認知心理学7. Memory Codes: 記憶コード

心理学 認知心理学

認知心理学もくじ

Recap/preview

memory code

  • The format of the representation used to store the memory.
    • Formats: Acoustic (phonemic), Semantic, Visual.
  • The format (physical, phonemic, semantic) of information encoded into memory (CP 130)
  • Why doesn’t our mind create memories in just one code. Simple.
    • Advantages to multiple codes. Deficit in one can still learn with others.
  • How do we store information in LTM?

Multi-Store Theory of Memory

  • Alternative to multi-store model memory.
  • Atkinson and Shiffrin
    • Used separate stores to explain memory differences.
    • Emphasized rehearsal.
  • Criticisms:
    • Capacity of STM more variable then 7+/-2.
      • Criticisims of multi-store: Variable capacity in STM. Can recall up to 20 words when repeating sentence. What exactly is a chunk? Abstract.
    • Evidence that both STM and LTM use multiple codes.
      • Both STM (mainly phonemic) and LTM (mainly semantic) multiple codes. Form chunk (FBI) in STM requires semantic. Need acoustic code in LTM for pronunciation.
    • Best evidence for separate stores is decay rates.
      • Decay rates: Fast in STM 30 sec. Slow in LTM minutes to years.

levels of processing

  • had an impact on memory research. Experiments have provided a lot of support. Fit well with information-processing stages.
  • A theory that proposes that "deeper" (semantic) levels of processing enhance memory (CP 132)
  • ''Craik & Lockhart (1972)''

Encoding

  • maintenance rehearsal
    • Rehearsal that keeps information active in STM (CP 135)
  • ''Craik and Watkins (1973)'' didn't quite agree with rehearsal.
    • Remember the last word presented that begins with a g
    • daughter oil rifle garden grain table football anchor giraffe
    • After 27 lists, their memory for all words was tested.

Watkins (1973)はこのリハーサルについて、維持リハーサル(maintenance rehearsal)と精緻化リハーサル(elaborative rehearsal)の2つのタイプのリハーサルが存在すること、これらのリハーサルのうち、維持リハーサルは短期貯蔵庫内に情報をためる機能をし、精緻化リハーサルは情報を短期記憶に転送する機能を持つと仮定している。
（情報処理理論と幼児教育）

維持リハーサル
維持リハーサルとは、機械的リハーサルまたは機械的反復とも呼ばれるように、英単語を声に出して何度も暗唱することで記憶を保持するやり方です。
維持リハーサルは、短期記憶の中に保持する機能を持ちますが、長期記憶にうつす効果は弱いようです。
いくら単語を暗誦しても、しばらく経ったら忘れてしまうという経験はないでしょうか。
それは、維持リハーサルのレベルであったために、長期記憶に残らなかったためなのです。

精緻化リハーサル
精緻化リハーサルとは、統合化リハーサルとも呼ばれ、記憶したい内容の意味づけをしたり、他の内容と関連づけることで記憶の構造化を図ったりすることです。
具体的に言えば、次のようなものが精緻化リハーサルの例です。

  • 語呂合わせ
  • カテゴリー分け（white,black,red・・・という単語を「色」というカテゴリーで覚えるなど）
  • 関連付け（徳川家康→江戸幕府など）
  • チャンキング（複数の記憶をまとめて１つの記憶にすること。たとえば「犬」と「猫」について覚えたいときは「犬と猫が喧嘩している様子」として覚えるなど）

精緻化リハーサルは、短期記憶を長期記憶にうつすのに効果的です。
語呂合わせは、特に覚えようとしたわけでもないのになぜか覚えているものですよね。それが精緻化リハーサルの効果といえます。
この精緻化リハーサルを積極的にとりいれていくことが、暗記には非常に効果的です。
(偏差値80の勉強法)

Remember, Atkinson and Shiffrin emphasized rehearsal as the dominant way info is moved into LTM.
Craik and Lockhart: Rehearsal doesn’t necessarily result in learning.
Rehearsal used for maintainence in STM. Need to maintain things in STM often but we don’t need the info in LTM.
Effectiveness of rehearsal depends on the level that it is processed.
Rehearsal that results in learning is due to attending to the meaning during rehearsal.

Craik and watkins experiment.

