Mental navigation in the primate entorhinal cortex.

A cognitive map is a suitably structured representation that enables novel computations using previous experience; for example, planning a new route in a familiar space1. Work in mammals has found direct evidence for such representations in the presence of exogenous sensory inputs in both spatial2,3 and non-spatial domains4-10. Here we tested a foundational postulate of the original cognitive map theory1,11: that cognitive maps support endogenous computations without external input. We recorded from the entorhinal cortex of monkeys in a mental navigation task that required the monkeys to use a joystick to produce one-dimensional vectors between pairs of visual landmarks without seeing the intermediate landmarks. The ability of the monkeys to perform the task and generalize to new pairs indicated that they relied on a structured representation of the landmarks. Task-modulated neurons exhibited periodicity and ramping that matched the temporal structure of the landmarks and showed signatures of continuous attractor networks12,13. A continuous attractor network model of path integration14 augmented with a Hebbian-like learning mechanism provided an explanation of how the system could endogenously recall landmarks. The model also made an unexpected prediction that endogenous landmarks transiently slow path integration, reset the dynamics and thereby reduce variability. This prediction was borne out in a reanalysis of firing rate variability and behaviour. Our findings link the structured patterns of activity in the entorhinal cortex to the endogenous recruitment of a cognitive map during mental navigation.

A large-scale neural network training framework for generalized estimation of single-trial population dynamics.

Achieving state-of-the-art performance with deep neural population dynamics models requires extensive hyperparameter tuning for each dataset. AutoLFADS is a model-tuning framework that automatically produces high-performing autoencoding models on data from a variety of brain areas and tasks, without behavioral or task information. We demonstrate its broad applicability on several rhesus macaque datasets: from motor cortex during free-paced reaching, somatosensory cortex during reaching with perturbations, and dorsomedial frontal cortex during a cognitive timing task.

Mixtures of strategies underlie rodent behavior during reversal learning.

In reversal learning tasks, the behavior of humans and animals is often assumed to be uniform within single experimental sessions to facilitate data analysis and model fitting. However, behavior of agents can display substantial variability in single experimental sessions, as they execute different blocks of trials with different transition dynamics. Here, we observed that in a deterministic reversal learning task, mice display noisy and sub-optimal choice transitions even at the expert stages of learning. We investigated two sources of the sub-optimality in the behavior. First, we found that mice exhibit a high lapse rate during task execution, as they reverted to unrewarded directions after choice transitions. Second, we unexpectedly found that a majority of mice did not execute a uniform strategy, but rather mixed between several behavioral modes with different transition dynamics. We quantified the use of such mixtures with a state-space model, block Hidden Markov Model (block HMM), to dissociate the mixtures of dynamic choice transitions in individual blocks of trials. Additionally, we found that blockHMM transition modes in rodent behavior can be accounted for by two different types of behavioral algorithms, model-free or inference-based learning, that might be used to solve the task. Combining these approaches, we found that mice used a mixture of both exploratory, model-free strategies and deterministic, inference-based behavior in the task, explaining their overall noisy choice sequences. Together, our combined computational approach highlights intrinsic sources of noise in rodent reversal learning behavior and provides a richer description of behavior than conventional techniques, while uncovering the hidden states that underlie the block-by-block transitions.

Validating model-based Bayesian integration using prior-cost metamers.

There are two competing views on how humans make decisions under uncertainty. Bayesian decision theory posits that humans optimize their behavior by establishing and integrating internal models of past sensory experiences (priors) and decision outcomes (cost functions). An alternative hypothesis posits that decisions are optimized through trial and error without explicit internal models for priors and cost functions. To distinguish between these possibilities, we introduce a paradigm that probes the sensitivity of humans to transitions between prior-cost pairs that demand the same optimal policy (metamers) but distinct internal models. We demonstrate the utility of our approach in two experiments that were classically explained by Bayesian theory. Our approach validates the Bayesian learning strategy in an interval timing task but not in a visuomotor rotation task. More generally, our work provides a domain-general approach for testing the circumstances under which humans explicitly implement model-based Bayesian computations.

Low-dimensional dynamics for working memory and time encoding.

Our decisions often depend on multiple sensory experiences separated by time delays. The brain can remember these experiences and, simultaneously, estimate the timing between events. To understand the mechanisms underlying working memory and time encoding, we analyze neural activity recorded during delays in four experiments on nonhuman primates. To disambiguate potential mechanisms, we propose two analyses, namely, decoding the passage of time from neural data and computing the cumulative dimensionality of the neural trajectory over time. Time can be decoded with high precision in tasks where timing information is relevant and with lower precision when irrelevant for performing the task. Neural trajectories are always observed to be low-dimensional. In addition, our results further constrain the mechanisms underlying time encoding as we find that the linear "ramping" component of each neuron's firing rate strongly contributes to the slow timescale variations that make decoding time possible. These constraints rule out working memory models that rely on constant, sustained activity and neural networks with high-dimensional trajectories, like reservoir networks. Instead, recurrent networks trained with backpropagation capture the time-encoding properties and the dimensionality observed in the data.

Neural Encoding and Representation of Time for Sensorimotor Control and Learning.

The ability to perceive and produce movements in the real world with precise timing is critical for survival in animals, including humans. However, research on sensorimotor timing has rarely considered the tight interrelation between perception, action, and cognition. In this review, we present new evidence from behavioral, computational, and neural studies in humans and nonhuman primates, suggesting a pivotal link between sensorimotor control and temporal processing, as well as describing new theoretical frameworks regarding timing in perception and action. We first discuss the link between movement coordination and interval-based timing by addressing how motor training develops accurate spatiotemporal patterns in behavior and influences the perception of temporal intervals. We then discuss how motor expertise results from establishing task-relevant neural manifolds in sensorimotor cortical areas and how the geometry and dynamics of these manifolds help reduce timing variability. We also highlight how neural dynamics in sensorimotor areas are involved in beat-based timing. These lines of research aim to extend our understanding of how timing arises from and contributes to perceptual-motor behaviors in complex environments to seamlessly interact with other cognitive processes.

Neural coding of uncertainty and probability.

Organisms must act in the face of sensory, motor, and reward uncertainty stemming from a pandemonium of stochasticity and missing information. In many tasks, organisms can make better decisions if they have at their disposal a representation of the uncertainty associated with task-relevant variables. We formalize this problem using Bayesian decision theory and review recent behavioral and neural evidence that the brain may use knowledge of uncertainty, confidence, and probability.

Engineering recurrent neural networks from task-relevant manifolds and dynamics.

Many cognitive processes involve transformations of distributed representations in neural populations, creating a need for population-level models. Recurrent neural network models fulfill this need, but there are many open questions about how their connectivity gives rise to dynamics that solve a task. Here, we present a method for finding the connectivity of networks for which the dynamics are specified to solve a task in an interpretable way. We apply our method to a working memory task by synthesizing a network that implements a drift-diffusion process over a ring-shaped manifold. We also use our method to demonstrate how inputs can be used to control network dynamics for cognitive flexibility and explore the relationship between representation geometry and network capacity. Our work fits within the broader context of understanding neural computations as dynamics over relatively low-dimensional manifolds formed by correlated patterns of neurons.

Integration of speed and time for estimating time to contact.

To coordinate movements with events in a dynamic environment the brain has to anticipate when those events occur. A classic example is the estimation of time to contact (TTC), that is, when an object reaches a target. It is thought that TTC is estimated from kinematic variables. For example, a tennis player might use an estimate of distance (d) and speed (v) to estimate TTC (TTC = d/v). However, the tennis player may instead estimate TTC as twice the time it takes for the ball to move from the serve line to the net line. This latter strategy does not rely on kinematics and instead computes TTC solely from temporal cues. Which of these two strategies do humans use to estimate TTC? Considering that both speed and time estimates are inherently uncertain and the ability of the human brain to combine different sources of information, we hypothesized that humans estimate TTC by integrating speed information with temporal cues. We evaluated this hypothesis systematically using psychophysics and Bayesian modeling. Results indicated that humans rely on both speed information and temporal cues and integrate them to optimize their TTC estimates when both cues are present. These findings suggest that the brain's timing mechanisms are actively engaged when interacting with dynamic stimuli.

