Craik and Lockhart (1972) an alternative to the duplex model of memory; no natural stages in memory; emphasis on coding .

 Main Idea - Stimuli can be encoded at varying depths from shallow to deep.

  Shallow processing corresponds to activation of sensory analyzers. Several degrees of depth at this level ; sensing, pattern recognition, attention, rote repetition

 Deep processing involves understanding or interpreting the stimulus; a conceptual analysis. E.g. reading a novel, few perceptual features activate many perceptual units which activate larger conceptual units (schema for a love story)

Levels describes processes versus structures.

 Assumption: a stimulus can be processed in many ways: physical, acoustic, semantic each operation requiring a deeper degree of processing. The different levels of processing do not refer to memory stores; it refers to the coding processes engaged in to get information into memory. Information is processed through an ordered hierarchy of levels beginning with physical features to phonemic to semantic.

  Assumption: deeper processing leads to better memory: Hyde and Jenkin's study; Craik and Tulving (1975)(capitol letters, rhymes with, the girl placed the --- on the table). Findings: Ss remembered the words encoded with the sentences best. Criticism: takes longer to perform the semantic task. Experiment 2: lengthened the physical task - is this a c-v-c-v-c-v word (RAVEL)? Still found memory performance best on the semantic task.

Self-reference effect - (Rogers, Kuiper, & Kirker, 1977) when asked to relate words to themselves, participants remember the words very well (we have self-schema rich with a networks of associations -Bellezza, 1992). Similarly, when we generate our own cues recall is enhanced (Greenwald & Banji, 1989)

 
  Criticisms:
1)asking Ss to do two different things with the encoding and recall, perhaps not asking the right question
2).Shallow processing may lead to durable learning

  Bransford and Franks (1977) type of recall task determines whether deeper processing improves memory. Asked Ss to perform a physical recall task after they performed a physical encoding task. (TAP)

Kolers(1976) transformed prose - people had better memory for information much later (Ss read inverted print, backward print or normal print. The best practice for reading an inverted sentence was having read the inverted sentence suggesting that Ss remembered the operations that they engaged in while reading)

   LT memories of any sort can be established as a result of primary memory processing. This is more likely to occur with deeper stimulus processing. Analyzing a stimulus at the semantic level is more likely to leave a LT memory.

 Assumption: Processing at deeper levels, not continuous processing at the same level enhances memory. Craik & Tulving (1975) attempted to constrain Ss to different levels of processing. They found no differences in LTM thus while semantic processing led to better LTM there is no evidence that different levels of shallow processing improved performance. Argues against the notion that maintenance rehearsal facilitates memory (Nelson (1977) - two encodings per word - does the word train have an r, does the word knight have an r, half of the Ss were then asked to read through the words again (in a new order)  - Ss who saw the words twice recalled them better. C & T reasoned that they remembered the words encoded twice better- the levels argument would be that Ss engaged in two separate encodings, one deeper than the other.
 (repetition priming)

 Making a semantic decision leads to a richer, more elaborate semantic code than is involved in simply registering the meaning of a word"

  Recent studies that give the depth assumption trouble:    A) maintanance rehearsal does not lead to learning (Nelson study,  Glenberg, Smith, and Green )

  repetition priming - repetition may cause a substantial increment in the ability to recall an item, repeated presentation may prime the representation of a word increasing its accessibility

 deeper processing leads to better retention?. There is evidence that superficial processing may be retained over long periods (e.g. visual characteristics - Kolers 76; vocal characteristics; initial letters -Nelson & Brooks, 73; mode of testing - Bransford & Franks 77

   information is processed through an ordered hierarchy of levels beginning with physical to phonemic to semantic level? Kleiman (75) studying reading found evidence that phonemic encoding may occur later than rather than prior to semantic processing. Semantic encoding may occur when Ss are unaware of orthographic or phonemic features

   Marcel; 1978 - pattern masking experiment - presented words tachistoscopically followed by a pattern. Ss sometimes guessed a word that was semantically similar but orthographically different from the stimulus (e.g. the word was red and they guessed blue).

  In experiment 2, after presenting a word, the Ss were asked ( at varying retention intervals) 1) did a word precede the mask; 2) did the word visually resemble (Hill, Hilt or Cigarette); 3) did the word semantically resemble (Hill, Mountain or Machine). Found: the 3rd condition was best under the shortest interval - just the opposite of the levels prediction)

  Dyslexic patients (Marshall & Newcombe, 66,73)    - semantically process words that they are able to pronounce but unable to read (pile read as heap)

   the problem of measuring depth? . Is it the time it takes to process the info? No. Refer back to the Bransford findings. Also, can produce tasks that are shallow that take a lot of time. Criak & Tulving (75) CVC study. No evidence for slower processing leading to better memory.

   The Compatibility effect Findings: questions that evoke a positive (yes) response are better retained than those that evoke a negative response.( Does dog rhyme with log ?- with button?) This is called the COMPATIBILITY EFFECT. Craik & Tulving conclude that positive instances "are related more closely to the question hence creating a more integrated and retrievable unit". They conclude that simple depth of processing is not as important as degree of elaboration. - a semantic encoding leads to a richer and more elaborate semantic code rather than simply registering the word's meaning. Under slow presentation conditions, when Ss were aware of subsequent recall test, a phonemic encoding led to poorer retention. The conditions were so slow that the Ss had to be aware of the meaning of the words but they failed to elaborate on the meaning. A semantic code leads to a richer, more elaborate memory code than simply registering the meaning of the word ( personal relevance)

Back to Handout Page