bioRxiv Subject Collection: Neuroscience's Journal

Reinforcement learning of state representation and value: the power of random feedback and biological constraints
How external/internal 'state' is represented in the brain is crucial, since appropriate representation enables goal-directed behavior. Recent studies suggest that state representation and state value can be simultaneously learnt through reinforcement learning (RL) using reward-prediction-error in recurrent-neural-network (RNN) and its downstream weights. However, how such learning can be neurally implemented remains unclear because training of RNN through the 'backpropagation' method requires downstream weights, which are biologically unavailable at the upstream RNN. Here we show that training of RNN using random feedback instead of the downstream weights still works because of the 'feedback alignment', which was originally demonstrated for supervised learning. We further show that if the downstream weights and the random feedback are biologically constrained to be non-negative, learning still occurs without feedback alignment because the non-negative constraint ensures loose alignment. These results suggest neural mechanisms for RL of state representation/value and the power of random feedback and biological constraints.

Terminal nucleotidyltransferase Tent2 microRNA tailing regulates excitatory/inhibitory balance in the hippocampus
One of the post-transcriptional mechanisms regulating the stability of RNA molecules involves the addition of non-templated nucleotides to their 3' ends, a process known as RNA tailing. To systematically investigate the physiological consequences of terminal nucleotidyltransferase TENT2 absence on RNA 3' end modifications in the mouse hippocampus we developed a new Tent2 knockout mouse. Electrophysiological measurements revealed increased excitability in Tent2 KO hippocampal neurons, and behavioral analyses showed decreased anxiety and improved fear extinction in these mice. At the molecular level, we observed a significant contribution of TENT2 to the monoadenylation of various classes of miRNAs, but found no effect of the enzyme's loss on the total poly(A) tail length of mRNAs, as measured by Direct Nanopore RNA sequencing. Alterations in the monoadenylation of a large population of microRNAs affected the overall mRNA abundance, particularly transcripts related synaptic transmission, which were downregulated in the hippocampus of Tent2 knockout mice. These changes explain the observed behavioral and electrophysiological alterations. Our data thus establish a link between TENT2-dependent microRNA tailing and the balance of inhibitory and excitatory neurotransmission.

Retrosplenial cortex encodes both local and global space in connected environments
In everyday life, mammals have to find their way between interconnected spaces. How the brain processes spatial information in these complex environments is a crucial question for understanding how we navigate. Theoretical frameworks have proposed that the brain presents simultaneous activities that are either maintained in connected spaces (global reference frame) or distinguished in each space (local reference frames). The different head direction signals discovered in the retrosplenial cortex (RSC) suggest that this brain area supports both local and global spatial activity. We tested this hypothesis by recording RSC neurons in environments consisting of 2 or 4 connected rooms that are either identical or different. We observed two types of head direction activity simultaneously: one showing opposite or orthogonal directional activity in the 2 and 4 connected spaces and thus anchored to local reference frames, and one maintaining a stable head direction activity across connected spaces and thus anchored to the global reference frame. We also found that other RSC neurons show a spatial pattern that repeats oppositely or orthogonally in the 2 and 4 connected spaces respectively. In contrast, hippocampal place cells activity never shows opposite or orthogonal patterns, but rather a spatial remapping or a translational repeated activity. Overall, the results show that only the RSC has the simultaneous neuronal activity necessary to form a spatial representation in connected environments.

Sustained activity in a descending neuron is associated with flight saccades in Drosophila
Flies perform rapid turns termed saccades to change direction during flight. Evasive turns can be elicited by looming stimuli mimicking an approaching object such as a potential predator. Whereas projection neurons of the optic lobes responsive to looming stimuli have been well described, how this information is transmitted to the motor system to elicit a saccade, is not well understood. Here we describe activity of the descending neuron DNp03 in Drosophila, which receives direct input from looming-sensitive visual interneurons and projects to wing motor areas within the ventral nerve cord. Whole-cell patch-clamp recordings from this neuron during head-fixed flight confirm that DNp03 is responsive to looming stimuli on the side ipsilateral to its dendrites. In addition, activity of this neuron is state-dependent as looming stimuli only elicit spikes during flight and not rest. The behavior in response to the looming stimulus is variable, which allowed us to study how activity of DNp03 relates to the execution of a saccade. Our analysis revealed that sustained activity in DNp03, persisting even after the visual stimulus ended, was the strongest predictor of saccade execution, therefore reflecting the behavioral decision of the fly to respond to the stimulus.

