As I've described here, the Neural Engineering Framework (NEF) has some functional similarities to Artificial Neural Networks (ANNs). Naturally, there is also an overlap in applications between NEF-based neural networks and ANNs. For example, Spaun, which was built using the NEF, was used to recognize digits from the MNIST data set. Anecdotally, I've also seen an NEF-based network used for gesture recognition.
At both of these tasks, the ANNs vastly surpassed the NEF-based networks. So what's the point on "taking-on" the ANNs on their own turf, aside from biological plausibility? What does a NEF-based network offer application-wise that can't be filled more adequately by an ANN?