Quantitative Biology (Neurons and Cognition)

ALT Energy consumption is a major concern for modern artificial intelligence (AI) applications. The brain is thought to achieve energy-efficient computation through sparse coding schemes. By emulating the spiking dynamics of the brain, neuromorphic chips hold the potential to realize energy-efficient AI. In this work, we use birdsong to demonstrate that auditory object recognition can be achieved using ultra-sparse coding. Feature-detecting neurons tuned to short temporal features (∼37 ms), sampled from exemplar syllables, transform continuous auditory signals into streams of ultra-sparse spikes that mark the onsets and offsets of features within the signal. Spike patterns elicited by the exemplars are stored as templates. During recognition, template matching is performed using axonal delays between feature-detecting and template-detecting neurons. Recognition is further enhanced by an additional group of specialized binary neurons trained to discriminate between syllable categories with

ALT Brains do not simply add up every signal they receive: they select which routes dominate. Yet connectomes are usually analyzed with tools that sum over all paths and place neural states in Euclidean spaces, blurring the feedback circuits that may sustain internal states. Here we treat neural signaling as routing instead of diffusion and describe it with max-plus algebra, in which signals accumulate along a path and the strongest route prevails. We propose a model in which the functional connectome of Caenorhabditis elegans emerges from the time-varying combination of electrical, synaptic, monoaminergic, neuropeptidergic and hormonal layers, and analyze five networks within this framework. Each collapses onto a handful of dominant feedback loops, which recover well-characterized circuits and point to new candidates for experimental study. Unlike linear spectra, these structures keep the biological meaning of the data, and their order-based description suggests that brain states live in

ALT To function effectively in a changing world, agents must detect changes and act on them in a timely fashion. We study this in the context of a change point detection task in which clicks arrive from a Poisson process whose rate changes at an unknown time. An agent must declare the change within a detection window of fixed width or the trial ends in a miss. Here we prove a window sufficiency theorem, which states that the posterior probability of a change depends only on the clicks within the most recent window, regardless of the prior on the change time. This result follows from combining survival conditioning, which rules out change times whose windows have already expired, with cancellations driven by independent increments of a known baseline process. Window sufficiency generalizes to an unknown post-change process, as long as the baseline process is known. For an exponential prior on the change time, we further derive the posterior in closed form as a function of the elapsed times

ALT Predictive Coding (PC) is an influential account of how brain representations emerge through hierarchical, local, self-supervised prediction-based learning, without requiring globally propagated error signals. Despite extensive evidence supporting PC, a striking paradox persists: computational models built on these biologically grounded principles have consistently failed to match state-of-the-art Deep Neural Networks in performance and brain-alignment. Here, we show that this paradox can be overcome. We introduce Predictive Brain Networks (PBNs), a hierarchical recurrent framework that implements local PC-based learning and achieves representational alignment with the human ventral visual stream matching or exceeding established vision models. Importantly, PBN top-down predictions encode unique, brain-aligned information not captured during bottom-up feature extraction, with this contribution increasing in locally trained variants. Further, PBNs substantially outperform existing PC-in

ALT The Wilson-Cowan model is a classic example of a “neural mass model” - a simplified yet powerful way to describe the activity of large networks of neurons. It is particularly interesting because, even in its relatively simple form, it can produce a surprising variety of behaviours, from steady rhythms to complex oscillations. A key factor influencing these dynamics is the presence of time delays, reflecting the time it takes for signals to travel between neurons. This article offers an accessible introduction to the Wilson-Cowan model, focusing on how these time delays affect its behaviour. We explore the core concepts behind the model, survey the existing literature, and introduce some numerical techniques specifically designed for analysing such delay differential equations. We also provide a practical, hands-on tutorial on using the DDE-BIFTOOL software package, demonstrating its application to the Wilson-Cowan model. Downloadable code allows the reader to reproduce our results and

