You can edit almost every page by Creating an account. Otherwise, see the FAQ.

Practopoiesis

From EverybodyWiki Bios & Wiki






And can I suggest that discussion of the article (as opposed to editing) be done on the associated talk page? --Brian Josephson (talk) 09:14, 26 May 2017 (UTC)

Practopoiesis is a theory by Danko Nikolić proposing that life can be understood as a hierarchy of adaptive processes; adaptive processes at lower levels of organization (such as evolution by natural selection) determine properties of the adaptive mechanisms at higher levels of adaptive organization (such as genes).[1] The theory of practopoiesis states that life can be explained by adaptive mechanisms that operate at a total of four levels of organization.

These levels are (from lowest to highest) :

Evolution → Gene expression → Cell anapoiesis → Interaction with the environment

Cell anapoiesis consists of internal homeostatic mechanisms that change the properties of the cell but do not directly involve gene expression. Anapoietic mechanisms do not directly affect the environment of the cell.

The theory has been used to generate new counterintuitive empirically testable predictions[2][3], to propose improvement of control systems [4][5] and artificial intelligence[6][7], and to guide creation of efficient methods for acquiring new languages.[4]

Notably, the theory introduced the level of anapoiesis, an adaptive process not previously formulated as a specific adaptive mechanisms despite evidence of numerous homeostatic mechanisms. Anapoiesis stands for "reconstruction of knowledge"—transforming knowledge from a general form to a specific one.[citation needed]

Practopoiesis was initially developed the explain the functioning of the brain and the emergence of mental phenomena. Its main tenet holds that all our semantic knowledge and procedural knowledge (skills) are stored in a form of "fast learning" mechanisms at the level of anapoiesis. Mental operations therefore occur through application of anapoietic mechanisms. The theory proposes that these fast adaptive mechanisms have been acquired (or poietically created) over an individual's lifetime through interaction with the environment and expression of genes, i.e., by neural plasticity.

That way, practopoiesis challenges current neuroscience doctrines based on a widespread assumption that thinking is synonymous with neural spiking activity (cell function). In contrast, practopoiesis asserts that mental operations primarily occur at the anapoietic level — i.e., that minds emerge from fast homeostatic mechanisms. In other words, it is the homeostatic "fast learning" that is responsible for implementing mental operations such as perception, attention, recall from memory and decision-making. Importantly, it follows that neural spiking activity operates at a level higher than anapoiesis (cell function), and thus, is not directly responsible for mental operations. Spiking activity is necessary for closing sensory-motor loops that create behavior. Anapoietic mechanisms dynamically reorganize these loops, and, in doing so, form a mind.

Hierarchical interactions with the environment[edit]

Hierarchical interaction between adaptive mechanisms forming a practopoietic loop of causation. According to the theory, each mechanism must receive specific feedback from the external world.

Practopoiesis provides an explanation how different levels of adaptive mechanisms mutually interact: Each adaptive level must both affect the environment and receive feedback from the environment. A lower level indirectly affects the environment—by first affecting the mechanisms at higher levels of organization. Each lower level must operate at a slower time scale than the level above it. These multi-level interactions are illustrated in the figure on the right.

Anapoiesis[edit]

File:FastVsSlowAdaptiveMechanisms.png
Slow adaptive mechanism based on growing synaptic connections (left) and fast adaptive mechanism based on changing membrane properties (right). Red color indicates increased difficulty of passing signals.

Anapoiesis refers collectively to all homeostatic mechanisms that do not involve expression of genes. In the case of anapoiesis, the cell uses its existing proteins to adjust its functionality, requiring no immediate synthesis of proteins. These adaptations can thus occur much faster than those based on gene expression. One can think of anapoietic mechanisms as cell control mechanisms already prepared through gene expression, waiting to be quickly engaged as soon as there is a need. Membrane proteins should play important role in anapoiesis.

