imitation

Optimizing CD45-Targeted Astatine-211-Radioimmunotherapy for Malignant and Non-Malignant Blood Disorders

ABSTRACT CD45 is expressed on almost all normal and neoplastic hematopoietic cells but not on non-blood cells and has, therefore, been pursued as a drug target. Initially centered on augmenting conditioning before hematopoietic cell transplantation (HCT) for blood cancers, there is increasing interest in expanding CD45-directed therapies into other settings, with radioimmunotherapy (RIT) being the major therapeutic modality so far. Investigators at our institution pioneered CD45 RIT with b-emitters such as iodine-131 (131I) using the murine monoclonal antibody (mAb), BC8. A phase 3 trial testing 131I-BC8 (131I-apamistamab [Iomab-B]) with allogeneic HCT in older adults with relapsed/refractory acute myeloid leukemia showed improved outcomes over conventional care, validating this approach. More recently, attention has shifted toward a-emitters that deliver substantially higher decay energies over much shorter distances than b-emitters, rendering them more suitable for precise and potent target cell killing. In our work, we focus on astatine-211 (211At) for its ideal half-life and decay without a-emitting daughters. For clinical application, mAbs are conjugated with the bifunctional boron cage molecule, isothiocyantophenethyl-ureido-closo-decaborate(2-) (B10-NCS), to enable stable protein astatination. Three early-phase trials testing 211At-BC8-B10 as augmentation of HCT conditioning for patients with malignant and non-malignant blood disorders are ongoing, with emerging data indicating significant anti-tumor efficacy. Nonetheless, relapses still occur. Other important limitations include marked infusion toxicities and human antimouse antibody (HAMA) responses related to the murine nature of BC8 and dimer formation after 211At labeling of mAb-B10 conjugates with tissue residualization from 211At atom oxidation. The latter may contribute to the risk of liver cell injury, the dose limiting extramedullary toxicity of CD45 RIT. As a first step toward our goal of optimizing CD45 RIT, we have raised new, fully human CD45 mAbs as basis for novel therapeutics. In preliminary in vivo studies in immunodeficient mice, we found some of these mAbs to have greater anti-tumor efficacy than a humanized version of BC8 (HuBC8) we generated as a reference mAb. We will now conduct comparative in vivo CD45+ cell targeting (“biodistribution”) and anti-tumor efficacy studies to select a lead candidate mAb for clinical application and use protein engineering to maximize the selectivity and efficacy of targeted radiation delivery. We will use immunodeficient mice xenotransplanted with human leukemia cells for this purpose as no human approaches are available and in vitro testing is inadequate to measure both the targeting and biologic RIT effects on human leukemia cells. Mice provide the in vivo milieu needed for comprehensive evaluation. Development of improved mAb astatination methodologies to minimize off-target toxicities of 211At-RIT will further increase therapy specificity and reduce toxicity. In parallel, we will conduct genome-scale, unbiased target identification/validation studies to identify partner drugs for rational combination therapies aimed at enhancing the anti-tumor efficacy of 211At-CD45 RIT.

Systems Biology of Early Atopy: Role of Human Milk (SunBEAm-Milk)

Surprisingly little is known about the effect of breastfeeding (BF) on infant immune system development besides an effect on the gut microbiome, but its impact on metabolites and Tregs could support protection against food allergy (FA). BF is currently recommended to prevent the development of allergic diseases, especially asthma/recurrent wheezing and AD in early childhood, but firm conclusions could not be drawn regarding FA due to high heterogeneity and low quality of studies. Reverse causation, recall bias and the poor accuracy of outcome assessment are significant limitations. Most are inadequately powered to specific FA; however, a recent study showed that exclusively BF infants had lower odds of egg, sesame, and peanut allergies. Importantly, immunomodulatory composition of HM varies between mothers, which has not been taken into consideration. For over two decades we have been developing methods to assess immunomodulatory factors in the complex matrix of HM and their association with infant FA. We have shown that high levels of HM total and specific IgA are associated with protection against cow’s milk allergy, but it is unclear whether HM IgA is responsible for or is a biomarker of the vertical transfer of protection. Infant fecal and systemic IgA levels during breastfeeding and after weaning are also elevated in infants at low risk for atopic disease raising the question of whether HM factors such as cytokines can promote IgA production in infants. Consistent with this, we showed that HM cytokines, such as APRIL, induce IgA production in naïve infant B cells, and infants receiving HM with higher levels of APRIL had lower incidence of allergic disease. Finally, lower levels of several HM fatty acids including short-chain fatty acids and DHA were associated with FA. While some these factors were are associated with maternal atopic disease, several of them are not and suggest a role for diet instead. The System Biology of Early Atopy (SunBEAm) population-based cohort of 2500 mother-infant pairs is >50% recruited and provides an unprecedented opportunity to assess association of HM feeding and immune factors in HM with development of infant immune system and FA/AD. The Common Sample comprises a subset of 100 dyads with FA, 100 with FA+AD, 100 with AD, 100 with no FA or AD and more extensively profiled biological data. Utilizing all 2-month HM samples available in the Common Sample, we will assess levels of immune factors in HM and their association with maternal/infant characteristics (Aim 1). Utilizing data from the whole cohort, we will assess the association between HM vs formula feeding on well-defined FA/AD further adjusted based on high vs low levels of HM immune components in the Common Sample (Aim 2b). Finally, we will examine the immune cell and epithelial effects of HM on infant immune markers and intestinal organoids (Aim 3). Key findings will be validated in an independent birth cohort. The ultimate goal is to uncover protective properties of BF and HM in FA and subsequent design of policies and prevention strategies to address the increasing rates of FA.

Linking Single-Cell Transcriptomic, Morphological, and Temporal Signatures of Vulnerability in Neurodegeneration

Neurodegeneration involves complex cellular phenotypes and molecular changes that vary widely among the cells of the nervous system. Current methodologies permit either detailed molecular profiling (e.g., single-cell transcriptomics) or functional phenotyping (e.g., live imaging of neuronal activity), but not both in the same cells. Thus, it is difficult to directly link a neuron's functional state or fate with its gene expression profile. To address this limitation, we developed an innovative technology, VISTA-FISH (Video Imaging with Spatial- Temporal Analysis by FISH), that couples prospective live-cell imaging with high-resolution spatial transcriptomic profiling of the same cells. This approach enables in situ comparisons of gene expression in neurons that exhibit divergent behaviors or outcomes. Using VISTA-FISH, we will profile iPS-derived human neurons to link single-cell gene expression, morphology, and temporal phenotypes to study molecular pathways driving resilience as well as susceptibility. After exposing neurons carrying TDP43 and C9orf72 mutations to a stimulus inducing TDP43 aggregation, we will jointly record TDP43 localization and neuron activity using live-cell microscopy, then measure single-cell gene expression of the same cells (Aim 1). We will also combine live-cell measurements of TDP43 half-life with CRISPR screening and single-cell gene expression (Aim 2). These rich datasets will enable us to determine transcriptomic changes associated with differences in protein aggregation, protein synthesis, and protein degradation in individual cells, providing an unprecedented molecular perspective on factors responsible for vulnerability and resilience to neurodegeneration.

Characterization and functional impact of somatic numtogenesis in the human cortex

Project Summary This project focuses on studying nuclear mitochondrial insertions (numts), which are fragments of mitochondrial DNA that get integrated into the nuclear DNA of human cells. While this process, called numtogenesis, occurs naturally and can be passed down to future generations, it has also been observed to occur somatically in our bodies. Historically the function of numts has been difficult to study because they are repetitive and difficult to map with short read sequencing technologies, but there is emerging evidence that they can influence cell function and play a role in diseases, aging, and even complicate genetic studies. Our recent research discovered numts in the human brain’s cortex, and their presence appeared to be linked with earlier death, suggesting they may play a role in aging. However, due to limitations in the data we used, we could not fully explore the extent or impact of these insertions across different tissues or individuals. This project aims to map and study numts in more detail, especially in the human cortex, to further explore this ongoing transfer of DNA from the mitochondria to the nuclear genome and their potential to impact aging and brain function. We will accomplish this by 1) improving sequencing methods to detect numts, 2) comparing their presence across different tissues, and 3) investigating how they affect gene expression and DNA structure. By the end of the project, we aim to provide a model for how such somatic variation may occur and impact cellular function at the tissue level.

