Neuroscience

eLife Assessment

The article presents important findings of a dissociation between phasic and tonic pain functions in adaptive behavior, combining immersive VR, computational modeling, skin conductance, and EEG data. The methodology used is convincing. Its ecological design and sophisticated computational modeling are major strengths.

eLife Assessment

This valuable study draws on large-scale multimodal MRI measurements of human brain structure across the lifespan to offer a new perspective on visual cortex architecture. The data provide compelling evidence for two cortical architectural gradients that show distinct functional, cytoarchitectural, behavioural, and lifespan profiles. One gradient captures a broad early-to-higher-level visual cortical hierarchy in which cortical thickness tissue density covary; the other reflects more localised divergence from this relationship, notably predicting putative anterior temporal visual field representations that have not previously been described.

eLife Assessment

This important study uses a feedback-driven recurrent neural network framework to explore the dynamics underlying learning of BCI decoder perturbations. With convincing evidence, the authors demonstrate that behavioral learning trajectories that match those of primates learning within-manifold and outside-manifold perturbations are likely tied to the dynamical controllability of the network and input-driven learning. This work is likely to motivate a new generation of BCI and learning experiments combining large-scale neural recordings with latent dynamical systems analyses.

eLife Assessment

This is a valuable paper that compares various deep learning models, trained with different objective functions, on their ability to predict fMRI data collected during naturalistic video gameplay. The data and analysis provide solid within-distribution evidence that models trained with PPO and imitation learning outperform untrained models and standard convolutional networks. However, the evidence for brittleness in out-of-distribution encoding remains incomplete, as the claim that this stems from the networks' training rather than from alternative causes-like overfitting of ridge regression parameters-is not yet fully supported.