ALIVE LIBRARY
CALM TECHNOLOGY

The ethics of variable rewards

Last updated: June 2026

Variable rewards are design patterns that deliver unpredictable reinforcement, creating strong behavioral hooks by exploiting the brain’s dopamine-driven anticipation system — the same mechanism that makes slot machines powerfully addictive.

01

The Principle

B.F. Skinner’s mid-20th century experiments established that variable-ratio reinforcement schedules — rewards delivered after an unpredictable number of responses — generate the highest response rates and greatest resistance to extinction. Slot machines embody this: each pull carries the possibility of a jackpot, keeping players engaged through uncertainty rather than consistent payout.

The neuroscience goes deeper. Wolfram Schultz’s research on reward prediction error shows that dopamine neurons fire most strongly not on the reward itself, but when outcomes are better than expected. This anticipatory surge creates a powerful learning signal that reinforces the preceding behavior. Digital products borrow this directly: pull-to-refresh, infinite scroll, like notifications, algorithmic feeds, and loot boxes all trigger the same prediction-error loop. The brain stays hooked on the possibility of the next rewarding moment.

In my own work I have implemented these patterns more than once. At first they felt like clever engagement tools. Only later, reviewing usage logs and hearing users describe feeling “addicted” or “unable to stop scrolling,” did the ethical dimension become impossible to ignore. What I had seen as harmless gamification was shaping behavior through the same mechanisms long studied in behavioral psychology and neuroscience.

02

Why It Matters for Design & Building

Variable rewards are among the most potent tools in the designer’s toolkit precisely because they work so well. This potency demands serious ethical scrutiny in calm technology. Used sparingly and transparently they can support useful habits. Deployed indiscriminately they erode autonomy, create compulsive cycles, and prioritize platform metrics over human wellbeing.

As a Design Engineer, this has become one of the areas I examine most rigorously. The central question is whether the uncertainty serves the user’s goals or primarily serves retention numbers. In one social-feature project, replacing unpredictable notification rewards with predictable, user-controlled digests dramatically reduced compulsive checking while preserving connection. The product felt calmer overall.

In calm technology, variable rewards should be used with restraint. They should be transparent, optional, and aligned with deliberate user intent rather than engineered compulsion. When a design deliberately keeps users in a state of anticipation to boost time-on-site or session frequency, it crosses from persuasion into manipulation. The ethical line is whether a user would willingly choose to keep the mechanism active if they fully understood how it operates.

03

Real-World Examples

Loot boxes in mobile games represent one of the clearest and most criticized applications. Players spend time and money chasing unpredictable rare rewards, with the variable schedule keeping them engaged long after rational value diminishes. Many jurisdictions now regulate these as gambling mechanics.

Infinite scroll and algorithmic feeds on major social platforms demonstrate the mechanism at massive scale. Each new piece of content arrives with variable relevance and emotional impact, triggering repeated dopamine anticipation and making stopping feel effortful. Users often report entering the platform for a specific purpose and emerging hours later wondering where the time went.

A news app I reviewed for a client used variable “breaking news” push notifications and personalized story recommendations. While it increased session frequency, many users described feeling anxious and overwhelmed. Shifting to scheduled digests and user-defined thresholds preserved timeliness for important events while reducing the constant low-level pull.

Related entries

References

  1. Skinner, B.F. (1953). Science and Human Behavior.
  2. Schultz, W. (2016). "Dopamine Reward Prediction Error Coding." Dialogues in Clinical Neuroscience.
  3. Eyal, N. (2014). Hooked: How to Build Habit-Forming Products.
  4. Case, A. (2015). Calm Technology. O'Reilly Media.
  5. Gray, C. M., et al. (2018). "The Dark (Patterns) Side of UX Design." Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems.