Tensor network theory

Last updated

Tensor network theory is a theory of brain function (particularly that of the cerebellum) that provides a mathematical model of the transformation of sensory space-time coordinates into motor coordinates and vice versa by cerebellar neuronal networks. The theory was developed by Andras Pellionisz and Rodolfo Llinas in the 1980s as a geometrization of brain function (especially of the central nervous system) using tensors. [1] [2]

Contents

Metric tensor that transforms input covariant tensors into output contravariant tensors. These tensors can be used to mathematically describe cerebellar neuronal network activities in the central nervous system. Metrictensor.svg
Metric tensor that transforms input covariant tensors into output contravariant tensors. These tensors can be used to mathematically describe cerebellar neuronal network activities in the central nervous system.

History

Neuronal network schematic. The sensory inputs get transformed by the hidden layer representing the central nervous system which in turn outputs a motor response. Neuronal Network scheme.JPG
Neuronal network schematic. The sensory inputs get transformed by the hidden layer representing the central nervous system which in turn outputs a motor response.

Geometrization movement of the mid-20th century

The mid-20th century saw a concerted movement to quantify and provide geometric models for various fields of science, including biology and physics. [3] [4] [5] The geometrization of biology began in the 1950s in an effort to reduce concepts and principles of biology down into concepts of geometry similar to what was done in physics in the decades before. [3] In fact, much of the geometrization that took place in the field of biology took its cues from the geometrization of contemporary physics. [6] One major achievement in general relativity was the geometrization of gravitation. [6] This allowed the trajectories of objects to be modeled as geodesic curves (or optimal paths) in a Riemannian space manifold. [6] During the 1980s, the field of theoretical physics also witnessed an outburst of geometrization activity in parallel with the development of the Unified Field Theory, the Theory of Everything, and the similar Grand Unified Theory, all of which attempted to explain connections between known physical phenomena. [7]

The geometrization of biology in parallel with the geometrization of physics covered a multitude of fields, including populations, disease outbreaks, and evolution, and continues to be an active field of research even today. [8] [9] By developing geometric models of populations and disease outbreaks, it is possible to predict the extent of the epidemic and allow public health officials and medical professionals to control disease outbreaks and better prepare for future epidemics. [8] Likewise, there is work being done to develop geometric models for the evolutionary process of species in order to study the process of evolution, the space of morphological properties, the diversity of forms and spontaneous changes and mutations. [9]

Geometrization of the brain and tensor network theory

Around the same time as all of the developments in the geometrization of biology and physics, some headway was made in the geometrization of neuroscience. At the time, it became more and more necessary for brain functions to be quantified in order to study them more rigorously. Much of the progress can be attributed to the work of Pellionisz and Llinas and their associates who developed the tensor network theory in order to give researchers a means to quantify and model central nervous system activities. [1] [2]

In 1980, Pellionisz and Llinas introduced their tensor network theory to describe the behavior of the cerebellum in transforming afferent sensory inputs into efferent motor outputs. [1] They proposed that intrinsic multidimensional central nervous system space could be described and modeled by an extrinsic network of tensors that together describe the behavior of the central nervous system. [1] By treating the brain as a "geometrical object" and assuming that (1) neuronal network activity is vectorial and (2) that the networks themselves are organized tensorially, brain function could be quantified and described simply as a network of tensors. [1] [2]

Example

Six rotational axes about which the extraocular muscles turn the eye and the three rotational axes about which the vestibular semicircular canals measure head-movement. According to tensor network theory, a metric tensor can be determined to connect the two coordinate systems. VOR coordinates.PNG
Six rotational axes about which the extraocular muscles turn the eye and the three rotational axes about which the vestibular semicircular canals measure head-movement. According to tensor network theory, a metric tensor can be determined to connect the two coordinate systems.

