Helmets are not designed to protect against rotational injury, when the blood vessels and nerves attached to the brain are stretched or ruptured by the brain's inertia during sudden jerks of the skull.
Response: There is no consensus among the medical community on the threshold of rotational injury or how to measure it, although there is probably a rotational component in most serious head injuries. The standards-making community believes that a helmet that protects well against straight through (translational) impacts also reduces the effects of rotational injury, since rotational motion comes from off-center translational impacts. But the damage to the interior of the brain is not simply a function of a turning motion. It results from different parts of the brain moving in different directions at different speeds after even a translational impact. Reducing the severity of the impact reduces that type of damage.
There is potential for improvement of current helmets eventually when rotational injury is better understood and means of predicting it in a crash are developed and accepted. Because this is an area requiring further research it has always been an easy target for raising doubt about the performance of current helmets. But in fact current helmets are a long way from perfect in almost any respect, and this is just one element that can be improved. That does not negate the benefits of wearing today's helmets. They work very well despite their imperfections!