Craik and Watkins

  • daughter, oil, rifle, garden, grain, table, football, anchor, giraffe
  • If rehearsal automatically leads to learning - should be more likely to remember grain than garden

http://dl.dropbox.com/u/3770752/wiki/cognitive/07/Craik.png

  • no significant difference here
    • rehearsal does not affect for learning so well.
    • rehearsalは単にSTMのdurationを増やすためだけ。

Test faces. Old new and caricatures.

People encode features in such a way to make them more distinct then the original.

more Supporting Evidence: ''Hyde and Jenkins (1969)''

  • Hyde and Jenkins (1969)
    • incidental learning
      • A task that requires people to make judgments about stimuli without knowing that they will later be tested on their recall of the stimuli (CP 136)
      • judgment task
    • Four groups:
    1. Told to remember words
    2. Does the word contain an ‘e’?
    3. How many letters in the word?
    4. Is the word pleasant?

Hyde and Jenkins, published before Craiks an Lockharts levels of processing. Probably influenced it
incidental learning task: judgment task and a surprise recall task.
Experiment. Lists of primary associate (words that are strongly associated with each other, as typically measured by asking people to provide associations to words (CP 136)), words that are highly associated (bed, pillow)
Conditions told to remember
Contain ‘e’: focus on spelling of word.
How many letters: focus on spelling
Rate Pleasant/unplasent: focuses is on meaning.

  • clustering
    • Percentage of occasions in which a word is followed by its primary associate during the free recall of words (CP 137)
  • Results:
Recall Cluster
‘e’ 9.4 words 26% shallow
+ letters: 9.9 words 31% shallow
Pleasant: 16.3 words 68% deep
Remember 16.1 words 64% deep
  • Incidental learning as effective as intentional learning when meaning of word was considered.
  • Remember the words pairs highly associated. Words easier to recall if clustered, need to focus on meaning. Number of associated pairs recalled together divide by total. Same pattern as recall. Groups differed in the use of meaning to aid recall.

Types coding

  • structural coding:
    • focus is on surface characteristics
    • memory code that emphasizes the physical structure of the stimulus (CP 138)
      • Does the word have the letter 'e'?
  • phonemic coding:
    • focus is on pronunciation
    • A memory code that emphasizes the pronunciation of the stimulus (CP 138)
      • Does the word rhyme with "gate"?
  • semantic coding:
    • focus is on the meaning of words.
    • A memory code based on the meaning of the stimulus (CP 138)
      • Does the word fit in a given sentence.

Craik and watinks experiment example of the how effectiveness of different coding strategies.

Craik & Tulving, 1975

われわれは日常さまざまな出来事を経験し学習するが、そのすべてを意図的に学習しているわけではない。このような、意図せずに行なわれる学習を、偶発学習という。Craik & Tulving（1975）は、この偶発学習を用いて、記憶の保持量と情報処理の深さの関係について、実験を通して検討している。

　彼らは、被験者に記憶するようには告げずにいくつかの単語を呈示した。その際、被験者の単語に対する処理の深さを制御するための方向づけ課題を、単語を呈示する前に与えた。方向づけ課題は、その単語が「大文字で書かれているか（形態的処理）」、「韻を踏んでいるか（音韻的処理）」、「文脈にあてはまるか（意味的処理）」を判断させるものであった。このような手続き後、被験者に単語の保持テストを実施したところ、「形態的処理＜音韻的処理＜意味的処理」の順で単語の保持率が高かった。このことから、情報処理の水準には段階があり、意味的課題（「文脈にあてはまるか」）ではより深い水準の認知処理が必要とされるため、記憶にも影響を与えるのではないかということが示唆されている（処理水準説：記憶は覚えるべき項目がどのようにして処理されるのかということにより規定され、また記憶の研究はそうした処理と記憶の関係を明らかにするべきであるという提案の意）。
(偶発学習における被験者の自己関連づけ効果に関する一考察)

  • Why does semantic encoding lead to more accurate recall (vs. surface or acoustic)?
    • structure level, acoustic model, or semantic level?
  • incidental learning task.
  • Present question that focuses on 1 of the 3 levels of processing and then a word.
  • Results:

http://dl.dropbox.com/u/3770752/wiki/cognitive/07/Craik%20%26%20Tulving%201975.png
reaction time.

  • We see that processing at the semantic level is best for recall and that it takes the most time.
  • So is processing time
  • processing time related to recall
  • Recall was best for semantic processing and it also took the longest
  • So is time of processing the reason for better retention?
  • Respond yes only to words that contain two consonants followed by two vowels followed by a consonant.