Vestibulo-ocular reflex

In 1986, Pellionisz described the geometrization of the "three-neuron vestibulo-ocular reflex arc" in a cat using tensor network theory. [10] The "three-neuron vestibulo-ocular reflex arc" is named for the three neuron circuit the arc comprises. Sensory input into the vestibular system (angular acceleration of the head) is first received by the primary vestibular neurons which subsequently synapse onto secondary vestibular neurons. [10] These secondary neurons carry out much of the signal processing and produce the efferent signal heading for the oculomotor neurons. [10] Prior to the publishing of this paper, there had been no quantitative model to describe this "classic example of a basic sensorimotor transformation in the central nervous system" which is precisely what tensor network theory had been developed to model. [10]

Here, Pellionisz described the analysis of the sensory input into the vestibular canals as the covariant vector component of tensor network theory. Likewise, the synthesized motor response (reflexive eye movement) is described as the contravariant vector component of the theory. By calculating the neuronal network transformations between the sensory input into the vestibular system and the subsequent motor response, a metric tensor representing the neuronal network was calculated. [10]

The resulting metric tensor allowed for accurate predictions of the neuronal connections between the three intrinsically orthogonal vestibular canals and the six extraocular muscles that control the movement of the eye. [10]

Applications

Neural Networks and Artificial Intelligence

Neural networks modeled after the activities of the central nervous system have allowed researchers to solve problems impossible to solve by other means. Artificial neural networks are now being applied in various applications to further research in other fields. One notable non-biological application of the tensor network theory was the simulated automated landing of a damaged F-15 fighter jet on one wing using a "Transputer parallel computer neural network". [11] The fighter jet's sensors fed information into the flight computer which in turn transformed that information into commands to control the plane's wing-flaps and ailerons to achieve a stable touchdown. This was synonymous to sensory inputs from the body being transformed into motor outputs by the cerebellum. The flight computer's calculations and behavior was modeled as a metric tensor taking the covariant sensor readings and transforming it into contravariant commands to control aircraft hardware. [11]

Related Research Articles

Ataxia is a neurological sign consisting of lack of voluntary coordination of muscle movements that can include gait abnormality, speech changes, and abnormalities in eye movements, that indicates dysfunction of parts of the nervous system that coordinate movement, such as the cerebellum.

<span class="mw-page-title-main">Neural Darwinism</span> Theory in neurology

Neural Darwinism is a biological, and more specifically Darwinian and selectionist, approach to understanding global brain function, originally proposed by American biologist, researcher and Nobel-Prize recipient Gerald Maurice Edelman. Edelman's 1987 book Neural Darwinism introduced the public to the theory of neuronal group selection (TNGS) – which is the core theory underlying Edelman's explanation of global brain function.

<span class="mw-page-title-main">Cerebellum</span> Structure at the rear of the vertebrate brain, beneath the cerebrum

The cerebellum is a major feature of the hindbrain of all vertebrates. Although usually smaller than the cerebrum, in some animals such as the mormyrid fishes it may be as large as it or even larger. In humans, the cerebellum plays an important role in motor control. It may also be involved in some cognitive functions such as attention and language as well as emotional control such as regulating fear and pleasure responses, but its movement-related functions are the most solidly established. The human cerebellum does not initiate movement, but contributes to coordination, precision, and accurate timing: it receives input from sensory systems of the spinal cord and from other parts of the brain, and integrates these inputs to fine-tune motor activity. Cerebellar damage produces disorders in fine movement, equilibrium, posture, and motor learning in humans.

<span class="mw-page-title-main">Sense of balance</span> Physiological sense regarding posture

The sense of balance or equilibrioception is the perception of balance and spatial orientation. It helps prevent humans and nonhuman animals from falling over when standing or moving. Equilibrioception is the result of a number of sensory systems working together; the eyes, the inner ears, and the body's sense of where it is in space (proprioception) ideally need to be intact.

<span class="mw-page-title-main">Brainstem</span> Posterior part of the brain, adjoining and structurally continuous

The brainstem is the stalk-like part of the brain that interconnects the cerebrum and diencephalon with the spinal cord. In the human brain, the brainstem is composed of the midbrain, the pons, and the medulla oblongata. The midbrain is continuous with the thalamus of the diencephalon through the tentorial notch.

<span class="mw-page-title-main">Vestibulocochlear nerve</span> Cranial nerve VIII, for hearing and balance

The vestibulocochlear nerve or auditory vestibular nerve, also known as the eighth cranial nerve, cranial nerve VIII, or simply CN VIII, is a cranial nerve that transmits sound and equilibrium (balance) information from the inner ear to the brain. Through olivocochlear fibers, it also transmits motor and modulatory information from the superior olivary complex in the brainstem to the cochlea.

<span class="mw-page-title-main">Neural pathway</span> Connection formed between neurons that allows neurotransmission

In neuroanatomy, a neural pathway is the connection formed by axons that project from neurons to make synapses onto neurons in another location, to enable neurotransmission. Neurons are connected by a single axon, or by a bundle of axons known as a nerve tract, or fasciculus. Shorter neural pathways are found within grey matter in the brain, whereas longer projections, made up of myelinated axons, constitute white matter.