Importantly, practopoietic theory asserts that anapoiesis is primarily responsible for mental operations. For example, when we perceive a person, the brain is proposed to generate not only a pattern of neural activity in response to that person, but also that the neurons quickly adjust, some by quickly habituating and others by quickly sensitizing. This pattern of habitation and sensitization, according to the theory, are fast homeostatic mechanisms that prepare us for the interaction with that person. This is the process of anapoiesis. And, when the preparation is successful—resulting in sufficiently good preparation for interacting with that person—we have, according to the theory, correctly perceived someone. The same process is also responsible for attention towards that person, deciding whether to interact with that person, thinking about that person, and so on along the entire palette of cognitive operations.

Moreover, these adaptive mechanisms (such as sensitization and habituation) are not fixed and shared across all neurons. Rather, each neuron learns individually what is an appropriate adaptation action in a given situation. Hence, each neuron has a unique knowledge of when to sensitize or habituate. This knowledge on when to sensitize vs. habituate represents our semantic knowledge acquired over lifetime. Therefore, according to practopoiesis, knowledge stored in synapses is not the only important one.

Wang[4] proposed that anapoietic skills grow primarily as a result socio-cultural interactions, and thus reflect socio-cultural knowledge. This is in contrast to corporeal-temporal-spatial knowledge that lays immediately bellow (created by expression of genes) and semiotic knowledge above the anapoietic level (behavior and interactions).

Advantages of adaptive hierarchies[edit]

An adaptive hierarchy can take full advantage of inductive bias; when a learning system is biased such that its assumptions approximate reality well, learning is effective. In a practopoietic hierarchy of adaptive components, components lower on the hierarchy shape the biases (assumptions) of the components higher on the hierarchy. For example, evolution determines the learning biases of gene expression mechanisms; gene expression shapes the assumptions of the anapoietic mechanisms; anapoiesis shapes the assumptions of the processing based on neural activity.

As a result, a hierarchical architecture achieves high levels of intelligence with a relatively small system size[8]. Such a system does not need to store complete information on how to adapt in all possible situations. Instead, only general rules for entire classes of environments are learned. These rules are then applied through poietic hierarchy. For example, a neural network needs not learn how to interact specifically with all possible chairs, but instead anapoietic mechanisms contain knowledge of rules on how to interact with chairs in general.

A unique approach to explaining mind[edit]

File:Tri-traversal theory of mind.png
Tri-traversal theory of mind

The key difference from the classical brain theory is that the primary brain mechanism for implementing mental operations is not presumed to rely on the electrochemical activity executed by neural circuitry (i.e., on the network "computation" based on excitation and inhibition). Instead, according to practopoiesis, the emergence of the mind depends critically on the mechanisms making quick adaptive changes to that circuitry: anapoietic mechanisms. These fast adjustment mechanisms are hypothesized to be closely related to the well-known phenomenon of (fast) neural adaptation which adjust routing and computational properties of excitatory-inhibitory networks.

File:Practopoiesis Little Red Riding Hood.png
Illustrated routing of neural activity by a pattern of membrane adaptations. A percept of Little Red Riding Hood is favoured over a grandmother.

For example, when deciding which person one perceives—e.g., Grandmother or Little Red Riding Hood—the fast adaptive mechanisms of neurons may route neural activity in the unique pattern that prescribes interaction with that person just as they would when one interacts with Grandmother, or with Little Red Riding Hood. This way, neural adaptations determine the percept. At the same time, they determine the decision on how to behave, the direction of attention, contents of working memory, and so on.

Relationship to biology and cybernetics[edit]

Practopoiesis is closely related to a number of biological theories[1] and cybernetic concepts[4]. These include evolution by natural selection, and homeostasis and allostasis. For example, practopoiesis asserts that Lamarckian inheritance cannot possibly work because a species would not be able to accumulate general knowledge that requires a larger number of generations. Past knowledge would be quickly lost over new one.[1]

The theory is founded in two theorems of cybernetics: practopoietic hierarchy is organized such to maximize the requisite variety of responses that an organism generates in its interaction with the environment.[4] Through its hierarchy, practopoiesis also respects the fact that such successful interaction requires the organism to effectively become a model of its environment.

