Cue-Dependent Forgetting: Why a Reminder Brings Back What Felt Lost

Forgetting can happen in two quite different ways. The memory can decay or be overwritten, so that nothing is left to retrieve; or the memory can be intact and the retrieval fail, because the cues present at the moment do not connect to it. Endel Tulving and Stephen Madigan drew the distinction between trace-dependent and cue-dependent forgetting in a 1970 review of memory research, and Tulving's 1974 article argued that a great deal of everyday forgetting is of the second kind. The evidence is the reminder: when a cue brings back something that a moment ago could not be recalled at all, the memory was there the whole time.

The mechanism is in how memories are stored and reached. Tulving and Pearlstone's 1966 distinction between availability and accessibility is the frame. What is available is everything held in the store; what is accessible is the subset that the cues of the moment can reach. Free recall, with no cues beyond the instruction to remember, reaches only part of what is available. Supply a cue that was present at encoding, such as the category a word belonged to, and more of the store comes within reach. The memory did not change; the route to it did.

Which cues work depends on what was encoded alongside the memory. Tulving and Thomson stated this as the encoding specificity principle in 1973: a cue is effective at retrieval to the extent that the information it carries was stored with the memory at the time of learning. A category name works in the demo above because the category names were on the list when it was read. A cue that was never part of the original episode, however closely related to the word, works less well. This is why the same fact can be unreachable in one setting and immediate in another.

Context is a cue like any other. Godden and Baddeley had divers learn word lists either on land or underwater and recall them in the same or the other environment; recall was reliably better when the environment at recall matched the one at learning, and the drop when it did not match was substantial. A 2021 replication by Jaap Murre with sixteen divers did not find the same-context advantage, so the example shows that environmental cues can matter rather than that they always will; how much they matter depends on the conditions. Rooms, moods, times of day and the people present all become part of the episode and can all serve as cues to it later. Whether this is what happens with the errand lost on walking through a doorway is less clear: Radvansky, Krawietz and Tamplin found in 2011 that walking through a doorway impaired memory for objects just picked up, but that returning to the original room did not restore it, and they favored an account in which the doorway marks the boundary of an event rather than removing cues.

Endel Tulving and Zena Pearlstone published "Availability versus accessibility of information in memory for words" in the Journal of Verbal Learning and Verbal Behavior in 1966. High school students studied lists of words organized by category, with the category names shown during study alongside the words, and the lists varied in length, 12, 24 or 48 words, and in how many words each category contained, one, two or four. The students were told they would be asked for the words, not the category names.

Recall was tested in two ways. Some students recalled freely, with no help beyond the instruction. Others were given the category names on the recall sheet as cues. Cued recall exceeded free recall, and the advantage grew with list length: on a short list the cues added little, because free recall already reached most of the words, while on the 48-word lists the gap was large. The students who had recalled without cues were then given the category names and asked to recall again, and they recovered words they had just failed to produce. Those words had been available throughout; they had not been accessible until the cue arrived.

Tulving and Joseph Psotka extended the result in 1971 to the question of why learning more lists seems to erase earlier ones. Students learned several lists of categorized words in turn, and free recall of the earlier lists fell as more lists were added, the classic pattern of retroactive interference. When the category names were supplied as cues, recall of the earlier lists was largely restored. What looked like memories being overwritten was mostly memories becoming harder to reach, and the right cue reached them. Tulving's 1974 article in American Scientist drew the two studies together under the heading of cue-dependent forgetting, a term he and Stephen Madigan had already used in their 1970 review.

The pattern shows up wherever something cannot be recalled in one setting and returns easily in another. The rows are illustrations, not findings from the cited studies.

Cue-dependent forgetting is countered at both ends: by storing more routes when learning, and by supplying cues when recall fails.

• Organize what you learn. Material stored under categories or headings can later be reached through them, as the demo above shows; an unorganized list offers no such handles.
• Generate your own cues when stuck. For a name, run through the alphabet; for a fact, recall the place, the book or the person it came with. A cue that was part of the original episode is the one most likely to work.
• Match the contexts. If you will need something in a particular setting, learn or rehearse it there, or at least imagine the setting while learning. Varying the places you study widens the set of contexts that can serve as cues.
• Return to the scene. When something learned in a place will not come, going back there can supply cues that were part of the encoding. Treat this as a possible help rather than a guarantee: in one doorway study, returning to the room did not bring back what had been forgotten.
• Treat a blank as unreachable rather than gone. The memory may well be available; the useful move is to look for a route to it rather than to conclude it has been lost.

Cue-dependent forgetting is a failure of access to a memory that is still stored. Its neighbors are other things that can go wrong between storing and recalling.

• It is not the tip-of-the-tongue phenomenon, which is the particular feeling of knowing a word while being unable to say it, often with its first letter or syllable count available. A tip-of-the-tongue state is one experience that cue-dependent forgetting can produce; cue-dependent forgetting also covers the many cases where nothing feels close at all until the cue arrives.
• It is not the context effect, the finding that recall is better when the setting at recall matches the setting at learning. The context effect is one form of cue dependence, in which the environment is the cue; cue-dependent forgetting also covers semantic cues such as the category names in the demo.
• It is not absent-mindedness, in which attention was never on the thing at all, so it was never properly encoded. The two overlap in prospective memory, where a missed cue is the failure, as when you walk into a room and the errand is gone; the pure case of absent-mindedness has no retrieval route to restore, because the memory was not stored in the first place.