SMIDSY: why drivers don't see motorcycles — and what actually fixes it

SMIDSY is motorcyclists' shorthand for the apology they hear after being knocked off: "Sorry mate, I didn't see you." It describes one crash above all others — a car pulling out of a side road, or turning right across oncoming traffic, into the path of a motorcycle that had right of way. Crash researchers call the same thing a looked-but-failed-to-see (LBFTS) error, and it has been the leading cause of car-versus-motorcycle collisions in every major in-depth study from the Hurt Report (fieldwork 1976–77) to MAIDS.

The Hurt Report in California (1981) found that roughly three-quarters of motorcycle crashes involved a collision with another vehicle, that a junction was the most likely place for it, with the other vehicle violating the motorcycle's right of way, and that "the failure of motorists to detect and recognize motorcycles in traffic is the predominating cause of motorcycle accidents". Europe's most comprehensive in-depth study, MAIDS (921 motorcycle and moped crashes investigated on-scene in France, Germany, the Netherlands, Spain and Italy), found that in 50% of cases the primary contributing factor was a human error by the other vehicle's driver, and that the most frequent error by far was a failure to perceive the motorcycle: in 36.6% of all the crashes investigated, that single failure was the primary factor. It also found something that matters for everything below: car drivers who also held a motorcycle licence were roughly half as likely to be the one who failed to see the bike (26% against 51%).

The RSA's Motorcyclist Spotlight Report for 2021 to 2025, published on 26 June 2026, found that 64% of motorcyclist deaths involved a collision with another vehicle, and that one in four riders was killed at a junction. Two-thirds of serious injuries happened in multi-vehicle collisions, and in those the action most often recorded — for the other driver as well as for the rider — was "failure to observe". Of the 117 riders killed, 116 were men. 2025 was the worst year of the five: 30 motorcyclists died, close to one in every six people killed on Irish roads that year, and the highest motorcyclist toll since 2007.

Kevin Williams' Science of Being Seen project sorted the research into what actually goes wrong, and the useful conclusion is that SMIDSY is not one error. The driver can fail at any link in the chain we described in the gorilla article — look, see, recognise, respond — and a motorcycle makes every link weaker than a car does.

The rest of this article takes failures 2 to 5 in turn, because those are the ones that happen to careful drivers — including the careful driver you think you are.

The A-pillar. The pillars either side of your windscreen have grown thick for crash protection — they house airbags and hold the roof up in a rollover — and the type-approval rules allow each one to block up to 6 degrees of the driver's binocular field of view (Directive 77/649/EEC). Six degrees sounds like nothing. At 50 metres it is a wedge more than 5 metres wide, room to hide a motorcycle several times over, and a motorcycle approaching from a distance is narrow enough to sit completely inside it. This is not a new discovery: Olson (1989), re-reading the Hurt Report's Los Angeles crash data, pointed out that in at least 48% of the 457 cases where a driver violated a motorcycle's right of way, the driver's view of the motorcycle had been blocked to some degree. Measured in time: when the UK's Transport Research Laboratory reconstructed real crashes in which the pillar was suspected, the motorcycles had been hidden for two to four seconds — longer than a typical "look right, look left, look right again".

Constant bearing. This is the part few drivers know. If you are creeping towards a junction and a motorcycle is approaching on a genuine collision course, its position relative to you does not change — it stays at the same angle, in the same spot on your windscreen, getting bigger but not moving sideways. That is exactly the object that stays behind the pillar, because it never drifts out from behind it until it is close enough to "grow" around the edges. Sailors and pilots are trained to fear a constant bearing. Almost no driver is ever told about it.

The blind sweep. The brain suppresses much of what the eyes take in while they are jumping quickly between two points — a phenomenon called saccadic suppression — and what it suppresses first is exactly the coarse motion signal that flags an approaching vehicle. Vision does its real work in the pauses. A fast flick of the head to "check the junction" is mostly jump and hardly any pause: between the fixation at each end you are, for practical purposes, blind. Anything sitting where the eyes stop gets seen, and a car is big enough to be caught by peripheral vision even when it is not. A motorcycle in the middle of the sweep is not.

This is the failure the gorilla experiment made famous, and a study by Australian researchers published in 2018 ran a photograph-based driving version of it. Pammer, Sabadas and Lentern showed drivers a series of photographs taken from the driver's seat and asked them to judge whether each scene was safe or unsafe. In the ninth photograph an extra vehicle had been added on the right of the scene, near a junction: a taxi for one group of participants, a motorcycle for the other.

A second experiment made the point sharper: when the motorcycle was turned to face the junction — heading somewhere the driver would have to reckon with — the miss rate fell to 33%. The brain sees what it judges relevant to its own path. The mechanism is expectation. Attention finds what it has been told to find, and at a junction a driver's search template is, in effect, car-shaped. Beanland, Lenné and Underwood (2014) showed why that template forms: in a driving simulator, drivers detected whichever vehicle type was common faster than the one that was rare — and on Irish roads motorcycles are rare. Your brain has learned, from thousands of junctions, that the thing coming will be a car. It is a good bet, and it is the bet behind these crashes.

