Gandhinagar/New Delhi: One of the authors of this piece is sitting in a chair with brown upholstery, in the middle of a small, unremarkable room in an equally unremarkable building in Gandinagar.
There is what looks like a television in front of him, a tight cap with electrodes is constricting his skull, and there are more electrodes on his face to record every twitch of the skin and movement of the eye.
This equipment in the Directorate of Forensic Science (DFS) in Gandhinagar has helped put many a criminal behind bars. And now, the Supreme Court has ruled that its use is unconstitutional without the consent of the subject.
Questions surrounding the use of technologies such as narco analysis (where answers are obtained after putting subjects in a chemically induced stupor) and brain mapping are becoming more relevant with little public knowledge of the methods involved, even though investigative agencies increasingly rely on these techniques.
Methods such as the brain electrical oscillation signature (BEOS) and another American technique called brain fingerprinting have traditionally polarized both academic and legal circles, although in India, investigations refer to them with the utmost awe—of the sort that is usually reserved for absolute truth.
The apex court’s order on Wednesday has stopped short the march of such techniques in India—but only just. The order may have been prompted by issues related to individual rights, but some neuro-psychologists and other experts insist that not enough is known about the brain and how it stores and processes memory to allow the use of these technologies to convict people. They point to the fact that BEOS, which is in use in Gujarat and Mumbai, has not been peer-reviewed. A Union government panel in 2008 studied the method and ruled that it was “unscientific” and unfit for use in forensic applications, say panel members.
The method’s proponents, meanwhile, say that the method’s benefits offset concerns over the technology and that a validation exercise conducted by the department of science and technology (DST) on 120 subjects showed an accuracy rate of 95%.
Illustration: Jayachandran / Mint
Meanwhile, in the drab room in DFS, one of the authors is getting his hair wiped down with spirits to get any traces of oil or dust off. On the table in front of him, a red light on the webcam winks. He is, to use popular parlance, about to get brain mapped.
Response and crime
BEOS is the brainchild of Champadi Raman Mukundan, a former professor of psychology at Bangalore’s National Institute of Mental Health and Neuro Sciences (Nimhans), who evolved a method to interpret variance in the brain’s electrical activity as a forensic tool in criminal investigations. This system is underpinned by the theory that there are two kinds of knowledge: experiential knowledge, which comes from deliberate participation, and conceptual knowledge, or knowledge from hearsay, reading or third-person accounts.
BEOS works by comparing the electrical pulses given off when the subject is introduced to an experience that he or she has undergone in the past, with the pulses given off to stimuli that directly relate to the case and which only the criminal is likely to know.
Subjects are administered a sequence of stimuli, called probes, which could be in the form of spoken sentences piped into the room, or visual stimuli displayed on TV. The technology doesn’t require responses from subjects—something BEOS creators tout as an advantage. Still, lawyers say the technique infringes a suspects’ right against incriminating themselves.
The probes relating to the case are interspersed with probes that are more general, and serve to establish a pattern of electrical activity.
For instance, the subject is administered probes that are universally known to be true and which are neutral to the subject. Other probes may be specific to the subject, but neutral in nature, such as questions on his past, answers to which are already known to the investigators. These serve as test patterns for detecting experiential knowledge of the crime. If similar patterns are seen when crime-related probes are administered, the subject is said to have experiential knowledge of the crime.
Axxonet System Technologies Pvt. Ltd, headquartered out of Bangalore, is the only Indian firm that develops systems based on BEOS. In 2006, it installed its first systems in the Gujarat State Forensic Laboratory and later at the Mumbai forensic lab in Kalina.
Chetan Mukundan, the professor’s son and one of Axxonet’s directors, says the proof of the efficacy of the technology lies in the investigative agencies’ whole hearted acceptance of this technology.
Victim or the crime?
Scientists argue that there are three major problems with basing a suspect’s guilt on readings taken through skull electrodes. The electrodes in the skull cap nestle close to the scalp. Beneath the scalp is the skull itself. Beneath the bone is the dura mater, a tough leather like substance that holds in cerebrospinal fluid, and beneath that is the brain itself. Each of these layers end up distorting or blurring the electrical activity in the brain that the electrodes pick up, they claim.
Brain activity typically measures in tens of microvolts (which is one millionth of a volt). The typical AA battery puts out 1.5 volts.
“There is too much noise,” says Mahesh Jayachandra, a neuroscientist and expert on the use of electroencephalography (EEG) for mapping brain activity, adding that all the layers in the middle could distort these small pulses.
Measuring as it does in microvolts, it is hard to gauge how much of the electrical activity is unique to a particular stimulus, he adds.
The problem is that despite all the advances in measuring brain activity, scientists’ understanding of the concept of memory is characterized more by what they don’t know.
“We do not have a general theory of the brain. We know how the plumbing works. We haven’t figured out where it (memory) is stored,” Jayachandra says.