ALT Individual neurons exhibit diverse molecular, morphological, and electrophysiological properties that shape their distinct roles in neural circuits. While these multimodal features are largely shared across individuals, they remain an underutilized source of information in analyzing neural population dynamics. We posit that these shared biological profiles provide a common reference for relating neural dynamics across subjects and recording sessions. Our framework leverages this structure by routing neurons into latent groups based on their biological profiles and analyzing population dynamics through group-level representations. Because these assignments are anchored in biological properties that are consistent across subjects, the learned organization supports transfer to unseen animals without retraining or neuron-wise correspondence. We evaluate our framework on whole-brain calcium imaging in larval zebrafish and large-scale electrophysiology in the mouse visual cortex. Across both

ALT Cognition is implemented by distributed neural populations whose collective activity evolves through structured, low-dimensional state spaces. In this review, we survey the analytical and conceptual tools used to study latent population dynamics in biological neural networks. We begin with single-unit coding, which remains essential for interpreting neuronal selectivity, and then review population-level approaches, including population vectors, cosine similarity, principal component analysis and its mnemonic variants, multivariable linear regression axes, demixed principal component analysis, and single-trial latent-variable models, and distinguish operationally defined population subspaces from neural manifolds with reproducible geometry or topology. We use these methods to discuss sensory coding, working memory, decision-making, temporal processing, and sensorimotor transformations. Two organizing principles are emphasized throughout. First, cognitive variables can be expressed in se

ALT Neurobiological circuits may be energy efficient in part because there is a division of labor: different subsystems compute or represent different things. In light of this, it is interesting that a major difference between subsystems is the time scale on which they typically vary. Here, we argue that temporal smoothness constraints—the idea that it may be energetically costly for systems to vary much more quickly than their typical time scale—imply a certain division of labor with respect to temporal stimulus features. In particular, slow subsystems ought to represent slowly varying stimulus features, and fast subsystems ought to represent quickly varying features. We formulate a novel efficient coding model that we use to investigate this claim, and exactly solve for the optimal division of labor. In addition to finding that slower subsystems ought to represent more slowly varying temporal features, we find surprising structure of mathematical interest: optimal codes have subsystems w

ALT Learning algorithms are usually analyzed under the assumption that informative gradients are continuously available. But this is not always true—learning signals in biological neural networks, for example, may be sparse, local, and only intermittently available at any given synapse. In this work, we show that if gradient information is noisy and only intermittently available, learning may be unstable. In particular, even if average learning dynamics converge to the correct optimum, variance in learning may be asymptotically infinite. We explore this point in a number of analytically tractable settings, including single-layer and multi-layer linear networks, and derive explicit conditions under which learning is stable. These conditions complement, but are conceptually distinct from, existing conditions related to the stability of stochastic gradient descent. Our work demonstrates that unbiased credit assignment is not enough for reliable learning; sufficiently frequent and informative

ALT Networks of delay-coupled neural mass models alternate between synchrony and anti-phase locking as the delay between brain regions increases. The origin of these delay-dependent bands has remained unclear. Here, we show that a single function, the phase interaction function H of an isolated node, predicts where the bands occur and explains why they repeat. For a Wilson–Cowan model with intra-nodal delay τ and inter-nodal delay ρ, we compute the limit cycle and phase response, and hence H, by harmonic balance. Each band boundary is a zero of H' shifted by the phase lag Ωρ, so the bands repeat in ρ with the oscillation period. Simulations of the full delay-differential networks confirm these predictions, from two nodes, where predicted frequencies agree to second order in the coupling strength, to rings, where the theory also identifies coexisting stable twisted states that random initial conditions rarely reach. At weak coupling, the delays at which synchrony gains or loses stability ar

