Warning! Hazard perception can seriously damage your driving.
I failed! I felt mortified. Needing a pass mark of 57 out of 75, I scored a miserable 45. I had prepared well for the test, I thought, and I had followed all the official advice on how to approach it. How could I, a man with 39 years’ accident‐free driving, a man regarded by BSM Driving School in Edinburgh as one of their best, one of their most experienced, most reliable and most successful driving instructors – how could I fail my Hazard Perception Test? And by such an appalling margin?
The Hazard Perception Test was introduced by the Driving Standards Agency (DSA) in November 2002 as part of the Theory Test that every learner driver has to pass. You sit in front of a computer screen and you see a video clip featuring an everyday road scene, videoed from the front of a moving car. Whenever you see a hazard, you click the mouse. You don’t need to click the mouse on anything: you just click it. A hazard means anything that may require you to change speed or direction. It may be for example a car parked on the left, or a pedestrian on the pavement, or a traffic light at green. Many hazards will appear on the clip, but from time to time one of them will develop: the parked car will give a right signal, or the pedestrian will approach the kerb, or the traffic light will change from green to amber. Whenever you see a hazard developing, you click again. Each hazard that develops is marked out of 5, and to score the maximum 5 you must click as early as possible. Extra, unnecessary clicks are not penalised unless you really go berserk with them. The test as a whole consists of several such clips, including a total of 15 developing hazards, and hence a maximum score of 75. The pass mark for a learner driver is 44 out of 75.
In 2004 the DSA decided that all existing instructors must take the Hazard Perception Test, with a higher pass mark of 57 out of 75. That is why I was taking it.
I was really shocked to get as low a score as 45, and I could not understand it. Many of my pupils, coached by myself in their preparation for the test, were scoring 60 or more. It was obvious that I had a big problem, and I determined to research it from all angles.
As I researched it, a picture began to emerge. I found that most young instructors have little difficulty with the test, but that older instructors like myself often have difficulty. I learned that some candidates find they have to reject the DSA’s advice to imagine that they are in a real driving situation, and they have to treat the test instead as a video game. But the biggest part of the picture emerged when I began to read up on the psychology of visual perception.
Over the last 25 years, it has come to be recognised that visual processing in the brain proceeds in parallel down many separate pathways, including two major ones. One of these two major pathways, the dorsal pathway, is very fast and deals with what is often called “vision for action”. The other, the ventral pathway, is much slower. It is characterised in the literature by many names: “vision for perception”, “vision for cognition”, “vision for recognition”, “vision for identification”, “vision for consciousness”, “vision for conscious identification”, or “phenomenal vision”. As there is a recent trend towards using the terms “perception” and “cognition” with a wide scope that would include vision for action, here I will use the phrase “vision for conscious identification”, which seems to me to be its best characterisation in modern terms.
Now, I have Asperger Syndrome (AS). AS is an autism spectrum disorder. It is currently diagnosed on the basis of impairments in social interaction and non‐verbal communication, and restricted patterns of behaviour. It is recognised however that this diagnostic convention may not reflect the true nature of the disorder. It is known to be of neural origin, and abnormalities of sensory perception are increasingly being found within it. In particular, when I began to look at the psychology of visual perception, I was intrigued to read that individuals with autism spectrum disorders seem to have a specific deficit in the dorsal, “vision for action”, pathway.
Certainly I have known since childhood that my vision for action is weak. As a child I found it hard to catch a ball. To this day I cannot ride a bicycle. Any two‐year‐old can beat me at “Snap”: the child knows at a glance whether the two pictures match, whereas I have to look at them one by one and compare them consciously. And yet, unlike many individuals with autism spectrum disorders, I had no difficulty learning to drive a car at age 17. What I believe happened is that I learned to drive, atypically, by using my slow but intact “vision for conscious identification” pathway, almost to the exclusion of my potentially fast but deficient “vision for action” pathway. Paradoxically, I think that that helped me to become a good driving instructor. It means I have always thought about what I do when I drive, and I store my knowledge in representational, declarative terms. Consequently it has always been instinctive for me to be able to instruct a pupil as to what to do.
After further reading, experiment and introspection, I concluded that my preferred slow ventral pathway, while adequate for the real world of driving, is on its own too slow for the specialised and artificial setting of the Hazard Perception Test.
Armed with that knowledge, I set out to find a strategy for my second attempt at the test. I was allowed an unlimited number of attempts, but time was short as I had left it late, and now I had to pass the test within the next 3 weeks. I used the same practice DVD as I had used before my first attempt, but now I used it very differently.
First of all, I experimented on each of the practice clips to discover the exact scoring system. Published information told me that the scoring window runs from just before the hazard starts to develop, and is divided into 5 equal parts during which the available score is 5, 4, 3, 2 and then 1. I discovered that the length of the window varies a lot from clip to clip, and on average it is about 4 seconds. I also discovered that in at least one clip the scoring window runs from when the hazard first appears, contrary to the published advice that it always runs from just before the hazard starts to develop. That is why it is essential to click when you first see a hazard, and to click again if it develops.