Two more findings sharpen the picture. Crundall and colleagues (2012) tracked the eyes of 74 drivers on video approaches to T-junctions and found that novices, experienced drivers and "dual drivers" (who both drive and ride) all found the approaching motorcycle equally quickly — the problem was not the looking. Strikingly, it was the experienced car drivers, not the novices, who showed the eye-movement signature of a look-but-fail-to-see error: their gaze on an approaching motorcycle was shorter than their gaze on a car, as though the look had ended before the brain had registered what it was looking at. The authors put this down to over-learned search habits and years of expecting cars. Novices were not safer overall — they pulled out sooner — but their problem was a different one. Dual drivers had the safest responses of the three groups, which echoes what MAIDS saw in real crashes. Clarke and colleagues' study of 1,790 UK police crash files, 1,003 of them examined in detail (2007), adds an uncomfortable detail: these junction crashes "often seem to involve older drivers with relatively high levels of driving experience". Experience is not a defence.

In 2019 a University of Nottingham team put drivers through junctions in a high-fidelity simulator and, at the moment they began to pull out, stopped the simulation, blanked the screen and asked them what had been approaching. Drivers failed to report an approaching motorcycle on 13% to 18% of occasions. In the third study they failed to report a motorcycle 16 times, and a car 3 times — and on 11 of those 16 occasions the eye-tracker showed the driver had already looked straight at the bike.

The explanation the researchers propose is not attention but short-term memory. Pulling out of a junction is a busy manoeuvre: you look right, then left, then check the road ahead, then the mirror, and by the time you move, the small item you registered a few seconds ago has been overwritten by everything you looked at since. Crucially, forgetting was not linked to how long the driver had looked at the bike — it was linked to how much looking came after, and to how long passed between seeing the bike and moving. The more head movements after the glance at the bike, the more likely it was to go unreported.

The last failure happens after everything else has gone right. The driver sees the motorcycle, knows it is a motorcycle, judges there is time to go — and there isn't. Horswill, Helman, Ardiles and Wann (2005) showed drivers video of vehicles approaching a junction and asked them to press a button at the moment each would have reached them. Across speeds and viewing times, the motorcycle was judged to arrive later than a car or van at the same speed. The brain uses an object's size as a shortcut for distance and arrival time: a smaller object is judged to be further away and to arrive later. Horswill's second experiment showed the bias survives even when the bike's rate of growth on the retina is plainly visible — it is not that a narrow object fails to "loom", it is that the brain discounts it because it is small.

The consequence shows up in the gaps drivers accept. The observation Horswill's team set out to explain was that "drivers adopt smaller safety margins when pulling out in front of motorcycles compared with cars". A rider travelling faster than the surrounding traffic makes the consequence of the illusion worse — the driver's late estimate is wrong by an even bigger margin — which is one reason motorcyclists carry responsibility here too, and why the rider section below is not an afterthought.

The honest summary of the conspicuity research is that being bright helps — meaningfully, on the best evidence — but does not solve the problem, and what counts is standing out from the background rather than the colour itself. The best evidence, a New Zealand case-control study (Wells et al., BMJ 2004), found riders wearing reflective or fluorescent clothing had a 37% lower crash-injury risk, a white helmet 24% lower than black, and a daytime headlight about 27% lower (a borderline result). Worth having — but every one of the five failures above happens to riders in full hi-vis, because a pillar hides yellow as well as black, and inattentional blindness is about what the brain is looking for, not what colour the thing it misses happens to be.

• Assume you have not been seen until the driver's behaviour proves otherwise — wheels stopped, eyes on you. A driver looking in your direction is worth nothing; that is what "looked but failed to see" means.
• Break the constant bearing. A small, deliberate change of line within your lane on the approach to a junction — out and back — breaks the constant bearing and gives the driver's peripheral vision the sideways motion it responds to; Roadcraft notes that the eye's peripheral receptors are "particularly good at sensing movement". Combined with a change of speed it can bring you out from behind the driver's pillar, but do not rely on a metre of movement alone: at 50 metres a pillar can hide more than five metres of road. A rider on a fixed line at a fixed speed is about as hard an object to detect as the road produces.
• Close the speed gap. The size-arrival illusion already makes drivers think you are slower than you are; riding faster than the traffic around you makes their estimate worse, not better. Near junctions, ride at the speed drivers expect.
• Cover the brakes at every side road and every right-turner, and plan the escape before you need it. Reaction time you have already spent is the only kind that counts.
• Get out of the driver's blind arc. Behind a car that could turn right, sit where the driver's mirror shows you, not in the space their pillar covers — and never alongside a car that is signalling.

Pulling out in front of a motorcycle on test — anything that makes the rider brake or swerve — is a Grade 3, the dangerous-or-potentially-dangerous fault that is an immediate fail, because the tester marks observation on whether you took effective observation and reacted to what was there, not on whether your head turned. The head-flick that catches cars and misses bikes is exactly the "inadequate observation" at junctions that the RSA lists first among the most common reasons for failing the test. Everything in the driver section above is also test technique: the paused look, the second late look, the car-sized gap.

It cuts the other way too. A learner who hesitates because a motorcycle is somewhere in the road gets marked for progress. The skill the test is asking for is the one this whole article is about — see the bike, judge it correctly, and then either go or wait with confidence. That is why we train junctions with commentary: you cannot say "bike, right, 60 metres, wait" about a motorcycle your brain skipped.