PSYB57H3 Lecture Notes - Sough, Connectionism, Spreading Activation

10 views6 pages
Published on 15 Apr 2013
School
Department
Course
Professor
Page:
of 6
Chapter 8: Associative theories of long-term memory
-Long term memory is huge and we can easily retrieve information meaning its really
fast and efficient.
-connections aid with putting the new things we learn with the old things we already
know
The Network Notion
-The connections are not simply just the pathways but our actual memory.
-Knowledge is represented by a sprawling network of connections which are a vast set
of associations.
How might the network work?
-Start with nodes which are individual ideas and then they are tied to each other via
connections aka associations or association links.
-To remember something you start at a node and travel via connections to the desired
information. However not all associations are equally strong, some are super fast and
some have no associations but we can get from point A to point B nevertheless just
through intermediate channels.
-The more attention you pay and the active intellectual engagement you have the
stronger you association links
The more connections(multiple connections) you make during the learning process the
more retrieval paths you are building for conjuring up that learned material.
Spreading activation
-Nodes are activated when it receives strong input signals, which means that the
association links act as an activation carrier that carry theses input signal from one
node to the next.
-Once a node has been activated it can therefore activate other nodes if they are
connected through the association links. Nodes gets activation from other nodes, the
more they receive the stronger they get so which increases they activation level to the
point where it reaches it response threshold after this happens the node fires.
-when it fires it activates other nodes and brings attention to itself which illustrates the
idea of finding a node in a network.
-sometimes they are at sub-threshold activation meaning they are not quite ready to fire,
what happens is that another node at this activation level can join (summate) with the
other one and then fire. Also because its at sub-threshold level its already warmed up
meaning weak inputs can set it off.
-works inline with the ideas of how neurons function so understanding nodes in a
biological context is fairly easy to imagine.
-spreading activation is the idea that each node gets activated, fires and activates other
nodes through association links thereby spreading the activation.
-We don’t get to choose what links get triggered or what paths to take, our minds just go
on this journey depending on the number of existing associations we have. Some paths
are more effective than others depending on how often and how strong these paths are.
Chapter 8: Associative theories of long-term memory
Evidence Favoring the Network Approach
Hints
-e.g. asked the capital city of south dakota, when you are asked this questions, the
nodes pertaining to your knowledge of south dakota start to fire and spread along the
association links, however if you don’t know much about south Dakota or haven’t
thought about it in a long time your associations links would be weak and you wont
remember but once given a hint you have more information. for instance the hint that its
a man’s name.
-Pierre is the capital and once the hint has been put out there then its being activated by
2 nodes simultaneously the node for south dakota and the one for men’s names making
it easier to retrieve
Context Reinstatement
-Recall in the same context you learned, it will help. Being underwater will conjure up
the same thoughts you had when you were done there, the environment serves as the
connections aka the association links.
-the nodes representing the thoughts of the context become connected to the nodes for
the target information.
More direct tests of the network claims
-The network approach makes its own predictions about memory, allowing a more direct
test of the claim that memory does indeed have an associative base.
Spread of Activation and Priming
-The idea that a node with sub-threshold activation can combine with another one at that
level to produce a connection. This idea explains why hints and context reinstatement
are helpful for reaching threshold. We can study this through the lexical-decision task.
-words vs letter strings pairs are presented on a screen and the person has to say yes
for words and no for letter strings.
-sometimes the 2 words in the pair were related sometimes not and they wanted to see
how this affected performance.
-when the words were related, for instance bread and butter, performance was high aka
response time was short. When recognizing that the first word bread is actually a word
the nodes are activates as well as other other nodes like butter, loaf, wheat etc so when
you see the 2nd word as being butter you can easily identify it because it was partially
activated just by looking at the word bread.
-performance showed that the task was 100ms faster with related words that primed
others.
Sentence Verification
-when you search for the word you go from node to node like travel the farther you go
the longer it takes. by measuring the time we can see how the process unfolds.
Chapter 8: Associative theories of long-term memory
-the sentence verification task is an experiment that involved people reading sentences of
a screen. e.g. a robin is a bird, a cat has claws with false sentences like a cat is a
bird.They had to hit true or false as quickly as possible pertaining to whatever was true
or false.
-they theorized that to see if its true they travelled from node to node and thought about
association links, the shorter time it to took for travel the closer the nodes are in
association(directly linked).
-slower connection for robin is an animal because they are indirectly connected, 2 step
connection.
-also cats have hearts, dogs have hearts are more associated with the animal node than
with the individual animals themselves making connection slow
-response time depend on the number of associative links it has to travel to get a
response. not the whole story some connections can be made more quickly than others.
The more typical the example the faster it is recognized. for instance a robin is a more
typical bird than a peacock.
-The principle of non-redundancy explains why we look at birds having feathers rather
than in each individual bird having feathers kinda like the cat, heart, animal thing but
some birds have prominent feathers like peacocks and dont follow the non-redundancy
principle like peacocks-->birds-->have feathers.
Degree of Fan
The speed of searching is influenced by the degree of fan which is the amount of nodes
being activated by one initial node. For instances you know more about robins than you
do about aardvarks so degree of fan is high for robins and low for aardvarks.
-degree of fan is important because once a node is activated it spreads along many
connections and your attention is thinly spread across all of these different things and
can kinda slow down the process on the other hand w aarkvarks lets say they have 5
connections then 20% goes to each connection and its faster to get the information you
need, granted it might not be much.
-An experiment to measure this was conducted. It showed sentences that had people in
certain places aka the doctor is in the bank, lawyer is in church, lawyer is in park,
fireman in the park. Some examples had the people in more than one place aka lawyer
was in the park and the church and other examples had more than one person in the
location aka fireman and lawyer were in the park. Just by going off the degree of span
idea they tested to see how much people remembered, due to the fact that the node for
lawyer was spread out aka wider fan it took longer for the people to accurately identify
where the lawyer was and that the sentence had been exposed to them in the prior
exposure round.
-so basically if activation were a fixed quantity--> More divisions-->less activation for
individual nodes-->longer time spent for activation to reach threshold-->longer time for a
response aka firing.
Retrieving information from a network.
-Searching through the network via associative links

Document Summary

Long term memory is huge and we can easily retrieve information meaning its really fast and efficient. Connections aid with putting the new things we learn with the old things we already know. The connections are not simply just the pathways but our actual memory. Knowledge is represented by a sprawling network of connections which are a vast set of associations. Start with nodes which are individual ideas and then they are tied to each other via connections aka associations or association links. To remember something you start at a node and travel via connections to the desired information. However not all associations are equally strong, some are super fast and some have no associations but we can get from point a to point b nevertheless just through intermediate channels. The more attention you pay and the active intellectual engagement you have the stronger you association links.