Neural Computation

This paper introduces a sensorimotor learning framework for a corticocerebellar network, grounded in the perspective of population dynamics. Using an optimal control theory approach, the cerebellum model enhances preparatory activity through premotor input, allowing the motor cortex to reach the desired initial conditions for movement more efficiently. Unlike traditional motor learning approaches that focus on acquiring new skills, this paradigm emphasizes automatization of already executable behaviors through repetition driven by intrinsic motivation. The proposed model is evaluated using a center-out reaching task, demonstrating that the role of the cerebellum is to shorten the preparatory period required for the successful execution of the movement. These findings suggest that corticocerebellar interactions play a crucial role in optimizing motor preparation, offering insight into the neural mechanisms underlying movement efficiency.

A novel mathematical framework to define the threshold of action potentials in excitable cells is presented. Unlike previously applied methods that rely on approximations or bifurcations, the approach focuses on the geometry of membrane potential trajectories. The changes in concavity during the upstroke of an action potential can be directly obtained from a time series of voltages. The concavity criterion is then extended to models based on autonomous dynamical systems where the changes in concavity can be obtained analytically from a curve of inflection points in phase space. The inflection point manifold defines a region required for excitability: all the orbits that cross it contain action potentials, and all the trajectories that contain action potentials are in it. This analytical principle can then be used to define excitability in a dynamical system, and also a measure of excitability that enables quantification and comparisons of excitability across dynamical system. The measure provides a way to compare the excitabilities of systems that model neurons with different electrophysiological phenotypes and consider different stimulus conditions. The traditionally vague physiological concept of electrical excitability is transformed into a rigorous analytical description by considering the time-dependent curvature of the membrane potential. The criterion is robust across smooth, single compartment models of electrical excitability and can be can be extended to single compartment models in higher dimensions, and multicompartment models as well.

Flexibly recombining computational modules is essential for biological and artificial neural networks to rapidly adapt to changing environments. This requires modules to be shared across tasks rather than rigidly segregated, yet what determines this organization remains unknown. Previous work suggests that weight initialization shapes whether networks learn task-specific or generic representations, but it is unclear whether this extends to recurrent networks and, more importantly, to network connectivity. Here, we systematically vary the initial weight variance of recurrent neural networks and study them using a framework that allows us to identify the functionally relevant connectivity subspaces for each computational module. We find that networks with low initial weight variance converge to solutions in which different subtasks rely on largely overlapping weight subspaces, whereas high-variance networks implement subtasks in higher-dimensional, more segregated weight subspaces. Our results also provide mechanistic insights with implications for interpreting biological neural circuits and for designing efficient recurrent architectures.

Action-selection is determined by a combination of goal-directed and habitual processes. Habits are defined as the reward-independent, stimulus-response relationships which form when an action is regularly executed in the same context, regardless of outcome. An influential computational model proposes that habit formation is driven by action prediction errors which occur when non-habitual actions are taken. It has been further suggested that action prediction errors are encoded in activity of specific dopamine neurons, and it has been recently observed that dopamine activity in the tail of the striatum follows a pattern consistent with the action prediction errors. However, the original models capture changes in habits across trials, but do not describe the time-course of action prediction errors within trials, hence it is difficult to directly compare them with dopamine activity. We begin by outlining the temporal-difference action learning algorithm, which uses biologically-plausible mechanisms to determine how dynamic changes in action intensity influence the resultant prediction errors across near-continuous time. We then demonstrate that dopaminergic data recently collected from the tail of the striatum is better represented by action prediction errors than reward prediction errors. Overall, our results support the existence of value-free action prediction errors and associated habitual behaviour in dopaminergic signals. Author summaryWhenever we choose one action over another, there are two ways that the selection can be made. We could take the time to consider what we want to achieve, calculate which action is the most likely to give us that outcome and balance it against the possible negative consequences. These goal-directed calculations are very time-consuming and our brains could not possibly do it for every choice. Instead, we often rely on the second method, habits, which learn to copy the actions that were most often chosen in the past. In this paper, we present a new model of learning that is based on biologically plausible brain networks and applies action prediction errors to update our habits across continuous time. Using simulations, we reveal testable predictions that are specific to our temporal-difference action learning model and build an intuition for its behaviour. Finally, this model is tested against real dopaminergic data from the tail of the striatum, and we show that our model provides better explanation for these data, than classic reward-based reinforcement learning models.

