You’re driving down the road when the car in front of you suddenly slams on the breaks. You see their break lights go on followed by the car rapidly getting closer to you and filling your visual field. Your brain registers what is happening and immediately plans a motor response in reaction – take your foot off the gas and apply pressure to the brake. Your car then screeches to a halt – but will it stop quickly enough to avoid colliding with the car in front of you?

A recent study looks at the ability to predict when a driver is about to brake in an emergency breaking situation. They used a driving simulator and monitored the gas pedal, the brake, the muscle activity in the subjects’ legs (EMG), the speed of the car, the distance to the car in front, and the EEG activity (electrical brain activity) of the driver. What they found is not surprising, but quantifies the time intervals involved.

The EEG reveals that the visual stimulus is recognized first, followed by the intention to perform a motor action (240 ms – miliseconds). Then the motor action begins (335 ms), resulting in stopping the gas (430 ms) and then applying pressure to the brake (595 ms). There is actually a range of figures for each event, based upon the accuracy of prediction. The longer you wait, and the more data you gather, the higher the accuracy. Predictive accuracy of >95% was generally achievable using information from EMG and EEG. To boil it all down, the authors estimate that a system using EMG and EEG can predict a driver’s intention to brake in an emergency setting 130 ms prior to when they would actually brake. This potentially could result in a 3.66 meter shorter breaking distance, which can mean the difference between a crash and near miss in many cases.

The researchers demonstrate that there is a window for improved reaction time in emergency breaking (130 ms) and that window is large enough to make a practical difference is crash avoidance. They further speculate that a driving assistance system based upon this information could therefore reduce car crashes.

A major limitation of the system, of course, is that the driver would have to be fitted with EMG and EEG leads while driving. This is a non-trivial limitation – who wants to wear a cap full of electrodes while they’re driving? The technology to measure brain waves accurately without such leads is not available or even on the drawing board.

Such as application also assumes that we can engineer a computer system that would respond faster than the driver themselves with sufficient accuracy that it will not cause more accidents than it prevents (by, for example, registering a false positive and applying the brakes).

Another option for achieving the same goal is to take the driver out of the loop – with collision avoidance systems. These are already in use. Essentially sensors on the car detect the rapidly approaching obstacle and apply the brakes themselves.

This research, however, does highlight a couple of interesting points. The first is that, in an emergency breaking situation, miliseconds make a difference. Further, the time it takes for the driver to realize that they need to break is significant. There is a certain minimal amount of time necessary for the brain to process the sensory information and plan an action. This cannot be avoided, and is precisely why you should not tailgate. However – this assumes that the driver is paying attention and is prepared for the possibility that they will have to urgently brake. If the driver is distracted, even a little – by, say, talking on a cell phone – that will add precious miliseconds to the overall reaction time and significantly increase the risk of accident.

I don’t think we’ll be driving around wearing an EEG cap anytime soon (if ever), but there are ongoing efforts to add passive accident prevention to cars to make them safer. Human drivers are slow to react (compared to the fast speeds at which we often drive), are easily distracted, cannot multitask, and have a hard time maintaining continuous attention. Computers, however, have none of these failings. They are tireless and focused. They are not as smart as humans, but if given a very specific task we can develop the programming to assist or even take over some simple functions, like very quickly applying the brakes to avoid a head on collision.