate and bind information while the modular architecture, present at different hierarchical levels, allows separate processing of various kinds of information while preventing wide‑scale spreading of activation. Hubs play critical roles in information processing and are involved in many brain diseases. Long‑distance connections are crucial to reduce the number of processing steps. Connectomes show more long‑distance connections than would be expected if energy minimisation would be a primary goal of network optimisation. Indeed, both C. elegans and the macaque show non‑optimal component placement. In a recent paper we show that this non‑optimal organisation also occurs for the human connectome. Moreover, such a spatial layout has benefits in network dynamics allowing for easier switching between brain states. Therefore, there might be distinct cognitive benefits; conversely, we might expect cognitive deficits in brain network disorders to be linked to an altered spatial organisation of brain networks. Finally, I will outline how information about topological and spatial connectome changes can be a starting point for personalised interventions with non‑invasive brain stimulation using focused ultrasound.

Analyzing artificial neural networks to understand the brain

Dr Grace Lindsay

Thursday 10 November 2022, 14:00-15:00

In the first part of this talk I will discuss work demonstrating that recurrent neural networks can replicate broad behavioral patterns associated with dynamic visual object recognition in humans. An analysis of these networks shows that different types of recurrence use different strategies to solve the object recognition problem. The similarities between artificial neural networks and the brain presents another opportunity, beyond using them just as models of biological processing. In the second part of this talk, I will discuss a proposed research plan for testing a wide range of analysis tools frequently applied to neural data on artificial neural networks. I will present the motivation for this approach as well as the form the results could take and how this would benefit neuroscience.

Network-based approaches towards studying context-specific cell signalling

Dr Evangelia Petsalaki

 Thursday 27 October 2022, 15:00-16:00

It is well established that signalling responses happen through complex networks. However, most signalling research still uses linear pathways as the ground truth. Moreover, signalling responses are highly dependent on context, such as tissue type, genetic background etc and therefore these static pathways are not always suitable. There is also a high bias in the literature towards kinases and pathways for which reagents and prior knowledge is readily available. This leaves a huge dark space in our understanding of cell signalling and significantly hinders studies of its general principles. Our group uses data‑driven and network‑based approaches to understand and describe the organisation principles of cell signalling that allow the diverse and context‑specific cell responses and phenotypes. In this talk I will showcase different network‑based methods that we have developed and/or use to extract phenotype‑specific networks from omics data and use it to study different diseases. First, I will present a method that combines paired transcriptomics and imaging data to extract context‑specific signalling networks, with the context in this case cell shape in breast cancer. The method is generalisable to any paired transcriptomics/phenotype data, and I will briefly mention how we have extended it to study the disease progression of non‑alcoholic fatty liver disease. I will finally present a project where we integrated RNAseq, ATAC‑seq and ChIP‑seq data to create a network representative of endothelial dysfunction and performed in silico perturbations to identify potential targets for the condition.

Network function in human cerebral organoids as a platform for mechanistic and therapeutic advances in cognitive disorders

Dr Susanna Barrett Mierau

Thursday 30 June 2022, 15:00-16:00

Human cerebral organoids offer an extraordinary in vitro cellular model for studying human brain development and early disturbances in neurologic disease. Microelectrode array (MEA) recordings are commonly used to compare neuronal activity in 2D and 3D cultures. Yet, MEA recordings can also reveal cellular‑scale network activity, including patterns or motifs in network function seen across spatial scales from cellular to whole brain networks. We have used MEA recordings from human air‑liquid interface cerebral organoids to study network function and maturation. We have also demonstrated intact neuronal network function development with MEA recordings in a human cerebral organoid model of amyotrophic lateral sclerosis with frontotemporal dementia. To facilitate investigations of network development in organoids and the impact of disease‑causing perturbations, we created a MATLAB network analysis pipeline (MEA‑NAP) for batch analysis of MEA experiments to compare network function over time and conditions (e.g., genetic mutation or drug treatment). This user‑friendly, open source diagnostic tool can process raw voltage time‑series acquired from single‑ or multi‑well MEAs and automatically infer key network properties from organoids or 2D human (or murine) neuronal cultures. Our pipeline enables users to perform MEA analysis beyond standard measures of activity or correlation alone to identify differences in network topology and roles of individual nodes in network activity. Our analyses of network function in organoids demonstrate that they can serve as a platform for investigating disease mechanisms and screening new therapies.

