🔄 Absolute Pitch May Not Be So Absolute

📋 Study Overview

🎯 Research Question

Is absolute pitch a fixed, immutable ability, or can AP possessors' pitch categories change with listening experience?

The paper opens by naming the view it is testing: "the general consensus is that once AP note categories are established in early life, they cannot be changed (Ward & Burns, 1982)" (p. 1496). The experiments asked whether the tuning of those categories could be systematically shifted by exposure to detuned music — and, if so, how far the shift travels.

🔬 Methodology

Two Experiments

Experiment 1 (N = 13 AP possessors)

• 7 female, 6 male; mean age 26 (SD 11). All self-reported AP, then verified as AP1 — the highest level — on the University of California, San Francisco online AP test
• Pre-listening: Violin tones from C4 to B4 inclusive, each in three versions — 33 cents flat, in tune, 33 cents sharp. (33 cents is about a third of a semitone, the amount by which an AP possessor's own tuning is reported to vary between days — though the paper cites that figure as evidence of stability, and never links it to the choice of increment; the comparison is ours. It is reported to drift from one day to the next; Abraham, 1901.) On every trial participants typed the note name and rated its intonation on a 1–3 scale (1 = out of tune, 3 = perfectly in tune). The paper does not mention feedback
• Exposure: Listened to Brahms's Symphony No. 1 in C minor. Across the first movement (~15 min) the recording was detuned downward at roughly 2 cents per minute — a rate below the just-noticeable difference for pitch change — until it sat 33 cents flat. Movements 2–4 (~30 min) were then heard at that stable, detuned pitch. To keep participants attending, they were asked to notate melodic themes
• Post-listening: The same task again, plus a block of piano notes — a timbre the symphony does not use — to test whether any shift crossed instruments
• Key detail: the rate of change was below the threshold for noticing; the final 33 cents is well above it. That is why the shift worked and no participant reported detecting it when questioned afterward
• What was measured: every result the paper reports is about the 1–3 intonation rating. No note-naming accuracy appears anywhere in the article, so "categories shifted" here means which version of a note sounds in tune changed — not that anyone started calling C something else

Experiment 2 (N = 14 AP possessors)

• 8 female, 6 male; mean age 24 (SD 7). All AP1 on the same UCSF test. The paper does not say whether anyone took part in both experiments
• Same three-part design, but the listening material was four pieces of phase music built on only five notes — G, A, B♭, C and D — detuned to 33 cents flat over the first 15 minutes and then held there for about 45 minutes
• The question: whether the shift stays on the five notes actually heard or spreads to the rest of the octave. Experiment 2 was designed to test “both the note and the timbre generalizability” — the new question is across notes, and timbre is tested again with a second instrument
• In the post-listening session a block of French horn notes replaced Experiment 1's piano block, testing a second timbre — one with a far more fluid tuning system than the piano

📊 Key Findings

1. Intonation judgments shifted with detuned exposure (Experiment 1)

Main finding: after as little as 45 minutes of music detuned to 33 cents flat, participants' intonation ratings moved in the direction of the detuning (a significant Session × Note Type interaction).

• Notes that were objectively 33 cents flat were now rated closer to "in tune" than before — and the in-tune notes lost ground
• The shift occurred even though no participant reported detecting the detuning when questioned afterward
• The effect was confined to the violin timbre heard during exposure. The piano block showed nothing: "the judgments for piano notes did not indicate that there was an assimilative context effect on note categorization" (p. 1498)
• What moved was the interaction, not the overall level. The main effect of session was not significant in either experiment (Exp. 1: F(1, 12) = 1.42, p > .25; Exp. 2: F(1, 13) = 2.27, p > .15) — participants did not simply become more lenient about tuning across the board

2. The shift generalized across notes — but not across timbres (Experiment 2)

Experiment 2 detuned music built on only five notes (G, A, B♭, C, D). Afterwards, flat versions of all the notes in the octave were judged more in tune, including notes that never appeared in the detuned music (Session × Note Type interaction F(2, 26) = 7.87, p < .01; the three-way interaction that would show an effect restricted to the notes actually heard was not significant, F(2, 26) = 0.54, p = .59). The authors read this as evidence that the musical scale is represented as a linked whole rather than as twelve independent frequencies.

