Can a machine read your mind to prove guilt?

The police wanted to hook Abdul Karim Telgi up to a machine that reads brain waves. The court said yes—but only if he agrees.

The year was 2004. Telgi, the mastermind behind India's biggest fake stamp paper scam—a fraud that had cheated the government of crores in revenue—was in custody. The Special Investigation Team (SIT) had documents. They had witnesses. But they wanted more. They wanted to go inside his head.

So they asked a special court for permission to run three tests on Telgi and his co-accused: a lie-detector test (polygraph), a narco-analysis test (where a drug is injected to make the person semi-conscious and supposedly unable to lie), and a P300 brain fingerprinting test (a machine that measures brain waves to check if a person recognises specific details of a crime). The special court said yes. The accused appealed to the Bombay High Court.

And that is where the real fight began—a fight that would eventually reach the Supreme Court and reshape how Indian courts think about science, the human mind, and the right to remain silent.

When the machine promised certainty

The Bombay High Court, in Ramchandra Reddy and Ors. v. State of Maharashtra, was asked a simple question: could the State force an accused person to undergo brain fingerprinting and narco-analysis?

The accused argued that these tests violated their fundamental right against self-incrimination (the constitutional protection that says no person can be forced to be a witness against themselves). The tests, they said, were unreliable, unscientific, and essentially a form of torture.

The State argued differently. Modern crime required modern tools. The fake stamp paper scam was vast—hundreds of crores, multiple states, dozens of accused. Traditional investigation had hit a wall. These scientific tests could unlock the truth.

The Bombay High Court agreed. It observed that modern technology necessitates the use of scientific methods to aid investigation and judgment formation. The court upheld the legality of using these complex techniques and maintained that evidence procured under the effect of truth serum is also admissible.

The tests could go ahead—but only with the accused's consent. The court said yes, but only if he agrees.

The drug that loosens the tongue

Narco-analysis is not a gentle procedure. A drug called sodium pentothal—sometimes called "truth serum"—is injected into the accused. The drug puts the person in a semi-conscious, highly suggestible state. The theory is that in this state, a person cannot construct lies. They can only tell the truth.

The accused was placed in a semi-conscious state. The interrogators leaned in, waiting for the truth to spill out. But here is the problem: the person is also hallucinating, confused, and mixing memory with fantasy. The "truth" that comes out is often a jumble of real memories, imagined scenarios, and things the interrogator suggests. It is, as the Supreme Court would later note, deeply unreliable.

Brain fingerprinting works differently. The accused is shown images or words related to the crime—a photograph of the victim's house, the weapon used, the amount stolen. A cap with electrodes measures electrical signals in the brain. If the brain recognises the detail, it produces a specific wave pattern called P300. The machine says: this person knows this detail.

The technician adjusted the electrode cap. On the monitor, a flat line suddenly spiked—P300. The machine had found a match. But knowing a detail is not the same as committing a crime. A police officer who investigated the case also knows the details. A journalist who read the case file knows them. The machine cannot tell the difference between a guilty memory and an innocent one.

Why the Supreme Court changed its mind

The Bombay High Court's decision in Ramchandra Reddy was not the final word. A few years later, in Selvi & Ors v. State of Karnataka & Anr, the Supreme Court revisited the entire question. This time, the court looked not just at whether these tests were useful, but whether they could be forced on an accused person.

The Supreme Court asked a deeper question: is a brain wave a "testimonial" act? The Constitution protects a person from being forced to give "testimonial evidence"—that is, evidence that comes from their own mouth or their own mind. A blood sample is physical evidence; you can be forced to give it. But a statement—even a statement made through brain waves—is testimonial. You cannot be forced to give it.

The court held that the compulsory usage of brain mapping and polygraph tests in a criminal investigation would be unconstitutional. Consequently, these complex tests cannot be made a part of the compulsory investigation process.

