André Moraes Bastos

Predictive processing theories propose that the brain builds internal models of its environment by reducing the discrepancy between internally generated predictions and external sensory signals. Prior work has linked these processes to oscillatory activity in gamma (40–100 Hz) and alpha/beta (10–30 Hz) frequency ranges. Current computational approaches face a trade-off: abstract predictive-processing models can implement self-supervised computations but often omit oscillatory spiking dynamics, whereas biophysically constrained spiking models can generate neural rhythms but often require extensive manual tuning. Here, we introduce the Genetic Stochastic Delta Rule (GSDR), an evolutionary optimization framework for fitting nonlinear neural models to electrophysiological objectives. We first evaluate GSDR in simplified optimization settings, then apply it to spiking-network objectives involving firing rates, beta/gamma spectral ratios, and empirical macaque stimulus-evoked gamma dynamics from visual cortex. We show that GSDR can search constrained synaptic parameter spaces, reduce reliance on manual tuning, and reproduce spectral and circuit-level phenotypes associated with predictive routing. We also used Izhikevich simulations as a model-class robustness analysis, showing that the approach is not limited to the original Hodgkin-Huxley-style implementation. These results position GSDR as a methodological framework for automated, multi-objective exploration of oscillatory neural models, with predictive routing serving as a motivating case study rather than as a completed functional proof.

Anesthetic-induced unconsciousness may arise partly from a change in how brain areas can manipulate each other’s activity. To quantify this change, we use control-theoretic tools that can precisely characterize how easily subsystems in complex networks can control each other. These tools rely on the Jacobian of the dynamics, an object that fully specifies how inputs to a function affect the outputs. We built on JacobianODE, a method for data-driven Jacobian learning, to enable its application to neural recordings. We developed a deep learning framework that recovers directional, nonlinear control from partially observed multi-area recordings by combining delay-coordinate embedding, a volume-preserving invertible encoder, and latent dynamics via JacobianODE. We validated our framework on the Lorenz system and a partially observed working-memory recurrent neural network. We then applied it to local field potential recordings from posterior parietal (PPC), superior temporal gyrus (STG), frontal eye fields (FEF), and ventrolateral prefrontal cortex (vlPFC) in two non-human primates, comparing wakefulness with propofol anesthesia. Anesthesia pervasively reduced the ability of areas to control each other, both in terms of driving towards novel states and stabilizing along existing trajectories. This change in ease of control was driven by a decrease in the magnitude of interareal coupling. Directional ease of driving control from PPC to vlPFC and FEF to vlPFC was increased under anesthesia, providing a potential mechanism for paradoxical excitation observed during propofol infusion. Together, these results recast anesthetic unconsciousness as a directional breakdown of cortical control.

Our proposal introduces a novel therapeutic approach to treat neurodegenerative disorders such as Parkinson's disease and Alzheimer's disease. Our proposed method combines solid scientific evidence about misfolded protein aggregates' connection to these disorders with a fundamental physical principle stating that an object's resonance frequency is determined by its mechanical structure and mass distribution. Applied to molecular systems, this implies that misfolded proteins, due to their distinct tertiary and quaternary structures, will necessarily exhibit different vibrational and resonance profiles compared to their native forms. The treatment we propose uses frequency-adjusted high-intensity focused ultrasound or microwave systems to target and denature certain pathological protein clusters. Our approach intends to eliminate misfolded proteins such as alpha-synuclein from Parkinson's disease and amyloid beta from Alzheimer's disease through their specific resonance frequencies. These distinct resonance frequencies reflect variations in their structural conformations and post-translational modifications when compared to healthy proteins. This method aims to neutralize dysfunctional proteins with selective precision to prevent damage to adjacent healthy tissue. We acknowledge significant challenges ahead, notably the precise determination of misfolded protein resonance frequencies in vivo, ensuring targeted specificity, and managing the potential overload of biological clearance mechanisms.