Timbre was the boundary the effect did not cross. In Experiment 1 the piano notes showed no shift; in Experiment 2 the French horn notes showed none either, despite the horn's far more fluid tuning. The paper is explicit: exposure to altered notes "generalizes to other notes, but apparently not to different timbres" (p. 1500). The recalibration travels across the scale but stays inside the instrument that was heard.

3. Plasticity persists in adulthood

Adult AP possessors' tuning could be recalibrated, without their noticing, by as little as 45 minutes of detuned music — hard to square with a model in which the categories are frozen once a critical period closes. The authors' own reading is narrower than "AP is not developmental": note representations "may initially be established by early experience", but "cultural conventions of tuning standards are likely necessary to hold these representations in place" (p. 1498). Acquisition may well be early; what this study shows is that maintenance depends on continuing input.

💡 Main Conclusions

"Here we provide the first evidence that the note categories of adults with AP can change with listening experience. Participants with AP showed shifts in perception in direct accord with prior exposure to music detuned by a fraction of a semitone. This suggests that the apparent stability of AP categories is conferred not by early experience but rather by the cultural norms adopted for tuning music." — Hedger et al., 2013 (Abstract)

Key Implications:

• AP is not immutable: the tuning of pitch categories can adapt to recent listening, even in adulthood
• Established early, held in place by input: the paper does not deny an early developmental origin. It argues that a consistent cultural tuning standard is what keeps the categories where they are — which challenges the strict reading of "critical period" as locked, not the critical period itself
• Tuning is stored per timbre: the shift never crossed instruments. The authors connect this to the known phenomenon of "absolute piano" — listeners who pass AP tests only when the notes have a piano timbre — "as if each instrument voice maintains its own separate tuning" (p. 1501)
• What this does and does not show about training: every participant already had AP at the highest level tested (AP1). The study shows that existing note categories can be re-tuned by listening; it does not test, and does not claim, that adults without AP can build such categories from scratch
• Practical implications: the authors note that if music did not converge on the A = 440 Hz standard, AP representations "might change constantly" (p. 1498). Extrapolating that to historical tunings such as Baroque A = 415 Hz goes beyond what was tested — 415 Hz is roughly a semitone down, three times the shift used here, and large enough to be noticed consciously

⚠️ Limitations & Context

Study Limitations

• Duration of the effect: the shift did not decay across the ~45-minute post-listening session (F(3, 36) = 1.49, p > .20), and hearing perfectly in-tune notes during that session did not reset it — which the authors read as evidence that tuning is set by pitch relations within music rather than by absolute frequency. Beyond that session nothing was measured, and the authors themselves describe the manipulation as having "temporarily shifted note intonation judgments in listeners with AP" (p. 1501)
• Detuning magnitude: the manipulation was a single value — 33 cents, about a third of a semitone. The study does not establish a threshold: whether smaller deviations also shift categories, or whether larger ones would be consciously detected and therefore rejected, was not tested
• The timbre boundary is delimited, not explained: the authors leave open whether the effect might cross between more acoustically similar instruments — "It remains unclear whether the effect might generalize between instruments that have similar acoustic properties yet distinguishable qualities" (p. 1500)
• Small sample: 13 participants in Experiment 1 and 14 in Experiment 2, all of them AP1. The paper does not raise this itself
• Laboratory setting: a single controlled exposure may differ from real-world tuning variability

Historical Context (2013 vs 2020s)

🔗 Related Research

• Theoretical frame: the authors read their results as support for Wilson et al.'s (2012) model, in which AP requires not only the early learning of tone–label mappings but their consistency over time to stay in place
• Elsewhere on this site (editorial links, not citations in this paper): Miyazaki (1988) - identified precise AP, imprecise AP and non-possessors using microtonal stimuli
• Same first author, later work: Van Hedger et al. (2015) - auditory working memory predicts AP learning in adults
• Brain plasticity: Loui et al. (2011) - enhanced connectivity in AP possessors (but does training induce connectivity?)
• Adult training: Bongiovanni et al. (2023) trained 177 adults and found that learning transferred poorly across octaves (d′ = 0.23), concluding that training a single target frequency is an inappropriate measure of AP learning; Wong et al. (2025) closed six such loopholes and still found learning (13.9% → 31.7%, chance 8.3%)

📖 Access Full Study

📚 Full Citation