ALT Altered resting-state functional connectivity (FC) is a candidate marker for the diagnosis of Parkinson's disease (PD). Publicly available PD datasets, however, are small and fragmented across sites, scanners and clinical criteria. Transfer learning mitigates data scarcity by reusing a model trained on larger cohorts, but represents their knowledge primarily through a single set of weights shared by all individuals. Motivated by precision neuroimaging, we propose a retrieval-augmented classification (RAC) framework that additionally places each target individual in the context of similar individuals from external cohorts. A graph convolutional autoencoder (GAE), trained on external PD and control cohorts, encodes these cohorts into a neuro-knowledge base. Each target individual is encoded using the pretrained GAE, and its nearest neighbors are retrieved from the external knowledge base. Their embeddings are combined with the target embedding using mean aggregation or attention for clas

ALT Animals have a remarkable capacity for autonomy—driven by intrinsic motivation—without external sensory cues or extrinsic objectives, yet the computational mechanisms behind this ability largely remain a mystery. We combine whole-brain recordings in the fruit fly (Drosophila melanogaster), which reveal neural representations of spontaneous exploratory decisions more than 10 seconds in advance, with a neuromechanical digital twin to test whether optimizing intrinsic goals yields realistically embodied artificial agents that quantitatively match fly behavior and internal representations of future volitional action during exploration. Our hierarchical policy selects from a discrete library of low-level motor primitives, consistent with ethological theories of behavioral syllables. We train agents without external reward using 20 intrinsic motivation algorithms spanning curiosity, entropy, and competency methods. The agents are evaluated in a virtual tethered-ball assay modeled on the fly

ALT Identifying prodromal Parkinson's Disease is a precondition for any future treatment that modifies the course of the disease, since such therapies would ideally act on the initial pathogenic events, early enough to delay the clinical onset of the disease rather than only slow its progression. In this work, a visual exploration model is retrained on an expanded corpus. Without any further training or threshold recalibration, it is then tested on an enriched cohort of patients with high risk of being in the prodromal phase of Parkinson's Disease, testing whether the learned signal follows a continuum of disease rather than only separating patients from controls. The resulting score distribution is analysed as an external evaluation of the hypothesis that the oculomotor signature targeted by the model is already partly present before clinical diagnosis. With every subject scored on data not used to train the model, the scores are ordered as controls < high-risk patients < Parkinson's Dise

ALT The flip tilt illusion is the misperception of the orientation of the alignment between a pair of dots by 90 degrees. It occurs when the dot-pair contains one black dot and one white dot, is typically visible only in the peripheral visual field. It is predicted by the central-peripheral dichotomy (CPD) theory. Along the visual pathway from retina to downstream stages, our brain admits only a tiny fraction of visual input information for recognition due to a processing bottleneck that starts from the output of the primary visual cortex (V1) along the visual pathway. Accordingly, perception downstream from V1 relies on impoverished information, selected largely by gaze shifts to center the selected content at fovea and additionally by the capacity of the bottleneck even in the central visual field, likely making perceptual hypotheses about the visual scene ambiguous or illusory. To overcome, downstream areas send feedback to upstream areas such as V1 to query for additional information,

ALT The head-direction (HD) system is a neuronal compass that represents head orientation in 3D. Its encoding of 3D head orientation can be decomposed into a multiplexed representation of 2D head tilt relative to gravity and 1D head orientation in the horizontal plane (azimuth). Defining azimuth during 3D motion is not trivial, and recent studies have indicated that the HD system implements a so-called dual-axis/tilted-azimuth (DA/TA) rule, a representation that is well supported experimentally but remains mathematically and mechanistically difficult to interpret. Here, I introduce a new formulation of DA/TA based directly on the elements of a rotation matrix describing 3D head orientation. In this formulation, TA is represented by a 2D vector, whose direction encode TA and whose norm depends on tilt, consistent with a tilt-dependent gain of HD tuning reported by experimental studies. I provide a geometrical interpretation of this formulation and use it to characterize the topology of the

