Traffic: Why We Drive the Way We Do (and What It Says About Us) - Tom Vanderbilt (2008)

Chapter 3. How Our Eyes and Minds Betray Us on the Road

Keep Your Mind on the Road:
Why It’s So Hard to Pay Attention in Traffic

Any man who can drive safely while kissing a pretty girl is simply not giving the kiss the attention it deserves.

—Albert Einstein

Here is a common traffic experience: You are driving, perhaps down a mostly empty highway, perhaps on the quiet streets around your house, when you suddenly find yourself “awake at the wheel.” You realize, with a mixture of wonder and horror, that you cannot remember what you have been doing for the past few moments—nor do you know how long you have been “out.” You may find yourself sitting in your driveway and asking, as the Talking Heads once did, “How did I get here?”

This phenomenon has been called everything from “highway hypnosis” to the “time-gap experience,” and while it has long puzzled people who study driving, it is still not fully understood. What is known is that it usually happens in fairly monotonous or familiar driving situations. Some scientists suggest that it’s related to drowsiness, and that we may even be taking what are called “microsleeps” at the wheel.

What is also unclear is how much attention we were actually paying to the road while under the spell of highway hypnosis versus to what extent we have simply forgotten everything that happened during that period. You may have wondered why you did not drift off the side of the road. Perhaps you were lucky; one study that had subjects drive for several (boring) hours in a driving simulator found that the roughly one in five drivers who succumbed to “driving without awareness”—as measured by EEG readings and eye movements—drifted out of their lane one-third of the time. You may have wondered what would have happened if a car (or bike or small child) had veered into the lane while you were zoning out. Would you have responded in time? Did a near accident almost happen during that period, one that you have since forgotten about?

Think back to the blank stares of drivers monitored by DriveCam. Why is it so hard to pay attention while we are driving? How and why do our eyes and mind betray us on the road?

Driving, for most of us, is what psychologists call an “overlearned” activity. It is something we’re so well practiced at that we’re able to do it without much conscious thought. That makes our life easier, and it is how we become good at things. Think of an expert tennis player. A serve is a complex maneuver with many different components, but the better we become at it, the less we think of each individual step. This example comes from Barry Kantowitz, a psychologist and “human factors” expert at the University of Michigan; he has spent years studying the safest and most efficient ways for humans to interact with machines, working with everyone from NASA pilots to operators of nuclear power plants. “One of the interesting things about learning and attention is that once something becomes automated, it gets executed in a rapid string of events,” he says. “If you try to pay attention, you screw it up.” This is why, for example, the best hitters in baseball do not necessarily make the best hitting coaches. Coaches need to be able to explain what to do; Charley Lau, the legendary batting coach and author of the classic book The Art of Hitting .300, never actually hit .300 himself.

The more overlearned an activity becomes, the less cognitive workload it imposes—though studies suggest that even the most mundane activities, like switching gears, never become fully automatic. The task always costs something. Having less workload is, on the one hand, a good thing. If, while driving, we were to really process every potential hazard, carefully analyze every motion and decision, and break down each maneuver into its component parts, we would quickly become overwhelmed. People who bring test subjects into driving simulators find something like this happening. “We’re not going to get a driver to be one hundred percent vigilant to the driving task, because we would all get out of the car sweating,” according to Jeffrey Muttart, a crash investigator and researcher at the University of Massachusetts. “If you see people get out of a driving simulator test, almost the first thing they do is take a deep, cleansing breath. Because I’m frying their brains. This is a ten-minute drive, and they want to try hard to do well.”

Too little workload has its own problems. We get bored. We get tired. We lapse into highway hypnosis. We may make errors. Anyone who has (like me) put on mismatched socks or run the coffeemaker without adding coffee or water will be aware of this phenomenon. The absolute ease of the activity allows the mind to wander. A classic psychological principle, the Yerkes-Dodson law, posits that the ability to learn is harmed by too little—or too much—“arousal.” This idea applies as well to human performance. Driving in North Dakota is on the low side of the curve, driving in Delhi on the high side. The ideal conditions presumably lie somewhere in between.