AI-enabled methods for de novo design of functional peptides

PROJECT SUMMARY Macrocyclic peptides offer unique therapeutic potential, particularly for targeting intracellular protein-protein interactions considered ‘undruggable’ with traditional therapeutic modalities. Additionally, peptides can combine the benefits and bridge the gap between conventional small molecule therapeutics and large biologics. However, developing new peptide-based therapeutics using traditional approaches, such as natural product discovery or high-throughput library screening, has remained slow and challenging. Moreover, these conventional approaches cover a small fraction of the chemical and structural space, are restricted to a few starting peptide scaffolds, and typically fail to optimize for multiple therapeutic properties simultaneously. Our central hypothesis is that structure-guided deep learning methods can rapidly explore the chemical and structural space beyond natural products and enable precise, rapid, and custom design of functional peptides simultaneously optimized for target binding, selectivity, and membrane permeability. In our recent work, we developed physics-based methods for designing constrained peptides and macrocycles and, more recently, introduced deep learning methods for structure prediction, sequence redesign, and de novo design of peptide monomers and targeted binders. Here, we propose to develop a new generation of structure-guided deep learning (DL) tools to address the current limitations of computational and experimental methods and enable accurate, accessible, and broadly applicable design of macrocycles. Specifically, we will pursue the projects focused on: (i) leveraging DL methods to systematically enumerate the chemical and structural space of constrained peptides and membrane-traversing peptides to develop scaffolds and core design principles for functional peptide design; (ii) high-throughput design and data collection to improve design selection, filtering metrics, and sequence design algorithms; (iii) developing generative DL methods that expand beyond current capabilities and allow sequence and structure design with vast chemical space of non-canonical amino acids; and (iv) use those new generative methods to design macrocyclic binders against different therapeutically-relevant targets, including the critical fusion and attachment proteins from viruses of pandemic concern. Our preliminary work in these proposed areas demonstrates the feasibility of this approach. The proposed computational tools, scaffold sets, and designed peptides will significantly advance therapeutic design beyond the state-of-the-art and enable rapid and custom design of drug- like peptides tailored for addressing complex therapeutic, diagnostic and research challenges.

Structure-function and mechanistic studies of a specific glycosyltransferase complex in fusion-driven pediatric gliomas

Abstract Glycosylation is a co/post-translational modification involved in cell-matrix interactions, antigen-antibody interactions, tumor invasion, and cell motility. Abnormal glycosylation is a hallmark of cancer, with various glycosylation-related genes linked to glioma prognosis and tumor heterogeneity. Pediatric low-grade gliomas (pLGGs) stand as the most common childhood central nervous system tumor, accounting for 30%-40% of all CNS tumors in children. Despite its relatively low mortality rate, pLGGs are associated with devastating lifelong morbidity. The most common alteration found in 75% of tumors is the KIAA1549:BRAF fusion, causing an aberrant activation of the MAPK/ERK signaling pathway. Current treatments, such as traditional chemotherapies and targeted therapies, have limitations such as resistance, lack of specificity, toxicity and paradoxical activation of the MAPK pathway. This highlights the urgent need for novel therapeutic approaches. Investigations into KIAA1549:BRAF-driven pLGGs identified their dependency on the protein-O-mannosyl transferase (POMT) complex for survival. In contrast, BRAFV600E-mutant cells did not show dependency, suggesting the POMT complex as a vulnerability and promising target in KIAA1549:BRAF-driven pLGGs. Therefore, our goal is to characterize the POMT complex structurally and biochemically and study its roles in KIAA1549:BRAF-driven pLGGs. In this proposal, we aim to 1) determine the high-resolution structures of the complex in its unbound, substrate-bound, and inhibitor-bound forms and 2) elucidate the POMT complex mechanisms in KIAA1549:BRAF-driven pLGGs. We will define the critical functional domains, active sites, interaction interfaces and translational modifications crucial for enzymatic activity using cryo-EM techniques, mutagenesis, and functional studies. To study biological pathways and molecular events modulated by the POMT complex, we will implement global proteomics and transcriptomics analysis in well-characterized disease models. In parallel, we will assess the effect of the POMT complex on the MAPK/ERK signaling pathway. This study will guide the structure-based design of probes and drugs targeting the POMT complex and will unveil glycosylation-mediated oncogenesis in pediatric gliomas. It will aid in the development of new targeted therapies and the identification of new biomarkers for pLGGs harboring the KIAA1549:BRAF fusion. The research will be conducted in the Fischer lab at Dana-Farber Cancer Institute, which provides a collaborative and resource-rich environment. The career development plan includes training in scientific writing, mentoring, and presentation skills, as well as interdisciplinary networking with experts in structural biology and pediatric oncology. The candidate’s career goal is to establish an independent research laboratory focused on developing new therapeutic modalities for pediatric neurooncology. The training provided through this fellowship represents a critical step toward achieving this goal.

Targeted Prodrug Cytokines for Metastatic Breast Cancer Immunotherapy

Project Summary. Our approach directly addresses key limitations in targeting and treating metastatic breast cancer, where we propose the selective activation of modular immune-modulating cytokines within the hypoxic and ROS-active TME for delivery across the BBB, providing the necessary pre-clinical data for future clinical translation. The in vitro and in vivo investigations of this novel immunotherapeutic in immunocompetent models will allow our team to study the interplay between tumor-driven immune activation, cytokine signaling, and anti-tumor immunity in both primary and metastatic sites, and establish a robust groundwork for subsequent clinical validation within the OSUCCC. This proposal addresses two key challenges in developing a novel immunotherapy strategy for breast cancer by answering two hypotheses: (1) can a modular immunotherapy platform with tumor-selective activation of prodrug recombinant cytokines overcome these limitations in drug delivery, and (2) can the development of nanobody-cytokine fusions that can selectively target primary breast cancer tumors and cross the BBB to reach metastatic tumor sites? The first hypothesis focuses on achieving tumor environment-specific activation of prodrug-based recombinant cytokines. Protein cytokines are highly potent, and while others have tried to block their activity using a fused genetic linker to ‘mask’ functionality, no one has yet attempted to use a non-canonical-based chemical strategy to achieve this inhibition. Immune-modulating cytokines will be recombinantly expressed with integrated ncAAs that block cytokine activity until the function is regenerated in the breast cancer TME. Once the cytokine activity is controlled, our second hypothesis will be to achieve selective delivery of the cytokine via fusion to nanobodies. While success has been found in targeting primary tumors in drug and protein delivery, a key challenge remains in reaching secondary metastatic tumors in hard-to-reach sites (i.e., brain). Engineered nanobodies, with affinity for breast cancer tumors and the ability to bind to BBB transcytosis receptors, will enable selective delivery to metastatic breast-to-brain tumors, resulting in tumor- specific activation, immune responses, and improved therapeutic outcomes. This system can significantly improve therapeutic outcomes for patients with mBC by integrating selective activation and delivery mechanisms to reduce off-target effects and enhance tumor-specific immune responses in both primary and secondary metastatic tumor sites. Optimizing drug delivery systems to tune immune responses could offer more effective and less invasive treatment options when compared to traditional and engineered cell-based approaches. Our momentum towards precision medicine and targeted therapies holds significant promise for improving outcomes for mBC patients, and has the potential to serve as a pan-cancer treatment for aggressive metastatic cancers from the following aims: (1) generating a modular platform for tumor-specific activation of prodrug cytokines, (2) evaluating cytokine delivery and anti-cancer immune phenotypes in mBC.