ALT Connectomes now map the neurons and synapses of an adult fly's brain and nerve cord, but how much of behaviour follows from the wiring is unknown. Here we turn one such connectome into a spiking network of 172,433 neurons with a published brain model's untuned parameters, and run it alone and inside a simulated body. Two fixed rules join them: odours and looming objects drive sensory neurons, and motor neurons move the legs. Alone, the network passes a signal to the motor neurons it is known to drive when the pathway supplies about 2% or more of their input. In the body, the fly navigates around obstacles to 95% of the odour sources. A rewiring that keeps each neuron's connection counts and signs reaches 18%, and the fly's leg commands replayed with the senses off reach almost none, so navigation needs the measured wiring and sensory feedback. Dopamine-gated weakening of the mushroom body's output synapses turns approach to an odour into avoidance, which requires the pairing and is odo

ALT Binocular video-oculography records separate signals for each eye, yet human visual processing and oculomotor control operate relative to a single, virtual cyclopean eye. This study addresses the estimation of the cyclopean eye during binocular free visual exploration, a target-free paradigm widely used for the screening of neurological disorders. Using eye-tracking data collected at 1 kHz from 113 participants across three cohorts (patients with Parkinson's Disease, age-matched healthy controls, and young controls), we compare two different methods: one based on an Independent Component Analysis and another based on Factor Analysis. The extracted cyclopean signals were evaluated on their ability to preserve clinically critical fixation and saccadic events using both velocity-threshold and pixel-dispersion event detectors, and were validated against references via Dynamic Time Warping. Results suggest that the technique based on Factor Analysis guarantees full identifiability across al

ALT Hierarchical and modular organization defines brain architecture, yet how it supports flexible computation and functional specialization remains unclear. Although dynamic range is a hallmark of efficient information processing, its distribution across modules and hierarchical levels remains largely unexplored. Here, we show that hierarchical-modular networks naturally generate heterogeneous modular dynamics and firing rates, extending regimes of critical-like dynamics with enhanced intrinsic timescales, reflecting critical slowing down, and diverse scaling exponents without fine-tuning near criticality. Intrinsic timescales increase across hierarchical levels, while their dispersion within levels emerges only when connectivity profiles and firing rates are heterogeneous. We introduce modular dynamic range to quantify how individual modules respond to external inputs and show that heterogeneity enhances both global and module-specific sensitivity, with modular intrinsic timescales incre

ALT Network neuroscience has traditionally relied on node-centric approaches, which are inherently limited in capturing the brain's overlapping functional architectures and higher-order edge-to-edge interactions. The emerging edge-centric paradigm addresses this gap by shifting the analytical unit from nodes to edges. However, while edge-centric models have been established for DTI and fMRI, comparable EEG methodologies have yet to be developed. In this study, we proposed a Phase-based Edge-centric Coupling (PEC) algorithm for EEG network modelling, where edge time series are derived from instantaneous phase differences of EEG signals and edge-to-edge coupling is quantified via normalized inner product. Simulation tests demonstrated the robustness of PEC algorithm, achieving high accuracy (<10% relative error) in coupling estimation and discriminative capacity (AUROC>0.7) against spurious coupling.When provided sufficient data length(>=16 cycles), PEC outperformed traditional Phase Node-ce

ALT We study the dynamics of recurrent neural networks with random couplings, sublinear odd activation functions, and temporally correlated noise. Using large N methods, we determine stationary correlations and distinguish the loss of local linear stability from the onset of chaos, characterized by a positive largest Lyapunov exponent. These thresholds delimit an intermediate regime in which local instability coexists with asymptotic stability under a common noise history. Their separation depends on both the noise correlation time and the asymptotic growth of the activation function. We show that internal noise with any finite correlation time cuts off the divergence of the network's intrinsic correlation time taucorr at the edge of chaos; it exhibits a critical scaling taucorr,max propto 1/sqrtD₀, where D₀ is the integrated noise strength. Static random input instead produces coincident thresholds and a divergent correlation time. We also calculate the frequency-dependent participation d