Dec 10, 2019

The Tip of the Tongue Phenomenon

A review of the literature surrounding the tip of the tongue phenomenon, primarily centered on the transmission deficit theory.

An Account of the Tip of the Tongue Phenomenon: The Transmission Deficit Theory

The tip-of-the-tongue phenomenon (TOT) is particularly intriguing because it is an experience; a conscious phenomenon. There are few other examples of this kind of conscious metacognition, where there is a direct awareness of some kind of unconscious process, in this case memory retrieval. The phenomenon is also unusual in that there is a fairly clear causal relation between memory and the experience; the upstream causes of TOT aren't hidden behind layers of interpretation as they are with most every other aspect of conscious experience. TOT provides a unique window into understanding more about both human consciousness and memory at the functional level, as well as the relation between the two.

Before examining possible mechanisms behind TOT, it will be helpful to define it phenomenologically. One of the first descriptions comes from William James (1890), who outlined TOT in The Principles of Psychology:

Suppose we try to recall a forgotten name, [t]he state of our consciousness is peculiar. There is a gap therein; but no mere gap. It is a gap that is intensely active. A sort of wraith of the name is in it, beckoning us in a given direction, making us at moments tingle with the sense of our closeness, and then letting us sink back without the longed-for term. If wrong names are proposed to us, this singularly definite gap acts immediately so as to negate them. They do not fit into its mould. And the gap of one word does not feel like the gap of another, all empty of content as both might seem necessarily to be when described as gaps. (p. 259)

James' description of TOT is a good starting point as it reveals a few important details about the experience. First, a TOT state is one produced by the inability to remember something specific. James gives the example of a name, which can be broadened to the more general category of words. The target memory is quite distinct in this regard; it is usually not a long sentence or string of words, numbers, and facts, but rather a single word or common phrase. Next, the state has certain phenomenal properties, such as the feeling of activity; the feeling of knowing. There is also the added ability of those in a TOT state to distinguish between other false yet similar words and the target word, a kind of negative affirmation that memory usually does not provide.

One characteristic of James' description that proves useful for studying TOT is the feeling of knowing. In a TOT state, the feeling of knowing (FOK) is the most salient subset of the overall experience. From an experimental perspective, FOK is useful because it can be produced by asking subjects to judge their FOK for a variety of different tasks such as answering trivia questions (e.g. Metcalfe 1986), providing a gradient that can be useful in the construction of a model behind the effect. FOK can also be inferred from subjects' choices to respond to a prompt in a particular way, such as when they elect to make a split-second decision to retrieve the answer to a question as opposed to spending more time thinking it out (e.g. Reder & Ritter 1992). FOK is useful in that it can be independent of TOTs—anyone can be asked to judge their FOK for any piece of unretrieved information regardless of if the person is in a particular state.

At a high level and without the examination of any experiments or data, there are two prevalent accounts of TOT. First, there is the possibility that TOTs arise when memory itself is damaged in some way and the relevant information is simply missing (e.g. Brown & McNeill, 1966). The second explanation is that the issue is in the retrieval of the information; that there is some kind of deficit in the ability to access the memory that causes a TOT state (e.g. Burke et al., 1991). The transmission deficit hypothesis is a better-defined version of the second; that there is a difficulty in accessing memories that causes the TOT phenomenon, specifically that connections between lexical and phonological nodes are eroded enough as to prevent activation from reaching all phonological nodes and hence retrieval of the word (Burke et al., 1991). In this view, activation spreads first to lexical nodes then to phonological nodes in a model called node structure theory (Burke et al., 1991; Barresi et. al., 2000; Cohen & Faulkner, 1986). Activation of all the nodes would result in full access to the word and its meaning, while activation of some of the nodes such as just the lexical nodes or a few phonological nodes would result in a failure to retrieve the word, but still preserve access to some of the word's properties.

From the results of a number of experiments related to TOT and FOK, it will be argued that the transmission deficit account of TOT is the most suitable mechanism to explain the phenomenon. James' description of TOT's three main components will be used as an additional framework for evaluating each experiment's findings, to ensure that the mechanism as described gives an accurate account of the phenomenological aspects of TOT.

