Curiosity Gap

Canonical
WendelKahneman

Confidence

77%

Cognitive Load

Low

Evidence

production validated

Impact

feature

Ethical Guardrail

Agents must never use fake curiosity. Always deliver on the promise once the gap is closed. Never withhold information the user needs to make an informed decision.

Design Intent

The human brain cannot tolerate an information gap. When we know something is almost known but not quite, we feel a powerful itch to close the gap. Curiosity Gap deliberately creates a small, intriguing information void right before a desired action.

Psychology Principle

The human brain cannot tolerate an information gap. When we know something is almost known but not quite, we feel a powerful itch to close the gap.

Description

Deliberately creates a small information void right before a desired action, leveraging the brain's drive to resolve uncertainty.

When to use

Any prompt, teaser, or entry point where engagement or click-through is low -- feature discovery, content feeds, notifications.

Example

Duolingo Lesson Tease: 'Your friends are learning Spanish... you're only 3 lessons behind them' + one-tap 'Catch up now'.

Autonomy Compatibility

Suggest

Behavioral Objective

Users take the target action to resolve the curiosity gap they feel.

  • Higher open and click-through rates on prompts
  • Increased voluntary engagement with new features
  • Stronger sense of discovery and delight

Target Actor

role

Everyday user

environment

Content feeds, notification-heavy apps

emotional baseline

Mild boredom mixed with desire for novelty

ai familiarity

medium

risk tolerance

low

Execution Model

1

tease_creation

Give just enough context to spark a specific question.

User feels no emotional pull to learn more.

2

gap_highlight

Make the missing piece feel tantalizingly close.

User sees the tease but is not curious enough to act.

3

easy_closure

One-tap or one-second action to resolve the gap.

User wants to close the gap but the action feels like work.

Failure Modes

Gap feels like clickbait

Always deliver real high-value information

micro

Curiosity is too vague

Make the gap personally relevant and specific

micro

Reveal is disappointing

Ensure the payoff exceeds the tease

feature

Overuse causes curiosity fatigue

Limit gaps to 2-3 per session

micro

Gap hides critical information

Never withhold safety, legal, or high-stakes details

feature

Agent Decision Protocol

Triggers

  • Engagement on a new feature or prompt is low
  • User is scrolling past opportunities
  • User says 'I'm not sure if I need this'

Escalation Strategy

L1: Diagnose low engagement via behavioral_signals

L2: Nudge -- add a personalized curiosity tease with one-tap closure

L3: Restructure -- redesign the entry point with a specific, relevant gap

L4: Constrain -- limit to 2-3 curiosity gaps per session to prevent fatigue

L5: Yield -- flag for human behavioral designer review

Example

User hasn't reviewed their weekly summary -> agent shows 'Your project saved 14 hours last week... here's the one change that made it possible' with one-tap reveal.

Behavioral KPIs

Primary

  • Click-through rate on curiosity-teased elements
  • Time from tease to resolution
  • Session depth after closing a gap

Risk

  • User reports of 'tease and disappoint'
  • Dismissal rate of curiosity prompts

Trust

  • User-reported excitement after resolving gaps
  • Autonomy Dial usage when agent creates gaps

Behavioral Signals

gap_ignored

curiosity_tease_shown=true AND gap_closed=false AND dwell_time < 2s

tease_dismissed=true AND repeat_dismiss_count > 2

gap_effective

curiosity_tease_shown=true AND gap_closed=true AND time_to_close < 5s

payoff_revealed=true AND session_depth_after > 3

gap_fatigue

curiosity_tease_shown_count > 3 AND gap_closed_rate < 20%

user_feedback_contains='clickbait' OR user_feedback_contains='teasing'

Decay Monitoring

Revalidate when

  • User base becomes desensitized to common teaser styles
  • Platform algorithms change content visibility
  • New content types are introduced

Decay signals

  • Declining click-through on teased content
  • User feedback that 'everything feels like a tease'
  • Drop in perceived value after resolving gaps

Pattern Relationships

Related Patterns

Canonical Implementation

Duolingo Lesson Tease: 'Your friends are learning Spanish... you're only 3 lessons behind them' + one-tap 'Catch up now'

Telemetry Hooks

curiosity_tease_showngap_closedpayoff_revealed

Tags

engagementdiscoveryagent-ready