Some Results on Activation and Scaling of SparseDistributed

It has been suggested that in certain situations it would make sense to use diierent activation probabilities for writing and reading in SDM (Sparse Distributed Memory). However, here we model such a situation and nd that, at least approximately, it is optimal to use the same probabilities for writing and reading. We also investigate the scaling up of SDM, in connection with some observations made by Sjj odin, see 4]. It is shown that the original SDM (here in Jaeckel's version) does not scale up if the reading address is disturbed, but that this can be remedied by using a kind of sparse SDM.