cognitive-biases

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Cognitive Biases - Psychology for Product Design

认知偏差 - 产品设计心理学

Understanding psychological patterns that influence human decision-making, first systematically studied by Kahneman and Tversky. Essential for creating user experiences that work with human psychology.
了解影响人类决策的心理模式,这一领域最早由卡尼曼和特沃斯基进行系统性研究。这对于打造契合人类心理的用户体验至关重要。

When to Use This Skill

何时使用这项技能

  • Designing user onboarding flows
  • Improving conversion rates ethically
  • Analyzing why users behave unexpectedly
  • Reviewing designs for dark patterns
  • Planning pricing and positioning strategies
  • Understanding decision-making in user research
  • 设计用户引导流程
  • 以符合伦理的方式提升转化率
  • 分析用户异常行为的原因
  • 审查设计中的暗黑模式
  • 规划定价与定位策略
  • 在用户研究中理解决策机制

Foundation: Dual-Process Theory

基础:双加工理论

┌─────────────────────────────────────────────────────────────────┐
│                     HUMAN DECISION-MAKING                        │
├────────────────────────────┬────────────────────────────────────┤
│       SYSTEM 1 (95%)       │          SYSTEM 2 (5%)             │
├────────────────────────────┼────────────────────────────────────┤
│ Fast                       │ Slow                               │
│ Automatic                  │ Deliberate                         │
│ Intuitive                  │ Analytical                         │
│ Unconscious                │ Conscious                          │
│ Associative                │ Logical                            │
│ Low effort                 │ High effort                        │
│ Emotional                  │ Rational                           │
├────────────────────────────┼────────────────────────────────────┤
│ "Feels right"              │ "Let me think about this"          │
└────────────────────────────┴────────────────────────────────────┘

Most user interactions happen through System 1.
Design for intuition, not just logic.
┌─────────────────────────────────────────────────────────────────┐
│                     HUMAN DECISION-MAKING                        │
├────────────────────────────┬────────────────────────────────────┤
│       SYSTEM 1 (95%)       │          SYSTEM 2 (5%)             │
├────────────────────────────┼────────────────────────────────────┤
│ Fast                       │ Slow                               │
│ Automatic                  │ Deliberate                         │
│ Intuitive                  │ Analytical                         │
│ Unconscious                │ Conscious                          │
│ Associative                │ Logical                            │
│ Low effort                 │ High effort                        │
│ Emotional                  │ Rational                           │
├────────────────────────────┼────────────────────────────────────┤
│ "Feels right"              │ "Let me think about this"          │
└────────────────────────────┴────────────────────────────────────┘

Most user interactions happen through System 1.
Design for intuition, not just logic.

Core Cognitive Biases

核心认知偏差

1. Anchoring Bias

1. 锚定偏差

What it is: The brain latches onto the first piece of information as a reference point for all subsequent decisions.
Pricing Example:

❌ Without anchor:
   "Pro plan: $49/month"
   User thinks: "Is that expensive?"

✅ With anchor:
   "Enterprise: $199/month" (shown first)
   "Pro plan: $49/month"
   User thinks: "That's a great deal!"
Product applications:
  • Show premium/enterprise tier first in pricing tables
  • Display original price crossed out before sale price
  • Set high initial expectations, then exceed them
定义: 大脑会将接收到的第一条信息作为后续所有决策的参考点。
Pricing Example:

❌ Without anchor:
   "Pro plan: $49/month"
   User thinks: "Is that expensive?"

✅ With anchor:
   "Enterprise: $199/month" (shown first)
   "Pro plan: $49/month"
   User thinks: "That's a great deal!"
产品应用场景:
  • 在定价表格中优先展示高级/企业套餐
  • 在售价前显示划掉的原价
  • 设定较高的初始预期,然后超出用户预期

2. Loss Aversion

2. 损失厌恶

What it is: Humans feel losses 2x more intensely than equivalent gains.
Framing comparison:

Gain frame (weaker):    "Save $100 with annual billing"
Loss frame (stronger):  "You're losing $100 by paying monthly"

Progress frame:
Weaker:  "Complete setup to unlock features"
Stronger: "Don't lose your progress - 80% complete"
Product applications:
  • Free trials that create ownership feeling
  • Progress indicators showing what users might lose
  • "Save" vs "Spend" framing in messaging
定义: 人类对损失的感受强度是同等收益的2倍。
Framing comparison:

Gain frame (weaker):    "Save $100 with annual billing"
Loss frame (stronger):  "You're losing $100 by paying monthly"