I3-BC: Image-Based Infiltrating Immune Cell Detection and Outcomes in Breast Cancer Clinical Trials

PROJECT SUMMARY Tumor infiltrating lymphocytes (TILs) represent an accessible biomarker of the tumor-immune microenvironment (TIME) in breast cancer, demonstrating consistent association with response to neoadjuvant chemotherapy and outcomes in HER2-positive and triple-negative breast cancer. Despite efforts to standardize TIL enumeration from hematoxylin and eosin stained tumor slides, TILs have not gained widespread adoption due to inter- observer variability, and time limitations in pathologic assessment, among others. Further, other key elements of the microenvironment, such as tumor-associated macrophages (TAMs), do not yet have standardized approaches for quantification or characterization. As a result, there is no assessment of the TIME for the vast majority of breast cancers diagnosed in the US and around the world. However, the rapid growth of digital pathology offers the potential to leverage computational approaches to overcome these limitations and democratize access to TIL and TAM enumeration. The overall goal of this project is to determine if computational approaches to TILs (existing) and TAMs (to be developed within this grant) are comparable to pathologist- enumerated TILs and TAMs and, further, associated with relevant patient outcomes from two phase III breast cancer clinical trials. Prior to project initiation, we have developed both a compute-intensive artificial intelligence- based TILs approach, an open source software (QuPath)-based TILs approach, and expertise in RNAseq-based immune quantification. We will first focus on TILs - benchmarking the two computational and RNAseq immune approaches against pathologist TIL counts (‘gold standard’) then evaluating association of each with event-free survival in two completed clinical trials (Aim 1). In parallel, we will develop a novel computational approache to enumerate and phenotype TAMs by using immunohistochemical staining for macrophage markers on the same slide with standard H&E, then apply in the same two clinical trials (Aim 2). Our approach is innovative because we will benchmark diverse approaches at scale in relevant clinical studies. The study is significant because we will determine if computational approaches to TILs/TAMs align with pathologist estimates and clinical outcomes, then ensure these algorithms are available to the community. Our long-term goal is to democratize computational TIL and TAM enumeration as pathology decision-support to facilitate integration of accessible tumor-immune microenvironment into clinical trials and care.

Developing a novel technology for studying T cell differentiation in vivo

Summary CRISPR-based genetic screens have revolutionized our understanding of gene functions and molecular mechanisms across various biological processes. In the field of T cell biology, CRISPR screens have played a pivotal role in identifying genes that impact critical aspects, such as T cell development, differentiation, and function. However, traditional screens have struggled to distinguish genes with diverse mechanisms of action, necessitating further investigations. To address this challenge, researchers have harnessed the power of CRISPR screens combined with single-cell sequencing (scCRISPR-seq), enabling the simultaneous assessment of genetic perturbations and high-dimensional phenotypes at the single-cell level. While scCRISPR- seq has predominantly been performed in vitro using immortalized cell lines, its physiological relevance is limited due to oversimplified biological context and disparities compared to primary cells. This limitation highlights the urgent need for large-scale in vivo scCRISPR-seq with primary T cells. However, various challenges have discouraged its widespread adoption. The use of viral vectors for sgRNA delivery compromises physiological relevance, as the in vitro activation conditions fail to faithfully represent the intricate T cell priming process in vivo. Moreover, viral vector components and continuous Cas9 expression can trigger immunogenicity and cytotoxicity, leading to cell depletion and hindering long-term studies. Additionally, current scCRISPR-seq methods face technical limitations, including low editing efficiency and inadequate perturbation identity recovery rates, which impede efficient large-scale in vivo applications. Fortunately, recent advances in ribonucleoprotein complex (RNP) transfection have addressed many of these challenges. This cutting-edge technology enables efficient gene editing in primary T cells without the need for in vitro activation or permanent Cas9 expression. Leveraging the high editing efficiency of RNP transfection, the investigator’s team aims to develop a novel strategy for in vivo T cell CRISPR screens. This innovative approach involves arrayed RNP transfection and co- transfer of T cells that recognize the relevant antigens. Instead of traditional genetic barcodes, the strategy utilizes congenic markers (CD45.1/45.2 and CD90.1/CD90.2) from donor TCR transgenic T cells as "external barcodes." These markers facilitate the recovery of gene perturbation identity at the single-cell level through the application of CITE-seq. Importantly, this RNP-based strategy seamlessly integrates with existing single-cell sequencing protocols, enabling the comprehensive assessment of transcripts, epitopes, and chromatin accessibility simultaneously. To demonstrate the efficacy of this strategy, the team plans to develop two benchmarking approaches: RNP-CET-seq to investigate the role of TCR regulators in T cell exhaustion and RNP-CATE-seq to map the gene regulatory atlas of exhausted CD8 T cells. In summary, the proposed RNP- based scCRISPR-seq strategy overcomes the limitations of current approaches, enabling large-scale, multi- module in vivo genetic screens within a physiologically relevant context across various disease models.

Uncovering genetic determinants of carbapenem resistance in Klebsiella pneumoniae

Carbapenem-resistant Klebsiella pneumoniae represents an urgent global health threat due to its increasing prevalence and high mortality rates, necessitating a comprehensive understanding of its resistance mechanisms. While key resistance mechanisms and their genetic determinants are known, such as beta- lactamases and porin mutations, the cause of resistance in many strains remains elusive. Moreover, other strains that carry known genetic carbapenem-resistance factors have been found to still be susceptible to carbapenems for unclear reasons. Further, strains can carry genetic elements which, while not conferring resistance directly, can promote resistance indirectly by accelerating its acquisition, such as through mutations in DNA repair systems or mobile genetic elements. To address these knowledge gaps, we propose a genome-wide association study (GWAS), with the aim of maximizing the discovery of gene variants associated with meropenem resistance, with experimental validation of candidates to identify true causal variants. We will overcome limitations of prior studies in the following ways: 1) We have compiled an expanded data set of publicly available K. pneumoniae genomes from strains isolated across a wide distribution of countries, with in hand access to >100 isolates upon which experimental validation studies will be performed. 2) We will perform comprehensive capture of genetic variants by employing a reference-free GWAS, utilizing unitigs, stretches of DNA sequence that represent the entire spectrum of genetic variation. 3) We will enhance statistical power to detect genetic variants with even subtle effects on resistance by using a quantitative, continuous minimum inhibitory concentration (MIC) phenotype to meropenem rather than a binary designation of resistant or susceptible. 4) We will reduce the number of false positives arising from correlation, or linkage disequilibrium (LD), with known carbapenemase and other known resistance factors by performing a conditional GWAS, where known factors are included as covariates. 5) We will further mitigate confounding effects due to population structure and LD, which cause non-random relationships between variants, by utilizing a pangenome-wide regression with an elastic net penalty. 6) Crucially, we will functionally validate our findings, which will include genetic variants associated with increased resistance, whether through direct or indirect mechanisms, as well as those that may restore susceptibility in strains already possessing known resistance factors. We will bridge the gap between GWAS findings and functional validation by leveraging our high-throughput experimental capabilities. This integrated approach promises to uncover novel mechanisms of carbapenem resistance, its acquisition, and susceptibility in K. pneumoniae, with the potential to inform the development of future diagnostics or therapeutic strategies.