The appearance at the anatomical level of a canonical laminar microcircuit suggests that each six-layer column of granular cortex may mediate a canonical computation. Hypotheses for such computations include predictive coding, predictive routing, efficient coding, and others. However, single-neuron recordings capture only the individual elements of the hypothesized laminar microcircuit, while local field potentials (LFPs) from a laminar probe offer insight into the broader population activity. Through the Allen Institute’s OpenScope Brain Observatory, data in mice performing a visual oddball task during multi-area laminar recording was used to test predictive processing hypotheses in the spectral domain. Histological labeling of the cortical laminae enabled a fine-grained examination of their roles in the task, and frequency bands capturing both feedforward and feedback effects were analyzed. ɣ-band local-field potential (LFP) oscillations conveyed feedforward prediction errors in lower sensory areas of cortex; ⍺/β-band oscillations weakened in unpredictable conditions compared to predictable ones; and θ-band oscillations additionally signalled slower, longer-scale temporal prediction errors. In combination with the previous findings, predictive routing explains these experiments where neither ubiquitous predictive coding nor feedforward adaptation can.

The appearance at the anatomical level of a canonical laminar microcircuit suggests that each six-layer column of granular cortex may mediate a canonical computation. Hypotheses for such computations include predictive coding, predictive routing, efficient coding, and others. However, single-neuron recordings capture only the individual elements of the hypothesized laminar microcircuit, while local field potentials (LFPs) from a laminar probe offer insight into the broader population activity. Through the Allen Institute’s OpenScope Brain Observatory, data in mice performing a visual oddball task during multi-area laminar recording was used to test predictive processing hypotheses in the spectral domain. Histological labeling of the cortical laminae enabled a fine-grained examination of their roles in the task, and frequency bands capturing both feedforward and feedback effects were analyzed. ɣ-band local-field potential (LFP) oscillations conveyed feedforward prediction errors in lower sensory areas of cortex; ⍺/β-band oscillations weakened in unpredictable conditions compared to predictable ones; and θ-band oscillations additionally signalled slower, longer-scale temporal prediction errors. In combination with the previous findings, predictive routing explains these experiments where neither ubiquitous predictive coding nor feedforward adaptation can. Significance: Cortical columns robustly signal perceptual features through the firing rates of spiking neurons. In accordance with this rate coding, predictive processing theories hypothesized that neuronal firing rates ubiquitously signal surprise. However, a recent large-scale study of spike rates did not support this conjecture. An alternate model, predictive routing, suggests that neuronal oscillations rather than spike rates could encode surprise. These neuronal oscillations, which can affect the timing but not rate of spiking, formed coherent ɣ rhythms which consistently signaled both simpler and more complex forms of surprise in mouse visual cortex. Together with the findings on spike-rates in the same experiment, our findings suggest that cortical circuits encode surprise in the rhythmic timing of spikes rather than in their rate.

Importance GABAergic inhibitory interneurons have been implicated in the pathophysiology of schizophrenia. However, there is conflicting evidence regarding the nature and extent of the deficits across brain areas and interneuron subtypes. Objective The primary objective was to determine the extent of changes in parvalbumin, somatostatin, calbindin, and calretinin interneurons across brain regions and cortical layers in schizophrenia compared to healthy controls. The secondary objective was to examine differences in neuronal density and neuronal mRNA expression of GABAergic interneurons. Data Sources A comprehensive search was conducted from Fall 2024 to Spring 2025, including terms related to schizophrenia, interneurons, immunohistochemistry and mRNA. Only post-mortem human studies containing neuroanatomical data of parvalbumin, somatostatin, calbindin, and calretinin interneurons were considered. Study Selection We selected immunohistochemistry and mRNA studies that examined parvalbumin, somatostatin, calbindin, and calretinin interneuron density or expression in schizophrenia patients. Data Extraction and Synthesis Data were extracted following PRISMA guidelines and were pooled with a random-effects model. A t-score statistic was applied to obtain distributions of effects. A two-sample t-test and Cohen’s effect size was used to compare across areas, layers, cell types and methods. Main Outcome(s) and Measure(s) Primary measurements/outcome were laminar interneuron density (assessed by immunohistochemistry) and gene expression (assessed by mRNA). Results Data from 28 immunohistochemistry studies (362 control participants, 335 individuals with schizophrenia) revealed that the hippocampus and prefrontal cortex were most consistently characterized by alterations in GABAergic interneurons; parvalbumin and somatostatin interneuron density was reduced in the hippocampus, while data from 18 mRNA studies (524 control participants, 519 individuals with schizophrenia) indicated reduced parvalbumin and somatostatin expression in prefrontal cortex. Layer-specific analysis demonstrated that parvalbumin interneurons were most affected in the superficial layers of prefrontal cortex, while somatostatin interneurons exhibited the strongest deficits in layers 2 and 5. Conclusions and Relevance Our results show that GABAergic interneurons in the PFC and hippocampus are particularly affected. Parvalbumin and somatostatin showed the largest deficits, involving superficial layers and layer 5. We also identified significant reductions in parvalbumin and calretinin interneuron density in subcortical areas. Together, these data have important implications for the pathophysiology and computational models of circuits deficits in the disorder. Question What are the alterations in GABAergic interneurons in specific brain regions and cortical layers in ScZ compared to HCs? Findings This meta-analysis of 46 post-mortem studies showed significant reductions in PV and SST interneurons in the PFC and hippocampus. PV interneuron density was reduced in superficial layers of the PFC, while SST mRNA expression was reduced in layers 2 and 5. Meaning These findings suggest that specific GABAergic interneuron deficits, particularly PV and SST in the PFC and hippocampus, may contribute the most to the pathophysiology of ScZ.

