Myo← All posts
Mechanism·

Your feed is a variable-ratio schedule

By the Myo team

The most powerful reinforcement schedule in behavioural psychology was mapped in 1957. Your phone runs it by default, and the cost isn't in how much you use it.

In 1957, Charles Ferster and B.F. Skinner published a book cataloguing what happens when you vary how a reward arrives while holding constant what it is.

The finding that outlived everything else: the schedule matters more than the reward.

That sounds like a lab curiosity about pigeons. It is also the best single explanation for why you pick up your phone to check one thing and put it down twenty minutes later without remembering deciding to stay.

The four basic shapes

Reinforcement can be scheduled on time or on actions, and predictably or unpredictably. That gives four combinations.

Fixed interval. The reward becomes available after a set time. It produces bursts of activity right before the deadline and very little otherwise. Studying for an exam works like this.

Fixed ratio. The reward comes after a set number of actions. You get steady work and a short pause after each payoff. Piecework.

Variable interval. The reward comes after an unpredictable amount of time. You get slow, steady, persistent checking. Waiting for a reply.

Variable ratio. The reward comes after an unpredictable number of actions. This one is different in kind.

In Ferster and Skinner's animal data, ratio schedules produced high, steady response rates, and variable ratio produced the steadiest of all, with almost no pause after a payoff. Behaviour trained on intermittent reward also took longer to extinguish than behaviour rewarded every time. When the reward stops, a variable-ratio learner keeps going, because "nothing yet" looks exactly like a normal dry stretch. There is no signal that says stop.

This is the slot machine. It is not a coincidence that it's also the feed.

What actually runs on variable ratio

Almost everything designed to be checked.

Scroll a feed: most items are nothing, and occasionally something lands. The payoff per swipe is unpredictable. Refresh an inbox, a notification tray, a comment section, a dashboard or a marketplace and you get the same structure. Each action might pay. Usually it doesn't.

The design didn't have to be deliberate to have the effect. Any system where interesting content arrives irregularly and you can check at will is a variable-ratio schedule, whether or not anyone intended it. Pull-to-refresh simply makes the lever easier to reach.

Notifications add a second layer. A buzz is a variable-interval cue: something arrived at a time you couldn't predict. You check, and then the feed behind the notification runs its own variable-ratio loop. One schedule hands you over to the other.

That combination produces the thing people find hardest to explain about their own behaviour: checking that isn't enjoyable and isn't decided. Variable ratio trains persistence, and persistence is a separate output from pleasure. If the gap between "I keep doing this" and "I don't even like this" sounds familiar, wanting and liking run on different systems, and that post explains why the gap exists.

The mechanism, one level down

Why would unpredictability train harder than reliability? The best-supported answer comes from reward-prediction error.

Schultz, Dayan and Montague (1997) recorded single dopamine neurons in monkeys. Those neurons fired to a reward that came unexpectedly. Once the monkey learned that a cue predicted the reward, the firing moved to the cue, and the reward itself produced little response. When an expected reward failed to arrive, firing dipped below its resting rate. The signal tracks the difference between what you expected and what you got. Schultz (2016) reviews the same pattern across monkeys, rodents and humans.

Put that next to a schedule. A fully predictable reward stops generating much error, so the learning signal goes quiet. A reward that might or might not arrive on the next action keeps generating error, every time it lands and every time it doesn't. The schedule keeps the teaching signal switched on.

Two caveats belong here. The recordings come from primates with simple cues, and nobody has recorded dopamine neurons in someone scrolling a feed. And prediction error is a learning signal. It is not "dopamine equals pleasure", which is the version that circulates online and which the same research contradicts.

What the research on feeds and phones shows

The schedule idea has moved from pigeons to people in a few useful studies.

Social reward follows learning rules. Lindström et al. (2021) analysed more than 1 million posts from over 4,000 users across several platforms. The timing of posts fitted a reinforcement-learning model: people posted more often when likes arrived faster. An online experiment with 176 participants then showed that social reward causally shaped posting. The limit is important. This models posting, the active side of social media, and it contains no dopamine measurement. Passive scrolling was not tested.

Alerts cost attention. Kushlev, Proulx and Dunn (2016) ran a within-subject field experiment with 221 participants. In the week with alerts on and the phone within reach, people reported more inattention and hyperactivity than in the week with alerts off and the phone put away. These were self-reported symptoms in students and community adults, not diagnoses.

Batching beats both extremes. This is the study closest to the schedule fix. Fitz et al. (2019) randomized 237 people for two weeks. Those who received notifications batched three times a day reported better attention, mood, productivity and sense of control, and lower stress. Hourly batching did little. And the contradicting result: people who had notifications removed entirely reported more anxiety and more fear of missing out. Going to zero was worse than giving the reward a timetable.

Friction changes behaviour at scale. The Danish Competition and Consumer Authority (2025) ran a six-week field experiment with 269 teenagers aged 13 to 17. Friction interventions cut daily social-media activity by 31 to 36% across the arms, and a six-second wait with a calming animation alone cut use by more than a third. It is a government report, not peer-reviewed, and it reports no follow-up after the interventions stopped, so whether the change outlasts the friction is unknown.