Determining the amount of information a neuron receives per unit of time is key to understanding brain connectivity and how neural networks encode and transmit information. In particular, estimating this information flow by distinguishing between excitatory and inhibitory synaptic contributions is critical to understanding neural network function, as maintaining the excitation-inhibition (E/I) balance regulates neuronal excitability and circuit stability, whereas its disruption can lead to a plethora of brain disorders, including neurodegenerative and psychiatric conditions. However, because synaptic conductances cannot be measured directly, inverse methods are required to infer them from the membrane potential --a readily measurable quantity. Although partial solutions have been proposed, accurately estimating these conductances remains a significant challenge due to the complexity and diversity of the inputs. This is particularly true in the spiking regime, where neurons actively fire. In this work, we introduce a novel computational strategy that combines two critical metrics extracted from the time course of the membrane potential recording: the amplitude of the spike and the interspike interval. By using these quantities, the proposed method enables the accurate separation of excitatory and inhibitory contributions, yielding highly favorable results in the spiking regime. Author summaryQuantifying the continuous stream of inputs a neuron receives is key to understanding brain connectivity. Inside the brain, individual cells must maintain a tight balance between excitation and inhibition (E/I) to process information correctly, as any disruption in this equilibrium can impair its functionality. However, directly measuring the underlying excitatory and inhibitory synaptic conductances is technically challenging, and existing mathematical tools often fail when neurons enter their active firing regime. In this work, we introduce a novel computational strategy designed to extract and separate these time-varying conductances directly from the neurons spiking activity. By dynamically tracking just two accessible metrics - the amplitude of the spikes and the time intervals between them - our algorithm estimates both conductance profiles with high precision. Furthermore, we demonstrate that this procedure is highly robust against realistic experimental noise and data variability, providing an accessible framework that does not require complex hardware or an unfeasible number of repetitive experimental trials. By tracking changes in the E/I ratio of the synaptic input, this method provides an efficient approach to detecting pathological imbalances and understanding how local connectivity shapes cellular functionality.

Classical neurogeometric models describe the primary visual cortex as a fibered structure in which retinal position and local orientation are coupled through the geometry of the roto-translation group. We extend this approach to the visuomotor cortex by modeling it as an assemblage of visual and motor cortical geometries. The model combines orientation-selective representations, analogous to those of the primary visual cortex, with movement-direction-selective representations, analogous to those of the primary motor cortex, in order to describe the mixed visual and motor selectivity observed in the visuomotor cortex. We introduce a coupled visuomotor structure in which visual orientation and motor direction coexist over a common spatial plane and interact through a relative-orientation constraint. Neural responses are modeled by orientation- and direction-dependent profile functions, and preference maps are obtained from vectorized population responses. Numerical simulations generate visual, motor, and mixed visuomotor response maps. A competition rule between visual and motor responses produces incidence ratios close to experimental observations in macaque visuomotor cortex. This framework provides a first neurogeometric approximation of visuomotor functional architecture and a mathematical setting for studying visually guided action.

Biological neurons can perform nonlinear computations within their dendrites and support branch-localized plasticity. This raises the possibility that single cells can store memories more efficiently and with less interference by confining synaptic modifications to specific dendrites. We study a parallel-dendrite model performing online familiarity detection and compare three dendrite-update rules during learning: (i) independent thresholding, (ii) an interacting rule that adapts the target local dendritic activation per item, and (iii) an interacting n-winners-take-all (WTA) rule that constrains the number of updated branches per item. The interacting rules substantially improve capacity by limiting variance in memory responses and decorrelating weights across branches -- even when inputs are strongly correlated. These results suggest that competition among dendrites, consistent with resource-limited plasticity mechanisms, can enhance single-cell memory beyond non-interacting schemes.

