Thursday, 15 August 2013

Attention: Humans driving


Reaction times correlate with lifespan and with intelligence. In evolutionary terms this makes sense: quick reactions allow you to spot dangers, avoid predators and live another day.

In contemporary life, which is now very safe, one of the few causes of premature death are traffic accidents. Motorbikes are the worst. Without a protective cage to absorb impact, riders cannot survive. Helmets don’t help much because necks break as easily as skulls. Pushbikes don’t mix well with motor traffic. All cyclist deaths are premature and preventable, if only by taking public transport, which is safe. Cars are now safe for their occupants. Car braking systems very rarely fail. Human drivers fail every so often. Young over-confident incompetents are the worst, as are drunks of all ages, but careless idiots still cause injury, and each of us will be a careless idiot at some time in our driving lives.

Here is a reaction time test which is a better measure of attention while driving. Unlike the “BBC poisoned arrow sheep” and the various traffic light tests it has face validity, because it lasts for 5 minutes, which simulates the real life requirement of sustained attention. Of course, 5 minutes is not very long, and half an hour would be more realistic. The simulation is also unrealistic in that the hazards happen frequently, whereas in real life they happen rarely, with long boring bits in between. Furthermore, in this test one can begin to anticipate when it is time for the next hazard to show up. The results give the average time of response, but merely record the number of crashes and “false starts” without any further comment. The site constructors have a particular emphasis on fatigue caused by lack of sleep, which is certainly a factor when lorry drivers are sleep deprived, but probably not such a big issue with ordinary car drivers.

My own results “need to be seen in context” (were bad). On the first trial I hadn’t properly understood the instructions. After each hazard the reaction time comes up, which is distracting, but is probably a good example of the many distractions while driving. My average reaction time was 0.31 with one accident and 4 false starts. In other words, death. After a brief pause I tried to redeem myself. Second trial: average reaction time 0.32 with o accidents, 0 false starts. Clear evidence of learning in a dead person.

Let me know how you get on.


  1. "whereas in real life they happen rarely, with long boring bits in between": this is a key problem in industrial safety systems - most humans don't cope well in those circumstances. It's an attraction of automation.

    You could use a variant of the technique used for steam engines before the introduction of Watt's governor: small boys were meant to regulate steam flow by pulling on a string. To ensure that they didn't fall asleep they were sat on one-legged stools. The equivalent for a car driver would be .....

  2. Reaction times may correlate with mortality mainly because they (RTs) are a better index of mutation load than IQs.

  3. Mutation load is certainly a possibility. Interestingly, virtually all of the reaction time studies have little psychological data on their subjects. Once we get full genomic data some things will become clearer.

  4. My first attempt was appalling. Reaction time of .37 s, with 2 crashes and 2 false starts.

    WTF is a false start? Changing lane when the obstacle wasn't actually in my own lane? I wasn't aware of doing that, but I can well believe I misjudged the location of the obstacle. But then wouldn't that have resulted in a crash?

    The other possible meaning is swerving when there was no obstacle at all - but I'm quite sure I never did that.


  5. Second attempt, shortly after the first: 0.34s, 1 crash, 1 false start.

    Evidently I'm still a danger on the roads.

    Though in real life, in any 5-minute period taken at random, I usually don't crash. In fact, I occasionally complete quite long journeys without crashing.

    But what the hell is a "false start"?


  6. "False start" is not a particularly well chosen phrase. As you surmise, it means a mistaken application of the brakes when there is no real hazard. This changes you to the other lane. Your real life 5 minutes comparison is not representative. This test is trying to cram many hazard events into a short period. It probably represents the number of full scale emergency stops you would make in a year or more.