But where? Most driving rarely requires our full workload. So we listen to the radio, look out the window, or, increasingly, talk on the cell phone or read text messages—in the case of one fatal crash in California, the driver may have been operating a laptop computer as he drove. Or we may change the way we drive—we speed up because driving does not seem overly taxing. To the extent that this keeps us in the middle of the Yerkes-Dodson curve, it’s a good thing. But the problem with driving is that we never know for sure when things are going to change very quickly, when that nice empty road—seemingly safe for a cell phone conversation—is going to turn into an obstacle course. We may also be unaware of just how much workload our secondary activity is consuming.

“Let’s say you’re driving on a straight road. It’s relatively easy. I could ask you to do arithmetic at the same time and it wouldn’t mess up your driving,” Kantowitz said. “If you’re driving on a curved road, especially if it’s sharp curve, that takes more attention if you’re to keep the car operating safely within the lane. If I ask you to do mental arithmetic on a curve you’ll do it more slowly and you’ll screw it up. Or if you do it well you’ll screw up the driving.” A study by a Danish researcher found that those same types of arithmetic problems took longer to do when driving in a village than on a highway.

This raises another point: Researchers look at how driving is affected when people do other things, but research also shows that secondary tasks suffer as well. We become worse drivers and worse talkers. This is obvious to anyone who has listened to the wandering, interrupted musings of a driver talking on a phone (journalists know that people calling from their cars give terrible interviews). As Kantowitz put it, “There’s no free lunch.”

“My basic belief after almost forty years of studying this stuff is that people can’t time-share at all,” Kantowitz told me. “You only get the appearance. It’s like speed-reading. You think you can read really fast but your comprehension disappears. You can give the illusion of time-sharing if it’s simple information, but in general we’re not built for time-sharing.” Think of the annoying crawl type found on the bottom of the screen on CNN and other news networks. We are led to believe that this is how people now process information, as if we are suddenly genetically programmed to multitask. Studies have shown, however, that the more information there is on the screen, the less we actually remember.

The relative ease of most driving lures us into thinking we can get away with doing other things. Indeed, those other things, like listening to the radio, can help when driving itself is threatening to cause fatigue. But we buy into the myth of multitasking with little actual knowledge of how much we can really add in or, as with the television news, how much we are missing. As the inner life of the driver begins to come into focus, it is becoming clear not only that distraction is the single biggest problem on the road but that we have little concept of just how distracted we are.

In the largest study to date of the way we actually drive today, the Virginia Tech Transportation Institute, working with NHTSA, equipped one hundred cars in the Washington, D.C., and northern Virginia area with cameras, GPS units, and other monitoring devices, and then set about recording a year’s worth of what it calls “pre-crash, naturalistic driving data.” After poring over forty-three thousand hours of data and more than two million miles of driving, the study found that almost 80 percent of crashes and 65 percent of the near crashes involved drivers who were not paying attention to traffic for up to three seconds before the event.

That period of time is critical. “A total time of two seconds looking away from the forward roadway is when people start to get in trouble,” explained Sheila “Charlie” Klauer, a researcher at VTTI and the study’s project manager. “That’s when they get to the point when they are starting to lose track of what’s going on in front of them.” The two-second window is not technically related to the “two-second rule” for following distance, but the comparison is instructive. The point is that a lot can happen in two seconds—like colliding with the car in front if it came to a stop or slowed—but drivers, lulled by the expectancy that it will not stop, drive as if the world will not have changed when they return their eyes to the road after that two seconds. They drive as if the world is a television show viewed on TiVo that can be paused in real time—one can duck out for a moment, grab a beer from the fridge, and come back to right where they left off without missing a beat. For many of the crashes, Klauer found that “the eye glance happened to be at exactly the wrong time. If they had not chosen to look away at that very second they would have probably been okay.”

