2nd/10_Wiki/Topics/AI_and_ML/Cognitive-Therapy-in-CBT.md

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-cognitive-therapy-cbt	Cognitive Therapy in CBT	10_Wiki/Topics	verified	self	CBT cognitive behavioral therapy cognitive distortion automatic thoughts ABC model Aaron Beck third wave	none	A	0.9	applied	psychology cbt cognitive-therapy mental-health cognitive-distortion mindfulness ai-mental-health		2026-05-10	pending	language applicable_to psychology Mental Health Apps Self-Help Cognitive Worker Wellness

Cognitive Therapy in CBT

매 한 줄

"매 situation X — 매 thought 의 변화 의 emotion + behavior". Aaron Beck (1960s) 의 founding. 매 가장 evidence-based therapy. 매 modern: 매 AI mental health (Woebot, Wysa) 의 base. 매 cognitive worker 의 self-care 의 leverage.

매 핵심

Beck's Cognitive Triad

1. Self: "I am worthless".
2. World: "World is unfair".
3. Future: "Things will never improve".

→ 매 depression 의 typical.

ABC Model (Ellis)

• A: Activating event.
• B: Belief (interpretation).
• C: Consequence (emotion + behavior).
• 매 B 의 매 modify 의 C 의 변화.

Cognitive Distortions (Beck / Burns)

1. All-or-Nothing: 매 흑백.
2. Overgeneralization: 매 single 의 always.
3. Mental Filter: 매 negative 만.
4. Disqualifying Positive: 매 positive 의 dismiss.
5. Jumping to Conclusions (mind-reading / fortune-telling).
6. Magnification / Minimization: 매 catastrophize / dismiss.
7. Emotional Reasoning: 매 feel = real.
8. Should statements: 매 imperative.
9. Labeling: 매 self / others 의 stigmatize.
10. Personalization: 매 self-blame.

CBT 의 process

1. Identify automatic thought.
2. Identify distortion.
3. Examine evidence.
4. Generate alternative thought.
5. Test in behavior (behavioral experiment).
6. Address core belief (deeper).

Third-wave CBT

• DBT (Dialectical Behavior Therapy): 매 emotion regulation, 매 acceptance + change.
• ACT (Acceptance & Commitment Therapy): 매 value-driven action.
• MBCT (Mindfulness-Based Cognitive Therapy): 매 meditation 결합.
• CFT (Compassion-Focused Therapy).
• Schema Therapy.

매 효과

• Depression: 매 strong evidence.
• Anxiety: 매 strong.
• PTSD: 매 effective (CBT-T, CPT).
• Insomnia: 매 CBT-I (gold standard).
• Chronic pain: 매 CBT for pain.
• OCD: 매 ERP.
• Eating disorders.

매 modern AI 의 응용

• Woebot: 매 first chatbot CBT.
• Wysa: 매 chatbot.
• Replika: 매 companion (caution).
• Mind tools / journaling apps.

Limit

• 매 nuance / empathy 의 limit.
• 매 crisis (suicide ideation) 의 X.
• 매 adjunct, 매 substitute X.
• 매 EU AI Act 의 high-risk.

매 cognitive worker 의 응용

• Imposter syndrome: 매 distortion 의 challenge.
• Catastrophizing on PR review.
• Burnout prevention.
• Negotiation / interview anxiety.

💻 패턴

Thought record (CBT 표준)

class ThoughtRecord:
    def log(self, situation, automatic_thought, emotion, intensity_0_100,
             evidence_for, evidence_against, alternative_thought, new_intensity):
        return {
            'situation': situation,
            'automatic_thought': automatic_thought,
            'distortion': self.identify_distortion(automatic_thought),
            'emotion': emotion,
            'before_intensity': intensity_0_100,
            'evidence_for': evidence_for,
            'evidence_against': evidence_against,
            'alternative_thought': alternative_thought,
            'after_intensity': new_intensity,
        }
    
    def identify_distortion(self, thought):
        patterns = {
            'all_or_nothing': ['always', 'never', 'completely', 'totally'],
            'overgeneralization': ['everyone', 'no one', 'every time'],
            'catastrophize': ['disaster', 'awful', 'terrible', 'ruined'],
            'should': ['should', 'must', 'have to'],
            'labeling': ['I am stupid', 'I am loser', 'I am fail'],
        }
        found = []
        for distortion, keywords in patterns.items():
            if any(kw in thought.lower() for kw in keywords):
                found.append(distortion)
        return found

Cognitive restructuring (LLM-assisted)

def cbt_restructure(automatic_thought, situation):
    prompt = f"""You are a CBT-informed cognitive coach (NOT a therapist).
    
Situation: {situation}
User's automatic thought: "{automatic_thought}"

Help by:
1. Identify the cognitive distortion (Beck/Burns categories).
2. Ask 3 evidence-examining questions.
3. Suggest 1 balanced alternative thought (NOT toxic positivity).