This study presents improvements to the Hodgkin-Huxley (HH) models of ionic conductance and action potential generation. Sodium and potassium conductances are expressed by a single analytical formula describing the impulse response of a convolution of exponential distributions within a short-memory integration space. Treating transmembrane ion transit duration as a random variable, conductance profiles are interpreted as realizations of the probability density functions governing ionic movements. Applying the central limit theorem, the lognormal distribution emerges as the asymptotic profile of ionic conductances, constituting a fundamental primitive for such biosignals. A temporal state-transition paradigm describes the action potential waveform through four successive membrane potential transitions. Applied to electrophysiological recordings from lamprey reticulospinal neurons, this framework enables indirect estimation of key physiological quantities, including depolarization threshold, Nernst potentials, and net ion fluxes across the membrane. These advances open new perspectives for parameter estimation from experimental data and neuronal network simulation.

As silicon-based computing approaches fundamental physical limits, neurocomputing offers an energy-efficient alternative by leveraging the intrinsic non-linear dynamics of biological systems. To harness these dynamics, it is vital to understand the structure-function relationship governing how neural cultures process complex spatio-temporal information and how to appropriately decode the resulting neural electrophysiological activity. We investigated this utilizing a closed-loop electrophysiology platform, the CL1, to implement reservoir computing in human iPSC-derived neuronal networks. To systematically evaluate the variables driving neurocomputational capacity, we explored how cellular composition (cortical vs. hippocampal lineages), and the physical architecture (unstructured monolayers, 3D neural organoids, and modular networks confined by microfluidic devices) influenced electrophysiological properties and interacted with different decoding methodologies. Using a spatio-temporal version of a handwritten digit pattern recognition task (MNIST), we analyzed how these biological and analytical factors influenced classification accuracy. To ensure robust interpretation this required us to first demonstrated that reservoir computing decoding methods require strict artifact control and trial-based cross-validation to distinguish network computation from artifactual signal separability or temporal data leakage. Applying this validated frequency-domain pipeline, we suggest a clear functional hierarchy where structural modularity acts as a vital functional regularizer. Modular cortical cultures significantly outperformed unconstrained monolayers and organoids on MNIST. Furthermore, decoding frequency information from raw signals proved superior to typical time-bin decoding implementations. These findings establish that maximizing the computational potential of Synthetic Biological Intelligence, while avoiding false positives, requires a synergistic optimization of cellular identity, structural governance, and rigorous decoding logic. In doing so, this work provides a critical base establishing the criteria under which to evaluate neurocomputing implementations.

Time compaction theory is a general framework explaining how a brain can efficiently deal with dynamic situations occurring in, e.g., sports games. It involves a geometric representation of the time dimension, which enables effective learning and strategic action planning. The theory has recently received experimental support in humans. However, its current computational model has an important limitation: it does not account for deliberate waiting and speed modulation, behaviors ubiquitous in natural environments. This work substantially extends the original model formulation by a dimensional lifting of an n-D workspace into (n + 1)-D mental space, where time remains geometrically embedded. The proposed biologically inspired computational model can generate adaptive behavior across increasingly complex situations, from navigation in everyday social environments to competitive sports. Furthermore, by actively conditioning the expected responses of other agents and stabilizing future predictions, we introduce the concept of uncertainty points in sequences of generalized cognitive maps to support the generation of adaptive strategies in interactive environments, where future prediction has a limited time horizon. Thus, we provide a mechanism for chaining short-term solutions into long-term strategies, which is illustrated by simulating the behavior of a player in a real football game. Author summaryHumans often anticipate future interactions in dynamic environments. Many behaviors, such as avoiding other pedestrians, letting someone pass through a narrow corridor, or reproducing the kind of dribbling maneuvers performed by elite football players, require deciding not only where to move but also when to move. Existing theories suggest that the brain simplifies such situations by representing future interactions as static spatial maps, making them easier to learn and recall. However, current computational models cannot naturally account for common behaviors such as waiting, slowing down, or modulating speed. Here we show that these behaviors readily emerge if the model space is extended by an additional virtual coordinate that encodes accumulated waiting rather than physical time. The proposed model simultaneously admits a wide variety of behaviors, including speed modulation, multigoal decisions, and compound actions, while preserving the principles of time compaction. We illustrate the model in everyday situations and by reproducing two real football plays, comparing the observed behaviors with model simulations. Our results suggest computational principles through which the human brain may efficiently represent, memorize, and exploit dynamic situations.

