Long Term Memory
Long Term Memory
Long Term Memory
& Retrieval
Dr. Meenakshi Shukla
Encoding: Getting Information into Long-Term Memory
• There are a number of ways of getting information into long-term memory, some of
which are more effective than others. One example is provided by different ways of
rehearsing information. Consider, for example, holding a phone number in your memory
by repeating it over and over. If you do this without any consideration of meaning or
making connections with other information, you are engaging in maintenance rehearsal.
• Typically, this type of rehearsal results in little or no encoding and therefore poor
memory, so you don’t remember the number when you want to call it again later.
• But what if, instead of mindlessly repeating the phone number, you find a way to relate
it to something meaningful. As it turns out, the first three numbers are the same as your
phone number, and the last four just happen to be the year you were born! Coincidence
as this may be, it provides an example of being able to remember the number by
considering meaning or making connections to other information. When you do that,
you are engaging in elaborative rehearsal, which results in better memory than
maintenance rehearsal.
• This contrast between maintenance rehearsal and elaborative rehearsal is one example
of how encoding can influence the ability to retrieve memories.
Levels of Processing
• The results indicated that deeper processing is associated with better memory.
• Gordon Bower and David Winzenz (1970) decided to test whether using visual
imagery—generating images in your head to connect words visually—can enhance
memory. They used a procedure called paired-associate learning, in which a list of word
pairs is presented. Later, the first word of each pair is presented, and the participant’s
task is to remember the word it was paired with.
• Why are participants more likely to remember words they connect to themselves? One
possible explanation is that the words become linked to something the participants know
well—themselves. Generally, statements that result in richer, more detailed
representations in a person’s mind result in better memory.
• Generating material yourself, rather than passively receiving it, enhances learning
and retention. Norman Slameka and Peter Graf (1978) demonstrated this effect, called
the generation effect, by having participants study a list of word pairs in two different
ways:
1. Read group: Read these pairs of related words. king–crown; horse–saddle; lamp–shade;
etc.
2. Generate group: Fill in the blank with a word that is related to the first word. king–cr
_______; horse–sa _______; lamp–sh _______; etc.
• After either reading the pairs of words (read group) or generating the list of word pairs
based on the word and first two letters of the second word (generate group),
participants were presented with the first word in each pair and were told to indicate
the word that went with it.
• Participants who had generated the second word in each pair were able to reproduce
28 percent more word pairs than participants who had just read the word pairs.
Organizing Information
• Folders on your computer’s desktop, computerized library catalogs, and tabs that
separate different subjects in your notebook are all designed to organize information so
it can be accessed more efficiently. The memory system also uses organization to access
information. This has been shown in a number of ways.
• Look at the list you created and notice whether similar items (for example, apple, plum,
cherry; shoe, coat, pants) are grouped together. If they are, your result is similar to the
result of research that shows that participants spontaneously organize items as they recall
them (Jenkins & Russell, 1952).
• One reason for this result is that remembering words in a particular category may serve as
a retrieval cue—a word or other stimulus that helps a person remember information
stored in memory.
• In this case, a word in a particular category, such as fruits, serves as a retrieval cue for
other words in that category. So, remembering the word apple is a retrieval cue for other
fruits, such as grape or plum, and therefore creates a recall list that is more organized
than the original list that you read.
• If words presented randomly become organized in the mind, what happens when
words are presented in an organized way during encoding? Gordon Bower and
coworkers (1969) answered this question by presenting material to be learned in an
“organizational tree,” which organized a number of words according to categories. For
example, one tree organized the names of different minerals by grouping together
precious stones, rare metals, and so on.
• Another group of participants also saw four trees, but the words were randomized, so that
each tree contained a random assortment of minerals, animals, clothing, and
transportation. These participants were able to remember only 21 words from all four
trees. Thus, organizing material to be remembered results in substantially better recall.
• What was that all about? Although each sentence makes sense, it was probably difficult
to picture what was happening, based on the passage. Bransford and Johnson’s
participants not only found it difficult to picture what was going on, but they also found
it extremely difficult to remember this passage.
