This research focuses on handling nonlinear noise in recurrent neuronal networks. Correlated noise passing through nonlinear functions complicates the analysis of complex phenomena such as transients and bifurcations.

The proposed method replaces nonlinear functions of Ornstein-Uhlenbeck (OU) noise with a Gaussian-equivalent process matched in mean and covariance. For expansive nonlinearities, it combines this with a lognormal moment closure to derive a closed dynamical mean-field theory for recurrent neuronal networks.

The resulting theory captures order-one transients, fixed points, and noise-induced shifts of bifurcation structure, outperforming standard linearization-based approximations in the strong-fluctuation regime. This approach offers a tractable route to noise-dependent phase diagrams in computational neuroscience models.

👉 Read the full article on arXiv

This study proposes a machine learning approach to improve the diagnosis of non-amnestic Alzheimer's disease (atAD), which is often misdiagnosed compared to typical AD (tAD).

The method utilizes a combination of clinical testing batteries and MRI data, including hippocampal volume and comprehensive brain-wide features, to classify atAD from non-AD cognitive impairment. The machine learning model significantly enhances the detection rate (recall) of atAD cases compared to using hippocampal volume alone.

Performance improvements were observed from 52% to 69% for the NACC dataset and 34% to 77% for the ADNI dataset. This approach holds significant implications for improving diagnostic accuracy for non-amnestic atAD in clinical settings using standard clinical tests and MRI.

👉 Read the full article on arXiv

This research introduces a computational framework for schema formation and utilization, which are crucial for generalizing experience and adapting to new situations. Schema-based hierarchical active inference (S-HAI) integrates predictive processing and active inference with schema-based mechanisms.

In S-HAI, a higher-level generative model encodes abstract task structure, while a lower-level model encodes spatial navigation. Simulations demonstrate that S-HAI reproduces key behavioral signatures of rapid schema-based generalization in spatial navigation tasks, including the flexible remapping of abstract schemas onto novel contexts.

Furthermore, S-HAI replicates prominent neural codes reported in rodent medial prefrontal cortex, such as task-invariant goal-progress cells and goal-identity cells. These findings suggest that schema formation and generalization may arise from hierarchical predictive processing principles implemented across cortical and hippocampal circuits.

👉 Read the full article on arXiv