Biology subjects

Cooke, J. R. H.

Publications and source records attributed to Cooke, J. R. H..

6 recordsLinked to original sources

bioRxiv2026

Task-Dependent Modulation of Feedback Control in Human Steering

We examined whether human steering behavior conforms to optimal feedback control (OFC) principles when driving a vehicle through sequences of upcoming gates varying in width (narrow/wide) relative to the vehicles size, while occasional lateral velocity perturbations elicited corrective steering responses. In 24 participants, three predictions of OFC were tested: (1) greater positional variability when passing wide gates; (2) reduced corrective steering (lower feedback gains) to perturbations for wide compared to narrow gates (minimum intervention); and (3) modulation of feedback gains according to the relative spatial configuration of upcoming gates. Steering data were analyzed in terms of vehicle position and control velocity, the latter representative of feedback gain modulations. Results supported all predictions. When no perturbations occurred, participants showed wider distributions of passing locations through wide gates, consistent with allowing more task-relevant variability. During perturbed trials, feedback gains were lower for wide gates, confirming context-dependent control. When both gate types were center aligned, feedback responses were symmetric, whereas alignment of inner posts produced asymmetric feedback gains, being stronger for outward perturbations in narrow gates. Overall, these findings indicate that participants adapt their feedback control strategies to task-specific spatial constraints, demonstrating flexible optimization of steering control behavior.

neuroscience↗

bioRxiv2025

A dynamic causal inference framework for perception-action loops

Causal inference, the process of inferring the causes of our sensory input, is central to multisensory perception. While most computational models of causal inference focus on static perceptual tasks with no temporal or motor components, real-world behavior unfolds dynamically and often involves closed-loop control. Here, we introduce and validate a general modeling framework that unifies multisensory causal inference with optimal feedback control by casting the problem as inference and action in a switching linear dynamical system (SLDS). Our framework combines approximate inference over latent causal structures with a mixture-of-controllers approach, in which each controller is optimized for a specific causal model and weighted by the current belief in that model. We show that classical static models of multisensory perception are special cases of our framework, and extend them to dynamic, action-oriented settings where inference and control evolve over time. Using simulations of previously published behavioral tasks, including visuo-vestibular heading discrimination and path integration with interception, we show that the model reproduces key empirical findings, such as S-shaped biases in heading estimation and motor strategies based on inferred object motion. We also demonstrate the versatility of our approach in novel task extensions, including motion in depth, latent switching dynamics, and multi-objective motor control. Together, our results provide a principled, general-purpose computational account of how causal inference and motor behavior interact in dynamic, uncertain environments, offering a bridge between theories of perception and action. Author summaryWhen we interact with the world, our brain constantly combines information from different senses -- like visual and vestibular sensory input - to determine what is happening and guide an appropriate response. However, these signals do not always originate from the same source in the world. For example, a moving object might not match the background movement we feel during self-motion. In such situations, the brain faces the challenge of "causal inference": do these signal arise from a common source or from separate events. Most previous research has studied this problem in simplified, perceptual settings. But in everyday life, we move through the world and make decisions in real time based on noisy and often conflicting sensory input. In this work, we develop and validate a new computational framework that combines causal inference with models of movement control. This approach allows us to simulate more realistic situations where perception and action interact continuously. We show that our model not only reproduces results from earlier experiments but also makes predictions about how people behave in more complex tasks, like intercepting moving targets or navigating uncertain environments. Our framework offers a new way to study how the brain integrates sensory information to guide behavior in the real world.

neuroscience↗

bioRxiv2024

Visuoinertial and visual feedback in online steering control

Multisensory integration has primarily been studied in static environments, where optimal integration relies on the precision of the respective sensory modalities. However, in numerous situations, sensory information is dynamic and changes over time, due to changes in our bodily state and the surrounding environment. Given that different sensory modalities have different delays, this suggests that optimal integration may not solely depend on sensory precision but may also be affected by the delays associated with each sensory system. To investigate this hypothesis, participants (n = 22, 16 female) engaged in a continuous steering task. Participants sat on a motion platform facing a screen that displayed a cartoonish traffic scene, featuring a car traveling along a road. In the visuoinertial condition, where vestibular and somatosensory feedback were available, they were tasked with counteracting an external multi-frequency perturbation signal, which laterally perturbed the platform and the car, such that the car was kept within the center of the road. In the visual condition, the visual car was perturbed, while the motion platform remained stationary. We show that participants compensate better for the perturbation in the visuoinertial than the visual condition, particularly in the high frequency range of the perturbation. Using computational modelling, we demonstrate that this enhanced performance is partially due to the shorter delay of the vestibular modality. In this condition, participants rely more on the vestibular information, which is less delayed than the more precise but longer delayed, visual information. Author summaryNavigating the world effectively requires the continuous integration of sensory information related to our self-motion, forming a cohesive representation of our current location and direction. Although studies on multisensory integration indicate that sensory inputs are combined optimally based on their precision, these findings have largely been derived from static experimental settings. In real-life situations, the sensory environment changes continuously, as does the state of our body. In such scenarios, how are sensory inputs combined? We developed an experimental paradigm where participants were tasked with controlling a simple vehicle while counteracting a time-varying perturbation signal. To assess the role of different sensory inputs we manipulated whether visual and inertial (primarily vestibular) cues were available or solely visual cues. Our results show that the presence of both visual and inertial cues significantly improved task performance. Modeling suggests this improvement is likely due to the shorter delay associated with the vestibular system compared to the visual system. Collectively, our results suggest that sensory precision alone is insufficient for effective state estimation and control in naturalistic scenarios, where accounting for the different delays of the sensory systems is critical for optimal task performance.

