When is the generalized delta rule a learning rule? a physical analogy

The authors show that under some conditions the weights and threshold obtained under the linear generalized delta rule can be calculated a priori. The analysis is illustrated by a physical analogy. The steady-state weight vector produced by the generalized delta rule can be equated to the center of mass of a collection of particles placed at corners of a hypercube defined by the weights and threshold. The result is a direct mapping from the input and target signals onto the weight-threshold hypercube.<>