Memory &Cognition
1978.
Vol.
6 (3), 283-287
Memory
performance andsubject-defined
depth of processing
JOHN G. SEAMON and SUSAN VIROSTEK
Wesleyan University, Middletown, Connecticut 06457
This experiment found
that
subjects could order a series of classification questions in terms
of their relative depth of processing. The subject-defined processing depth was used to predict
performance in an incidental learning task which employed the same questions with different
subjects. Asignificant rank correlation was obtained between question depth of processing
and stimulus word free recall. These data provide an independent assessment of processing
depth and memory performance and thereby satisfy a necessary,
but
not sufficient condition
for the depth-of-processing hypothesis. Little success was achieved, however, in attempting
to define the scheme relating processing depth and performance. While memory is related
to encoding coditions, the search for the relational scheme continues.
Several years ago, Craik and Lockhart (1972)
hypothesized that memory performance varies directly
with the level or depth
of
stimulus encoding. Better
performance in studies
of
incidental learning is typically
found when stimuli are classified on the basis
of
meaning
or conceptual relationships than visual or phonological
features. The former tasks are felt to reflect deep
semantic processing, while the latter are thought to
entail surface-structural analysis.
Evidence for the difference between structural and
semantic processing tasks appears to be strong. Hyde
and Jenkins (1969) found associative clustering to be
greater following a semantic than a structural orienting
task. Elias and Perfetti (1973) showed false recognitions
on acoustically or semantically related distractors to
vary directly with encoding conditions. Finally, Seamon
and Murray (1976) found that stimulus meaningfulness
directly affected the recall and recognition
of
words
analyzed on a semantic basis,
but
had no effect on items
classified on a physical basis.
The above studies clearly show that structural and
.semantic orienting tasks entail different kinds
of
processing, but as Nelson
(l977)
effectively argues,
different kinds of processing need not reflect different
depths
of
processing. Differential depth
of
processing is
an assumption based on different types
of
processing
which produce different measures of memory
performance.
The purpose
of
this research is to determine if a large
number
of
classification tasks can be ordered in terms
of
processing depth to be used subsequently as a basis
The present study was supported by a Wesleyan University
Research Grant awarded to the first author. Requests forreprints
should be sent to John G. Seamon, Department of Psychology,
Wesleyan University, Middletown, Connecticut 06457. Appreci-
ation is expressed to Thomas O. Nelson and Fergus I. M. Craik
for comments on an earlier version. Susan Virostek is now at
Columbia University.
for predicting performance on a free recall task. As
stated by Nelson (1977,
p.165),
"the
derivation
of
the
empirical ordering from the theoretical ordering must
occur independently for memory performance and for
depth of processing so that the examination
of
data
can potentially falsify the principle." Anonsignificant
rank correlation between the ordering
of
classification
tasks and subsequent memory performance would
constitute evidence contrary to the depth-of-processing
hypothesis. Alternatively, a significant rank correlation
would satisfy a necessary,
but
not sufficient condition
for the Craik and Lockhart
(1972)
position.
PART 1
Method
Subjects. The subjects were 17 Wesleyan undergraduates
who were enrolled in the senior author's memory seminar
during the spring semester of 1977. At the time of the
experiment, the subjects had read and discussed sensory
memory, attention, and pattern recognition as topics in
the course. The students were not yet familiar with the levels-
of-processingapproach to memory .
Materials. The 13 classification tasks listed by Nelson (1977,
Table 3) and shown in Table I were used in this study. The 13
questions were printed separately on pieces of paper, shuffled,
and placed in an envelope for each subject. The numbers
associated with each question in Table I will be used for later
data presentations in Part 2 and were not available to the
subjects. Table I merely lists the questions with an associated
numerical code.
Procedure. The first part of this experiment was conducted
in a classroom with each of the 17 students seated at a separate
desk. On a blackboard in front of the room, the word FROG
was printed in uppercase letters. Each subject was given an
envelope containing the 13 classification questions and asked
to order the questions in terms of their relative processing depths
from low to high on the surface of the desk. Depth of processing
was defined as the amount of processing or degree of difficulty
associated with each question. No mention of ease of learning
was made to the subjects at any time. The emphasis was placed
on the difficulty of different kinds of processing per se. Subjects
were told that there was no "correct" sequence; they could
283