The analogy has limits. Ferster and Skinner never tested phones. "Social media is a slot machine" is an analogy built on animal schedules, and the human studies above measure behaviour and self-report, never dopamine. Treat the schedule as the most useful lens available, not as a proven mechanism for your feed.

The counterintuitive part

The instinct is to reduce total consumption. Cut the hours, delete the app, set a limit.

Schedule logic suggests a different lever. If unpredictability is what trains hardest, then converting one frequent unpredictable reward into a predictable one can do more than removing several predictable ones, even though removal looks like more discipline. The Fitz result points the same way: three scheduled batches helped, and total removal backfired.

Checking messages twice at set times is a fixed-interval schedule. Same messages, same total minutes, a very different training signal. The behaviour it builds is different because the schedule it runs on is different.

That's why screen time as a metric misses something. Two hours on variable ratio and two hours on a predictable schedule are not the same two hours, and a number that counts only duration can't tell them apart.

It is also why fixed-length abstinence tends to disappoint. Switch everything off, then switch it back on, and the old schedule is waiting exactly where you left it. Why the 30-day challenge fails covers the timing side of that.

What to do with this

Don't audit your hours. Audit your unpredictability. Here are five changes you can make this week, each aimed at the schedule.

  1. Find your worst loop. For two days, note each time you check something without deciding to. At the end, pick the one source you checked most often. That is your highest-frequency variable-ratio loop, and it gets all your effort first.

  2. Give it a timetable. Turn that source into a fixed-interval schedule: two or three set times a day, written down. Fitz's best result came from three batches, and hourly did little, so don't make the windows too frequent.

  3. Batch, don't silence. Use your phone's scheduled-summary or focus settings to deliver non-urgent notifications in batches, and keep people who genuinely need to reach you on a direct line. Removing everything was the condition that raised anxiety.

  4. Add a few seconds of friction. Move the app off your home screen, log out, or use a delay tool before it opens. The Danish six-second wait suggests small friction matters, and it works on the trigger rather than on your willpower.

  5. Replace the slot with something effortful. When the urge hits at a scheduled-off time, have one effortful option ready: a page of a book, a short walk, a task with an end. Effort-gated rewards behave differently, which the effort post explains. If the minutes after you stop feel flat, that's expected, and opponent process is the reason.

For a fuller plan aimed at the scrolling habit itself, the doomscrolling guide puts these together. If you want to see which loops dominate your day first, the 60-second audit is a quick place to start.

How Myo uses this

Every reward you log in Myo carries a schedule, and the schedule changes its weight. Fixed sits at the base. Variable interval weighs more. Variable ratio weighs most, because a high frequency of unpredictable reward is the pattern the model treats as the strongest drag on your baseline.

The ordering, variable ratio above variable interval above fixed, comes from the reinforcement literature. The specific multipliers are Myo's modelling choices, not measured constants, and the science page labels them that way. Myo models from what you log. It does not measure dopamine, and no app can. If you were hoping for a number from inside your head, what a dopamine self-check can and can't tell you is worth reading first.

FAQ

Why is variable-ratio reinforcement so hard to stop?

Because the absence of a reward looks normal. On a variable-ratio schedule, many actions in a row produce nothing, so a long dry stretch gives you no information that the rewards have ended. In Ferster and Skinner's animal work, intermittently reinforced behaviour took longer to extinguish than behaviour rewarded every time. The result is checking that continues well past the point where it pays, and well past the point where you enjoy it.

Is social media really like a slot machine?

Structurally, yes. Both pay out after an unpredictable number of actions, which is the definition of a variable-ratio schedule. As a scientific claim it is an analogy, though. The schedule research comes from pigeons and rats, and studies of social media such as Lindström et al. (2021) measure behaviour like posting, not brain chemistry. It is a useful way to understand the pull, and it is not a measured equivalence.

What is the difference between variable ratio and variable interval?

Variable ratio pays after an unpredictable number of actions, so checking more often gets you more payoffs and the behaviour speeds up. Variable interval pays after an unpredictable amount of time, so checking faster does not help much and the result is slow, persistent checking. A feed is closer to variable ratio, because every swipe might pay. Waiting for a specific reply is closer to variable interval.

Do notification batching settings actually help?

There is decent evidence that they can. In a two-week randomized trial with 237 people, Fitz et al. (2019) found that batching notifications three times a day improved attention, mood and sense of control, and lowered stress. Hourly batching did little. People who had notifications removed entirely felt more anxious and more worried about missing out, so a timetable did better than silence.

Does turning off notifications reduce dopamine?

No study shows that, and the question is framed around a misunderstanding. Dopamine is a signalling chemical your brain uses all the time, not a supply you can lower by switching settings. What turning off or batching notifications changes is the schedule your checking runs on, which affects how strongly the habit is trained. The likely benefit is behavioural: fewer triggers and more predictable rewards.

Sources

Myo models dopaminergic state from behavior; it does not measure dopamine. Not medical advice.

Keep reading

More in this guide →Take the 60-second audit →