MODELING ITEM LENGTH EFFECTS IN MEMORY SPAN - NO REHEARSAL NEEDED

It is widely assumed that word length effects on memory span reflect the operation of subvocal rehearsal, and that span will be the amount of material that can be rehearsed in a fixed time interval. However recent data question both these conclusions. Furthermore, the relevant models are often under-specified and it is not always clear what item length effects they predict. We report an implementation of a simple trace decay memory, without a rehearsal process. The model exhibits item length effects due to opposing effects of decay and redintegration, and lexicality effects that occur only during redintegration. Further simulations demonstrate that item length effects can also be obtained when forgetting is due to interference or imperfect trace registration instead of trace decay. It is concluded that many of the data that have hitherto been taken as evidence for subvocal rehearsal can be explained in terms of simple models without rehearsal. (C) 1995 Academic Press, Inc.