One important component of the transmission deficit hypothesis is that access to lexical attributes of the word triggering a TOT state is preserved (Burke et al., 1991). This indicates that semantic aspects of the word, such as its meaning and certain letter components may be accessible in a TOT state, even if the word itself cannot be produced. A study conducted by Brown and McNeill (1966) examined the ability to reject incorrect words while in a TOT state. The researchers created a list of relatively unused but likely familiar words to their subjects. They presented the definitions of these words to subjects, who were instructed to guess the word, number of syllables, initial letter, and similar words in both sound and meaning. The subjects also had to signal if they were in a TOT state. The researchers were particularly interested in the words that the subjects wrote down as similar in sound and similar in meaning (SS and SM, respectively) to the actual word, as well as the number of syllables the subjects guessed. They found that the number of syllables reported was the best indicator of the TOT state, where subjects were most accurate about the number of syllables when in TOT states but were unable to recall the precise word.

Brown and McNeill's findings support the notion that TOT states allow some access to semantic attributes of a word, like its meaning. Syllabic information and the first letter is stored in phonological nodes, which the transmission deficit hypothesis suggests are not fully activated. Given fact that the subjects were able to accurately report the number of syllables, it seems likely that some phonological nodes are being activated, but not all. Additionally, the first syllables of SS words produced by subjects were found to most often match the first syllables of the target word, and 48% of the SS words produced had the exact same number of syllables as the target word, as compared to just 20% for SM words. This finding further supports the idea that some syllabic information is being accessed during a TOT state. Taken with the fact that over 70% of the words produced by subjects in a TOT state were SS words rather than SM words, it seems reasonable to propose that subjects are accessing some phonological nodes during retrieval. When they produced SM words, subjects were not activating any phonological nodes, as the only information they had to produce a similar word was the meaning of the word, which they were given or knew through activation of lexical nodes.

It might be argued that the subjects' tendency to report SS words could be attributed to a complete activation of all the available nodes, including phonological, and the failure to retrieve the word is due to the word being completely missing from memory (e.g. Brown & McNeill, 1966). Such a possibility seems unlikely when considering that TOTs can be resolved without any additional prompting; simply ruminating for some time can lead to a subject recalling the word (Burke et al., 1991). More time spent thinking about the word cannot lead to the completion of a missing memory—it must be the case that the information is already known and eventually becomes activated. The transmission deficit theory allows for such a possibility; if enough activation reaches the relevant phonological nodes through a weak connection, the word will still be recalled.

One related question is how activation best reaches the relevant nodes to trigger the recall of a word. The transmission deficit theory suggests that a large amount of activation is needed to overcome weak connections between nodes, so it seems likely that activating words with similar phonological characteristics would assist in the resolution of TOTs. This is because activating phonological nodes shared between the target word and a word with similar characteristics would increase the amount of activation getting to the relevant nodes, given the node structure theory's model of memory (Burke et al., 1991; Abrams et al., 2003). A study by Abrams et al. (2003) aimed to address this hypothesis by presenting subjects in a TOT state with phonologically similar words. The researchers found that when subjects read words with the same first syllable and words with shared phonemes, the TOT state was more often resolved than when presented with words that shared first letters or other syllables. This result bolsters the transmission deficit theory's account of a weakened connection that is resolved through greater activation levels. More specifically, it suggests that there are certain phonological nodes that are more closely tied to the nodes that must be activated for a full retrieval. Activating these related shared nodes spreads more activation across the weakened connections, enough to result in resolution of the TOT state.

The transmission deficit hypothesis also provides an account of why activation does not reach all phonological nodes during a TOT state. It suggests that the connection between lexical and phonological nodes is degraded to a point that a large amount of activation is necessary to pass through and reach phonological and propositional nodes, while a small amount of activation will not, producing a TOT state (Burke et al., 1991). An important consideration is how such a connection might be weakened, and what factors might contribute. Burke et al. (1991) ran a study in which subjects carried a journal with them and noted down when they experienced TOTs. They recorded the target word (after the effect) and contextual information such as the number of syllables they thought the target had, and words similar to the target word. The researchers analyzed the characteristics of the words, as well as the participants, who were divided into young, middle-age, and old. They found that the words that brought on TOT were similar in their relatively low frequency and low recency of use. In a second lab-based study similar to that of Brown and McNeill (1966), the researchers found that when given prompts for low-frequency words, the same two factors were good predictors of the TOT effect.

The fact that words that triggered TOTs were found to be low frequency and to have not been used recently fits into the transmission deficit theory's account of TOTs. It seems probable that over time, connections between nodes are weakened with a lack of use. It also seems likely that lower frequency words would have both fewer connections and weaker connections. The node structure theory suggests that this is the mechanism that leads to activation failing to pass between nodes (Burke et al., 1991; Barresi et. al., 2000; Cohen & Faulkner, 1986).