Multi-modal Micro Electrode Fluidic Array (MEFA) Shells for Brain Organoids

Abstract Brain organoids (BOs) derived from human stem cells bridge the gap between monolayer cell culture studies and animal models, which have well-documented limitations. Monolayer cell culture models fail to accurately replicate the 3D interconnectivity in the brain; animal models, while helpful, are limited due to interspecies differences, with most research focusing on rather phenotypical rather than mechanistic aspects. Concurrent with the advancement of BO models is the urgent need to develop 3D micro instrumentation supporting these organoids to investigate brain development and disease in their accurate physiological environment. Conventional microelectrode arrays (MEAs) used for neuronal cell culture studies are planar, which limits recording access to a small fraction of cells on the bottom side of the organoid. Also, conventional microfluidics is inherently planar, and while recent advances in 3D MEAs and 3D microfluidics have enabled electrical and chemical interrogation in 3D, combining both features with tunability and precision to allow independent and simultaneous control is challenging. Recently, we reported new 3D micro instrumentation in the form of 3D shell MEAs and demonstrated its applicability for electrical recording from BOs. They feature lithographically patterned and chip-integrated electrodes and self-folding polymer shells that can be triggered to wrap around BOs to measure electrical activity from the entire organoid surface. The 3D MEA shell system is modeled on and resembles a miniaturized electroencephalography (EEG) cap; the process used to make them is size-scalable, chip-integrated, and mass- producible. In the research, we aim to develop and validate 3D Micro Electrode Fluidic Array (MEFA) shells with multi-modal electrical recording and biochemical control capabilities, offering high spatiotemporal resolution, tunability, and scalability. Since 3D spatiotemporal patterns of neurochemicals play a critical role in molecular and cellular events of neural development and disease, we propose to apply and validate the MEFA shells in two studies that mimic neurodevelopment and monitor the spatiotemporal effects in neurological disorders and their treatments in vitro. We anticipate that the proposed 3D MEFAs would revolutionize brain sciences by permitting real-time, in-situ studies of electrical and chemical stimulation and interrogation of BOs in a high- throughput manner. The proposed 3D scalable, reproducible, and tunable 3D micro instrumentation for BOs has broad relevance to understanding brain development in utero and the development of anatomically accurate drug and toxicity screening platforms for brain sciences and neurological disorders.

Adventures in Spin Labeling: Clinical Perfusion Imaging and the Path to Technical Innovation

Arterial spin labeling (ASL) MRI has become a vital tool in clinical neuroimaging, enabling noninvasive assessment of cerebral perfusion across a range of conditions including stroke, vascular malformations, and brain tumors. With broader clinical adoption, its practical strengths — as well as important limitations — have become increasingly clear.

Pharmacological exploitation of neurotrophins and their receptors to develop novel therapeutic approaches against neurodegenerative diseases and brain trauma

Neurotrophins (NGF, BDNF, NT-3) are endogenous growth factors that exert neuroprotective effects by preventing neuronal death and promoting neurogenesis. They act by binding to their respective high-affinity, pro-survival receptors TrkA, TrkB or TrkC, as well as to p75NTR death receptor. While these molecules have been shown to significantly slow or prevent neurodegeneration, their reduced bioavailability and inability to penetrate the blood-brain-barrier limit their use as potential therapeutics. To bypass these limitations, our research team has developed and patented small-sized, lipophilic compounds which selectively resemble neurotrophins’ effects, presenting preferable pharmacological properties and promoting neuroprotection and repair against neurodegeneration. In addition, the combination of these molecules with 3D cultured human neuronal cells, and their targeted delivery in the brain ventricles through soft robotic systems, could offer novel therapeutic approaches against neurodegenerative diseases and brain trauma.

Localisation of Seizure Onset Zone in Epilepsy Using Time Series Analysis of Intracranial Data

There are over 30 million people with drug-resistant epilepsy worldwide. When neuroimaging and non-invasive neural recordings fail to localise seizure onset zones (SOZ), intracranial recordings become the best chance for localisation and seizure-freedom in those patients. However, intracranial neural activities remain hard to visually discriminate across recording channels, which limits the success of intracranial visual investigations. In this presentation, I present methods which quantify intracranial neural time series and combine them with explainable machine learning algorithms to localise the SOZ in the epileptic brain. I present the potentials and limitations of our methods in the localisation of SOZ in epilepsy providing insights for future research in this area.

Generative models for video games (rescheduled)

Developing agents capable of modeling complex environments and human behaviors within them is a key goal of artificial intelligence research. Progress towards this goal has exciting potential for applications in video games, from new tools that empower game developers to realize new creative visions, to enabling new kinds of immersive player experiences. This talk focuses on recent advances of my team at Microsoft Research towards scalable machine learning architectures that effectively capture human gameplay data. In the first part of my talk, I will focus on diffusion models as generative models of human behavior. Previously shown to have impressive image generation capabilities, I present insights that unlock applications to imitation learning for sequential decision making. In the second part of my talk, I discuss a recent project taking ideas from language modeling to build a generative sequence model of an Xbox game.

Generative models for video games

Developing agents capable of modeling complex environments and human behaviors within them is a key goal of artificial intelligence research. Progress towards this goal has exciting potential for applications in video games, from new tools that empower game developers to realize new creative visions, to enabling new kinds of immersive player experiences. This talk focuses on recent advances of my team at Microsoft Research towards scalable machine learning architectures that effectively capture human gameplay data. In the first part of my talk, I will focus on diffusion models as generative models of human behavior. Previously shown to have impressive image generation capabilities, I present insights that unlock applications to imitation learning for sequential decision making. In the second part of my talk, I discuss a recent project taking ideas from language modeling to build a generative sequence model of an Xbox game.

Attending to the ups and downs of Lewy body dementia: An exploration of cognitive fluctuations

Dementia with Lewy bodies (DLB) and Parkinson's disease dementia (PDD) share similarities in pathology and clinical presentation and come under the umbrella term of Lewy body dementias (LBD). Fluctuating cognition is a key symptom in LBD and manifests as altered levels of alertness and attention, with a marked difference between best and worst performance. Cognition and alertness can change over seconds or minutes to hours and days of obtundation. Cognitive fluctuations can have significant impacts on the quality of life of people with LBD as well as potentially contribute to the exacerbation of other transient symptoms including, for example, hallucinations and psychosis as well as making it difficult to measure cognitive effect size benefits in clinical trials of LBD. However, this significant symptom in LBD is poorly understood. In my presentation I will discuss the phenomenology of cognitive fluctuations, how we can measure it clinically and limitations of these approaches. I will then outline the work of our group and others which has been focussed on unpicking the aetiological basis of cognitive fluctuations in LBD using a variety of imaging approaches (e.g. SPECT, sMRI, fMRI and EEG). I will then briefly explore future research directions.

Spatially-embedded recurrent neural networks reveal widespread links between structural and functional neuroscience findings

Brain networks exist within the confines of resource limitations. As a result, a brain network must overcome metabolic costs of growing and sustaining the network within its physical space, while simultaneously implementing its required information processing. To observe the effect of these processes, we introduce the spatially-embedded recurrent neural network (seRNN). seRNNs learn basic task-related inferences while existing within a 3D Euclidean space, where the communication of constituent neurons is constrained by a sparse connectome. We find that seRNNs, similar to primate cerebral cortices, naturally converge on solving inferences using modular small-world networks, in which functionally similar units spatially configure themselves to utilize an energetically-efficient mixed-selective code. As all these features emerge in unison, seRNNs reveal how many common structural and functional brain motifs are strongly intertwined and can be attributed to basic biological optimization processes. seRNNs can serve as model systems to bridge between structural and functional research communities to move neuroscientific understanding forward.

Meta-learning functional plasticity rules in neural networks

Synaptic plasticity is known to be a key player in the brain’s life-long learning abilities. However, due to experimental limitations, the nature of the local changes at individual synapses and their link with emerging network-level computations remain unclear. I will present a numerical, meta-learning approach to deduce plasticity rules from either neuronal activity data and/or prior knowledge about the network's computation. I will first show how to recover known rules, given a human-designed loss function in rate networks, or directly from data, using an adversarial approach. Then I will present how to scale-up this approach to recurrent spiking networks using simulation-based inference.

