Sperling's partial-report technique showed that:

Iconic memory holds more information than people can report before it fades

Short-term memory holds seven plus or minus two chunks

Forgetting is caused by proactive interference

Iconic memory holds more information than people can report before it fades

Long-term memory is coded semantically

Divers recalled words better in the same environment — under water or on land — in which they had learned them. This finding supports:

Context-dependent memory and the encoding-specificity principle

The levels-of-processing framework

Jenkins and Dallenbach (1924) found that participants forgot less over eight hours of sleep than over eight hours awake. The result is usually taken to favour:

Interference theory over decay theory

Decay theory over interference theory

The multi-store model's sensory register

Assertion (A): Introducing a 30-second filled delay before recall eliminates the recency effect but leaves the primacy effect intact. Reason (R): Recency items are held in short-term memory, which the delay-filling task displaces, whereas primacy items have been rehearsed into long-term memory. Choose the correct option.

Both A and R are true, and R is the correct explanation of A

Both A and R are true, but R is not the correct explanation of A

H. M., after removal of his medial temporal lobes, showed which of the following? Select all that apply.

Normal short-term (digit-span) memory

Improvement in mirror drawing across sessions

Inability to form new declarative memories

Normal short-term (digit-span) memory

Improvement in mirror drawing across sessions

Inability to form new declarative memories

Loss of all procedural skills learned before surgery

According to Craik and Lockhart's levels-of-processing framework, which task should produce the BEST later recall of a word?

Deciding whether the word fits the sentence "She cooked the ___"

Deciding whether the word is printed in capital letters

Deciding whether the word rhymes with "cake"

Counting the letters in the word

Verbal Learning, Memory and Forgetting: Ebbinghaus and the Verbal-Learning Tradition; Encoding, Storage and Retrieval; Sensory, Short-Term (Working) and Long-Term Memory; and the Theories of Forgetting

This chapter takes the "Verbal learning" named in Unit 5's learning block together with its final block: "Memory and Forgetting. Memory processes: Encoding, Storage, Retrieval. Stages of memory: Sensory memory, Short-term memory (Working memory), Long-term Memory (Declarative – Episodic and Semantic; Procedural). Theories of Forgetting: Interference, Retrieval Failure, Decay, Motivated forgetting." The verbal-learning tradition founded by Ebbinghaus supplied the methods — nonsense syllables, savings, serial and paired-associate lists — on which memory research was built, so the two belong together. Savings scores and serial-position curves are worked with numbers, and each theory of forgetting is tied to the experiment that tests it.

1. Verbal learning

Hermann Ebbinghaus (Über das Gedächtnis, 1885), using himself as subject, learned lists of nonsense syllables (consonant–vowel–consonant trigrams) to eliminate prior meaning, and measured retention by the savings method — how much less effort relearning took. His forgetting curve showed rapid loss in the first hour and slower loss thereafter. Savings = (original trials − relearning trials) ÷ original trials × 100: a list first learned in 20 trials and relearned in 8 shows (20 − 8) ÷ 20 × 100 = 60 per cent savings.

The verbal-learning tradition
Method or finding	What it is
Serial learning	Items learned in order; produces the serial-position curve — better recall of early items (primacy, owing to more rehearsal and transfer to long-term memory) and of late items (recency, still in short-term memory). A 30-second filled delay removes recency but not primacy (Glanzer and Cunitz, 1966).
Paired-associate learning	Learning stimulus–response pairs (house–tree); the basis of interference research and of mnemonic mediation.
Free recall and clustering	Items recalled in any order tend to be grouped by category even when presented randomly (Bousfield) — evidence of organisation.
Meaningfulness	Glaze's association value and Noble's m: items that evoke more associations are learned faster.
Von Restorff (isolation) effect	A distinctive item in a list is remembered better.
Conditions of practice	Distributed practice beats massed practice (the spacing effect); overlearning slows forgetting; active recitation beats rereading; whole versus part learning depends on the material's length and meaningfulness.

Retention can be measured by recall (free, serial or cued), recognition (picking the correct item; usually the most sensitive), relearning (savings) and reconstruction (restoring the order of items). The same person may fail to recall something they can easily recognise, which is why these measures give different estimates of how much has been forgotten.

2. Memory processes: encoding, storage and retrieval

Encoding transforms information into a storable form — acoustic, visual or semantic. Craik and Lockhart's levels-of-processing framework (1972) held that deeper, meaning-based (semantic) processing produces more durable memories than shallow (structural or phonemic) processing: judging whether a word fits a sentence yields better recall than judging whether it is in capital letters. Storage is the maintenance of information over time, and consolidation stabilises new memories, which is why a blow to the head can erase the preceding minutes (retrograde amnesia). Retrieval brings stored information back into awareness, and depends on cues.

