Evolving Connectionist Systems (ECoS) learn local models from data through clustering the data and associating a local output function for each cluster represented in a connectionist structure (Stefan Schliebs & Nikola Kasabov, 2013). These clusters are created based on similarity between data samples either in the input space, or both in the input space and output space. These functions are represented in their connection weights.
ECoS traditionally uses the simple sigmoid model of a neuron. Evolving Spiking Neural Networks (§evolving_snn) architectures use a spiking neuron model.