The court also questioned the science itself. Because the scientific tests were inconclusive and the evidence unreliable due to the subject's semi-conscious state, the court ruled against mandatory usage. As the Supreme Court observed, "the person giving statements during narco-analysis was in a semi-conscious state, making the evidence unreliable."

The line between science and compulsion

The two cases—Ramchandra Reddy and Selvi—stand on opposite sides of a critical line. The Bombay High Court said: these tests are legal, and the evidence is admissible, as long as the accused agrees. The Supreme Court said: these tests cannot be forced, because they violate the right against self-incrimination, and because the science is not reliable enough to justify the intrusion.

Both courts agreed on one thing: the science is complex. Brain fingerprinting, narco-analysis, polygraph—these are not simple tools like a blood test or a fingerprint. They involve the mind, consciousness, memory, and the very nature of truth. When the subject matter is complex but the scientific method itself is insufficiently documented, empirical, or reliable, the court must reject its compulsory use to ensure correct judgment formation and safeguard fundamental rights.

There is a deeper question here that neither court fully resolved. If the science of brain fingerprinting improves—if one day the P300 test can reliably distinguish between a guilty memory and an innocent one—would the constitutional objection still hold? The Supreme Court in Selvi left that door slightly ajar. Its ruling rested not just on the unreliability of the science, but on the fundamental principle that no person should be forced to become a witness against themselves. Even a perfect mind-reading machine, the logic suggests, would still violate that right if used without consent.

But what about consent given under pressure? An accused person sitting in a jail cell, facing a life sentence, is not exactly in a position to refuse a "voluntary" test. The police want it. The court has approved it. The accused is told: if you agree, it will help your case. Is that truly free consent? The Bombay High Court did not address this. The Supreme Court, by banning compulsion outright, created a cleaner rule—but the grey area of "voluntary but pressured" consent remains.

Consider also the practical challenge for investigators. A narco-analysis session requires a trained anaesthesiologist, a forensic psychologist, a video recording setup, and a sterile medical environment. Brain fingerprinting requires a specialist technician and expensive equipment. Most police stations in India have none of these. The cost and complexity of these tests mean they are available only to the wealthy accused or the most high-profile cases. Telgi's case was exceptional. For the ordinary accused, the debate over brain fingerprinting is academic—they will never be offered the test.

And yet the principle matters. The Selvi judgment is not just about narco-analysis and brain mapping. It is about the limits of state power over the human mind. The Constitution draws a line around the body—you can be fingerprinted, photographed, and have your blood drawn. But it draws a tighter line around the mind. The state can observe your body. It cannot extract your thoughts.

This distinction is becoming harder to maintain as technology advances. Brain-computer interfaces, functional MRI scans that can detect lies, and AI-powered analysis of speech patterns are all on the horizon. The Selvi framework—consent is required, compulsion is banned—provides a starting point for judging these future technologies. But the court did not say how to handle cases where the accused voluntarily agrees but the science is still evolving. That question remains open.

What this means for lawyers and investigators

The practical result is clear. An investigator can ask an accused person to undergo a narco-analysis or brain fingerprinting test. The accused can say no. If the accused says yes, the test can be conducted, and the results can be used in court—but the court will weigh their reliability carefully. If the accused says no, the investigator cannot force it, and the court cannot draw an adverse inference (a negative conclusion) from the refusal.

THE PLAY: Never seek a court order to force narco-analysis or brain fingerprinting on an unwilling accused—the Supreme Court has closed that door permanently.

THE TEST: Before offering any neuroscientific test to an accused, ensure the consent is recorded on video, the accused has independent legal advice, and the procedure is conducted by qualified medical professionals in a proper facility. Any deviation risks the evidence being thrown out.

WHAT THIS MEANS: The Indian legal system has drawn a bright line: your mind is your own. The state cannot force its way in, no matter how sophisticated the key. But if you open the door yourself, what the state finds may be used against you—though the court will examine it with deep suspicion.

The machine cannot read your mind. But it can read your silence.