Practopoiesis is also related to autopoiesis; practopoiesis states that autopoiesis of an organism or a cell occurs through allopoietic interactions among its components.

Implications for philosophy of mind and artificial intelligence[edit]

Due to emphasizing interaction with the environment and the abandonment of representation, practopoiesis is compatible with the concepts of embodied and embedded cognition.[9]

Mittal and Rainey hypothesize that strong emergent systems are practopoietic systems. Moreover, emergent behavior should be controlled through practopoietic hierarchy. [5]

It is argued that only a hierarchy of slow and fast learning mechanisms can produce Pierce's abductive reasoning and Searle's understanding. Solutions to both problems are based in the proposal that semantics of our minds emerge through the very nature of anapoietic mechanisms, which necessarily possess knowledge in a more general form than do neural networks.[1]

Several proposals argue that the boosted adaptability that results from practopoietic hierarchy will be necessary for advances in artificial intelligence.[6] [10]and brain inspired computing[7]

Practopoietic hierarchy has been used to propose new methods for more efficient language acquisition.[4] Namely, enacting the communicated message and active behavioral interaction during communication should be critical for successfully learning a new language.

Ji proposes that practopoietic hierarchy can help understand Peircian semiotic triade.[11]

See also[edit]

References[edit]

  1. 1.0 1.1 1.2 1.3 Nikolić, Danko. Practopoiesis: Or how life fosters a mind. Journal of Theoretical Biology 373 (2015): 40-61.
  2. Eriksson, David (2017). A principle for describing and verifying brain mechanisms using ongoing activity. Frontiers in Neural Circuits, 11.
  3. Nikolić, Danko. (2016) Testing the theory of practopoiesis using closed loops. In: Closed Loop Neuroscience. Ed. Ahmed El Hady. Academic Press.
  4. 4.0 4.1 4.2 4.3 4.4 4.5 Wang, Jianfen (2016). An Ecology of Literacy: A Context-based Inter-disciplinary Curriculum for Chinese as a Foreign Language (Doctoral dissertation, The Ohio State University).
  5. 5.0 5.1 Mittal, Saurabh & Rainey, Larry (2015, July). Harnessing emergence: The control and design of emergent behavior in system of systems engineering. In Proceedings of the Conference on Summer Computer Simulation (pp. 1-10). Society for Computer Simulation International.
  6. 6.0 6.1 Miccoli, Anthony (2016). Posthuman Trajectories: Cartesian Logic and Ethical Technoprogressivism. Word and Text, A Journal of Literary Studies and Linguistics, 6(1), 114-129.
  7. 7.0 7.1 Yu, Shan (2016). New challenge for bionics—brain-inspired computing. Zoological Research, 37(5), 261.
  8. Faggella, Daniel (2016) Kindergarten for Robots Is a Lot Like Kindergarten for Kids. motherboard.vice.com
  9. Cornelis, H., & Coop, A. D. (2014). Afference copy as a quantitative neurophysiological model for consciousness. Journal of Integrative Neuroscience, 13(02), 363-402.
  10. Nikolić, D. (2017). Why deep neural nets cannot ever match biological intelligence and what to do about it?. International Journal of Automation and Computing, 14(5), 532-541.
  11. Ji, S. (2017). Wave-Particle Duality and Quantity-Quality Complementarity in Natural and Human Sciences: Implications for Credition Research. In Processes of Believing: The Acquisition, Maintenance, and Change in Creditions (pp. 417-433). Springer International Publishing.

External links[edit]

Rejected due to article by the same name, which redirects to the draft page.[edit]


This article "Practopoiesis" is from Wikipedia. The list of its authors can be seen in its historical and/or the page Edithistory:Practopoiesis. Articles copied from Draft Namespace on Wikipedia could be seen on the Draft Namespace of Wikipedia and not main one.