The Law on Brain Technologies in Warfare Series – The Brain Firing: Lawfulness and Liability of Weapon Control through BCI
Editors’ note: This post is an entry in our series on Air Commodore (ret.) William H. Boothby’s book, The Law on Brain Technologies in Warfare published by Elgar International.
Soldiers may soon direct weapon systems using only their thoughts. Brain-computer interfaces (BCIs) can connect a person’s brain to a computing device. The computer picks up neural signals generated by the brain and translates them into a command for a physical gadget, allowing the user to mentally control that gadget. Employing BCIs to direct weapons has already been researched and has even earned a patent.
After sketching possible BCI-enabled set-ups for weapon control, this post highlights legal challenges the technology poses to the law of armed conflict and to international criminal law.
Weapon Control through BCIs
Various scenarios for weapon control are conceivable: a commander may remotely direct a ground vehicle equipped with a gun, a weaponized drone, or a missile firing mechanism aboard a vessel, to name just a few. In principle, there are two types of BCIs: active and reactive.
Active BCIs require the soldier to focus on giving a certain command. For example, the computer may recognize the neural signals evoked when the soldier imagines pressing a button with his hand. Thereupon, the computer transmits the instruction to fire to the connected weapon. This manner of control demands an intense, conscious effort.
Reactive BCIs decipher neural signals generated subconsciously, before the operator becomes aware of them. Imagine that you accidentally place your hand on a hot stove top. Your brain will elicit signals to immediately retract it. These signals are an instantaneous, subconscious reaction. For military BCIs, such signals can be provoked by rapid serial visual presentation (RSVP). The soldier, tasked with spotting a target, is shown images of the battlefield at very short intervals. Upon recognizing the target, the soldier’s brain emits a subconscious neural response, which the BCI then decodes and transmits to the weapon.
Weapon control is arguably one of BCI’s most promising applications for militaries. It permits remote command without depending on a physical control station. Reactive BCIs, on top, enable ultrarapid decisions. Target engagement based on subconscious neural signals is possible in 250 to 500 milliseconds after visual perception, which is the fastest method currently imaginable to operate any device.
International Humanitarian Law
Because thought-controlled weapons are unprecedented, I first address their general legality under weapons law and then turn to relevant aspects of targeting law.
Weapons Law
Initially, the question arises whether such a system can be legally used in armed conflict under any circumstance. Considering the man-machine collaboration constitutes a weapon system, i.e. a means of warfare, a review under Article 36 of Additional Protocol I (AP I) (considered customary international law) is in order. Accordingly, the BCI must be capable of complying with international humanitarian law a priori, independent of a specific operational context. Closer scrutiny is demanded to determine whether the technology is inherently indiscriminate or incapable of ensuring proportionate targeting decisions.
The active BCI lets the operator consciously select targets much like a traditional remote-controlled system. To effectively discriminate between targets, the system must reliably respond to the user’s commands. Particularly, the computer must differentiate between mere ideas (“hazy intentions”) and the final resolution to engage. Even when the soldier is under immense stress, anxiety, or in pain, the correct intention ought to be identified. Provided comprehensive testing can ensure reliable functioning, the active BCI does not pose issues of legality.
The reactive BCI, in contrast, does not allow the soldier to consciously reflect or even to veto a targeting decision. Before the operator is aware of subconsciously identifying a target, the weapon will have struck it. Thus, the soldier lacks conscious control over the “moment of command.” This makes it hard to predict whether the user will distinguish lawful from unlawful targets as well as conduct a proper—subconscious—proportionality assessment, where necessary.
Having a human operator at the core of the targeting process, while simultaneously lacking control over decisions made, constitutes an unusually challenging set-up. It challenges predictability. With conventional weapons, the predictability of a targeting decision is ordinarily uncontroversial. As long as a manually operated weapon like a hand-held gun can be directed against a specific target, it is assumed the target selection is discriminate. Once certified that the gun strikes the envisaged target, the outcome of its use is sufficiently predictable. This is because the shooter controls the target selection process by aiming the gun. Differently put, a gun per se respects the principle of distinction because its targeting process is fully controlled by the soldier using it.
The reactive BCI cannot grant this amount of control. The underlying assumption of generating a predictable outcome fails. This compels a positive test for the system’s overall adherence to distinction and proportionality. Generally, where human supervision over the targeting process is absent, it is necessary to assess under Article 36 of AP I whether the weapon system can comply with the targeting principles.
One such test may be to ascertain how much conscious human control is yet present in the targeting process. This notion ensures that the soldier’s intention to strike lawful targets actually manifests in predictably striking lawful targets. There appears to be no definitive threshold of how much control, and thus how much predictability, is prescribed.
Similar to the discussion on lethal autonomous weapon systems, one could argue that human oversight at the stage of engineering and testing suffices to engender predictable results. Still, even with extensive testing in a simulation, a subconscious decision may be premature. Possibly, it would have been corrected had there been just one more conscious moment of deliberation.
I have discussed this issue in more depth elsewhere. As a final note for this post, one must keep in mind that the legality of BCI systems will highly depend on the employment scenario(s) for which they are built. The review is substantially altered, for example, in case the reactive BCI’s “normal or expected circumstances” of use are operations not endangering civilian targets whatsoever.
Targeting Law
Assuming one considers both BCI-controlled weapon systems permissible at all, they ought to obey the targeting principles, which are also customary in nature. Particularly relevant are distinction and proportionality (in the context of the specific attack) as well as the duty to take precautions.
Regarding the active BCI, its specific technicalities must be taken into account. BCIs may not only transmit the soldier’s thoughts but also supply input from the controlled device (so-called bi-directional BCIs). Detailed information from the battlefield could be transmitted directly to the user’s brain. This arguably influences a given target assessment. Alternatively, bi-directional BCIs permit altering the soldier’s mood. Such a modified state of mind could distort the operator’s judgment, leading to less precise distinction between, say, fighters of armed groups from civilians. Further, the soldier might adversely weigh the military necessity against incidental loss of civilian life.