Encoding specificity (Tulving and Thomson, 1973): a cue helps retrieval to the extent that it was encoded with the item.
Context-dependent memory: divers who learned words under water recalled them better under water, and those who learned on land better on land (Godden and Baddeley, 1975).
State-dependent memory: recall is better in the same internal state — mood or drug state — as at learning.
The tip-of-the-tongue state (Brown and McNeill, 1966): people can often report a word's first letter and number of syllables while unable to retrieve it — showing partial access.
Reconstruction: Bartlett's "War of the Ghosts" (1932) showed that recall is shaped by schemas; Loftus and Palmer (1974) showed that the verb in a question ("smashed" versus "hit") changed estimates of speed and led to false memories of broken glass.

3. Stages of memory: sensory, short-term (working) and long-term

Atkinson and Shiffrin's multi-store model (1968)
Store	Capacity	Duration	Coding and key evidence
Sensory — iconic	Large	About a quarter to half a second	Sperling (1960): with the partial-report technique people could report any cued row of a 12-letter display, though whole report gave only about 4 letters
Sensory — echoic	Large	About 2–4 seconds	Auditory partial-report studies (Darwin, Turvey and Crowder, 1972)
Short-term	7 ± 2 items (Miller, 1956), perhaps nearer 4 ± 1 (Cowan); expanded by chunking	About 18–30 seconds without rehearsal (Brown–Peterson task, 1958–59)	Mainly acoustic — acoustically similar letters are confused (Conrad, 1964)
Long-term	Effectively unlimited	Minutes to a lifetime	Mainly semantic — errors involve similar meanings (Baddeley, 1966)

Baddeley and Hitch (1974) replaced the single short-term store with working memory, an active workspace with a central executive (an attentional controller) and two slave systems: the phonological loop (a phonological store with an articulatory rehearsal process — hence the word-length effect, and the disruption caused by articulatory suppression) and the visuospatial sketchpad. Baddeley added the episodic buffer in 2000 to integrate information and link it to long-term memory. Long-term memory divides into declarative (explicit) memory — episodic (personal events located in time and place) and semantic (general knowledge), a distinction drawn by Tulving (1972) — and non-declarative (implicit) memory, including procedural skills, priming and classical conditioning (Squire).

📖 H. M. and K. F.

After bilateral removal of his medial temporal lobes in 1953, H. M. could no longer form new declarative memories (anterograde amnesia), yet his short-term memory was normal and he improved at mirror drawing without remembering having practised. K. F., studied by Shallice and Warrington, had the reverse pattern: a severely impaired short-term memory with normal long-term learning. Together they form a double dissociation supporting separate stores.

4. Theories of forgetting

The four theories the syllabus names
Theory	Claim	Key evidence
Decay	Memory traces fade with time and disuse (Thorndike's law of disuse).	Hard to test because time is always filled with other events. Jenkins and Dallenbach (1924) found less forgetting after sleep than after equal time awake, favouring interference; decay may apply to sensory and unrehearsed short-term memory.
Interference	Other learning disrupts memory. Proactive interference: earlier learning disrupts later learning. Retroactive interference: later learning disrupts earlier learning.	Greatest when materials are similar (McGeoch). Underwood (1957) showed that most forgetting of a list over 24 hours came from proactive interference from previously learned lists; Wickens showed release from PI when the category changes.
Retrieval failure (cue-dependent forgetting)	Information is available but not accessible without the right cue (Tulving).	Cued recall recovers items lost in free recall; context- and state-dependent effects; the tip-of-the-tongue state.
Motivated forgetting	We forget what is threatening or unwanted — Freud's repression (unconscious) and conscious suppression.	Laboratory evidence comes from directed-forgetting studies and Anderson and Green's (2001) "think/no-think" paradigm, in which repeatedly suppressing a response impaired its later recall; evidence for repression of traumatic memories remains contested.

🧠 Pro = forward, retro = backward

Proactive interference works forward in time (old → new): after years of writing last year's date, you write it in January. Retroactive works backward (new → old): learning a new phone number makes the old one hard to recall. In the classic design, Group E learns A, then B, and is tested on A — that tests retroactive interference.

Key takeaways

Ebbinghaus used nonsense syllables and the savings method (savings = (OL − RL) ÷ OL × 100) and found a steep early forgetting curve; serial learning shows primacy and recency.
Deeper, semantic encoding lasts longer (levels of processing); retrieval depends on cues matching the encoding context (encoding specificity; context- and state-dependent memory); recall is reconstructive (Bartlett; Loftus).
Iconic memory lasts under a second (Sperling), echoic a few seconds, short-term memory holds 7 ± 2 chunks for about 20 seconds, and long-term memory is unlimited and mainly semantic.
Working memory comprises a central executive, phonological loop, visuospatial sketchpad and episodic buffer; long-term memory is declarative (episodic, semantic) or non-declarative (procedural, priming).
Forgetting is explained by decay, proactive and retroactive interference, retrieval failure (available but not accessible) and motivated forgetting.

