WME cheat sheet

Phonological loop (Baddeley): temporary storage of speech-based information, roughly 2 seconds of decaying trace refreshed by subvocal rehearsal.

Digit span: forward span averages about 7±2 items in adults; backward span is typically 1-2 items shorter because it adds a manipulation component.

Word length effect: recall drops as spoken duration of items increases; lists of short words outperform lists of long words of equal count.

Phonological similarity effect: acoustically similar items (cat, hat, mat) are harder to recall than dissimilar items, implicating speech coding.

Articulatory suppression (repeating 'the, the, the') blocks rehearsal and removes both the word length and phonological similarity effects.

Irrelevant sound effect: background speech-like sound impairs serial recall of visually presented lists, but steady noise or music without changing speech tokens does not.

Serial position curve for auditory lists: primacy reflects long-term transfer, recency reflects the short-lived echoic/loop store; a filled delay abolishes recency.

Echoic memory: high-capacity, brief (about 2-4 s) sensory store for raw acoustic features, measured by partial-report procedures.

Chunking rule: capacity is counted in chunks, not elements, so recoding digits into dates or acronyms raises effective span without changing store size.

Iconic memory: brief visual sensory store, roughly 250-500 ms, fading faster than echoic memory; masked by a following pattern.

Visual long-term memory is capacious for distinct scenes and objects (thousands of exemplars), but recognition accuracy falls when foils are same-category lures.

Picture superiority effect: images are recalled better than their verbal labels, especially on free recall with delayed testing.

Dual coding account: pictures are encoded both visually and verbally, giving two retrieval routes; verbal-only items get one.

Change blindness: failure to detect a change when the transients are masked; signals that visual memory holds gist and sparse detail, not a pixel copy.

Boundary extension: viewers remember scenes as extending beyond the photographed frame, evidence that scene memory is schema-driven reconstruction.

Visual similarity effect: confusable visual forms (mirror images, similar symbols) reduce recognition accuracy in short-term visual tasks.

Encoding specificity applies: recognition is best when test format matches study format (same viewpoint, same size, same orientation).

Capacity estimate: about 3-4 objects (Cowan's 4±1), lower than the classic 7±2 figure, which reflects chunk-based verbal span.

Change detection with a single-feature probe and a brief retention interval is the standard capacity paradigm; guessing is corrected with Cowan's K = N×(hit − false alarm).

Slot models (fixed number of discrete slots) predict a plateau in performance once set size exceeds capacity; resource models predict graded decline.

Feature binding: integrated objects (color+shape) are remembered about as well as single features, but only when attention is available at encoding.

Contralateral delay activity (CDA) over posterior electrodes rises with set size and plateaus near each individual's capacity limit.

At set sizes beyond capacity, errors come mainly from misbinding features across objects rather than from losing whole objects.

Visual working memory is attention-gated: filtering cues that mark irrelevant items as distractors reduce effective load and improve recall of targets.

Precision versus capacity: continuous-report tasks separate how many items are stored from how precisely each is stored; precision declines gradually with load.

Immediate memory is tested with zero or near-zero delay (typically under about 30 s) and without intervening distraction.

Free recall, cued recall, and recognition are the three standard retrieval formats; recognition is easiest, free recall hardest.

Miller's 7±2 applies to immediate serial recall of chunks; Cowan's 4±1 applies when rehearsal and chunking are prevented.

Brown-Peterson paradigm: a brief filled delay (counting backward) sharply reduces recall of a small item set, showing decay plus interference.

Proactive interference: prior trials with the same item category depress immediate recall even though the delay is unchanged.

Working memory span tasks (reading span, operation span) combine storage with concurrent processing and correlate with fluid reasoning.

Recency advantage in immediate free recall is largest for the last few items and disappears if a filled delay or a suffix intervenes.

Modality effect: spoken presentation yields a larger recency boost than visual presentation because of the echoic/auditory store.

Delayed memory uses retention intervals of minutes to years with distraction between study and test; long-term consolidation is the assumed mechanism.

Forgetting curve (Ebbinghaus): retention drops steeply in the first hours, then flattens; spaced repetition flattens it further.

Consolidation stages: synaptic (minutes to hours, protein-synthesis dependent) and systems-level (hippocampal-neocortical transfer over weeks to years).

Sleep after learning improves delayed retention, especially slow-wave sleep for declarative material; the benefit grows with the interval.

Encoding specificity and transfer-appropriate processing: delayed test performance depends on the match between study and test cues, not on depth alone.

Levels-of-processing effect is larger on delayed tests than on immediate tests; semantic encoding shows the biggest advantage after a delay.

Retroactive interference (new learning after study) and proactive interference (old learning before study) both depress delayed recall.

Reminiscence bump: older adults recall disproportionately more events from ages roughly 10-30 on delayed autobiographical tests.

Retrieval practice (testing effect) raises delayed retention more than an equal amount of restudying, with the gap widening as the delay grows.