Understanding Neural Processing as an Integrated Intelligent System-Volume II

Understanding how the brain produces perception, memory, learning, and adaptive behavior requires models that link circuit structure, neural dynamics, and computation across levels of organization. Since the launch of Volume I (link here), the field has advanced rapidly, driven by (i) large-scale multimodal recordings that pair neural activity with anatomy, cell types, and behavior; (ii) expanding functional connectomics and cell-type–resolved circuit maps; and (iii) the adoption of modern machine learning, especially foundation-model and representation-learning approaches, to build generalizable models of neural activity and behavior.At the same time, the community is increasingly emphasizing causal and perturbation-based validation. Rather than evaluating models only by fit or decoding performance, there is growing demand for models that produce testable predictions under intervention, remain robust across subjects and sessions, and clarify what computations are implemented by specific circuits. This shift creates an ideal moment to refresh the scope with a Volume II that supports both mechanistic neuroscience and NeuroAI.Volume II will capture emerging frameworks that treat neural systems as integrated intelligent agents—systems that learn, predict, control, and adapt under biological constraints. We welcome empirical, theoretical, and methods contributions that connect circuit-level mechanisms to computation and behavior, including work that integrates heterogeneous datasets, introduces interpretable models, and establishes evaluation standards that promote generalization and causal validity.Topics of interest (indicative, not exhaustive):• Foundation models for neuroscience: scalable models of neural activity/behavior; self-supervised learning; transfer across subjects, tasks, and modalities• Multimodal and multiscale integration: linking activity with cell types, anatomy, connectivity, and behavior; unified analysis across spikes, calcium imaging, LFP/ECoG, and fMRI• Functional connectomics and circuit computation: how wiring and cell-type organization shape dynamics, learning, and representations• Causal modeling and perturbation-aware neuroscience: optogenetics/electrical stimulation/pharmacology/lesions; counterfactual modeling; closed-loop experiments• Neural dynamics and geometry: manifolds, attractors, sequences, oscillations, compositional dynamics; bridging dynamical systems and learning• Learning rules and credit assignment: biologically grounded learning, neuromodulation, multi-timescale plasticity, continual learning.• Inference, prediction, and control: predictive processing, Bayesian/probabilistic coding, reinforcement learning, active inference, and control-theoretic accounts with circuit implementations• Brain–machine interfaces and real-time intelligence: robust decoding/encoding, adaptive control, latency constraints, safety and stability• Interpretability and mechanistic AI: methods that transform predictive models into circuit hypotheses; model comparison and ablation strategies• Benchmarks, datasets, and standards: open resources, evaluation metrics aligned with scientific questions, reproducibility and reporting practices Topic editors: David Wack, Konstantinos Slavakis, Rajendra D Badgaiyan, Paul John Werbos Source: https://www.frontiersin.org/research-topics/86947/understanding-neural-processing-as-an-integrated-intelligent-system-volume-ii