A cup of coffee can boost your motor performance: Enhanced motor sequence learning with caffeine
Caffeine is consumed in various beverages, such as coffee, energy drinks, soda, and tea. The effects of caffeine have been shown to manifest in various ways across different studies. Some studies suggest that caffeine can enhance motor performance. Despite this, numerous motor learning studies do not control for participants' caffeine intake or do not record whether participants consumed caffeine. The purpose of this study was to compare the motor learning performance (i.e., online improvement) of individuals who consumed caffeine and those who did not before practice with a novel motor sequence task. Sixty-five right-handed healthy undergraduate students participated in this study. Individuals reported their caffeine consumption through a pre-experiment questionnaire prior to performing an eight-item serial reaction time task. The participants who consumed caffeine reported drinking one cup of coffee on average 158.8 minutes before performing the sequential task. The motor performance of individuals who consumed caffeine was compared to those who did not, with the caffeine-consuming group (n = 13) exhibiting faster response times during online learning than the non-caffeine group (n = 52). The findings of this study suggest that caffeine should be controlled for in future motor learning research.

chrna3 modulates biphasic response to alcohol
Alcohol use disorders (AUDs) are complex phenomena governed by genetics, neurophysiology, environment, and societal structures. New methods to understand the underlying neurogenetics are valuable for designing better prevention and interventional strategies. Here, we used a novel, two-choice self-administration zebrafish assay (SAZA) to isolate the function of nicotinic acetylcholine receptor (nAChR) subunit alpha3 ( chrna3 ) in alcohol response. Juvenile zebrafish exhibited a biphasic response when self-administering alcohol that transitioned from attraction to aversion within minutes, suggesting they can regulate exposure to alcohol. This inverted U-shaped self-administration mirrored the effect alcohol has on shoaling behaviour. Exposure to low concentration of alcohol reduced anxiety-like behaviours characterised by decrease in shoal cohesion, while sedative effects became prominent at higher concentrations resulting in reduced locomotion and uncoordinated swimming. In contrast, chrna3 mutants displayed blunted responses, exhibiting prolonged alcohol self-administration, and increased gregariousness when compared to wild-type fish. These findings suggest that chrna3 dysfunction increased tolerance to alcohol. In humans, genetic variants that decrease CHRNA3 function may increase the likelihood of developing alcohol dependence. Thus, our results provide new insights into chrna3 function, as a potential target for pharmacological manipulation, and highlight the use of non-rodent alternatives to study the neurogenetics of development of AUD.

Indiv-Brain: Individualized Brain Network Partition Learned from Individual fMRI Data using Deep Clustering with Vertex-level Attention
Individualized functional network partitioning is increasingly critical in elucidating individual differences in cognition, development, and behavior. The previous studies grouped brain regions that were pre-defined by common brain parcellations into individualized brain networks. The employment of common brain region parcellations ignores the individual variations in brain structure and reduces the spatial resolution of the results. Moreover, some studies trained models on a group of subjects to guide individual network partitioning. These methods largely depend on the sample size and encounter challenges in the case of limited subject data. In this paper, we propose Indiv-Brain which automatically partition brain vertices into different brain networks based on individual fMRI data without any prior brain parcellation. The Indiv-Brain consists of three sequential modules: stacked denoising autoencoder (SDAE) for mapping the raw fMRI data to a latent embedding space, masked vertex modeling (MVM) for learning attention-enhanced representations of brain vertices, and deep embedding clustering with spatial attention (DEC-A) for unsupervised clustering on the learned vertex representations. The experiments on Human Connectome Project (HCP) demonstrate that the accuracy of Indiv-Brain outperforms existing methods. We compared the model with methods like SDAE, DEC, IDEC, and k-means. For the results of 8 subjects, the accuracy of Indiv-Brain was consistently the highest, averaging 6.65 percentage points higher than the IDEC method. To the best of our knowledge, this is the first study to obtain individualized brain network partition based on individual fMRI data with deep learning models. Our study provides a novel insight into understanding individualized brain networks, especially suitable for special patients.