<span class="mw-page-title-main">Rodolfo Llinás</span> Colombian neuroscientist (born 1934)

Rodolfo Llinás Riascos is a Colombian and American neuroscientist. He is currently the Thomas and Suzanne Murphy Professor of Neuroscience and Chairman Emeritus of the Department of Physiology & Neuroscience at the NYU School of Medicine. Llinás has published over 800 scientific articles.

<span class="mw-page-title-main">Climbing fiber</span> Series of neuronal projections from the inferior olivary nucleus located in the medulla oblongata

Climbing fibers are the name given to a series of neuronal projections from the inferior olivary nucleus located in the medulla oblongata.

<span class="mw-page-title-main">Purkinje cell</span> Specialized neuron in the cerebellum

Purkinje cells, or Purkinje neurons, are a class of GABAergic inhibitory neurons located in the cerebellum. They are named after their discoverer, Czech anatomist Jan Evangelista Purkyně, who characterized the cells in 1839.

<span class="mw-page-title-main">Dentate nucleus</span> Nucleus in the centre of each cerebellar hemisphere

The dentate nucleus is a cluster of neurons, or nerve cells, in the central nervous system that has a dentate – tooth-like or serrated – edge. It is located within the deep white matter of each cerebellar hemisphere, and it is the largest single structure linking the cerebellum to the rest of the brain. It is the largest and most lateral, or farthest from the midline, of the four pairs of deep cerebellar nuclei, the others being the globose and emboliform nuclei, which together are referred to as the interposed nucleus, and the fastigial nucleus. The dentate nucleus is responsible for the planning, initiation and control of voluntary movements. The dorsal region of the dentate nucleus contains output channels involved in motor function, which is the movement of skeletal muscle, while the ventral region contains output channels involved in nonmotor function, such as conscious thought and visuospatial function.

Neural engineering is a discipline within biomedical engineering that uses engineering techniques to understand, repair, replace, or enhance neural systems. Neural engineers are uniquely qualified to solve design problems at the interface of living neural tissue and non-living constructs.

<span class="mw-page-title-main">Neural oscillation</span> Brainwaves, repetitive patterns of neural activity in the central nervous system

Neural oscillations, or brainwaves, are rhythmic or repetitive patterns of neural activity in the central nervous system. Neural tissue can generate oscillatory activity in many ways, driven either by mechanisms within individual neurons or by interactions between neurons. In individual neurons, oscillations can appear either as oscillations in membrane potential or as rhythmic patterns of action potentials, which then produce oscillatory activation of post-synaptic neurons. At the level of neural ensembles, synchronized activity of large numbers of neurons can give rise to macroscopic oscillations, which can be observed in an electroencephalogram. Oscillatory activity in groups of neurons generally arises from feedback connections between the neurons that result in the synchronization of their firing patterns. The interaction between neurons can give rise to oscillations at a different frequency than the firing frequency of individual neurons. A well-known example of macroscopic neural oscillations is alpha activity.

<span class="mw-page-title-main">Flocculus</span>

The flocculus is a small lobe of the cerebellum at the posterior border of the middle cerebellar peduncle anterior to the biventer lobule. Like other parts of the cerebellum, the flocculus is involved in motor control. It is an essential part of the vestibulo-ocular reflex, and aids in the learning of basic motor skills in the brain.

<span class="mw-page-title-main">Vestibulospinal tract</span> Neural tract in the central nervous system

The vestibulospinal tract is a neural tract in the central nervous system. Specifically, it is a component of the extrapyramidal system and is classified as a component of the medial pathway. Like other descending motor pathways, the vestibulospinal fibers of the tract relay information from nuclei to motor neurons. The vestibular nuclei receive information through the vestibulocochlear nerve about changes in the orientation of the head. The nuclei relay motor commands through the vestibulospinal tract. The function of these motor commands is to alter muscle tone, extend, and change the position of the limbs and head with the goal of supporting posture and maintaining balance of the body and head.

Models of neural computation are attempts to elucidate, in an abstract and mathematical fashion, the core principles that underlie information processing in biological nervous systems, or functional components thereof. This article aims to provide an overview of the most definitive models of neuro-biological computation as well as the tools commonly used to construct and analyze them.

