For telepaths, mentalists and psychics, technology may be approaching what supposed magical powers never reliably achieved. Brain computer interfaces are increasingly able to record and decode neural signals, but genuine mind reading remains a far more distant and complicated prospect.

These systems, known as BCIs, can convert brain activity into speech, movement or other intended actions. Their growing sophistication has made the language of mind reading tempting, even though the technology does not provide unrestricted access to a person’s thoughts.

Earlier claims of psychic ability have repeatedly failed under scrutiny. Uri Geller’s 1970s appearance on The Tonight Show with Johnny Carson and more rigorous scientific studies showed that purported mentalists generally performed no better at telepathic tasks than controls without claimed psychic powers.

BCIs have produced far more tangible results. Monkeys have played Pong using thoughts alone, while people with disabilities have used the technology to regain abilities that they could not otherwise perform.

One of the most common applications is restoring communication for people who cannot speak verbally. These devices typically use electrodes surgically implanted in the brain and can convert intended speech into articulation that sounds remarkably human.

In 2025, researchers presented a BCI that allowed a patient with ALS to alter the intonation of his computer generated voice. The patient could also sing short melodies through the system.

Such devices usually interpret signals from the motor cortex while a user physically attempts to speak. That means they execute an intended action rather than reveal an unrestricted internal monologue, stream of consciousness or deliberately private thought.

One breakthrough study went further by decoding words that participants imagined as well as words they attempted to verbalize. Its ability to detect inner speech prompted the designers to create a mental password that users could imagine to switch the system on and off.

“In terms of decoding speech, we're getting closer,” says Professor of Ethics Frédéric Gilbert from the University of Tasmania. “But it's not the equivalence of this concept of mind reading where you cannot lie to the neurotechnology. It’s not a case of you can’t lie to God.”

The distinction between decoding and reading is central. Current BCIs detect particular signals from a limited group of neurons, then identify patterns that statistically correlate with certain actions.

They cannot inspect a person’s complete brain state and simply announce what is on that person’s mind. “The term mind-reading is a metaphor that scientists often use in order to illustrate some very striking and amazing achievements,” says Gilbert.

“But technically, what this technology does is to record and decipher, not read minds,” Gilbert says. The technology achieves impressive results partly by reducing the immense complexity of human thought to a narrower decoding problem.

A major limitation is reverse inference, which assumes that a particular pattern of brain activity corresponds to a specific thought. Gilbert compares that reasoning to seeing wet grass and concluding that rain must have fallen, even though humidity or water from a neighbor washing a car could explain it.

Similarly, activity in one localized area does not automatically disclose the complete contents of someone’s mind. A system capable of genuine mind reading would have to avoid this reverse inference problem.

Neural signals may indicate the linguistic content of some thoughts, but they do not provide a simple cipher for private hopes, fears, beliefs or opinions. There is no direct relationship between a discrete piece of mental content and a single, definitive pattern of neural activity.

BCIs consequently need feedback from users to verify whether an extracted thought is accurate. “There's no capacity to check the veracity of that extraction without us confirming it,” says Gilbert.

The systems also struggle with thoughts on which they have not been specifically trained. Identifying whether someone instructed to think about either a cat or a dog is imagining one of those options is very different from deciphering unconstrained thought.

Words are only one element of mental life, and not everyone has an internal monologue. Even people who internally narrate their experiences also think through images, feelings and other private phenomena that are difficult to describe.

In 2023, researchers developed a BCI that could describe language inside people’s heads and the meanings of their thoughts as they listened to spoken dialog, told themselves stories and watched silent movies. Even with that relative success, reproducing the full range of a person’s thoughts remains a distant fantasy.

The possibility of extracting private information without consent has nevertheless prompted debate among philosophers and bioethics researchers. Proposed neurorights would create legal limits on how hypothetical systems might access thoughts, similar to protections against technologies that coerce people into revealing information.

“The very possibility that someone can extract private information that you never consented to be extracted is pretty scary,” says Gilbert. He argues that such coercive access should be illegal, while stressing that genuine mind reading may not be remotely possible.