The reactive BCI additionally raises concerns regarding the duty to take all feasible precautions in the choice of means and methods of attack with a view to avoiding, and in any event to minimizing, incidental loss of civilian life. Given the subconscious nature of targeting, the reactive BCI cannot, by design, constitute a choice of means that minimizes collateral harm compared to conscious decision-making. Conscious decision-making (as in active BCIs) will inherently produce more, or at least equally, discriminate and proportionate targeting decisions.
Hence, a feasible precaution is to ensure the soldier makes a conscious targeting decision. This naturally limits the employment of reactive BCIs to scenarios in which either striking civilian targets is precluded per se, or the military advantage of increased speed is sweeping once the attack has been decided upon.
International Criminal Law
BCIs may be used to commit international core crimes. To hold the operator accountable under international criminal law, the mentally-initiated strike of a weapon must constitute an actus reus, and the soldier must satisfy the mens rea. Both seem problematic. Subsequently, I map out the most elementary issues based on the Rome Statute.
Actus Reus
The actus reus (material element of the crime) refers to the action the perpetrator must exert to carry out a crime under the Rome Statute. At its core, the soldier must engage in specified conduct, or perform an act.
What such an act denotes is not defined in the Rome Statute. Frequently referred to in this context is the U.S. Model Penal Code describing an act or action as “a bodily movement.” Because the mental command of a BCI is no observable movement of muscles, the definition is not met.
This mismatch showcases the paradigm shift of BCI command: for as long as criminal law has existed, any criminal conduct has always entailed an externally visible bodily act. Where previously, a soldier needed to pull a trigger to shoot a person, BCIs would allow merely thinking about pulling a trigger, and yet shoot a person. As he induces the same effects in the physical world, it would be seriously perplexing if solely by switching to an active BCI the soldier was exonerated from criminal liability. The decisive factor cannot be whether the operator moves his body, but whether he creates a physical impact on the outside world. A consciously produced mental command thus constitutes an act.
But what precisely is the actus reus? Let us assume the BCI user has launched the connected weapon and thereby engaged in the war crime of directing an attack against the civilian population (Rome Statute, art. 8(2)(b)(i)). Has he directed an attack by activating the BCI? By contemplating the attack? By generating the neural signals the computer recognizes? By evoking the electrical activity within the computer to launch the weapon?
Ultimately, this question revolves around the necessary causal nexus between conduct and effect. The doctrinally sound answer points to the act of eliciting the neural signals the computer transmits. In the context of the active BCI, this is the temporally closest, voluntary action in the causal chain, which is necessary to incite the attack.
It remains to be evaluated if the same considerations apply to the reactive BCI. The soldier equally generates neural signals, albeit not consciously. Determinative, therefore, is whether the actus reus must be a voluntary act (based on conscious volition). So far, this question has been mostly academic and will need to be addressed once reactive BCIs for weapon control manifest. If one were to request a willed action, the last voluntary act with a causal connection to the attack would be the operator activating the BCI in the first place.
This controversy is highly relevant as the perpetrator must entertain the mental element at the time of the actus reus.
Mens Rea
The mens rea requires committing the actus reus with intent, both regarding the conduct and, where applicable, regarding a consequence (see Rome Statute, art. 30(1)-(2)).
Sticking to the war crime of directing attacks against the civilian population, the elements of crime include the consequence that “[t]he perpetrator intended the civilian population … to be the object of the attack.” Ergo, the mens rea exists when the operator launches the attack, while meaning to attack civilians or while being aware that civilians would be struck in the ordinary course of events. Presuming that no bi-directional information inflow or interference has affected the soldier’s volition, the active BCI does not preclude the mens rea.
If one, for the reactive BCI, identifies the subconsciously evoked neural signal as the actus reus, intent cannot be established in the same manner. There can be no intent without conscious thought. Conversely, if the actus reus was the activation of the BCI, the soldier may in theory—but likely not in practice—satisfy the mens rea. The threshold for oblique intent (the operator being aware that civilians would be attacked) is “virtual certainty.” Hence, the soldier would need to be aware, with virtual certainty, of subconsciously selecting civilian targets. Possibly, a superior commander having failed to intervene when the operator was expected to attack civilians could instead be held liable in this instance (Rome Statute, art. 28).
Nonetheless, there might be an accountability gap through the defense of “my brain did it, not me.” One may or may not consider this an acceptable outcome.
Final Thoughts
In the end, said determinations are not as much a question of lex lata as of lex ferenda. This post attempts no more than to reveal the primary shifts and challenges once BCI-controlled weapon systems emerge. The remedies for such issues might differ substantially based on the ensuing technological specificities. In the concluding observations of his book, Bill Boothby contemplated “whether future law will need to distinguish between thoughts.”
I suggest that future law will need to distinguish between brain signals. Without question, tackling such issues of neurowarfare compels a serious interdisciplinary effort in understanding how these technologies work and addressing the complications they bear.
***
Denise Koecke holds a degree in International Relations from TU Dresden and is currently a law student at European University Viadrina Frankfurt (Oder).
The views expressed are those of the author, and do not necessarily reflect the official position of the United States Military Academy, Department of the Army, or Department of Defense.
Articles of War is a forum for professionals to share opinions and cultivate ideas. Articles of War does not screen articles to fit a particular editorial agenda, nor endorse or advocate material that is published. Authorship does not indicate affiliation with Articles of War, the Lieber Institute, or the United States Military Academy West Point.
Photo credit: Getty Images via Unsplash+
