26 terms

Semantic Networks

Category, exemplar, concept
Category: group of objects that have something in common
Exemplar: instance of a category
Concept: mental representations of categories
Feature Based Theories
Categorize based off rules
Defining features- must have feature to be in category (necessary and sufficient)
Probabilistic View
Having certain features makes more likely that the instance belongs to a category
Graded membership: some are better members of a category than others
Fuzzy boundaries: there can be things that fall at the border of a category, where there is no clear cut off
Probabilistic Prototype
In order to decide if something is a member of the category, it is compared to the prototype (dealized member; "averaged model")
Method: Subjects are shown a set of faces. They are then given a list of some new some old faces and are asked to identify which face they had seen before. Among these new faces is the prototype.
Results: Subjects claim to have seen the prototyped face even though it is new.
Conclusion: False recognition of the prototype face (actual average of faces) occurred because it looked like the prototypical face in their mind (mental average of faces)
Probabilistic Exemplar
Judge category membership by comparing overlap with all members
Method: Subjects were asked to judge typicality of either "spoon" or "large spoon."
Results: Metal spoons were rated as more typical of "spoon" and wooden spoons were rated as more typical of "large spoon."
Conclusion: Because subjects had specific features in mind, it makes sense that they had an exemplar of "spoon" and "big spoon" in mind. An average would not have yielded these results.
Typicality Effect
Degree to which members are seen as central members of each category
Method: Subjects are asked to rate fruits by typicality on a scale of 1-7.
Results: Apples are rated as very typical. Pineapples are less so.
Conclusion: There are some fruits that are considered more fruity than others.
Typicality Effect
Typicality rating predicts instance generation, developmental patters, sentence verification speed.
Cannot explain times in which context and function influence categorization (shoe and hammer vs shoes and sock)
Knowledge of Function
Emphasizes operational function as opposed to multiple people only.
Method: Subjects were given information about two stimuli. Then they were shown a third image that shared physical similarly to one but a biological similarity to the other.
Results: __
Conclusion: The essence of the images were prioritized over the physical properties, suggesting that natural kinds are categorized by their biology rather than physical properties.
Method: Subjects were told two stories about change. One about a raccoon into a skunk and one about a coffeepot into a bird feeder.
Results: When the changed object was a natural kind, the subjects were much more resistant to say that they had changed. They were much more accepting when artifacts were changed.
Conclusion: Suggests that "natural kinds" are understood as different than artifacts
Distinction between natural kind and artifacts
Double dissociation study in which one patient had an impaired knowledge of inanimate objects but preserved knowledge of animals/flowers/etc and vice versa.
Suggests that "natural kinds" are categorized differently than artifacts
Semantic Networks Model
Visual model in which concepts, symbolically represented by nodes, and relations, represented by connections/pathways, form a network
Hierarchical Semantic Network
Method: Subjects performed a sentence verification task. The sentences varied in how distantly the two concepts were associated (a canary is a canary to a canary is a bird).
Results: There was a linearly increasing relationship between the broadness of the topic and the response time.
Conclusion: The longer RT can be attributed to the amount of time it takes to mentally travel through the links in the network.
Problem: This model cannot explain the typicality effect and it does not incorporate degrees of semantic relatedness.
Spreading Activation
All nodes have a baseline activation level. When one node is thought about, the activation level increases and the activation diffuses to surrounding nodes. All related nodes, within about two links, are also activated. Activation travels faster for shorter links (typicality effect).
Semantic Relatedness Model
Like the hierarchical network, but without strict levels. Link length represents relatedness (shorter for more related, longer for less related).
Semantic Priming
Method: Subjects perform a lexical decision task in which they must determine is a pair of words are both words. The pairs vary from both non-words, one non-word one word, two unrelated words, and two related words.
Results: RT is the shorter for related words compared to unrelated words.
Conclusion: Semantic priming occurs when recognition of a stimulus is influenced by previous presentation of similar stimuli. When subjects read the first word, the activation spreads to related words, by the time they read the second (related) word, the word has already been activated, making them faster to decide if it is a word.
