Memory Codes

1
Chapter 6

• Memory Codes

2
Two major topics

• Levels of processing
• Encoding specificity

3
Levels of Processing
Originally proposed by Craik and Lockhart (1972)
as an alternative to the modal model of memory
4
The levels of processing framework proposes that
there is a single memory system rather than three
Modal Model
Short Term Memory
Long Term Memory
Sensory Memory
Memory
NO!
YES!
5
The memory trace is a byproduct of processing
Processing
Memory
Memory Trace
Input
6
Processing can vary with regard to its level or
depth
Processing
Shallow
Deep
7
Levels of Processing

• Shallow Only the superficial, physical
  characteristics of the stimulus are processed
• Intermediate The meaning of the stimulus is
  identified
• Deep (elaborative) The stimulus is linked to
  other relevant information

8
Note Levels of processing should be thought of
as a continuum with shallow, intermediate and
deep representing relative points on that
continuum.
9
Since the memory trace is the byproduct of
processing, different levels of processing will
result in different memory traces
Processing
Memory
Trace
Shallow
Input
Intermediate
Trace
Trace
Deep
10
Deeper levels of processing not only result in
traces that are qualitatively different, but also
in traces that are more persistent
Processing
Memory
Transient Trace
Shallow
Memory Improves
Input
Intermediate
Trace
Persistent Trace
Deep
11
The Illusion of Short and Long-Term Memory
The levels of processing framework basically
looks at short and long-term memory as an illusion
12
The Illusion of Short and Long-Term Memory
The levels of processing framework basically
looks at short and long-term memory as an
illusion The illusion of short term memory occurs
when information is processed at a shallow level
13
The Illusion of Short and Long-Term Memory
The levels of processing framework basically
looks at short and long-term memory as an
illusion The illusion of short term memory occurs
when information is processed at a shallow
level The illusion of long term memory occurs
when information is processed at a deep level
14
Research that supports the levels of processing
approach to memory frequently uses the incidental
learning paradigm
15
The Incidental Learning Paradigm

• Participants are presented the learning materials
  under some pretext other than a study on memory.
• After the exposure to the learning material,
  participants are given a surprise memory test.

16
Example

• Participants given a list of words and asked to
  rate each in terms of its pleasantness. They are
  told this is part of a word norming study.
• The rating task is referred to as an orienting
  task since it governs how the participant
  interacts with the learning material.
• When the rating task is complete, the
  participants are asked to recall as many of the
  words as they can.

17
This is referred to as incidental learning, since
participants arent explicitly instructed to
learn. Presumably any learning has been
incidental as opposed to intentional
18
Test of the levels of processing framework

• The level of processing is manipulated by varying
  the orienting task in an incidental learning
  paradigm
• In some instances orienting tasks that require
  only shallow levels of processing are used
• In other instances orienting tasks that require
  deeper levels of processing are used

19
Examples of orienting tasks that require shallow
levels of processing

• Count the number of syllables
• Count the number of vowels
• Generate a word that rhymes
• Count the number of upper case letters

20
Examples of orienting tasks that require a deep
level of processing

• Generate a synonym
• Generate an antonym
• Rate the pleasantness of the word
• Indicate if the word will fit in a sentence

21
Typical Finding
Memory is significantly better when an orienting
task is used which requires deep levels of
processing. Often performance with these
orienting tasks equals that found with
intentional learning instructions.
22
Doubts about depth
As appealing as the levels of processing approach
may be, some have raised doubts about it.
23
Empirical data that is problematic
Consider the following experiment by Stein
(1978) Participants presented words in the
context of one of two orienting tasks Shallow
(non-semantic) Is the letter c capitalized?
-- roCk Deep (semantic) Does this word (roCk)
fit in the following sentence The __________
rolled down the hill.
24
Memory for the words is tested two
ways Non-semantic Test Pick the word just
seen from the following Rock, rOck, roCk,
rocK Semantic Test Pick the words just seen
from a larger set
25
This results in the following conditions
Test
Semantic
Non-semantic
Semantic
Orienting Task
Non-semantic
26
What were the findings?
Test
Semantic
Non-semantic
Semantic
Orienting Task
Non-semantic
27
Conclusion
Non-semantic information can persist as long as
semantic information. That is inconsistent with
levels of processing framework.
28
Question

• Why did a non-semantic orienting task produce
  better memory than a semantic orienting task when
  a non-semantic test was used?

29
One possible explanation can be found in the
concept of transfer-appropriate processing.
The idea of transfer appropriate processing is
that memory will be a function of the extent to
which processing at study is the same (or
similar) to the processing at test
30
A Transfer-Appropriate Processing Explanation
The non-semantic orienting task produced superior
performance on the non-semantic test because both
required the participant to process the
orthographic characteristics of the word.
31
Another Criticism of Levels of Processing
Levels of processing really doesnt explain why
semantic processing should result in better memory
32
Possible Explanations of Semantic Processing
Effects

• Elaboration Hypothesis
• Distinctiveness Hypothesis
• Transfer Appropriate Processing

33
Elaboration Hypothesis

• Deep, semantic processing results in elaboration
  where the item to be remembered becomes linked to
  other things that are related
• Those other things become potential retrieval
  cues which can facilitate access

34
Some supportive evidence for the elaboration
hypothesis
Craik and Tulving asked participants to judge
whether words fit into a sentence. They then
varied the complexity of the sentence
Simple She cooked the _______________ Medium
The ripe ______________ tasted delicious Complex
The small lady angrily picked up the red
_____________
35
Some supportive evidence for the elaboration
hypothesis
Finding Better recall was obtained for the
complex sentences, but only when the word was
judged to fit in the sentence.
36
Distinctiveness Hypothesis

• Deep, elaborative processing results in a
  distinctive memory trace (the more you know about
  something, the more distinct it becomes)
• Distinctive traces are less likely to be confused
  with other traces.