Counting on AR: EEG responses to incongruent information with real-world context
Augmented Reality (AR) technologies enhance the real world by integrating contextual digital information about physical entities. However, inconsistencies between physical reality and digital augmentations, which may arise from errors in the visualized information or the user's mental context, can considerably impact user experience. This study characterizes the brain dynamics associated with processing incongruent information within an AR environment. We designed an interactive paradigm featuring the manipulation of a Rubik's cube serving as a physical referent. Congruent and incongruent information regarding the cube's current status was presented via symbolic (digits) and non-symbolic (graphs) stimuli, thus examining the impact of different means of data representation. The analysis of electroencephalographic (EEG) signals from 19 participants revealed the presence of centro-parietal N400 and P600 components following the processing of incongruent information, with significantly increased latencies for non-symbolic stimuli. Additionally, we explored the feasibility of exploiting incongruency effects for brain-computer interfaces. Hence, we implemented decoders using linear discriminant analysis, support vector machines, and EEGNet, achieving comparable performances with all methods. The successful decoding of incongruency-induced modulations can inform systems about the current mental state of users without making it explicit, aiming for more coherent and contextually appropriate AR interactions.

Carbonic anhydrase inhibitors prevent presymptomatic capillary flow disturbances in a model of cerebral amyloidosis
To enhance early diagnosis and treatment of Alzheimer{middle dot}s disease (AD), understanding the pathological changes before symptoms arise is crucial. The continuum model of AD suggest that A{beta} beta (A{beta}) accumulation precedes symptoms by at least 15 years, with vascular changes detectable around this time. Disturbances in capillary flow dynamics have been linked to reduced oxygen delivery to brain tissue, but evidence in presymptomatic AD remains elusive. We examined capillary flow dynamics in presymptomatic Tg-SwDI mice and the capacity of carbonic anhydrase inhibitors (CAIs) to prevent capillary flow disturbances. Our study revealed capillary flow disturbances associated with alterations in capillary morphology, adhesion molecule expression, and A{beta} load in cognitively normal 9-10-month-old Tg-SwDI mice. Treated mice showed ameliorated capillary flow disturbances, enhanced oxygen availability, and reduced A{beta} load. These findings underscore the importance of capillary flow disturbances in presymptomatic AD and highlight CAIs{middle dot} potential for preserving vascular integrity in early AD.

Speech Synthesis from Electrocorticogram During Imagined Speech Using a Transformer-Based Decoder and Pretrained Vocoder
This study describes speech synthesis from an Electrocorticogram (ECoG) during imagined speech. We aim to generate high-quality audio despite the limitations of available training data by employing a Transformer-based decoder and a pretrained vocoder. Specifically, we used a pre-trained neural vocoder, Parallel WaveGAN, to convert the log-mel spectrograms output by the Transformer decoder, which was trained on ECoG signals, into high-quality audio signals. In our experiments, using ECoG signals recorded from 13 participants, the synthesized speech from imagined speech achieved a dynamic time-warping (DTW) Pearson correlation ranging from 0.85 to 0.95. This high-quality speech synthesis can be attributed to the Transformer decoder's ability to accurately reconstruct high-fidelity log-mel spectrograms, demonstrating its effectiveness in dealing with limited training data.