Progress frame:
Weaker:  "Complete setup to unlock features"
Stronger: "Don't lose your progress - 80% complete"
产品应用场景:
  • 能让用户产生拥有感的免费试用
  • 显示用户可能失去的内容的进度指示器
  • 消息文案中使用“节省” vs “花费”的不同表述

3. Availability Bias

3. 可得性偏差

What it is: We overestimate the likelihood of events we can easily recall.
Making success feel common:

"Join 50,000+ developers"        → Success is common
"Featured in TechCrunch"         → Credibility by association
"Sarah from NYC just signed up"  → Real-time social proof
"5 people viewing this now"      → Popularity signal
Product applications:
  • Social proof and testimonials prominently displayed
  • Recent activity feeds that influence behavior
  • Success stories that make outcomes feel achievable
定义: 我们会高估那些容易回忆起的事件发生的可能性。
Making success feel common:

"Join 50,000+ developers"        → Success is common
"Featured in TechCrunch"         → Credibility by association
"Sarah from NYC just signed up"  → Real-time social proof
"5 people viewing this now"      → Popularity signal
产品应用场景:
  • 突出展示社交证明和用户评价
  • 影响用户行为的近期动态信息流
  • 让成果看起来触手可及的成功案例

4. Confirmation Bias

4. 确认偏差

What it is: We seek information confirming existing beliefs and ignore contradictory evidence.
Personalization flow:

User selects: "I'm a developer"
Show: Developer-focused features
Hide: Marketing automation features
User thinks: "This product gets me"
Product applications:
  • Personalized onboarding based on user type
  • Customizable dashboards reflecting preferences
  • Content recommendations aligned with interests
定义: 我们会寻找能证实现有信念的信息,而忽略矛盾的证据。
Personalization flow:

User selects: "I'm a developer"
Show: Developer-focused features
Hide: Marketing automation features
User thinks: "This product gets me"
产品应用场景:
  • 基于用户类型的个性化引导流程
  • 反映用户偏好的可自定义仪表盘
  • 符合用户兴趣的内容推荐

5. Planning Fallacy

5. 规划谬误

What it is: We consistently underestimate how long tasks will take.
Setting realistic expectations:

❌ "Quick setup"           → User expects 1 min, takes 10
✅ "10-minute setup"       → User expects 10, finishes in 8

Progress that manages expectations:
┌────────────────────────────────────┐
│ Step 2 of 5 · About 4 minutes left │
│ ████████░░░░░░░░░░░░░░ 40%         │
└────────────────────────────────────┘
Product applications:
  • Realistic time estimates for user tasks
  • Progress indicators with time remaining
  • Break complex tasks into visible steps
定义: 我们会持续低估完成任务所需的时间。
Setting realistic expectations:

❌ "Quick setup"           → User expects 1 min, takes 10
✅ "10-minute setup"       → User expects 10, finishes in 8

Progress that manages expectations:
┌────────────────────────────────────┐
│ Step 2 of 5 · About 4 minutes left │
│ ████████░░░░░░░░░░░░░░ 40%         │
└────────────────────────────────────┘
产品应用场景:
  • 为用户任务提供真实的时间预估
  • 显示剩余时间的进度指示器
  • 将复杂任务拆分为可见的步骤

6. Framing Effect

6. 框架效应

What it is: How information is presented changes decisions, even when underlying data is identical.
Same data, different perception:

Negative frame: "10% of projects fail"
Positive frame: "90% success rate"

Feature absence: "No hidden fees"
Feature presence: "Transparent pricing"

Risk frame: "You might lose data"
Safety frame: "Your data is protected"
Product applications:
  • Positive framing in UI copy and messaging
  • Feature benefits vs feature absence language
  • Success-oriented progress messaging
定义: 信息的呈现方式会改变决策,即使底层数据完全相同。
Same data, different perception:

Negative frame: "10% of projects fail"
Positive frame: "90% success rate"

Feature absence: "No hidden fees"
Feature presence: "Transparent pricing"

Risk frame: "You might lose data"
Safety frame: "Your data is protected"
产品应用场景:
  • UI文案和消息中使用积极表述
  • 强调功能优势而非缺失的语言
  • 以成功为导向的进度提示

7. Sunk Cost Fallacy

7. 沉没成本谬误

What it is: We continue investing because of past investments, not future value.
Leveraging investment:

"You've been with us for 2 years"
"Don't lose your 500 saved items"
"Your profile is 80% complete"
"3,000 connections would miss you"
Product applications:
  • Progress saving and restoration features
  • Investment tracking showing accumulated value
  • Gentle reminders of past engagement
定义: 我们会因为过去的投入而继续投入,而非基于未来的价值。
Leveraging investment:

"You've been with us for 2 years"
"Don't lose your 500 saved items"
"Your profile is 80% complete"
"3,000 connections would miss you"
产品应用场景:
  • 进度保存与恢复功能
  • 显示累积价值的投入跟踪
  • 温和提醒用户过去的参与度

8. Social Proof

8. 社会认同

What it is: We look to others' behavior to determine correct actions.
Types of social proof:

Expert:     "Recommended by security researchers"
Celebrity:  "Used by Elon Musk"
User:       "500,000+ teams trust us"
Wisdom:     "Most popular plan"
Peers:      "Teams like yours use Premium"
Product applications:
  • Customer logos and testimonials
  • Usage statistics and popularity indicators
  • "Most popular" badges on pricing plans
定义: 我们会通过他人的行为来判断正确的行动方式。
Types of social proof:

Expert:     "Recommended by security researchers"
Celebrity:  "Used by Elon Musk"
User:       "500,000+ teams trust us"
Wisdom:     "Most popular plan"
Peers:      "Teams like yours use Premium"
产品应用场景:
  • 客户标志和用户评价
  • 使用统计数据和受欢迎程度指标
  • 定价套餐上的“最受欢迎”标识

9. Scarcity

9. 稀缺性

What it is: We value things more when they're rare or diminishing.
Scarcity signals:

Time:      "Sale ends in 2:34:12"
Quantity:  "Only 3 seats left"
Access:    "Invite-only beta"
Exclusivity: "Limited to 100 companies"

⚠️  Only use with REAL scarcity
Product applications:
  • Limited-time offers (when genuinely limited)
  • Stock/availability indicators
  • Waitlist and invite-only access
定义: 当事物稀有或逐渐减少时,我们会赋予其更高的价值。
Scarcity signals:

Time:      "Sale ends in 2:34:12"
Quantity:  "Only 3 seats left"
Access:    "Invite-only beta"
Exclusivity: "Limited to 100 companies"

⚠️  Only use with REAL scarcity
产品应用场景:
  • 限时优惠(仅限真实有限的情况)
  • 库存/可用性指示器
  • 等待列表和仅限邀请的访问权限

Bias Analysis Framework

偏差分析框架

Step 1: Identify Decision Points

步骤1:识别决策点

Map where users make decisions:
User Journey Decision Points:

Landing Page
├── Stay or bounce?           [Availability, Social Proof]
├── Which CTA to click?       [Framing, Anchoring]
Signup
├── Email or social login?    [Convenience, Trust]
├── Share optional data?      [Reciprocity]
Pricing
├── Which plan?               [Anchoring, Decoy]
├── Monthly or annual?        [Loss Aversion]
Onboarding
├── Complete or skip?         [Commitment, Sunk Cost]
├── Invite teammates?         [Social Proof]
Retention
├── Continue or churn?        [Sunk Cost, Loss Aversion]
└── Upgrade or stay?          [Anchoring, Social Proof]
绘制用户做出决策的节点:
User Journey Decision Points:

Landing Page
├── Stay or bounce?           [Availability, Social Proof]
├── Which CTA to click?       [Framing, Anchoring]
Signup
├── Email or social login?    [Convenience, Trust]
├── Share optional data?      [Reciprocity]
Pricing
├── Which plan?               [Anchoring, Decoy]
├── Monthly or annual?        [Loss Aversion]
Onboarding
├── Complete or skip?         [Commitment, Sunk Cost]
├── Invite teammates?         [Social Proof]
Retention
├── Continue or churn?        [Sunk Cost, Loss Aversion]
└── Upgrade or stay?          [Anchoring, Social Proof]

Step 2: Map Current Bias Usage

步骤2:梳理当前偏差的使用情况

Audit existing design:
ScreenDecisionBias UsedEthical?Effective?
PricingPlan selectionAnchoring
CheckoutAdd extrasScarcity⚠️ Fake
Trial endConvertLoss aversion
审核现有设计:
页面决策内容使用的偏差是否符合伦理?是否有效?
定价页套餐选择锚定偏差
结账页添加附加项稀缺性⚠️ 虚假
试用期结束转化为付费用户损失厌恶

