Memory in long-term memory (LTM) is organized in a systematic way that enables efficient
encoding, storage, and retrieval. This involves various frameworks, including hierarchies of
concepts, semantic networks, and schemas. Below is a more detailed explanation of how LTM
is organized:
1. Hierarchies of Concepts
Definition:
• A hierarchical structure organizes concepts at varying levels of generality, from
broad to specific categories.
Structure of Hierarchies
• Rosch et al. (1976) identified three levels:
1. Superordinate Level:
• The most general category, such as “furniture” or “animal.”
• Superordinate categories lack specific attributes and are abstract.
2. Basic Level:
• Intermediate categories like “chair” or “dog.”
• These categories are most commonly used in everyday life because they offer
the best balance between informativeness (specific enough) and distinctiveness (different from
other categories).
• People interact with items at this level in a similar manner (e.g., sitting on all
types of chairs in the same way).
3. Subordinate Level:
• Highly specific categories, such as “easy chair” or “Labrador retriever.”
• Subordinate categories provide additional details but may not always be
necessary for basic communication.
Key Research:
1. Naming Study by Rosch et al. (1976):
• Participants named objects in pictures. Basic-level names were overwhelmingly
used (e.g., “chair” 1,595 times
(e.g., “chair” 1,595 times), while superordinate names like “furniture” were used only once, and
subordinate names such as “easy chair” were used just 14 times. This highlights the preference
for basic-level categories in everyday communication.
2. Attributes Study:
• Participants listed attributes for categories at each level of the hierarchy.
• Superordinate level: Few attributes were listed because these categories are too
abstract.
• Basic level: A rich list of distinct and informative attributes was provided.
• Subordinate level: Attributes were detailed but often overlapped across
subordinate categories, making them less distinct.
3. Motor Interaction Consistency:
• At the basic level, interaction with category members is consistent. For example,
sitting on a chair is similar for all types of chairs, whereas interaction with a table (another
superordinate member of “furniture”) differs significantly.
Impact of Expertise:
• Familiarity influences hierarchical preferences. Experts are more likely to use
subordinate categories because they can distinguish finer details within a domain.
• For instance, birdwatchers frequently use subordinate names like “sparrow” or
“cardinal,” while non-experts stick to basic names like “bird” (Tanaka & Taylor, 1991).
Limitations of Hierarchical Models:
1. Arbitrary Levels:
• Real-world hierarchies often extend beyond three levels (e.g., living being >
mammal > dog > Labrador retriever > specific Labrador).
2. Task Dependence:
• The level at which a concept is categorized depends on the task or context. For
instance, people categorize faster at the superordinate level (e.g., “animal”) when images are
presented briefly or under poor visibility conditions (Praß et al., 2013).
2. Semantic Networks
Definition:
• Semantic networks represent concepts as nodes connected by links indicating
semantic relatedness or association.
Key Theory:
• Collins and Loftus (1975) proposed the spreading activation model, which
emphasizes the flexible and associative nature of semantic memory:
1. Nodes: Represent concepts (e.g., “fire”).
2. Links: Represent relationships between concepts (e.g., “fire” and “smoke”).
3. Spreading Activation: When a concept is activated (e.g., “fire”), activation
spreads to related nodes (e.g., “smoke,” “heat”). Closer nodes receive stronger activation, while
more distant nodes receive weaker activation.
Research Findings:
1. Word Association Tasks:
• Participants produce words related to a target word. For example, “fire” might
elicit responses like “smoke,” “flame,” or “heat.” These associations reveal the structure of
semantic networks.
2. Feature Overlap:
• Concepts that share features (e.g., “knife” and “fork” both involve eating) are
more closely linked in the network.
3. Semantic Ratings:
• Participants rate how similar two concepts are. For instance, “fire” and “burn”
might receive a high rating, while “fire” and “tree” might score lower.
Advantages:
• Explains semantic priming, where exposure to one word (e.g., “doctor”) facilitates
faster recognition of related words (e.g., “nurse”).
• Activation in semantic networks makes concept retrieval richer and more flexible.
Applications:
• Semantic networks explain how we infer meaning in ambiguous sentences (e.g.,
“The house was on fire” automatically activates concepts like “smoke” and “heat”).
3. Schemas
Definition:
• Schemas are cognitive frameworks that organize knowledge about events,
objects, and relationships, providing a structure for interpreting new information.
How Schemas Work:
• Schemas are built through experience and represent typical patterns or
sequences of events. For instance, a “restaurant schema” includes expectations like being
seated, receiving a menu, ordering, and paying.
Key Features of Schemas:
1. Encoding and Retrieval:
• Schemas help encode new information by focusing on relevant details and filling
in missing information during retrieval.
2. Filling in Gaps:
• Schemas allow us to infer information that isn’t explicitly stated. For example, if
someone mentions “visiting the zoo,” you might assume they saw animals even if it isn’t
mentioned.
Key Research:
1. Bartlett (1932):
• In his study “The War of the Ghosts,” participants distorted unfamiliar stories to fit
their existing schemas. This process, called rationalization, highlighted how schemas align
information with prior knowledge.