Sex and APOE4-specific links between cardiometabolic risk factors and white matter alterations in individuals with a family history of Alzheimer's disease
INTRODUCTION: White matter (WM) alterations are among the earliest changes in Alzheimer's disease (AD), yet limited work has comprehensively characterized the effects of AD risk factors on WM. METHODS: In older adults with a family history of AD, we investigated the sex-specific and APOE genotype-related relationships between WM microstructure and risk factors. Multiple MRI-derived metrics were integrated using a multivariate approach based on the Mahalanobis distance (D2). The links between WM D2 and cognition were also explored. RESULTS: WM D2 in several regions was associated with high systolic blood pressure, BMI, and glycated hemoglobin, and low cholesterol, in both males and females. APOE4+ displayed a distinct risk pattern, with LDL-cholesterol having a detrimental effect only in carriers, and this pattern was linked to immediate memory performance. Myelination was the main mechanism underlying WM alterations. DISCUSSION: Our findings reveal that combined exposure to multiple cardiometabolic risk factors negatively impacts microstructural health, which may subsequently affect cognition. Notably, APOE4 carriers exhibited a different risk pattern, especially in the role of LDL, suggesting distinct underlying mechanisms in this group.

Generation of humanized mouse models to support therapeutic development for SYNGAP1 and STXBP1 disorders
Heterozygous variants in SYNGAP1 and STXBP1 lead to distinct neurodevelopmental disorders caused by haploinsufficient levels of post-synaptic SYNGAP1 and pre-synaptic STXBP1, which are critical for normal synaptic function. While several gene-targeted therapeutic approaches have proven efficacious in vitro, these often target regions of the human gene that are not conserved in rodents, hindering the pre-clinical development of these compounds and their transition to the clinic. To overcome this limitation, here we generate and characterize Syngap1 and Stxbp1 humanized mouse models in which we replaced the mouse Syngap1 and Stxbp1 gene, respectively, with the human counterpart, including regulatory and non-coding regions. Fully humanized Syngap1 mice present normal viability and can be successfully crossed with currently available Syngap1 haploinsufficiency mouse models to generate Syngap1 humanized haploinsufficient mice. Stxbp1 mice were successfully humanized, yet exhibit impaired viability (particularly males) and reduced STXBP1 protein abundance. Mouse viability could be improved by outcrossing this model to other mouse strains, while Stxbp1 humanized females and hybrid mice can be used to evaluate target engagement of human-specific therapeutics. Overall, these humanized mouse models represent a broadly available tool to further pre-clinical therapeutic development for SYNGAP1 and STXBP1 disorders.

The neural mechanisms of fast versus slow decision-making
Not all decisions are created equal; factors such as the difficulties or associated costs affect the time spent to make decisions. This is variously interpreted as speed/accuracy, fast/slow, or impulsivity/deliberateness tradeoffs according to different models of behaviour1-5. Regardless, it is generally assumed that decision latency reflects the neural mechanisms underlying behavioural strategy and cognitive investment. However, such investigations have been difficult in mice which are consistently impulsive. Here, we show that manipulating cost, using a novel floating-platform paradigm, overcomes the natural impulsivity of mice, more closely matching human behaviour. Furthermore, this approach allowed us simultaneously to measure the flow of activity from medial to lateral frontal cortex (MFC[->]LFC) and record sequences of single neuron activity with 2-photon imaging. Surprisingly, MFC display a different mode of operation, with high vulnerability to optical inhibition compared to LFC. Furthermore, the balance in choice coding at the beginning of sequences in MFC correlated with trial history and behavioural strategy. We found that for optimal performance, slow sequences in MFC showed declining numbers of active neurons whereas the opposite was true in LFC. Our results suggest that while LFC acts as an integrative motor threshold, MFC plays a larger cognitive role in the selection and timing of decisions than previously thought. Our study offers a methodological and mechanistic framework in mouse frontal cortex to understand the neural basis of voluntary decision making.

