Biorobotics/Biomechanics

Spinal cord

Modelling

Neuron

Brain

Analytics

Intracellular

Cerebellum

For three decades our founders challenged the common consensus of one of the biggest scientific questions of our time: How does the brain work?

IntuiCell | Research

Born in the intersection
of AI and Neuroscience.

IntuiCell is a Lund University spin-off with presence in Lund and Stockholm. 

For three decades our founders challenged the common consensus 
of one of the biggest scientific questions of our time:

Today, we apply those groundbreaking neurophysiological insights 
to redesign artificial intelligence. Creating AI that is 

Explore our contrarian research on the mechanisms by which the 
brain predicts the world.

Research

CO Norrsken House
Birger Jarlsgatan 57C
SE-113 56 Stockholm

About

AI isn't actually intelligent. IntuiCell already is.

IntuiCell

The function of the brain function is defined by the interactions between its neurons. But these neurons exist in tremendous numbers, are continuously active and densely interconnected. Thereby they form one of the most complex dynamical systems known and there is a lack of approaches to characterize the functional properties of such biological neuronal networks. Here we introduce an approach to describe these functional properties by using its activity-defining constituents, the weights of the synaptic connections and the current activity of its neurons. We show how a high-dimensional vector field, which describes how the activity distribution across the neuron population is impacted at each instant of time, naturally emerges from these constituents. We show why a mixture of excitatory and inhibitory neurons and a diversity of synaptic weights are critical to obtain a network vector field with a structural richness. We argue that this structural richness is the foundation of activity diversity in the brain and thereby an underpinning of the behavioral flexibility and adaptability that characterizes biological creatures.

The function of the brain function is defined by the interactions between its neurons.

Neuronal networks quantified as vector fields

How biological brains can learn so quickly to become operational and achieve complex behavior remains an unresolved issue. Here we introduce a ‘neuromorphic’ learning strategy that resembles how immature biological brains learn by consisting of continual random activations of a complex mechanically coupled system with rich, dynamic intrinsic sensor dependencies, in this regard reminiscent of a biological body. Using a dynamic model of biological skin tissue with embedded sensors, we trained small, recurrent networks that emulated a primordial cortex and featured excitatory and inhibitory neurons with simultaneous independent learning in both types of synapses. Training with non-repetitive, random activations of the skin, where the recurrent network activity state was inherited between activations, autonomously led to rapid acquisition of remarkably generalizable representations of a predictive nature. The network could separate inputs and solve a kinematics task that had never been encountered, also after substantial parts of the sensor population were deleted. This strategy of focussing learning on the dominant regularities in dynamic sensory information can potentially explain complex brain operation and efficient learning.

How biological brains can learn so quickly to become operational and achieve complex behavior remains an unresolved issue.

Sensory dependencies rapidly and autonomously yield generalizable representations in recurrent cortical-like networks

Currently, it is accepted that animal locomotion is controlled by a central pattern generator in the spinal cord. Experiments and models show that rhythm generating neurons and genetically determined network properties could sustain oscillatory output activity suitable for locomotion. However, current central pattern generator models do not explain how a spinal cord circuitry, which has the same basic genetic plan across species, can adapt to control the different biomechanical properties and locomotion patterns existing in these species. Here we demonstrate that rhythmic and alternating movements in pendulum models can be learned by a monolayer spinal cord circuitry model using the Bienenstock–Cooper–Munro learning rule, which has been previously proposed to explain learning in the visual cortex. These results provide an alternative theory to central pattern generator models, because rhythm generating neurons and genetically defined connectivity are not required in our model. Though our results are not in contradiction to current models, as existing neural mechanism and structures, not used in our model, can be expected to facilitate the kind of learning demonstrated here. Therefore, our model could be used to augment existing models.

Currently, it is accepted that animal locomotion is controlled by a central pattern generator in the spinal cord.

The Bcm rule allows a spinal cord model to learn rhythmic movements

The mammalian skin is a densely innervated soft tissue where embedded neural mechanoreceptors report to the brain the mechanical events arising at its surface. Up to now, models of the transformations of these events into neural signals relied on quasistatic or viscoelastic mechanical models of the skin. Here, we employed a model which, in addition to elasticity and viscosity, accounted for mass to accurately reproduce the propagation of mechanical waves observed in vivo. Skin dynamics converted sensory inputs into rapidly evolving spatiotemporal patterns that magnified the information made available to a population of mechanoreceptors. Accounting for dynamics greatly enhanced the separability of tactile inputs and was efficient for a large range of mechanical parameter values. This advantage vanished when these parameters were set to approximate the quasistatic or viscoelastic cases.