Since my ventral pathway – since anybody’s ventral pathway – is too slow to give the early warning of a hazard that I needed, I had to engage my potentially much faster but deficient dorsal, “vision for action”, pathway to give that early warning, unaccustomed though it was. I reasoned that to engage it I had to give it a simple job, and to ensure that I was free of any other cognitive tasks at the same time. And the simplest possible job for the dorsal stream is to effect a simple motor response, such as clicking a mouse, whenever anything moves.
Therefore, I rejected all the official advice on how to approach the test. I rejected the advice to imagine that I am driving the vehicle: autism spectrum individuals are deficient on imagination and, for such individuals, using imagination on a cognitive task interferes with the task. I rejected the advice to scan all the time to look for and identify clues, even though scanning is what I teach my pupils to do and is what I myself do in the real driving situation with unimpeachable results. I rejected the advice to plan: to think about what might happen and to consider the options I might need to take to reduce risk. I rejected the advice to look for static clues such as road markings, traffic signs, road works, parked cars, and wet or muddy road surfaces. I rejected all these things, which in the real driving situation are second nature to me. Instead I learned to look fixedly at the centre of the screen and concentrate on picking up movement with my peripheral vision. I resisted the urge to look upwards and leftwards towards a non‐existent mirror. My strategy was: if anything moves, click. Afterwards look at it, identify it, decide if it is a hazard, and if it develops then click again. In effect I demarcated my two visual streams: I used my damaged “vision for action” merely for the initial click, ensuring that nothing interfered with it, and I did not use my reliable but slower “vision for conscious identification” until afterwards. I practised the same clips over and over again, rejecting the advice that doing so is counterproductive. In training, this strategy worked well. But would it work in the test, on clips that I had not seen before?
All too quickly, it seemed, the day arrived for my second attempt at the test.
On this my second attempt, I scored 58 out of 75, an improvement of 13 points over my first attempt. This works out at an average improvement of about 0.70 seconds in the time taken to respond to each one of the 15 developing hazards. Given that my simple reaction time – click when the light flashes – is unimpaired at around 0.16 seconds, that is a very great improvement indeed.
I was cock‑a‑hoop. I was also mightily relieved: I passed with one day to spare.
But the scariest bit was yet to come. I got into my car to drive home and I quickly found I was driving like a zombie. Was I checking my mirrors? – no. Was I planning? – no. Was I moving my eyes? – no. Was I signalling? – no. The overall feeling was one of unreality, like a video game. It was the same feeling that I remember having on the one and only occasion when I tried to drive with sunglasses on. I approached a roundabout, I looked right, and I saw a car coming round the roundabout; but my feet did not brake for me, at least not until I made a conscious effort of will as if I were a new learner. It isn’t that I was tired. Rather, it seems that in the artificial world of the Hazard Perception Test I had redeployed my two visual streams so radically that they would no longer work together harmoniously in the real world. There had not been the slightest hint of such a difficulty after my first attempt at the test.
Quite fazed, I pulled in at the side of the road, secured the car and turned off the ignition. I closed my eyes and rehearsed the proper way to drive. I imagined a traffic light ahead of me turning to amber, I moved my eyes to where the mirror would be, and I braked physically. With eyelids closed, I practised moving my eyes towards the mirrors and back, and physically operating the indicator stalk. I imagined a child in the road in front of me and I braked hard. With the engine restarted, I imagined reaching a bend to the left and I turned the wheel. I repeated this to the right. I recited aloud the “five habits” – look well ahead, move your eyes, spot the problems, keep space, be seen. I drove off again carefully and talked myself through everything: I was instructor and pupil at the same time. I got home without feeling too unsafe, and without making any serious driving errors that I am aware of. At least, nobody honked.
The next day, driving still felt a little odd. By the day after, everything had returned to normal and I felt quite safe again.
What are the lessons? First, I think, that one size does not fit all. I have special strengths and weaknesses, and I had to take account of these in planning for the test. Secondly, that it is not always best to practise what you preach. Thirdly, never to become blasé when driving, no matter how many years’ accident‐free experience you have behind the wheel. Fourthly, for a driving instructor, to remember that a pupil, too, may have unforeseeable specific difficulties that can surface at any time. But most important of all, to avoid driving when your mind is disturbed. Just as it is irresponsible to drive while unfit through the influence of drink or drugs, so too it is irresponsible to drive while unfit through the influence of the Hazard Perception Test, or any other video game. Hazard Perception can seriously damage your driving.
Eric P Smith
9 February 2008
Spencer J, O’Brien J, Riggs K, Braddick O, Atkinson J, & Wattam‑Bell J (2000). Motion processing in autism: Evidence for a dorsal stream deficiency. NeuroReport, Vol. 11, No. 12 (September 2000), pp2765–2767.
Full text can be downloaded from:
Scott F J, Baron‑Cohen S, & Leslie A (1999). ‘If pigs could fly’: A test of counterfactual reasoning and pretence in children with autism. British Journal of Developmental Psychology, Vol. 17 No. 3 (September 1999), pp349–362.
Full text can be downloaded from: