Synaptic promiscuity in brain development

BC-PMJRS: A Brain Computing-inspired Predefined Multimodal Joint Representation Spaces for enhanced cross-modal learning

Multimodal learning faces two key challenges: effectively fusing complex information from different modalities, and designing efficient mechanisms for cross-modal interactions. Inspired by neural plasticity and information processing principles in the human brain, this paper proposes BC-PMJRS, a Brain Computing-inspired Predefined Multimodal Joint Representation Spaces method to enhance cross-modal learning. The method learns the joint representation space through two complementary optimization objectives: (1) minimizing mutual information between representations of different modalities to reduce redundancy and (2) maximizing mutual information between joint representations and sentiment labels to improve task-specific discrimination. These objectives are balanced dynamically using an adaptive optimization strategy inspired by long-term potentiation (LTP) and long-term depression (LTD) mechanisms. Furthermore, we significantly reduce the computational complexity of modal interactions by leveraging a global-local cross-modal interaction mechanism, analogous to selective attention in the brain. Experimental results on the IEMOCAP, MOSI, and MOSEI datasets demonstrate that BC-PMJRS outperforms state-of-the-art models in both complete and incomplete modality settings, achieving up to a 1.9% improvement in weighted-F1 on IEMOCAP, a 2.8% gain in 7-class accuracy on MOSI, and a 2.9% increase in 7-class accuracy on MOSEI. These substantial improvements across multiple datasets demonstrate that incorporating brain-inspired mechanisms, particularly the dynamic balance of information redundancy and task relevance through neural plasticity principles, effectively enhances multimodal learning. This work bridges neuroscience principles with multimodal machine learning, offering new insights for developing more effective and biologically plausible models.

Neural circuit assembly: A sticky cue that connects neurons with a preference for hue

Neurons rely on molecular interactions typically mediated by transmembrane adhesion proteins to locate appropriate partners and establish connections. A recent study finds that a member of the cadherin family of cell adhesion proteins organizes color-preferring connections in the part of the retinal neural circuit designated for encoding light decrements.

Expanding ligand-receptor interaction networks for axon guidance: Structural insights into signal crosstalk and specificity

Guidance of nascent axons to their targets is mediated by attractive and repulsive cues that activate receptors on the axonal growth cone. The number of ligand-receptor interactions implicated in axon pathfinding is still expanding, and large-scale cell-surface and extracellular protein interactome studies have revealed extensive crosstalk between signaling axes once thought to act independently. This raises the question how the apparent promiscuity of molecular interactions is compatible with specific signaling outcomes and effects on growth cone steering. Structural studies have provided insights into the modularity of binding interactions and shown the capacity of receptors to engage multiple ligands. Here, we review recent findings about the complexity of ligand-receptor interaction networks for axon guidance, and how structures of ligand-receptor complexes reveal mechanisms that may specify signaling output.

Biased cell adhesion organizes the Drosophila visual motion integration circuit

Layer-specific brain computations depend on neurons synapsing with specific partners in distinct laminae. In the Drosophila lobula plate, axons of the four subtypes of T4 and T5 visual motion direction-selective neurons segregate into four layers, where they synapse with distinct subsets of postsynaptic neurons. Here, we identify a layer-specific expression of different receptor-ligand pairs of the Beat and Side families of cell adhesion molecules between T4/T5s and their postsynaptic partners. Developmental genetic analysis demonstrate that Beat/Side-mediated interactions are required to restrict innervation of T4/T5 axons and the dendrites of their partners to a single layer. We show that Beat/Side interactions are not required for synaptogenesis. Instead, they contribute to synaptic specificity by biasing cellular adjacency, causing neurons to segregate in discrete layers, restricting partner availability before synaptogenesis. We propose that the emergence of synaptic specificity relies on a competitive dynamic among postsynaptic partners with shared Beat/Side expression to adhere with T4/T5s.

Nervous system-wide analysis of all C. elegans cadherins reveals neuron-specific functions across multiple anatomical scales

Differential expression of cell adhesion proteins is a hallmark of cell-type diversity across the animal kingdom. Gene family-wide characterization of their organismal expression and function is, however, lacking. Using genome-engineered reporter alleles, we established an atlas of expression of the entire set of 12 cadherin gene family members in the nematode Caenorhabditis elegans, revealing differential expression across neuronal classes, a dichotomy between broadly and narrowly expressed cadherins, and several context-dependent temporal transitions in expression across development. Engineered mutant null alleles of cadherins were analyzed for defects in morphology, behavior, neuronal soma positions, neurite neighborhood topology and fasciculation, and localization of synapses in many parts of the nervous system. This analysis revealed a restricted pattern of neuronal differentiation defects at discrete subsets of anatomical scales, including a novel role of cadherins in experience-dependent electrical synapse formation. In total, our analysis results in previously little explored perspectives on cadherin deployment and function.

Spatial constraints and cell surface molecule depletion structure a randomly connected learning circuit

The brain can represent almost limitless objects to "categorize an unlabeled world" (Edelman, 1989). This feat is supported by expansion layer circuit architectures, in which neurons carrying information about discrete sensory channels make combinatorial connections onto much larger postsynaptic populations. Combinatorial connections in expansion layers are modeled as randomized sets. The extent to which randomized wiring exists in vivo is debated, and how combinatorial connectivity patterns are generated during development is not understood. Non-deterministic wiring algorithms could program such connectivity using minimal genomic information. Here, we investigate anatomic and transcriptional patterns and perturb partner availability to ask how Kenyon cells, the expansion layer neurons of the insect mushroom body, obtain combinatorial input from olfactory projection neurons. Olfactory projection neurons form their presynaptic outputs in an orderly, predictable, and biased fashion. We find that Kenyon cells accept spatially co-located but molecularly heterogeneous inputs from this orderly map, and ask how Kenyon cell surface molecule expression impacts partner choice. Cell surface immunoglobulins are broadly depleted in Kenyon cells, and we propose that this allows them to form connections with molecularly heterogeneous partners. This model can explain how developmentally identical neurons acquire diverse wiring identities.