Artificial neural networks (ANNs) are usually homogenous in respect to the used learning algorithms. On the other hand, recent physiological observations suggest that in biological neurons synapses undergo changes according to local learning rules. In this study we present a biophysically motivated learning rule which is influenced by the shape of the correlated signals and results in a learning characteristic which depends on the dendritic site. We investigate this rule in a biophysical model as well as in the equivalent artificial neural network model. As a consequence of our local rule we observe that transitions from differential Hebbian to plain Hebbian learning can coexist at the same neuron. Thus, such a rule could be used in an ANN to create synapses with entirely different learning properties at the same network unit in a controlled way.