2.1.1. Models of Memory


Key Definitions

Articulatory suppression: a research technique where participants are required to repeat a sequence of sounds while at the same time performing the experimental task

Capacity: the number of units of information that can be held in a memory store

Dual task technique: a research technique where participants are exposed simultaneously to two sets of stimuli, either of the same or different modalities

Duration: the amount of time for which information can be held in the memory store

Heuristic model (theory): a model or theory that inspires new hypotheses in a certain area

Memory: a cognitive process of encoding, storing, and retrieving information

Parsimonious model: a model that can explain a lot of observations with only a limited number of components

Phonological similarity effect: a memory phenomenon where stimuli that have a similar pattern of articulation are more likely to be confused in memory, even when they are presented visually

Primacy effect: a memory phenomenon where the first several words on the list are remembered better than words from they are presented visually

Recency effect: a memory phenomenon where the last several words on the list are remembered better than words from the middle of the list

Word length effect: a memory phenomenon where the estimated capacity of short-term memory depends on the length of the words on the list presented to be participants


Essential Understanding

  • Multi-store Memory Model (MSM) Theory

→ memory is a basic cognitive process used to encode, store, and retrieve information

→ a well-known classic memory model

↳states that:

↳ memory consists of 3 separate components: sensory memory store, short-memory store (STM), and long-memory store

↳ information flows from sensory memory to long-term memory if certain conditions are met: attention and rote rehearsal

  • →Research

    → the model is mostly simple which requires testing like the fact that the stores are separate or the idea that rehearsal is necessary and sufficient for the information from STM to LTM

    ↳ no single study can test the model in its entirety→ a collective effort by researchers is needed for the task

Ex: Glanzer and Cunitz (1966) supported that STM and LTM are two separate memory stores

↳ based on the disappearance of the latency effect (but not the primacy effect) after an interference task

  • →Evaluation

    → model was criticized for reasons for oversimplification and failing to explain some memory phenomena

    → it needs to be understood that there is a trade-off between simplicity of the model and its explanatory power

  • Working Memory Model Theory

    → proposed by Baddeley and Hitch (1974)

    → made to further elaborate the structure of STM

    → includes four components: visuospatial sketchpad, phonological loop, central executive, and episodic buffer

  • Research

    → explained some findings like

    ↳ phonological similarity effect

    ↳ word length effect

    → these effects could be explained by the introduction of the articulatory rehearsal component

    → further demonstrated that both effects disappear when articulatory suppression is used and material is presented visually


    Atkinson and Shiffrin (1968) Multi-store of Memory

↳ key: multi-store model presents the structure of memory as consisting of 3 separate stores and the process as a unidirectional flow of information from one store to another

  • Components

    → sensory memory store

    → short-term memory store (STM)

    → long-term memory store (LTM)

  • Characteristics

    → each one is described by:

    ↳ duration

    ↳ capacity

    ↳ conditions necessary for information to move from a store to another

  • Sensory Memory

Duration: depends on very short modality

↳ 1 seconds for visual stimuli

↳ 2-5 seconds for auditory stimuli

Capacity:

↳ everything in the perceptual field

Conditions required to move to the next store:

↳ attention (it doesn’t move to the next store without attending to a unit of information)

  • Short-term Memory (STM)

Duration: depends on the modularity but usually no longer than 30 seconds

↳ rehearsal increases STM duration

Capacity: 7 (plus or minus) 2 chunks (from Miller, 1956)

↳ a chunk is a combination of units

Conditions required to move to the next store:

↳ rehearsal as it gradually moves to long-term memory

  • Long-term Memory (LTM)

Duration: limit not established but potentially longer than human life

Capacity: limit has not been established but could be unlimited

Conditions required to move to the next store:

↳ N/A


Support for the Multi-Store Model of Memory

→ there are many aspects that require testing for this simple model

Some aspects that need testing include:
↳ number of memory stores

↳ memory stores being separate from each other

↳ rehearsal as a requirement for moving from STM to LTM

↳ direction of information flow: model suggests that it’s only a unidirectional process

→ Glanzer and Cunitz partially supports this model


Research for MSM Model

  • Glanzer and Cunitz (1966, Experiment 2)→ Primacy and Recency Effect

↳ key: the fact that inclusion of a filler task results in disappearance of recency effect but not primacy effect supports the idea that STM and LTM are separate memory stores

Aim: investigate the serial position effect with and without the inference from a filler activity

Serial position effect: tendency to recall the first and last items on a list better than the middle items

Method: experiment; repeated measures design

Participants: 46 army-enlisted men

Procedure:

→ a series of 15-word lists was read to participants

↳ afterwards, participants were required to do a free-recall task

→ 3 conditions:

↳ a free-recall task after immediately hearing the words on list

↳ a filler activity (i.e. counting backwards for 10 seconds), then free recall

↳ same filler activity for 30 seconds

→ Each participant was given 15 lists, 5 for each of the 3 conditions (conditions were randomized)

