(1989), Atkinson and Shiffrin (1968), Penfield (1969), and Raaijmakers and
Shiffrin (1981). Approaches to memory that may be characterized as predom-
inantly constructionist can be found in Bartlett (1932), Bransford et al. (1977),
Loftus (1980, 1982), Neisser (1967, 1984), and Schacter (1996). Constructionist
approaches to memory are also implicit in neural net (also known as con-
nectionist or parallel distributed processing) models of memory (e.g., Rumel-
hart, Hinton & Williams, 1986; Grossberg & Stone, 1986; see Collins & Hay,
1994, for a summary). Raaijmakers and Shiffrin (1992) provide a technical de-
scription of various contemporary memory models, while Bolles (1988) pro-
vides a nontechnical overview of a constructionist approach to memory written
by someone outside the field.
Historical Support for Record-Keeping Theories of Memory
Although I will champion the constructionist theory in this chapter, historically
it has been record-keeping metaphors that have dominated thinking about
memory (Roediger, 1980). The ancient Greek philosopher Plato, in the The-
aetetus dialogue, likened memory to a wax tablet on which experiences leave
an impression and likened the process of recollection to trying to capture birds
inanaviary.Wemaynotalwaysbeabletocapturetheoneweseek.Saint
Augustine (a.d.354–430), an important Christian theologian, and John Locke
(1631–1704), a British empiricist famous for his claim that there are no innate
ideas, both characterized memory as a storehouse containing records of the
past. More recently, cognitive psychologists have used libraries (e.g., Broad-
bent, 1971), keysort cards (e.g., Brown & McNeill, 1966), tape recorders (e.g.,
Posner & Warren, 1972), stores (e.g., Atkinson & Shiffrin, 1968), and file sys-
tems (e.g., Anderson & Milson, 1989) as metaphors for memory.
The modern era of memory research is usually said to have begun with the
publication of Hermann Ebbinghaus’sUber das Gedachtnis(On Memory) in 1885
(Ebbinghaus, 1885; Hoffman, Bringmann, Bamberg, & Klein, 1986). Ebbinghaus
presented himself lists of arbitrarily ordered words or syllables (but not non-
sense syllables, as is often claimed) and counted the number of recitations it
took him to recall the list perfectly. In some experiments he later attempted to
relearn those lists; the reduction in the number of trials to learn the list the
second time constituted another, more indirect, measure of memory.
From years of doing these experiments, Ebbinghaus established several im-
portant principles of memory. One principle, sometimes known as the Ebbing-
haus forgetting curve, is that most forgetting takes place within the first few
hours and days of learning (see figure 14.1). After a few days, the rate at which
information is lost from memory is very slow and gradual. He also showed that
as the number of syllables on a list increased, the number of trials to learn the
list increased exponentially. A list of 36 items took him 50 times the number of
repetitions to learn as a list of 7 items. Ebbinghaus did not just study arbitrarily
ordered lists; he also tried to memorize more meaningful information, specifi-
cally various sections of the poemDon Juan. He found that he needed only one
tenth as many recitations to memorize the poem as he needed to memorize the
equivalent number of arbitrarily ordered syllables. Meaningful information is
easier to memorize.
314 R. Kim Guenther