Another important finding by Burke et al. is that as age increases, the number of TOTs increases, as does the number of TOTs for names of people, places, and movies. This supports the idea that TOT is related to a deficit of activation; people, places, and movie names would all be expected to have fewer connections to other contexts and concepts. As the working memory of older adults is diminished there is less activation available, and given that nodes with names of people, places, and movies have relatively few connections, it would be expected that it would become more difficult for activation to spread to these nodes and pass through a weakened connection. Other work has replicated the findings that older adults are diminished in their ability to retrieve proper names and attribute this deficit to a lack of access to phonological nodes (e.g. Barresi et. al., 2000; Cohen & Faulkner, 1986). The same mechanistic account as incorporated into the transmission deficit hypothesis would explain why these particular words trigger more TOTs among older adults.

The final aspect of TOTs yet to have been accounted for in the transmission deficit theory's account is the experience of a high FOK. As mentioned previously, FOK can be studied in broader contexts due to its more general constraints—it relates to less specific knowledge, not just words. In experiments focused on FOK, the goal is often to gauge the subjects' awareness of knowing the answer to a question, dependent on a variety of experimental factors. Reder and Ritter (1991) studied FOK by giving subjects the option to either retrieve or calculate the answer to a math problem (incentivizing retrieval) and varying how often the subjects saw certain aspects of the problems, specifically the operator and operands. They found that the frequency with which subjects saw the same and similar problems had an effect on FOK; as subjects saw the same operands paired together more often, the more likely they were to try and retrieve the answer to the problem, even if mistaken that they had seen it before.

Reder and Ritter showed that a judgement about the question is made before any attempt to find the solution. This judgement was found to be based on the frequency of the elements of the question, suggesting that FOK is caused by the change in activation for a particular association between entities, in this case numbers, which would increase with a higher frequency.

These findings coincide with the mechanism proposed by the transmission deficit theory. In particular, a word's association with a related concept would be expected to be fairly strong, indicating that the words that trigger TOTs aren't purely low frequency. It would expected for people to have used these words in specific contexts and little elsewhere, or for them to have used these words fairly regularly in the past. These conditions would result in higher FOK for the word due to higher strength in memory. These predictions are confirmed by the results of Burke et al. (1991) who determined that words that triggered TOTs had a longer time since last use, specifically acquaintances that had not been contacted for a while—words that have been used regularly in the past but not recently.

A series of findings related to aspects of the TOT experience have been described, bolstering the transmission deficit account. James' description of the phenomenon has been a useful framework for identifying aspects of the experience that require mechanistic explanation. Summarizing the mechanism, a TOT state occurs when a particular word is sought from memory and its lexical nodes are activated, providing some information about the word. Further, some phonological nodes are also activated, providing some attributes such as the syllabic structure and perhaps even the first letter. However, a weakened connection between lexical and phonological nodes prevents activation from spreading to all phonological nodes. The word also must be quite particular; it should have a relatively low baseline activation level but with weakened connections due to interference, lack of use, or time. This allows for an FOK judgement about the word to be made without activating the word itself, leading to the very salient experience of feeling the answer is known, but not being able to retrieve it.


Abrams, L., White, K. K., &; Eitel, S. L. (2003). Isolating phonological components that increase tip-of-the-tongue resolution. Memory & Cognition, 31(8), 1153-1162.

Barresi, B. A., Nicholas, M., Tabor Connor, L., Obler, L. K., & Albert, M. L. (2000). Semantic degradation and lexical access in age-related naming failures. Aging, Neuropsychology, and Cognition, 7(3), 169-178.

Brown, R., & McNeill, D. (1966). The "tip of the tongue" phenomenon. Journal of verbal learning and verbal behavior, 5(4), 325-337.

Burke, D. M., MacKay, D. G., Worthley, J. S., & Wade, E. (1991). On the tip of the tongue: What causes word finding failures in young and older adults?. Journal of memory and language, 30(5), 542-579.

Cohen, G., & Faulkner, D. (1986). Memory for proper names: Age differences in retrieval. British Journal of Developmental Psychology, 4(2), 187-197.

James, W., Burkhardt, F., Bowers, F., & Skrupskelis, I. K. (1890). The principles of psychology (Vol. 1, No. 2). London: Macmillan.

Metcalfe, J. (1986). Feeling of knowing in memory and problem solving. Journal of Experimental Psychology: Learning, Memory, and Cognition, 12(2), 288.

Reder, L. M., & Ritter, F. E. (1992). What determines initial feeling of knowing? Familiarity with question terms, not with the answer. Journal of Experimental Psychology: Learning, memory, and cognition, 18(3), 435.