The sources of distraction inside a car have been painstakingly logged by researchers. We know that the average driver adjusts their radio 7.4 times per hour of driving, that their attention is diverted 8.1 times per hour by infants, and that they search for something—sunglasses, breath mints, change for the toll—10.8 times per hour. Research has further revealed just how many times we glance off the road to do these things and how long each glance takes: In general, the average driver looks away from the road for .06 seconds every 3.4 seconds. “On average, radio tuning takes seven glances plus or minus three,” said Linda Angell, a safety researcher at General Motors, in a conference room at the Technical Center in Warren, Michigan. “That’s for an oldish radio. We do better with the modern radio, which zeroes you in on the right region.” Most of these glances, Angell noted, do not take our eyes off the road for longer than 1.5 seconds. But there are exceptions, such as “intense displays”(e.g., lots of features) or looking for a button you have not pressed in a while. The iPod is changing the equation yet again: Studies have shown that scrolling for a particular song takes our eyes off the road for 10 percent longer than simply pausing or skipping a song—plenty of time for something to go wrong.

Even a succession of very short glances, less than two seconds each, can cause problems. Researchers talk of the “fifteen-second rule,” which indicates the maximum amount of time a driver should spend operating any kind of in-car device, whether navigation or radio, even as they are (at least occasionally) looking at the road. “What we believe is that task time is very important,” Klauer said. “The longer the task time, the more dangerous the task is, and the greater the crash risk.” And so a fifteen-second task might require only short glances at the device, but, Klauer said, “that risk increases every time the driver looks away.”

The study found that while dialing a cell phone put drivers at a greater crash risk, talking on a cell phone presented only a slightly higher risk than normal driving. “When a driver is talking or listening on their cell phone, at any given moment within that conversation what our odds ratio is telling us is they’re only at a slightly higher crash risk than an alert driver. Statistically speaking, it’s not different,” Klauer said. Does that mean talking on a cell phone is safe? Maybe it’s all that dialing we need to worry about. But the study also found that talking (or listening) on a cell phone was a contributing factor in as many crashes as dialing was. “We think that’s probably true because while dialing is a much more dangerous task while the driver’s doing it, the task is fairly short,” Klauer told me. “But drivers typically talk on their cell phone for a long period of time. Over that long period of time a lot more crashes and near crashes are more apt to occur. That slight increase in crash risk is starting to add up.” As more drivers talk for longer periods, Klauer said, “it’s going to become a lot more dangerous.”

The reason we talk for a long time on our cell phones is related to the reason we all think we are better drivers than we are, and to the thing that also makes us think we are better drivers on our cell phones than we are: lack of feedback. Cell phone users are not aware of the risk because, by all surface measures, they seem to be driving fine. Traffic affords us these illusions—until it does not, as the hundred-car study showed. “Cell phone conversations are particularly insidious because you don’t notice your bad performance, particularly the cognitive side,” John Lee argues. “So if you’re dialing the phone, you get immediate feedback because you don’t quite stay in the lane, because you’re punching the buttons.” Once the dialing is done, the driver can again look at the road. The weaving stops. They seem to be in control.

Drivers may confidently assume they can adequately compensate for talking on a cell phone or texting on a BlackBerry by lowering their speed or putting more space between their own car and the car ahead of them, but the evidence gleaned from the hundred-car survey suggests otherwise. One might think, for example, that rear-end collisions most commonly occur because the driver behind was following too closely. Yet the study found that the majority of rear-end crashes happened when the following car was more than two seconds away from the car it struck. “I think people compensated a little bit for their inattention,” Klauer said. “‘I need to answer this cell phone, I need to look at these papers on the seat next to me.’ So they back off the lead vehicle and give themselves some space. Then they start to engage in something else. Then something unexpected happens and they’re in trouble.”

