NEUROIMAGING 2011

🧠 Enhanced Cortical Connectivity in Absolute Pitch Musicians: A Model for Local Hyperconnectivity

Psyche Loui, Hui C. Charles Li, Anja Hohmann, Gottfried Schlaug

Journal of Cognitive Neuroscience, April 2011; 23(4):1015–1026

📅 Published: April 2011 👥 N=24 musicians (12 AP, 12 non-AP) 🔬 DTI + Tractography 📖 Open Access (PMC3012137)

🎯 Key Finding

Evidence that white matter connectivity — not just gray matter volume — differs in absolute pitch musicians. Using diffusion tensor imaging (DTI), Loui et al. found hyperconnectivity in bilateral superior temporal lobe structures linked to AP. Crucially, the volume of tracts connecting left posterior STG to left posterior MTG predicted AP performance accuracy (rs = −.52, p = .01), suggesting AP relies on efficient mapping from perception to categorization in the left hemisphere — the authors describe the pSTG→pMTG pathway as carrying information "from a fine-grained, controlled perceptual representation to an unconscious and automatic category-based representation."

⚠️ Historical Study (2011): This study found structural correlates of AP — but it is not an "innate/hardwired" paper, though it is sometimes read that way. Its authors conclude that "a combination of predispositions and environmental factors lead to the development of superb pitch categorization ability," and that their results “appear to support the hypothesis” that early musical training is “necessary but not sufficient” for AP. They split it: hyperconnectivity in both temporal lobes “may be one of the prerequisites”, while exposure “may influence and shape the hyperconnectivity in the left more than in the right hemisphere”. What later training studies add is something Loui 2011 never tested — whether adults can learn functional AP: Wong et al. (2025) found measurable adult learning, while Bongiovanni et al. (2023) found it transfers poorly across octaves.

📊 Study Design

Participants

  • N=24 musicians from Greater Boston area
  • 12 AP possessors (self-reported, verified by test)
  • 12 non-AP controls (matched musicians)
  • Groups matched for: age, gender, handedness, ethnicity, IQ, age of onset and years of musical training
  • All professional or amateur musicians
  • IQ measured via Shipley-Hartford Retreat Test

AP Verification Test

  • 52 sine wave tones (established test: Keenan et al., 2001; Zatorre & Beckett, 1989; Ward & Burns, 1982)
  • Range: F#3 (370 Hz) to F#4 (739.97 Hz)
  • Duration: 500ms per tone (50ms rise/decay)
  • Intertone interval: 2 seconds
  • Task: Label each pitch by letter name (+ accidentals)
  • Scoring: Correct if within ±1 semitone (chance = 3/11 = 27%)

🔬 Brain Imaging Protocol

Diffusion Tensor Imaging (DTI)

  • Scanner: 3-Tesla General Electric
  • Anatomical: T1-weighted MPRAGE (0.93 × 0.93 × 1.5 mm voxels)
  • DTI: 25 noncollinear directions + 1 baseline (b=1000 sec/mm²)
  • Analysis: Fractional anisotropy (FA) — the paper defines it only as “the degree of directional preference of water diffusion”, and lists four competing sources it cannot distinguish between
  • Software: MedINRIA 1.7 (tractography: FA threshold 0.2, smoothness 0.2, fibers >10mm)

Regions of Interest (ROIs)

  • Primary: Posterior superior temporal gyrus (pSTG) and posterior middle temporal gyrus (pMTG) — drawn bilaterally
  • Control: Corticospinal tracts (motor-related, unrelated to auditory processing)
  • Reliability: ROIs drawn by one coder, verified by a second (both blind to group). Intercoder reliability: 98% (volumes), 89% (tract locations)
  • ROI volumes: pSTG mean 250 mm³, pMTG mean 285 mm³ — no group differences

📈 Results

Behavioral Performance

AP Group (n=12)
97%
Range: 92–100% correct
Non-AP Group (n=12)
36%*
Range: 19–77% (chance = 27%)

Highly significant difference: t(22) = 5.8, p < .001

* The paper is internally inconsistent here: its body text reports 36% for the control group, while its own Table 1 reports 41% (same range, 19–77%). We quote the body text and flag the discrepancy rather than silently picking one.

