Jan 137 min read

Will Tomorrow's Tech Read Your Mind?

Everyone loves a good magic trick. Take mind-reading, for instance. We can enjoy the stagecraft and illusion safe in the knowledge that it isn’t actually possible for someone to pry into our innermost thoughts. But what if it was? We may be on the verge of a technological breakthrough that takes mind-reading from illusion to reality.

Brain scanning is one of the wonders of medical science. Its history goes back to 1924 and, over the years, the abstract signals and images collected from these scans have helped us to map the structure of the human brain and develop our understanding of how it works. This has been hugely beneficial for developing treatments for neurological conditions like epilepsy. Brain scans can read structures and activity (like blood flow) but they haven’t been able to decipher our thoughts – until now….

Recent trials have returned some jaw-dropping results. Aided by advances in artificial intelligence, brain scanning technology can now see what we’re seeing, hear what we’re hearing and possibly even tap into what we’re thinking.

Watch my short video to find out more...

How brain scans work

The human brain may be the only known thing in the universe sophisticated enough to carry out studies on itself. Neuroimaging technologies developed over the past century have helped scientists discover a huge amount about the workings of the brain. Typically, studies involve putting a participant in a scanner, presenting them with audio / visual stimuli, seeing which parts of the brain activate in response, and then interpreting that correlation in terms of mental states and cognitive processes.

Three key techniques used for brain scanning are tailored to map different types of activity:

EEG (short for electroencephalography)

This measures electrical activity, tracking the waves of electricity produced by brain cells when they communicate. This helps us identify which areas of the brain are active when we are learning or focusing attention. This non-invasive technology has been around for almost 100 years and is used for diagnosing epilepsy.

fMRI (functional magnetic resonance imaging)

fMRI measures the blood oxygen levels in different parts of the brain. This is useful because brain cells consume oxygen more quickly when they are active. While EEG scans can tell when activity is happening in the brain, fMRI scans can pinpoint where, helping to develop our understanding of the different regions of the brain. These two techniques are often used together.

PET (positron emission tomography)

PET is the most precise of these technologies. It detects the radiation given off by a substance injected into the body, called a radiotracer, and uses this information to create a 3D map of the brain. PET scans can be used to study the brain’s complex chemistry, and can help researchers understand the impact of drugs or diagnose diseases like Alzheimer’s. PET scans are often combined with CT (computed tomography) scans which use X-rays to take detailed pictures of the brain.

This TED ED video offers a great and easily digestible five-minute summary of the three technologies:

Recent breakthroughs

In a significant study last year, scientists were able to reconstruct the Pink Floyd song Another Brick in the Wall, Part 1 by listening in on people’s brainwaves. It was the first time a recognisable song had been decoded using recordings of electrical activity in the brain. You can hear the "brain scan" version of the song here.

The experiment used a version of EEG, with electrodes placed on the surface of people’s brains while they were undergoing surgery to treat epilepsy. On this basis, we’re clearly not going to be tuning into each other’s brain playlists any time soon, but scientists believe similar results will one day be achieved using much less invasive techniques. The results are already improving our understanding of how we perceive sound and could lead to the development of much better, and more sophisticated, devices for people with speech difficulties.

In another study, researchers showed videos and images to people undergoing fMRI scans and used the signals to piece them back together – essentially seeing the world through the subjects’ eyes.

This news clip includes some footage where you can see how the reconstructed "mind-reading" clips compare to the originals. Interestingly, the reconstructed clips tend to vary between different people – after all, we all see the world slightly differently.

The success of this study, carried out between the National University of Singapore and the University of Hong Kong, raises the tantalizing prospect of one day being able to record and share our dreams, something that researchers at the University of Texas are busy working on.

The role of AI

These recent breakthroughs all share a common factor: the rapid advancement of AI.

The process of trying to ascribe meaning to the myriad of data taken from brain scans is one of statistical analysis. The power of large language models like those which power ChatGPT or Stable Diffusion (used in the image and video study) is that they are enabling researchers to make much more educated guesses.