Important:
- NOT a substitute for professional therapy.
- If suicidal ideation or crisis is mentioned, suggest immediate professional help / crisis line.

Format: JSON."""
    return llm.generate(prompt)

Behavioral activation

def behavioral_activation_plan(low_mood_period):
    """매 depression 의 evidence-based intervention."""
    return {
        'pleasure_activities': [  # 매 매일 1+
            ('walk in nature', 'value: connection'),
            ('cook a meal', 'value: nourishment'),
            ('call friend', 'value: relationship'),
        ],
        'mastery_activities': [  # 매 매일 1+
            ('complete small task', 'value: accomplishment'),
            ('learn 10 min skill', 'value: growth'),
        ],
        'rule': 'Do BEFORE feeling motivated. Action precedes motivation.',
    }

Imposter syndrome (cognitive worker)

def imposter_check(thought):
    distortions = {
        'discount_positive': ["I just got lucky", "anyone could do it"],
        'overgeneralize_negative': ["I always mess up", "I'm always behind"],
        'mind_reading': ["they think I'm a fraud", "they'll find out"],
        'labeling': ["I'm an imposter", "I'm not real engineer"],
    }
    
    flags = []
    for d, examples in distortions.items():
        if any(e.lower() in thought.lower() for e in examples):
            flags.append(d)
    
    if flags:
        return {
            'flags': flags,
            'questions': [
                'What evidence (concrete projects, code shipped) supports your competence?',
                'Would you say this to a peer in the same situation?',
                'What is the difference between "feeling like" and "being"?',
            ],
        }
    return None

Behavioral experiment

class BehavioralExperiment:
    def design(self, prediction):
        return {
            'prediction': prediction,  # 매 e.g., "If I speak up, they'll laugh"
            'experiment': '...',  # 매 actually do it
            'expected_outcome': prediction,
            'observed_outcome': None,
            'learning': None,
        }
    
    def review(self, exp_id, observed):
        # 매 prediction vs reality
        ...

CBT-I (insomnia)

def cbt_i_protocol():
    return {
        'stimulus_control': [
            'Bed = sleep + sex only.',
            'If awake > 20 min, leave bed.',
            'Wake same time every day (even weekends).',
        ],
        'sleep_restriction': 'Restrict time-in-bed to actual sleep time + 30 min.',
        'cognitive_restructure': 'Challenge "I MUST sleep 8h or tomorrow is ruined" thought.',
        'sleep_hygiene': [
            'No caffeine after 2 PM.',
            'No screen 60 min before.',
            'Cool, dark, quiet room.',
        ],
        'paradoxical_intention': 'Try to STAY awake (reduces performance anxiety).',
    }

Mindfulness moment (MBCT)

def three_minute_breathing_space():
    return [
        ('1 min', 'Awareness — what is here? Thoughts, feelings, body.'),
        ('1 min', 'Gathering — focus on breath as anchor.'),
        ('1 min', 'Expanding — awareness to whole body, posture.'),
    ]

🤔 결정 기준

상황	Approach
Daily stress	Thought record + behavioral activation
Insomnia	CBT-I
Anxiety	CBT + exposure
Depression	CBT + behavioral activation + meds (severe)
Imposter syndrome	Distortion check + evidence
Burnout	DBT + boundary + recovery
Mild	Self-help (Burns "Feeling Good")
Moderate-severe	Therapist + adjunct

기본값: 매 self-help 의 mild + 매 therapist 의 moderate+. 매 AI 의 adjunct only.

🔗 Graph

• 부모: Psychology
• 변형: DBT
• 응용: Cognitive-Distortion · ERP
• 사상가: Aaron-Beck
• Adjacent: Cognitive Biases · Beliefs · Chronic-Pain-Management-Protocols · Brain-Derived Neurotrophic Factor (BDNF)

🤖 LLM 활용

언제: 매 self-help guidance. 매 product mental wellness feature. 매 cognitive worker self-care. 언제 X: 매 therapy substitute. 매 crisis (suicide / self-harm) 의 LLM 의 단독 의 X — 매 redirect to crisis line.

❌ 안티패턴

• AI 의 sole therapy: 매 EU AI Act 의 high-risk.
• Toxic positivity: 매 "just think positive" 의 invalidate.
• No professional referral: 매 crisis 의 miss.
• One-shot fix: 매 process 의 weeks.
• Cognitive only (no behavior): 매 behavioral activation 의 critical.
• Self-diagnosis: 매 specific disorder 의 일반 의 X.

🧪 검증 / 중복

• Verified (Beck, Ellis, Burns, Linehan, NICE guidelines).
• 신뢰도 A.
• Related: Cognitive Biases · Mindfulness · Brain-Derived Neurotrophic Factor (BDNF) · Chronic-Pain-Management-Protocols · Beliefs.

🕓 Changelog

날짜	변경
2026-05-08	Phase 1
2026-05-10	Manual cleanup — distortion catalog + ABC + 매 thought record / behavioral experiment / CBT-I code