Pattern Recognition and Pattern Resonance

Consider a line by line process like Wolfram's cellular automata. For each line, there are a number of cells. Each cell is labelled with a number, called a pattern process id. For each step, each pattern process reads the entire line, and writes into the cells with the matching label. This is a generalisation of Wolfram's automata. In the case where cell values are continuous, or at least many valued, we can approximate the workings of the brain. Essentially we can use 'spare' cells to store a pattern process's internal state, and by convention prevent other pattern processes from reading that internal state. For example, we can have a fatigue cell, whose value goes to 1 upon 'firing', and then reduces to zero in some way, and the pattern process will not fire when this fatigure cell is above a certain threshold. Then the neuron inputs can cause values in a buffer to increase, and decrease in line with how neurons work. The interesting thing then is the possibilities of circuits and cycles to form. For example, if process 1 reads a 1 from cell 1, and writes to cell 2, zeroes cell 1, and pattern process 2 does the opposite, we will get a simple tick-tock between the cells. We can arrange things in any cycle or network we like. Thus a pattern process has a number of input cells, and a number of output cells, and we specify that the output cells of different pattern process are disjoint. This is not actually necessary. We assume that each line starts identically zero (and even this we can change), and each pattern process adds to what is there before. Thus it makes no difference in which order the pattern processes run.