<span class="mw-page-title-main">Anatomy of the cerebellum</span> Structures in the cerebellum, a part of the brain

The anatomy of the cerebellum can be viewed at three levels. At the level of gross anatomy, the cerebellum consists of a tightly folded and crumpled layer of cortex, with white matter underneath, several deep nuclei embedded in the white matter, and a fluid-filled ventricle in the middle. At the intermediate level, the cerebellum and its auxiliary structures can be broken down into several hundred or thousand independently functioning modules or compartments known as microzones. At the microscopic level, each module consists of the same small set of neuronal elements, laid out with a highly stereotyped geometry.

<span class="mw-page-title-main">Cerebellar granule cell</span> Thick granular layer of the cerebellar cortex

Cerebellar granule cells form the thick granular layer of the cerebellar cortex and are among the smallest neurons in the brain. Cerebellar granule cells are also the most numerous neurons in the brain: in humans, estimates of their total number average around 50 billion, which means that they constitute about 3/4 of the brain's neurons.

<span class="mw-page-title-main">Granule cell</span> Type of neuron with a very small cell body

The name granule cell has been used for a number of different types of neurons whose only common feature is that they all have very small cell bodies. Granule cells are found within the granular layer of the cerebellum, the dentate gyrus of the hippocampus, the superficial layer of the dorsal cochlear nucleus, the olfactory bulb, and the cerebral cortex.

<span class="mw-page-title-main">Outline of the human nervous system</span> Overview of and topical guide to the human nervous system

The following diagram is provided as an overview of and topical guide to the human nervous system:

References

  1. 1 2 3 4 5 Pellionisz, A., Llinás, R. (1980). "Tensorial Approach To The Geometry Of Brain Function: Cerebellar Coordination Via A Metric Tensor" (PDF). Neuroscience. 5 (7): 1125––1136. doi:10.1016/0306-4522(80)90191-8. PMID   6967569. S2CID   17303132.{{cite journal}}: CS1 maint: multiple names: authors list (link)[ dead link ]
  2. 1 2 3 Pellionisz, A., Llinás, R. (1985). "Tensor Network Theory Of The Metaorganization Of Functional Geometries In The Central Nervous System". Neuroscience. 16 (2): 245–273. doi:10.1016/0306-4522(85)90001-6. PMID   4080158. S2CID   10747593.{{cite journal}}: CS1 maint: multiple names: authors list (link)[ dead link ]
  3. 1 2 Rashevsky, N (1956). "The Geometrization of Biology". Bulletin of Mathematical Biophysics. 18: 31–54. doi:10.1007/bf02477842.
  4. ↑ Palais, Richard (1981). "The Geometrization of Physics" (PDF): 1–107.{{cite journal}}: Cite journal requires |journal= (help)
  5. ↑ Mallios, Anastasios (August 2006). "Geometry and physics of today". International Journal of Theoretical Physics. 45 (8): 1552–1588. arXiv: physics/0405112 . doi:10.1007/s10773-006-9130-3. S2CID   17514844.
  6. 1 2 3 Bailly, Francis (2011). Mathematics and the Natural Sciences: The Physical Singularity of Life. Imperial College Press. ISBN   978-1848166936.
  7. ↑ KALINOWSKI, M (1988). "The Program of Geometrization of Physics: Some Philosophical Remarks". Synthese. 77 (2): 129–138. doi:10.1007/bf00869432. S2CID   46977351.
  8. 1 2 Kahil, M (2011). "Geometrization of Some Epidemic Models". Wseas Transactions on Mathematics. 10 (12): 454–462.
  9. 1 2 Nalimov, W (2011). "Geometrization of biological ideas: probabilistic model of evolution". Zhurnal Obshchei Biologii. 62 (5): 437–448. PMID   11605554.
  10. 1 2 3 4 5 6 Pellionisz, Andras; Werner Graf (October 1986). "Tensor Network Model of the "Three-Neuron Vestibulo-Ocular Reflex-Arc" in Cat". Journal of Theoretical Neurobiology. 5: 127–151.
  11. 1 2 Pellionisz, Andras (1995). "Flight Control by Neural Nets: A Challenge to Government/Industry/Academia". International Conference on Artificial Neural Networks.
This page is based on this Wikipedia article
Text is available under the CC BY-SA 4.0 license; additional terms may apply.
Images, videos and audio are available under their respective licenses.