Discovering and quantifying patterns in networks with coloured nodes

Dr Vincenzo Nicosia

 Thursday 16 June 2022, 15:00-16:00

Graphs with coloured nodes provide an informative model for complex systems whose units are associated to a discrete number of classes. Some relevant examples include social systems, geographic networks, and cell adjacency networks. Quite often, as in the case of geographical networks, the arrangement of those classes is an important aspect of the global organisation of the system. Hence, quantifying the existence of heterogeneity and correlations in the assignment of the nodes of a graph to classes is paramount to characterise the behaviour of a system. In this talk we will cover the basics of networks with coloured nodes, and we will show how a simple set of measures, based on random walks on the graph, can be effectively used to measure the existence of correlations and heterogeneity among classes. Interesting applications include the quantification of spatial segregations in cities, the identification of polarisation in social systems, the emergence of robust spatial organisation in plant tissues, and the incorporation of metadata in community detection tasks.

ENIGMA at 10: What have we learned from a decade of large scale collaborative meta‑analysis in neuroimaging

Prof Sarah Medland

Thursday 05 May 2022, 11:00-12:00

Abstract not available.

A Connectomic Hypothesis for the Hominization of the Brain

Prof Claus Hilgetag

University Medical Center Hamburg Eppendorf

Thursday 21 April 2022, 15:00-16:00

Abstract not available.

Inflammation and mental health: questions of causality

Prof Ed Bullmore

 University of Cambridge

Thursday 31 March 2022, 16:00-17:00

There is not much room for doubt that inflammation is associated with depression and other mental health disorders. However, there are many open questions about the causal relationships that might explain the increasing mass of largely correlational findings. In this somewhat speculative talk, I will explore the idea that early life adversity and other social stressors could epigenetically reprogramme immune response and brain developmental trajectories leading to long‑term changes in vulnerability to depression and other mental health disorders in adult life.

Title not available

Dr Marta Zlatic

Thursday 17 June 2021, 15:00-16:00

Abstract not available.

Human brain functional genomics and the mechanisms underlying genetic risk for schizophrenia

Dr Michael Gandal

Thursday 20 May 2021, 15:00-16:00

Our understanding of the pathophysiology of neuropsychiatric disorders, including schizophrenia, lags greatly behind other fields of medicine. Defining genetic contributions to disease risk can provide a rigorous foothold for mechanistic understanding, and over the past decade, large‑scale genetic studies have successfully identified hundreds of genetic variants robustly associated with schizophrenia. However, mechanistic insight and clinical translation continue to lag the pace of risk variant identification, hindered by the sheer number of targets and their predominant noncoding localization, as well as pervasive pleiotropy and incomplete penetrance. Successful next steps require identification of “causal” genetic variants and their proximal biological consequences; placing variants within biologically defined functional contexts, reflecting specific molecular pathways, cell types, circuits, and developmental windows; and characterizing the downstream, convergent neurobiological impact of polygenicity within an individual. Comprehensive transcriptomic profiling in human brain can provide a quantitative biological context for interpreting the molecular effects of disease‑associated genetic variants and for identifying shared and distinct molecular pathways disrupted across major neuropsychiatric disorders. Here, I will discuss our recent work as part of the PsychENCODE Consortium to generate a large‑scale functional genomic resource of the human cortex, integrating genotype and RNA‑seq data from more than 2000 samples, including over 500 derived from individuals with schizophrenia. We find pervasive differential splicing and expression, with changes at the transcript isoform‑level—as opposed to the gene level—showing the largest effect sizes, genetic enrichments, and disease specificity. Coexpression networks identify a glial‑immune signal demonstrating shared disruption of the blood‑brain‑barrier and up‑regulation of NFκB‑associated genes, as well as disease‑specific alterations in microglial‑, astrocyte‑, and interferon‑response modules. Finally, we leverage the transcriptome‑wide association approach to identify 64 high confidence candidate risk genes. This large‑scale integration of genomic data in human brain enables a comprehensive systems‑level view of the neurobiological architecture of major neuropsychiatric illness and provides a resource for mechanistic insight and therapeutic development.