Loss of Insulin Signaling in Microglia Impairs Cellular Uptake of Aβ and Neuroinflammatory Response Exacerbating Alzheimer-like Neuropathology
Insulin receptors are present on cells throughout the body, including the brain. Dysregulation of insulin signaling in neurons and astrocytes has been implicated in altered mood, cognition, and the pathogenesis of Alzheimers disease (AD). To define the role of insulin signaling in microglia, the primary phagocytes in brain critical for maintenance and damage repair, we created mice with an inducible microglia-specific insulin receptor knockout (MG-IRKO). RiboTag profiling of microglial mRNAs revealed that loss of insulin signaling results in alterations of gene expression in pathways related to innate immunity and cellular metabolism. In vitro, loss of insulin signaling in microglia results in metabolic reprograming with an increase in glycolysis and impaired uptake of A{beta}. In vivo, MG-IRKO mice exhibit alterations in mood and social behavior, and when crossed with the 5xFAD mouse model of AD, the resultant mice exhibit increased levels of A{beta}; plaque and elevated neuroinflammation. Thus, insulin signaling in microglia plays a key role in microglial cellular metabolism, neuroinflammation and the ability of the cells to take up A{beta}; such that reduced insulin signaling in microglia alters mood and social behavior and accelerates AD pathogenesis. Together these data indicate key roles of insulin action in microglia and the potential of targeting insulin signaling in microglia in treatment of AD.

The preference for surprise in reinforcement learning underlies the differences in developmental changes in risk preference between autistic and neurotypical youth
Risk preference changes nonlinearly across development. Although extensive developmental research on the neurotypical population has shown that risk preference is highest during adolescence, developmental changes in risk preference in autistic people, who tend to prefer predictable behaviors, have not been investigated. Here, we aimed to investigate these changes and underlying computational mechanisms. Using a game-like risk-sensitive reinforcement learning task, we found a significant difference in nonlinear developmental changes in risk preference between the autistic and neurotypical groups (N = 75; age range, 6-30 years). The computational modeling approach with reinforcement learning models revealed that individual preferences for surprise modulated such preferences. These findings indicate that for neurotypical people, adolescence is a developmental period involving risk preference, possibly due to lower surprise aversion. Conversely, for autistic people, who show opposite developmental trajectories of risk preference, adolescence could be a developmental period involving risk avoidance because of low surprise preference.

Reply to: Eigenmodes of the brain: revisiting connectomics and geometry
In Pang, Aquino et al. (2023), we presented multiple lines of evidence to indicate that brain geometry plays a previously under-appreciated role in shaping dynamics. Mansour et al. raise concerns about one specific analysis, in which we showed that eigenmodes derived from the geometry of the human cortex can reconstruct diverse activity maps generated with functional magnetic resonance imaging (fMRI) better than eigenmodes derived from connectomes estimated with diffusion MRI (dMRI). Here, we address their concerns and show how our findings and conclusions remain valid.

Social risk coding by amygdala activity and connectivity with dorsal anterior cingulate cortex
Risk is a fundamental factor affecting individual and social economic decisions, but its neural correlates are largely unexplored in the social domain. The amygdala, together with the dorsal anterior cingulate cortex (dACC), is thought to play a central role in risk taking. Here, we investigated in human volunteers (n=20; 11 females) how risk (defined as variance of reward probability distributions) in a social situation affects decisions and concomitant neural activity as measured with fMRI. We found social variance-risk signals in the amygdala. Activity in lateral parts of the amygdala increased parametrically with social reward variance of the presented options. Behaviorally, 75% of participants were averse to social risk as estimated in a Becker-DeGroot-Marschak auction-like procedure. The stronger this aversion, the more negative was the coupling between risk-related amygdala regions and dACC. This negative relation was significant for social risk attitude but not for the attitude towards variance-risk in juice outcomes. Our results indicate that the amygdala and its coupling with dACC process objective and subjectively evaluated social risk. Moreover, while social risk can be captured with a framework originally established by finance theory for individual risk, the amygdala appears to processes social risk largely separately from individual risk.

Broadband high frequency Activity initializes Distractor Suppression
Selective attention requires fast and accurate distractor suppression. We investigated if broadband high frequency activity (BHA; 80 to 150 Hz), indicative of local neuronal population dynamics in early sensory cortices, indexes rapid processing of distracting information. In the first experiment we tested whether BHA distinguishes targets from distracting information in a visual search paradigm using tilted gratings as targets and distractors. In the second experiment, we examined whether BHA distractor processing can be trained by statistical learning. In both experiments, BHA preceded the low-frequency target enhancement (NT) and distractor suppression (PD; 1 to 40 Hz) components and distinguished between targets and distractors. Only the BHA but not low-frequency component amplitude correlated with participants performance and was higher for lateral distractors versus lateral targets. Furthermore, BHA predicted the strength of the PD. These results indicate that BHA initiates stimulus discrimination via distractor suppression.