Step 3: Design Improvements

步骤3:设计改进方案

For each decision point:
Decision: Plan selection

Current state:
- Plans listed low to high
- No default highlighted
- Equal visual weight

Improved design:
- Anchor with Enterprise first (Anchoring)
- "Most popular" badge on target plan (Social Proof)
- "Recommended for you" personalization (Confirmation)
- Annual savings calculated (Loss Aversion)
针对每个决策点:
Decision: Plan selection

Current state:
- Plans listed low to high
- No default highlighted
- Equal visual weight

Improved design:
- Anchor with Enterprise first (Anchoring)
- "Most popular" badge on target plan (Social Proof)
- "Recommended for you" personalization (Confirmation)
- Annual savings calculated (Loss Aversion)

Output Template

输出模板

After completing analysis, document as:
markdown
undefined
完成分析后,按以下格式记录:
markdown
undefined

Cognitive Bias Analysis

Cognitive Bias Analysis

Product/Feature: [Name]
Analysis Date: [Date]
Product/Feature: [Name]
Analysis Date: [Date]

Decision Point Audit

Decision Point Audit

Decision PointCurrent BiasesEthical AssessmentRecommendations
[Point 1][Biases used][✅/⚠️/❌][Changes]
[Point 2][Biases used][✅/⚠️/❌][Changes]
Decision PointCurrent BiasesEthical AssessmentRecommendations
[Point 1][Biases used][✅/⚠️/❌][Changes]
[Point 2][Biases used][✅/⚠️/❌][Changes]

Recommended Improvements

Recommended Improvements

High Priority

High Priority

  • [Improvement 1]: Apply [bias] at [location] to [effect]
  • [Improvement 2]: Remove [dark pattern] from [location]
  • [Improvement 1]: Apply [bias] at [location] to [effect]
  • [Improvement 2]: Remove [dark pattern] from [location]

Medium Priority

Medium Priority

  • [Improvement 3]
  • [Improvement 4]
  • [Improvement 3]
  • [Improvement 4]

Ethical Checklist

Ethical Checklist

  • All scarcity claims are factual
  • Users can easily reverse decisions
  • No exploitation of vulnerable states
  • Transparent about pricing and terms
  • Personalization is controllable
  • All scarcity claims are factual
  • Users can easily reverse decisions
  • No exploitation of vulnerable states
  • Transparent about pricing and terms
  • Personalization is controllable

Success Metrics

Success Metrics

MetricCurrentTargetMeasurement
Conversion rateX%Y%Analytics
User satisfactionXYSurvey
Regret rateX%<Y%Cancellations
undefined
MetricCurrentTargetMeasurement
Conversion rateX%Y%Analytics
User satisfactionXYSurvey
Regret rateX%<Y%Cancellations
undefined

Ethical Guidelines

伦理准则

✅ Do: Enhance Experience

✅ 可以做:提升体验

Ethical bias application:

Reducing cognitive load:
├── Smart defaults (don't make users think)
├── Progressive disclosure (show what's relevant)
└── Clear visual hierarchy (guide attention)

Building trust:
├── Real testimonials with names/photos
├── Honest scarcity (actual inventory)
└── Transparent pricing (no surprises)

Helping decisions:
├── Comparison tables (reduce effort)
├── Recommendations (based on real fit)
└── Clear CTAs (obvious next steps)
Ethical bias application:

Reducing cognitive load:
├── Smart defaults (don't make users think)
├── Progressive disclosure (show what's relevant)
└── Clear visual hierarchy (guide attention)

Building trust:
├── Real testimonials with names/photos
├── Honest scarcity (actual inventory)
└── Transparent pricing (no surprises)

Helping decisions:
├── Comparison tables (reduce effort)
├── Recommendations (based on real fit)
└── Clear CTAs (obvious next steps)

❌ Don't: Exploit Users

❌ 不可以做:剥削用户

Dark patterns to avoid:

Fake urgency:
├── "Only 2 left!" (when unlimited)
├── "Sale ends soon!" (perpetual sale)
└── Countdown timers that reset

Hidden information:
├── Fees revealed at checkout
├── Auto-renewal buried in terms
└── Difficult cancellation flows

Manipulation:
├── Guilt-tripping copy
├── Confirm-shaming ("No, I don't want to save money")
└── Trick questions in opt-outs
Dark patterns to avoid:

Fake urgency:
├── "Only 2 left!" (when unlimited)
├── "Sale ends soon!" (perpetual sale)
└── Countdown timers that reset

Hidden information:
├── Fees revealed at checkout
├── Auto-renewal buried in terms
└── Difficult cancellation flows