Lifelong Learning AI via neuro inspired solutions

AI embedded in real systems, such as in satellites, robots and other autonomous devices, must make fast, safe decisions even when the environment changes, or under limitations on the available power; to do so, such systems must be adaptive in real time. To date, edge computing has no real adaptivity – rather the AI must be trained in advance, typically on a large dataset with much computational power needed; once fielded, the AI is frozen: It is unable to use its experience to operate if environment proves outside its training or to improve its expertise; and worse, since datasets cannot cover all possible real-world situations, systems with such frozen intelligent control are likely to fail. Lifelong Learning is the cutting edge of artificial intelligence - encompassing computational methods that allow systems to learn in runtime and incorporate learning for application in new, unanticipated situations. Until recently, this sort of computation has been found exclusively in nature; thus, Lifelong Learning looks to nature, and in particular neuroscience, for its underlying principles and mechanisms and then translates them to this new technology. Our presentation will introduce a number of state-of-the-art approaches to achieve AI adaptive learning, including from the DARPA’s L2M program and subsequent developments. Many environments are affected by temporal changes, such as the time of day, week, season, etc. A way to create adaptive systems which are both small and robust is by making them aware of time and able to comprehend temporal patterns in the environment. We will describe our current research in temporal AI, while also considering power constraints.

Learning Relational Rules from Rewards

Humans perceive the world in terms of objects and relations between them. In fact, for any given pair of objects, there is a myriad of relations that apply to them. How does the cognitive system learn which relations are useful to characterize the task at hand? And how can it use these representations to build a relational policy to interact effectively with the environment? In this paper we propose that this problem can be understood through the lens of a sub-field of symbolic machine learning called relational reinforcement learning (RRL). To demonstrate the potential of our approach, we build a simple model of relational policy learning based on a function approximator developed in RRL. We trained and tested our model in three Atari games that required to consider an increasingly number of potential relations: Breakout, Pong and Demon Attack. In each game, our model was able to select adequate relational representations and build a relational policy incrementally. We discuss the relationship between our model with models of relational and analogical reasoning, as well as its limitations and future directions of research.

Multi-level theory of neural representations in the era of large-scale neural recordings: Task-efficiency, representation geometry, and single neuron properties

A central goal in neuroscience is to understand how orchestrated computations in the brain arise from the properties of single neurons and networks of such neurons. Answering this question requires theoretical advances that shine light into the ‘black box’ of representations in neural circuits. In this talk, we will demonstrate theoretical approaches that help describe how cognitive and behavioral task implementations emerge from the structure in neural populations and from biologically plausible neural networks. First, we will introduce an analytic theory that connects geometric structures that arise from neural responses (i.e., neural manifolds) to the neural population’s efficiency in implementing a task. In particular, this theory describes a perceptron’s capacity for linearly classifying object categories based on the underlying neural manifolds’ structural properties. Next, we will describe how such methods can, in fact, open the ‘black box’ of distributed neuronal circuits in a range of experimental neural datasets. In particular, our method overcomes the limitations of traditional dimensionality reduction techniques, as it operates directly on the high-dimensional representations, rather than relying on low-dimensionality assumptions for visualization. Furthermore, this method allows for simultaneous multi-level analysis, by measuring geometric properties in neural population data, and estimating the amount of task information embedded in the same population. These geometric frameworks are general and can be used across different brain areas and task modalities, as demonstrated in the work of ours and others, ranging from the visual cortex to parietal cortex to hippocampus, and from calcium imaging to electrophysiology to fMRI datasets. Finally, we will discuss our recent efforts to fully extend this multi-level description of neural populations, by (1) investigating how single neuron properties shape the representation geometry in early sensory areas, and by (2) understanding how task-efficient neural manifolds emerge in biologically-constrained neural networks. By extending our mathematical toolkit for analyzing representations underlying complex neuronal networks, we hope to contribute to the long-term challenge of understanding the neuronal basis of tasks and behaviors.

A Framework for a Conscious AI: Viewing Consciousness through a Theoretical Computer Science Lens

We examine consciousness from the perspective of theoretical computer science (TCS), a branch of mathematics concerned with understanding the underlying principles of computation and complexity, including the implications and surprising consequences of resource limitations. We propose a formal TCS model, the Conscious Turing Machine (CTM). The CTM is influenced by Alan Turing's simple yet powerful model of computation, the Turing machine (TM), and by the global workspace theory (GWT) of consciousness originated by cognitive neuroscientist Bernard Baars and further developed by him, Stanislas Dehaene, Jean-Pierre Changeux, George Mashour, and others. However, the CTM is not a standard Turing Machine. It’s not the input-output map that gives the CTM its feeling of consciousness, but what’s under the hood. Nor is the CTM a standard GW model. In addition to its architecture, what gives the CTM its feeling of consciousness is its predictive dynamics (cycles of prediction, feedback and learning), its internal multi-modal language Brainish, and certain special Long Term Memory (LTM) processors, including its Inner Speech and Model of the World processors. Phenomena generally associated with consciousness, such as blindsight, inattentional blindness, change blindness, dream creation, and free will, are considered. Explanations derived from the model draw confirmation from consistencies at a high level, well above the level of neurons, with the cognitive neuroscience literature. Reference. L. Blum and M. Blum, "A theory of consciousness from a theoretical computer science perspective: Insights from the Conscious Turing Machine," PNAS, vol. 119, no. 21, 24 May 2022. https://www.pnas.org/doi/epdf/10.1073/pnas.2115934119

PET imaging in brain diseases

Talk 1. PET based biomarkers of treatment efficacy in temporal lobe epilepsy A critical aspect of drug development involves identifying robust biomarkers of treatment response for use as surrogate endpoints in clinical trials. However, these biomarkers also have the capacity to inform mechanisms of disease pathogenesis and therapeutic efficacy. In this webinar, Dr Bianca Jupp will report on a series of studies using the GABAA PET ligand, [18F]-Flumazenil, to establish biomarkers of treatment response to a novel therapeutic for temporal lobe epilepsy, identifying affinity at this receptor as a key predictor of treatment outcome. Dr Bianca Jupp is a Research Fellow in the Department of Neuroscience, Monash University and Lead PET/CT Scientist at the Alfred Research Alliance–Monash Biomedical Imaging facility. Her research focuses on neuroimaging and its capacity to inform the neurobiology underlying neurological and neuropsychiatric disorders. Talk 2. The development of a PET radiotracer for reparative microglia Imaging of neuroinflammation is currently hindered by the technical limitations associated with TSPO imaging. In this webinar, Dr Lucy Vivash will discuss the development of PET radiotracers that specifically image reparative microglia through targeting the receptor kinase MerTK. This includes medicinal chemistry design and testing, radiochemistry, and in vitro and in vivo testing of lead tracers. Dr Lucy Vivash is a Research Fellow in the Department of Neuroscience, Monash University. Her research focuses on the preclinical development and clinical translation of novel PET radiotracers for the imaging of neurodegenerative diseases.

Learning from others, helping others learn: Cognitive foundations of distinctively human social learning

Learning does not occur in isolation. From parent-child interactions to formal classroom environments, humans explore, learn, and communicate in rich, diverse social contexts. Rather than simply observing and copying their conspecifics, humans engage in a range of epistemic practices that actively recruit those around them. What makes human social learning so distinctive, powerful, and smart? In this talk, I will present a series of studies that reveal the remarkably sophisticated inferential abilities that young children show not only in how they learn from others but also in how they help others learn. Children interact with others as learners and as teachers to learn and communicate about the world, about others, and even about the self. The results collectively paint a picture of human social learning that is far more than copying and imitation: It is active, bidirectional, and cooperative. I will end by discussing ongoing work that extends this picture beyond what we typically call “social learning”, with implications for building better machines that learn from and interact with humans.

Neural circuits of visuospatial working memory

One elementary brain function that underlies many of our cognitive behaviors is the ability to maintain parametric information briefly in mind, in the time scale of seconds, to span delays between sensory information and actions. This component of working memory is fragile and quickly degrades with delay length. Under the assumption that behavioral delay-dependencies mark core functions of the working memory system, our goal is to find a neural circuit model that represents their neural mechanisms and apply it to research on working memory deficits in neuropsychiatric disorders. We have constrained computational models of spatial working memory with delay-dependent behavioral effects and with neural recordings in the prefrontal cortex during visuospatial working memory. I will show that a simple bump attractor model with weak inhomogeneities and short-term plasticity mechanisms can link neural data with fine-grained behavioral output in a trial-by-trial basis and account for the main delay-dependent limitations of working memory: precision, cardinal repulsion biases and serial dependence. I will finally present data from participants with neuropsychiatric disorders that suggest that serial dependence in working memory is specifically altered, and I will use the model to infer the possible neural mechanisms affected.