PDP (Parallel Distributed Processing) model is a massive network of interconnected nodes modeled after neuronal networks (making it more biologically plausible). It represents concepts in a distributed vs local way, using weight to represent the strength of a connection. Learning occurs when experience changes the weights. This can model several things because it accurately predicts human behavior, including learning, and it accounts for graceful degradation and partial information (the idea that one small part can be destroyed without ruining it all).
Schemas in memory for text
Method: Subjects were given Eskimo folk stories (unfamiliar but meaningful) and were asked to recall the exact story. The amount of time between reading and recall varied.
Results: Of the mistakes that were made, most tended to normalize (Westernize) the story.
Conclusion: Memory is reconstructive, not reproductive, and it is influenced by schemas (organized knowledge about the world). Gaps in memory are filled in using schemas.
Schemas in memory otherwise
Method: Subjects hang out in a grad student's office for a few minutes not knowing that they would later be asked to recall as many items from the room as they could.
Results: Subjects mostly recalled schema consistent items consistent (desk, chair, typewriter). They recalled some schema inconsistent items (picnic basket). Errors mostly tended to be schema consistent (the subjects would recall books though there were none).
Conclusion: Memory gaps are reconstructed using schemas.
Schemas give context
Method: subjects are read a paragraph, a vague description of an everyday task. Some were given context before the reading, some were given context after, and some not at all.
Results: Those who had context before reading had significantly higher rates of accurate memory recall than those who were given context after or not at all.
Conclusion: The title provided context, a schema, as framework for interpreting and organizing the information.
Schemas and visual memory
Method: Subjects were shown a zoomed in picture for a brief time. They were then asked to do a drawing recall.
Results: Subjects tended to recall more about the picture than what was there.
Conclusion: The knowledge of what the objects were influenced memory by making people think the picture was more complete than it actually was.
Identity in schemas
Method: Subjects were asked to adopt a specific identity, either homebuyer or burglar. They were then told the story of two boys who came across a house while playing hooky. The story contained 18 "burglar details" and 18 "homeowner details." The subjects were then asked to recall the story.
Results: The homeowners were more likely to remember homeowner details and the burglars were more likely to remember burglar details.
Conclusion: Despite all having read the same passage, their memory of it differed depending on the identity they adopted beforehand. They were more likely to remember items consistent with their frame of mind at the time.
A script is a type of schema that deals with knowledge about ordered daily experiences
Method: Subjects are asked to study a sequence of events and then recall. Some sequences aligned with scripts and some did not, were out of order.
Results: When the event did not correspond to the order of the script, the subjects fixed the difference when recalling the events.
Conclusion: Scripts, as schemas, influence memory reconstruction.
Scripts in Eyewitness Memory
Method: Subjects are asked to come up with a script of a convenience store robbery. They then view a slideshow that depicts a robbery. 3 typical events are omitted. They are then asked to do a recognition test in which they identify whether certain images were a part of the slideshow.
Results: Subjects recalled seeing events that did not actually happen (more so the central parts than the peripheral parts).
Conclusion: Expectations guide perception.
Schemata help and hurt memory
Method: All subjects read stories about Nancy, a college student. Some subjects are read only the stories. Some subjects are also read a background story that implies she had an affair. All subjects are asked to recall the story.
Results: Target condition subjects accurately recall more details that align with the background story but they also had more intrusions of false memory.
Conclusion: Schemata can both help and hurt memory. More memory overall, both true and false.
Reconstructive Errors in Retrieval
Method: The subjects were given a story about a couple that has argument about whether or not to have kids prior to getting engaged. After some time, the subjects were asked to recall the story. Some subjects were casually told that the couple ended up getting married. The rest were not given an update.
Results: Those who knew the outcome recalled errors that resolved conflict. Their confidence of these memories was higher than that of the accurate memories.
Conclusion: Accuracy =/= confidence
Reconstructive Autobiographical Memories
Does who you are now affect who you thought you were?
Method: Subjects were asked to identify their political party affiliation. Four years later they were asked to do so again.
Results: Of those who did not change, 96% reported no change. Of those who did change, 91% reported no change.
Conclusion: A large percentage of subjects falsely remembered not changing. Their belief of who they are now shaped their belief of who they were in the past.