Attention

William James opened his chapter on attention in the Principles of Psychology (1890) with a sentence that has aged like a provocation: “Everyone knows what attention is.” In the 135 years since, the science of attention has mostly been a sustained argument about what James thought everyone knew.

Here’s what the argument has established, at minimum: attention is not a thing you have. It’s an outcome.

In 1995, the neuroscientists Robert Desimone and John Duncan proposed what’s now called the biased competition model. Neural populations representing different objects or locations compete for processing resources through mutual inhibition. The winner gets amplified and routed to higher cortical areas.

Attention, in this model, is the bias — the thumb on the scale. Top-down signals from the prefrontal and parietal cortex can pre-load the competition in favor of whatever you’re looking for. Bottom-up salience — sudden movement, high contrast, a loud sound — can win the competition without any top-down involvement at all.

When you try to focus on something, you’re not directing a beam. You’re tilting an arena.

The default condition of that arena, absent deliberate tilting, is to wander. In 2010, Matthew Killingsworth and Daniel Gilbert at Harvard gave 2,250 people a smartphone app that interrupted them at random moments throughout the day and asked two things: what are you doing, and are you actually thinking about it?

Part of why the default is so persistent may be that attention has a temporal dimension people rarely consider. In a rapid stream of visual stimuli — images flashing at about ten per second — if you detect one target, you briefly lose the ability to detect a second target appearing right behind it.

For roughly a quarter of a second, a second target passes through the visual system, gets processed at some level, and simply does not make it to experience. The perceptual system sees it. You don’t.

The fullest demonstration of what attention does — and therefore what its absence costs — comes from people who have lost access to half of it. After damage to the right parietal cortex, patients develop hemispatial neglect: they stop attending to the left side of space.

Not because they’re blind in the left visual field. The eyes work. The information arrives. But attention stops being allocated there, and so the left side of the world effectively ceases to exist as a lived reality. Patients eat only the right side of a plate. Asked to copy a drawing, they render only the right half. Asked to bisect a line, they mark well to the right of center.

In 1978, the Italian neuropsychologists Edoardo Bisiach and Claudio Luzzatti asked neglect patients to describe a familiar piazza in Milan from memory — first imagined from one end, then from the other. From each imagined vantage point, patients reported only the buildings that fell in their attended hemispace. The inattended half didn’t just go unnoticed in the room; it was absent even in memory, even in imagination. Without attentional allocation, reality contracts.

Herbert Simon saw the implications of this contraction in 1971, before there was much to see it in. He was writing a chapter on designing organizations for a world that was about to drown in information — when the internet didn’t exist and the smartphone was forty years away.

The prediction was essentially complete. As information became abundant, attention would become the scarce resource — and whoever understood that scarcity would organize around exploiting it. The platforms that have restructured digital life in the decades since are not really in the news business, or the entertainment business, or the social business. They’re in the business of winning the biased competition — supplying the bottom-up salience that wins the arena without your consent.

What’s strange, in retrospect, is that when engineers building artificial intelligence needed the key operation for processing language, they independently rediscovered attention — and named it that.

In 2017, a team at Google Brain published a paper called Attention Is All You Need, introducing the transformer architecture. The attention mechanism, technically, is a way for a model to dynamically weight which parts of an input sequence matter for producing each output — a query matched against keys, a weighted sum over values, a learned selection over representations. The name was chosen because the operation resembles what biological attention does: selective weighting of what matters and suppression of what doesn’t.

It turned out to be the key to the most consequential AI systems in history.

The Buddhist tradition arrived at a similar conclusion from the opposite direction, with neither neuroscience nor neural networks to lean on.

In April 1942, Simone Weil wrote a letter to the poet Joë Bousquet that contained a definition no neuroscientist would have written, and no neuroscientist has improved on.

Weil had been thinking about attention for years — not as a cognitive resource but as a moral act. Genuine attention, for her, required suspending the self: setting aside one’s own projects, preoccupations, and ready interpretations in order to be truly available to what is in front of you. That is why it counts as giving. It costs the giver something real — the same scarce, finite capacity the neuroscientists measure in bits and milliseconds. To attend fully to another person is to spend, on them, the very thing the feed is built to take.

Simon and Weil are describing the same scarcity: a finite mind that can be fully present to only a sliver of what reaches it. But their moral orientations are inverted. Simon’s attention is extracted — a resource that information consumes. Weil’s attention is given — a gift the self extends outward. One is mined; one is offered. Both framings are accurate about the bottleneck the neuroscience describes.

The question they leave open — and it may be the most important question attention raises — is which verb you think should govern it. The science can tell you the resource is scarce, that the mind wanders half the time, that catching one thing blinds you to the next, that allocation is the difference between a world that exists and one that quietly doesn’t. It cannot tell you whether the scarce thing is yours to be harvested or yours to give away. That part was never a finding. It was always a choice.