Building on cortical models
Over the past decade our community has made substantial progress in the construction of anatomically detailed network models of the cortical tissue. Thanks to advances in computer hardware and simulation technology, researchers can now routinely work with these models at the natural density of neurons and synapses. Moreover, the availability of cloud services means that such investigations can be carried out without having to install either the model or the simulation software. A recent workshop analyzed the impact of a specific model of the cortical microcircuit, published ten years ago . The model has been reused in multiple contexts: for reproduction studies, validation of mean-field approaches, exploration of methods of model sharing, and as a building block for larger models. Although the model was less successful in inspiring further neuroscientific studies than the authors of the original work had hoped, it became a de facto benchmark for neuromorphic computing systems. It sparked a constructive race for ever shorter simulation times and lower energy consumption. The quantitative comparison of different platforms reveals qualitative differences between conventional and neuromorphic hardware and limits of speed-up. The structure of the model is based on light microscopy because these were the data available at the time. Guided by simulation results and physiological evidence, the original publication hypothesized a preference of excitatory neurons for inhibitory targets. Modern electron microscopy data of cortical volumes combined with AI based reconstruction techniques is capable of resolving individual synaptic connections. This advances the concept of digital twins of the cortical network to a new level of precision, and has already enabled us to confirm the assumption of target type specifity underlying earlier models. Maybe with the progress sketched here, our community is at a transition point where it becomes easier to cooperatively and incrementally work on models with a larger explanatory scope. Presented in the van Vreeswijk Theoretical Neuroscience Seminar series (formerly WWTNS) on 2025-12-10. Recording duration: 00:39:15.
Shaping connections through remote gene regulation
In the third of this year’s Brain Prize webinars, Oscar Marin (King's College London, UK), Leslie Griffith (Brandeis University, USA), and Kesley Martin (Simons Foundation, USA) will present their work on shaping connections through remote gene regulation. Each speaker will present for 25 minutes, and the webinar will conclude with an open discussion. The webinar will be moderated by the winners of the 2023 Brain Prize, Michael Greenberg, Erin Schuman and Christine Holt.
Lateral entorhinal cortex directly influences medial entorhinal cortex through synaptic connections in layer 1
Standard models of episodic memory suggest that lateral (LEC) and medial entorhinal cortex (MEC) send independent inputs to the hippocampus, each carrying different types of information. Here, we describe a pathway by which information is integrated between LEC and MEC prior to reaching hippocampus. We demonstrate that LEC sends strong projections to MEC arising from neurons that receive neocortical inputs. Activation of LEC inputs drives excitation of hippocampal-projecting neurons in MEC layer 2, typically followed by inhibition that is accounted for by parallel activation of local inhibitory neurons. We therefore propose that local circuits in MEC may support integration of ‘what’ and ‘where’ information.
How are nervous systems remodeled in complex metazoans?
Early in development the nervous system is constructed with far too many neurons that make an excessive number of synaptic connections. Later, a wave of neuronal remodeling radically reshapes nervous system wiring and cell numbers through the selective elimination of excess synapses, axons and dendrites, and even whole neurons. This remodeling is widespread across the nervous system, extensive in terms of how much individual brain regions can change (e.g. in some cases 50% of neurons integrated into a brain circuit are eliminated), and thought to be essential for optimizing nervous system function. Perturbations of neuronal remodeling are thought to underlie devastating neurodevelopmental disorders including autism spectrum disorder, schizophrenia, and epilepsy. This seminar will discuss our efforts to use the relatively simple nervous system of Drosophila to understand the mechanistic basis by which cells, or parts of cells, are specified for removal and eliminated from the nervous system.