Nageotte nodules in human DRG reveal neurodegeneration in painful diabetic neuropathy
Diabetic neuropathy is frequently accompanied by pain and loss of sensation attributed to axonal dieback. We recovered dorsal root ganglia (DRGs) from 90 organ donors, 19 of whom had medical indices for diabetic painful neuropathy (DPN). Nageotte nodules, dead sensory neurons engulfed by non-neuronal cells, were abundant in DPN DRGs and accounted for 25% of all neurons. Peripherin-and Nav1.7-positive dystrophic axons invaded Nageotte nodules, forming small neuroma-like structures. Using histology and spatial sequencing, we demonstrate that Nageotte nodules are mainly composed of satellite glia and non-myelinating Schwann cells that express SPP1 and are intertwined with sprouting sensory axons originating from neighboring neurons. Our findings solve a 100-year mystery of the nature of Nageotte nodules linking these pathological structures to pain and sensory loss in DPN.

Anxiety and risk-taking behavior maps onto opioid and alcohol polysubstance consumption patterns in male and female mice
Polysubstance use is prevalent in the population but remains understudied in preclinical models. Alcohol and opioid polysubstance use is associated with negative outcomes, worse treatment prognosis, and higher overdose risk; but underlying mechanisms are still being uncovered. Examining factors that motivate use of one substance over another in different contexts in preclinical models will better our understanding of polysubstance use and improve translational value. Here we assessed baseline anxiety like and locomotive behavior and then measured voluntary consumption of multiple doses of alcohol and fentanyl in group housed male and female mice using our novel Socially Integrated Polysubstance (SIP) system. Fifty six male (n=32) and female (n=24) adult mice were housed in groups of 4 for one week with continuous access to food, water, two doses of ethanol (5% and 10%) and two doses of fentanyl (5 ug/ml and 20 ug/ml). Our analyses revealed sex differences across multiple domains - female mice consumed more liquid in the dark cycle, had higher activity, a higher preference for both ethanol and fentanyl over water, and their fentanyl preference increased over the seven days. We then used machine-learning techniques to reveal underlying relationships between baseline behavioral phenotypes and subsequent polysubstance consumption patterns, where anxiety- and risk taking-like behavioral phenotypes mapped onto discrete patterns of polysubstance use, preference, and escalation. By simulating more translationally relevant substance use and improving our understanding of the motivations for different patterns of consumption, this study contributes to the developing preclinical literature on polysubstance use with the goal of facilitating better treatment outcomes and novel therapeutic strategies.

The Impact of Ketamine and Thiopental Anesthesia onUltraweak Photon Emission and Oxidative-Nitrosative Stress in Rat Brains
Anesthetics such as ketamine and thiopental, commonly used for inducing unconsciousness, have distinct effects on neuronal activity, metabolism, and cardiovascular and respiratory systems. Ketamine increases heart rate and blood pressure while preserving respiratory function, whereas thiopental decreases both and can cause respiratory depression. This study investigates the impact of ketamine (100 mg/kg) and thiopental (45 mg/kg) on ultraweak photon emission (UPE), oxidative-nitrosative stress, and antioxidant capacity in isolated rat brains. To our knowledge, no previous study has investigated and compared UPE in the presence and absence of anesthesia. Here, we compare the effects of ketamine and thiopental anesthetics with each other and with a non-anesthetized control group. Ketamine increased UPE, lipid peroxidation, and antioxidant enzyme activity while reducing thiol levels. Conversely, thiopental decreased UPE, oxidative markers, and antioxidant enzyme activity, while increasing thiol levels. UPE was negatively correlated with thiol levels and positively correlated with oxidative stress markers. These findings suggest that the contrasting effects of ketamine and thiopental on UPE are linked to their differing impacts on brain oxidative stress and antioxidant capacity. This research suggests a potential method to monitor brain oxidative stress via UPE during anesthesia, and opens up new ways for understanding and managing anesthetic effects.

