Neural network without neurotransmitters

  • Thread starter Secan
  • Start date
  • #26
...
And I think your assumption: "computers could be the model for neurons" may not be very useful. It really is the other way around. Concepts like parallel programming - or multitasking- are easily seen in modern mammalian brains. For which the primary design evolved 500 million years ago, long before John Von Neumann came up the the cpu concept.
...
I agree.
I thing of circuits as components connected by the wires being simplified neurons/synapses. Since synapses have hundreds of neurotransmitters, that can be used as a form of multichannel information flow. For passive components, resistors, capacitors, there can be only one channel of information in the circuit. But interconnected computers approach multichannel processing. Animals have had a 500 million year head start on our electrical technology.
 
  • #27
Laroxe
Science Advisor
314
300
I think the idea that the brain can be understood as if it was simply a complicated electric circuit with neurones acting as wires is hugely misleading. A neurone doesn't simply function as a switch, its been recognised for some time that neurons can compute a huge range of functions. Each branch of a dendrite acts like a little non-linear output device, summing up and outputting a local spike if that branch gets enough inputs at roughly the same time. The dendrites receive inputs from a very large number of other neurons from far and wide and it seems the distance and direction of the inputs is significant. We also know that neurons continuously send spikes to each other, that appear to be independent of particular inputs a lot of repeated movements are controlled by groups of neurons independently.

Any network is a mixture of excitatory and inhibitory neurons, with the latter making their target neurons less likely to fire a spike. These will tend to use different chemical transmitters, but the brain doesn't just have two, it uses a variety of different transmitters with different qualities, in addition in a nerve synapse there are a range of different binding sites again with different functions. A single neuron can have synaptic connections with up to 10,000 others and not just at the dendrites, connections can be made at the cell body or even along the axon.

We then have to consider that the brain sits in a neurochemical bath and if you alter the chemicals in the bath of fluid, you can change how communication works in specific areas with the right receptor sites or in the whole network at once.

There is an attempt to model some of these processes with Liquid State Machines which try to introduce continuous dynamics in randomly connected networks.

This is important is that the resulting network is almost guaranteed to have chaotic dynamics and chaotic dynamics give us a very rich playground in which to do stuff. Of course evolution could have equipped us with directly connected neurons, but it would be a massive downgrade.

Sorry I couldn't find anything which summarised all of this though there are plenty of individual papers around.
 
  • Like
Likes BillTre

Related Threads on Neural network without neurotransmitters

  • Last Post
Replies
1
Views
3K
  • Last Post
Replies
1
Views
5K
Replies
5
Views
959
Replies
11
Views
3K
Replies
3
Views
5K
Replies
1
Views
3K
Replies
1
Views
5K
  • Last Post
Replies
4
Views
1K
Replies
8
Views
1K
Replies
8
Views
4K
Top