2. Brewer and Treyens (1981):
• Participants recalled objects from an office setting. They remembered more
schema-consistent items (e.g., books) and even falsely recalled schema-consistent items that
weren’t present.
Benefits of Schemas:
1. Prediction:
• Schemas help us anticipate events, making the world more predictable.
2. Reducing Cognitive Load:
• By summarizing patterns, schemas simplify complex environments.
3. Filling in Missing Information:
• They enhance understanding of incomplete information, such as interpreting
fragmented stories or visuals.
Limitations:
• Schemas can lead to false memories or errors when they conflict with actual
details, especially over time.
Formation and Use of Schemas
Schemas develop through experience and represent typical patterns or sequences of events.
They act as mental shortcuts, enabling efficient interpretation of new information. For example,
a “restaurant schema” includes expected steps like being seated, ordering food, eating, and
paying.
How Schemas Operate in Memory
1. Encoding and Storage:
• Schemas help encode information by highlighting what is most important or
relevant.
• They store generalized knowledge, not exact details, which allows for flexibility in
applying them to various situations.
2. Retrieval and Influence on Memory:
• When retrieving information, schemas “fill in the gaps” using default knowledge.
• Bartlett (1932): In his famous study with “The War of the Ghosts,” participants
remembered stories by aligning them with their existing cultural schemas, leading to
rationalization (distortions that made the stories fit their cultural expectations).
3. Errors and False Memories:
• Schemas can lead to schema-consistent memory errors:
• Brewer and Treyens (1981): Participants falsely recalled objects consistent with
an office schema (e.g., books) that were not present in the actual room.
• Over time, schema-driven false memories increase, as shown by Lampinen et al.
(2001).
Benefits of Schemas:
1. Forming Expectations:
• Schemas make the world predictable by helping us anticipate events. For
example, in a meeting, you expect an agenda to be followed.
2. Preventing Cognitive Overload:
• By filling in gaps, schemas simplify complex situations, reducing the need for
exhaustive processing of details.
3. Enhancing Understanding:
• They allow inferences that make ambiguous or incomplete information
meaningful. For example, reading “Tom was driving when a tire blew, and he arrived home late”
leads us to infer the challenges Tom faced.
4. Facilitating Perception:
• Schemas guide the interpretation of visual scenes. Palmer (1975): Participants
identified context-appropriate objects (e.g., a loaf in a kitchen scene) more accurately than
unrelated ones (e.g., a mailbox).
4. Propositions and Event Schemas
Propositions:
• Concepts in LTM are organized into propositions, which represent relationships
between ideas (e.g., “A dog chased a cat”).
• Propositions are the building blocks of situations or events, which combine
multiple propositions to represent complex occurrences.
Event Schemas:
• Repeated sequences of propositions become stored as event schemas. For
example, a “birthday party schema” includes typical events like singing, cutting a cake, and
giving gifts.
• Zwaan & Radvansky (1998): Events are linked in memory based on dimensions
like:
• Time: Events happening simultaneously are grouped together.
• Causation: Actions with cause-and-effect relationships are integrated.
5. Stereotypes
Stereotypes are specific schemas that involve oversimplified generalizations about groups of
people. While they make social interactions more efficient, they can lead to biases and errors in
judgment.
Formation of Stereotypes:
• Stereotypes are created through cultural exposure, personal experience, and
social learning.
• They are automatically activated when encountering members of a stereotyped
group.
Stereotypes in Memory:
1. Schema-Driven Errors:
• Stereotypes distort memory to align with expectations:
• Sulin and Dooling (1974): Participants read a story about “Adolf Hitler” (or a
fictitious dictator). Those who read about Hitler falsely remembered stereotypical details, like
persecuting Jews, even when such details were absent.
2. Influence on Problem Solving:
• Stereotypes interfere with reasoning when they conflict with reality:
• Reynolds et al. (2006): Participants struggled to solve a riddle involving a female
surgeon due to the stereotype that surgeons are male.
3. Contextual Flexibility:
• Stereotypes are not fixed. Their activation depends on the context:
• Casper et al. (2010): Participants only activated a stereotype (e.g., “clumsy
women”) when paired with a relevant visual cue (e.g., a poorly parked car).
6. Integration of Approaches in LTM
To organize memory effectively, LTM combines hierarchies, semantic networks, schemas, and
event structures into a cohesive system:
• Hub-and-Spoke Model:
• This model explains how multiple types of information (verbal, sensory, motor)
are integrated. A central hub (e.g., the anterior temporal lobe) links specialized regions (spokes)
that process different types of knowledge (e.g., sensory input, motor functions).
• Interdependence of Systems:
• For example, semantic networks enrich schemas by linking related concepts. In
turn, schemas guide the retrieval of hierarchical information based on task requirements.
Conclusion
Memory in LTM is organized through hierarchies, networks, schemas, propositions, and
stereotypes to allow efficient retrieval and use of information. While these systems enhance
understanding, they can introduce biases and errors, especially when schemas or stereotypes
conflict with actual details. This dynamic interplay between efficiency and accuracy reflects the
complexity of memory organization in LTM.