Module 2 · Attention & digital life · Moderate evidence

Habit loops and unpredictable rewards

Cue, action, reward — and why a reward you can’t predict is the hardest one to put down.

7 min read · Level: Practising · Free — no account needed to read

The idea

Habits run on a loop: a cue (a buzz, boredom, a spare second), an action (checking the phone), and a reward (a message, something funny, a bit of news). Repeat the loop enough and it becomes automatic — your hand moves before you decide.

The strongest loops use rewards you can’t predict. Sometimes the check brings something exciting, often nothing. Psychologists have known for decades that unpredictable rewards keep behaviour going more stubbornly than reliable ones — the same principle that makes slot machines compelling. Pull-to-refresh and feeds work in a similar way.

You may hear this described as a “dopamine” effect. Dopamine is involved in wanting and anticipation — the pull toward a possible reward — more than in pleasure itself. That is why checking can feel urgent even when what you find isn’t very satisfying.

Why it works

Seeing the loop gives you three places to intervene: remove or change the cue, make the action harder, or find a different reward that meets the same need.

A real example

The situation

Tomás unlocks his phone roughly every ten minutes while studying. Most checks bring nothing, but he can’t seem to stop.

What happened

He maps the loop: the cue is getting stuck on a hard problem, the action is checking, the reward is a brief escape. He changes the reward — when he gets stuck, he stands and stretches for a minute instead. Checking drops noticeably within a week.

The takeaway

He didn’t fight the need for a break; he met it in a way that didn’t hijack the next hour.

Use this when you are…

• a habit feels automatic and hard to control
• you check something even when it rarely has anything new
• planning how to change a digital habit

Common mistakes

• Only fighting the action with willpower, and ignoring the cue and the reward.
• Believing popular claims that you can “reset” your brain’s chemistry by avoiding pleasure for a day.
• Removing a reward without replacing the need it was meeting.

What the evidence says

Moderate evidence: Several studies support it, but the effects are smaller, the studies are fewer or less rigorous, or results vary between groups.

The power of unpredictable rewards is long established in behavioural research, and habit loops are well studied. How strongly app design drives compulsive use varies between people, and brain-chemistry explanations in popular media are often oversimplified.

Where this comes from: Habit formation research; Research on digital media and wellbeing.

Fields that study this: Behavioural science, Neuroscience.

Try it now: Map one loop

1. Pick one habit you want to change — for example, checking a particular app.
2. Write the cue (what happens just before), the action, and the reward (what you get from it).
3. Choose one: remove the cue, add friction to the action, or swap the reward.

Practise with a guided technique

• Urge surfing — (3–5 min · Emerging evidence)
• Replace the scroll — (2 min · Practical evidence)
• Add friction to your phone — (5 min · Emerging evidence)

Questions to think about

• What need is your most frequent check really meeting — boredom, connection, escape, curiosity?
• What else could meet that need?

Try this today

Notice the cue the next three times you reach for your phone without deciding to.

Afterwards, ask yourself

What cues did you notice, and which one would be easiest to change?

Words in this lesson

• Habit loop — The cycle that keeps a habit going: a cue, a behaviour, and a reward.
• Variable reward — A reward that comes unpredictably. Unpredictable rewards keep behaviour going more strongly than predictable ones.
• Dopamine — A brain chemical involved in motivation and in wanting or anticipating rewards — more about the pull toward something than about pleasure itself.
• Cue — A trigger that prompts a habit — a time, place, feeling, person or preceding action.