Retrieval from Semantic Memory

220
in review
Chapter 6
Memory
FACTORS AFFECTING RETRIEVAL FROM LONG-TERM MEMORY
Process
Effect on Memory
Encoding specificity
Retrieval cues are effective only to the extent that they tap
into information that was originally encoded.
Context dependence
Retrieval is most successful when it occurs in the same
environment in which the information was originally learned.
State dependence
Retrieval is most successful when people are in the same
psychological state as when they originally learned the
information.
?
1. Stimuli called
help you recall information stored in long-term memory.
2. If it is easier to remember something in the place where you learned it, you have a
memory.
3. The tendency to remember the last few items in a list is called the
effect.
Context-dependent memories are those that are helped or hindered by similarities or differences in environmental context. This context-dependency effect is not
always strong (Smith, Vela, & Williamson, 1988), but some students do find it helpful
to study for a test in the classroom where the test will be given.
Sometimes, we also encode information about how we were feeling during a
learning experience, and this information, too, can act as a retrieval cue. When our
internal state influences retrieval, we have a state-dependent memory. For example, if people learn new material while under the influence of marijuana, they tend
to recall it better if they are also tested under the influence of marijuana (Eich et al.,
1975). Similar effects have been found with alcohol (Overton, 1984) and other psychoactive drugs (Eich, 1989), although memory is best when people aren’t using any
drugs during encoding or retrieval. Mood states, too, can affect memory (Eich &
Macaulay, 2000). College students are more likely to remember pleasant events when
they are feeling good at the time of recall (Bower, 1981; Ehrlichman & Halpern,
1988). Negative events are more likely to be recalled when people are feeling sad or
angry (Lewinsohn & Rosenbaum, 1987). These mood congruency effects are strongest
when people try to recall personally meaningful episodes (Eich & Metcalfe, 1989).
The more meaningful the experience, the more likely it is that the memory has been
colored by their mood. (See “In Review: Factors Affecting Retrieval from Long-Term
Memory.”)
Retrieval from Semantic Memory
context-dependent memories Memories that are helped or hindered by
similarities or differences between the
contexts in which they are learned and
recalled.
state-dependent memory Memory
that is helped or hindered by similarities or differences in a person’s internal
state during learning versus recall.
The retrieval situations we have discussed so far are relevant to episodic memory—our
memory for events. But how do we retrieve information from semantic memory, in
which we store our general knowledge about the world? Researchers studying this question typically ask participants general-knowledge questions, such as (1) Are fish minerals? (2) Is a beagle a dog? (3) Do birds fly? and (4) Does a car have legs? As you might
imagine, most people answer such questions correctly. But by measuring how long it
takes to answer them, psychologists gain important clues about how semantic memory
is organized and how we retrieve information from it.
One view of semantic memory suggests that virtually everything we know about, including concepts such as “bird” or “animal,” is represented in
a dense network of associations (Churchland, 1989). Figure 6.7 presents just a tiny part
Semantic Networks
221
Retrieving Memories
Living thing
is - a
is - a
Plant
Animal
is - a
is - a(n)
Bird
has
Mammal
is-not-a
is - a
is - a
is - a
Wings
has
has
Cat
Lion
is - a
is - a
Bat
can
Fly
can
Robin
Penguin
has - a
Red breast
can not
FIGURE 6 .7
Semantic Memory Networks
This drawing represents just a small part
of a network of semantic associations.
Semantic network theories of memory
suggest that networks like these allow us
to retrieve specific pieces of previously
learned information, to draw conclusions
about how concepts are related, and to
make new inferences about the world.
of what our semantic memory network might look like. In general, semantic network
theories suggest that information is retrieved from memory through the principle of
spreading activation (Medin, Ross, & Markman, 2001). In other words, when you
think about some concept, it becomes activated in the network, and this activation
begins to “spread” down all the paths that are related to it. So if you are asked if a robin
is a bird, the concepts of both “robin” and “bird” will become activated, and the spreading activation from each will intersect somewhere in these paths. When they do, you
know what answer to give.
Some associations in the network are stronger than others, as illustrated by the
thicker lines between some concepts in Figure 6.7. For instance, you probably have a
stronger association between “bat” and “wings” than between “bat” and “mammal.”
Spreading activation travels faster along stronger paths than along weaker ones. As a
result, you’d probably respond more quickly to “Can a bat fly?” than to “Is a bat a
mammal?”
Because of the tight organization of semantic networks and the speed at which activation spreads through them, we can gain access to an enormous body of knowledge
about the world quickly and effortlessly. We can retrieve not only facts we have learned
from others but also knowledge that allows us to draw our own conclusions and inferences (Matlin, 1998). For example, imagine answering these two questions: (1) Is a
robin a bird? and (2) Is a robin a living thing? You can probably answer the first question “directly,” because you probably learned this fact at some point in your life. However, you may never have consciously thought about the second question, so answering it requires some inference. Figure 6.7 illustrates the path to that inference. Because
you know that a robin is a bird, a bird is an animal, and animals are living things, you
infer that a robin must be a living thing. As you might expect, however, it takes slightly
longer to answer the second question than the first.
Figure 6.7 shows that concepts such as
“bird” or “living thing” are represented in semantic memory as unique sets of features
or attributes. As a result, there may be times when you can retrieve some features of a
concept from your semantic network, but not enough of them to identify the concept.
For example, you might know that there is an animal that has wings, can fly, and is not
a bird, and yet be unable to retrieve its name (Connor, Balota, & Neely, 1992). When
this happens, you are retrieving incomplete knowledge. (The animal in question is a bat.)
Retrieving Incomplete Knowledge
spreading activation In semantic
network theories of memory, a principle that explains how information is
retrieved.