Attention Works Like a Muscle. That Means You Can Train It.

There is a sentence Daniel Goleman wrote in 2013 that is worth reading slowly, because most engineers have quietly assumed the opposite for years:

“Attention works much like a muscle — use it poorly and it can wither; work it well and it grows.” — Daniel Goleman, 2013

Read that again with your own working day in mind. Most people who struggle to concentrate treat it as a fact about themselves. They say they have a short attention span, the way they might say they are tall or left-handed. It sounds like a description. It is actually a surrender, because it removes the possibility of doing anything about it.

Attention is not a fixed trait you are stuck with. It is trainable. And anything trainable is a skill.

Goleman’s framing puts the matter back in your hands. If attention behaves like a muscle, then the person who cannot sit with a problem for forty minutes is not defective. They are untrained. Those are very different situations, and only one of them has a remedy.

Why this matters more in the AI age

For most of an engineer’s career, knowing things was the advantage. You knew which subsystem handled the mapping. You knew what that error meant. You had read the datasheet. Knowledge was expensive to acquire, so the people who had it were valuable.

That advantage is thinning. A language model will produce a confident, well-organised, mostly correct explanation of nearly any error message in a couple of seconds. Recall has become cheap. What has not become cheap is the ability to look at that confident explanation and notice that it does not match what the hardware is actually doing. This is the same point we made in Information Is Free. Skill Is Not. — what you can look up was never the thing anyone was paying for.

In the AI age, trained attention is the difference between using the tools and being replaced by them.

Noticing requires presence. Presence requires attention that has been built up rather than let go. The engineer who is half-present will take the plausible answer and move on, and will be wrong in a way that costs the team a week. The engineer who is fully present will pause on the one detail that does not fit. Both had the same tool. Only one had the trained attention to use it well.

How to train your attention

The muscle comparison is useful because it sets the right expectations. You do not fix a weak muscle by reading about it, and you do not fix it in one session. You train your attention the same way you would train anything physical: load it, repeatedly, slightly beyond what is comfortable, and let it recover.

Everyone else is assuming their attention is fine. That assumption is the opening.

A few ways to load it deliberately:

• Work in fixed, unbroken blocks. Start with a length you will actually complete. Forty-five minutes finished honestly is worth more than a planned three hours that you abandon after ten. Extend the block only once the shorter one is reliable.
• Remove the channel, do not resist it. A muted notification still costs you, because some part of your mind is holding a place for it. Quit the chat client. Put the phone in a different room, not face-down on the desk. This is how you leave the state of continuous partial attention that most working days now default to.
• Treat the urge to check as the repetition. The pull to look at something else usually arrives at the exact moment the problem becomes hard. That moment is not an interruption of the work. It is the work. Stay another five minutes.
• Read the important thing twice, slowly. The error, the requirement, the register description, the function you are about to change. Slow reading is a form of attention training that also happens to prevent the bugs.
• Practise recovery, not just effort. A trained muscle needs rest that is genuinely rest. Scrolling a feed between sessions is not recovery; it is more of the same load, applied badly. Walk, or sit, or look at something far away.
• Measure it honestly. Note how long you stayed with one thing before you reached for something else. That number is your baseline. Watch it move over a month.

One warning worth repeating: effort that feels comfortable is usually not training. The same trap shows up in study habits, where rereading your notes feels like learning and mostly is not. If the session was pleasant throughout, you probably did not load anything.

Using AI without letting the muscle wither

AI assistants are not the enemy of attention, but they make the untrained state very comfortable. The discomfort of not knowing used to last long enough to force thinking. Now it can be ended in two seconds, every single time. Ending it every single time is exactly the pattern Goleman describes as using the muscle poorly.

Form your own answer first. Then let the tool test it. You keep the reasoning and outsource the typing.

In practice: sit with the problem, write down what you believe is happening and why you believe it, and only then bring in the assistant to check you, correct you, and handle the mechanical parts. You still get the speed. You just do not pay for it with the one capacity that was going to keep you employable. Used in that order, the tool is one more way to train your attention rather than a reason to let it wither.

At TECH VEDA we train this the way we train any technical skill — long, undistracted sessions on real systems, where the answer is not available until you have understood the machine in front of you.

What accumulates over the years

The reason so few engineers train their attention is that the early returns are small and private. Nobody praises you for the bug you did not create. The first month gives you almost nothing you can show anyone. This is exactly why the advantage survives: a benefit that takes years to appear is a benefit almost nobody will wait for.

But if you train your attention consistently, the gains do not add up in a straight line. They multiply, because each one makes the next one easier to earn.

Roughly what to expect, if you keep at it:

• In the first weeks, the careless errors stop. The off-by-one you would have written, the return value you would not have checked, the requirement you would have pattern-matched instead of read. These were never hard problems. They were absence. They go first, and they go quietly.
• Within a few months, your reasoning chains get longer. Holding four related facts in your head at once is what lets you solve a problem that cannot be solved by holding three. Every extra link you can hold moves a whole class of problem from impossible to merely difficult. This is where people notice that you have started finishing things others hand back.
• Over a year or two, your knowledge starts to stick. This is the part most engineers miss. Nothing enters your memory that you were not attending to when it passed. Half-present work deposits nothing. So one engineer finishes the year with a real, usable library of patterns from every system they touched, and another finishes the same year having been present for none of it. Both write the same number on their CV.
• After a few years, the problems come to you. You become the person who finds the thing nobody else could find. Harder problems get routed to you, and harder problems are better training than easier ones. The loop starts feeding itself: attention earns you the difficult work, and the difficult work builds the attention.
• Over a career, the gap becomes structural. The engineer with ten years of attended experience and the engineer with the same year lived ten times are not close to each other, and no amount of catching up in the last six months will fix it. One of them can be handed an unfamiliar system and reason about it. The other needs someone to have already written the answer down.

Attention is the gate on everything you learn. Whatever you were not present for, you did not keep — no matter how many years it took to not keep it.

The same compounding runs in reverse, and this is the uncomfortable half. Every session spent half-present is not neutral. It trains the habit of half-presence, weakens the tolerance for difficulty, and leaves the year’s experience undeposited. The engineer who feels stuck at the same level after six years is usually not lacking talent or opportunity. They have been practising something, faithfully, for six years. It was just not the thing they thought they were practising.

You are always training something. The only question is whether you chose it.

And the leverage of the tools compounds along the same lines. Give an AI assistant to an engineer with trained attention and it removes the mechanical work, so they spend their hours on the hard part and get better faster than they could have alone. Give the same assistant to an engineer without it and it removes the thinking, so the hard part never gets practised at all. Same tool, five years, two very different people at the end of it.

The encouraging part of Goleman’s sentence is the second half. Attention can wither, yes. But it also grows. Whichever is happening to yours right now, it is happening because of what you do every day — which means the compounding has already started, in one direction or the other. Train your attention from here and you decide which direction it runs.

— Raghu Bharadwaj