Files
2nd/10_Wiki/Topics/AI_and_ML/Neuroplasticity.md
T
koriweb d8a80f6272 chore(wiki): dangling 링크 canonical 정규화 (768파일/1200건)
이름만 다른(표기 변형) [[위키링크]]를 대상 문서의 canonical 제목으로 치환해
끊겼던 1,200개 링크를 연결. 제목/파일명 정규화 일치만 적용하고 별칭 매칭은
과병합 위험으로 제외(애매성 가드). 원본은 _link_reconcile_backup/ 에 백업.
도구: Datacollect/scripts/link_reconcile_apply.mjs

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-08 12:24:15 +09:00

4.7 KiB

id, title, category, status, canonical_id, aliases, duplicate_of, source_trust_level, confidence_score, verification_status, tags, raw_sources, last_reinforced, github_commit, tech_stack
id title category status canonical_id aliases duplicate_of source_trust_level confidence_score verification_status tags raw_sources last_reinforced github_commit tech_stack
wiki-2026-0508-neuroplasticity Neuroplasticity 10_Wiki/Topics verified self
Neuroplasticity
Brain Plasticity
Synaptic Plasticity
none A 0.9 applied
neuroscience
plasticity
hebbian
ltp
critical-period
learning
2026-05-10 pending
language framework
python brian2-pytorch

Neuroplasticity

매 한 줄

  • 신경가소성은 시냅스·회로 수준에서 경험에 따라 신경계 구조·기능이 변화하는 능력이며, LTP/LTD가 분자 기반이다.

매 핵심

  • Hebbian rule: "fire together, wire together" — pre/post synaptic 동시 발화 시 시냅스 강화.
  • LTP/LTD: NMDA receptor 매개 Ca²⁺ → CaMKII (LTP, 강화) vs calcineurin (LTD, 약화). hippocampus CA1, cortex L2/3에서 잘 연구됨.
  • Critical period: 시각피질 monocular deprivation 효과는 어린 시기 강함. parvalbumin GABA 성숙이 닫힘 trigger. 성인기 reopen에 chondroitinase, fluoxetine, dark exposure.
  • Adult plasticity: motor learning, taxi driver hippocampus 부피 증가, training-induced cortical map remodeling.
  • AI 연결: STDP(spike-timing dependent plasticity)는 SNN 학습 규칙, BCNN의 Hebbian feature learning에 영감.

💻 패턴

# STDP weight update rule
import numpy as np

def stdp(dt, w, A_plus=0.01, A_minus=0.012, tau_plus=20e-3, tau_minus=20e-3, w_max=1.0):
    if dt > 0:  # post after pre → LTP
        dw = A_plus * np.exp(-dt / tau_plus)
    else:
        dw = -A_minus * np.exp(dt / tau_minus)
    return np.clip(w + dw, 0, w_max)
# Hebbian learning in linear neuron (Oja's rule, normalized)
import numpy as np
def oja_update(w, x, y, lr=0.01):
    return w + lr * (y * x - y ** 2 * w)
# LTP induction: theta-burst stimulation pattern generator
def theta_burst(duration_s=2.0, burst_hz=5, pulses_per_burst=4, intra_hz=100):
    times = []
    t = 0.0
    while t < duration_s:
        for k in range(pulses_per_burst):
            times.append(t + k / intra_hz)
        t += 1 / burst_hz
    return times
# Synaptic scaling (homeostatic plasticity)
import numpy as np
def synaptic_scaling(W, target_rate, actual_rate, tau=1000.0):
    factor = target_rate / (actual_rate + 1e-6)
    return W * (1 + (factor - 1) / tau)
# Brian2: STDP synapse simulation
from brian2 import *
G = NeuronGroup(2, "dv/dt = -v/(10*ms) : 1", threshold="v>1", reset="v=0")
S = Synapses(G, G,
             """w : 1
                dApre/dt = -Apre/(20*ms) : 1 (event-driven)
                dApost/dt = -Apost/(20*ms) : 1 (event-driven)""",
             on_pre="""v_post += w
                       Apre += 0.01
                       w = clip(w + Apost, 0, 1)""",
             on_post="""Apost -= 0.012
                        w = clip(w + Apre, 0, 1)""")
S.connect(i=0, j=1)
# Cortical map plasticity index from receptive field overlap
import numpy as np
def map_plasticity(rf_pre, rf_post):
    overlap = np.minimum(rf_pre, rf_post).sum() / np.maximum(rf_pre, rf_post).sum()
    return 1 - overlap  # higher = more remodeling
# BDNF-dependent plasticity: serum BDNF as biomarker proxy
def plasticity_score(bdnf_ng_ml, exercise_min_week, sleep_hr):
    # toy index, not clinical
    return 0.4 * bdnf_ng_ml / 30 + 0.3 * min(exercise_min_week, 300) / 300 + 0.3 * min(sleep_hr, 8) / 8

매 결정 기준

  • 개입 시기: critical period 진행 중이면 sensory restoration(eg. amblyopia patching) 효과 큼.
  • 약리 보조: 성인 plasticity 재개에 SSRI(fluoxetine), tDCS, aerobic exercise(BDNF↑).
  • 학습 설계: spaced repetition(LTP consolidation), sleep 보장(systems consolidation).
  • 연구 모델: in vitro slice → LTP/LTD 측정. in vivo two-photon → spine turnover.

🔗 Graph

🤖 LLM 활용

  • 학습 곡선 데이터에서 plasticity phase 추정(initial vs consolidation).
  • 논문 요약: LTP 분자 경로 다이어그램 생성.
  • 실험 설계 review(빠진 control 식별).

안티패턴

  • "성인 뇌는 변하지 않는다" 신화 인용.
  • STDP를 단순 Hebbian과 동일시(타이밍 차이 핵심).
  • BDNF 보조제 임의 권고(증거 부족).

🧪 검증

  • LTP: fEPSP slope baseline 대비 +30% 30분 이상.
  • 행동: motor task 학습률, cortical map fMRI pre/post.

🕓 Changelog

  • 2026-05-08 Phase 1: 초안 자동 생성.
  • 2026-05-10 Manual cleanup: 본문 보강, STDP/Oja/Brian2 코드 추가, critical period reopen 약리 반영.