Memory Models and Forgetting: From the Multi-Store Model to Retrieval Failure

For the multi-store model, know the three stores and their properties (coding, capacity, duration). Know the key evidence: Sperling (1960) for the sensory register; Miller (1956) and Peterson & Peterson (1959) for STM; Conrad (1964) for acoustic coding in STM; Baddeley (1966) and patient H.M. for LTM. Then know the two main challenges: patient K.F. (breaks the serial gateway assumption) and Craik & Lockhart's levels of processing (undermines maintenance rehearsal as the route to LTM).

For the working memory model, know all four components and what each does — but especially be able to name and explain the specific evidence for the phonological loop (word length effect, phonological similarity effect, irrelevant speech effect). Know that the central executive is the most important but also the least specified — this is both a strength and the main weakness.

For forgetting, focus on the distinction between storage failure and retrieval failure, then know the three theories: trace decay (Ebbinghaus, but challenged by Jenkins & Dallenbach sleep study), interference (proactive and retroactive — Underwood, 1957 is the key study), and retrieval failure (Tulving & Pearlstone, 1966, Godden & Baddeley, 1975). The encoding specificity principle ties it all together.

Multi-store model: sensory register · STM · LTM · coding (acoustic/semantic) · capacity · duration · serial position effect · primacy effect · recency effect · chunking · maintenance rehearsal

Working memory model: central executive · phonological loop · phonological store · articulatory rehearsal process · visuospatial sketchpad · episodic buffer · word length effect · phonological similarity effect · dual-task paradigm

Forgetting: trace decay · forgetting curve · proactive interference (PI) · retroactive interference (RI) · retrieval failure · availability vs accessibility · encoding specificity principle · context-dependent memory · state-dependent memory

Apply to examples: Use patient cases to illustrate model structure. H.M. supports the STM/LTM distinction. K.F. (intact LTM learning, impaired STM) undermines the idea that STM is a necessary gateway to LTM. For forgetting: misremembering a password after learning a similar one = proactive interference; struggling to remember something without going back to where you learned it = context-dependent retrieval failure.

Compare the models: The working memory model is more complex than the multi-store model, but that complexity is earned — dual-task performance, K.F.'s dissociation, and individual difference research all require multiple components. The key comparison is *active processing* (WMM) vs *passive flow* (multi-store). For forgetting, compare what each theory says about whether the information is *gone* or just *inaccessible* — retrieval failure argues it's mostly the latter.

Critique: Multi-store model: K.F.'s pattern breaks the serial flow; it doesn't explain why deep processing (not rehearsal) drives long-term retention. Working memory: the central executive is underspecified — hard to test independently. Trace decay: Jenkins & Dallenbach (1924) showed sleep reduces forgetting more than wakefulness — time itself is not the cause. Interference: tested mainly with word lists, which limits ecological validity.

Memory models have direct applications for learning and study. The encoding specificity principle predicts that studying in conditions similar to the exam context — same room, same time of day — may improve retrieval. Spacing practice exploits the shape of the forgetting curve and dramatically improves long-term retention. Working memory capacity strongly predicts academic outcomes; children with dyslexia and ADHD show specific phonological loop deficits that can be addressed with targeted support. For clinical psychology, understanding the distinction between availability and accessibility is central to treating intrusive memories and amnesia.

Getting the coding wrong: STM uses acoustic coding (Conrad, 1964); LTM uses semantic coding (Baddeley, 1966). Students often reverse these or say both use "visual" coding.

Forgetting what the working memory components actually do: listing the four components without explaining what each is for (and the evidence that separates them) is insufficient. The phonological loop evidence — word length effect, phonological similarity — is particularly important to know in detail.

Treating trace decay as the default theory of forgetting: the Jenkins & Dallenbach sleep study is strong evidence that the passage of time per se doesn't cause most forgetting. Interference and retrieval failure together give a better account. Presenting decay uncritically will lose marks.

Confusing proactive and retroactive interference: proactive = *old* learning interferes with *new* (P comes before R in the alphabet and in time). Retroactive = *new* learning interferes with *old*. Students frequently reverse them.

Ignoring the availability/accessibility distinction: when answering questions about forgetting, make clear whether you're arguing the memory trace is *lost* or merely *inaccessible without the right cue*. Most retrieval failure evidence supports the latter.