The advantage of processing for meaning is not limited to verbal informa-
tion. Subjects were better at recognizing pictures of faces if they previously
thought about whether each face seemed friendly than if they previously
thought about whether each face had a big nose (Smith & Winograd, 1978) and
if they assessed faces for honesty rather than for the sex of the face (Sporer,
1991). In general, thinking about the meaning of a stimulus or elaborating on
the stimulus is likely to permit the stimulus to be assimilated by a greater por-
tion of a cognitive system, and so create more possibilities for reconstructing a
memory of the stimulus later on. Elaboration may also help make information
more distinctive (Craik & Lockhart, 1986; Winnograd, 1981).
Processing the meaning of a stimulus improves memory only when that
processing connects the stimulus to relevant knowledge. For instance, asking a
person whether a shirt is a type of clothing enhances memory for the word
shirt, as opposed to the case where the person is asked whether the wordshirt
contains more vowels than consonants. However, asking a person whether a
shirt is a type of insect does not promote very good memory forshirt(Craik &
Tulving, 1975). In the latter case, answering the question does not encourage
the person to connectshirtwith knowledge of shirts (see Schacter, 1996).
Levels of processing research has been used to challenge the duplex model of
short-term memory (see Klatzky, 1980). There is an important qualification to
the general finding that thinking deeply about information promotes better
memory than does thinking in a shallow manner about the information. The
qualification is that it depends on how memory is tested. If the memory testing
procedure matches the manner in which information is originally learned, then
memory for that information is better than if there is a mismatch.
An example comes from a study by Morris, Bransford, and Franks (1977).
Subjects were required to decide for each of a group of words whether the
word could have a particular semantic property (e.g., ‘‘Does a train have a sil-
ver engine?’’) or whether the word rhymes with another word (e.g., ‘‘Doestrain
rhyme withrain?’’ ). The semantic task was the ‘‘deep’’ task and the rhyming
task was the ‘‘shallow’’ task. Later, some subjects were given a standard recog-
nition task in which they had to pick out the target word from a list of dis-
tractors. Subjects who had made the semantic judgment did better on the
recognition task than did subjects who had made the rhyming judgment. But
other subjects were given a very different test of memory in which they had to
pick out from a list of words which word rhymed with one of the words pre-
viously studied. Now it was the subjects who had originally made the rhyming
judgments who did better. This finding, usually calledtransfer appropriate pro-
cessing, is discussed again later in this chapter.
Individual Differences in Memory
Why does one person have a better memory than another person? Record-
keeping theories, especially those that liken human memory to the memories of
computers or libraries, imply that there is an all-purpose memory system for
storing every kind of experience. According to the record-keeping theory, the
reason some people have better memories than others is that some people have
more efficient mechanisms for storing or retrieving records. Even Plato talked
about some people having a purer kind of wax tablet for storing experiences.
330 R. Kim Guenther