Part-Set Cueing Effect: Why Reminders Can Block the Rest of a List

The effect is a problem of retrieval, not of storage: the uncued words are still in memory, and the cues make them harder to reach. Three accounts of how have been argued over since 1968, and the evidence now favors a mix of the last two.

The first account, retrieval competition, was set out by Rundus in 1973. Seeing some list words again strengthens them, so when the search for the rest begins, the strengthened words keep coming to mind first. Each time a cue word surfaces instead of a new one, the search has produced a failure, and a search that produces too many failures in a row is abandoned.

The second, strategy disruption, was developed by Basden and Basden in 1995. During study, a person builds a private organization of the list: a running order, a few groupings, an image or two. A reminder sheet imposes a different order and a different grouping, and recall that follows the sheet's structure abandons the one that would have worked. On this account the cues hurt because they change the plan.

The third, retrieval inhibition, treats the cues as a form of forced practice. Bäuml and Aslan showed in 2004 that re-exposing part of a list as cues lowered recall of the rest, as did asking people to retrieve those same items, while re-exposing them simply for extra learning did no harm. In their experiment, people recalled 77.3 percent of the target words when nothing had been re-exposed and about 71 percent when four words from each list had been shown again as cues, with or without a delay. Their reading is that cues prompt covert retrieval of the cue words, and retrieving some items inhibits their unretrieved neighbors, which is the mechanism behind retrieval-induced forgetting.

Whatever the balance between these accounts, the boundary is clear. Cues that arrive after a person has already written down everything they can recall unaided do not hurt. The effect belongs to the moment when a general search of memory is about to start and part of the answer is placed in front of it.

Norman Slamecka published "An examination of trace storage in free recall" in the Journal of Experimental Psychology in 1968. The question behind it was whether words learned together are stored as a linked network, in which case any list word should serve as a doorway to the others, or as independent traces, in which case cues would do nothing. Six experiments gave participants a 30-word list, read aloud twice, and then asked them to recall it. Experimental groups received a written subset of the list, from 5 to 29 words depending on the experiment, and were told to recall the rest; control groups recalled the whole list from a blank page. The rarity of the words and the strength of the associations between them were varied across experiments, and the fourth experiment used a list of six words from each of five categories, with 5, 15 or 25 words given as cues.

The cued groups recalled a smaller proportion of the non-cue words than the controls, and the deficit did not depend on the composition of the list or the number of cues. Slamecka read this as evidence against associative storage: if the traces were linked, cues should have helped. He also tried to explain the deficit away. His last two experiments tested whether the cued groups were simply losing time, or losing the last few words from short-term memory while they read their cue sheets. When cues came only after four minutes of unaided recall, or when controls were kept busy with a filler task while the cued group read its cues, the deficit disappeared, and he suggested the effect might be a short-term artifact.

That suggestion did not survive. Slamecka's own 1969 follow-up delayed every recall by 30 seconds and still found no help from cues, and a slight deficit in one experiment. Roediger, Stellon and Tulving reported in 1977 that after two hearings of 48 unrelated words, people given 16 or 32 of them as cues recalled fewer of the 16 uncued words than controls, across a ten-minute test that gave everyone time to run out of words. By 1984 Raymond Nickerson could review dozens of studies under the title "Retrieval inhibition from part-set cuing: A persisting enigma in memory research," and the enigma was the reliability of the effect rather than its existence.

The rows below are illustrations of the situation the experiments model, a partial list placed in front of a search of memory. They are not findings from the studies cited on this page.

The experiments point to one rule: do the unaided search first.

• When you need to recall a list, write down everything you can before you look at any partial version of it. Cues that come after exhaustive recall did not hurt in Slamecka's fifth experiment, and they can still help you check.
• In meetings, have everyone write their ideas or recollections privately before anything goes on the board. The board becomes a checking tool rather than a set of cues.
• If someone starts listing items to help you, ask them to wait until your own search has run dry.
• When you cue yourself, use the kind of cue that targets one item, such as a first letter or a category name, rather than a handful of the items themselves. Bäuml and Aslan's experiment used category-plus-first-letter cues to test recall, and that kind of cue is a door to one word rather than a crowd in front of the others.

Part-set cueing is one of several ways retrieval can fail while the memory itself is intact. Its neighbors differ in what blocks the search and when.

• It is not cue-dependent forgetting, where recall fails because the right cue is absent and returns when it is supplied. Here cues are present and they make things worse.
• It is not memory inhibition of the retrieval-induced kind, where practicing retrieval of some items weakens their unpracticed neighbors. The two are closely related, and Bäuml and Aslan argued that cues work by prompting that same covert retrieval, but part-set cueing needs no practice phase: the cues are simply shown.
• It is not the tip-of-the-tongue phenomenon, where a single word is blocked, often by a wrong word that keeps arriving. Part-set cueing lowers recall across a whole set of items at once.