Representation of accumulating evidence for a decision in two parietal areas.

Decisions are often made by accumulating evidence for and against the alternatives. The momentary evidence represented by sensory neurons is accumulated by downstream structures to form a decision variable, linking the evolving decision to the formation of a motor plan. When decisions are communicated by eye movements, neurons in the lateral intraparietal area (LIP) represent the accumulation of evidence bearing on the potential targets for saccades. We now show that reach-related neurons from the medial intraparietal area (MIP) exhibit a gradual modulation of their firing rates consistent with the representation of an evolving decision variable. When decisions were communicated by saccades instead of reaches, decision-related activity was attenuated in MIP, whereas LIP neurons were active while monkeys communicated decisions by saccades or reaches. Thus, for decisions communicated by a hand movement, a parallel flow of sensory information is directed to parietal areas MIP and LIP during decision formation.

Time in Cortical Circuits.

Time is central to cognition. However, the neural basis for time-dependent cognition remains poorly understood. We explore how the temporal features of neural activity in cortical circuits and their capacity for plasticity can contribute to time-dependent cognition over short time scales. This neural activity is linked to cognition that operates in the present or anticipates events or stimuli in the near future. We focus on deliberation and planning in the context of decision making as a cognitive process that integrates information across time. We progress to consider how temporal expectations of the future modulate perception. We propose that understanding the neural basis for how the brain tells time and operates in time will be necessary to develop general models of cognition. SIGNIFICANCE STATEMENT: Time is central to cognition. However, the neural basis for time-dependent cognition remains poorly understood. We explore how the temporal features of neural activity in cortical circuits and their capacity for plasticity can contribute to time-dependent cognition over short time scales. We propose that understanding the neural basis for how the brain tells time and operates in time will be necessary to develop general models of cognition.

Epidemiology of venomous and semi-venomous snakebites (Ophidia: Viperidae, Colubridae) in the Kashan city of the Isfahan province in Central Iran.

BACKGROUND: Information on the epidemiology of venomous snake species responsible for envenomation to humans in Iran has not been well documented. In the Kashan city, venomous snakebite remains a recurring medical problem. Information providing the correct identification of snake species responsible for envenomation in this geographic region would be useful to regional medical clinics and personnel for the effective and optimal management of the patients. MATERIALS AND METHODS: In this cross-sectional study, all patient data was collected from Kashan city and its suburbs. The specific data relating to the taxonomic identification of snakes responsible for envenomation were evaluated. A general approach to the diagnosis and management of patients was also provided. Snakes responsible for bites were transported to a laboratory, where their taxonomic classification was confirmed based on key anatomical features and morphological characteristics. RESULTS: A total of 46 snakes were examined. Of these, 37 (80%) were non-venomous species, and 9 (20%) were identified as venomous. Seven of the nine venomous snake species (78%) were of the family Viperidae, and two specimens (22%) were in the family Colubridae. Specifically, the viperid species were Macrovipera lebetina obtusa, Pseudocerastes persicus, Pseudocerastes fieldi, and Echis carinatus. The two colubrid species were Malpolon monspessulanus insignitus and Psammophis schkari. CONCLUSION: Five different species of venomous snakes responsible for envenomation in the Kashan city region were confirmed. The viper, P. fieldi, was reported for the first time in the central part of Iran.

Keeping time and rhythm by internal simulation of sensory stimuli and behavioral actions.

