Seeking Examples of Learning With Fixed Weights

Two months ago I posted the following query, which got zero response:
-----------------------------------------------------------------
In most applications of ANNs, learning consists of changing the weights
until desired behavior is reached.  In these applications no learning
takes place after the weights are fixed.  In a feed-forward network no
such learning is possible.  However, a recurrent network has the
possibility of learning even after the weights are fixed.

I would like to know about any examples where this has been
accomplished.
-----------------------------------------------------------------
Since then I have learned of only one relevant item:

Evolving Adaptive Neural Networks with and Without Adaptive Synapses
(2003) by Kenneth O. Stanley, Bobby D. Bryant, and Risto Miikkulainen
 
It can be found here: http://nn.cs.utexas.edu

And, I have also created an ANN that learns with fixed weights.  It
solves a simple puzzle requiring active use of memory.  The memory is in
the state of this highly recurrent ANN.  Complete detailed explanation
and C software is available now for free download from the URL below. 
(this is public domain software)

Mitchell Timin

-- 
"Many are stubborn in pursuit of the path they have chosen, few in
pursuit of the goal." - Friedrich Nietzsche

http://annevolve.sourceforge.net is what I'm into nowadays.
Humans may write to me at this address: zenguy at shaw dot ca