Probabilistic computation in natural vision

A central goal of vision science is to understand the principles underlying the perception and neural coding of the complex visual environment of our everyday experience. In the visual cortex, foundational work with artificial stimuli, and more recent work combining natural images and deep convolutional neural networks, have revealed much about the tuning of cortical neurons to specific image features. However, a major limitation of this existing work is its focus on single-neuron response strength to isolated images. First, during natural vision, the inputs to cortical neurons are not isolated but rather embedded in a rich spatial and temporal context. Second, the full structure of population activity—including the substantial trial-to-trial variability that is shared among neurons—determines encoded information and, ultimately, perception. In the first part of this talk, I will argue for a normative approach to study encoding of natural images in primary visual cortex (V1), which combines a detailed understanding of the sensory inputs with a theory of how those inputs should be represented. Specifically, we hypothesize that V1 response structure serves to approximate a probabilistic representation optimized to the statistics of natural visual inputs, and that contextual modulation is an integral aspect of achieving this goal. I will present a concrete computational framework that instantiates this hypothesis, and data recorded using multielectrode arrays in macaque V1 to test its predictions. In the second part, I will discuss how we are leveraging this framework to develop deep probabilistic algorithms for natural image and video segmentation.

What is Cognitive Neuropsychology Good For? An Unauthorized Biography

Abstract: There is no doubt that the study of brain damaged individuals has contributed greatly to our understanding of the mind/brain. Within this broad approach, cognitive neuropsychology accentuates the cognitive dimension: it investigates the structure and organization of perceptual, motor, cognitive, and language systems – prerequisites for understanding the functional organization of the brain – through the analysis of their dysfunction following brain damage. Significant insights have come specifically from this paradigm. But progress has been slow and enthusiasm for this approach has waned somewhat in recent years, and the use of existing findings to constrain new theories has also waned. What explains the current diminished status of cognitive neuropsychology? One reason may be failure to calibrate expectations about the effective contribution of different subfields of the study of the mind/brain as these are determined by their natural peculiarities – such factors as the types of available observations and their complexity, opportunity of access to such observations, the possibility of controlled experimentation, and the like. Here, I also explore the merits and limitations of cognitive neuropsychology, with particular focus on the role of intellectual, pragmatic, and societal factors that determine scientific practice within the broader domains of cognitive science/neuroscience. I conclude on an optimistic note about the continuing unique importance of cognitive neuropsychology: although limited to the study of experiments of nature, it offers a privileged window into significant aspects of the mind/brain that are not easily accessible through other approaches. Biography: Alfonso Caramazza's research has focussed extensively on how words and their meanings are represented in the brain. His early pioneering studies helped to reformulate our thinking about Broca's aphasia (not limited to production) and formalised the logic of patient-based neuropsychology. More recently he has been instrumental in reconsidering popular claims about embodied cognition.

From natural scene statistics to multisensory integration: experiments, models and applications

To efficiently process sensory information, the brain relies on statistical regularities in the input. While generally improving the reliability of sensory estimates, this strategy also induces perceptual illusions that help reveal the underlying computational principles. Focusing on auditory and visual perception, in my talk I will describe how the brain exploits statistical regularities within and across the senses for the perception space, time and multisensory integration. In particular, I will show how results from a series of psychophysical experiments can be interpreted in the light of Bayesian Decision Theory, and I will demonstrate how such canonical computations can be implemented into simple and biologically plausible neural circuits. Finally, I will show how such principles of sensory information processing can be leveraged in virtual and augmented reality to overcome display limitations and expand human perception.

“Mind reading” with brain scanners: Facts versus science fiction

Every thought is associated with a unique pattern of brain activity. Thus, in principle, it should be possible to use these activity patterns as "brain fingerprints" for different thoughts and to read out what a person is thinking based on their brain activity alone. Indeed, using machine learning considerable progress has been made in such "brainreading" in recent years. It is now possible to decode which image a person is viewing, which film sequence they are watching, which emotional state they are in or which intentions they hold in mind. This talk will provide an overview of the current state of the art in brain reading. It will also highlight the main challenges and limitations of this research field. For example, mathematical models are needed to cope with the high dimensionality of potential mental states. Furthermore, the ethical concerns raised by (often premature) commercial applications of brain reading will also be discussed.

The bounded rationality of probability distortion

In decision-making under risk (DMR) participants' choices are based on probability values systematically different from those that are objectively correct. Similar systematic distortions are found in tasks involving relative frequency judgments (JRF). These distortions limit performance in a wide variety of tasks and an evident question is, why do we systematically fail in our use of probability and relative frequency information? We propose a Bounded Log-Odds Model (BLO) of probability and relative frequency distortion based on three assumptions: (1) log-odds: probability and relative frequency are mapped to an internal log-odds scale, (2) boundedness: the range of representations of probability and relative frequency are bounded and the bounds change dynamically with task, and (3) variance compensation: the mapping compensates in part for uncertainty in probability and relative frequency values. We compared human performance in both DMR and JRF tasks to the predictions of the BLO model as well as eleven alternative models each missing one or more of the underlying BLO assumptions (factorial model comparison). The BLO model and its assumptions proved to be superior to any of the alternatives. In a separate analysis, we found that BLO accounts for individual participants’ data better than any previous model in the DMR literature. We also found that, subject to the boundedness limitation, participants’ choice of distortion approximately maximized the mutual information between objective task-relevant values and internal values, a form of bounded rationality.

Qualitative Structure, Automorphism Groups and Private Language

It is generally agreed upon that qualities of conscious experience instantiate structural properties, usually called relations. They furnish a representation of qualities (or qualia, in fact) in terms of a mathematical space Q (rather than a set), which is crucial to both modelling and measuring of conscious experience." "What is usually disregarded is that “only such structural properties generalize across individuals” (Austen Clark), but that qualities themselves as differentiated by stimulus specifications, behavior or reports do not. We show that this implies that only the part of Q which is invariant with respect to the automorphism group has a well-defined referent, while individual elements do not. This poses a prima facie limitation of any theory or experiment that aims to address individual qualities. We show how mathematical theories of consciousness can overcome this limitation via symmetry groups and group actions, making accessible to science what is properly called private language.

Norse: A library for gradient-based learning in Spiking Neural Networks

We introduce Norse: An open-source library for gradient-based training of spiking neural networks. In contrast to neuron simulators which mainly target computational neuroscientists, our library seamlessly integrates with the existing PyTorch ecosystem using abstractions familiar to the machine learning community. This has immediate benefits in that it provides a familiar interface, hardware accelerator support and, most importantly, the ability to use gradient-based optimization. While many parallel efforts in this direction exist, Norse emphasizes flexibility and usability in three ways. Users can conveniently specify feed-forward (convolutional) architectures, as well as arbitrarily connected recurrent networks. We strictly adhere to a functional and class-based API such that neuron primitives and, for example, plasticity rules composes. Finally, the functional core API ensures compatibility with the PyTorch JIT and ONNX infrastructure. We have made progress to support network execution on the SpiNNaker platform and plan to support other neuromorphic architectures in the future. While the library is useful in its present state, it also has limitations we will address in ongoing work. In particular, we aim to implement event-based gradient computation, using the EventProp algorithm, which will allow us to support sparse event-based data efficiently, as well as work towards support of more complex neuron models. With this library, we hope to contribute to a joint future of computational neuroscience and neuromorphic computing.

