Models of memory

Seite 1

Multi-Store Model of Memory [MSM] (Atkinson & Shiffrin)

Inspired by computer science, sees brain as an ‘info processing unit’ (input -> processing -> output; like a computer)

Modular information processing model of encoding, storage, & retrieval of memory

i.e. Process of memory broken into distinct parts that interact with each other

3 stages (stores of memory): sensory memory [SM], short-term memory [STM], and long-term memory [LTM]

SM: Unlimited capacity, very short duration (<1s)

STM: Limited capacity (~7 ±2 items), short duration (<1min)

LTM: Unlimited capacity & duration

All modalities (types) of memory stored in the same store of memory (‘location’) regardless of modality

Memory stores interact via control processes:

Attention (sensory -> STM): SM info selected thru attention, transferred to STM for processing

Rehearsal (STM -> STM/LTM): STM info information repeated (rehearsed) to maintain in STM; transferred to LTM if rehearsed enough

Retrieval (LTM -> STM): when information is recalled from LTM to STM

MSM’s core assertions have been supported by a variety of research:

Glanzer & Cunitz (1966)’s findings support the concept of separate short-term & long-term memory stores, the short term store having limited duration

Peterson & Peterson (1959)’s findings suggest a distinct short-term memory store with duration of short-term memory was no greater than 18s (for remembering consonant triplets—language, aka echoic memory)

Evaluating the theory:

One of the first memory models proposed

Has predictive power

Core assertions are testable; in fact, research supports core assertions (separate stores of memory, STM having limited duration/capacity, etc.)

However, in many respects, MSM could be considered overly reductionist

Research (e.g. Quinn & McConnel (1996)) indicates presence of different, separate memory stores for diff types of memory, not just one unitary store of ‘short-term memory’

Assumes all memory is equally easy/difficult to recall & involves same mechanisms in same ways (may not be, e.g. FMs)

Doesn’t account for human ability to ‘chunk’ memory (e.g. considering “51” as a single ‘item’ rather than two, as some animals have demonstrated in exps)

Assumes memory is a passive process of storage & recall & that info in memory stores not impacted by other mental processes/representations (e.g. emotion, schemas, etc.)

Assumes info in LTM is ‘untouchable’ (no degradation/loss/modification); research indicates otherwise (that long-term memories degrade too over time, e.g. Talarico & Rubin (2003))

Assumes memory not affected by any mental representations, cognitive processes, etc. when research has proven they are (e.g. Loftus & Palmer (1974))

Doesn’t explain confabulated memory, i.e. memories believed to be true based on fabricated, distorted, or misinterpreted knowledge of events (e.g. Shaw & Porter (2015))

All in all, reflects understanding of memory in psychological research in mid-20th century, but psychology has moved on since then

Seite 2

Working Memory Model [WMM] (Baddeley & Hitch)

Proposed as a response to MSM; challenges assertion of unitary stores of memory.

Develops concept of ‘short-term memory’ to include diff modalities of storage (i.e. different types of info processed/stored differently) w/ concept of Working Memory [WM] consisting of several stores/mechanisms all interacting with each other:

Central Executive: ‘Manager’ of WM; responsible for controlling attention & resource allocation between all other components of WM

Visuospatial Sketchpad: Stores visual and spatial info

Phonological Loop: Stores auditory info; consists of—

Phonological Store (‘Inner ear’): Passively stores sound for short duration

Articulatory Rehearsal Component (‘Inner voice’): Repeats auditory info & can change modality of info (e.g. visual speech info to auditory info by subvocally ‘saying’ words to oneself as one read them)

Episodic Buffer: Integrates info from visuospatial sketchpad and phonological loop into episodic memories for storage in LTM (like a video editor); added in a later version of WMM (not in original)

All stores interact w/ long-term memory (similarly to MSM; not modeled explicitly in WMM)

Research supports assertions of WMM (in contrast to MSM):

Baddeley & Hitch (1976) found that performance on tasks decreased with presence of a difficult suppression task; R asserts this supports presence of CE

However, unclear if CE was present or if humans have limited info processing capability

Quinn & McConnel (1996) found that visual & auditory memory were separately influenced by modality-specific distractors, implying separate mechanisms for them

Evaluating the theory:

Re: STM, more comprehensive than MSM; research supports some of its elements/assertions (e.g. modality-specific memory stores)

However, research on other elements remains unclear (e.g. CE); in fact, existence & nature of CE as a component has been debated extensively (some argue that it’s an overly vague/untestable concept)

However, only models STM & not SM/LTM; not comprehensive in that regard

Somewhat predictive, but unclear/vague; exact function of each component (esp CE) in given situations still unclear, model has been criticized for emphasizing structure over processing

Still doesn’t/only vaguely accounts for various aspects of memory, e.g. confabulated

memory (might be CE overloaded therefore mistaken?), long-term memory decay, different types of memory (e.g. FMs) being easier to recall, etc.