With 302 neurons and a rigorously characterized connectome, the nematode Caenorhabditis elegans represents a powerful model organism to study the fundamental roles of neuronal circuits in behavior. However, despite the breadth of research, many questions remain unanswered regarding how these organisms are able to successfully navigate their environment. Here, we present a biologically grounded dynamical circuit model for the investigation of sensory-guided behavior during C. elegans chemotaxis. Our mathematical model consists of the chemosensory neuron AWA, interneurons RIM and RIA, motor neurons, including SMDs and RMDs, and body wall muscles that provide proprioceptive feedback through stretch receptors. After optimization with an evolutionary algorithm, the model locomotes effectively toward a chemical attractant, realistically capturing nematode chemotactic behavior. Chemotaxis is ensured by sharp turns, which resemble the omega turns of living nematodes, as a key emergent property of the model. The sharp turning behavior is triggered by decreases in the concentration of the attractant. These result in reduced AWA activity, which in turn triggers disinhibition of RIM and subsequent changes in RIA oscillations. The ensuing coordinated changes in downstream motor neurons activity patterns produce sharp turns, which correct the nematodes path, so that the model worm heads toward the attractant, and remains at its proximity, after it reaches the gradient peak. The proposed framework, along with its emergent dynamics, provides new insights into the minimum requirements for C. elegans circuitry to display major features of its chemotactic behavior, including omega turns. In parallel, it generates experimentally testable hypotheses with respect to the participating neuronal elements.

Decision formation is commonly described as the accumulation of noisy evidence in a low-dimensional decision variable, but it remains unclear how this latent computation is implemented by heterogeneous neural responses. Here, we propose that ramping and sequentially firing neurons form complementary population codes for the same decision variable. Inspired by Laplace-domain neural representations of time, exponential receptive fields in a ramping population give rise to a translatable edge-like activity profile; localized receptive fields in a sequential population give rise to an aligned bump-like profile. We construct a continuous attractor neural network that dynamically maintains these complementary representations while implementing evidence accumulation along a shared latent manifold. At the behavioral level, simulations show that the circuit closely reproduces the single-trial trajectories, choice probabilities, and reaction-time statistics of a standard diffusion decision model while generating heterogeneous ramping and sequential neural responses. Our framework connects latent behavioral dynamics, population geometry, and recurrent circuit mechanisms. More broadly, it provides a circuit-level realization of computation in the Laplace domain that may support the representation and updating of continuous cognitive variables across decision making, timing, memory, and spatial cognition.

Confidence evaluates the likely accuracy of a current decision. However, to be maximally informative about accuracy, confidence judgments should incorporate information about ones broader decision tendencies, such as their propensity to favor specific alternatives. We distinguish bias-aware confidence, which considers such tendencies, from bias-blind confidence, which relies only on evidence available on the current trial. To adjudicate between bias-aware and bias-blind confidence, we identified a signature of bias-aware confidence: the down-weighting of confidence for alternatives that a participant is biased toward. We then used a large dataset (N = 200) spanning 4- and 8-choice digit-classification tasks to show that humans reliably exhibit this signature of bias-aware confidence. This effect was reduced under speed pressure and could not be explained by guessing. In contrast to the human results, artificial neural networks (ANNs) trained for object recognition lacked this signature of bias- aware confidence. Importantly, augmenting ANNs with a metacognitive module that allows confidence to take the networks biases into account led to the emergence of human-like bias- aware confidence. These findings show that human confidence incorporates not only information from the current trial but also longer-term decision tendencies, and that this capacity - absent in standard ANNs - can be conferred through specialized metacognitive mechanisms.