The mammalian skin is a densely innervated soft tissue where embedded neural mechanoreceptors report to the brain the mechanical events arising at its surface.

The Import of Skin Tissue Dynamics in Tactile Sensing

The dynamics of the human body can be described by the accelerations and masses of the different body parts (e.g., legs, arm, trunk). These body parts can exhibit specific coordination patterns with each other. In human walking, we found that the swing leg cooperates with the upper body and the stance leg in different ways (e.g., in-phase and out-of-phase in vertical and horizontal directions, respectively). Such patterns of self-assistance found in human locomotion could be of advantage in robotics design, in the design of any assistive device for patients with movement impairments. It can also shed light on several unexplained infrastructural features of the CNS motor control. Self-assistance means that distributed parts of the body contribute to an overlay of functions that are required to solve the underlying motor task. To draw advantage of self-assisting effects, precise and balanced spatiotemporal patterns of muscle activation are necessary. We show that the necessary neural connectivity infrastructure to achieve such muscle control exists in abundance in the spinocerebellar circuitry. We discuss how these connectivity patterns of the spinal interneurons appear to be present already perinatally but also likely are learned. We also discuss the importance of these insights into whole body locomotion for the successful design of future assistive devices and the sense of control that they could ideally confer to the user.

The dynamics of the human body can be described by the accelerations and masses of the different body parts (e.g., legs, arm, trunk).

Whole Body Coordination for Self-Assistance in Locomotion

Control of musculoskeletal systems depends on integration of voluntary commands and somatosensory feedback in the complex neural circuits of the spinal cord. It has been suggested that the various connectivity patterns that have been identified experimentally may result from the many transcriptional types that have been observed in spinal interneurons. We ask instead whether the muscle-specific details of observed connectivity patterns can arise as a consequence of Hebbian adaptation during early development, rather than being genetically ordained. We constructed an anatomically simplified model musculoskeletal system with realistic muscles and sensors and connected it to a recurrent, random neuronal network consisting of both excitatory and inhibitory neurons endowed with Hebbian learning rules. We then generated a wide set of randomized muscle twitches typical of those described during fetal development and allowed the network to learn. Multiple simulations consistently resulted in diverse and stable patterns of activity and connectivity that included subsets of the interneurons that were similar to “archetypical” interneurons described in the literature. We also found that such learning led to an increased degree of cooperativity between interneurons when performing larger limb movements on which it had not been trained. Hebbian learning gives rise to rich sets of diverse interneurons whose connectivity reflects the mechanical properties of the system. At least some of the transcriptomic diversity may reflect the effects of this process rather than the cause of the connectivity. Such a learning process seems better suited to respond to the musculoskeletal mutations that underlie the evolution of new species.

Control of musculoskeletal systems depends on integration of voluntary commands and somatosensory feedback in the complex neural circuits of the spinal cord.

A model for self-organization of sensorimotor function: spinal interneuronal integration

Recent spinal cord literature abounds with descriptions of genetic preprogramming and the molecular control of circuit formation. In this paper, we explore to what extent circuit formation based on learning rather than preprogramming could explain the selective formation of the monosynaptic projections between muscle spindle primary afferents and homonymous motoneurons. We adjusted the initially randomized gains in the neural network according to a Hebbian plasticity rule while exercising the model system with spontaneous muscle activity patterns similar to those observed during early fetal development. Normal connectivity patterns developed only when we modeled β motoneurons, which are known to innervate both intrafusal and extrafusal muscle fibers in vertebrate muscles but were not considered in previous literature regarding selective formation of these synapses in animals with paralyzed muscles. It was also helpful to correctly model the greatly reduced contractility of extrafusal muscle fibers during early development. Stronger and more coordinated muscle activity patterns such as observed later during neonatal locomotion impaired projection selectivity. These findings imply a generic functionality of a musculoskeletal system to imprint important aspects of its mechanical dynamics onto a neural network, without specific preprogramming other than setting a critical period for the formation and maturation of this general pattern of connectivity. Such functionality would facilitate the successful evolution of new species with altered musculoskeletal anatomy, and it may help to explain patterns of connectivity and associated reflexes that appear during abnormal development.

Recent spinal cord literature abounds with descriptions of genetic preprogramming and the molecular control of circuit formation.

