Gervain et al. 2013 - Valproate Reopens Critical-Period Learning of AP

📊 Study Design

Participants

N=24 healthy adult males
Median age: 23 (range 18-27)
Right-handed, English monolinguals
No absolute pitch
Little/no musical training (mean 2.4 years, started mean age 12 - well after critical period)
Mean IQ: 110

Method

Design: Randomized, double-blind, placebo-controlled crossover
2 treatment arms (15 days each)
VPA regimen: 500mg days 1-3, 1000mg days 4-14
Training: Days 8-14 (7 days, ~10 min/day)
Task: Associate 6 pitch classes with 6 proper names
Test: Day 15 - identify 18 synthesized piano tones
Washout: 2-4 weeks between arms

🔬 Training Protocol

What Participants Learned

6 pitch classes (e.g., C, D, E, F#/Gb, G#/Ab, A#/Bb) paired with 6 proper names (e.g., Sarah, David, Francine, Jimmy, Karen, Leo).

Why Proper Names Instead of Musical Note Names?

Made task equally novel for participants with/without musical training
Diverted attention from music theory aspect
Avoided interference with existing knowledge of actual note names

Training Videos (3 blocks)

Block 1: Each name shown with 3 examples (same pitch class, 3 consecutive octaves, lowest→highest)
Block 2: Same but octaves scrambled (not necessarily lowest→highest)
Block 3: Semi-random order, one tone at a time paired with name

Test Format

18 trials: Hear synthesized piano tone → identify associated name
Chance performance: 3/18 correct
Constraint: Successive tones separated by >1 octave interval
Stimuli: 500ms duration, 3750ms silence between tones

Training Compliance

Participants completed average 4.63/7 training sessions (SD=2.06, range 0-7). No significant correlation between number of sessions and performance (r=0.13, p=0.55).

📈 Results

First Treatment Arm (Most Reliable Data)

VPA Group (n=11)

5.09 / 18

Significantly above chance (p=0.002)

Placebo Group (n=12)

3.50 / 18

At chance level (p=0.21)

Significant effect of Condition: F(1,21) = 6.37, p = 0.02

Second Treatment Arm (Carry-Over Effects)

VPA Group (n=8)

2.75 / 18

At chance (p=0.33)

Placebo Group (n=10)

3.33 / 18

At chance (p=0.71)

No significant difference: F(1,16) = 0.452, ns

Crossover Analysis (n=18 who completed both arms)

Main effect of Order: F(1,15) = 6.06, p = 0.03 (higher scores for those who took VPA first)
Interaction Condition × Order: F(1,15) = 8.85, p = 0.009
VPA first group: Significant effect of treatment F(1,8) = 20.25, p = 0.002 (VPA > placebo)
Placebo first group: No difference F(1,7) = 1.13, p = 0.32

Interpretation: Carry-over effects from first arm likely interfered with learning in second arm (memory conflict between pitch classes/names used in each arm).

Error Distribution Analysis

Errors plotted as deviations from correct pitch (in whole tones):

VPA group: Peaked distribution at 0 (correct responses), errors appeared random
Placebo group: Flatter distribution, greater randomness
Implication: Learning was "absolute" - categories didn't represent nearness from one to another (true AP-like behavior)

💊 Valproate (VPA) Pharmacology

What is Valproate?

Commonly used anticonvulsant (epilepsy treatment)
Mood stabilizer for bipolar disorder
HDAC inhibitor (histone deacetylase inhibitor)
Modulates epigenome to promote neuroplasticity

How Does It Work?

Inhibits HDAC - an enzyme that acts as epigenetic "brake" on critical-period learning
Reopens chromatin structure → enables adult neuroplasticity
Previous animal studies: reopened visual cortex plasticity (amblyopia recovery), auditory learning in mice
This is first human study showing reopening of perceptual critical period

Blood Levels in Study

Mean concentration: 567 µmol/L (SD=165.53)
Active range: 350-700 µmol/L
1 participant below range (261 µmol/L)
3 participants above range (708, 732, 854 µmol/L)
8 participants within active range
No correlation between VPA blood levels and AP performance (r=0.36, p=0.28)

🧠 Cognitive & Mood Effects

No General Cognitive Enhancement

Extensive battery of mood and cognitive tests showed NO differential effect of VPA vs placebo on:

RAVLT (auditory verbal learning)
Stroop Task
Digit Span Test
Visual acuity, depth perception
Depression scores (BDI-II)
Most mood measures (VAS scales)

One Mood Effect: Mania Scale

Altman Self-Rating Mania Scale: Significant Time × Treatment interaction F(1,16) = 4.54, p = 0.049

Both conditions showed decrease in mania scores post-treatment
Decrease was greater in VPA condition (p=0.03 with Bonferroni correction)
Expected finding - VPA is clinically used as mood stabilizer for bipolar disorder
Cannot explain AP results: Would predict general cognitive improvement (not observed)

Critical Implication

VPA effect was specific to AP task - not due to general changes in mood or cognition. This specificity strengthens the interpretation that VPA reopened the critical period for the trained perceptual task.