Neural speech tracking is stronger for attended speech, yet its common correlation-based readout is scale invariant, so interpreting this effect only as response gain is incomplete. We tested whether attention improves representational fidelity, defined here as scale-invariant agreement between a speech envelope and its neural reconstruction. A leakage-resistant analysis evaluated held-out trials or story parts in three public electroencephalography datasets (52 participants). Fidelity was Fisher-transformed reconstruction-envelope correlation; projection slope quantified scale-dependent gain. In the spontaneous Auditory Attention Switching Dataset, nine odd- numbered participants were used for discovery and nine even-numbered participants for split- sample validation. Fidelity was higher for attended speech in the validation sample and exceeded 5,000 within-trial circular label shifts. The effect replicated under story-part-disjoint validation in KUL and trial-disjoint validation in DTU. A KUL crossover compared the same clean speech sources in attended and ignored states. Across 4,819 isolated spontaneous switches, fidelity did not differ from baseline before the report but shifted toward the newly reported target 0.25-1 s afterward. Passive keypresses altered nonspecific decoder energy. Gain advantages also occurred in all datasets. Selective attention was therefore evident in the scale-invariant preservation of target dynamics, while gain remained a complementary feature.

The human visual system integrates both static and dynamic information to support form and shape perception, yet the computational principles underlying the integration of motion for object recognition remain unclear. Artificial neural networks (ANNs) offer a computational framework for developing and testing hypotheses about these principles: if ANNs trained on motion-related tasks develop representations that align with brain activity and support object categorization, this would suggest that the training objectives and architectural constraints of these networks may capture key aspects of motion processing in biological visual systems in general, and motion processing for object recognition, in particular. Here, we investigated this question using "object kinematograms", stimuli in which object form is conveyed solely through motion cues. We measured neural responses of two higher regions of the lateral and the dorsal visual pathways, respectively, with strong sensitivity to dynamic cues from objects: lateral occipitotemporal cortex (LOTbio), and left supramarginal gyrus (SMGlh), as well as primary visual cortex (V1). We compared brain responses to representations extracted from two neural networks: SlowFast, a dual-pathway architecture trained on action recognition that processes slow- and fast-varying visual information with cross-pathway integration, and DorsalNet, a model of the primate dorsal visual pathway trained on embodied self-motion estimation. Representational similarity analysis revealed distinct representational profiles across brain areas, demonstrating functional specialization in motion-based form processing. LOTbio was best characterized by the slow pathway of the SlowFast model, whereas SMGlh showed strong similarity to both models. Critically, we found that representations aligned with brain activity also better supported behavioral function: the full SlowFast model, incorporating both slow and fast pathways, outperformed other models in few-shot categorization of object kinematograms and showed the highest similarity to human perceptual judgments. These findings demonstrate that with appropriate inductive biases, specifically, dual-pathway architectures for multi-scale motion processing and training objectives focused on dynamic visual tasks, ANNs can develop functionally useful representations of motion-defined forms that exhibit better alignment with the visual regions involved in processing dynamic visual signals.

Solitary animals face a tradeoff when detecting threats: faster detection means accepting more false alarms. We show that groups can manage this tradeoff better by treating an undisturbed neighbor as evidence against a threat, becoming both faster and more accurate than lone individuals. Modeling each animal as a noisy evidence-accumulator that flees when its belief crosses a threshold, we find that a neighbor's flight signals danger while its stillness signals safety. A naive responder reacts only to flights and inflates false alarms as the group grows; a Bayesian responder weighs both, approximated by a single social discounting rate that interpolates between these limits. This yields closed-form expressions for group performance, including cascade branching ratios that stay strongly subcritical in safety and turn supercritical under threat, so the rate at which an animal discounts a threat while its neighbors stay still can be inferred from behavior alone, and it sets a ceiling on how many neighbors an animal can attend before discounting alone can no longer hold its false-alarm rate. Wild sulphur molly shoals under bird attack are best described by discounting rates well above what individually Bayesian updating supplies over any neighborhood they could plausibly attend, and the same model, at the inferred value, predicts a false-alarm rate that stays constant as shoals grow.

