ONE BY one, the disciplines of science have lost their innocence. For chemistry, the defining moment came during World War I, when the Germans unleashed an asphyxiating wall of chlorine gas on French troops. For physics, it was the 1945 obliteration of Hiroshima in a bright flash of nuclear fission. Knowledge brings power, and power can mean new ways to kill.

Now, looking ahead at the future of warfare, it appears that neuroscience will be next. Over the last decade, there has been remarkable progress in deciphering the signals the brain uses. The work is bringing new insights into disease, but could also be used to create a frightening new family of neurochemical weapons, designed to shut down particular brain circuits – or, eventually, to create highly agile killing machines, controlled by the mind.

“We are on the cusp of a major revolution here, and we need to identify some of the issues that may arise,” says Rod Flower, chair of a British Royal Society committee that has just issued a report identifying the military applications of neuroscience.

Brain science is a celebrated science. But it is time to confront the threats it poses. University neuroscience programs should insist that their students learn about the potential social impact of their work, and the specific ways it could be used by militaries, or terrorists, to create weapons. Our international treaties, particularly the chemical and biological weapons conventions, need to be revised to reflect the new scientific realities.

The chemical weapon convention bans the use of the kinds of horrific weapons deployed in World War I and, more recently, in the Iran-Iraq war. But the convention allows the development of non-lethal “riot control agents” for domestic use. In an earlier age, this sounded like a reasonable exception, but it has become a dangerous loophole which could set off a new arms race.

The problem is that “non-lethal” gases are only non-lethal until they aren’t. This was on dramatic display in 2002, when a gang of Chechen separatists took over Moscow’s Dubrovka theater and claimed some 800 hostages. Several days into the siege, Russian Special Forces flooded the ventilation system with a gas, knocking everyone inside unconscious, then stormed in. All the terrorists were killed, but more than 100 hostages also died of overdoses, and many others suffered side effects from the gas. If the dose is large enough, a non-lethal gas will kill.

The Russian government has never said what compound it employed, but it was likely the kind of “non-lethal” gas allowed under international law. The danger is that governments will use insights from neuroscience to design new gases, allowable under international treaties, which could also be used on the battlefield to immobilize or kill.

Another dark possibility, highlighted in the Royal Society report, is a future where weapons are hooked directly into an operator’s brain, boosting performance. This admittedly veers into the realm of sci-fi, but not as far as you might think. Scientists have already created a system, called “Braingate,” which allows a person to control a cursor with their thoughts alone. DARPA, a research arm of the Pentagon, has shown it can dramatically improve the performance of analysts looking at satellite imagery by monitoring their brains for signs they’ve spotted something unconsciously. If a mind-controlled plane drops a bomb on civilians, who would be held responsible?

When assessing rapidly moving areas of science, one is trapped by what Martha Farah, director of the Center for Neuroscience and Society at the University of Pennsylvania, calls a central paradox of the endeavor: Speculation means identifying dangers that will not materialize; but the failure to speculate means bad surprises are surely in store.

Neuroscience has become such an area, and if something terrible were to happen, nobody can claim to have not been warned.

UPDATE: I recently called on Boston scientists to boycott the publisher Elsevier, which supports legislation to restrict the public’s access to biomedical research. In an encouraging development, members of Congress have reintroduced a piece of bipartisan legislation, the Federal Research Public Access Act, which would expand free access to taxpayer-funded research. More than 6,000 researchers have joined the Elsevier boycott, including 50 at MIT, 47 at Harvard, 14 at Brandeis University, 13 in the University of Massachusetts system, eight at Boston University, six at Tufts, and four at Boston College.