Effective behavior often requires synchronizing our actions with changes in the environment. Rhythmic changes in the environment are easy to predict, and we can readily time our actions to them. Yet, how the brain encodes and maintains rhythms is not known. Here, we trained primates to internally maintain rhythms of different tempos and performed large-scale recordings of neuronal activity across the sensory-motor hierarchy. Results show that maintaining rhythms engages multiple brain areas, including visual, parietal, premotor, prefrontal, and hippocampal regions. Each recorded area displayed oscillations in firing rates and oscillations in broadband local field potential power that reflected the temporal and spatial characteristics of an internal metronome, which flexibly encoded fast, medium, and slow tempos. The presence of widespread metronome-related activity, in the absence of stimuli and motor activity, suggests that internal simulation of stimuli and actions underlies timekeeping and rhythm maintenance.

Dynamics of macaque MT cell responses to grating triplets.

Neurons in area MT are sensitive to the direction of motion of gratings and of plaids made by summing 2 gratings moving in different directions. MT component direction-selective (CDS) neurons respond to the individual gratings of a plaid. Pattern direction-selective (PDS) neurons on the other hand, combine component information and respond selectively to the resulting pattern motion. Adding a third grating creates a "triplaid," which contains 3 grating and 3 plaid motions and is perceptually multistable. To examine how direction-selective mechanisms parse the motion signals in triplaids, we recorded MT responses of anesthetized and awake macaques to stimuli in which 3 identical moving gratings whose directions were separated by 120° were introduced in 3 successive epochs, going from grating to plaid to triplaid. CDS and PDS neurons-selected based on their responses to gratings and plaids-had strikingly different tuning properties in the triplaid epoch. CDS neurons were strongly tuned for the direction of motion of individual gratings, but PDS neurons nearly lost their selectivity for either the gratings or the plaids in the stimulus. We explain this reduced motion selectivity with a model that relates pattern selectivity of PDS neurons to a broad pooling of V1 afferents with a near-cosine weighting profile. Because PDS neurons signal both component and pattern motion in gratings and plaids, their reduced selectivity for motion in triplaids may be what makes these stimuli perceptually multistable.

The spread of attention across features of a surface.

Contrasting theories of visual attention have emphasized selection by spatial location, individual features, and whole objects. We used functional magnetic resonance imaging to ask whether and how attention to one feature of an object spreads to other features of the same object. Subjects viewed two spatially superimposed surfaces of random dots that were segregated by distinct color-motion conjunctions. The color and direction of motion of each surface changed smoothly and in a cyclical fashion. Subjects were required to track one feature (e.g., color) of one of the two surfaces and detect brief moments when the attended feature diverged from its smooth trajectory. To tease apart the effect of attention to individual features on the hemodynamic response, we used a frequency-tagging scheme. In this scheme, the stimulus features (color and direction of motion) are modulated periodically at distinct frequencies so that the contribution of each feature to the hemodynamics can be inferred from the harmonic response at the corresponding frequency. We found that attention to one feature (e.g., color) of one surface increased the response modulation not only to the attended feature but also to the other feature (e.g., motion) of the same surface. This attentional modulation was evident in multiple visual areas and was present as early as V1. The spread of attention to the behaviorally irrelevant features of a surface suggests that attention may automatically select all features of a single object. Thus object-based attention may be supported by an enhancement of feature-specific sensory signals in the visual cortex.

A new perceptual illusion reveals mechanisms of sensory decoding.

Perceptual illusions are usually thought to arise from the way sensory signals are encoded by the brain, and indeed are often used to infer the mechanisms of sensory encoding. But perceptual illusions might also result from the way the brain decodes sensory information, reflecting the strategies that optimize performance in particular tasks. In a fine discrimination task, the most accurate information comes from neurons tuned away from the discrimination boundary, and observers seem to use signals from these 'displaced' neurons to optimize their performance. We wondered whether using signals from these neurons might also bias perception. In a fine direction discrimination task using moving random-dot stimuli, we found that observers' perception of the direction of motion is indeed biased away from the boundary. This misperception can be accurately described by a decoding model that preferentially weights signals from neurons whose responses best discriminate those directions. In a coarse discrimination task, to which a different decoding rule applies, the same stimulus is not misperceived, suggesting that the illusion is a direct consequence of the decoding strategy that observers use to make fine perceptual judgments. The subjective experience of motion is therefore not mediated directly by the responses of sensory neurons, but is only developed after the responses of these neurons are decoded.