neuroscience↗

bioRxiv2021

Subjective confidence reflects representation of Bayesian probability in cortex

What gives rise to the human sense of confidence? Here, we tested the Bayesian hypothesis that confidence is based on a probability distribution represented in neural population activity. We implemented several computational models of confidence, and tested their predictions using psychophysics and fMRI. Using a generative model-based fMRI decoding approach, we extracted probability distributions from neural population activity in human visual cortex. We found that subjective confidence tracks the shape of the decoded distribution. That is, when sensory evidence was more precise, as indicated by the decoded distribution, observers reported higher levels of confidence. We furthermore found that neural activity in the insula, anterior cingulate, and prefrontal cortex was linked to both the shape of the decoded distribution and reported confidence, in ways consistent with the Bayesian model. Altogether, our findings support recent statistical theories of confidence and suggest that probabilistic information guides the computation of ones sense of confidence.

neuroscience↗

bioRxiv2018

Causal inference for spatial constancy across whole-body motion

The brain can estimate the amplitude and direction of self-motion by integrating multiple sources of sensory information, and use this estimate to update object positions in order to provide us with a stable representation of the world. A strategy to improve the precision of the object position estimate would be to integrate this internal estimate and the sensory feedback about the object position based on their reliabilities. Integrating these cues, however, would only be optimal under the assumption that the object has not moved in the world during the intervening body displacement. Therefore, the brain would have to infer whether the internal estimate and the feedback relate to a same external position (stable object), and integrate and/or segregate these cues based on this inference - a process that can be modeled as Bayesian Causal inference. To test this hypothesis, we designed a spatial updating task across passive whole body translation in complete darkness, in which participants (n=11), seated on a vestibular sled, had to remember the world-fixed position of a visual target. Immediately after the translation, a second target (feedback) was briefly flashed around the estimated \"updated\" target location, and participants had to report the initial target location. We found that the participants responses were systematically biased toward the position of the second target position for relatively small but not for large differences between the \"updated\" and the second target location. This pattern was better captured by a Bayesian causal inference model than by alternative models that would always either integrate or segregate the internally-updated target position and the visual feedback. Our results suggest that the brain implicitly represents the posterior probability that the internally updated estimate and the sensory feedback come from a common cause, and use this probability to weigh the two sources of information in mediating spatial constancy across whole-body motion.\n\nAuthor SummaryA change of an objects position on our retina can be caused by a change of the objects location in the world or by a movement of the eye and body. Here, we examine how the brain solves this problem for spatial updating by assessing the probability that the internally-updated location during body motion and observed retinal feedback after the motion stems from the same object location in the world. Guided by Bayesian causal inference model, we demonstrate that participants errrors in spatial updating depend nonlinearly on the spatial discrepancy between internally-updated and reafferent visual feedback about the objects location in the world. We propose that the brain implicitly represents the probability that the internally updated estimate and the sensory feedback come from a common cause, and use this probability to weigh the two sources of information in mediating spatial constancy across whole-body motion.

neuroscience↗

bioRxiv2017

Bayesian adaptive stimulus selection for dissociating models of psychophysical data

Comparing models facilitates testing different hypotheses regarding the computational basis of perception and action. Effective model comparison requires stimuli for which models make different predictions. Typically, experiments use a predetermined set of stimuli or sample stimuli randomly. Both methods have limitations; a predetermined set may not contain stimuli that dissociate the models whereas random sampling may be inefficient. To overcome these limitations, we expanded the psi-algorithm (Kontsevich & Tyler, 1999) from estimating the parameters of a psychometric curve to distinguishing models. To test our algorithm, we applied it to two distinct problems. First, we investigated dissociating sensory noise models. We simulated ideal observers with different noise models performing a 2-afc task. Stimuli were selected randomly or using our algorithm. We found using our algorithm improved the accuracy of model comparison. We also validated the algorithm in subjects by inferring which noise model underlies speed perception. Our algorithm converged quickly to the model previously proposed (Stocker & Simoncelli, 2006), whereas if stimuli were selected randomly model probabilities separated slower and sometimes supported alternative models. Second, we applied our algorithm to a different problem; comparing models of target selection under body acceleration. Previous work found target choice preference is modulated by whole body acceleration (Rincon-Gonzalez et al., 2016). However, the effect is subtle making model comparison difficult. We show that selecting stimuli adaptively could have led to stronger conclusions in model comparison. We conclude that our technique is more efficient and more reliable than current methods of stimulus selection for dissociating models.\n\nData AvailabilityAll data and code will be posted on our institutional repository system following acceptance. In the meantime feel free to contact the authors if you would like any of the code.

neuroscience↗