Nengo

• If you want to represent a number, say between 0 and 100, you can use a bunch of neurons (say 1000), and see how often they fire in say 100ms.
• This can then represent a number (it's a type of rate coding). So, 100 spikes might represent .1, 1000 1, 10000 10, and 100000 100.
• If you then do this for a vector of (say) 50 numbers, you can represent the vector with (say) 50000 neurons.
• You can then wire up the connections so that they combine vectors by addition, multiplication and other operations.
• This is the basic idea of Nengo. It's a project run by Chris Eliasmith.
• I've got a book, How to Build a Brain by him that talks about this.
• He's done some more stuff with this, like Spaun, and there are a scad of papers.