week 11: models in memory Flashcards
threshold model of recognition
- idea that there is a threshold of activation that a memory trace must exceed for it to be considered ‘old’ or before it can be recognized
- has recognition and rejection thresholds
descrete vs continuous modles for the threshold model
discrete: assumes features retrieved from memory are in an all or none manner
cont: info is retrieved along a continuum (based on signal detection theory ie hit, miss, correct rejection etc)
generate recognize model of recall
- unlike recognition, recall searches memory
- assumes recall in two stage process: generate memory traces (cues>memory traces>more connections) and recognize
recognition failure
- ## sometimes items can be recalled but not recognized
recall uses ____ to prompt retrieval where recognition uses ____
cues, the item itself
network model are about items aka ___ and their associations aka ____
nodes, links
spreading activation
- related nodes are activated through strength of associative links
adaptive control of thought model
- prepositions are made up of idea units (often 2 nodes and links)
- type (general) vs token (specific) nodes
- activation spread through the network but is a limited resource
- production (mental maps) memory vs declarative memory (prepositions)
- more practice = better production
hierarchical vs relatedness
- hierarchical: each item is connected but there is a built in natural hierarchical relationship w their own features
- relatedness: still a structure but more agonistic to do w hierarchy and more to do with how the concepts relate
latent semantic analysis
- massive training set of millions of text and from this, we can derive a stat estimating co occurrence (how often do two words appear together)
global matching models
- memories are assessed through processes that consider the entire set of available traces
- relation in memory occurs at retrieval
- patterns of info imposed on a common framework
multiple trace models
- many records for each type of memory
- memory traces are activated in parallel
distributed storage models
- many different memories are imposed on the same structure
- emphasizes associative memory
- common underlying framework and memories are connections between nodes
- theory of distributed associative memory, composite holographic associative retrieval model
search of associative memory model
- memories are stored in traces that contain content, associative and contextual info
- remembering = cue overlaps w features in a trace (activates it and can be retrieved)
- stronger is first
SAM recognition process
- use test item and context as cues
- determine recognition threshold (bias)
- is familiarity greater than criterion (yes or no judgement based on familiarity)
- doesnt have to just be one strong memory, can be many weak ones
SAM recall process
- question
- retrieval plan
- assemble probe cues in STM
- search (set selection)
- sampling trace
- retrieve the trace
- evaluation
- decide to or not to produce a response
MINERVA 2
- assumes memories are a string of features
- each trace can be conceptualized as a set of number, one per feature indicating presence/absence
- activates many memory traces which each have their own set of numbers
minerva 2 echo general def
- weighted composite of all activated traces
- sum total
- can produce schemas
- many echos narrow on trace
minerva 2 echo intensity
- strength of which theyre active
- weak= not known
minerva 2 echo content
- weighted average of the features of all active traces
parallel distributed processing model
- output, hidden, and output units
- translates input memory through middle to the output, takes on same form in output
- alters weights (strength of which input and output relate)
- interconnected units, loosly corresponding to neurons and axons
- translate input memory
- each memory has a different pattern
dual process model
- assumes theres two processes: familiarity and recollection
in dual process familiarity and recollection
- familiarity: faster, quantitative
- recollection: more conscious/effortful, qualitative
neural basis for remember and know processes
- remember: hippocampus only gets activated w highest confidence (yes or no is hippo driven)
- know: more based on familiarity and has linear relationship in media temporal lobe that surrounds hippo (linear scale)