A model for self-organization of sensorimotor function: the spinal monosynaptic loop

Biological cortical networks are potentially fully recurrent networks without any distinct output layer, where recognition may instead rely on the distribution of activity across its neurons. Because such biological networks can have rich dynamics, they are well-designed to cope with dynamical interactions of the types that occur in nature, while traditional machine learning networks may struggle to make sense of such data. Here we connected a simple model neuronal network (based on the 'linear summation neuron model' featuring biologically realistic dynamics (LSM), consisting of 10 of excitatory and 10 inhibitory neurons, randomly connected) to a robot finger with multiple types of force sensors when interacting with materials of different levels of compliance. Scope: to explore the performance of the network on classification accuracy. Therefore, we compared the performance of the network output with principal component analysis of statistical features of the sensory data as well as its mechanical properties. Remarkably, even though the LSM was a very small and untrained network, and merely designed to provide rich internal network dynamics while the neuron model itself was highly simplified, we found that the LSM outperformed these other statistical approaches in terms of accuracy.

Biological cortical networks are potentially fully recurrent networks without any distinct output layer, where recognition may instead rely on the distribution of activity across its neurons.

Dynamic neuronal networks efficiently achieve classification in robotic interactions with real-world objects

Hodgkin and Huxley's seminal neuron model describes the propagation of voltage spikes in 

 features crucial for understanding the neural code. We consider channel current fluctuations in a trisection of the Hodgkin-Huxley model, allowing an analytic-mechanistic explanation of these features and yielding consistently excellent matches with 

 recordings of cerebellar Purkinje neurons, which we use as model systems. This shows that the neuronal encoding is described conclusively by a soft-thresholding function having just three parameters.

Hodgkin and Huxley's seminal neuron model describes the propagation of voltage spikes in axons, but it cannot explain certain full-neuron features crucial for understanding the neural code. 

Channel current fluctuations conclusively explain neuronal encoding of internal potential into spike trains

Recurrent circuitry components are distributed widely within the brain, including both excitatory and inhibitory synaptic connections. Recurrent neuronal networks have potential stability problems, perhaps a predisposition to epilepsy. More generally, instability risks making internal representations of information unreliable. To assess the inherent stability properties of such recurrent networks, we tested a linear summation, non-spiking neuron model with and without a “dynamic leak”, corresponding to the low-pass filtering of synaptic input current by the RC circuit of the biological membrane. We first show that the output of this neuron model, in either of its two forms, follows its input at a higher fidelity than a wide range of spiking neuron models across a range of input frequencies. Then we constructed fully connected recurrent networks with equal numbers of excitatory and inhibitory neurons and randomly distributed weights across all synapses. When the networks were driven by pseudorandom sensory inputs with varying frequency, the recurrent network activity tended to induce high frequency self-amplifying components, sometimes evident as distinct transients, which were not present in the input data. The addition of a dynamic leak based on known membrane properties consistently removed such spurious high frequency noise across all networks. Furthermore, we found that the neuron model with dynamic leak imparts a network stability that seamlessly scales with the size of the network, conduction delays, the input density of the sensory signal and a wide range of synaptic weight distributions. Our findings suggest that neuronal dynamic leak serves the beneficial function of protecting recurrent neuronal circuitry from the self-induction of spurious high frequency signals, thereby permitting the brain to utilize this architectural circuitry component regardless of network size or recurrency.

Recurrent circuitry components are distributed widely within the brain, including both excitatory and inhibitory synaptic connections.

Born in the intersection
of AI and Neuroscience.

IntuiCell is a Lund University spin-off with presence in Lund and Stockholm.

For three decades our founders challenged the common consensus
of one of the biggest scientific questions of our time:

How does the brain work?

Today, we apply those groundbreaking neurophysiological insights
to redesign artificial intelligence. Creating AI that is actually intelligent.

Explore our contrarian research on the mechanisms by which the
brain predicts the world.

All Publications.

Born in the intersectionof AI and Neuroscience.

IntuiCell is a Lund University spin-off with presence in Lund and Stockholm.

For three decades our founders challenged the common consensus of one of the biggest scientific questions of our time:

How does the brain work?

Today, we apply those groundbreaking neurophysiological insights to redesign artificial intelligence. Creating AI that is actually intelligent.

Explore our contrarian research on the mechanisms by which the brain predicts the world.

All Publications.

Born in the intersection
of AI and Neuroscience.

For three decades our founders challenged the common consensus
of one of the biggest scientific questions of our time:

Today, we apply those groundbreaking neurophysiological insights
to redesign artificial intelligence. Creating AI that is actually intelligent.

Explore our contrarian research on the mechanisms by which the
brain predicts the world.