Statistical analysis and optimality of biological systems

Prof Gasper Tkacik

Thursday 22 April 2021, 15:00-16:00

Normative theories and statistical inference provide complementary approaches for the study of biological systems. A normative theory postulates that organisms have adapted to efficiently solve essential tasks and proceeds to mathematically work out testable consequences of such optimality; parameters that maximize the hypothesized organismal function can be derived ab initio, without reference to experimental data. In contrast, statistical inference focuses on the efficient utilization of data to learn model parameters, without reference to any a priori notion of biological function. Traditionally, these two approaches were developed independently and applied separately. Here, we unify them in a coherent Bayesian framework that embeds a normative theory into a family of maximum‑entropy “optimization priors.” This family defines a smooth interpolation between a data‑rich inference regime and a data‑limited prediction regime. I will illustrate how this framework can productively guide our thinking on several neuroscience and non‑neuroscience examples.

Transdiagnostic mapping in neurodevelopmental disorders

Dr Duncan Astle

Cambridge CBU

Thursday 25 March 2021, 15:00-16:00

Abstract not available.

Structure-function coupling in brain networks

Dr Bratislav Misic

McGill University

Thursday 11 March 2021, 15:00-16:00

Structure‑function relationships are a fundamental principle of many naturally occurring systems, including brain networks. Collective communication among connected neuronal populations is thought to support patterned neural activity, as well as flexible cognitive operations and complex behavior. Cognitive dysfunction, due to aging or disease, may arise from perturbations in structure‑function coupling. Traditional accounts assume uniform structure‑function coupling throughout the brain, and often leave out important biological detail. Here I will focus on three new directions to study structure‑function coupling: (1) multiplexed models that allow regionally heterogeneous structure‑function relationships, (2) brain networks annotated with micro‑architectural features, and (3) redefining function as a computational property. Altogether, I hope to show that structural network reconstructions enriched with local molecular and cellular metadata, in concert with more nuanced representations of functions and properties, hold great potential for a truly multiscale understanding of the structure–function relationship.

Connectomics and Network Neuroscience: From Topology to Dynamics

Prof Olaf Sporns

Tuesday 12 January 2021, 15:00-16:00

Networks (connectivity) and dynamics are two key pillars of network neuroscience – an emerging field dedicated to understanding structure and function of neural systems across scales, from neurons to circuits to the whole brain. In this presentation I will review current themes and future directions, including structure/function relationships, use of computational models to map information flow and communication dynamics, and a novel edge‑centric approach to functional connectivity. I will argue that network neuroscience represents a promising theoretical framework for understanding the complex structure and functioning of nervous systems.

Combined imaging-genomics approaches to investigate psychiatric risk

Dr Aaron Alexander‑Bloch

Tuesday 15 December 2020, 15:00-16:00

This talk focuses on recent attempts to combine brain imaging and genetics in psychiatric research. I will discuss some of my prior work in psychiatric neuroimaging of individuals with schizophrenia, as well as the limitations of this work, and argue that links with psychiatric genomics are of increasing importance for neuroimaging research. Then, I will briefly discuss three ongoing imaging‑genomics studies: 1) a transcriptome‑wide association study of imaging phenotypes in the UK Biobank; 2) an investigation of copy number variants in the Philadelphia Neurodevelopmental Cohort; 3) an analysis of the correspondence between developmental trajectories derived from imaging and brain gene expression.

How humans build models of the world through temporal contingencies

Prof Danielle Bassett

Tuesday 01 December 2020, 15:00-16:00

Human learners acquire not only disconnected bits of information, but complex interconnected networks of relational knowledge. The capacity for such learning naturally depends on the architecture of the knowledge network itself. I will describe recent work assessing network constraints on the learnability of relational knowledge, and a free energy model that offers an explanation for such constraints. I will then broaden the discussion to the generic manner in which humans communicate using systems of interconnected stimuli or concepts, from language and music, to literature and science. I will describe an analytical framework to study the information generated by a system as perceived by a biased human observer, and provide experimental evidence that this perceived information depends critically on a system’s network topology. Applying the framework to several real networks, we find that they communicate a large amount of information (having high entropy) and do so efficiently (maintaining low divergence from human expectations). Moreover, we also find that such efficient communication arises in networks that are simultaneously heterogeneous, with high‑degree hubs, and clustered, with tightly‑connected modules—the two defining features of hierarchical organization. Together, these results suggest that many real networks are constrained by the pressures of information transmission to biased human observers, and that these pressures select for specific structural features.

Probing and Rescuing Dysfunctional Brain Circuits in Depression

Dr Conor Liston

Tuesday 17 November 2020, 15:00-16:00

Abstract not available.