The AI models used are trained on data from each individual. That’s to say, an AI trained on data from a scan of my brain, wouldn’t be able to decipher thoughts from a scan of yours.

Even with the huge progress we’ve made mapping and monitoring the human brain, there is still plenty that remains a mystery. And that’s before we even get into the "hard problem" of trying to explain consciousness.

Does a dead salmon think?

Because brain scanning is an inexact science, based on approximations, it can sometimes churn out wild and surprising results.

Among the thousands of data points produced by a brain scan, some will inevitably be noise, errors or the result of random chance. And, with such a large amount of data to analyse, there is a risk that some of these readings may be over-analysed and lead to false positives. This is commonly known as the multiple comparisons problem.

Neuroscientist Craig Bennett demonstrated this to comedic effect when he put a supermarket-bought salmon through an fMRI scan and asked it to perform a task – watch a series of images of human faces and determine their emotions. You won’t be surprised to learn that this exercise didn’t have the immediate appearance of being a success given that the subject was a) a salmon and b) dead.

The results of the scan were locked away for a couple of years until one day a conversation about multiple comparisons prompted Bennett to take another look.

I ran the fish data through my [statistical] processing pipelines and couldn’t believe what I saw. Sure, there were some false positives. [...] Rather, it was where the false positives occurred that really floored me. A cluster of three significant voxels [data points] were arranged together right along the midline of the salmon’s brain.

If the results of this analysis were to be believed, the salmon had actually been thinking about those images Bennett displayed in the fMRI machine.

The findings of the study were published in full in the Journal of Serendipitous and Unexpected Results and later won Bennett and his team an Ig Nobel prize.

The lesson of the dead salmon experiment is not to discredit fMRI scans but to highlight the need for caution when analysing results.

Brain scans are generating some incredible findings. But, if the next evolution of brain scanning puts the technology to more use in everyday life we will need to be very aware of its limitations.

From mind-reading AirPods to a brain-hacking dystopia?

In July last year Apple quietly announced its latest innovation, not with one of its famously slick events, but with a filing at the US Patent & Trademark Office.

The patent application details a next generation version of the company’s AirPod earphones with built in electrodes that would enable it to function as an EEG scanner, measuring electrical activity in the wearer’s brain.

An image from Apple’s patent showing an earphone equipped with electrodes

It offers a glimpse of the potential future of this technology where we may soon be able to control our digital devices with our minds.

But there’s another, more scary, version of where this may lead to: our devices and the organisations behind them gain unprecedented access to our thoughts.

Professor Nita Farahany, a leading academic on the ethical, legal and social implications of emerging technologies, believes that "a new dawn of brain-tracking and hacking is coming." In her 2023 book The Battle for Your Brain, Farahany states:

Imagine a world where your brain can be interrogated to learn your political beliefs, thought crimes are punishable by law, and your own feelings can be used against you. A world where people who suffer from epilepsy receive alerts moments before a seizure, and the average person can peer into their own mind to eliminate painful memories or cure addictions.

She continues:

Neuroscience has already made all of the above possible today, and neurotechnology will soon become the 'universal controller' for all of our interactions with technology. This can benefit humanity immensely, but without safeguards, it can severely threaten our fundamental human rights to privacy, freedom of thought, and self-determination.

Farahany is not ideologically opposed to neurological interventions and has been open about using EEG headgear to battle her migraines. But she is one of many who say that governments have been much too slow in regulating technology firms which already have the means to predict, shape and monetise our thoughts.

Shweta Singh, an assistant professor of information systems and management at the University of Warwick, warns that:

In essence, TikTok doesn’t necessarily read your mind, it in fact makes your mind. These algorithms encourage you to form opinions and nudge you towards a herd and therefore have immense power to reinforce and worsen herd mentality.

The evolution of mind-reading from the realm of magic tricks to the real world could bring huge benefits, from tracking and treating neurological conditions to preventing accidents caused by tiredness and fatigue. But if we’re to avoid it leading to the kind of dystopian future some experts fear, we’ll first need to be confident that we can control who has access to our thoughts.