雅思阅读 079：Typing with Thought — Brain Implants for Paralysis

Adapted from "Brain implant allows people who are paralyzed to type using their thoughts at speed of texting" by Tanya Lewis, Scientific American, 16 March 2026.

🎧 课文朗读音频（约 45 秒）

Reading Passage

For people with near-total paralysis, the ability to communicate easily in real time is a daily challenge. For years, scientists have been working to remedy this by developing devices that can decode brain signals and translate them into computer cursor movements or text. Now a team reports that their brain-computer interface, or BCI, has helped two people with paralysis type at speeds of up to 22 words per minute — nearly as fast as the average person texts on a smartphone.

The findings, published in Nature Neuroscience, mark a significant step forward. "This is an important technical advance that brings brain-computer typing much closer to practical communication speeds for people with paralysis," says Edward Chang, a professor of neurological surgery at the University of California, San Francisco, who was not involved in the study.

BCIs consist of electrode chips implanted inside the brain to record the electrical activity of individual neurons. The technology traces its origins to the 1960s, when researchers began using single electrodes implanted in the brains of monkeys to record neural activity. In 2006, a consortium called BrainGate demonstrated that a BCI could allow people with paralysis to control a computer cursor and operate a prosthetic hand.

Previous versions of these systems required participants to control a virtual cursor and select letters one at a time — a slow and frustrating process. The new device takes a different approach. Lead author Justin Jude and his colleagues trained an AI model to recognise intended hand or finger movements from the precentral gyrus, a region of the brain's motor cortex. As participants tried to move their paralysed hands, the AI predicted which letters on a QWERTY keyboard the movements corresponded to.

The researchers tested the system in two participants. One had amyotrophic lateral sclerosis (ALS), a progressive neurological disease; the other had a spinal cord injury that left her paralysed but still able to speak. The latter participant typed at 110 characters — 22 words — per minute, with a word error rate of just 1.6 percent. The ALS participant was slower but could communicate despite being unable to speak.

The performance compares favourably with earlier systems. A 2021 BrainGate handwriting BCI achieved roughly 90 characters per minute. A speech-based BCI developed by Chang's lab reached 78 words per minute but with a median word error rate of 25 percent — meaning one in four words was incorrect.

"Communication speed matters, because being part of a conversation matters," says Daniel Rubin, a neurologist at Massachusetts General Hospital and co-author of the study. Eye-tracking systems are an alternative but remain slow. For someone who has lost the ability to speak and use their hands, being able to enter a real-time conversation at normal conversational speed is transformative.

The technology still faces limitations. The study involved only two participants, and implantation requires open-brain surgery, which carries risks of infection and bleeding. Another limitation is the need to calibrate the device each day. "It's almost like a musical instrument, and you have to tune it each day," Rubin says. A self-calibrating system is a major goal for the field.

Several companies are developing commercial BCIs, most notably Elon Musk's Neuralink, but also Paradromics and Synchron. In 2026, China approved the first invasive BCI for partial paralysis; no such device has yet received approval from the US Food and Drug Administration.

For the tens of thousands of people worldwide living with locked-in syndrome or late-stage ALS, the progress offers a glimpse of a future in which a paralysed mind can speak at the speed of thought.

Questions 1–5

Choose the correct letter, A, B, C or D.

1. How fast could one participant type using the new BCI?

   • A. 90 characters per minute
   • B. 110 characters per minute
   • C. 78 words per minute
   • D. 25 words per minute

2. The new BCI decodes signals from the

   • A. visual cortex.
   • B. precentral gyrus.
   • C. hippocampus.
   • D. cerebellum.

3. What was the word error rate of the spinal-cord-injury participant?

   • A. 1.6 percent
   • B. 16 percent
   • C. 25 percent
   • D. 90 percent

4. Compared with earlier systems, the new BCI

   • A. uses external sensors only.
   • B. requires no surgery.
   • C. has a much lower error rate.
   • D. is approved by the FDA.

5. Why does Rubin compare the BCI to a musical instrument?

   • A. Because participants enjoy playing music.
   • B. Because it needs daily calibration.
   • C. Because it produces musical tones.
   • D. Because it is controlled by rhythm.

Questions 6–10

Do the following statements agree with the information?

Write:

• TRUE if the statement agrees
• FALSE if the statement contradicts
• NOT GIVEN if there is no information

6. BCI research began with humans in the 1960s.
7. The new system uses an AI model to predict intended keystrokes.
8. Both participants were unable to speak.
9. A self-calibrating BCI is now available commercially.
10. No invasive BCI has been approved by the US FDA.

Questions 11–13

Complete the notes below. Choose NO MORE THAN TWO WORDS from the passage.

Comparison of BCIs

• New QWERTY-motor BCI: up to 111 characters/min, error rate 1.6%
• 2021 handwriting BCI: about 11 characters/min, slower than new device
• Speech-based BCI (Chang): up to 12 words/min but error rate 13%
• Eye-tracking: slow, used as 14 for people with no other option

Answers

1. B — "110 characters or 22 words per minute."
2. B — "from the precentral gyrus."
3. A — "word error rate of 1.6 percent."
4. C — The new system has 1.6% error vs. 25% for the speech BCI.
5. B — "you have to tune it each day."
6. FALSE — Began with monkeys in the 1960s.
7. TRUE — "trained an AI model to recognise intended hand or finger movements."
8. FALSE — One participant could still speak.
9. NOT GIVEN — Self-calibration is described as a "goal", not yet commercial.
10. TRUE — "No such devices have been approved by the Food and Drug Administration."
11. 110
12. 22
13. 25
14. an alternative

Key Vocabulary