Racing the Devil


The incident below is, on the face of it, a classic example of the perversity of human behaviour, but watch closely to the end…

So who was expecting the sudden appearance of a train coming the other way? After the event, it’s obvious there’s a finite probability of two trains meeting at the railway crossing, but of course that’s a ‘post facto’ argument (1).

While we may dismiss the drivers behavior as reprehensible and egregious this near miss actually provides a good example of the complex interrelationship between human decision making, error, uncertainty and risk (2)(3).

In this instance the driver clearly perceived the train proceeding in the same direction, but not the oncoming train. The most likely reason for this lack of perception is, understandably, a high level of task fixation coupled with the inherent difficulty in perceiving the other train given it’s angle of approach and cluttered sightline. Human beings tend to focus upon what they believe to be important and look for what they expect to see (4).

Underlying this failure to perceive is a biased assessment of the likelihood (and risk) of a second train passing through the crossing at the same time. Now in situations of uncertainty where hard data is not available human beings use rules of thumb, such as the availability heuristic, to evaluate probability. These rules of thumb can also result in biases in assessing probability (5).

The other part of this story is of course a decision to violate the normal or socially approved rules of road behavior. These rules in this instance are designed to deconflict road and rail operations.

As we transition from an approved, deterministic and safe operation to one which is based on ‘accepted’ or tolerated practice which ‘may be’ safe we start having to make judgements of probability and risk based on imperfect heuristics. In this case the driver clearly discounted the likelihood of a second train, incorrectly as it turns out.

Further the cognitive workload of assessing whether a task is ‘safe’ or ‘unsafe’ within a specific environment and context (and usually against a time constraint) is so much greater. This can in turn lead to operator ‘situation’ overload where the operator (driver) perceives but cannot process new information.

As time available for these decisions decreases or a situation becomes more uncertain greater reliance is in turn placed on such heuristics. So under time pressure and uncertainty we become most vulnerable to our cognitive biases.

And of course eventually the devil may turn up…


1. The second train accident is also a significant risk at crossings that are light protected. A Queensland study (Ferreira, Tey 2010) indicated that 9% of drivers would move off before the lights at a crossing had finished flashing.

2. Using the error taxonomy proposed by James Reason the drivers action constitutes a ‘violation’ of procedure. That is the wilful departure from some approved procedure or ‘law’ for a perceived local benefit

3. An Australian study of train drivers observations of public behaviour at rail crossings put ‘racing the train’ at the top of their list of risk taking behaviour (Ford, Matthews 2002). Note however that this is a subjective evaluation and does not necessarily reflect actual risk.

4. This failure to perceive is termed inattentional blindness. In a classic psychological study of this affect test subjects were asked to watch a basketball game and count the number of times the ball was bounced by each team. They became so focused on the game that they then completely failed to notice a man in a gorilla suit walking through the game!

5. Known as the availability heuristic. Lack of personal experience or biased reporting (in terms of which stories get more air time) can cause false high or low estimates.


Davey, Jeremy D. and Wallace, Angela M. and Stenson, Nicholas J and Freeman, James E., 2007, Train drivers’ ratings of perceived risk associated with illegal behaviours at level-crossings. Journal of Occupational Health and Safety, Australia and New Zealand 23(5): pp. 445-450.

Tey, Li-Sian and Ferreira, Luis (2010). Driver compliance at railway level crossings. In: , Australasian Transport Research Forum 2010 Proceedings. Australasian Transport Research Forum 2010, Canberra, Australia, (1-13). 29 September – 1 October 2010.