The drivers were redistributing workload. With more of their attention devoted to a cell phone conversation, they may have had to work just a bit harder to stay in their lane; similarly, the narrower the lane, the more mental energy it takes to stay in that lane (my own theory is that cell phones in cars have contributed to the seeming death of signaling for turns). Driving closer to someone also requires more mental energy, as does driving fast. We can usually feel this starting to take a toll, so we do things like drop back from a car in front of us or slow down. Clearly we do not always compensate enough, and there is evidence to suggest that we hardly compensate at all for our cell phone impairment when we’re doing things like changing lanes.

Something similar happens with very new drivers on highways: So much of their mental concentration is devoted to simply staying in the lane, they have trouble paying attention to their speed. And it is not only drivers who suffer, as anyone who has walked behind someone talking on a mobile phone has noticed. When psychologists have asked people to walk around a track while memorizing words that were shown to them, walking speeds slowed as the mental task got harder. Similarly, researchers in Finland have found that pedestrians using mobile devices walked more slowly and were less able to interact with the device, pausing occasionally to “sample the environment.” But pedestrians on cell phones do not sample the environment as often as they should, as a study of a Las Vegas crosswalk showed: Those talking on cell phones were less likely to look at traffic while crossing and took longer to do so.

Our attention, like a highway dropping from three lanes to two lanes, suffers from a bottleneck, one theory claims: Only so much can get through at once. Trying to squeeze more mental “cars” past the bottleneck means we have to slow them all down, space them out—or it means that some of those cars might drive off the road. In the hundred-car study, something else was also happening when drivers got on their cell phones. They began to look almost exclusively straight ahead, much more so than they did when they were not on their cell phones. They were, by external measures, “paying attention.” But keeping one’s eyes on the road is not necessarily the same thing as keeping one’s mind on the road.

Consider for a moment the incredibly complex question of what it even means to pay attention while driving. There are an infinite number of things we could notice if we chose to, or had the spare mental capacity. But through practice and habit we learn to expertly analyze complicated scenes and extract only the information we need, ignoring the rest. New drivers, as we have seen, look rather rigidly ahead and near the front of the car, using “foveal” rather than peripheral vision to help them stay in their lane. As drivers get more experienced, they cast their eyes farther out along the road, barely registering the pavement markings. This happens without their even noticing. Experiments have been done in which researchers pulled over drivers on the highway and asked them if they recalled having seen certain traffic signs. The recall rates were as low as 20 percent. Were drivers simply not seeing things? One study found that the remembered signs were not necessarily the most visible ones but the signs that drivers judged most important (e.g., speed limit). This suggests that drivers saw enough of the signs to process what they were, at some subconscious level, and then effectively forgot most of them.

We do this sort of thing all the time—and for good reason. Remembering traffic signs we have seen is not useful to our lives. Steven Most, a psychologist at the University of Delaware, compares the flow of information and images we get in daily life to a stream passing through our heads. Unless we stop to “scoop up” some of that water—or “capture” it with our attention—it will flow in and out of our minds. “Sometimes, you attend to things enough to be aware of them in the moment, but that encoding process isn’t necessarily taking place,” he told me. “The awareness is there but not the memory of the awareness. When attention is distracted enough, it’s even questionable whether we have that momentary awareness.”

The reason we notice things like signs while driving is not as simple as it might seem. The average driver, asked why he saw a stop sign, might say, “Because it was there” or “Because it’s the color red, and humans are hard-wired to see red more easily.” But often we see a sign simply because we know where to look for one. This curious fact was explained by Carl Andersen, a vision specialist with the Federal Highway Administration, in a laboratory filled with eye-catching prototype warning signs in bold new colors like “incident pink.” “If drivers are in an area that they already know, they almost don’t even see the sign, because they already know it’s there,” Andersen said. This is known as “top-down processing.” We see something because we are looking for it. To see things that we are not looking for, like unexpected stop signs, we need to rely on “bottom-up processing.” Something has to be conspicuous enough to catch our attention. “If you’re on one of those divided state highways, the older highways, you’re not expecting to stop,” Andersen said. “You’d better have advance signing and reduce the speed to prepare people for it.”

Drivers actually look at most traffic signs at least twice: once for “acquisition” and again for “confirmation.” Curiously, we do not really read things like stop signs. “Studies have been done where they intentionally misspell ‘stop,’” Andersen said. “Everybody stops and then they drive off. They query the people later and the vast majority never saw that it was misspelled.” (In fact, they may not have even seen it; it’s estimated that one-fifth of our viewing time is interrupted by blinks and what are known as saccades, or our eyes’ rapid movements, during which we are, as one expert puts it, “effectively blind.”) Other studies, in driving simulators, have done things like change “No Parking” signs briefly to stop signs, and then back again. When the signs were at intersections, where stop signs usually are, drivers were more likely to notice the change. When they popped up elsewhere (e.g., at mid-block), drivers hardly ever noticed the change. When the drivers did see the sign change from “No Parking” to “Stop” at the intersection, they did not see it change back to “No Parking.” Their decision to stop, the researchers noted, had already been made.

What does this have to do with real driving? After all, traffic signs do not change capriciously. A lot of things in traffic do change, however, and the question of whether we will notice those things depends not just on how visible they are but, indeed, on whether or not we are looking for them and how much spare capacity we have to process them. In a now-famous psychological experiment, a group of researchers had subjects view a video that showed a circle of people passing a basketball around. Half wore white shirts, half wore black. The subjects were asked to count the number of passes. What at least half the subjects did not notice was that a person wearing a gorilla suit passed right through the middle of the circle of basketball players. They were suffering from what has been called “inattentional blindness.”

The idea that people could not see something as striking as a gorilla in a group of basketball players, although their eyes were locked on the video screen, suggests just how unstable and selective attention is—even when we are giving something our “undivided” attention. “There’s an unlimited amount of information in the world, but our capacity for attending to information is pretty limited,” explained Daniel Simons, a psychologist at the University of Illinois and the coauthor of the gorilla study. “If you’re limited in how many things you can pay attention to, and attention is a gateway to consciousness, then you can only be aware of a limited subset of what’s out there.”

Inattentional blindness, it has been suggested, is behind an entire category of crashes in traffic, those known as “looked but did not see accidents.” As with the gorilla-experiment subjects, drivers were looking directly at a scene but somehow missed a vital part—perhaps because they were looking for something else, or perhaps because something came along that they were not looking for. All too often, for instance, cars collide with motorcycles. One of the most frequently cited reasons is “failure to see,” and these events are so common that motorcyclists in England have taken to calling them SMIDSYs, for “Sorry, Mate, I Didn’t See You.”

Many people assume that “failure to see” means that the motorcycle itself was difficult to see, because of its size or its single headlight. But it may also be that car drivers tend to be on the lookout for other cars when entering an intersection or turning across a lane of oncoming traffic. They may be in a sense “looking through” the motorcycle, because it does not fit their mental picture of the things they think they should be seeing. This is why safety campaigns (e.g., “Watch for motorcycles” or the United Kingdom’s “Take longer to look for bikes”) stress the idea of drivers simply being aware that motorcycles are out on the road. “The common intuition is that we first see things in the world and then interpret the scene in front of us,” said Most. “What this work shows is that it’s possible that the idea you have in mind actually precedes the perception and affects what you see. Our expectations and knowledge of what’s in a scene influence what we see in a scene.”

These expectations might also help explain the troublingly high numbers of emergency vehicles that are struck on the highway, even as they sit on the shoulder with their lights flashing brightly (and despite the fact that most places have laws requiring drivers to change lanes or slow down in the presence of an ambulance). These incidents are so common that the term “moth effect” has been coined for them. The idea is that drivers are lured to the lights, like moths to a flame.