37
Some Evidence Supports the Distinctiveness
Hypothesis
In a study of face recognition, Winograd had
participants study faces in one of two
conditions Elaborative coding Participants
scanned each picture in an effort to find the
most distinctive feature Distinctive encoding
Participants focused on the most distinctive
feature (determined by earlier ratings)
38
Some Evidence Supports the Distinctiveness
Hypothesis
Finding both groups produced equal levels of
recognition performance, suggesting that
elaboration is effective because it produces
distinctive encodings.
39
Transfer-Appropriate Processing
Typically, tests of memory are based on meaning.
Therefore, semantic processing at the time of
encoding will typically be effective than
non-semantic processing.
40
Concluding thoughts about levels of processing

• The notion that semantic processing typically
  leads to better memory seems to be widely
  accepted (referred to as a levels of processing
  effect)
• It does not appear that the levels of processing
  framework has replaced the modal model
• Levels of processing has focused attention on the
  importance of how information is processed

41
Encoding Specificity
Similar to the notion of transfer appropriate
processing in that it emphasizes the relationship
between study and test conditions.
42
Encoding Specificity
Specifically, encoding specificity asserts that
events are encoded in a specific context. Memory
for the event depends on the extent to which the
original context is reinstated at the time of
test.
43
Encoding specificity is best understood in the
context of retrieval cues and memory
44
Retrieval cues are prompts that help someone to
remember a past event.
45
A Demonstration
Read each of the following word pairs
lake - VALLEY ruby - DIAMOND carrot -
TOMATO hammer - SAW tree - GRASS strawberry -
JAM book - MAGAZINE paper - PENCIL
car - BUS sun - SWIM road - STREET shirt -
SHOES house - YARD pet - DOG television -
RADIO chicken - BEEF
sky - BIRD mind - BODY frog - SNAKE cloud -
SKY soup - CHICKEN bell - SOUND light -
POST table - CHAIR
46
Write down as many of the words that were printed
in CAPITAL LETTERS as you can. If you are not
sure of a word, feel free to guess.
47
For a second time, write down as many of the
words printed in CAPITAL LETTERS as you can. Use
the following words to help prompt your recall.
lake ruby carrot hammer tree strawberry book paper
car sun road shirt house pet television chicken
sky mind frog cloud soup bell light table
48
Did you recall more words the second time? Why?
49
What makes a cue effective?

• Associative strength theory an effective cue is
  one that has a strong pre-experimental
  association with the item to be remembered
• Encoding specificity theory an effective cue is
  one that was encoded with the item to be
  remembered (i.e., was part of the encoding
  context).

50
A test to distinguish between the theories
Cue at Test
Strong
Weak
Study Lake - River Test Lake - ?
Study Lake - River Test Fast - ?
Strong
Cue at Study
Study Fast - River Test Lake - ?
Study Fast - River Test Fast - ?
Weak
51
Encoding specificity provides a nice explanation
of how context influences memory and a phenomenon
referred to as state-dependent memory
52
State dependent memory refers to instances when
memory improves when the individual is placed
back in the original physical or emotional state
in which the event occurred.
53
Encoding specificity also provides an explanation
for why recognition is typically easier than
recall.
54
Examples of recognition and recall

• Recognition
• Picking out a suspect in a police line up
• Hearing a familiar song on the radio
• A multiple choice test (?)
• Recall
• Telling a friend what you did Friday night
• Saying the Pledge of Allegiance
• An essay test (?)

55
Two Theories

• Generation-recognition theory Underlying
  processes are different.
• Recall involves two processes generation and
  recognition
• recognition involves one process recognition
• Encoding specificity theory The underlying
  processes are the same.
• The difference is in the quality of the cues that
  are provided.

56
Why is recognition memory easier than recall
memory?

• Generation-recognition theory Recognition does
  not involve the generation process required by
  recall. Therefore it is easier
• Single process theory With recognition the
  individual gets the best cue possible -- the item
  itself.

57
A test of the two theories
The logic
According to the generation-recognition theory,
one should never be able to recall a word that
they cant recognize. Thats because, recall
implies recognition. Recognition failure of a
recallable word would be a finding that would be
inconsistent with the generation-recognition
theory
58
Recognition failure of recallable words

• Phase 1 Study words pairs (blue - butter),
  receive cued recall test (blue - ?)
• Phase 2 Study new word pairs (road - shrimp),
  receive cued recall test (road - ?)
• Phase 3 Study new word pairs (glue - chair), no
  test
• Phase 4 Word association task (write down four
  words that come to mind in response to table)
• Phase 5 Circle any word generated on the word
  association task that was also on the last list.
• Phase 6 Cued recall test (glue - ?)

59
Result
Participants frequently could not recognize words
(phase 5) that they were able to recall (phase 6)
60
Result
Participants frequently could not recognize words
(phase 5) that they were able to recall (phase 6)
In other words, the experiment generated
recognition failure of recallable words. These
results are inconsistent with the
generation-recognition theory.
61
Encoding specificity theory provides an
explanation
According to encoding specificity theory, the
most effective cues will be those that were part
of the original encoding context.
62
Encoding specificity theory provides an
explanation
According to encoding specificity theory, the
most effective cues will be those that were part
of the original encoding context. Since glue
was part of the original encoding context, it is
an effective cue for table. On the recognition
test (phase 5), table occurs in a new context
therefore, it is difficult to remember (i.e., the
cues were ineffective)