Multilevel Development of Cognitive Abilities for Artificial Intelligence

Motivation

In biological intelliegent systems there are multiple mechanisms working in congruence on multiple levels, both at the structural and neurobiological level to develop complex cognitive abilities. What remains unknown is which mechanisms are necessary and sufficent to synthetically replicate these cognitive abilities for artificial intelligence. A neurocomputational model is offered of the devloping brain that spans the sensorimotor, cognitive, and conscious levels. We will look at how we can replicated this model artifically using known computational algorithms and data strucutres.

The ending goal for the final model is to solve three tasks that increase in complexity. The first task is visual recognition, then cogintive manipulation, then maintenance of consious percepts. These tasks contain two fundemental mechainisms for the multilevel development of cognitive abilities in biolgical neural networks.

The first fundemental mechanism of coginition seen in biolgical neural networks is epigensis with Hebbian learning at the local scale and with reinforcment learning at the global scale. Again, by modeling local and global scales, the hope is to better replicate biological multilevel congruence of mechanisms to acheive higher advanced cognition for artifical intelligence.

Neurobiological Models

Local Epigenesis of STDP Regulated Synapses

Image Source: Multilevel development of cognitive abilities in an artificial neural network \(^{[2]}\)

Adaptive Topologies

To continue.

To continue.

Structural Models

Prelimenaries

Topological Neuroscience

You can read about Topological Neuroscience on my Topological Neuroscience page.

A Large-Scale Model of the Functioning Brain

Image Source: A Large-Scale Model of the Functioning Brain \(^{[1]}\)

System Models

Reinfrocment Learning

To continue.

Glossary

Hebbian Theory

Hebbian theory attempts to explain synaptic plasticity or adaptation of brain neurons during the learning process. The neuropsychological theory introduced first by Donald Hebb, claims that the increase of synaptic efficacy is caused by the presynaptic cells repeated and persistent stimulation of a postsynaptic cell. An excerpt from Hebb's book, The Orginization of Behavior, explains his idea further:

"When an axon of cell A is near enough to excite a cell B and repeatedly or persistently takes part in firing it, some growth process or metabolic change takes place in one or both cells such that A’s efficiency, as one of the cells firing B, is increased."