The Second Hearing

The second hearing is a new measurement

This essay treats each hearing as a measurement: the listener takes a new reading of the same music. A repeat changes the conditions of listening even when the sound file stays the same, because the listener carries a trace of the first hearing. On the second pass a rhythm may be easier to follow. A phrase can start to feel predictable, and an unfamiliar texture may now have a name.

That observation is a frame for the essay and makes no claim about neuroscience. The evidence becomes useful only once the outcome is named. Liking, recognition, perceived unpredictability, shared neural timing and memory interference are different measurements, and a change in one cannot stand in for a change in the others. Shared neural timing here means how closely the brain signals of different listeners move in step. Memory interference means how much a song gets in the way of a later memory task.

Four different forms of familiarity

This essay separates four forms of familiarity: piece familiarity, style familiarity, within-piece repetition and across-hearing repetition. With piece familiarity, the listener has heard this exact recording or composition before. Style familiarity is the set of expectations a listener has learned from related music, and it applies even when the piece is new. Within-piece repetition happens during one hearing, when a rhythm, phrase, chorus, timbre (the color of a sound) or form comes back. Across-hearing repetition is the next meeting with the same excerpt or piece.

These distinctions matter because a new piece can belong to a familiar style, and a piece heard many times can stay hard to predict. A 2017 naturalistic study, one built to resemble ordinary listening, found that familiarity with similar music strongly predicted liking, though participants did not know the chosen examples at the start. That is style familiarity at work without piece familiarity. A 2019 study used EEG, a recording of electrical activity at the scalp. It sorted excerpts into familiar and unfamiliar styles before it tested immediate repetition, and its result comes two sections later.

The path of liking depends on how people listen

In three controlled experiments published in 2004, the same number of plays had no single fixed meaning across listening conditions. The experiments varied the amount of exposure and the ecological character of the music, meaning how close it came to music heard in daily life. They also varied whether people listened with focus or only incidentally, while doing something else. Music heard more often was liked more when people heard it incidentally. With focused listening, liking for the most true-to-life music rose and later declined. For less true-to-life music that rise and fall was weaker or absent. Recognition generally rose more with focused listening than with incidental listening.

The 2017 study used a different schedule. Fifteen adults rated 40 examples they did not know at the start. They gave ratings after the 1st, 10th, 19th and 28th presentation, that is, after each of those plays, over about four weeks. Liking rose with more plays at all four complexity levels, as rated by experts. A standard statistical test gave the presentation effect as F(3,42)=12.35, p<0.00001. That p value means a rise this large would occur by chance less than once in 100,000 tries if plays made no difference. Omega-squared, an effect-size measure, was 0.197: the number of plays accounted for about a fifth of the variation in ratings. The presentation-by-complexity interaction was not significant, so the study found no reliable sign that complexity changed that rise.

Because the studies differ in listening mode, material, schedule and setting, neither result overrides the other. Read together, they show only that repeated exposure can change liking and that its path depends on the conditions of exposure.

Attention is not liking

In the 2019 study, 40 participants in two groups heard instrumental excerpts three times in a row. The researchers measured inter-subject EEG correlation: how closely the brain signals of different listeners rise and fall together. They proposed that engagement with music affects this correlation. It fell across repeats for familiar-style excerpts and held up better for unfamiliar-style ones. The pattern was tested in both groups, each with its own set of excerpts.

The strongest caveat is that the researchers did not check this EEG measure against listener behavior to confirm that it tracks musical engagement. So a falling correlation cannot be read as boredom, or a sustained one as quality. Attention may change while liking or recognition moves in another direction.

Memory can move without preference

A 2023 experiment played four new sung choruses to 44 participants, each chorus between one and four times. The next day each song played again while participants did a serial-recall task, which asks people to recall items in order, and they kept doing the task after the song ended. Earlier work had found that an earworm, a tune that gets stuck in the mind, takes up working memory when the music is sung, as a kind of inner singing. Working memory is the short-term store people use to hold items in mind. Worse recall can therefore be read as a sign that a song is still running in the mind. Greater exposure tended to increase interference with the task during and after some songs, and the effects varied by song. Ratings of familiarity and of the wish to sing along rose between the two sessions.

The experiment separates outcomes that everyday talk often treats as one. A song can grow more familiar, easier to hear again in the mind or more disruptive to another memory task without growing more enjoyable. Four songs and a modest sample cannot establish that repetition reliably creates an earworm.