• To make sense of this passage, look at the figure
below and then reread the passage. When you
do this, the passage makes more sense.
Bransford and Johnson’s (1972) participants
who saw this picture before they read the
passage remembered twice as much from the
passage as participants who did not see the
picture or participants who saw the picture
after they read the passage.
• For Group 2 the study part of the study-test sequence was changed. Once a pair was
recalled correctly in a test, it was no longer studied in next study sessions. However, all
of the pairs were tested during each test session until performance reached 100
percent. This group therefore studied less of the pairs as the experiment progressed.
• For Group 3 the test part of the study-test sequence was changed. Once a pair was
recalled correctly, it was no longer tested during the next test sessions. This group was
therefore tested on less of the pairs as the experiment progressed.
• When tested a week later, Groups 1 and 2 recalled 81 percent of the pairs, but Group 3
only recalled 36 percent of the pairs. This result shows that being tested is important
for learning because when testing was stopped for Group 3 once items were recalled
correctly, performance decreased. In contrast, the results for Group 2 show that
cessation of studying did not affect performance. The enhanced performance due to
retrieval practice is called the testing effect.
Retrieval: Getting Information Out of Memory
Retrieval Cues
• Retrieval cues are words or other stimuli that help us remember information stored in
our memory. Location can serve as a retrieval cue. When we forget something, returning
to the place where we had originally thought about that something can help in retrieving
the original thought.
• Many other things besides location can provide retrieval cues. Hearing a particular song
can bring back memories for events you might not have thought about for years. Or
some smells can trigger memories (Proust phenomenon).
• The operation of retrieval cues has also been demonstrated in the laboratory using a
technique called cued recall. We can distinguish two types of recall procedures. In free
recall, a participant is simply asked to recall stimuli. These stimuli could be words
previously presented by the experimenter or events experienced earlier in the
participant’s life.
• In cued recall, the participant is presented with retrieval cues to aid in recall of the
previously experienced stimuli. These cues are typically words or phrases.
• One of the most impressive demonstrations of the power of retrieval cues was provided
by Timo Mantyla (1986), who presented his participants with a list of 504 nouns, such as
banana, freedom, and tree. During this study phase, participants were told to write
three words they associated with each noun. For example, three words for banana
might be yellow, bunches, and edible.
• In the test phase of the experiment, these participants were presented with the three
words they had generated (self-generated retrieval cues) for half the nouns, or with
three words that someone else had generated (other-person-generated retrieval cues)
for the other half of the nouns. Their task was to remember the noun they had seen
during the study phase.
• You might think it would be possible to guess banana from three properties like yellow,
bunches, and edible, even if you had never been presented with the word banana. But
when Mantyla ran another control group in which he presented the cue words
generated by someone else to participants who had never seen the 504 nouns, these
participants were able to determine only 17 percent of the nouns.
• The results of this experiment demonstrate that retrieval cues (the three words) provide
extremely effective information for retrieving memories, but that retrieval cues are
significantly more effective when they are created by the person whose memory is
being tested.
Matching Conditions of Encoding and Retrieval
• Returning to the location of encoding can help recall things. This example illustrates the
following basic principle: Retrieval can be increased by matching the conditions at
retrieval to the conditions that existed at encoding.
• There are three specific situations in which retrieval is increased by matching conditions
at retrieval to conditions at encoding. These different ways to achieve matching are:
(1) encoding specificity—matching the context in which encoding and retrieval occur;
(2) state-dependent learning—matching the internal mood present during encoding and
retrieval; and
(3) transfer-appropriate processing—matching the task involved in encoding and retrieval.
Encoding Specificity
• The principle of encoding specificity states that we encode information along with its
context.
• Another example of how matching the conditions at encoding and retrieval can influence
memory is state-dependent learning—learning that is associated with a particular
internal state, such as mood or state of awareness.
