Insight-based computing (mathematical theory of human thinking)

 

Human brain is an idea generator. It is asynchronous (insights come when appropriate) and universal (ideas may be of different types, represent: events in the outer world, internal states of the body, own actions, even system organization of the brain itself). Very flexible device for real-world conditions.

The principle of insight is justified in different aspects. Perception faces the infinite data flow. In order to ensure elementary workability, it is necessary to reduce it to some reasonable amount. Sensory insights capture a finite portion from the infinite input. On the other hand, let's consider internal data encoding. It uses patterns of neuronal activity, but this activity requires resources. Out of existing theoretical concepts, a Finite-State Automaton is the most relevant to describe human computing. If FSA is implemented as a mechanical product, its states lock themselves using a trigger. In the brain, such triggers need to be specially implemented using a positive biofeedback. When a valuable idea emerges, it fixes itself and neurons spend energy for some period of time.

In principle, insight is a state. Then why not use the well elaborated formalism of a Finite-State Automaton? Because the brain as a whole is not FSA. You will need to define many such blocks, link them with each other, and consider a complicated network of FSAs. Instead, insight provides a universal framework. Just define a typology. For an in-depth description, you may associate each type with the corresponding part of the brain and study neuronal mechanisms, but this is not necessary. Insights in some part form a condition for definite insights in another. You can describe everything on functional level.

The concept of insight looks like a panacea, but unfortunately it is of small practical use without concrete recipes of implementation. It is just a universal term which makes it possible to denote anything that you may ever need. To turn it into a workable computer, we need to describe how different types of insight operate in the nervous system. This typology of low-level insights and relations between them may serve as an axiomatic system for a mathematical theory describing human thinking. Otherwise, it forms a rather simple foundation of the human computing system. Out of them, 3 levels of human software - firmware, system, and application - are created.

 

Typology

  Sensory

Sensory insights ensure understanding of the world. They capture a portion of reality and represent it as a discrete concept.

  Motor and motivational

While perception is the input of the brain, the motor system is its output. This part begins with motivation which sets the goal. Various ideas about necessary actions follow suit.

    Algorithms

Algorithms are complex actions. When computers construct algorithms, they usually check various combinations and different order of elementary actions. In humans, all this emerges as a single insight.

  Emotional

Emotional images provide a state of the brain and of the organism as a whole best suitable for a particular activity. Tonic emotions are such states which may persist for hours, days, and even longer. Phasic emotions - swift pulses of activation which prompt switching of states. They also may serve as asynchronous substitutions of pacemakers which provide insights about timing.

  Linguistic

Language is a separate sensory-motor channel. In this case, discrete nature and relation to semiotics is especially obvious. Again, humans often can't explain how they choose appropriate words to express particular situations.

 

Mechanisms

Insight is essentially non-algorithmic. Moreover, it has problems with cause-consequence relations. It goes in the opposite direction. Insight is tightly related to reflex, but relations are complicated. Reflex is a direct expression of a cause-consequence pair. Instead, insight searches for a solution which fits some requirements. It goes from the consequence. The simplest implementation of this non-algorithmic tool is the trial & error method. You generate various images in random order and constantly monitor them. As soon as you feel that it is fine, the process stops.

On the other hand, many types of insight are variants of reflex. The idea may be directly prompted by some event which literally serves as a stimulus. In other cases, there are no visible stimuli, but there is a context which replaces them. As a more complicated variant, there is no stimulus, no appropriate context, but you was busy. Insight happened as soon as you was released and got necessary computational resources.

In this aspect, insight is a generalization of reflex. It does not require any sharp, distinct stimulus. Moreover, it may describe cause-consequence relations of internal events, while reflex is primarily for the behavioral level. Linking different insights via associations, you will get a chain of images that is an inference engine.

 

 

Copyright (c) I. Volkov, May 14, 2018

 

www.000webhost.com