Limitations on Depth of Processing

  • Explanation for different rates of forgetting (explain levels of processing account. STM acoustic, ltm semantic.) was too simple.
  • Why does semantic encoding lead to more accurate recall (vs. surface or acoustic)?
  • Also, why aren’t all semantic task performances equal?
  • Elaboration- Craik & Tulving, 1975 (experiment 7)
    • “does the given word fit into the sentence?”

Elaboration

  • When we store an item, we also store information related to the item.
  • Happens at the semantic level.
  • Explains why there are differences between two semantic tasks.

Increase semantic elaboration with a richer more detailed context.

Craik and Tulving (1975) Experiment 7

    • Ss read sentences and to decide if the word fit.
    • noncued recall task.
      • recall that occurs without hints or cues provided by the experimenter (CP 141)
    • cued recall task.
      • Recall that occurs with hints or cues, such as providing the questions asked during the judgment phase of a task (CP 141)
  • Elaboration conditions:
    • Conditions differed in the amount and detail of contextual information
  • Simple: The ( ) is torn.
  • Medium: The ( ) frightened the children.
  • Complex: The great bird swooped down and carried off the struggling ( ).
  • Results
    • Free recall v. Cued recall

http://dl.dropbox.com/u/3770752/wiki/cognitive/07/result.png

Notice that the type of encoding didn’t help for ‘no’ responses.
Sometimes the word didn’t fit in with the sentences- seeing the word RAIN instead of TOWEL in the example above.
So just mere elaboration didn’t help, just like mere repetition didn’t help with retention.
''Relevance is important''.
The word must be consistent.
Elaboration does help for recall

Not just elaboration…

  • Relevance is important.
  • Stein and Bransford (1979)
    1. The fat man read the sign. (control group)
    2. The fat man read the sign that was 2 feet high.
      1. imprecise elaboration:
        1. no relation between the adjective fat and the height of the sign.
        2. Provision or generation of additional material unrelated to remembered material (CP 142)
    3. The fat man read the sign warning about thin ice.
      1. precise elaboration:
        1. the adjective fat can be associated with the danger of thin ice.
        2. Provision or generation of additional material closely related to remembered material (CP 142)
    4. The fat man read the sign...(generated elaboration - own)
  • After reading the sentences they had to recall word.
  • Result
    • fill in the blank question: The ( ) man read the sign…

http://dl.dropbox.com/u/3770752/wiki/cognitive/07/result2.png

  • self-generation
    • Generation of items by participants in an experiment, rather than the provision of these items by the experimenter (CP 142)

Distinctiveness

  • distinctive item
    • Item that is distinct from other items
    • An item different in appearance or meaning from other item (CP 142)
  • Schmidt (1991)
    • primary distinctiveness:
      • An item distinct from other items in the immediate context (CP 143)
      • SPOON FORK TREE KNIFE GLASS
        • Treeを覚えている
      • GREEN HOUSE STREET BEER
        • Houseを覚えている (ほかは全部EE)
  • secondary distinctiveness:
    • Item that is distinct from other items in LTM
    • An item distinct from items stored in long term memory(CP 143)
    • orthographic distinctiveness:
      • Lowercase words that have an unusual shape (CP 143)
  • emotional distinctiveness:
    • Items that produce intense emotions.
    • Items that produce an intense emotional reaction (CP 143)
    • 9/11とか
  • processing distinctiveness:
    • Depends on how we use the memory code to process information.
    • Creation of a memory code that makes that memory distinct from other memories (CP 145)
    • Ex. Elaboration
    • “House Bunny”
  • Light & Carter-Sobell
    • Studied word pairs. (e.g. strawberry (stumulus) - jam(response))
    • Recognition of second word with a cue
      • Strawberry - jam
      • Traffic - jam
      • Grape - jam

result

  • Original pairs = .6
    • 60% recall rate.
  • Same meaning = .4
  • Different meaning = .2
  • We exaggerate distinct features.
  • Mauro & Kubovy (1992)
    • Ss viewed 100 identi-kit faces.
    • Recall originals from 300 test faces.
  • Results
    • Caricatures were recognized better
    • 特徴が普通より大げさに覚えられている。

Test faces. Old new and caricatures.

People encode features in such a way to make them more distinct then the original.

認知心理学もくじ