Practice questions (10)

Attempt each one before opening the answer. Every explanation names the tempting wrong option as well as the right one, because that is where marks are lost.

A list originally took 20 trials to learn and, a week later, 8 trials to relearn. What is the savings score, in per cent? Type a number.
Numerical answer — type the value.
Show answer
Answer: 60
Savings = (original − relearning) ÷ original × 100 = (20 − 8) ÷ 20 × 100 = 12 ÷ 20 × 100 = 60 per cent.
Sperling's partial-report technique showed that:
1. Short-term memory holds seven plus or minus two chunks
2. Forgetting is caused by proactive interference
3. Iconic memory holds more information than people can report before it fades
4. Long-term memory is coded semantically
Show answer
Answer: C — Iconic memory holds more information than people can report before it fades
When cued to report any one row after the display disappeared, people could report most of it, so nearly the whole display was briefly available; whole report gave only about four letters because the icon faded while reporting. The 7 ± 2 capacity is Miller's finding.
Learning a new computer password makes it hard to remember your old one. This is an example of:
1. Repression
2. Retroactive interference
3. Trace decay
4. Proactive interference
Show answer
Answer: B — Retroactive interference
New learning working backward to disrupt older memory is retroactive interference. Proactive interference is the opposite — the old password intruding when you try to recall the new one.
Which of the following are components of Baddeley's working-memory model? Select all that apply.
1. Iconic store
2. Central executive
3. Phonological loop
4. Visuospatial sketchpad
Show answer
Answer: B — Central executive; C — Phonological loop; D — Visuospatial sketchpad
Working memory comprises the central executive, the phonological loop, the visuospatial sketchpad and (since 2000) the episodic buffer. The iconic store is a sensory register in the Atkinson–Shiffrin model.
Remembering what you had for breakfast this morning draws mainly on:
1. Semantic memory
2. Echoic memory
3. Procedural memory
4. Episodic memory
Show answer
Answer: D — Episodic memory
Episodic memory stores personally experienced events located in time and place. Semantic memory holds general facts, procedural memory skills, and echoic memory holds sounds for a few seconds.
Divers recalled words better in the same environment — under water or on land — in which they had learned them. This finding supports:
1. Motivated forgetting
2. Trace-decay theory
3. Context-dependent memory and the encoding-specificity principle
4. The levels-of-processing framework
Show answer
Answer: C — Context-dependent memory and the encoding-specificity principle
Godden and Baddeley (1975) showed that reinstating the learning context provides retrieval cues, as encoding specificity predicts. Levels of processing concerns depth of encoding, not the match between learning and test contexts.
Jenkins and Dallenbach (1924) found that participants forgot less over eight hours of sleep than over eight hours awake. The result is usually taken to favour:
1. Interference theory over decay theory
2. Repression
3. Decay theory over interference theory
4. The multi-store model's sensory register
Show answer
Answer: A — Interference theory over decay theory
If forgetting were simple decay, equal time would produce equal loss. Less forgetting during sleep, when fewer new events occur, points to interference from waking activity.
Assertion (A): Introducing a 30-second filled delay before recall eliminates the recency effect but leaves the primacy effect intact. Reason (R): Recency items are held in short-term memory, which the delay-filling task displaces, whereas primacy items have been rehearsed into long-term memory. Choose the correct option.
1. Both A and R are true, and R is the correct explanation of A
2. Both A and R are true, but R is not the correct explanation of A
3. A is true, but R is false
4. A is false, but R is true
Show answer
Answer: A — Both A and R are true, and R is the correct explanation of A
Both are true and R explains A: Glanzer and Cunitz's dissociation is the standard evidence that the two ends of the serial-position curve reflect two different stores.
H. M., after removal of his medial temporal lobes, showed which of the following? Select all that apply.
1. Normal short-term (digit-span) memory
2. Improvement in mirror drawing across sessions
3. Inability to form new declarative memories
4. Loss of all procedural skills learned before surgery
Show answer
Answer: A — Normal short-term (digit-span) memory; B — Improvement in mirror drawing across sessions; C — Inability to form new declarative memories
H. M. had severe anterograde amnesia for declarative material but normal short-term memory and intact procedural learning, which improved without conscious recollection. His old procedural skills were preserved, not lost.
According to Craik and Lockhart's levels-of-processing framework, which task should produce the BEST later recall of a word?
1. Deciding whether the word fits the sentence "She cooked the ___"
2. Deciding whether the word is printed in capital letters
3. Deciding whether the word rhymes with "cake"
4. Counting the letters in the word
Show answer
Answer: A — Deciding whether the word fits the sentence "She cooked the ___"
Judging sentence fit requires semantic (deep) processing, which yields the most durable memory. Case and letter-counting judgements are structural, and rhyme judgements phonemic — both shallower.