Neural Foundations of Mental Simulation: Future Prediction of Latent Representations on Dynamic Scenes.

Humans and animals have a rich and flexible understanding of the physical world, which enables them to infer the underlying dynamical trajectories of objects and events, plausible future states, and use that to plan and anticipate the consequences of actions. However, the neural mechanisms underlying these computations are unclear. We combine a goal-driven modeling approach with dense neurophysiological data and high-throughput human behavioral readouts that contain thousands of comparisons to directly impinge on this question. Specifically, we construct and evaluate several classes of sensory-cognitive networks to predict the future state of rich, ethologically-relevant environments, ranging from self-supervised end-to-end models with pixel-wise or object-slot objectives, to models that future predict in the latent space of purely static image-pretrained or dynamic video-pretrained foundation models. We find that "scale is not all you need", and that many state-of-the-art machine learning models fail to perform well on our neural and behavioral benchmarks for future prediction. In fact, only one class of models matches these data well overall. We find that neural responses are currently best predicted by models trained to predict the future state of their environment in the latent space of pretrained foundation models optimized for dynamic scenes in a self-supervised manner. These models also approach the neurons' ability to predict the environmental state variables that are visually hidden from view, despite not being explicitly trained to do so. Finally, we find that not all foundation model latents are equal. Notably, models that future predict in the latent space of video foundation models that are optimized to support a diverse range of egocentric sensorimotor tasks, reasonably match both human behavioral error patterns and neural dynamics across all environmental scenarios that we were able to test. Overall, these findings suggest that the neural mechanisms and behaviors of primate mental simulation have strong inductive biases associated with them, and are thus far most consistent with being optimized to future predict on reusable visual representations that are useful for Embodied AI more generally.

Parametric control of flexible timing through low-dimensional neural manifolds.

Biological brains possess an unparalleled ability to adapt behavioral responses to changing stimuli and environments. How neural processes enable this capacity is a fundamental open question. Previous works have identified two candidate mechanisms: a low-dimensional organization of neural activity and a modulation by contextual inputs. We hypothesized that combining the two might facilitate generalization and adaptation in complex tasks. We tested this hypothesis in flexible timing tasks where dynamics play a key role. Examining trained recurrent neural networks, we found that confining the dynamics to a low-dimensional subspace allowed tonic inputs to parametrically control the overall input-output transform, enabling generalization to novel inputs and adaptation to changing conditions. Reverse-engineering and theoretical analyses demonstrated that this parametric control relies on a mechanism where tonic inputs modulate the dynamics along non-linear manifolds while preserving their geometry. Comparisons with data from behaving monkeys confirmed the behavioral and neural signatures of this mechanism.

Recurrent neural networks with explicit representation of dynamic latent variables can mimic behavioral patterns in a physical inference task.

Primates can richly parse sensory inputs to infer latent information. This ability is hypothesized to rely on establishing mental models of the external world and running mental simulations of those models. However, evidence supporting this hypothesis is limited to behavioral models that do not emulate neural computations. Here, we test this hypothesis by directly comparing the behavior of primates (humans and monkeys) in a ball interception task to that of a large set of recurrent neural network (RNN) models with or without the capacity to dynamically track the underlying latent variables. Humans and monkeys exhibit similar behavioral patterns. This primate behavioral pattern is best captured by RNNs endowed with dynamic inference, consistent with the hypothesis that the primate brain uses dynamic inferences to support flexible physical predictions. Moreover, our work highlights a general strategy for using model neural systems to test computational hypotheses of higher brain function.