Opponents of brain mapping also says that the terms “experiential” and “conceptual knowledge” do not fall in any generally known categories of memory.
“There is no accepted division on what constitutes so-called experiential and conceptual knowledge. Neural activity is today almost complete noise,” says N. Pradhan, a professor in Nimhans’ psycho-pharmacology department and Mukundan’s ex-colleague.
In defence, C.R. Mukundan says that his technology is already well published in previous research. “ There’s nothing completely new in our approach. Beginning from the 60s, there are 17-18 studies that associate neural signatures with remembrance and knowing. I’ve in a sense built on this, got some new analyses of electrical pulses on frequency and time domain and created a software”. He adds that he plans to publish his results more explicitly next year.
Pradhan was part of the so-called Nagaraj (a former director of Nimhans) panel, which in 2008 disputed BEOS’s claims. Mint could not independently verify the report’s contents.
Let’s say the stimulus is a sound that goes “bip, bip, bip” at regular intervals, elaborates Jayachandra. Throw in a “boop” at random intervals. There would be crazy electrical activity at the “boop”, because the brain recognizes the difference in the stimulus. And this is just with auditory stimulus. If the stimulus were a picture with the colour red, or a picture of the crime scene, the electrical activity in the brain would be huge, he adds. Prior experiments have shown this sort of reaction to the colour red, he adds.
“It is a perfectly valid research tool,” says Jayachandra. But perhaps not good enough to decide somebody’s guilt.
Others point to the huge leap of faith it requires to uniquely associate a specific memory to a train of electrical pulses.
For instance, says Narayanan Srinivasan, a cognitive scientist at Allahabad University, you could statistically show that men in general are taller than women. “But if I randomly picked a person’s height, I can’t be sure if it’s a man or a woman. There’s a huge probability assumption we have to make.”
Chetan Mukundan, however, says most scientists who critique his firm’s technology haven’t bothered to understand it. The Nagaraj panel, he says, didn’t visit its labs.
“We weren’t given a chance to demonstrate our technology,” he adds. “We were sent some questionnaires, which we filled and duly returned. But none of the scientists bothered to visit the labs.”
We know what you did
Meanwhile, the brain mapping exercise on one of the authors shows that he seems to have experiential knowledge of walking towards a cupboard—part of a sequence of stimuli that included walking towards the cupboard; opening it; retrieving a rod; and breaking a piggy bank with it.
A scientist conducting the experiment, who does not want to be identified, says showing experiential knowledge in isolated probes does not indicate involvement in a crime. The subject has to be shown to have experiential knowledge in the entire sequence of crime-related probes, that are interspersed with other neutral probes.
On the other side of the test room, monitors show the dense squiggles of EEG. The EEGs are isolated into 10-second bursts, with markers for the questions. A video of the author as the probe is administered is also coded into the software. In parallel, dozens of head maps such as those seen in TV medical dramas show his brain activity through swirls of bright red, green and blue. Red indicates lots of activity and blue shows no activity.
Mukundan’s system uses a statistical tool to study deviances in the brain’s electrical activity from patterns established through the neutral probes. It discounts for what are called artefacts, which is when the brain’s pulses may be disrupted by the twitch of an eyelid, a facial tic or even the simple act of swallowing.
Too many variables
Scientists argue that when it comes to the human brain, there are too many things they still don’t know.
A former director of the Mumbai forensic lab, Rukmani Krishnamurthy, who supervised the installation of BEOS systems at the lab, says several training sessions were conducted and results of the efficacy of the systems were detailed in several conferences. However, independent scientists say they haven’t read any of this information.
“The biggest problem with the technique is that there are no peer reviewed publications,” says Srinivasan.
While Srinivasan admits he didn’t personally visit any of the labs where BEOS was used, he says written information furnished by the companies was insufficient for him to be convinced of the validity of this technique.
The ministry of home affairs, which set up the central panel, didn’t accept the report and chose to dissolve the panel.
Axxonet’s secrecy on the technology, says Mukundan, is triggered entirely by fears that their technology will be stolen. “We’ve filed for patents, but revealing too much about the tech would mean somebody could easily reverse-engineer our technique and walk away with it.”
Though he admits that results describing the technology in detail are yet to be published in peer-reviewed journals, he says a study sponsored by DST’s technology information, forecasting and assessment council (TIFAC), and conducted in the Gujarat forensic lab, is proof of the technology’s efficacy.
The study, conducted in 2007, essentially says that on a sample of 120 people who participated in a validation study of the experiment, the brain mapping exercise returned a 95% accuracy. J.M. Vyas, the director of the Gandhinagar lab, refused to share the contents of the report citing confidentiality deals with TIFAC.
Even as the Supreme Court has ruled on the legality of such tests, the debate over their scientific validity continues.
Interestingly, in the US, there had been no explicit ruling on the validity of brain mapping. However, the Government Accountability Office and the Federal Bureau of Investigation have said that several years are needed before it can be used as a useful aid for investigation.