1. Higher FA Values in AP Musicians

Aggregated over the four ROIs (bilateral pSTG + pMTG), the AP group showed slightly higher fractional anisotropy. No individual ROI reached significance on its own.

  • Main effect of group: F(1, 88) = 4.0, p < .05 — on a difference of 0.02 in FA
  • AP group: Mean FA = 0.28 (var = 0.002)
  • Non-AP group: Mean FA = 0.26 (var = 0.002)
  • In the paper's words: "individual t test contrasts did not reveal significant between-group differences in FA values of specific regions," and there was no group × ROI interaction (F = 0.34, p = .80) — the authors describe this as all ROIs contributing "small effects" equally, not as four separate findings
  • The strongest effect in this ANOVA was regional, and it points right: F(1, 88) = 5.7, p = .001, with the highest FA in right pMTG (0.30) and right pSTG (0.29) against 0.25 in both left ROIs. The leftward story of AP in this paper is about tract volume, not regional FA

2. Larger pSTG-to-pMTG Tract Volume in AP

The central finding: White matter tracts connecting pSTG to pMTG were significantly larger in AP musicians.

  • Main effect of group: F(1, 44) = 16.6, p < .001
  • Left hemisphere: t(22) = 3.8, p = .001
  • Right hemisphere: t(22) = 2.3, p = .03
  • Total (L+R): t(22) = 3.9, p < .001
  • Fiber count: Also higher in AP, t(22) = 2.4, p < .05 (significant in left hemisphere only)
  • Tract FA: no difference between groups. The difference lies in the volume and fibre count of the tract, not in the average FA of the reconstructed tract

3. Left Tract Volume Predicts AP Accuracy

Key correlation: Left pSTG-to-pMTG tract volume correlated with AP performance as a continuous variable.

  • Left tract: rs = −.52, p = .01 (larger volume → smaller pitch deviation = better AP)
  • Right tract: rs = −.14, p = .53 (not significant)
  • Implication: left pSTG–pMTG tract volume is a neural correlate of AP ability — the paper's own article, and it stresses that only "a small fraction of all possible tracts in the human brain" was investigated

4. AP1 vs AP2 Subcategories

AP possessors were split into two performance-based subcategories by a median split at 97%. The AP-1/AP-2 labels come from Baharloo et al. (1998), where they are score bands (AP-2 combines a pure-tone and a piano-tone cut-off) on one test — not two different kinds of absolute pitch:

  • AP1 (n=6): Mean 99.7% — near-perfect, robust to scoring method
  • AP2 (n=6): Mean 94.6% — drops to 77.7% under the secondary scoring, which gives half credit for a one-semitone error. Under the same scheme AP1 barely moves (98.4%, range 95.2–100%)
  • Left tract volume varies with AP category: F(2, 21) = 10.2, p < .001, surviving Bonferroni correction. The group means fall in the order AP1 > AP2 > non-AP, but the paper reports the omnibus ANOVA — it does not report pairwise contrasts, so the ordering is a pattern in the means, not three established differences
  • Right tract: F(2, 21) = 2.1, p = .14 (not significant)

5. Control Analysis: Corticospinal Tracts

To rule out global connectivity differences, the researchers tested motor-related tracts:

  • No group difference: F(1, 44) = 0.02, p = .87
  • No hemisphere effect, no interaction
  • Reading: the pattern looks specific to auditory/temporal regions rather than whole-brain — though the authors call it “suggestive” and say “future research is needed to define the specificity and focality”

6. Early Musical Training Effect

  • Early onset musicians had larger left tract volume: F(1, 22) = 5.23, p = .03
  • Right tract: no difference (F = 0.44, p = .52)
  • No effect of tone language or Asian ethnicity on tract morphology
  • What actually rules out early training as the whole story: a partial correlation between AP category and left tract volume stayed significant after partialling out age of musical onset and tone language — r = −.67, p = .001. Group matching alone would not carry this, since the paper found an onset effect within the matched sample (the line above). In the authors' words, the tract differences, once age of musical onset is partialled out, "suggests that the AP morphology can exist over and above early musical training"