Hebbian Plasticity Supports Predictive Self-Supervised Learning of Disentangled Representations
Discriminating distinct objects and concepts from sensory stimuli is essential for survival. Our brains accomplish this feat by forming meaningful internal representations in deep sensory networks with plastic synaptic connections. Experience-dependent plasticity presumably exploits temporal contingencies between sensory inputs to build these internal representations. However, the precise mechanisms underlying plasticity remain elusive. We derive a local synaptic plasticity model inspired by self-supervised machine learning techniques that shares a deep conceptual connection to Bienenstock-Cooper-Munro (BCM) theory and is consistent with experimentally observed plasticity rules. We show that our plasticity model yields disentangled object representations in deep neural networks without the need for supervision and implausible negative examples. In response to altered visual experience, our model qualitatively captures neuronal selectivity changes observed in the monkey inferotemporal cortex in-vivo. Our work suggests a plausible learning rule to drive learning in sensory networks while making concrete testable predictions.
From a by-stander to an influencer: How microglia adapt to altered environments and influence neuronal activity
Microglia, traditionally classified as immune-responsive, adjust synaptic connections during development and disease. However, their role in the adult nervous system has been mostly diminished to an observer. In my research group, we are interested in how microglia are involved in establishing and maintaining accurate neuronal circuit function in the retina and in the visual cortex. In my talk, I will introduce our strategies how to decipher the microglia’s functional identity and how this information guided us to microglia enabled extracellular matrix remodeling and reinstatment of juvenile-like plasticity in the adult brain.
Neuro-Immune Coupling: How the Immune System Sculpts Brain Circuitry
In this lecture, Dr Stevens will discuss recent work that implicates brain immune cells, called microglia, in sculpting of synaptic connections during development and their relevance to autism, schizophrenia and other brain disorders. Her recent work revealed a key role for microglia and a group of immune related molecules called complement in normal developmental synaptic pruning, a normal process required to establish precise brain wiring. Emerging evidence suggests aberrant regulation of this pruning pathway may contribute to synaptic and cognitive dysfunction in a host of brain disorders, including schizophrenia. Recent research has revealed that a person’s risk of schizophrenia is increased if they inherit specific variants in complement C4, gene plays a well-known role in the immune system but also helps sculpt developing synapses in the mouse visual system (Sekar et al., 2016). Together these findings may help explain known features of schizophrenia, including reduced numbers of synapses in key cortical regions and an adolescent age of onset that corresponds with developmentally timed waves of synaptic pruning in these regions. Stevens will discuss this and ongoing work to understand the mechanisms by which complement and microglia prune specific synapses in the brain. A deeper understanding of how these immune mechanisms mediate synaptic pruning may provide novel insight into how to protect synapses in autism and other brain disorders, including Alzheimer’s and Huntington’s Disease.
Memory, learning to learn, and control of cognitive representations
Biological neural networks can represent information in the collective action potential discharge of neurons, and store that information amongst the synaptic connections between the neurons that both comprise the network and govern its function. The strength and organization of synaptic connections adjust during learning, but many cognitive neural systems are multifunctional, making it unclear how continuous activity alternates between the transient and discrete cognitive functions like encoding current information and recollecting past information, without changing the connections amongst the neurons. This lecture will first summarize our investigations of the molecular and biochemical mechanisms that change synaptic function to persistently store spatial memory in the rodent hippocampus. I will then report on how entorhinal cortex-hippocampus circuit function changes during cognitive training that creates memory, as well as learning to learn in mice. I will then describe how the hippocampus system operates like a competitive winner-take-all network, that, based on the dominance of its current inputs, self organizes into either the encoding or recollection information processing modes. We find no evidence that distinct cells are dedicated to those two distinct functions, rather activation of the hippocampus information processing mode is controlled by a subset of dentate spike events within the network of learning-modified, entorhinal-hippocampus excitatory and inhibitory synapses.
Memory, learning to learn, and control of cognitive representations
Biological neural networks can represent information in the collective action potential discharge of neurons, and store that information amongst the synaptic connections between the neurons that both comprise the network and govern its function. The strength and organization of synaptic connections adjust during learning, but many cognitive neural systems are multifunctional, making it unclear how continuous activity alternates between the transient and discrete cognitive functions like encoding current information and recollecting past information, without changing the connections amongst the neurons. This lecture will first summarize our investigations of the molecular and biochemical mechanisms that change synaptic function to persistently store spatial memory in the rodent hippocampus. I will then report on how entorhinal cortex-hippocampus circuit function changes during cognitive training that creates memory, as well as learning to learn in mice. I will then describe how the hippocampus system operates like a competitive winner-take-all network, that, based on the dominance of its current inputs, self organizes into either the encoding or recollection information processing modes. We find no evidence that distinct cells are dedicated to those two distinct functions, rather activation of the hippocampus information processing mode is controlled by a subset of dentate spike events within the network of learning-modified, entorhinal-hippocampus excitatory and inhibitory synapses.