On the same wavelength: The relationship between neural synchrony and cognitive ability during movie watching in late childhood and early adolescence
Development during late childhood and early adolescence is associated with vast improvements to thinking and reasoning abilities. Coinciding with developing cognitive abilities, the environments children navigate become more complex, with an expanding social circle giving rise to richer and more elaborate experiences. If cognitive development is associated with more complex experiences, do individual differences in cognitive abilities influence how children and adolescents experience their world? To address this question, we investigated the relationships between intersubject correlation (ISC) during movie watching and cognitive scores in children and adolescents (aged 7 -15, N = 309). Data for the current study was obtained from the Healthy Brain Biobank. As part of the Healthy Brain Biobank protocol, participants watched a 10-minute clip of Despicable Me while in the fMRI scanner. They also completed the Weschler Intelligent Scale for Children (WISC). We compared the degree to which participants brain activity synchronized to other participants brain activity during the movie, as measured by ISC, and investigated if scores on the WISC predicted this synchrony. We found adolescents (ages 11-15) with higher cognitive scores showed greater ISC during movie watching in brain networks associated with social processing and executive functions compared to those with below average cognitive scores. These networks included the frontoparietal and default mode networks. This pattern was not evident in children (ages 7-11) who differed in their cognitive scores. These results suggest that adolescents with more mature cognitive abilities may have more similar experiences of naturalistic stimuli. Our results also suggest that children may be less reliant on the frontoparietal network when processing movies, compared to previous findings in adults.

Rapid Changes in Risk Preferences Originate from Bayesian Inference on Parietal Magnitude Representations
Risk preferences - the willingness to accept greater uncertainty to achieve larger potential rewards - determine many aspects of our lives and are often interpreted as an individual trait that reflects a general 'taste' for risk. However, this perspective cannot explain why risk preferences can change considerably across contexts and even across repetitions of the identical decisions. Here we provide modelling and neural evidence that contextual shifts and moment-to-moment fluctuations in risk preferences can emerge mechanistically from Bayesian inference on noisy magnitude representations in parietal cortex. Our participants underwent fMRI while choosing between safe and risky options that were either held in working memory or present on the screen. Risky options that were held in working memory were less likely to be chosen (risk aversion) when they had large payoffs but more likely to be chosen (risk-seeking) when they had small payoffs. These counterintuitive effects are mechanistically explained by a computational model of the Bayesian inference underlying the perception of the payoff magnitudes: Options kept in working memory are noisier and therefore more prone to central tendency biases, leading small (or large) payoffs to be overestimated (or underestimated) more. Congruent with the behavioural modelling, fMRI population-receptive field modelling showed that on trials where intraparietal payoff representations were noisier, choices were also less consistent and less risk-neutral, in line with participants resorting more to their prior belief about potential payoffs. Our results highlight that individual risk preferences and their puzzling changes across contexts and choice repetitions are mechanistically rooted in perceptual inference on noisy parietal magnitude representations, with profound implications for economic, psychological, and neuroscience theories of risky behaviour.

Bidirectional locomotion induces unilateral limb adaptations
Humans can acquire and maintain motor skills throughout their lives through motor learning. Motor learning and skill acquisition are essential for rehabilitation following neurological disease or injury. Adaptation, the initial stage of motor learning, involves short-term changes in motor performance in response to a new demand in the person's environment. Repeated adaptation can improve skill performance and result in long-term skill retention. Locomotor adaptation is extensively studied using split-belt treadmill paradigms. In this study we explored whether bidirectional walking (BDW) on a split-belt treadmill can induce short-term gait adaptations. Twelve healthy volunteers participated in our single session, starting with 2 minutes of normal walking (NW), followed by four 5-minute blocks of BDW with a 1-minute passive rest in between blocks, and ending with another 2-minute of NW. We recorded body kinematics and ground reaction forces throughout the experiment. Participants quickly adapted to BDW with both legs showing decreased step lengths. However, only the backward-walking leg exhibited aftereffects upon returning to NW, indicating short-term adaptation. Notable kinematic changes were observed, particularly in hip extension and pelvis tilt, though these varied among participants. Our findings suggest that BDW induces unilateral adaptations despite bilateral changes in gait, offering new insights into locomotor control and spinal CPG organization.

