Reverse Neuroscience: How to Take Your Attention Back

You pick up your phone to check one thing. The weather, maybe, or a message you’re waiting on. Twenty minutes later you’re three videos deep into something you didn’t choose, and you can’t remember what you picked the phone up for.

It’s tempting to call that a lack of willpower. It isn’t. The app in your hand was designed by people who know exactly how your brain decides what to look at next, and they tuned it to keep you looking. You’re not losing a fair fight. You’re losing a fight you didn’t know you were in.

Cognitive neuroscientist Dr. Sahar Yousef, who teaches at UC Berkeley and runs the Becoming Superhuman Lab, put it neatly in a recent conversation on The Diary Of A CEO: “They use neuroscience to get us hooked, and I think we can use reverse neuroscience to reverse this entire thing.” That idea is worth taking seriously. If the problem was engineered, the fix can be engineered too.

How did your attention get taken?

Yousef compares it to the food industry, and the comparison holds up. Ultra-processed food works on three levers: more stimulation (sugar, salt, fat), less friction (it’s everywhere, and it’s cheap), and no point where you feel full. You don’t stop because you’re satisfied. You stop because the bag is empty.

Feeds work the same way. Colour, motion and sound turn up the stimulation. Infinite scroll and autoplay remove every bit of friction, so there’s never a natural place to stop. And because there’s always one more video, you never reach the moment where you feel you’ve had enough.

None of this is an accident. Yousef says plainly that the default settings on your devices are designed to maximise engagement, and that’s true even for work tools, not just social media. Her response is the line she gives to people who feel helpless about it: “I am better than the default settings.” In other words, the defaults are someone else’s choice. You’re allowed to make your own.

What does the science actually say?

This is where it pays to be careful, because attention research gets quoted loosely. Some of the numbers you’ll hear hold up well. Others are bigger than the evidence behind them.

The famous one is 47 seconds. It comes from Gloria Mark’s research, which logged how long office workers stayed on one screen before switching. The average fell from about two and a half minutes in 2004 to around 47 seconds in studies from 2016 onward. Here’s the thing, though: that measures time on one window, not time before you lose focus. Plenty of switches are deliberate, like jumping from a document to the data you need for it. What the research does show is that frequent switching goes along with more stress and less output.

The cost of switching between tasks is real and well established. Every time you change tasks, your brain has to drop one set of rules and load another, and that reconfiguration costs time and accuracy. Part of your attention also stays stuck on the task you just left. So “I’ll just answer this one message” is never quite as cheap as it feels.

Grayscale is a good example of a tip that works, just not as dramatically as it’s often sold. Turning your phone black-and-white makes it less rewarding to look at. In one study of US college students, it cut screen time by about 39 minutes a day. A pre-registered Dutch study found about 20 minutes a day, mostly from image-heavy apps, though people unlocked their phones just as often. That’s a real result for a setting you can change in ten seconds. It’s not the full hour a day you’ll sometimes hear quoted.

Breaks matter too, and what you do in them matters more. In a 2024 experiment with 308 employees, a four-minute break looking at nature restored people’s energy completely. A four-minute break scrolling short videos helped only partly, and people rated it less relaxing than simply doing nothing. Scrolling feels like rest. Your brain doesn’t treat it as rest.

The reverse moves

If the attention economy works by turning three dials, reverse neuroscience means turning them back. Yousef’s practical advice lines up neatly with each one.

Turn the stimulation down. Grayscale is the easy start. The bigger win is notifications. Yousef keeps a short VIP list: her partner, her boss’s boss, her top clients and a few keywords. Everything else waits “in the mailbox” until she chooses to look. Most notifications are someone else deciding that their thing is more important than yours right now.

Turn the friction back up, on distractions. Put your phone in another room for one work session. It sounds too simple to matter, but every bit of effort between you and a distraction is a moment where you get to choose. Out of sight is the closest thing to portion control your attention gets.

Get back the feeling of being full. Finishing a long article, a book chapter, or an hour of real conversation leaves you satisfied in a way a feed never does. Long-form content has an end. When you reach it, you’re done, and you know it.

Give your attention a shape. Yousef sorts the day’s work into two piles. There are the little tasks: an email, a signature, reordering something online. They take minutes and you get to cross them off. Then there are the one to three most important tasks that actually move your work and your life forward. She calls the first kind “fast dopamine” and the second “slow dopamine”, and she has a good image for the difference. The important task is a rat in a maze with a big hunk of cheese at the very end: a long, hard journey before any reward. The little tasks are crumbs lying right in front of you. Can you just eat that crumb? Sure. And another one.

That’s the trap in one picture. A feed is nothing but crumbs, an endless supply of small rewards with no maze at all, and your inbox isn’t much better. Your most important work doesn’t lose because you’re lazy. It loses because its reward comes late. Treat “fast” and “slow” dopamine as her metaphor, not as brain chemistry: dopamine does a lot of work in motivation, but the episode’s own fact-check notes that the direct evidence for little dopamine bursts when you finish a step is limited. What is well supported is the fix. Her advice is to “take that big hunk of cheese and sprinkle it throughout the maze”: break the important task into small, specific steps, so each one feels like a crumb, but a crumb you earned on the work that matters.

That’s the heart of what she calls a focus sprint, and it comes in five steps:

1. Block the time in advance, ideally when your energy is highest.
2. Set the goal as an output, and break it into a timed list. Not “work on the deck” but “slide three, seven minutes”. You don’t want to be working out the rules of the game in the middle of the game.
3. Cut the distractions. Phone away, notifications off, headphones if it’s noisy.
4. Set a timer, and let it count down. Watching progress makes you work faster.
5. Take a real break afterwards. Outside, or doing nothing, not scrolling.

