Neuron Field Structure
The Cerebro typically consists of several areas, each with its own functional focus. These areas are interconnected not only functionally but also spatially.This application simulates both aspects using so-called neuron fields. Technically, neuron fields are containers for neurons, but unlike their biological counterparts, they have no spatial orientation. As containers, they represent a coordinate-free space. In addition, they are responsible for implementing neurotransmitter functionality.
The neuron field is organized as a tree structure, with the Cerebro serving as the root. Several predefined fields fulfill specific roles:
- 1) Receptor Fields
Receptor fields contain neurons that act as input interfaces for reinforcement learning. From a biological perspective, they correspond to sensory systems such as vision, hearing, or the sense of smell.
- 2) Effector Fields
Effector fields contain neurons designated for output and serve as response interfaces for reinforcement learning. Biologically, they represent neurons that control, for example, muscular activity.
- 3) Feedback Fields
Feedback neurons function similarly to input interfaces, but they receive feedback signals from the reinforcement learning system. These neurons can be placed within dedicated feedback fields.
- 4) Other Fields
All remaining fields have no predefined purpose and are collectively referred to as other. However, they may be assigned names inspired by biological structures, such as:
- Sensors: retinal receptors, auditory cortex, somatosensory area
- Speech: speech center, Broca’s area, Wernicke’s area
- Emotions: amygdala, limbic system, reward circuit, striatum
- Integration: thalamus, basal ganglia, cerebellum
- Other: prefrontal cortex, executive control, working memory
The structure of the fields can—and must—be redefined for each scenario, providing maximum flexibility. For example, the Cerebro field may contain receptor fields, effector fields, feedback fields, and additional nested substructures. Thanks to the tree architecture, any field can be recursively subdivided. Neuron fields also handle the distribution of neurotransmitters. Neurons may signal their field when they experience a surplus or deficit. Further details are described in the section on Neuroplasticity.
// NeuronField interface example
public interface NeuronField extends CachedObject {
void add(Neuron neuron);
List getChildren();
void add(List neurons);
void visitFields(boolean breadth, Consumer visitor);
....
// Neurotransmitter functionality
....
}