🔍 Study Limitations & Methodological Considerations

1. No Baseline AP Test

Issue: Couldn't establish baseline performance (associations were learned during training)

Why not test early? No precise info on VPA time course, so training only started day 8 (full dose)

Concern addressed: If high-ability group assigned to VPA by chance, they should outperform in second arm too → didn't happen

2. Carry-Over Effects in Second Arm

Problem: Memory conflict between pitch classes/names from first vs second arm

Evidence: Order-dependent effect - only VPA-first group showed benefit

Future recommendation: Use RCT design (between-participants) to avoid carry-over

3. Reaction Times Not Measured

Issue: Participants had 4s to respond, no instruction to be fast

Why relevant: AP possessors recognize pitch faster than non-AP (Levitin & Rogers 2005)

Future work: Should measure RTs to distinguish true AP from quasi-AP (single tone reference)

4. Relatively Small Sample

First arm: n=11 VPA, n=12 placebo

Second arm: n=8 VPA, n=10 placebo (6 dropouts)

Context: Appropriate for VPA studies with healthy participants (comparable to Bell et al. 2005: n=12 per group)

Consequence: May explain lack of correlation between training compliance and performance

5. Duration of Effect Unknown

Question: How long does improved AP perception last after training ends?

Not tested: Study didn't retest participants days/weeks later

Future research needed: Longitudinal follow-up essential

6. Blinding Partially Compromised

17/18 participants (94%) correctly guessed which arm was VPA

Cues used: Mild side effects (drowsiness, nausea)

Why not a major concern:

Participants could guess but not be certain
Naïve to study hypothesis - couldn't voluntarily influence behavior in expected direction

7. Stimuli Choice: Piano Tones vs Pure Tones

Used: Synthesized piano tones (like Deutsch 2006)

Rationale: Instrument-generated sounds contain timbre cues beyond frequency

Trade-off:

✅ Pro: Intermediate/poor AP possessors perform much worse with pure tones (Lockhead & Byrd 1981, Bermudez & Zatorre 2009)
⚠️ Con: Extra cues may make task easier than "true" AP
Decision justified: Expected only moderate improvement, wanted to avoid floor effects

💡 Scientific Context & Implications

Why This Study is Revolutionary

First pharmacological intervention to change AP ability in humans
Proof-of-concept that critical-period plasticity can be chemically restored in adult brain
Translates animal research (HDAC inhibitors reopen plasticity in mice) to humans
Mechanism-driven: Based on molecular understanding of critical period closure (epigenetic "brakes")

Previous Adult AP Training Attempts (Why They Failed)

Prior studies (Meyer 1899, Mull 1925, Cuddy 1968, Brady 1970, Russo 2003) showed improvement only under restricted conditions:

Musically proficient participants (already better than chance pre-training)
Single tone training (not multiple pitch classes) → "quasi-AP" not true AP
Extensive training (weeks to months)

This study: Musically naïve participants, 6 pitch classes, only 7 days training → still got improvement with VPA

Epigenetic Basis of Critical Periods

How critical periods close:

Maturational processes + experience → neurophysiological/molecular changes
HDAC acts as epigenetic "brake" dampening neuroplasticity
Myelin-related signaling (Nogo receptor pathway) also closes critical periods

How VPA reopens them:

Inhibits HDAC → removes brake
Reopens chromatin structure → enables gene transcription changes
Mimics Nogo receptor deletion (Yang et al. 2012 - acoustic preference in mice)

Broader Implications for Neuroscience

Psychiatric drug mechanisms: VPA's therapeutic action may involve facilitating brain reorganization/rewiring via epigenome modulation
Behavioral paradigm: AP task could become standard test for assessing plasticity-inducing drugs
Cell-type specificity: Epigenetic treatments affect different cell types dramatically differently (future work needed)
MRI studies needed: Does HDAC inhibition induce hyperconnectivity (like natural AP possessors - Loui 2011)?

⚠️ Ethical & Safety Considerations

Why This Study is Controversial

1. Using Psychiatric Drugs for Skill Enhancement

VPA approved for epilepsy/bipolar disorder, NOT cognitive enhancement
Raises "pharmacological enhancement" debate (like Adderall for studying)
Ethical question: Is it appropriate to take drugs to learn AP?

2. Known Side Effects

In this study: Drowsiness, nausea (17/18 participants experienced)
Serious risks: Liver damage (hepatotoxicity), pancreatitis, weight gain
Teratogenic: Severe birth defects if taken during pregnancy (why only males studied)
Mood effects: Can affect mania/depression (though used therapeutically for this)

3. Risk-Benefit Analysis

Benefit: Modest improvement in pitch identification (5.09 vs 3.50 / 18)
Risk: Potentially serious medication side effects
Question: Do benefits justify risks for non-medical goal (learning AP)?