Purpose: In clinical trials, treatment efficacy is commonly assessed by comparing control and treatment groups. However, in large samples, even small and clinically trivial differences may achieve statistical significance. Accordingly, the Minimal Clinically Important Difference (MCID) is used as a threshold to determine whether statistically-significant effects are also clinically meaningful to patients. The objective of this study was to estimate the MCID for the Childhood Autism Rating Scale Second-Edition (CARS2) using the Patient Impression of Change (PIC) as an external anchor. Methods: Single-item PICs are not well suited to characterizing improvement in a multifaceted disorder such as ASD. Accordingly, the 77-item Autism Treatment Evaluation Checklist (ATEC) was used as a multi-item PIC. CARS2 and ATEC were administered concurrently to 62 children with ASD, aged 1.8-7.9 years, with assessments conducted six months apart. Results: The correlation between changes in CARS2 and ATEC total scores was 0.41-0.44 (p<0.0001), supporting the use of ATEC as an anchor measure. Two anchor-based methods yielded MCIDs of 2.39-2.82 CARS2 points. Two distribution-based methods produced MCIDs of 1.33-3.32. Conclusions: Taken together, these approaches suggest that a between-group difference of 2.6 CARS2 points (the midpoint of the anchor-based estimates) may serve as the MCID in children with ASD.

Previous work showed that observers integrate audiovisual duration cues optimally when cue-conflict is small. Does causal inference lead to a breakdown of audiovisual integration when duration conflicts are large? We addressed this by testing a wide range of duration cue-conflicts. Participants compared the auditory durations of a test and a standard stimulus. Audiovisual durations were consistent in the test stimulus, but differed by seven conflict durations (up to 250 ms) in the standard. Two levels of auditory noise were tested. Auditory duration percepts shifted systematically toward the visual duration, especially with high auditory noise. The shift was proportional to cue-conflict magnitude, inconsistent with causal inference. We compared several models. A heuristic model in which the observer probabilistically switches between the visual and auditory cues was preferred for most participants, although performance differences across models were small. Within the tested conflict range, the forced fusion, causal inference, and probabilistic cue switching models produced overlapping, near-linear shifts as a function of cue-conflict. Model simulations further revealed that given the measured sensory noise, forced fusion and causal inference can be discriminated only with unreasonably large conflicts. Together, while our results suggest that observers do not rely on causal inference when judging auditory durations under our conditions, high sensory encoding noise in auditory duration limits the discriminability of competing computational models.

Behavioral experiments are often infeasible when stimulus spaces have many dimensions or when testing time is limited. One way to address this challenge is adaptive stimulus selection, where informative stimuli are chosen dynamically based on participants responses. However, in high-dimensional spaces, identifying such stimuli is computationally demanding. Here, we describe High-dimensional Online Particle Estimation (HOPE), which selects informative stimuli in less than a second for up to 50 dimensions, enabling efficient estimation of high-dimensional psychometric functions. We validate HOPE through simulations and a face-categorization experiment in an 18-dimensional parameter space with human participants. Compared to uniform stimulus presentation, HOPE reduces uncertainty over model parameters two-to three-times faster, reaching the same certainty in half the trials or fewer. This efficiency enables psychophysical studies that were previously impractical due to the exponential scaling of trial requirements.

Scaling laws help determine the optimal data size for training large models but are established in domains where the target is deterministic. Physiological signals are different: heartbeat sequences are stochastic, so part of the error is irreducible even with large amounts of data. Metrics such as MAE do not account for non-deterministic behavior, and therefore assessing scaling requires evaluating distributional calibration (measuring how well predicted probability densities capture true conditional characteristics). We formulate a scaling law metric(n) = E + A n- and evaluate it with five metrics: accuracy (MAE, RMSE), distributional calibration (KS distance, goodness-of-fit), and training objective (negative log loss) using a neural temporal point process trained on a cohort of four-ECG datasets. The law fits all five metrics. While point accuracy is near saturation at n = 183, KS distance and goodness-of-fit improve by 6% and 12% respectively when extrapolated to 10,000 subjects, showing that scaling decisions in stochastic domains must be guided by distributional calibration rather than point accuracy.