💡 Main Conclusions

"Using diffusion tensor imaging and tractography, we observed hyperconnectivity in bilateral superior temporal lobe structures linked to AP possession. Furthermore, volume of tracts connecting left superior temporal gyrus to left middle temporal gyrus predicted AP performance." — Loui et al., 2011 (Abstract)

Key Implications:

  • White matter matters: evidence that connectivity — not just gray matter volume — differs in AP. The authors claim no priority; they write that such regional hyperconnectivity “has not been described in normal healthy individuals with AP”
  • Left hemisphere dominance: Only left pSTG→pMTG tract predicted AP performance, consistent with left-lateralized pitch labeling
  • Graded differences within AP: the two subgroups show graded connectivity differences. Loui et al. are careful here: they write that “it remains to be seen whether the phenotype of AP is categorical or continuous, but our results suggest that both views have merit”
  • Specificity: Corticospinal tracts showed no differences — hyperconnectivity is local to auditory regions
  • Model for hyperconnectivity: AP proposed as a model for understanding enhanced local connectivity in conditions like autism, synesthesia, and Williams syndrome

🧠 Theoretical Framework: Local Hyperconnectivity

AP as a Model for Hyperconnectivity

The authors proposed AP as an ideal model to study hyperconnectivity in the human brain:

  • Quantifiable behavior: AP accuracy provides a continuous behavioral measure that correlates with connectivity
  • Focal brain region: Effects are localized to temporal lobe auditory/association cortices
  • Link to developmental disorders: AP has elevated incidence in autism, Williams syndrome, and synesthesia — all associated with abnormal local connectivity
  • Three-node network, as a proposal: the authors write that the pSTG–pMTG connection measured here, “if identified and linked to musical behavior, would represent the completion of a three-node network between STG, MTG, and IFG”. The frontal connections were measured in a different study of theirs — Loui, Alsop & Schlaug 2009 — which is about musical disorders, not AP
📖 What Has Changed Since 2011:
This study found structural brain differences associated with AP — without endorsing a purely innate model. The authors' own conclusion is that early musical training is "necessary but not sufficient," and that the hyperconnectivity is likely shaped by early exposure: "early exposure to pitched information may have elicited rapid development of connections between auditory perception and association cortices." Recent adult training studies (Wong et al. 2025, Bongiovanni et al. 2023) measure adult learning without producing AP as the field defines it: Wong reports 13.9%→31.7% with 2 of 12 reaching AP-level performance, and Bongiovanni finds that single-note learning generalises poorly across octaves. This raises questions: Does adult AP training also induce connectivity changes? Or can functional equivalence be achieved via alternative neural pathways? Gervain et al. 2013 reported suggestive, unreplicated evidence that showed that HDAC inhibitors can reopen plasticity — future DTI studies of adult AP learners would be transformative.

🔍 Study Limitations

Items 3, 4 and 5 are caveats the authors state in the paper. Items 1, 2 and 6 are ours — reasonable, but not things this paper says about itself.

1. Correlation, Not Causation

Cannot determine if hyperconnectivity causes AP or results from AP-related brain use. Cross-sectional design (no longitudinal data).

2. Small Sample Size

N=24 (12 per group) limits statistical power for subgroup analyses. AP1/AP2 split yields only n=6 per subcategory.

3. Limited Tract Investigation

Only pSTG→pMTG and corticospinal tracts were analyzed. Other tracts (e.g., arcuate fasciculus, corpus callosum) may also differ. Authors acknowledge "we have only investigated a small fraction of all possible tracts."

4. FA Interpretation Ambiguity

Higher FA could reflect increased intra-axonal viscosity, increased myelination, decreased crossing fibers, OR increased fiber straightness. The imaging method cannot distinguish which factor drives the differences.

5. AP2 vs Heightened Tonal Memory

The distinction between "true AP" (AP1) and "heightened tonal memory" (AP2) remains unclear. AP2 subjects scored 94.6% overall but dropped to 77.7% when a one-semitone error earns only half credit, while AP1 held at 98.4% — are they truly AP possessors or exceptionally good relative pitch users?