Manipulation:
├── Guilt-tripping copy
├── Confirm-shaming ("No, I don't want to save money")
└── Trick questions in opt-outs

Ethical Decision Framework

伦理决策框架

Before applying a bias, ask:

1. Is this helping the user?
   YES → Continue
   NO  → Stop

2. Would I be comfortable if this was exposed?
   YES → Continue
   NO  → Stop

3. Does this create long-term value?
   YES → Continue
   NO  → Stop

4. Would this work on an informed user?
   YES → Continue (persuasion)
   NO  → Stop (manipulation)
Before applying a bias, ask:

1. Is this helping the user?
   YES → Continue
   NO  → Stop

2. Would I be comfortable if this was exposed?
   YES → Continue
   NO  → Stop

3. Does this create long-term value?
   YES → Continue
   NO  → Stop

4. Would this work on an informed user?
   YES → Continue (persuasion)
   NO  → Stop (manipulation)

Real-World Examples

真实案例

Amazon: Ethical Anchoring

亚马逊:符合伦理的锚定

Product page:

List Price:    $79.99  ──→ Anchor (if real MSRP)
Price:         $49.99
You Save:      $30.00 (38%)

✅ Ethical if list price is genuine
❌ Unethical if inflated for appearance
Product page:

List Price:    $79.99  ──→ Anchor (if real MSRP)
Price:         $49.99
You Save:      $30.00 (38%)

✅ Ethical if list price is genuine
❌ Unethical if inflated for appearance

Spotify: Positive Framing

Spotify:积极表述

Subscription conversion:

"Get 3 months free"
    vs
"Pay for 9 months, get 12"

Same value, different perception.
Ethical because both options are clearly available.
Subscription conversion:

"Get 3 months free"
    vs
"Pay for 9 months, get 12"

Same value, different perception.
Ethical because both options are clearly available.

Duolingo: Commitment + Loss Aversion

Duolingo:承诺 + 损失厌恶

Streak system:

"🔥 15 day streak!"
"Don't break your streak - practice now"

✅ Ethical: Creates positive habit
⚠️ Watch for: Anxiety-inducing pressure
Streak system:

"🔥 15 day streak!"
"Don't break your streak - practice now"

✅ Ethical: Creates positive habit
⚠️ Watch for: Anxiety-inducing pressure

Integration with Other Methods

与其他方法的整合

MethodCombined Use
Five WhysWhy do users behave unexpectedly?
Graph ThinkingMap bias influences across user journey
Business CanvasBias impact on value proposition
Jobs-to-be-DoneAlign bias use with user goals
A/B TestingValidate bias effectiveness ethically
方法结合使用场景
五个为什么分析用户异常行为的原因
图形思维绘制偏差在用户旅程中的影响路径
商业模式画布偏差对价值主张的影响
用户待办任务使偏差的使用与用户目标保持一致
A/B测试以符合伦理的方式验证偏差的有效性

Quick Reference

快速参考

BIAS CHEAT SHEET

Acquisition:
├── Social Proof → "Join 50,000+ users"
├── Anchoring → Show premium first
└── Scarcity → "Limited beta access"

Activation:
├── Commitment → Small first steps
├── Planning Fallacy → Realistic time estimates
└── Loss Aversion → Show progress at risk

Retention:
├── Sunk Cost → "Your history, connections"
├── Confirmation → Personalized experience
└── Social Proof → "Your team uses this"

Revenue:
├── Anchoring → Price comparison
├── Framing → Annual savings highlighted
└── Loss Aversion → "You're losing $X/month"

Referral:
├── Social Proof → "X friends joined"
├── Reciprocity → Give before asking
└── Scarcity → "Exclusive invite codes"
BIAS CHEAT SHEET

Acquisition:
├── Social Proof → "Join 50,000+ users"
├── Anchoring → Show premium first
└── Scarcity → "Limited beta access"

Activation:
├── Commitment → Small first steps
├── Planning Fallacy → Realistic time estimates
└── Loss Aversion → Show progress at risk

Retention:
├── Sunk Cost → "Your history, connections"
├── Confirmation → Personalized experience
└── Social Proof → "Your team uses this"

Revenue:
├── Anchoring → Price comparison
├── Framing → Annual savings highlighted
└── Loss Aversion → "You're losing $X/month"

Referral:
├── Social Proof → "X friends joined"
├── Reciprocity → Give before asking
└── Scarcity → "Exclusive invite codes"

Resources

资源