The listener is part of the measurement

A 2024 study built statistical models of how listeners perceive melodies. It recruited 96 nonmusicians and analyzed their ratings of short Western-classical melodies. Two participants with missing age data were excluded. The analyses kept the observations that listeners rated unfamiliar, so the results describe melodies the listeners said they did not know. The researchers computed properties of each melody and asked how listeners perceived its balance, contour (how the melody's line moves up and down), symmetry, complexity and unpredictability. Those perceptions mediated several links between the computed properties and liking. In these models, mediated means the link passed through a middle step, here the listener's perception. Differences between participants and between melodies explained a substantial share of the variation.

The models show why a formal feature, a property of the notes themselves, cannot be treated as though every listener receives it in the same way. They were correlational: they track what moves together and cannot show that a perceptual judgment causes liking. Short synthesized piano excerpts cannot establish rules that hold across cultures either.

A 2018 review asked which brain areas respond most reliably to familiar music. It was a systematic review, which gathers every study that meets rules set in advance, and it found 23 eligible neuroimaging (brain-scan) studies of adults. Eleven of them, with 212 participants, entered its activation-likelihood meta-analysis. That method pools brain-imaging studies and looks for locations where reported activity peaks cluster. The main analysis found no brain location where reported peaks consistently converged. That null result means studies with mixed designs did not support one simple familiarity center; it does not show that familiarity lacks a neural basis.

A 2026 set of four preregistered experiments asked which sounds people identify as music. Preregistered means the researchers registered their plans before collecting data. In the reported models, across 735 online Western participants, higher-level perceptual judgments separated music, not-music and ambiguous sounds more clearly than low-level acoustic features, the measurable properties of the sound wave. The clearest judgments were recognizing a musical instrument, detecting a melody and inferring that a sound was made on purpose. The task tested how people sort sounds into categories. Because it did not test liking, it cannot serve as a result about preference under repeated listening.

What replay count cannot say

A replay count, the number of times a track was played, records behavior and a chance to learn. It does not reveal whether the listener paid attention, chose the play, finished the work, remembered it later or valued it. Recommendation systems, habit, context, access, social cues and deliberate study can all produce another play.

A count also cannot establish artistic quality, durable preference, intention or universal taste. Those claims go beyond what the measure can support. Honest use of replay data begins by naming its denominator and setting: what the count is out of and where it came from. That means saying whose plays were counted and over what period. It also means naming the recommendation and access conditions, along with the rule that counted a play as finished.

A four-note listening card

This listening card is an editorial exercise. It is not a tested intervention, meaning a method shown in a study to change how people listen. The card has four notes: record your first response; name one detail you can hear for the first time; note one prediction or surprise; name what still resists you. It aims to keep liking, recognition, attention and prediction apart, so they do not blur into one vague impression, and it is not meant to force appreciation.

The card may be most useful when it records a null, a result in which a measure shows no change. A null is a legitimate observation and needs no ranking. Nothing new may become audible, or liking may stay where it was. A detail may also grow clearer while the work remains unappealing.

The honest measurement

Another play changes the listener's evidence, what they have heard and noticed so far, and leaves the work's rank where it was. The second hearing may make structure easier to perceive, show where a prediction was wrong, strengthen a memory trace or leave the response unchanged.

A question more specific than whether the work was liked keeps both the evidence and the listener in view: what changed between hearings, which measure changed and what remains unknown?

Sources

1. Liking and memory for musical stimuli as a function of exposure. American Psychological Association via PubMed. peer-reviewed primary research; three controlled exposure experiments. Published 2004-03-01; observed 2026-09-01.
2. Repeated Listening Increases the Liking for Music Regardless of Its Complexity. Frontiers in Neuroscience. peer-reviewed primary naturalistic repeated-listening experiment. Published 2017-03-31; observed 2026-09-01.
3. Neural Correlates of Familiarity in Music Listening: A Systematic Review and a Neuroimaging Meta-Analysis. Frontiers in Neuroscience. peer-reviewed systematic review and activation-likelihood meta-analysis. Published 2018-10-05; observed 2026-09-01.
4. Music synchronizes brainwaves across listeners with strong effects of repetition, familiarity and training. Scientific Reports. peer-reviewed primary EEG research. Published 2019-03-05; observed 2026-09-01.
5. The song that never ends: The effect of repeated exposure on the development of an earworm. Quarterly Journal of Experimental Psychology. peer-reviewed primary memory-interference experiment. Published 2023-01-09; observed 2026-09-01.
6. Perceptual representations mediate effects of stimulus properties on liking for music. Annals of the New York Academy of Sciences. peer-reviewed primary perceptual-modeling study. Published 2024-02-06; observed 2026-09-01.
7. Music is a distinct perceptual category with subjective grounds. Scientific Reports. peer-reviewed preregistered primary categorization research. Published 2026-05-27; observed 2026-09-01.

Corrections

No corrections recorded.