Desirable Difficulty: Does Friction Build Better Memory?

· product-design

The Observation

Nano Banana-style AI image summaries are everywhere now. Visual explanations of complex ideas—clean, digestible, shareable. But do they actually help people learn?

There's a difference between:

  • Understanding what it is — grasping the concept in the moment
  • Deeply memorizing something — building knowledge that your mind can retrieve and connect to future situations

The first feels like learning. The second is learning.

The Question

Would some friction in understanding be necessary to build better memory and intuition?

What Cognitive Science Says

This is the desirable difficulty hypothesis (Bjork):

Approach Feels like Actual learning
Easy, clear presentation "I get it!" Often shallow
Effortful processing Slower, harder Deeper encoding

Examples of desirable difficulties:

  • Spacing: Spread learning over time (vs. cramming)
  • Interleaving: Mix topics (vs. blocked practice)
  • Generation: Produce answers before seeing them (vs. passive reading)
  • Testing: Retrieve from memory (vs. re-reading)

The visual summary removes all friction. You don't generate, you consume. You don't struggle, you glide.

Contrasts

Nano Banana approach Opposite approach
Instant clarity Productive confusion
Pre-digested Self-constructed
Passive consumption Active struggle
Memorable image Memorable effort

Alternative approaches for deep learning:

  • Feynman technique: explain it yourself, find gaps
  • Socratic dialogue: questions that force you to reason
  • Worked examples with faded steps: gradually remove scaffolding
  • Teaching others: the act of explaining encodes

The Tension

Visual summaries are great for:

  • Breadth: Quickly scanning many ideas
  • Sharing: Spreading concepts virally
  • First contact: Getting oriented on a topic

But potentially harmful for:

  • Depth: Actually integrating knowledge
  • Retention: Building retrievable memory
  • Application: Using knowledge in novel contexts

The illusion of learning: The clarity of the image substitutes for the clarity of your own understanding. You mistake recognition for recall.

Application to Product Design

If you're building learning products, you face a tradeoff:

  • Reduce friction → higher engagement, feels good, spreads well
  • Add friction → lower engagement, feels hard, builds real skill

The best learning products might introduce strategic friction:

  • Start with struggle, end with clarity (not the reverse)
  • Make the user generate before revealing
  • Interrupt fluency when stakes are high

Open Questions

  • Can visual summaries be designed to include desirable difficulty? (e.g., interactive elements, hidden steps, prediction prompts)
  • Is there a "sweet spot" where explanation is clear enough to not lose people, but incomplete enough to require effort?
  • For what kinds of knowledge does this matter? (Procedural vs. declarative? Novel vs. reinforcing?)

Thread Draft

Do Nano Banana-style explainers actualy help people learn?

"I get it" in the moment ≠ "I can use this later"

The first feels like learning. The second is learning.

When everything is pre-digested, you consume but don't construct. You recognize but can't recall.

Genuine ask:

  • Do you actually remember things from visual explainers?
  • Or do they just feel satisfying in the moment?
  • What has made concepts stick for you?

Curious if others notice this too, or if I'm overthinking it.

Related