The Upgrade Tree

The branches · The link to machines · at the top of the ladder; nothing left to reach

A headset worn on the scalp controls a device without surgery

In wide use

Electrodes pressed against the scalp pick up the faint electrical hum of the brain through the skull. The signal is coarse, but it is enough to pick a letter, steer a toy or tell when a person is calm, and it needs no surgeon. Headsets that do this have been sold to the public since 2007, as toys, meditation aids and gaming controllers, in the hundreds of thousands. It is the only link to a machine on this branch that is in wide use, and by far the weakest.

What would count

A device worn on the scalp, with no surgery, reads the wearer's brain signals and lets them deliberately control something (a cursor, a toy, a game, a sound), at a rate better than chance; the rung in wide use needs such a device on general sale with a public count of units in the hundreds of thousands or more, from the maker or an independent audit.

Who settles it: Peer-reviewed reports for the first human; the makers' own sales statements or audited figures for the count

Where it stands

In wide use. Reached on 28 August 2014, on the record of store.neurosky.com. At the top of the ladder; nothing left to reach.

The last reading

Once a week every upgrade is read against the news, against what would count above. Open a line to see what that week held. Every reading of this upgrade is on this week's reading.

  • 16 October 2026 Nothing new

    Read monthly, as an entry at the top of the ladder. No withdrawal and no new count in the window; market reports give revenue, not units, and no dated NeuroSky statement of a million chips was found, so that claim stays on the list. On the other claim still to check: secondary sources (Wikipedia's account and Wolpaw's 2011 review) describe Vidal's 1977 paper in Proceedings of the IEEE as volunteers steering through a maze on a screen in real time by attending to corners of a flashing checkerboard, read from scalp signals, which would meet this entry's first-human test eleven years before the 1988 speller; the paper's own text was not read, so the claim is supported but not settled.

Changes to this entry

  • 2026-10-16, stage_reached_on: 2015-12-31 → 2014-08-28. The maker's page is undated and its footer says 2015, so the first date was a guess at the end of that year. The Internet Archive holds a dated copy of the same page, taken on 2014-08-28 (the earliest copy of it), that already carries the sentence 'with over one million in circulation' (https://web.archive.org/web/20140828125802/http://store.neurosky.com/products/eeg-tgam); the date is now that of the earliest dated copy. The statement may be older: a datasheet for the same chip, copyrighted 2011, has the same sentence but carries no date of its own.
  • 2026-10-18, earlier_stages: first-human, 1988-12-01, https://doi.org/10.1016/0013-4694(88)90149-6 → first-human, 1977-05-01, https://doi.org/10.1109/PROC.1977.10542. Jacques Vidal's paper of May 1977 (Proceedings of the IEEE 65(5), pages 633 to 641) was read in full on 2026-10-18. It reports volunteers with electrodes on the scalp, no surgery, steering a symbol through a maze on a screen in real time, with the computer picking the right one of four directions more than 90 per cent of the time (pages 636 to 638), which meets this entry's test as written. The date is the paper's own issue month, by the rule that a study is dated by its publication; the paper dates none of its sessions. The 1988 Illinois report (the first brain-controlled speller) is no longer the first human, and is kept in the notes.

Notes

The first-human record is Jacques Vidal's report from the University of California, Los Angeles (Proceedings of the IEEE 65(5), May 1977, pages 633 to 641, https://doi.org/10.1109/PROC.1977.10542, read in full on 2026-10-18). Volunteers with electrodes on the scalp and ears, and no surgery, moved a symbol through "a simple maze" on a screen in real time (page 637). Each chose a direction by fixing their gaze on one of four dots (left, right, up, down) around a flashing checkerboard; the computer sorted the brain's response to each single flash into one of four classes and moved the symbol accordingly (pages 637 and 638). With four choices, chance is one in four. The paper says the computer was right more than 90 per cent of the time with average subjects, and more than 80 per cent even when early sessions that went badly are counted (page 638). Its Figure 6 shows thirteen sessions by seven lettered subjects (page 638). What the paper does not give: a date for any session (it was received on 1976-07-19, page 633, and some figures carry data labels that read like a subject code and a date in June and July 1976, which the paper does not explain); an account of anyone getting out of the maze; or a test of whether the response needs the eyes to move. The stage is dated by the paper, as the schema dates a study. The 1988 Illinois report of volunteers spelling words by attending to flashing letters (https://doi.org/10.1016/0013-4694(88)90149-6) was this entry's first-human record until 2026-10-18; it is still the first brain-controlled speller. A review in Communications of the ACM (McFarland and Wolpaw, May 2011, page 61, https://doi.org/10.1145/1941487.1941506) describes Vidal's maze work, and Vidal's own 1973 review (Annual Review of Biophysics and Bioengineering 2, pages 157 to 180) only proposed it and said direct brain-computer communication "still lies somewhat in the future". Consumer headsets based on the same kind of signal reached shops in 2007 (NeuroSky), 2009 (a Mattel toy, Emotiv) and 2014 (Muse). The wide-use count is NeuroSky's own statement that over a million of its sensor chips are in circulation; the maker of the Muse headset says it has over half a million users. The hardware is cheap and the control is coarse, which is why this is the one upgrade on the branch anyone can buy.