Biology subjects

Ijspeert, A. J.

Publications and source records attributed to Ijspeert, A. J..

3 recordsLinked to original sources

bioRxiv2024

Artificial Embodied Circuits Uncover Neural Architectures of Vertebrate Visuomotor Behaviors

All brains evolve within specific sensory and physical environments1. Traditionally, neuroscience has focused on studying neural circuits in isolation, yet holistic characterization of their function requires integrative brain-body testing2,3. To investigate the neural and biomechanical mechanisms of sensorimotor transformations, we constructed realistic neuromechanical simulations (simZFish) of the larval zebrafish optomotor response, a visual stabilization behavior4,5. By computationally reproducing the body, physical body-water interactions, visual environments, and experimentally derived neural architectures, we closely replicated the behavior of real zebrafish6. Through systematic manipulation of physiological and circuit features, impossible in biological experiments, we demonstrate how embodiment shapes neural circuit architecture and behavior. When challenged with novel visual stimuli, simZFish predicted neuronal response types, which we identified via calcium imaging in the brain of real zebrafish and used to update the simZFish neural network. In virtual rivers, simZFish performed rheotaxis by using current-induced optic flow as navigational cues, compensating for the simulated water flow. Finally, a physical robot (ZBot) validated the role of embodied sensorimotor circuits in maintaining position in a real river with complex fluid dynamics and visual environments. Together, by iterating between simulations, behavioral observations, neural imaging, and robotic testing, we demonstrate the power of an integrative approach to investigating sensorimotor processing. Research HighlightsO_LIDeveloped simZFish, an open-source neuromechanical simulation modeling the zebrafish optomotor response (OMR). C_LIO_LIDemonstrated that simZFish visuomotor neural circuits are sufficient to maintain position in virtual water currents without additional sensory modalities. C_LIO_LIInvestigated how variations in eye geometry influence neural circuit functionality and behavior. C_LIO_LIConducted optic flow analysis of simulated visual input to identify retinal connectivity requirements, uncovering the factors that shape pretectal neurons preference for the lower posterior visual field. C_LIO_LIsimZFish predicted new neural response types confirmed via real zebrafish calcium imaging. C_LIO_LIValidated simZFish circuits with a physical robot, ZBot, performing rheotaxis in a natural river with rich visual and complex flow dynamics compared to idealized lab and simulation experiments. C_LI O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=137 SRC="FIGDIR/small/629427v1_ufig1.gif" ALT="Figure 1"> View larger version (63K): org.highwire.dtl.DTLVardef@11f2acforg.highwire.dtl.DTLVardef@133b1f3org.highwire.dtl.DTLVardef@e808bcorg.highwire.dtl.DTLVardef@1959616_HPS_FORMAT_FIGEXP M_FIG C_FIG Graphical abstractCombining data-driven neuromechanical simulations and robotic testing, we established an integrative framework for investigating embodied neural circuit functions. Leveraging the behavioral, neurobiological, and theoretical foundations of the visually guided optomotor response in larval zebrafish (blue, left), we developed a virtual zebrafish, simZFish (red, middle), to replicate realistic hydrodynamic, anatomical, sensory, neural, and behavioral aspects, permitting the investigation of emergent properties driven by embodiment. Validating these findings using a physical, zebrafish-inspired robot, ZBot (grey, right), in naturalistic environments informs further experiments, hypotheses, and robotic design.

neuroscience↗

bioRxiv2024

Sensory and central contributions to motor pattern generation in a spiking, neuro-mechanical model of the salamander spinal cord

This study introduces a novel neuromechanical model employing a detailed spiking neural network to explore the role of axial proprioceptive sensory feedback, namely stretch feedback, in salamander locomotion. Unlike previous studies that often oversimplified the dynamics of the locomotor networks, our model includes detailed simulations of the classes of neurons that are considered responsible for generating movement patterns. The locomotor circuits, modeled as a spiking neural network of adaptive leaky integrate-and-fire neurons, are coupled to a three-dimensional mechanical model of a salamander with realistic physical parameters and simulated muscles. In open-loop simulations (i.e., without sensory feedback), the model replicates locomotor patterns observed in-vitro and in-vivo for swimming and trotting gaits. Additionally, a modular descending reticulospinal drive to the central pattern generation network allows to accurately control the activation, frequency and phase relationship of the different sections of the limb and axial circuits. In closed-loop swimming simulations (i.e. including axial stretch feedback), systematic evaluations reveal that intermediate values of feedback strength increase the tail beat frequency and reduce the intersegmental phase lag, contributing to a more coordinated, faster and energy-efficient locomotion. Interestingly, the result is conserved across different feedback topologies (ascending or descending, excitatory or inhibitory), suggesting that it may be an inherent property of axial proprioception. Moreover, intermediate feedback strengths expand the stability region of the network, enhancing its tolerance to a wider range of descending drives, internal parameters modifications and noise levels. Conversely, high values of feedback strength lead to a loss of controllability of the network and a degradation of its locomotor performance. Overall, this study highlights the beneficial role of proprioception in generating, modulating and stabilizing locomotion patterns, provided that it does not excessively override centrally-generated locomotor rhythms. This work also underscores the critical role of detailed, biologically-realistic neural networks to improve our understanding of vertebrate locomotion. Author summaryIn this paper, we developed a computational model to investigate how salamanders move, both while swimming and walking. Unlike previous studies that often oversimplified the dynamics of these complex neural networks, our model includes detailed simulations of the classes of neurons that are considered responsible for generating movement patterns. The locomotor circuits, modeled as a spiking neural network, are coupled to a three-dimensional mechanical model of a salamander with realistic physical parameters and simulated muscles. The neural model integrates axial proprioceptive sensory feedback from the bodys movements to modulate the locomotor gaits. Our simulations suggest that this sensory feedback plays a major role in controlling the rhythm and coordination of movements. This has implications for understanding not only how salamanders move but also provides insights into the evolution of locomotion in vertebrates. By investigating how central and sensory mechanisms interact to produce efficient and adaptable movement, our work contributes to the broader field of neuroscience and robotics, offering potential strategies for designing more effective biomimetic robots.

neuroscience↗

bioRxiv2023

FARMS: Framework for Animal and Robot Modeling and Simulation

The apparent ease with which animals move hides the complexity of the neuromechanical systems underlying locomotion. This behavior remains difficult to understand and replicate, motivating sustained interest across biology, neuroscience, robotics, and computer animation. Despite extensive progress, existing approaches often focus on isolated aspects of locomotion, limiting the integration of neural control, biomechanics, and environmental interactions within a unified modeling framework. Bridging biological insight, robotic implementation, and physically based character animation therefore requires shared, extensible computational tools that promote reuse, while supporting physically grounded modeling and control across animals and robots. Our key technical contribution is the development of FARMS (Framework for Animal and Robot Modeling and Simulation), an open-source, interdisciplinary Python framework that integrates and extends existing open-source software for 3D modeling, experiment design, optimization, and data analysis within a modular and extensible workflow. FARMS lowers the barrier for constructing and simulating whole-body animal and robot models, enabling the definition of skeletal structures, musculature, and neural circuits at multiple levels of abstraction within a unified workflow. FARMS has already been used to model and study locomotor behaviors in a variety of animals and robots, and we demonstrate the frameworks versatility and effectiveness through a series of simulation case studies. We show that FARMS enables the reuse of model components across morphologies of different complexities, while capturing mechanical interactions between the body and diverse environments, including terrestrial and aquatic settings. This allows the study and synthesis of walking, swimming, and locomotor transitions within a single, unified framework.

animal behavior and cognition↗