Psychology 342  Learning & Memory

Short-Term (Working) Memory

I.  Short-Term Memory 

       A.  Capacity of primary memory 

       George Miller (1956)—“The magic number seven, plus or minus two"

       Two main findings: 

       (1) Ss are unable to identify a set of items that vary along only one dimension if there are more than a few items

(2) Chunking 

  Increase capacity of short-term memory by grouping individual units of information into meaningful units.

II.  Duration of the Trace in Short-Term Memory 

Brown-Peterson Paradigm  

        A.  Purpose—determine rate at which information decays in STS

        B.  Method—experimenter reads a consonant trigram and then 3-digit number.  S had to count backward by 3 from the 3-digit number for a certain amount of time then had to recall the trigram. 

C.  Results-- rapid rate of forgetting after as little as 18 s    

Forgetting in short-term memory--decay or interference?

What is proactive interference (PI)?

III.  Atkinson & Shiffrin (1968)

        Distinguished between structural and processing components of memory.

        The structure is thought to be those parts of the memory system that do not change

  Three structures: (1) sensory registers, (2) STS, (3) LTS

     Information transfer from STS to LTS under Ss control (control processes)

Transfer begins and continues during the entire time an item is in STS because learning takes place even when the S is not trying to remember the material.   

Repetition of an item (even if unaware) leads to better performance

Two main control processes in STS—retrieval and rehearsal.

 Because of decay, retrieval process must be both very fast and highly efficient

 Atkinson and Shiffrin suggested the concept of a human memory system that keeps track of incoming information and exercises some control over its information-processing capacity

IV.  Immediate or Working Memory

 A.  Definition

Attkinson & Shiffrin’s model describes STS as a working memory 

A system that temporarily holds and manipulates information as part of a wide range of cognitive tasks (i.e., learning, reasoning, comprehending). 

Does STS really serve as a general, unitary working memory where storage and manipulation compete for space in a limited capacity STS?

       B.  Evidence

 Use dual-task technique--S is required to perform one task that absorbs most of the capacity of his/her working memory, while at the same time performing learning, reasoning and comprehension tasks.

Assumption: Performing a concurrent STM task should lead to a dramatic impairment in performance.

       Baddeley (1986)--Ss were required to remember number sequences ranging from zero to eight digits in length, while at the same time performing a reasoning test. 

     Effect of concurrent memory load on the speed and accuracy with which Ss performed the syntactic reasoning test:

Results: (1) reasoning time increases clearly and systematically with concurrent memory load (2) The effect, however, is not catastrophic

If STS is a unitary system, a concurrent load of eight digits should cause reasoning performance to break down completely.  It does not.

Earlier concept of a unitary STS is challenged and replaced by a multi-component working memory model. 

V.  Working Memory model (Baddeley, 1986)

A.  Revised definition of STS 

The limits of digit span may be set by one of a number of subsystems, leaving other components of working memory relatively unimpaired.

     Baddeley proposed a model of working memory in which a controlling attentional system supervises and coordinates a number of subsidiary slave systems. 

Three components—(1) central executive, (2) articulatory or phonological loop, and (3) visuo-spatial sketchpad

 .

B.  The Phonological Loop  

Phonological loop activity

1. What kind of substitution errors did you make?

2. Did you substitute an item that was acoustically similar to the correct item?

3. If you recorded a number in place of a letter, was it a letter acoustically similar to the number?

4. Did you confuse an M with N?

5. Were you fairly accurate in recalling W on the 2nd trial? Why might that be?

 Purpose: account for evidence regarding importance of speech coding in STM 

Phonological loop =  phonological store + articulatory control process

 Helps to explain the following phenomena:

(1) Phonological similarity effect-- memory is worse for items that sound alike than for items that differ. 

PGTCD is harder to recall from memory than the sequence RHXKW

Assumed to occur because the store is based on a phonological code, so similar items will have similar codes.  Recall will require discriminating among the memory traces.  Similar traces will be harder to discriminate, leading to a lower level of recall.

       (2) Articulatory suppression--Ss say a word, such as “the,” over and over out loud. 

         Assumption: Repeating “the” occupies the articulatory control process. Should be no covert rehearsal using the articulatory control process.  Without covert rehearsal, visual information cannot be translated into a phonological code and so cannot be placed into the phonological store. Should be no phonological similarity effect.       

Should still be a phonological similarity effect for auditory items.   

 Results support predictions

         (3) Unattended or irrelevant speech effect—unattended verbal material impairs recall

Colle and Welsh (1976) had two groups of Ss recall visually presented consonants, but one group saw the consonants while some irrelevant speech was played in the background. 

 Relative to the Ss in the quiet condition, Ss who heard irrelevant speech were not as successful in recalling the information. 

 Assumption: phonemes from the irrelevant speech enter into the phonological store and interfere with the information about the visually presented items. 

 **The unattended material (nonsense syllables disrupt performance to the same extent as words) was gaining access to the phonological store, a store that holds phonological but not semantic information.

What about unattended music? (Salame & Baddeley, 1989)

(4)  Word-length effect--short words (man, dog) are recalled better than long words (gentleman, canine). 

 Baddeley, Thomson, and Buchanan (1975)--word length and pronunciation time. 

 As the number of syllables in the words increased, the proportion of words that could be recalled decreased.

        Results--linear relationship between how long it takes to pronounce a word and the level of recall.

         Working Memory explanation--number of items that can be immediately recalled depends on how often each item can be subvocally rehearsed by the articulatory control process

The shorter the items, the more items can be rehearsed before a particular trace decays. 

       Immediate memory is where “cognitive work” is performed