Unveiling microglial heterogeneity from single-cell transcriptomics in neurodegenerative diseases
Microglia are key players in maintaining brain homeostasis and responding to pathological conditions. Their multifaceted roles in health and disease have garnered significant attention in the context of neurodegeneration. In recent years, single-cell transcriptomic techniques have provided unprecedented insights into microglial heterogeneity, revealing distinct subpopulations and gene expression patterns associated with neuroprotection or neurotoxicity. Here, the transcriptomic landscape of microglia has been dissected by leveraging human single-nuclei RNA sequencing datasets of neurodegenerative conditions, encompassing amyotrophic lateral sclerosis, frontotemporal dementia, Alzheimer's disease, aging, and Parkinson's disease. Results have led to the identification of distinct cell subpopulations, representative of the functional heterogeneity of the brain microglia. Moreover, distinct gene signatures and regulatory networks linked to inflammation and neurodegeneration have been identified. Overall, the study provides an improved portrait of microglia in the context of neurodegenerative disorders, and it holds promise for developing a more targeted research aimed at modulating microglial function to mitigate disease progression and foster neuroprotection.

Detecting behavioural oscillations with increased sensitivity: A modification of Brookshire's (2022) AR-surrogate method
A core challenge of cognitive neuroscience is to understand how cognition changes over time within the same individual. For example, the tendency for behavioural responses in a range of cognitive domains to oscillate over time has been studied extensively. Recently, however, the phenomenon of behavioural oscillations has been called into question by indications that past findings might reflect aperiodic temporal structure rather than true oscillations. Brookshire (2022) proposed methods to control for aperiodic temporal structure while examining oscillations in behavioural time-courses and found no evidence of behavioural oscillations in reanalyses of four published datasets. However, Brookshire's (2022) method has been criticised for having low sensitivity to detect effects of realistic magnitude, so it is currently unclear whether these findings suggest that behavioural oscillations are not present in these and perhaps many other datasets, or whether they are false negatives. Here, we present a modification of Brookshire's (2022) AR-surrogate method with increased sensitivity to detect effects of realistic magnitude, adequate control of false positives, and other desirable properties such as the ability to increase statistical power by adding more participants. Using this method, we reanalyse the same publicly available datasets and show significant behavioural oscillations in each of them, suggesting oscillations in behaviour are a robust phenomenon upon which to draw theoretical inferences. The participant-level AR-surrogate method is currently the most sensitive method available for analysing behavioural oscillations while controlling for the contribution of aperiodic data fluctuations.

Boundaries in the eyes: measure event segmentation during naturalistic video watching using eye tracking
During naturalistic information processing, individuals spontaneously segment their continuous experiences into discrete events, a phenomenon known as event segmentation. Traditional methods for assessing this process, which include subjective reports and neuroimaging techniques, often disrupt real-time segmentation or are costly and time intensive. Our study investigated the potential of measuring event segmentation by recording and analyzing eye movements while participants view naturalistic videos. We collected eye movement data from healthy young adults as they watched commercial films (N=104), or online Science, Technology, Engineering, and Mathematics (STEM) educational courses (N=44). We analyzed changes in pupil size and eye movement speed near event boundaries and employed inter-subject correlation analysis (ISC) and hidden Markov models (HMM) to identify patterns indicative of event segmentation. We observed that both the speed of eye movements and pupil size dynamically responded to event boundaries, exhibiting heightened sensitivity to high-strength boundaries. Our analyses further revealed that event boundaries synchronized eye movements across participants. These boundaries, can be effectively identified by HMM, yielded higher within-event similarity values and aligned with human-annotated boundaries. Importantly, HMM-based event segmentation metrics responded to experimental manipulations and predicted learning outcomes. This study provided a comprehensive computational framework for measuring event segmentation using eye-tracking. With the widespread accessibility of low-cost eye-tracking devices, the ability to measure event segmentation from eye movement data promises to deepen our understanding of this process in diverse real-world settings.