Deriving local synaptic learning rules for efficient representations in networks of spiking neurons

How can neural networks learn to efficiently represent complex and high-dimensional inputs via local plasticity mechanisms? Classical models of representation learning assume that input weights are learned via pairwise Hebbian-like plasticity. Here, we show that pairwise Hebbian-like plasticity only works under specific requirements on neural dynamics and input statistics. To overcome these limitations, we derive from first principles a learning scheme based on voltage-dependent synaptic plasticity rules. Here, inhibition learns to locally balance excitatory input in individual dendritic compartments, and thereby can modulate excitatory synaptic plasticity to learn efficient representations. We demonstrate in simulations that this learning scheme works robustly even for complex, high-dimensional and correlated inputs. It also works in the presence of inhibitory transmission delays, where Hebbian-like plasticity typically fails. Our results draw a direct connection between dendritic excitatory-inhibitory balance and voltage-dependent synaptic plasticity as observed in vivo, and suggest that both are crucial for representation learning.

In vitro bioelectronic models of the gut-brain axis

The human gut microbiome has emerged as a key player in the bidirectional communication of the gut-brain axis, affecting various aspects of homeostasis and pathophysiology. Until recently, the majority of studies that seek to explore the mechanisms underlying the microbiome-gut-brain axis cross-talk relied almost exclusively on animal models, and particularly gnotobiotic mice. Despite the great progress made with these models, various limitations, including ethical considerations and interspecies differences that limit the translatability of data to human systems, pushed researchers to seek for alternatives. Over the past decades, the field of in vitro modelling of tissues has experienced tremendous growth, thanks to advances in 3D cell biology, materials, science and bioengineering, pushing further the borders of our ability to more faithfully emulate the in vivo situation. Organ-on-chip technology and bioengineered tissues have emerged as highly promising alternatives to animal models for a wide range of applications. In this talk I’ll discuss our progress towards generating a complete platform of the human microbiota-gut-brain axis with integrated monitoring and sensing capabilities. Bringing together principles of materials science, tissue engineering, 3D cell biology and bioelectronics, we are building advanced models of the GI and the BBB /NVU, with real-time and label-free monitoring units adapted in the model architecture, towards a robust and more physiologically relevant human in vitro model, aiming to i) elucidate the role of microbiota in the gut-brain axis communication, ii) to study how diet and impaired microbiota profiles affect various (patho-)physiologies, and iii) to test personalised medicine approaches for disease modelling and drug testing.

Understanding Perceptual Priors with Massive Online Experiments

One of the most important questions in psychology and neuroscience is understanding how the outside world maps to internal representations. Classical psychophysics approaches to this problem have a number of limitations: they mostly study low dimensional perpetual spaces, and are constrained in the number and diversity of participants and experiments. As ecologically valid perception is rich, high dimensional, contextual, and culturally dependent, these impediments severely bias our understanding of perceptual representations. Recent technological advances—the emergence of so-called “Virtual Labs”— can significantly contribute toward overcoming these barriers. Here I present a number of specific strategies that my group has developed in order to probe representations across a number of dimensions. 1) Massive online experiments can increase significantly the amount of participants and experiments that can be carried out in a single study, while also significantly diversifying the participant pool. We have developed a platform, PsyNet, that enables “experiments as code,” whereby the orchestration of computer servers, recruiting, compensation of participants, and data management is fully automated and every experiment can be fully replicated with one command line. I will demonstrate how PsyNet allows us to recruit thousands of participants for each study with a large number of control experimental conditions, significantly increasing our understanding of auditory perception. 2) Virtual lab methods also enable us to run experiments that are nearly impossible in a traditional lab setting. I will demonstrate our development of adaptive sampling, a set of behavioural methods that combine machine learning sampling techniques (Monte Carlo Markov Chains) with human interactions and allow us to create high-dimensional maps of perceptual representations with unprecedented resolution. 3) Finally, I will demonstrate how the aforementioned methods can be applied to the study of perceptual priors in both audition and vision, with a focus on our work in cross-cultural research, which studies how perceptual priors are influenced by experience and culture in diverse samples of participants from around the world.

Multiphoton imaging with next-generation indicators

Two-photon (2P) in vivo functional imaging of genetically encoded fluorescent Ca2+indicators (GECIs) for neuronal activity has become a broadly applied standard tool in modern neuroscience, because it allows simultaneous imaging of the activity of many neurons at high spatial resolution within living animals. Unfortunately, the most commonly used light-sources – tunable femtosecond pulsed ti:sapphire lasers – can be prohibitively expensive for many labs and fall short of delivering sufficient powers for some new ultra-fast 2P microscopy modalities. Inexpensive homebuilt or industrial light sources such as Ytterbium fiber lasers (YbFLs) show great promise to overcome these limitations as they are becoming widely available at costs orders of magnitude lower and power outputs of up to many times higher than conventional ti:sapphire lasers. However, these lasers are typically bound to emitting a single wavelength (i.e., not tunable) centered around 1020-1060 nm, which fails to efficiently excite state of the art green GECIs such as jGCaMP7 or 8. To this end, we designed and characterized spectral variants (yellow CaMP = YCaMP) of the ultrasensitive genetically encoded calcium indicator jGCaMP7, that allows for efficient 2P-excitation at wavelengths above 1010nm. In this talk I will give a brief overview over some of the reasons why using a fiber laser for 2P excitation might be right for you. I will talk about the development of jYCaMP and some exciting new experimental avenues that it has opened while touching on the prospect that shifting biosensors yellow could have for the 2P imaging community. Please join me for an interesting and fun discussion on whether “yellow is the new green” after the talk!

Neural codes in early sensory areas maximize fitness

It has generally been presumed that sensory information encoded by a nervous system should be as accurate as its biological limitations allow. However, perhaps counter intuitively, accurate representations of sensory signals do not necessarily maximize the organism’s chances of survival. We show that neural codes that maximize reward expectation—and not accurate sensory representations—account for retinal responses in insects, and retinotopically-specific adaptive codes in humans. Thus, our results provide evidence that fitness-maximizing rules imposed by the environment are applied at the earliest stages of sensory processing.

Application of Airy beam light sheet microscopy to examine early neurodevelopmental structures in 3D hiPSC-derived human cortical spheroids

The inability to observe relevant biological processes in vivo significantly restricts human neurodevelopmental research. Advances in appropriate in vitro model systems, including patient-specific human brain organoids and human cortical spheroids (hCSs), offer a pragmatic solution to this issue. In particular, hCSs are an accessible method for generating homogenous organoids of dorsal telencephalic fate, which recapitulate key aspects of human corticogenesis, including the formation of neural rosettes—in vitro correlates of the neural tube. These neurogenic niches give rise to neural progenitors that subsequently differentiate into neurons. Studies differentiating induced pluripotent stem cells (hiPSCs) in 2D have linked atypical formation of neural rosettes with neurodevelopmental disorders such as autism spectrum conditions. Thus far, however, conventional methods of tissue preparation in this field limit the ability to image these structures in three-dimensions within intact hCS or other 3D preparations. To overcome this limitation, we have sought to optimise a methodological approach to process hCSs to maximise the utility of a novel Airy-beam light sheet microscope (ALSM) to acquire high resolution volumetric images of internal structures within hCS representative of early developmental time points.

Understanding and treating epilepsy in tuberous sclerosis complex

Tuberous sclerosis complex (TSC) and focal cortical dysplasia type II (FCDII) are caused by mutations in mTOR pathway genes leading to mTOR hyperactivity, focal malformations of cortical development (fMCD), and seizures in 80-90% of the patients. The current definitive treatments for epilepsy are surgical resection or treatment with everolimus, which inhibits mTOR activity (only approved for TSC). Because both options have severe limitations, there is a major need to better understand the mechanisms leading to seizures to improve life-long epilepsy treatment in TSC and FCDII. To investigate such mechanisms, we recently developed a murine model of fMCD-associated epilepsy that recapitulates the human TSC and FCDII disorders. fMCD are defined by the presence of misplaced, dysmorphic cortical neurons expressing hyperactive mTOR – for simplicity we will refer to these as “mutant” neurons. In our model and in human TSC tissue, we made a surprising finding that mutant neurons express HCN4 channels, which are not normally functionally expressed in cortical neurons, and increased levels of filamin A (FLNA). FLNA is an actin-crossing linking molecule that has also multiple binding partners inside cells. These data led us to ask several important questions: (1) As HCN4 channels are responsible for the pacemaking activity of the heart, can HCN4 channel expression lead to repetitive firing of mutant neurons resulting in seizures? (2) HCN4 is the most cAMP-sensitive of the four HCN isoforms. Does increase in cAMP lead to the firing of mutant neurons? (3) Does increase in FLNA contribute to neuronal alterations and seizures? (4) Is the abnormal HCN4 and FLNA expression in mutant neurons due to mTOR? These questions will be discussed and addressed in the lecture.

