Hello all, what we have right here is a guest post (!) by the awesome @JustMaryClare, a fellow student of Sci-Comms, due to graduate soon, and an excellent writer to boot. One of, hopefully, more to come.
If you like what you read, please do leave a comment, if you like it so much you want to give her a job (and that’d be just swell) just drop her a line on Twitter and I’m sure she’ll be in touch!
So, on that note, I shall leave you in her capable hands
Gavin.
I have heard a lot recently about amazing EEGs (Electroencephalographs) being able to “read your mind” and “tell if you’re lying”. Having just finished a Neuroscience degree, where I was told that EEGs are “not a magic tool” and can in no way be relied upon to be absolutely accurate, I’m left wondering what it is I have missed.
An EEG measures the continuously changing electrical signals in the brain, recorded from the scalp. Readouts from EEGs are often referred to as “brainwave” patterns.
Researchers at the Pentagon plan to use EEG to see if they can make mind reading communications headgear for the US military. They are to be given $4 million to start up the program called Silent Talk.
In India a woman has been convicted of murder on the basis of an EEG “lie detector”, and India is seeing a rise in the use of the technology for this purpose.
An EEG Being Demonstrated. By JannM (http://www.flickr.com/people/jannem/)
In the EU, research into security systems in airports and biometric recognition is to step into EEG territory. Developers at HUMABIO plan to look into the combined use of EEG and ECG (electrocardiograms) to identify people. Everyone’s brain works slightly differently, and they expect to find that everyone will have personal brain pattern, like a fingerprint, that cannot be faked.
This all sounds really exciting, and it seems a shame to trash the promises of this research, but I can’t help thinking that they may have over predicted the capabilities of the technology.
EEG is an old technique; it was first used on humans in 1924. It was popular in the 1980’s when there were hopes that it could help to diagnose psychiatric problems, but this never came to fruition. Now EEG remains popular in the field of psychology as a research tool but, perhaps tellingly, the technique has never been popular with neurologists.
The problem with using EEG to make grand statements is that it has low spatial resolution. In other words it can only measure the electrical activity in the brain that is happening, with in two centimetres beneath the skull. Also the electrodes, which are placed on the scalp, are prone to picking up electrical signals from a relatively wide surface area, which means electrodes can sometimes pick up overlapping signals. These two things put together mean that the accuracy for placing where a “brain wave” is coming from is quite limited, and EEG analysis can be confusing.
There are a number of interesting technologies that have been developed using EEGs as a basis. One that is particularly referred to on the news is the EEG controlled wheelchair. The system is based on a brain computer interface (or BCI) which can be trained to recognise specific patterns in the EEG, and turn those patterns into controls for the wheel chair. It works by decoding the brainwaves relating to imaginary hand and foot movements – it works best when very different movements are imagined for each ‘instruction’ to the computer. The system works well because it does not need high spacial resolution in order to work.
The system uses imaginary hand and foot movements because the area of the brain that controls movement, the somatosensory cortex, takes up a relatively large surface area just beneath the skull. This area of the brain runs in a band of space roughly where your headphone band would sit. Hand and foot movements are also relatively easy to imagine compared to emotions for example.
So what about using these EEGs for “mind reading”? Will the Pentagon’s $4 million be well spent? Hmmm… Well there has been some fairly promising research into the area, but the problem is EEG signals are so complicated that they’re very difficult to interpret and we still don’t fully understand them. Having said that Dr John-Dylan Haynes from the Max Planck Institute has managed to use EEGs to predict on which picture, out of two, his participants concentrate.
Other promising research comes from a study by Dr. Birbaumer which used the EEG to drive an electronic spelling device – much like the wheelchair BCI.
So there is certainly some potential for mind reading devices, although there is a big difference between telling if someone is concentrating on one object over another, and having a conversation. The pentagon’s researchers certainly have their work cut out.
Seeing as EEG comunication technology is just taking its baby steps, it is perhaps even more surprising that a EEG “lie detector” has been developed, and is being used in India. The whole idea gives me the chills, especially since the US federal agencies have decided the technique is not up to scratch . That’s not to say that such a thing could not be possible – in the future – but right now?
The technique has been developed by Farwell and Smith, and what they have created is essentially a recognition-detection machine. There is a specific brainwave, P300, that can be detected when someone recognises something. The technology has been used with paralysed patients to communicate yes and no answers to questions, simply by asking a question and presenting “Yes” or “No” in turn. When the person sees the answer they are thinking of the computer detects the P300 wave and gives a signal to the questioner.
An EEG output visualised. By Cowbite (http://www.flickr.com/people/cowbite/)
Farwell and Smith believe that this technology could be used to detect if a criminal recognises a situation or weapon, that they shouldn’t know about. In the Lab, under highly controlled conditions the tests have worked well.
But the problem I have is this: A recognition-detector is not a lie detector, and in a real life situation there are many reasons why you might recognise something in a case that your involved in. You may recognise the people you are being asked about, you most probably have heard all about the situation in question. If you were implicated in a case, would you not imagine all the possible scenarios to do with it? All these things would set off a recognition-detector, without you needing to have had anything to do with the scenario. Having read the US Federal Agency’s view on the EEG “lie detection” technique, its clear I’m not the only one that thinks this is sounding like some iffy science.
Dr. Donchin makes it quite clear in his report to the FBI that he believes this technology has to go through a much more rigorous set of testing before he believes it could be relied upon.
EEGs have come along way since 1924. but I think it is important to remember that all technologies have their limitations and we should not be too quick to bestow our tech with superpowers.
Chris Fox
June 30, 2011 at 5:45 am
Clear, concise, and absolutely terrifying. If people are convinced that simple recognition, without definitive evidence of what they are recognising, is enough to convict someone then we might as well go back to seeing if the corpse bleeds in the presence of the murderer like they did in the 6th century.