Hi @iampaulq,
I may have answered your question in another post, but I’ll reply here as well just in case.
In Nengo, you can do neuron-neuron connections by using the .neurons attribute of any Nengo ensemble. As an example, if you had two populations (ens_a and ens_b) and you wanted to define a full connection weight matrix between the two, you’d do:
nengo.Connection(ens_a.neurons, ens_b.neurons, transform=weight_matrix)
Note that you are not restricted to creating fully connected weight matrices. Nengo allows you to mix and match the pre and post of the connections, as long as the given transformation matrix is sized appropriately.
For example:
nengo.Connection(ens_a, ens_b.neurons, transform=weight_matrix)
connects the decoded output of ens_a directly to the neurons of ens_b, bypassing the encoders of ens_b, but using the decoders of ens_a, while:
nengo.Connection(ens_a.neurons, ens_b, transform=weight_matrix)
connects the direct neuron output of ens_a to the encoded input of ens_b, bypassing the decoders of ens_a, but using the encoders of ens_b, and of course (for completeness):
nengo.Connection(ens_a, ens_b, transform=weight_matrix)
connects the decoded output of ens_a to the encoded input of ens_b, using both the decoders of ens_a and the encoders of ens_b.