This thesis investigated the interaction of perception and memory
when visually processing stimuli of varying familiarity. Specifically, it
assessed whether advantages for processing famous (highly learned) versus
non-famous (recently learnt) images were evident. A change-detection
methodology was constructed, which required two briefly displayed images
to be compared in memory. Each of the images contained two items, one of
which changed into a different item. Experiments were conducted on a
number of object classes: faces, landmarks, and consumer products, with the
factor of main interest being whether the images contained famous or nonfamous items. All categories of object benefited from the presentation of a
famous item, while the exact pattern of effects differed depending on the
object category. The divergence in the pattern of effects for certain object
classes is explained by the degree of structural representation developed and
maintained in conceptual short-term memory before transferral into
traditional short-term memory (Potter, 1976, 1993, 1998). The advantage
found for famous items, termed the visual Jame effect, is explained in terms of
efficient encoding mechanisms as described by robust representations (Tong
& Nakayama, 1999) and the population-encoding hypothesis (Perret, Osram,
& Ashbridge, 1998).