Summary: Our study (Mendoza-Halliday et al., 2024) made two contributions: (1) discovery of a ubiquitous cortical motif and (2) a tool derived from it—the Frequency-based Layer Identification Procedure (FLIP and vFLIP). Mackey et al. critique the tool, questioning its advantage over classic current source density (CSD) analysis, and reason backwards to challenge the motif’s ubiquity. In our rebuttal, we confirm the spectrolaminar motif in diverse cortical areas using data from multiple research groups (who joined us in this rebuttal) as well as Mackey et al.’s own dataset. Additionally, we introduce vFLIP2, an improved version of our tool that addresses their comments. It reliably identified and localized the motif in our data and Mackey et al.’s data. Our findings reaffirm the motif’s ubiquity. We value Mackey et al.’s comments, which helped refine our tool.

Predictive processing theories propose that the brain builds internal models of its environment by reducing the discrepancy between internally generated predictions and external sensory signals. Prior work has linked these processes to oscillatory activity in gamma (40-100 Hz) and alpha/beta (10-30 Hz) frequency ranges. Current computational approaches face a trade-off: abstract predictive-processing models can implement self-supervised computations but often omit oscillatory spiking dynamics, whereas biophysically constrained spiking models can generate neural rhythms but often require extensive manual tuning. Here, we introduce the Genetic Stochastic Delta Rule (GSDR), an evolutionary optimization framework for fitting nonlinear neural models to electrophysiological objectives. We first evaluate GSDR in simplified optimization settings, then apply it to spiking-network objectives involving firing rates, beta/gamma spectral ratios, and empirical macaque stimulus-evoked gamma dynamics from visual cortex. We show that GSDR can search constrained synaptic parameter spaces, reduce reliance on manual tuning, and reproduce spectral and circuit-level phenotypes associated with predictive routing. We also used Izhikevich simulations as a model-class robustness analysis, showing that the approach is not limited to the original Hodgkin-Huxley-style implementation. These results position GSDR as a methodological framework for multi-objective exploration of oscillatory neural models. Author summary: In predictive processing theories, the brain is hypothesized to build internal models of its environment. Empirical and theoretical studies suggest that neuronal oscillations are important components of this process, and abnormal oscillations are also linked to disorders such as schizophrenia. To study such mechanisms, computational neuroscience needs models that can express biologically meaningful spiking and oscillatory dynamics while also being trainable without extensive manual tuning. We developed the Genetic Stochastic Delta Rule (GSDR), a self-supervised evolutionary optimization framework for fitting nonlinear neural models to objectives. GSDR combines objective-guided search, stochastic exploration, genetic selection/deselection, and an activity-dependent MCDP update term. We show that GSDR can tune spiking networks toward beta/gamma spectral objectives and empirical stimulus-evoked gamma dynamics. The results do not prove predictive routing or identify a unique biological circuit; rather, they show that GSDR can identify candidate circuit configurations and can generalize beyond the original Hodgkin-Huxley-style model to Izhikevich simulations.