When an Answer Feels Close: Tip-of-the-Tongue States and Wanting to Know

Failing to produce a name is not quite the same as feeling that the name will arrive at any moment. Both leave a gap in the conversation, but only the second carries a compelling sense of proximity. That feeling may encourage another retrieval attempt or a request for the answer. It is not, however, a certificate that the sought information is correct. The tip-of-the-tongue state is interesting because it combines present failure with anticipated access. Understanding its connection to curiosity requires asking about the feeling as well as scoring the response. [1]

Janet Metcalfe, Bennett Schwartz and Paul Bloom investigated that connection in 2017. After attempting general-information questions, participants reported whether they were in a tip-of-the-tongue state and whether they wanted to see the correct answer later. The study separated objective response accuracy from a subjective sense of unresolved access. In 2023, Linus Holm and Michael Wells also examined the relationship in a word-pair learning experiment. Reading the two findings together helps preserve a distinctive observation without promoting it into a rule for every information-seeking situation. [1][2]

What people predict they would want

The 2017 article includes a separate MTurk survey with 127 respondents. They chose the situation in which they would have the strongest urge to search for an answer online. The alternatives distinguished being able to produce the answer, being certain of not knowing it, feeling able to recognize it, and experiencing a tip-of-the-tongue state. Twenty-five percent selected the tip-of-the-tongue option, compared with 39 percent for definite nonknowledge, 31 percent for a feeling of knowing, and 5 percent for being able to produce the answer. [1]

These responses were predictions about hypothetical situations, not records of actual searches. The survey respondents were also not the same sample as the students in the subsequent experiment. It would therefore be misleading to describe the comparison as showing how particular individuals failed to understand their own behavior. One measurement concerned a general expectation; the other concerned choices made around specific retrieval attempts. Their contrast is informative only if those differences remain visible. [1]

Catching the feeling near the attempt

The main experiment involved 46 Columbia University and Barnard College undergraduates aged 18 to 29. Each faced 82 general-information questions. A question appeared on screen and was read aloud by the experimenter. Once the reading finished, a five-second response period began. Participants then reported whether they were in a tip-of-the-tongue state, immediately followed by a choice about seeing the answer later. This timing was intended to capture a state close to the retrieval attempt rather than reconstruct it after a long completed task. [1]

Half the questions included an additional memory requirement. Participants had previously seen four digits that they needed to retain; on the other trials they saw asterisks. After answering and making the two judgments, they recalled the digits or stated that asterisks had appeared. Question sets were switched between load conditions across two groups, and question order was randomized for each participant. None of that means the researchers randomly assigned the tip-of-the-tongue feeling itself. The feeling remained a participant report, while timing and the additional memory task were features of the procedure. [1]

Participants were told that they could later see answers to at most 10 percent of the questions. This restriction was not enforced: they were ultimately allowed to see all the answers. The instruction addressed a pilot-testing problem in which many people requested every answer when no restriction was mentioned. Consequently, the observed choice was made under an apparent limitation. It was not unrestricted browsing, and participants did not actually purchase information. The setup asked them to prioritize answers they wanted to see later. [1]

Wanting confirmation after a correct answer

Across the experiment, the reported probability of wanting to see the answer was 0.44 when a tip-of-the-tongue state was reported and 0.18 when it was not. For omission errors alone, the corresponding values were 0.45 and 0.24. Commission errors, including partial or wrong responses, also showed the association. Separating these response types mattered: not producing an answer and producing an incorrect one are different outcomes, even though both can leave the question unresolved. [1]

The more unusual observation concerned correct responses. Participants sometimes reported a tip-of-the-tongue state after producing the correct answer. In that subset, the probability of wanting the answer later was 0.24 with the feeling and 0.03 without it. The researchers suggested that uncertainty about an otherwise correct response might help explain the phenomenon. That is a plausible interpretation, not a separately demonstrated mechanism showing that low confidence generated every such report. Getting the answer right and feeling that the retrieval episode has ended are not identical measures. [1]

The conditional analyses also do not provide equally dense data from every participant. Not everyone reported tip-of-the-tongue states in every combination of response type and memory-load condition. The article consequently reports reduced degrees of freedom for some comparisons. The correct-response finding is notable, but should not be narrated as something every student displayed in every condition. Its significance comes from the separation of response accuracy and reported experience, not from an invented universality. [1]

A different test with a waiting cost

Holm and Wells's 2023 study primarily investigated satisfaction in retrieval, while also testing whether tip-of-the-tongue states predicted requests for feedback. English-speaking participants learned 100 Swahili–English word pairs over five sessions across a week. Of 70 people who completed the first session, 30 completed all sessions and met the inclusion criteria. In the final test, participants attempted each English translation and made several judgments, including a yes-or-no report of a tip-of-the-tongue experience. [2]

Someone requesting the correct translation waited five seconds before seeing it. Someone declining feedback proceeded directly to the next item. Here information seeking carried an actual time cost. The average probability of requesting feedback was 0.61 on tip-of-the-tongue trials and 0.52 on other trials. Although the numerical difference pointed in the same direction as the earlier result, it was not statistically reliable; the reported p value was 0.076. This is neither evidence of proven equivalence nor a demonstration that the 2017 finding must be discarded. [2]

Several conditions differed at once. Newly learned translations replaced general knowledge, an online repeated-learning procedure replaced an experimenter-managed session, and waiting for immediate feedback replaced selecting an answer to see later. The studies did not isolate which of these differences explained their contrasting strength of evidence. Declaring the waiting cost the sole cause would be as unsupported as assuming that the earlier association must appear everywhere. Substantial participant attrition in the later study is an additional boundary on generalization. [2]

The absolute probabilities should not be treated as points on a universal curiosity scale, either. A request made under an apparent answer quota and one made in exchange for five seconds are choices in different environments. Their denominators refer to the respective conditions and samples. Comparing their patterns can motivate further experiments, but simply placing 0.44 beside 0.61 does not reveal which population is more curious in general. [1][2]

A meaningful signal, not a complete explanation

The 2017 study supports an association between an immediate feeling of near access and wanting information, including when the response was objectively correct. It does not establish an evolutionary function, an experimentally isolated causal effect of the feeling, or a guarantee of improved long-term learning. Participants were not assigned to feel the state at random. The possibility that curiosity reflects an urge to resolve discomfort, or anticipation of pleasure on finding the answer, belongs to the authors' explanatory discussion rather than to a separate biological measurement. [1]

Together, the studies suggest that unanswered questions need not occupy the same subjective distance. A person can have supplied the correct response while still feeling unsettled, and a strong sense of near retrieval need not reliably increase willingness to wait in another task. Objective accuracy, subjective proximity and the conditions of obtaining information can be examined separately. The tip-of-the-tongue state is therefore more than a label for failed recall: under specified circumstances, it offers a metacognitive signal associated with the next information-seeking choice. Recognizing its evidential limits makes that observation clearer, not less interesting. [1][2]

Sources

[1] Metcalfe, J., Schwartz, B. L. and Bloom, P. A. (2017). The tip-of-the-tongue state and curiosity. Cognitive Research: Principles and Implications, 2:31. CC BY 4.0.

[2] Holm, L. and Wells, M. (2023). Reliable retrieval is intrinsically rewarding: Recency, item difficulty, study session memory, and subjective confidence predict satisfaction in word-pair recall. PLOS ONE, 18(10): e0292866. Creative Commons Attribution.

These studies are credited to their authors and discussed in independently written explanations and comparisons, without direct quotations. CC BY 4.0. Supplementary data were not reanalyzed.