4. Study Safeguards Implemented

Only healthy males (avoid pregnancy risk)
Medical screening pre-enrollment (physician exam, blood tests)
Monitored hepatic enzymes, platelets, amylase, ammonia throughout
Mood/psychological state monitored (BDI-II, RNBI-24 to catch adverse effects)
No adverse effects found in any participant
6 dropouts: none due to side effects (scheduling conflicts, concussion from unrelated accident, etc.)

Authors' Perspective

"If further studies continue to reveal specificity of VPA to the AP task (or to tasks on which training or intervention is provided), critical information will have been garnered concerning when systemic drug treatments may safely be used to reopen neural plasticity in a specific, targeted way."

They emphasize targeted, specific plasticity reopening - not blanket cognitive enhancement.

🔮 Future Research Directions

Immediate Next Steps

RCT design: Between-participants to avoid carry-over effects
Measure reaction times: Distinguish true AP from quasi-AP
Longitudinal follow-up: How long do improvements last?
Larger sample sizes: Increase power, detect training compliance effects
Dose-response: Optimize VPA dosage for plasticity (current: based on epilepsy protocols)

Mechanistic Studies

fMRI during/after training: Does VPA induce hyperconnectivity like natural AP (Loui 2011)?
Cellular actions: Which cell types affected? (TK Hensch unpublished: VPA effects differ dramatically by cell type)
Myelin-related signaling: Does VPA affect myelinated long-range connections?
Epigenetic markers: Measure chromatin changes directly

Broader Applications

Other perceptual skills: Does VPA reopen critical periods for language, vision, etc?
Clinical applications: Stroke recovery, amblyopia treatment, language disorders
Alternative HDAC inhibitors: Safer compounds with fewer side effects?
Combination approaches: VPA + optimized training protocols (like Wong 2025)?
Biomarkers: Can we predict who will respond to VPA-enhanced training?

🎓 Historical Significance

Why 2013 Was a Turning Point

For decades, AP was considered an immutable trait with a fixed critical period (~age 6). This study fundamentally challenged that dogma by showing the critical period is not permanently closed - it's chemically locked, and the lock can be picked.

Before Gervain 2013

❌ "AP can only be acquired before age 6"
❌ "Adults cannot learn true AP"
❌ "Critical period closure is irreversible"
❌ "Training adults is futile"

After Gervain 2013

✅ "Critical period can be chemically reopened"
✅ "HDAC inhibitors enable adult plasticity"
✅ "Epigenetic brakes are reversible"
✅ "Adult training + VPA shows promise"

Legacy

Gervain 2013 didn't solve adult AP training - improvements were modest and side effects concerning - but it proved the barrier is biochemical, not insurmountable. It opened the door to a decade of research (Van Hedger 2019, Bongiovanni 2023, Wong 2025) showing adults CAN learn AP with optimized methods, even without drugs.

📚 Related Studies

💭 Critical Analysis

Strengths

First human pharmacological reopening of perceptual critical period
Rigorous double-blind, placebo-controlled, crossover design
Mechanism-driven (based on HDAC/epigenetics research in animals)
Specific effect (no general cognitive enhancement - rules out confounds)
Extensive safety monitoring (no adverse events)
Thoughtful stimulus design (proper names, piano tones, semi-randomization)
Thorough analysis of error distributions (true absolute learning, not relational)

Weaknesses

Small sample (n=11-12 first arm, further reduced in second arm)
Carry-over effects compromised second arm (order-dependent results)
No baseline AP test (though concern addressed by crossover analysis)
Blinding partially compromised (94% guessed correctly via side effects)
No duration-of-effect data (how long do improvements last?)
No reaction time measurement (can't distinguish AP speed from accuracy)
Half as many trials as Deutsch 2006 (18 vs 36 - may reduce reliability)
Only males tested (limits generalizability, though justified by teratogenic risk)

Impact & Influence

Citations: Highly cited in neuroplasticity literature (exact count varies by database)

Field impact:

Shifted AP research from "impossible in adults" to "how can we optimize it?"
Inspired behavioral training studies (Van Hedger 2019, Bongiovanni 2023, Wong 2025)
Contributed to broader "critical period reopening" research (vision, language)
Raised ethical debates about pharmacological enhancement

Clinical translation: Potential (but unrealized) for stroke recovery, sensory disorders, learning disabilities

🔗 Access & Resources

📄 Full Text

Open Access: Yes (CC BY license)

📊 Data & Methods

DOI: 10.3389/fnsys.2013.00102
Study registration: University of British Columbia ethics + Health Canada approval
Conflicts: None declared
Funding: HFSP, Michael Smith Foundation, Quinn Fellowship

🎯 Key Finding