Bad things about prototype theory
- Conceptual confirmation and conceptual combination
- typicality gradients not stable
- no map of the conceptual structure of the theory and typicality not predicted
- Failure to provide good decision theory
- we don’t know how prototypes are generated.
Spatial metaphor for memory – functions to increase capacity for our memory of
things, is that memory is searched sequentially, through a space in which objects
have a stable spatial relation, and that those objects are stable.
You can use method of loci to increase memory capacity.
Lots of support that memory functions like the spatial metaphor but there is proof it
1)We rapidly know when we don’t know something , very hard to explain through
2)Things don’t seem to have a stable difference between locations, red and shoe not
close but red and blue are and blue and shoe are
3) problem is that memory doesn’t seem to involve stable objects, memory
reconstructed, does not store all the info it encounters, stores relevant instructions.
Memory is integrative, reconstructive to make you smart, allow you to adjust. All of
these things enhance memory to make you intelligent but makes you less accurate.
Memory is already behaving in a more sophisticated fashion not like it was
proposed as simple.
One feature of the reconstructive process is that it adjusts itself to the context of
retrieval, memory works more on what is relevant to you and what is relevant to the
context, memory will adjust accordingly. Experiment to show this
Memory reconstructive or reproductive?
Loftus and Zannil 1975, showed how inaccurate memory is , wanted to show ppl
how much memory is influenced by context. Showed subjects a car accident
question 1) did you see the broken headlight? 2) Did you see a broken headlight ?
Subjects asked form 1 were more likely to report seeing the headlight than form 2 ,
regardless of weather there was a broken headlight. Ppl pick up on this information,
think it’s relevant and alter their memory. What is predictive of what participants
say, is which article was used in the question.
Putnon 1979 follow up experiment. You can improve people’s memory by hypnosis.
Repeated the experiment but now accident involving a car and bicycle, same thing
using the two types of questions, either the article “a” is used or the article “the” is
used. Difference is hypnotize some subjects and don’t hypnotize others. Replicated loft’s result, more ppl said yes once again to statement 1, hypnosis increased that
error, more pliable to the context. Hypnosis increases the ease at which memory will
adjust itself to the context at hand.
This became a political football – Loftus was challenging the claim that people can
reduce these repressed memories from childhood, ppl were reporting that they
were uncovering repressed memories in therapy. Somebody is asking you questions
in therapy and they are very suggestive. Many convictions occurring and she was
arguing that this kind of evidence is not very reliable. So she was hit to fight against
that, since ppl did not like that.
Important feature and issue that is facing memory. We have lots of information out
there. You have lots of information in your head. Problem you can only hold about 4
chucks of information. Bottleneck problem for memory, your working memory can
only hold 4 chunks but you want to get out a lot of LTM. You have to do important
selection; don’t want all your memory active at the same time. Working memory is a
higher order relevance filter, zero in on what is most relevant. Memory to deal with
this bottleneck its going to try to grab the gist, cant remember all the info presented.
A chunk is a gestalt an integrative whole. Functions as a unit/ item. I can get 3 items
from 8 words. What we see is memory is facing similar issues to categorization
trying to select relevant information and integrate that information together in a
We should predict that LTM is largely not going to pay attention to the surface form
of information but trying to get at the meaning. Lots of evidence to support this
notion for memory doing this process (chunking, gisting, meaning, integration) all
trying to point to the fact that memory is trying to zero in on the surface features.
Surface for vs integration form ?
Experiment conducted by Sachs 1967, have ps read a passage and sentence, then
ask the subject did you read the sentence A. if the test sentence comes right after the
sentence in the text, ppl are accurate. If the test sentence came a couple of minutes
after the target sentence subjects were not very accurate about the structure of the
sentence. When you switch from the active to the passive ppl cant tell. Coz ppl care
more about the meaning. If there is a change in meaning however they can tell.
People cannot tell the change in surface form.
If what memory is doing is picking up on the meaning to deal with the bottleneck it
might be that our confidence in what we are remembering not be as accurate as we
might think. Many ppl are confident about remembering something. Memory is
organized for meaning not accuracy, does confidence track accuracy or
How did they test that ? Bransford and Franks 1971 – reproduce Saks experiment, going to give ppl
sentences and give them new versions of the sentences. all that is changed is the
surface form not the meaning in the new sentence.
Old version boy hit the dog , new version dog hit by the boy.
4 sentences that form a set. 1) the rock hit the hut 2) the tiny hut was by the river
3) The rock hit the tiny hut by the river. 4)The rock rolled down the mountain and
hit the tiny hut by the river.
All single sentence, numbered according to the number of propositions each
sentence contains. Sentence – any string of words that is considered meaningfully
complete, proposition is a claim that can be evaluated for true or false.
Sentence 1 has 1 proposition , making 1 claim, where as the second sentence has 2
claims, the hut is tiny and it was by the river. 3 sentence has 3 claims and 4th has 4
On way to measure how meaningful a sentence is, is by counting how many
propositions it contains. As you go from 1-4 you get more and more
propositions. Subjects heard all of these statements but never ever heard new
version of sentence 4 but heard the new version of 1,2,3. Had ppl graph their
confidence, how confident are you that you saw it before? 5 being very confident
and -5 being not as accurate.
Graph shows that people could not determine changes of surface form, can’t
remember surface forms. Replicates Sachs since 2 graphs are overlapping a
graph for new and old. The important finding is that people are most confident
that they have heard the new version of 4 and could not distinguish between old
and new form. Even though at no time were they given that info. What happened
is that memory is creating an integration of information. That is what is most
highly remembered. Confidence is tracking how meaningfully integrated the
Bring all the average faces together, and people will find it attractive not average.
Coz more familiar = more attractive. The other thing is that people find these
faces easier to process coz information is so integrated.
Memory seems to be really active behind the scene even when our conscious is
not doing anything. Things are happening in the there. We tend to think that
memory is a container and what is it does is hold information, that is a passive
storage. But instead we see that memory is an active integrative processes.
Challenged a bunch of ideas, sequential spatial, reproductive vs. reconstructive,
surface form vs. integration of info, now we can challenge the idea that memory
is passive storage.
Active or passive storage?
Craik and Lockhart levels of processing 3 statements a) is table in capital letters? b ) does market rhyme w