What is attention — and can you actually train it?

1 · You are what you attend to

Start with the stakes, because they're higher than "focus better at work." Over a life, the contents of your mind are built from what you repeatedly attend to. Attend to threat, and the world comes to look dangerous. Attend to what's missing, and you live in lack. Attend to another person — really — and that is most of what love is in practice. William James said it in 1890: my experience is what I agree to attend to. Nothing has overturned it. It's also why the fight for your attention is now a global industry: whole companies are engineered to harvest it, because it is the scarcest valuable thing you have.

What attention actually is, in plain terms: your brain is flooded with far more signal than it can use, so it constantly decides how much weight to give each one — turning the volume up on some, down on others. That's attention. The technical name is "precision-weighting" (Feldman and Friston, 2010), but the volume-knob picture is accurate: to attend to something is to tell the brain this matters, act on it. Which is why attention changes what you literally perceive, not just what you notice — turn a faint sensation up and it becomes vivid; turn it down and it can vanish. Stage magicians live in that gap; so do chronic worriers, with the volume jammed up on threat. And crucially, the knob is not fully automatic. Some of it you can move.

2 · The part that's real: attention can be trained

The best evidence comes from the oldest technology for it — meditation, tested properly. In the Shamatha Project, one of the most rigorous studies of its kind, people did three months of intensive focused-attention practice — the unglamorous work of putting attention on one thing, noticing it has wandered, and returning it, thousands of times. Katherine MacLean and colleagues (Psychological Science, 2010) found real improvement in sustained attention — holding focus, catching faint signals — still detectable five months later, most in those who kept practising. The encouraging headline: attention behaves partly like a trainable skill, and the training that works is the boring kind.

3 · The part that's oversold: "brain training"

Now the other side, because fairness cuts both ways. The multi-billion-dollar "brain-training" industry promises its games make you sharper across your whole life. In 2016 a large group led by Daniel Simons reviewed the entire evidence base (Psychological Science in the Public Interest) and concluded bluntly: you get better at the specific game you practise, and there's little to no good evidence it transfers to real-world attention, work or thinking. The key distinction — carry it everywhere — is near transfer (getting better at the practised thing, and things like it: easy, real) versus far transfer (a narrow exercise improving your life broadly: rare, mostly unproven). Practising a puzzle makes you good at the puzzle, not at being a focused person.

4 · Show the work — the argument the field is actually having

Here's where a real review has to name the dispute, because "meditation trains attention" is not as settled as the wellness world implies.

In 2018 a group of contemplative-science researchers — not opponents, insiders — published "Mind the Hype" (Van Dam and colleagues, Perspectives on Psychological Science). Their charge, from inside the field: "mindfulness" is defined so loosely it means different things in different studies; much of the research uses weak designs (no active control, small samples, self-report); effects are often modest; and overselling can harm — a minority of people have adverse reactions, and hype sets people up to fail. An earlier large review (Goyal and colleagues, JAMA Internal Medicine, 2014) had already found only moderate evidence that meditation programmes improve anxiety, depression and pain, and little good evidence they beat active alternatives for attention and mood.

So hold both: attention is trainable (§2), and the meditation-improves-everything literature is weaker and hypier than its reputation. The honest position isn't "meditation works" or "meditation is bunk" — it's that the specific, boring practice of returning a wandering mind has real support, while the broad wellness claims run ahead of the evidence.

5 · Underseen / foreign-language: a Chinese programme that ran the tighter experiments

The Western account fixates on American mindfulness (MBSR) and misses one of the better-controlled attention-training programmes, developed in China: Integrative Body-Mind Training (IBMT), studied in a series of randomised trials by Yi-Yuan Tang and Michael Posner. IBMT is deliberately not effortful focus — it emphasises a relaxed, accepting, whole-body awareness guided by a coach. Their randomised studies report improvements in attention and self-regulation after as few as five short sessions (20–30 minutes), alongside changes in the brain's self-regulation circuitry (anterior cingulate). It's rigorous, it's fast, it uses active control groups — and it's largely invisible in the English-language popular story, partly because it doesn't fit the "sit and grind your focus" template. It also suggests the trainable ingredient may be state-setting (getting into a receptive mode) as much as effortful holding — which matters for what "training attention" even means.