Specific goals broken into steps and visible progress both have solid research behind them. The “double your output” claim you may have heard doesn’t have a published study yet, so don’t hold yourself to it. Use the sprint because it makes the work feel manageable, and save the crumbs for the gaps between sprints.

It’s not about another tool

Here’s where most productivity advice goes wrong. When you can’t focus, the reflex is to add something: a new app, a new system, a new dashboard to track it all. Yousef’s advice runs the other way. The best way to get to what matters, she says, “is to remove obstacles as opposed to adding.” She calls it digital hygiene: “There are obstacles in the way of productivity and performance and focus. Forget becoming superhuman. We just need to become human.”

She’s also blunt about why the next tool rarely fixes it. Software companies renew contracts by showing engagement, so “a productivity tool’s incentive actually isn’t your own productivity. It is your engagement with the tool.” That’s true of the boring stuff too: email, chat, the apps your whole team runs on. A tool that measures its success by how much you use it can’t also be the one that gets you out of it.

So what does help? Look at friction again, because it runs in two directions. You want more friction between you and your distractions, and less between you and your work. Most of your day is set up exactly backwards. Opening a chat app takes one tap. Getting into a piece of real work means hunting for the right file, finding the last email, remembering where you left off, and explaining the context to yourself all over again. No wonder the chat wins.

Getting something done is about your attention on one topic and the obstacles between you and it. Take away the obstacles and the attention follows. Add another tool and you’ve usually added another obstacle.

What this looks like for a software developer

Software development makes the point especially clearly, because a developer switches between levels of the same work all day. There’s the architecture: how the pieces fit, where the data lives, what could go wrong with security. There’s the organisation around the work: planning, tickets, reviews, keeping track of what’s done. And there’s the code itself. Each level needs a different kind of attention, and every jump between them costs you the same switching price as jumping between apps. A morning that moves from a design question to a ticket to a failing test and back again is three kinds of thinking fighting over one head.

This is where an AI coding harness changes the shape of the day, and not by being one more tool to check. It takes over most of the typing, the kind of work AI should handle so you don’t have to. Written procedures organise the work: what gets planned, reviewed and tested, in which order. What’s left for the developer is the part that needs a human most: the overall architecture, the security decisions, and reviewing what the harness built against what was intended. You still own every decision. You just stop paying the switching cost for the parts that didn’t need you.

We build LumaVista this way, and the difference is hard to overstate. Staying at one level for hours is deep work, and deep work is where the important decisions get made well. The product gets the benefit: more thought goes into the architecture and the security, because the attention isn’t spent on everything around them. If you want the other side of this, the risks of handing code to AI tools, we wrote about protecting your code and your clients.

Built for your focus, not for engagement

That’s the test we hold LumaVista to, for everyone, not only developers. Its job is to take friction out of your work, not to be one more place you have to check. Today, the Home screen gives you one place to put an idea the moment you have it. Type it in where you are, and LumaVista helps you shape it into something you can act on, so you don’t lose the thought while you decide which app it belongs in.

Here’s what we’re building next, and each one is about taking something away:

• No deciding what’s next. Home will show your next items sorted by what they need from you: something for deep focus, a few light tasks, and anything that’s genuinely time-sensitive. You set when you do your deep work, and Home leads with it then.
• No switching between unrelated things. Related work comes as one batch, such as three replies to the same client or the weekly reports that are all due, so you stay in one headspace longer.
• No hunting before you start. Before you pick up a topic, a short briefing is already waiting: where things stand, what changed since you last looked, and everything you need, linked. It skips what you already know.
• No setting up a sprint by hand. Start a focus sprint on one item and LumaVista brings the briefing, drafts the goal and the timed steps, and keeps that step list in view on every page while a countdown runs. Interruptions wait, and the break at the end comes before anything that was held back.
• No needless interruptions. When something you delegated finishes, LumaVista interrupts you only if it matters more than what you’re doing right now. Everything else waits for your next break.
• No feed to fall into. A short daily list of long-form pieces on what you care about, from sources you can trust, that ends when it’s done. It offers pieces a little longer than the ones you finish, so the length you can hold grows over time.
• No relearning what you already learned. Spaced-repetition practice on the topics you choose, offered as a quick activity when it fits your day. It’s AI that trains your brain instead of replacing it.

None of that is about doing more, or about adding a tool to your stack. It’s about getting your attention back, and spending it on the things you’d choose if nobody was competing for it.

What to do now

1. Turn on grayscale tonight. It takes ten seconds in your phone’s accessibility settings, and published studies show 20 to 40 fewer minutes of screen time a day.
2. Cut notifications to a VIP list. Pick the handful of people and topics that genuinely can’t wait. Everything else can wait in the mailbox.
3. Move your phone out of sight for one work block. Another room is best. A drawer will do.
4. Name tomorrow’s one to three most important tasks tonight. Give them the first uninterrupted block you have.
5. Take one break outside instead of on your phone. Four minutes of nature does more for your energy than four minutes of scrolling.
6. Read one long piece all the way to the end. Notice how it feels to finish something.
7. Remove one obstacle from your most important work. Pick the task you keep putting off and ask what stands between you and starting it: a missing file, an unclear next step, a tool you have to open first. Fix that one thing before you add anything new.