6. Self-Selection Bias

AP possessors may differ in unmeasured ways beyond connectivity (motivation, practice habits, specific training experiences). Groups were matched for observable variables but not all confounds can be controlled.

💭 Critical Analysis

Strengths

  • Targeted white-matter connectivity in AP with DTI. The authors claim no priority; what they write is that such regional hyperconnectivity "has not been described in normal healthy individuals with AP"
  • Groups carefully matched on 7 variables (age, gender, handedness, ethnicity, IQ, onset + years of training)
  • Blind ROI analysis with high intercoder reliability (98%/89%)
  • Control analysis (corticospinal tracts) confirms specificity of temporal lobe findings
  • Multiple analysis approaches: categorical (AP vs non-AP), subcategorical (AP1/AP2), and continuous (Spearman correlation)
  • Proposed testable theoretical model (local hyperconnectivity) with broader implications
  • Applied the AP-1/AP-2 terminology of Baharloo et al. (1998) to a neuroimaging sample, via a median split at 97%

Weaknesses

  • Small sample (n=12 per group, n=6 per AP subcategory)
  • Cross-sectional — cannot establish causality or track developmental trajectory
  • Limited tract analysis (only 2 regions tested out of many possibilities)
  • FA is a composite measure — cannot identify specific biological mechanism
  • Sine tones only — a deliberate control, not an oversight: the authors chose pure tones "to avoid biasing our results in favor of participants who have training in the instrument chosen as the testing timbre." It does limit generalisation to real-world AP use, which involves instrument timbres
  • Self-reported AP status as initial screening (though verified by behavioral test)
  • No longitudinal component — cannot track whether connectivity changes with training

Impact & Influence

Field impact:

  • Extended neuroimaging AP research from gray matter (Schlaug 1995, Zatorre 1998) to white matter connectivity
  • Introduced "local hyperconnectivity" model for AP, linking it to autism/synesthesia research
  • Cited by Gervain et al. (2013), which asks whether HDAC inhibition induces similar hyperconnectivity
  • Carried the AP-1/AP-2 distinction — originally defined as score bands by Baharloo et al. (1998), not as two kinds of AP — into the brain-imaging literature
  • Referenced by Gervain et al. 2013 as motivation for asking whether pharmacological intervention could induce similar connectivity

📚 Related Studies

NEUROIMAGING

Schlaug et al. 1995 - Planum Temporale Asymmetry

Exaggerated leftward planum temporale asymmetry in musicians with AP — in that sample driven mainly by a smaller right PT, not a larger left one. A grey-matter finding that Loui 2011 extends to white matter
NEUROIMAGING

Zatorre et al. 1998 - PET Imaging of AP

Functional imaging showing differential brain activation in AP vs RP during pitch tasks — complementary functional data to Loui's structural findings
CLINICAL TRIAL

Gervain et al. 2013 - Valproate Reopens Critical Period

Pharmacological reopening of AP plasticity — explicitly asks whether HDAC inhibitors could induce hyperconnectivity like Loui 2011 found
TRAINING STUDY

Wong et al. 2025 - 13.9% → 31.7% After 8 Weeks

2 of 12 adults reached AP-level performance after training — the neuroplasticity question raised by Loui remains: does training induce connectivity changes?

🔗 Access & Resources

📄 Full Text

Open Access: Yes (NIH Public Access)

📊 Study Details

  • DOI: 10.1162/jocn.2010.21500
  • Institution: Beth Israel Deaconess Medical Center, Harvard Medical School
  • Funding: NIDCD RO1 DC009823-01, Grammy Foundation
  • IRB: Beth Israel Deaconess Medical Center

📚 Full Citation

Loui, P., Li, H. C. C., Hohmann, A., & Schlaug, G. (2011). Enhanced cortical connectivity in absolute pitch musicians: A model for local hyperconnectivity. Journal of Cognitive Neuroscience, 23(4), 1015–1026. https://doi.org/10.1162/jocn.2010.21500