A good friend of mine

visited me today and we came to talk about my tiny little spiking neural network simulation. He asked about what could be done next with it, something useful, detecting something, or controlling something, or both.

I said: Good point! That is a good question. It might be the final question at all.

He asked why.

I said I was not sure.

He went to his computer to run some system updates.

"What can it do ?" has always been a good question, I said.

The simulation of the spiking neurons is a simulation of your brain. It will be capable to do exactly the same as your brain is doing right now. It shouldn't be too hard to find out by observing your own thoughts.

And so I ask: What happens in your brain right now ?

A: I'm asking myself the question, what currently happens in my brain.

Q: So what goes on in your brain ?

A: I'm asking myself a question.

Q: When would you have found the answer ?

A: It's when I understood, what in my brain is happening now.

Q: What would be the answer to that question "what in my brain is happening now ?" ?

A: It is: "I'm asking myself a question".


In the very moment, when you could understand, what's happening in your brain and how it works, you would have an answer and there won't be no longer a question anymore.

There would be an answer and no question anymore. Because the answer was the answer to that question. There will never be a question again, because you already know what the answer is.

Yes, there will only be answer and there will never be a question again. It will be the last and only answer because it was the final question. There won't be another question afterwards, because you would have the answer already.

(: IF THERE IS ANSWER THEN THERE IS NO QUESTION :)

And when there is One answer and No question, and also when there can't be another answer, because there will not be another question, this answer will be the answer to everything.

Q: Wait.. why can't there be any other new question about something different?

A: Because I already have the answer. It's the answer to the question: "What happens in my brain ?"

Q: No no, I mean a different question: How high is the moon ?

A: That would also be a question which happens in my brain, like any other question, and I've got the answer.

.: O, I forgot..

A: My answer is the final answer for everything, the universe and the rest. This answer means the end of all questions and it means the end of all science, because there will never be a question again. Finding out what's happening in the brain, will be the final science.

So let's find out..


I.) Be aware of what Douglas Adams said. That in the very moment you find out, the world will be replaced by something much more complicated.

II.) Beware: We might all turn into happy hippies then.

Firing frequency and Plastizity (some thoughts on)

When you define different firing frequencies or frequency dividers for the neurons, then you build up distinct channels for communication between neurons, especially when it comes to 'fire together, wire together'.

Let's say, we only use fixed point dividers (multipliers) for the frequencies. A neuron on channel 1 (divider = 1) can fire together and interconnect with all other neurons. A neuron firing with a prime number as it's frequency divider will only see other neurons from the same channel and multiple of this channel and from channel 1.

Channel 1 could mean: This is live sensor data, process with high priority
Channel 4 could mean: This is long term information
Channel 8 could mean: This is short term memory

Long and short term information can catch all events from live data.
Both interact only a little (each 4th iteration)

This way, visual data could be separated from audio data, left ear from right ear, and so on.
This also solves the big problem of circular references, without using a supervising authority.

Neurons are not intelligent. Not at all.

They can't make decisions, they can't encode or decode complex data. They are bare stupid and if they could speak they would ask: "Who am I ? Where do I come from ? Why ?" (Same as any living thing in universe). But neurons can't ask such question. Because they are not intelligent. Remeber: they are protozoa.

* Intelligence emerges from the interaction of many neurons
* All brain models, that require any intelligence in the neurons, are wrong.

SNN's and learning (some thoughts on)

(1) What is input, what is output ?

When you use a classical Backpropagation Network, you have one layer of neurons for input. and another one for output. Same it is in nature, when it comes to sense organs for the input and nerves to the muscles as the output. Same in robotics.

But what, when the input is of ideas and the expected output is of the same type ?
Everything happens in the mind then, without direct relation to the real physical world.

In such a case, input and output are patterns, to be more specific: A (current) state of the entire brain.

(2) Initial optimization of topology

The neuron model should be transformed into an euklidean space by using an algorithm, that can optimize the graph for shortest links (axons).

Nature did this over million years during evolution. When we start with an empty brain and no evolution at all, we should simply 'sort' it for shortest links.

This step also groups the neurons by meaning in space, e.g. groups synonym words so that they become neighbours.

(3) Working with Brain-Patterns

When thinking about making something useful out of SNN's, then categorization (grouping, clustering) of data comes first into my mind. E.g. a SPAM filter software that can categorize emails and so detect spam.

This applicaton has these components:

* the neuron network, that has run for a while and has built up new connections by playing the ancient game of 'what fires together wires together'.
* Input data that is used to trigger some neurons. This leads to a 'brain state' after some iterations.
* Comparison data that was first run same as 'input data', in order to compare both patterns to see, how much they have in common, how much the rate of convergence is.

(4) Representing brain patterns

Biological brains don't store their pattern states at all. The current state is the current pattern, that's all.

When using computers, you should look for imaging algorithms with data reduction / compression.
You could represent your brain's state by using wavelets and DWT. This also has the advantage, that you can process it with different level of details.

You then build an algorithm that calculates rate of convergence for two patterns.

This is all about mind. When you build roboters or you want to control machines, there should be some neurons that do the job and fire their signal to an output. Senors are connected with neurons that fire, when the sensor signal is above threshold.

I published a windows software with a simplified neuron model, please see
http://www.kroll-software.ch/research/thebrain_e.asp

Spiking networks cannot work without inhibiting neurons

Given a spiking network of 5 neurons: a, b, c, d, e

[a, b, c] build a cluster or pattern, that fires together.
[c, d, e] build another cluster that fires together.

When a and b fire, c will be stimulated.
When d and e fire, c will also be stimulated.
(c is shared by both patterns)

When applying the hebbian learning rule and start to connect the neurons with 'what fires together wires together', this will lead to all neurons being interconnected soon. The distinction of both patterns gets lost.

What we need to not loose both pattern states are two additional inhibiting neurons:
i1 fires together with a and c or b and c and inhibits d and e.
i2 fires together with d and c or e and c and inhibits a and b.

In nature, these inhibiting neurons are probably the Stellate cells.

So the next step in computing artificial spiking networks is: Build a network that can distinguish between different patterns by automatically applying the hebbian learning rule and with using inhibiting neurons.

The Interference of Association

When a neuron is stimulated, it will fire it's spikes somewhen in the next milliseconds and seconds. It will never be exactly at the same time. In combination with millions or billions other neurons, this causes an interference.

The same stimulation will lead to a distinct chain of association every time. It will never be the same, because of interference.