I have spent a good part of the past week thinking about driver perception, prompted partly by some thoughtful comments in Linkedin on my last post. One theme kept returning: people use “saw” to mean two different things, and in a collision case the difference decides who is at fault.

Did the driver see the hazard in time? That single sentence hides two separate questions, and only one of them belongs to the driver.

The first is whether the hazard was physically available to be perceived, from that seat, at that instant. This is a matter of geometry, occlusion, lighting, headlight range and the other party’s conspicuity. It is measurable. Given the sightline, the speed, the light available and the contrast of the object against its background, I can reconstruct how many metres and how many seconds the hazard sat within the driver’s field of view before impact. That figure is a fact about the road and the night. It is not a fact about the driver.

The second is whether a reasonable driver, giving the attention the law requires, would have perceived it in time. That is not a measurement. It is a standard, and it belongs to the court.

Running the two together is a common error in how these cases are argued.

The High Court kept them apart, deliberately, in Manley v Alexander [2005] HCA 79. At about 4.15am, Manley was driving a tow truck along Middleton Beach Road in Albany when he struck Alexander, who was lying on the carriageway in dark clothing after a night’s heavy drinking. Manley’s account was that he could not have seen Alexander any sooner, because his attention was on Alexander’s companion, Turner, who was at the roadside “moving around a fair bit like he had been drinking.” Manley watched Turner for two to three seconds, held his speed at about 55 to 60 km/h, and veered toward the centre of the road, presumably to widen his margin from Turner in case he stepped out.

On the first question, availability, the facts were barely in dispute. A street light illuminated the area where Alexander lay. The contour of the road gave Manley an uninterrupted view ahead for a distance considerably greater than the reach of his low-beam headlights. In the physical sense, the object was there to be seen. Availability was established by geometry and lighting, not asserted by a witness.

Availability does not necessarily decide the case. It sets the stage on which the reasonable-attention question can be asked at all. You cannot ask whether a driver should have perceived a hazard until you have shown the hazard was there to be perceived. A majority of the Court (Gummow, Kirby and Hayne JJ) then answered the question it had reserved for itself: watching Turner for two to three seconds, while holding speed and changing direction, was not reasonable care. A driver must attend to all that is happening on and near the roadway, and recognising one source of danger does not license giving it exclusive attention. (Callinan and Heydon JJ dissented, taking the view that a driver may properly prioritise an apparently live danger at the roadside over an apparently empty road.) In Manley the two questions lined up, availability established and attention found wanting. They need not. Keeping them distinct is what lets you see the harder case, where a hazard was plainly available and a reasonable driver still could not have perceived it in time.

Here is the part the comments pushed me to make explicit. Availability and perception come apart because of how human vision actually works. High-acuity sight occupies a tiny patch: only the central one to two degrees of the visual field, the fovea, resolves fine detail, and everything outside it is progressively blurrier. To take in a scene, the eyes dart across it about three times a second, the brain suppresses vision during each jump, and then assembles the still frames into what feels like a seamless, detailed picture. It is largely a construction. Away from the point of fixation, the periphery supplies coarse gist, good at motion and shape, poor at detail. When the eyes settle on one feature, everything else is held from older samples and that low-resolution gist, and the longer they dwell, the staler the model of the rest of the scene becomes. The fog creeps back into the places you have not looked at recently.

Manley’s own account fits this exactly. For two to three seconds his fovea was on Turner. The carriageway ahead, where Alexander lay, was covered only by his last samples and his periphery. The cue was available. His eyes were elsewhere.

Manley was 22. He was also, in the Court’s own description, an experienced driver of large vehicles. His failure was not a learner’s deficit. It was the ordinary limitation of a visual system that can resolve detail in only one place at a time, and it can defeat a competent, experienced driver as readily as a new one.

Where novices are genuinely worse is upstream of that limit, in deciding where to point the fovea and how fast to make sense of what lands on it. The eye goes where experience has taught it to look, and a reduced useful field (the area a driver actually extracts information from in a glance) is associated with higher at-fault crash risk. Expectancy is the other half: a body lying on a road at 4am, where the Court accepted a motorist would scarcely expect to find a pedestrian, violates the template a driver builds from thousands of hours of what roads usually contain. The unexpected object is detected later and comprehended more slowly. A learner has barely started building either the scan or the template, which is the part of driving the system leaves to chance. [VERIFY] the comparative claim that novices scan less efficiently and anticipate hazards more slowly than experienced drivers is well established in the young-driver literature (e.g. hazard-perception and eye-movement training studies), but confirm a specific source before publication.

This is where the forensic method earns its keep, and where it has to stop. I quantify availability: from the sightline, the lighting, the speed and the conspicuity of the other party, I can state how long the hazard was within the field of view and at what point detection became physically possible. Borrowing Endsley’s account of situation awareness, I can separate whether the cue could be detected at all (perception) from whether it could be recognised as a hazard (comprehension) and whether its path could be anticipated (projection). Each is a different question with a different answer, and collapsing them is how false certainty gets into a report.

What I do not do is answer whether a reasonable driver would have perceived and responded in time. That is the ultimate issue, and in Manley v Alexander the High Court reserved it to itself. An expert who answers it has stopped giving evidence and started delivering a verdict. The restraint is not modesty. It is the boundary that keeps the quantified part of the evidence both admissible and worth having.

So the question is not whether the driver saw the hazard. It is whether the road, the conditions and a reasonable standard of attention made it perceivable in time. Of those three, only the last is the driver’s. And whether the driver met it is the court’s to decide, not mine.