The Multi-Store Model

Atkinson and Shiffrin (1968) developed the concept of the multi-store model to describe how memory is organised.

Strengths:

• Clive Wearing - separate systems for STM and LTM.
• Murdock's serial positioning curve - separate STM and LTM systems - words at the beginning (LTM) and at the end (STM) can be remembered, but the middle words were forgotten (displacement). 

Weaknesses: 

• Baddely and Hitch - STM in the Multi-Store Model is too simple - Working Memory Model.
• Levels of Processing Approach dissagrees with the idea that the amount of rehersal determins what stays in the memory. Levels of Processing Approach suggests that it is how deeply the information is processed.

Encoding - information in the sensory memory is relatively unprocessed, and is encoded in the same form that it entered the body, from the senses.

Capacity - Sensory memory has a large capacity. George Sperling (1960) found that information in the sensory memory disappears rapidly.

Duration - This depends on the encoding. Visual may remain in the sensory memory for 0.5 seconds, but sounds may stay in the sensory memory for 4 seconds.

Encoding - primarily acoustic, but can be stored in different ways.                                        Baddely's study involved four different conditions: acoustically similar (55% accuracy), acoustically dissimilar (75% accuracy), semantically similar and semantically dissimilar (slight detrimental effect). Participants were asked to recall five words in the correct order immediately after hearing the words. Encoding of STM is primarily acoustic.

Capacity - limited capacity.                                                                                              Baddely measured the reading speed of participants. Then they were presented five words taken from one of two sets: one-syllable/polysyllabic and asked to write the words in serial order immediately after. Participants could recall as many words as they were able to articulate in 2 seconds. Immediate memory span is 2 seconds articulation.

Duration - 15-30 seconds. Can be lengthened with rehearsal.                                               Peterson and Peterson showed participants a consonant trigram. Interference task (counting backwards in threes) lasted for different intervals, then the participants were asked to recall the trigram. After 3 seconds, 80% accuracy, after 18 seconds, >10% accuracy. STM information decay worsens over time.

Memory is lost from STM through displacement.

Encoding - predominantly semantic and visual.                                                                  Baddely's study involved four different conditions: acoustically similar, acoustically dissimilar (same results), semantically similar (55% accuracy)and semantically dissimilar (85% accuracy). Participants were asked to recall ten words in the correct order after 20 minutes (they were given another task). Encoding of LTM is primarily semantic.

Capacity - No upper limit. Because of this, LTM has to be organised in subgroups:        Episodic: memory for personal information and past experiences.                                      Semantic: memory for facts and knowledge.                                                                      Procedural: memory for how to do things.

Duration - Can be for a whole lifetime.                                                                               Bahrick et al. showed 392 graduates photographs from their yearbooks. Recognition group (with a list of names to chose from) could remember 90% after 14 years, 80% after 25 years, 75% after 34 years and 60% after 47 years. Recall groups (without cues) could remember 60% after 7 years and <20% after 47 years. This shows that some information is retained for a lifetime. VLTM is better with cues.

Memory is lost from LTM through trace decay.