An unsupervised-learning algorithm which endows a logical neural network model with the ability to separate patterns belonging to different pattern classes while at the same time creating a topology-preserving mapping of the input space is examined in this paper. Emphasis is placed in the storage efficiency of the algorithm and its ability to maximise the number of pattern classes that can be stored in the network. To achieve that objective, the introduction of a decay mechanism in the learning rule is proposed. The development of this mechanism has been based on a detailed analysis of the self-organisation process. Experimental results indicate that this forgetting mechanism maximises the storage capacity of the network, considerably improving the system performance.