• For example, Eric Eich and Janet Metcalfe (1989) demonstrated that memory is better
when a person’s mood during retrieval matches his or her mood during encoding. They
did this by asking participants to think positive thoughts while listening to “merry” or
happy music, or depressing thoughts while listening to “melancholic” or sad music.
Participants rated their mood while listening to the music, and the encoding part of the
experiment began when their rating reached “very pleasant” or “very unpleasant.” Once
this occurred, usually within 15 to 20 minutes, participants studied lists of words while in
their positive or negative mood.
• After the study session ended, the participants were told to return in 2 days (although
those in the sad group stayed in the lab a little longer, snacking on cookies and chatting
with the experimenter while happy music played in the background, so they wouldn’t
leave the laboratory in a bad mood).
• Two days later, the participants returned, and the same procedure was used to put
them in a positive or negative mood. When they reached the mood, they were given a
memory test for the words they had studied 2 days earlier. The results showed that
they did better when their mood at retrieval matched their mood during encoding.
Mood Congruence
• Mood congruence means that you recall material more accurately if it is congruent with
your current mood (Fiedler et al., 2003; Joorman & Siemer, 2004; Schwarz, 2001). For
example, a person who is in a pleasant mood should remember pleasant material
better than unpleasant material, whereas a person in an unpleasant mood should
remember unpleasant material better.
• Laura Murray and her colleagues (1999) tested one group of students who did not have
tendencies toward depression, and one group with depressive tendencies. The
participants were instructed to look at a series of 20 positive- and 20 negative-trait
words. Later, the participants recalled as many words as possible from the original list.
Matching the Cognitive Task: Transfer-Appropriate Processing
• Donald Morris and coworkers (1977) did an experiment that showed that retrieval is
better if the same cognitive tasks are involved during both encoding and retrieval. The
procedure for their experiment was as follows:
Part I. Encoding
• Participants heard a sentence with one word replaced by “blank,” and 2 seconds later
they heard a target word. There were two encoding conditions. In the meaning condition,
the task was to answer “yes” or “no” based on the meaning of the word when it filled in
the blank. In the rhyming condition, participants answered “yes” or “no” based on the
sound of the word. Here are some examples:
• The question Morris was interested in was how the participants’ ability to retrieve the
target words would be affected by the way they had processed the words during the
encoding part of the experiment. There were a number of different conditions in this part
of the experiment, but we are going to focus on what happened when participants were
required to process words in terms of their sounds.
• Participants in both the meaning group and the rhyming group were presented with a
series of test words, one by one. Some of the test words rhymed with target words
presented during encoding; some did not. Their task was to answer “yes” if the test word
rhymed with one of the target words and “no” if it didn’t. In the examples below, notice
that the test words were always different from the target word.
• The key result of this experiment was that the participants’ retrieval performance
depended on whether the retrieval task matched the encoding task.
• Participants who had focused on rhyming during encoding remembered more words in
the rhyming test than participants who had focused on meaning. Thus, participants
who had focused on the word’s sound during the first part of the experiment did better
when the test involved focusing on sound. This result—better performance when the
type of processing matches in encoding and retrieval—is called transfer-appropriate
processing.
Design and results for the Morris et al. (1977) experiment. Participants who did a rhyming-based encoding task did
better on the rhyming test than participants who did a meaning-based encoding task. This result would not be
predicted by levels of processing theory but is predicted by the principle that better retrieval occurs if the encoding
and retrieval tasks are matched.
• Transfer-appropriate processing is like encoding specificity and state-dependent
learning because it demonstrates that matching conditions during encoding and
retrieval improves performance. But, in addition, the result of this experiment has
important implications for the levels of processing theory.
• The main idea behind levels of processing theory is that deeper processing leads
to better encoding and, therefore, better retrieval. Levels of processing theory
would predict that participants who were in the meaning group during encoding
would experience “deeper” processing, so they should perform better.
• Instead, the rhyming group performed better. Thus, in addition to showing that
matching the tasks at encoding and retrieval is important, Morris’s experiment
shows that deeper processing at encoding does not always result in better
retrieval, as proposed by levels of processing theory.