Why a Simple Puzzle After Learning Can Erase New Facts From Memory

Forgetting has long been treated as the enemy of learning, but psychologists increasingly view it as an adaptive feature of the mind, clearing away irrelevant information so that behaviour can remain flexible. Yet the precise mechanisms that cause memories to fade remain fiercely debated. One of the oldest candidates is retroactive interference, the phenomenon in which new information or activity disrupts a memory that has already been formed. A new study published in Current Psychology by Luke P. Fisher of Aston University and Tom Mercer of the University of Wolverhampton now provides some of the strongest evidence yet that even a brief, unrelated distraction after learning can measurably impair our grip on meaningful, factual information.

The story begins more than a century ago. In 1900, the German researchers Georg Müller and Alfons Pilzecker reported that people who memorised verbal material and then completed a picture-description task recalled far fewer of the original items than those given an unfilled delay. Only 24 percent of the verbal stimuli were recovered after the picture task, compared with 56 percent in the interference-free condition. The striking implication was that the interfering activity did not need to resemble the original material in any way; a mentally effortful task of a completely different kind was enough to induce forgetting. This became known as non-specific retroactive interference, or NRI, and it has haunted memory research ever since.

The dominant theoretical account of NRI is consolidation theory. According to this view, a newly encoded memory trace is fragile and requires a period of stabilisation, involving biological processes that unfold over minutes, before it becomes resistant to disruption. If attention and cognitive resources are diverted to a distracting task during this vulnerable window, consolidation is effectively put on hold, and the trace may be weakened or lost. Wakeful rest, by contrast, is thought to allow consolidation to proceed unimpeded, which is why quiet rest after learning has repeatedly been shown to boost retention. An alternative framework, temporal distinctiveness theory, offers a different explanation. In models such as SIMPLE, developed by Gordon Brown and colleagues, memories are harder to retrieve when they are temporally crowded by neighbouring events. A filled delay introduces additional psychological neighbours that reduce the distinctiveness of the target memory, making it less discriminable at retrieval, even when the total retention interval is identical.

Despite the theoretical appeal of NRI, the recent literature has been strikingly inconsistent. Some studies have found interference from visual search tasks, spot-the-difference puzzles, video games, and even internal autobiographical thinking. Others have found nothing. Fatania and Mercer reported NRI in children but not adults, and when children were given more generous encoding and retrieval time, the effect vanished there too. Martini and colleagues, testing word lists and stories with delays filled by Raven’s matrices, music, or puzzles, frequently found no advantage for wakeful rest over distraction. These null results raised an uncomfortable possibility: perhaps NRI is such a mild form of forgetting that it is easily masked, appearing only under specific conditions that researchers had yet to identify.

Fisher and Mercer suspected that two factors might explain the contradictions. The first was the nature of the material. Most prior work used simple word lists, whereas meaningful, semantically rich information, the kind people actually encounter in education and everyday life, had received far less attention. Salient or highly imageable material might resist interference, as some earlier findings suggested. The second factor was condition order. Many experiments manipulate rest and distraction within the same participants, counterbalancing which comes first, and several studies had hinted that interference effects are stronger when the distracting condition occurs in the second memory task rather than the first. Proactive interference, in which material learned earlier disrupts later learning, could also contaminate within-subject designs.

To test these ideas, the researchers ran three experiments with large samples, a notable departure from earlier NRI studies that often included fewer than 40 participants. In the first experiment, 121 undergraduates learned 20 trivia facts, such as that the word squirrel comes from a Greek term meaning shadow-tail, and then recalled them after a five-minute delay filled either with quiet wakeful rest or with spot-the-difference puzzles. The order of the two conditions was counterbalanced. Intriguingly, the overall NRI effect was absent: recall averaged around a quarter of the facts in both conditions. But a significant interaction with condition order emerged. When the distracting task came second, recall dropped markedly, whereas the control condition showed no such time-on-task decline. Interference, it seemed, was far more damaging when it arrived late in the session.

The second experiment, with 100 new participants, made two changes. The fact lists were shortened to 14 items each, equated for interest using the ratings collected earlier, and the concurrent interest and mind-wandering probes were removed from the encoding phase. The researchers reasoned that rating each fact for interest had forced deeper processing, creating more robust memories that could shrug off distraction, and that the dual demands of rating and encoding may have depressed baseline recall to the point where interference could not be detected. The results were dramatic. This time there was a clear overall NRI effect, with recall of roughly 42 percent in the rest condition versus 35 percent after distraction, and the late-occurring interference condition once again produced the worst performance of all. When distraction followed rest, the drop in recall within participants was large; when distraction came first, it was negligible.

The third experiment tackled a lingering confound. Because the first two studies manipulated condition within participants, a decline from the first to the second memory test could reflect fatigue or proactive interference rather than interference per se. So 135 participants recruited online were assigned to experience either two rest conditions or two distraction conditions. The pattern was unambiguous. The distraction group recalled consistently fewer facts than the rest group, an effect that was moderate early on and large in the later test. Moreover, a proportional analysis of the change from the first to the second test showed that the interference group declined by a median of about 21 percent, compared with roughly 6 percent for the control group, a difference that was statistically significant under the directional hypothesis the authors had set out. Both groups declined over time, but distraction made the decline substantially steeper.

Taken together, the three experiments suggest that NRI is real, that it applies to meaningful factual material rather than only to artificial word lists, and that its magnitude depends sensitively on experimental context. The authors argue that neither consolidation theory nor temporal distinctiveness theory alone can fully explain the pattern. Consolidation accounts why a distracting task immediately after learning is harmful, but struggles to explain why rest before learning also confers benefits, which fits more naturally with the idea of temporal isolation. The condition-order effects point to a role for cumulative proactive interference or for cognitive load and mental fatigue, which may make people more susceptible to distraction as a session wears on. Rehearsal during rest is another candidate explanation, though prior studies have generally found only weak or non-significant correlations between self-reported rehearsal and recall, and some researchers have challenged rehearsal as the mechanism underpinning the rest effect.

The practical implications are tantalising. If a few minutes of simple puzzles can erode freshly learned facts, then the modern habit of reaching for a phone or scrolling social media immediately after studying may be quietly undermining retention. The flip side is equally appealing: a brief period of quiet rest after learning, something as simple as sitting with closed eyes, might offer a cheap and effortless boost to memory. The authors caution that their study was not designed to adjudicate between theories, that the meaningful trivia facts were not controlled for truthfulness or prior knowledge, and that the proportional decline analysis in the final experiment was significant only under a one-tailed test. But the central message stands. Non-specific retroactive interference appears to be a low-level but pervasive force in forgetting, one that may accumulate across successive memory challenges, and one that any serious account of how, and why, we forget must now reckon with.

Subject of Research: Non-specific retroactive interference effects on memory for meaningful factual material

Article Title: Non-specific retroactive interference for meaningful material

Article References: Fisher, L. P., & Mercer, T. (2026). Non-specific retroactive interference for meaningful material. Current Psychology, 45(18), Article 1512. https://doi.org/10.1007/s12144-026-10052-5

Image Credits: AI Generated

DOI: 10.1007/s12144-026-10052-5

Keywords: retroactive interference, memory consolidation, wakeful rest, forgetting, temporal distinctiveness, semantic memory, free recall, proactive interference, cognitive load, spot-the-difference task, long-term memory, psychology