→ Dependent variable was the proportion of words correctly recalled, separately for each of the 15 positions of the word on the list

Results:

→ In condition without filler task, both aspects of the serial position effect could be observed: primacy and recency effect

→ In condition with filler task, the primacy effect stayed, but not the recency effect

↳ more prevalent in 30-sec condition than 10-sec condition

Conclusion:

↳ results fit well into the MSM theory, especially the following ideas:

↳ STM and LTM are two separate memory stores

↳ Information moves form STM to LTM if it is rehearsed, gradually decays if not

↳ Duration of STM is about 30 seconds

→ as participants recall word immediately after the list presentation, the last words may be easy to recall since they have just entered STM

↳ with a task interference, these traces decay and recency effect disappears


Evaluation of the MSM Theory

→ model is parsimonious (can explain a lot of observed data with few components) and heuristic (inspired numerous research studies)

→ has been criticized for simplicity and inability to explain some memory phenomena

  • Criticism

    ↳ model emphasizes structure over process: doesn’t pay enough attention to how information flows between the three components

    ↳rote rehearsal is the only mechanism that enables the transfer of info from STM to LTM→ Shown in further research that people also use semantic encoding

    ↳ model only explains the flow of information in one direction; arguably, though, information can also flow in the opposite direction

    ↳ has been argued that both STM and LTM are not unitary stores and should be further subdivided

  • Strengths

    ↳ one of the most influential models of memory that explains multiple observed phenomena


Baddeley and Hitch (1974)→ Working Memory Model

↳key: model is a zoom-in on the structure of STM with four components

  • Background

→ researchers felt that this was needed because some findings were not consistent with the MSM where STM is a unitary store

  • The Model→ working memory in this model consists of four components with different functions

    visuospatial sketchpad (inner eye): hold visual and spatial information

    phonological loop: hold auditory information; divided into two parts

    inner ear: holds sound in a passive manner

    articulatory rehearsal component (inner voice): rewords to increase their duration in memory; can turn visual speech information into sounds, changing the modality of perceived speech material

    central executive: responsible for the allocation of resources between visuospatial sketchpad and phonological loop

    episodic buffer: (added later) integrates information from the visuospatial sketchpad and phonological loop and links it to LTM


    Research for Working Memory Model

→ this model provided an explanation for some phenomena that MSM could not explain like the phonological similarity effect, word length effect, and effects of articulatory suppression

  • Phonological Similarity Effect

    ↳ first discovered by Conrad and Hull (1964): showed that lists of phonologically similar letters were more difficult than lists of letters that don’t have similar sounds

    ↳ this is presumably because acoustically encoded traces of rhyming letters are easier to confuse with each other

    ↳ this effects also appeared when stimulus was present in writing

  • Word Length Effect

    ↳ first discovered by Baddeley, Thompson, and Buchanan (1975)

    ↳ showed that capacity of STM is greater for short words than longer ones

    → there was nothing in the MSM to explain why STM capacity should change from the standard 7(plus or minus) 2 units depending on the word length

  • Explanations of the Findings

    ↳ introducing the articulatory rehearsal component (inner voice) in memory explained these effects

    → Phonological similarity effect: assuming that all speech material (and in writing) is sub vocally pronounced and encoded as an articulation pattern

    ↳ similar articulation patterns are easier to become confused with

    → Word length effect: articulation patterns of longer words are also longer

    ↳ in a given amount of time you can sub vocally pronounce fewer long words then short words

→ Further testing was also done by suppressing the inner voice

↳ if the explanation is correct, suppressing the inner voice should result in disappearance of both the phonological similarity effect and word length effect

↳ was observed in experiments that used articulatory suppression

  • Further Research: Effects of Articulatory Suppression

→ Articulatory suppression: technique that requires participants to repeat a sequence of sounds while at the same time performing the experimental task

↳ used to block the “inner voice” to see how human memory performs without it

→ Shown that the phonological similarity effect disappears under articulatory suppression when material is presented visually (Murray 1968; Baddeley, Lewis, Vallar 1984)

↳ In terms of the Working Memory Model, information cannot enter the phonological loop so it then goes to the visuospatial sketchpad, storing it visually

→ Also shown that in articulatory suppression, word length effect disappears with visual presentation (Baddeley, Thomson, Buchanan 1975)

↳ by this logic, information cannot enter the phonological loop through inner voice, so it also enters the visuospatial sketchpad instead, no matter the word length

Key: these evidence supports the idea that visual and auditory information is processed in separate stores within working memory


Evaluation of the Working Memory Model

↳key: working memory model has large explanatory power, but also has complexity issues

  • Strengths

    → explanatory power

    → explains phenomena that the MSM model can’t explain including phonological similarity effect, word length effect, and their disappearance during articulatory suppression

  • Limitations

    → its complexity makes it difficult to be empirically tested

    → does not include in LTM or sensory memory