Psychology › Section H · Memory › Topic H.1
How memory functions
Everything you learn depends on memory, yet memory is not one thing. This topic breaks it into three processes and three stores, then sorts long-term memory into the kinds you can put into words and the kinds that show only in what you do.
Three jobs: encode, store, retrieve
Memory is a set of processes that keep information available over time. Encoding turns incoming information into a form the brain can keep. Storage holds it, for seconds or for decades. Retrieval brings it back into awareness when needed. A failure at any step feels the same: "I can't remember." Forgetting a new classmate's name a minute after hearing it usually means it was never encoded, because you were planning what to say. The computer comparison (input, saving, opening a file) is a useful start, but later topics show that human memory rebuilds rather than replays.
Encoding: effort, form and depth
Some encoding is automatic: without trying, you pick up where you sat in class or how often a word keeps appearing. Other encoding is effortful and needs attention, like learning a formula. Information can be encoded visually, acoustically (why jingles stick) or semantically, by meaning. Fergus Craik and Endel Tulving (1975) asked people one question about each word: about its print (capital letters?), its sound (does it rhyme with a given word?) or its meaning (does it fit a given sentence?). In a surprise test, words judged for meaning were remembered best and words judged on print worst. Meaning linked to yourself is often deeper still, the self-reference effect (Rogers, Kuiper and Kirker, 1977).
The Atkinson-Shiffrin model
Richard Atkinson and Richard Shiffrin (1968) proposed that information passes through three stores. Sensory memory holds a brief, detailed copy of what the senses take in: a visual trace (iconic memory) lasts well under a second, and a sound trace (echoic memory) a few seconds. Attention selects a small part of this flood for short-term memory; the rest is lost almost at once. Information that is rehearsed or encoded well enters long-term memory. The model simplifies, since the stores trade information in both directions, but it remains the standard map for describing where a memory failed.
Short-term memory: brief and small
Without rehearsal, short-term memory (STM) keeps information for only about 15 to 30 seconds. Lloyd and Margaret Peterson (1959) gave people three-letter strings, then had them count backwards to stop rehearsal; recall fell steeply within 18 seconds. Later work showed that interference from earlier strings, not decay alone, explained much of the loss. George Miller (1956) famously put capacity at about seven items, plus or minus two. More recent estimates, notably by Nelson Cowan, place the limit nearer four meaningful chunks when rehearsal and grouping are prevented. Repeating information, called maintenance rehearsal, keeps it active in STM.
Working memory: a mental workspace
Alan Baddeley and Graham Hitch (1974) argued that short-term memory is a workspace for thinking, not a waiting room. Their working memory model has a phonological loop for speech sounds (the inner voice repeating a number), a visuospatial sketchpad for images and locations, and a central executive that directs attention. Baddeley added an episodic buffer in 2000, which binds information from the other parts and from long-term memory into single episodes. The evidence: two verbal tasks done at once interfere badly, while a verbal task paired with a visual one interferes much less, suggesting separate subsystems.
Long-term memory as a network
Long-term memory (LTM) has no known capacity limit and holds information for minutes or a lifetime. It is organised largely by meaning. In the semantic network model of Allan Collins and Elizabeth Loftus (1975), concepts are nodes joined by links that strengthen with use. Thinking of one concept partly activates its neighbours, called spreading activation, so reading "doctor" makes "nurse" quicker to recognise. Richly connected knowledge has many routes back to it. Making a new memory stable is called consolidation; the next topic shows which brain areas do it.
Explicit and implicit memory
Explicit, or declarative, memory is what you can consciously recall and put into words. Endel Tulving (1972) split it into episodic memory, for personal events tied to a time and place, such as your first day at a new school, and semantic memory, for facts and concepts, such as the boiling point of water. Implicit, or non-declarative, memory shows itself in performance. It includes procedural memory for skills such as cycling; priming, where an earlier exposure changes a later response without your awareness; and classically conditioned emotional reactions, such as unease at a smell you cannot explain.
Retrieval: recall, recognition, relearning
Recall means producing information with few or no cues, as in an essay question. Recognition means identifying it when you meet it again, as in a multiple-choice test; it is usually easier than recall. Relearning means learning material faster the second time. Hermann Ebbinghaus measured this as savings: if a list took 20 minutes to learn and 5 minutes to relearn a month later, 15 of the 20 minutes were saved, a 75% saving. Something remained even if not one item could be recalled, which makes relearning the most sensitive of the three measures.
Worked example: a code through the model
Sam's neighbour reads out a garage door code, 19690720, while Sam is carrying shopping. 1. Sensory memory: the spoken digits leave an echoic trace lasting a few seconds. 2. Attention: Sam is listening, so the digits pass into short-term memory. 3. Capacity: eight separate digits exceed about four chunks, so they are at risk. 4. Encoding: Sam notices they form a date, 20 July 1969, the first Moon landing. Eight items become one meaningful chunk, encoded semantically. 5. Duration: a chat about the weather follows, but one linked chunk survives it. 6. Check: next week Sam recalls the code with no cue, so encoding, storage and retrieval all worked.
Worked example: which memory system?
First ask: can the person consciously report it? 1. Describing a cousin's wedding: reportable and tied to a time and place, so explicit episodic memory. 2. Knowing that Canberra is Australia's capital, without remembering when you learned it: a reportable fact, so explicit semantic memory. 3. Touch-typing fast while unable to say which finger presses q: shown in performance, so implicit procedural memory. 4. After an advert shows the word "yellow", completing BAN___ as "banana": an earlier exposure shaping a response, so implicit priming. 5. Check: every explicit case can be put into words; every implicit case shows only in behaviour.
Common mistakes to avoid
First, calling forgetting yesterday's events "short-term memory loss": short-term memory lasts seconds, so yesterday is long-term memory. Second, treating seven as a fixed law: four chunks is the better modern guide. Third, confusing semantic encoding, a way of getting information in, with semantic memory, a type of stored knowledge. Fourth, assuming that failing to recall means nothing is stored: recognition and relearning often reveal it. Fifth, reading the three-store model as three physical boxes in the head: it describes processes, not places.
Key takeaways
Memory means encoding, storage and retrieval. Encoding may be automatic or effortful; meaning gives the deepest processing (Craik and Tulving, 1975). In the Atkinson-Shiffrin model (1968), sensory memory lasts seconds at most, STM about 15 to 30 seconds, and LTM has no known limit. STM holds about seven items (Miller, 1956), or nearer four chunks by recent estimates. Working memory (Baddeley and Hitch, 1974) has a phonological loop, visuospatial sketchpad, central executive and episodic buffer. LTM is a semantic network. Explicit memory is episodic or semantic; implicit memory includes procedural memory, priming and conditioned responses. Retrieval shows as recall, recognition or relearning.
Flashcards
Flashcard
Getting information into memory in a form the brain can keep
Test your understanding
Question 1 of 5
In a study like Craik and Tulving's (1975), which question asked about each word should lead to the best memory for the words in a surprise test?
Question 2 of 5
A list took 30 minutes to learn. A month later it took 12 minutes to relearn. What is the savings score?
Question 3 of 5
Jordan touch-types quickly but cannot say which finger presses the letter q. This skill is an example of:
Question 4 of 5
You follow spoken directions while picturing the route on a mental map. In the working memory model, which part holds the mental map?
Question 5 of 5
When rehearsal and grouping are prevented, about how much can short-term memory hold, according to more recent estimates such as Cowan's?