6 · The proposition (my synthesis, not established science)

Put the volume-knob model, the near/far-transfer split, the "return" that improves with practice, and IBMT's fast state-setting together:

Attention is the allocation of a limited precision budget, and the trainable part is the policy that re-allocates it — the return — not the raw hold. You will never stop your mind wandering; that's how the machine works. What improves with practice is noticing the drift sooner and re-pointing the budget. This predicts why meditation transfers better than brain-training games: meditation trains the general re-allocation policy in messy conditions, while a game trains a fixed task. Far transfer should appear only when what you train is the policy, not a specific task.

Flagged as interpretation — and, usefully, testable.

7 · The testable version — a study you could actually run

Modelled on the Optionality and self studies in this programme.

The build. An agent with a fixed "precision budget" it must spread across several signals to do tasks. Give it two trainable regimes: (A) practise one specific task ("brain-training game"); (B) practise re-allocating the budget after a simulated lapse ("the return").

Two pre-registered predictions:

• P1 (near beats far for task-training): regime A improves the trained task but shows little transfer to new tasks. Refuted if training one task broadly transfers.
• P2 (training the policy is what transfers): regime B — training the re-allocation policy itself — produces measurable transfer to new tasks, where regime A does not. Refuted if training the return doesn't transfer either, or if it's no better than task-training.

Why run it. If it holds, "the trainable thing is the return, not the hold" survives its first test and reframes most focus advice (which aims at the wrong target); it's then worth looking for the signature in real attention-training data. If it fails, the proposition is wrong. Cheap, falsifiable, a candidate for the next registered simulation and the paid Proposition. (A simulation can't prove human attention works this way — it tests whether the idea is coherent enough to take to data.)

8 · Why it touches your life — and the limits

• Train the return, not the grip. The skill is noticing sooner and coming back more kindly, not clenching harder. Trying harder tightens the very tension that scatters you.
• "Just focus" is the wrong instruction — and IBMT's result hints why: a receptive, relaxed state may set attention better than effortful holding.
• Where you point it is a moral choice, not only a productivity one. If you become what you attend to, then reclaiming attention from the machines built to seize it, toward the people and work that matter, is one of the few genuinely high-leverage freedoms you have.

Two honest limits. Some attention difficulties are not a training problem at all — conditions like ADHD are real differences in the machinery, and "just meditate" is not a treatment and can be one more thing to fail at. And the attention-training literature, like most of psychology, is stronger on direction than on size: it shows attention improves with practice more clearly than it pins down how much, for whom, and how lasting. Take the good news as real and bounded.

Through the RFT lens

(My own framework — a way of seeing, not established science.)

Recursive Field Theory treats attention as the setting that decides which signals a system feeds back into itself — and therefore which patterns get reinforced and which fade. "Training attention" becomes choosing what you recursively feed, which is why a small steady practice can, over months, change a person more than a big one-off effort. §6–7 make that testable. Offered as interpretation.

What to ask next

1. The testable one (§7): does training the re-allocation policy transfer where training a task does not?
2. What's the smallest daily dose that reliably changes anything? IBMT says maybe very little — the honest number matters more than the promise.
3. Can you protect attention structurally, not by willpower — changing the environment so the volume knob isn't hijacked in the first place?

How this was made / what I checked

Verification pass before publishing. The meditation evidence was not left in its flattering form: §4 puts the insiders' own "Mind the Hype" critique and the Goyal 2014 moderate-evidence finding directly against the encouraging Shamatha result, because "meditation trains attention" is the claim most likely to survive an unchecked edit. Near- vs far-transfer is stated per Simons 2016, not overstated. The IBMT effect sizes are given as "improvement after as few as five sessions" rather than a precise figure, pending confirmation against Tang & Posner's trials. The proposition in §6–7 is labelled mine, with refutation conditions.

CPD — structured learning

Format & time. Reflective structured learning · ~30–45 minutes.

Learning outcomes. After this you will be able to: explain attention as active signal-weighting rather than a passive spotlight; use the near/far-transfer distinction to judge any "trains your brain" claim; and state honestly what the evidence supports about training attention (and the field's own "Mind the Hype" caution).

Reflection prompts (write 3–5 lines each, keep for your CPD record):

• Track, for one day, what pulls your attention most. Did you choose those things, or were they chosen for you? By what?
• A client says "I just need to focus more." Using "train the return, not the grip," how would you reframe the goal into something trainable?
• Name one attention or brain-training claim you've half-believed. Does it promise near transfer (plausible) or far transfer (be sceptical)? What evidence would you want?

CPD-eligible structured learning; self-certified, not statutory-regulator endorsement. Log only genuine time spent. Not medical advice; attention difficulties can have clinical causes best assessed by a professional.