Restless engrams: the origin of continually reconfiguring neural representations
During learning, populations of neurons alter their connectivity and activity patterns, enabling the brain to construct a model of the external world. Conventional wisdom holds that the durability of a such a model is reflected in the stability of neural responses and the stability of synaptic connections that form memory engrams. However, recent experimental findings have challenged this idea, revealing that neural population activity in circuits involved in sensory perception, motor planning and spatial memory continually change over time during familiar behavioural tasks. This continual change suggests significant redundancy in neural representations, with many circuit configurations providing equivalent function. I will describe recent work that explores the consequences of such redundancy for learning and for task representation. Despite large changes in neural activity, we find cortical responses in sensorimotor tasks admit a relatively stable readout at the population level. Furthermore, we find that redundancy in circuit connectivity can make a task easier to learn and compensate for deficiencies in biological learning rules. Finally, if neuronal connections are subject to an unavoidable level of turnover, the level of plasticity required to optimally maintain a memory is generally lower than the total change due to turnover itself, predicting continual reconfiguration of an engram.
Cortical interneuron wiring in health and disease
The establishment of synaptic connections is essential for normal brain function, yet the molecular mechanisms responsible for the precise connectivity of specific neural circuits remain largely unknown. Previous work has shown that the assembly of cortical circuits requires specific functions of molecular signalling complexes at different classes of synapses. In this talk, I will describe the molecular logic through which specific pyramidal cell-interneuron circuits are established in the cerebral cortex of the mouse, and how alterations in some of these connectivity motifs might be liked to disease.
Towards multipurpose biophysics-based mathematical models of cortical circuits
Starting with the work of Hodgkin and Huxley in the 1950s, we now have a fairly good understanding of how the spiking activity of neurons can be modelled mathematically. For cortical circuits the understanding is much more limited. Most network studies have considered stylized models with a single or a handful of neuronal populations consisting of identical neurons with statistically identical connection properties. However, real cortical networks have heterogeneous neural populations and much more structured synaptic connections. Unlike typical simplified cortical network models, real networks are also “multipurpose” in that they perform multiple functions. Historically the lack of computational resources has hampered the mathematical exploration of cortical networks. With the advent of modern supercomputers, however, simulations of networks comprising hundreds of thousands biologically detailed neurons are becoming feasible (Einevoll et al, Neuron, 2019). Further, a large-scale biologically network model of the mouse primary visual cortex comprising 230.000 neurons has recently been developed at the Allen Institute for Brain Science (Billeh et al, Neuron, 2020). Using this model as a starting point, I will discuss how we can move towards multipurpose models that incorporate the true biological complexity of cortical circuits and faithfully reproduce multiple experimental observables such as spiking activity, local field potentials or two-photon calcium imaging signals. Further, I will discuss how such validated comprehensive network models can be used to gain insights into the functioning of cortical circuits.
Hippocampal disinhibitory circuits: cell types, connectivity and function
The concept of a dynamic excitation / inhibition ratio, that can shape information flow in cortical circuits during complex behavioural tasks due to circuit disinhibition, has recently arisen as an important and conserved processing motif. It has been also recognized that, in cortical circuits, a subpopulation of GABAergic cells that express vasoactive intestinal polypeptide (VIP) innervates selectively inhibitory interneurons, providing for circuit disinhibition as a possible outcome, depending on the network state and behavioural context. In this talk, I will highlight the latest discoveries on the dynamic organization of hippocampal disinhibitory circuits with a focus on VIP-expressing interneurons. I will discuss the neuron types that can be involved in disinhibition and their local circuit and long-range synaptic connections. I will also discuss some recent findings on how hippocampal VIP circuits may coordinate spatial learning.