Bodner, G. E. and
Lindsay, D. S. (2002). Remembering and
knowing in context. Journal of
Memory and Language, 48, 563-580.
Remember and know judgments are
often used in memory studies as a means for measuring the phenomenological experience
of the participants involved. Past
research has shown that the two judgments can function independently. Some experimental manipulations may affect
only one type of judgment, whereas others may affect them both, either in an
opposite or a parallel manner. For
example, deep levels of processing (LOP) at encoding led to more remember
responses than shallow LOP, but this manipulation had no effect on the amount
of know judgments. Bodner & Lindsay
have taken this LOP effect and added it to a list context manipulation to
determine its effect on the subjective experience of memory.
Tulving (1985) draws the
distinction between the two judgments in that “remembering” is often
accompanied by the specific recollection of episodic details from the encoding
phase, and “knowing” is the result of semantic memory. So to explain the LOP effect, we would have
to assume that deep LOP increases activation in episodic memory leading to more
remember judgments, but semantic activation would remain equivalent with both
deeper and shallower processing, thus no effect of know judgments. Rajaram (1996, 1998) suggests that that
remembering increases when the distinctiveness or salience of items is
increased, and knowing increases when fluency is maximized.
Dual process models can also be
used to explain the data from remember/know procedures. This approach suggests that people base
recognition judgments on either recollection or familiarity, which forms a
distribution of memory strength. A
threshold is set along this line at the point where participants will accept
the word as old (if it falls above) or label it as new (if it falls
below). Furthermore, at the point where
subjects will consider a word old, they set another criterion to divide these
recognition judgments into remember and know.
Gruppuso et al. (1997) proposes
that recollection and familiarity should be defined in functional terms. Recollection would occur when a person’s
memory is sufficient for performing a necessary task. Familiarity, on the other hand, is when memory is sufficient
enough to alert one that a stimulus is familiar, but insufficient to accomplish
the task at hand. Lastly, some kinds of information contributes to familiarity
rather than to recollection, and vice versa.
To test the influence of context
on remember/know judgments, Bodner & Lindsay created a paradigm consisting
of three LOP tasks (shallow, medium, and deep). These were combined to make two groups for each experiment: medium-with-shallow
and medium-with-deep. After
both lists were presented, subjects were given a R/K recognition test with
words from both lists as well as an equal number of new words.
Results: Both
test-list context and LOP encoding factors influenced whether or not subjects
were willing to claim remember or know to recognized items. More specifically, medium LOP items were
more likely to be classified as remembered and less likely to be classified as
known when mixed with shallow rather than deep LOP items.
The procedure used was identical
to that of experiment one except that only one of the two sets of LOP items was
presented along with the new items on the test. This was designed to test whether the effect of list context
arises at test rather than during study.
Results: The
overall level of recognition for medium items was similar to that found in
experiment 1, and was not affected by whether the other LOP task provided at
study was the shallow or deep LOP task.
More importantly, the LOP task had no influence on the rate of R/K
judgments given to the medium items.
The same encoding process was
used as in experiments 1 and 2. During
the recognition test, however, subjects were given a list-source task rather
than a R/K task. The new task included
determining whether each word was presented in list 1, list 2, one of the two
but uncertain about which one, or neither.
Results: The proportion of correct source judgments
for medium items did not differ between the medium-with-shallow and medium-with-deep
groups. Also, subjects showed a
tendency to label new words as having been presented in the shallower of the
two lists.
Experiment 4 had two main
goals. First, it aimed to examine if
subjects’ reported strongest recollections would differ as a function of the
test-list context. To accomplish this,
subjects were asked to rate the extensiveness of their recollection for the
current item then report it to the experimenter in addition to the R/K
judgments. The secondary goal was to
ensure that the medium-with-shallow group were not just placing high
confidence responses that were not associated with any recollected details into
the remember category.
Results: The medium-with-shallow group was
again more likely than the medium-with-deep to judge their recognition
of medium items as an experience of remembering. Furthermore, the medium-with-shallow group was more likely
to report thoughts and associations as their strongest recollection whereas the
medium-with-deep group was more likely to report recollections for list
source.
· The influence of
test-list context cannot be explained by an account that attributes remember
responses to episodic memory and know responses to semantic or perceptual
memory.
· The test-list context
effect of experiments 1 and 4 can be explained by Rajaram’s
distinctiveness/fluency theory for R/K judgments assuming that medium items
were processed more distinctively when tested in the presence of shallow items
than in the presence of deep items.
· The signal detection
model can accommodate the list-context effects if it is assumed that medium-with-deep
group sets a more conservative R/K criterion.
However, this does little to aid in the understanding of the state of
conscious awareness that accompanies recognition.
·Therefore, Gruppusso et
al’s theory that Ss not only respond differently on R/K tasks, but have
different phenomenological experiences all together is the most applicable to
Boder and Lindsay’s results.