Long-term memory
Recognition and retrieval: Retrieval as reconstruction: Hierarchical network:
- Regnition task - identification of a present item - Knowledge is used to reconstruct - Collins & Ouillian (1969) - sub concepts have
as known or unknown, list of items presented events during recall all features of super-concepts plus
and ppts asked to memories the list, - Problem of source management differentiating features
presentation of old and new items, ppts had to - McDermott (1966) - presented word list - Typically effects can’t be explained - an
decide whether the item is old or new to ppts, recognition test with old words additional assumption is necessary
- Measurement of RT and correctness for each from middle of list and new word, - Hierarchy doesn’t always work - ppts are
decision delayed recognition test 2 days later, faster to decide that an ostrich is an animal
- Shepard (1967) - 540 words in learning phase, ppts probably activate a concept that than a bird
60 test pairs of old and new words, 88% links the words together (e.g. ‘sleep’ for
correct decisions, same exp with 612 pics - ‘bed’, ‘nap’, ‘pillow’) when hearing the
97% correct words, activation of concept during Spreading activation model:
- Performance in recognition tasks usually encoding make it impossible to - Concepts in memory are linked to each other
better than recall tasks differentiate it from other concepts - Concepts are the nodes, semantic relations are
- Recognition performance remains high even activated by external stimulation the links
after longer retention intervals - Loftus & Palmer - after “smashed” more
- Problem in recognition taks are distractors - ppts reported seeing broken glass even Spreading activation:
new stimuli are often considered as known though there wasn’t any, misleading info - Evidence - fragment completion, stem completion,
stimuli if they’re similar to old stimuli, events leads to false memories but ppts might lexical decision
following critical event can falsify recognition be quite confident about their memories - 2 types of priming:
- Recognition performance is high but can easily (problem for eyewitnesses) - Repetition - priming effects depending on
be deceived whether the word was heard before or not,
evidence for activation of the nodes
- Semantic priming - priming effects for words
Cued recall: Forgetting: linked to the primed word in the network,
- Retrieval of info from memory depends on = loss of info and impossibility to access info evidence for the spreading activation over the
cues - Records damaged by accident = relations between nodes
- Tulving & Thomson (1973) - cues used during pathological forgetting - Problems of spreading activation models;
initial learning are more effective during later - Retrieval cues = cues that make it - Vague assumptions
retrieval than novel cues, most effective possible to reach a node in LTM - Can’t determine location or strengths of links
retrieval cues are those that were stored along - Memories fade away or decay gradually if - Can’t determine how far the activation spreads
with the memory of the experience unused
- Godden & Baddeley (1975) - words heard - Limits of decay theories - time alone
underwater are best recalled underwater, doesn’t cause anything, it’s processes
words heard on land are best recalled on land occurring in time that make things happen
- Grant et al - similar ^
Retrieval for understanding:
- Memory retrieval is used to structure the world, its how we
make sense of what we see and its the basis of understanding
- Perception depends on knowledge
- Understanding depends on knowledge
Recognition and retrieval: Retrieval as reconstruction: Hierarchical network:
- Regnition task - identification of a present item - Knowledge is used to reconstruct - Collins & Ouillian (1969) - sub concepts have
as known or unknown, list of items presented events during recall all features of super-concepts plus
and ppts asked to memories the list, - Problem of source management differentiating features
presentation of old and new items, ppts had to - McDermott (1966) - presented word list - Typically effects can’t be explained - an
decide whether the item is old or new to ppts, recognition test with old words additional assumption is necessary
- Measurement of RT and correctness for each from middle of list and new word, - Hierarchy doesn’t always work - ppts are
decision delayed recognition test 2 days later, faster to decide that an ostrich is an animal
- Shepard (1967) - 540 words in learning phase, ppts probably activate a concept that than a bird
60 test pairs of old and new words, 88% links the words together (e.g. ‘sleep’ for
correct decisions, same exp with 612 pics - ‘bed’, ‘nap’, ‘pillow’) when hearing the
97% correct words, activation of concept during Spreading activation model:
- Performance in recognition tasks usually encoding make it impossible to - Concepts in memory are linked to each other
better than recall tasks differentiate it from other concepts - Concepts are the nodes, semantic relations are
- Recognition performance remains high even activated by external stimulation the links
after longer retention intervals - Loftus & Palmer - after “smashed” more
- Problem in recognition taks are distractors - ppts reported seeing broken glass even Spreading activation:
new stimuli are often considered as known though there wasn’t any, misleading info - Evidence - fragment completion, stem completion,
stimuli if they’re similar to old stimuli, events leads to false memories but ppts might lexical decision
following critical event can falsify recognition be quite confident about their memories - 2 types of priming:
- Recognition performance is high but can easily (problem for eyewitnesses) - Repetition - priming effects depending on
be deceived whether the word was heard before or not,
evidence for activation of the nodes
- Semantic priming - priming effects for words
Cued recall: Forgetting: linked to the primed word in the network,
- Retrieval of info from memory depends on = loss of info and impossibility to access info evidence for the spreading activation over the
cues - Records damaged by accident = relations between nodes
- Tulving & Thomson (1973) - cues used during pathological forgetting - Problems of spreading activation models;
initial learning are more effective during later - Retrieval cues = cues that make it - Vague assumptions
retrieval than novel cues, most effective possible to reach a node in LTM - Can’t determine location or strengths of links
retrieval cues are those that were stored along - Memories fade away or decay gradually if - Can’t determine how far the activation spreads
with the memory of the experience unused
- Godden & Baddeley (1975) - words heard - Limits of decay theories - time alone
underwater are best recalled underwater, doesn’t cause anything, it’s processes
words heard on land are best recalled on land occurring in time that make things happen
- Grant et al - similar ^
Retrieval for understanding:
- Memory retrieval is used to structure the world, its how we
make sense of what we see and its the basis of understanding
- Perception depends on knowledge
- Understanding depends on knowledge
