ISR is normally studied using short sequences of list
items. At very short list lengths, recall is typically perfect,
but as the list length increases to around 4–6 words, so perfect
performance breaks down, and the subsequent serial
position curves are characterised by extended primacy
effects (the recall advantage of early list items, e.g.,
Drewnowski & Murdock, 1980).
By contrast, IFR is normally studied using longer sequences
of 10 or more list items. At these list lengths recall
is rarely, if ever, perfect, and the resultant serial position
curves are characterised by modest primacy effects and
larger and more extended recency effects (the recall advantage
of later list items, e.g., Murdock, 1962; Roberts, 1972).