Schizophrenia has been associated with separate irregularities in several neural oscillatory frequency bands, including theta, alpha, and gamma. Our multivariate classification suggests that instead of irregularities in many frequency bands, schizophrenia-related EEG differences may better be explained by an overall shift in neural noise, reflected by a change in the 1/f slope of the power spectrum. Significance statementUnderstanding the neurobiological origins of schizophrenia, and identifying reliable biomarkers, are both of critical importance in improving treatment of that disease. While we lack predictive biomarkers, numerous studies have observed disruptions to neural oscillations in schizophrenia patients. This literature has, in part, lead to schizophrenia being characterized as disease of disrupted neural coordination. We report however that changes to background noise (i.e., 1/f noise) are a substantially better predictor of schizophrenia than oscillatory power. The observed alterations in neural noise are consistent with inhibitory neuron dysfunctions associated with schizophrenia, allowing for a direct link between noninvasive EEG and neurobiological deficits.

For all animals the decision to explore comes with a risk of getting less. For example, a foraging bee might find less nectar, or hunting hawk less prey. This loss is often formalized as regret. It's been mathematically proven that exploring an uncertain world with a specific goal always has some regret. This is why exploration-exploitation can be a dilemma. Given this proof we wondered if the common advice to "focus on learning and not the goal" might have mathematical merit. So we re-imagined exploration in the dilemma as an open ended search for any new information. We then developed a new minimal description of information value, which generalizes existing ideas like curiosity, novelty and information gain. We use this description to model the dilemma as a competition between strategies that maximize reward and information independently. Here we prove this competition has a no regret solution. When we study this solution in simulation -using classic bandit tasks -it outperforms standard approaches, especially when rewards are sparse.

I demonstrate theoretically that calcium waves in astrocytes can compute anything neurons can. A foundational result in neural computation was proving the firing rate model of neurons defines a universal function approximator. In this work I show a similar proof extends to a model of calcium waves in astrocytes, which I confirm in a series of computer simulations. I argue the major limit in astrocyte computation is not their ability to find approximate solutions, but their computational complexity. I suggest some initial experiments that might be used to confirm these predictions.

One Sentence Summary: Novel technology (Path-seq) discovers cell wall remodeling program 24 during Mycobacterium tuberculosis infection of macrophages 26 Abstract:The success of Mycobacterium tuberculosis (MTB) stems from its ability to remain hidden from the 28 immune system within macrophages. Here, we report a new technology (Path-seq) to sequence miniscule amounts of MTB transcripts within up to million-fold excess host RNA. Using Path-seq we have 30 discovered a novel transcriptional program for in vivo mycobacterial cell wall remodeling when the pathogen infects alveolar macrophages in mice. We have discovered that MadR transcriptionally 32 modulates two mycolic acid desaturases desA1/A2 to initially promote cell wall remodeling upon in vitro macrophage infection and, subsequently, reduces mycolate biosynthesis upon entering dormancy. We 34 demonstrate that disrupting MadR program is lethal to diverse mycobacteria making this evolutionarily conserved regulator a prime antitubercular target for both early and late stages of infection. 36 38 40 Main Text: 42Mycobacterium tuberculosis (MTB) infection occurs by inhalation of bacilli-containing aerosols. 44Alveolar macrophages, which line the airway, are the first host cells to phagocytize the bacteria. This initial contact of MTB with alveolar macrophages begins a complex battle between bacterial virulence 46 and host immunity, orchestrated in large part by intricate gene regulatory pathways(1, 2). As such, measuring gene expression in vivo is central to our understanding of TB disease control and 48 progression(3).RNA-seq provides a sensitive method for global gene expression analysis. Specific for infection 50 biology, dual RNA-seq methods have allowed simultaneous profiling of host and pathogen RNA. However, the striking excess of eukaryotic over bacteria RNA limits the coverage of pathogen transcripts 52 in dual RNA-seq studies(4-8), and methods to partially enrich for bacterial transcripts have had limited success(9, 10). It is clear more sensitive approaches are needed to profile the transcriptional state of the 54 pathogen during infection, especially in vivo.To improve the coverage of pathogen transcripts, we made use of biotinylated oligonucleotide 56 baits that are complementary to the pathogen transcriptome. The baits are hybridized to mixed hostpathogen RNA and used to enrich pathogen transcripts for sequencing. We applied our pathogen-58 sequencing (Path-seq) method to explore transcriptional changes in MTB following infection in mice.Here Path-seq has led to discovery that MTB transcriptionally regulate mycolic acids during infection of 60 host cells, influencing virulence and persistence of the pathogen. 62RESULTS and DISCUSSION 64Development of Path-seq To enrich the bacterial pathogen transcripts, we used Agilent eArray(11) to create a custom bait library 66 that covers all MTB transcripts at even intervals. Our MTB library contains 35,624 probes, each with biotinylated oligonucleotides of 120 base lengths. The bait library composition is modul...

Deep reinforcement learning can match or exceed human performance in stable contexts, but with minor changes to the environment artificial networks, unlike humans, often cannot adapt. Humans rely on a combination of heuristics to simplify computational load and imagination to extend experiential learning to new and more challenging environments. Motivated by theories of the hierarchical organization of the human prefrontal networks, we have developed a model of hierarchical reinforcement learning that combines both heuristics and imagination into a “stumbler-strategist” network. We test performance of this network using Wythoff’s game, a gridworld environment with a known optimal strategy. We show that a heuristic labeling of each position as hot or cold, combined with imagined play, both accelerates learning and promotes transfer to novel games, while also improving model interpretability