A developmental-cognitive perspective on the impact of adolescent social media use

Concerns about the impact of social media use on adolescent well-being and mental health are common. While the amount of research in this area has increased rapidly over the last 5 years, most outputs are still marred by a multitude of limitations. These shortcomings have left our understanding of social media effects severely limited, holding back both scientific discovery and policy interventions. This talk discusses how developmental, cognitive and neuroscientific approaches might provide a new and improved way of studying social media effects. It will detail new studies in support of this idea, and raise potential avenues for collaborative work across the Cambridge Neuroscience community. As the digital world now (re)shapes what it means for us to live, communicate and develop, only an interdisciplinary approach will allow us to truly understand its impacts.

Exploring fine detail: The interplay of attention, oculomotor behavior and visual perception in the fovea

Outside the foveola, visual acuity and other visual functions gradually deteriorate with increasing eccentricity. Humans compensate for these limitations by relying on a tight link between perception and action; rapid gaze shifts (saccades) occur 2-3 times every second, separating brief “fixation” intervals in which visual information is acquired and processed. During fixation, however, the eye is not immobile. Small eye movements incessantly shift the image on the retina even when the attended stimulus is already foveated, suggesting a much deeper coupling between visual functions and oculomotor activity. Thanks to a combination of techniques allowing for high-resolution recordings of eye position, retinal stabilization, and accurate gaze localization, we examined how attention and eye movements are controlled at this scale. We have shown that during fixation, visual exploration of fine spatial detail unfolds following visuomotor strategies similar to those occurring at a larger scale. This behavior compensates for non-homogenous visual capabilities within the foveola and is finely controlled by attention, which facilitates processing at selected foveal locations. Ultimately, the limits of high acuity vision are greatly influenced by the spatiotemporal modulations introduced by fixational eye movements. These findings reveal that, contrary to common intuition, placing a stimulus within the foveola is necessary but not sufficient for high visual acuity; fine spatial vision is the outcome of an orchestrated synergy of motor, cognitive, and attentional factors.

Microneurography And Microstimulation Of Single Tactile Afferents In The Human Hand

Microneurography is a method, invented by Ake Vallbo and Karl-Erik Hagbarth in the late 1960, with which we can record the activity from single, identified nerve fibres in awake human participants. In this talk, I will then discuss the method, its advantages and limitations, and some of the key discoveries regarding coding of tactile events in the signalling from receptors in the human skin. An extension of the method is to stimulate single afferents, and record the resulting tactile sensations reported by the participants, so-called microstimulation. The first experiments were done in the 1980s, but the method has recently seen a revival, and is currently being combined with high-resolution brain imaging in the study of the relationship between tactile nerve signals, sensations, and processing of tactile information in the brain.

E-prop: A biologically inspired paradigm for learning in recurrent networks of spiking neurons

Transformative advances in deep learning, such as deep reinforcement learning, usually rely on gradient-based learning methods such as backpropagation through time (BPTT) as a core learning algorithm. However, BPTT is not argued to be biologically plausible, since it requires to a propagate gradients backwards in time and across neurons. Here, we propose e-prop, a novel gradient-based learning method with local and online weight update rules for recurrent neural networks, and in particular recurrent spiking neural networks (RSNNs). As a result, e-prop has the potential to provide a substantial fraction of the power of deep learning to RSNNs. In this presentation, we will motivate e-prop from the perspective of recent insights in neuroscience and show how these have to be combined to form an algorithm for online gradient descent. The mathematical results will be supported by empirical evidence in supervised and reinforcement learning tasks. We will also discuss how limitations that are inherited from gradient-based learning methods, such as sample-efficiency, can be addressed by considering an evolution-like optimization that enhances learning on particular task families. The emerging learning architecture can be used to learn tasks by a single demonstration, hence enabling one-shot learning.

The consequences and constraints of functional organization on behavior

In many ways, cognitive neuroscience is the attempt to use physiological observation to clarify the mechanisms that shape behavior. Over the past 25 years, fMRI has provided a system-wide and yet somewhat spatially precise view of the response in human cortex evoked by a wide variety of stimuli and task contexts. The current talk focuses on the other direction of inference; the implications of this observed functional organization for behavior. To begin, we must interrogate the methodological and empirical frameworks underlying our derivation of this organization, partially by exploring its relationship to and predictability from gross neuroanatomy. Next, across a series of studies, the implications of two properties of functional organization for behavior will be explored: 1) the co-localization of visual working memory and perceptual processing and 2) implicit learning in the context of distributed responses. In sum, these results highlight the limitations of our current approach and hint at a new general mechanism for explaining observed behavior in context with the neural substrate.

A New Approach to the Hard Problem of Consciousness

David Chalmers’s (1995) hard problem famously states: “It is widely agreed that experience arises from a physical basis, but we have no good explanation of why and how it so arises.” Thomas Nagel (1974) wrote something similar: “If we acknowledge that a physical theory of mind must account for the subjective character of experience, we must admit that no presently available conception gives us a clue about how this could be done.” This presentation will point the way towards the long-sought “good explanation” -- or at least it will provide “a clue”. I will make three points: (1) It is unfortunate that cognitive science took vision as its model example when looking for a ‘neural correlate of consciousness’ because cortical vision (like most cognitive processes) is not intrinsically conscious. There is not necessarily ‘something it is like’ to see. (2) Affective feeling, by contrast, is conscious by definition. You cannot feel something without feeling it. Moreover, affective feeling, generated in the upper brainstem, is the foundational form of consciousness: prerequisite for all the higher cognitive forms. (3) The functional mechanism of feeling explains why and how it cannot go on ‘in the dark’, free of any inner feel. Affect enables the organism to monitor deviations from its expected self-states in uncertain situations and thereby frees homeostasis from the limitations of automatism. As Nagel says, “An organism has conscious mental states if and only if there is something that it is like to be that organism—something it is like for the organism.” Affect literally constitutes the sentient subject.

High precision coding in visual cortex

Single neurons in visual cortex provide unreliable measurements of visual features due to their high trial-to-trial variability. It is not known if this “noise” extends its effects over large neural populations to impair the global encoding of stimuli. We recorded simultaneously from ∼20,000 neurons in mouse primary visual cortex (V1) and found that the neural populations had discrimination thresholds of ∼0.34° in an orientation decoding task. These thresholds were nearly 100 times smaller than those reported behaviourally in mice. The discrepancy between neural and behavioural discrimination could not be explained by the types of stimuli we used, by behavioural states or by the sequential nature of perceptual learning tasks. Furthermore, higher-order visual areas lateral to V1 could be decoded equally well. These results imply that the limits of sensory perception in mice are not set by neural noise in sensory cortex, but by the limitations of downstream decoders.

Deep imitation learning for neuromechanical control: realistic walking in an embodied fly

COSYNE 2025

Testing the power and limitations of predictive connectomics in the fly visual system

COSYNE 2025

Use of pCASL MRI sequence to study brain perfusion and vascular permeability after Blood Brain Barrier opening: limitations and perspectives

Statistical limitations in the reconstruction of low-dimensional neural trajectories from high-dimensional activity recordings

Differences in neural activation patterns within the action observation network during imitation of point-light displays and fully visible manipulative actions

FENS Forum 2024

Brain-like visual surround suppression in generic CNNs: successes and limitations

Neuromatch 5