What could cause a moth effect? There are many theories, ranging from arguments that we tend to steer where we look (which raises the question of why we do not drive off the road every time we see something interesting) to the idea that humans instinctively look toward light (ditto). Other researchers have argued that the fixation of attention on the roadside leaves drivers less able to judge their position in the lane. Many moth effect crashes involve alcohol-impaired drivers, perhaps no surprise in light of work that suggests that alcohol has a particularly deleterious effect on our eyes’ ability to perceive depth or direction while we are moving.

The simplest explanation may be that most drivers, upon seeing a car on the highway, assume that it is moving at the same high speed as everyone else—and cars with flashing lights are usually moving even faster than that. One study, conducted in a driving simulator, showed that drivers reacted more quickly when stopped police cars were parked at an angle to oncoming traffic, rather than straight ahead in the direction of traffic. As the two vehicles were essentially equally conspicuous, the reason the angled car was seen sooner had less to do with visibility than in how the drivers interpreted what they saw: a car that was obviously not moving in the direction of traffic. (This ability to interpret seemed to be a by-product of driving experience, as novice drivers had the same reaction times for both cars.)

Even when we see an unexpected hazard, the fact that it’s outside our “attentional set” means we are slower to react to it. This is demonstrated in a classic psychological test for what is known as the “Stroop effect.” Subjects are shown a list of color names; these words are printed in the same color as the name as well as in other colors. Naming the color a word is printed in, it turns out, typically takes longer when the word does not match the color; that is, it takes longer to say “red” when the word printed in red is “yellow” than when it’s “red.” One argument for why this happens is that while reading is for us an “automatic” activity, naming colors is not. The automatic gets in the way of the less automatic (as with the stereotyping studies in Chapter 1). But other theories suggest that attention is involved. That we can name the correct color when the word itself is “wrong” suggests that we can train our attention on certain things; yet the fact that it takes us longer to do it shows that we cannot always screen out the things on which we are not focused (i.e., the word itself).

What this means for traffic was highlighted in a study by Most and his colleague Robert Astur. Drivers on a computer driving simulator, navigating through an ersatz urban environment, were asked to look for an arrow at every intersection telling them where to turn. For some drivers, the arrow was yellow and for others it was blue. At one intersection, an approaching motorcycle, itself either blue or yellow, suddenly veered in front of the driver and stopped. Drivers’ reaction times to slam on the brakes were slower—and their collision rates were higher—when the motorcycle was a different color than the arrow. In a purely bottom-up form of processing, we might expect the motorcycle to stand out because it is different; but because we are looking at the scene from a top-down perspective, the odd-colored motorcycle is less visible because it is different from those things for which we are searching.

This attention disorder could also help explain the “safety in numbers” phenomenon of traffic, as described by Peter Lyndon Jacobsen, a public-health consultant in California. You might think that as there are more pedestrians or cyclists on a street, the more chances there are for them to be hit. You are right. More pedestrians are killed by cars in New York City than anywhere else in the United States. But as Jacobsen found, these relationships are not linear. In other words, as the number of pedestrians or cyclists increases, the fatality rates per capita begin to drop. The reason, as Jacobsen points out, is not that pedestrians begin to act more safely when surrounded by more fellow pedestrians—in fact, in New York City, as a stroll down Fifth Avenue will reveal, the opposite is true. It is the behavior of drivers that changes. They are suddenly seeing pedestrians everywhere. The more they see, typically, the slower they drive; and, in a neatly perpetuating cycle, the more slowly they drive, the more pedestrians they effectually see because those pedestrians stay within sight for a longer period.

And so New York City, when one considers how many pedestrians it has, is actually one of the safest cities in the country for walkers. (One study, looking at 1997-98 figures, found the Tampa-St. Petersburg-Clearwater area to be the most dangerous for pedestrians.) To cite another instance, the Netherlands has a much lower fatality rate per mile traveled for cyclists than does the United States. It is not likely that Dutch cyclists are any more visible in terms of pure conspicuity; they rarely wear reflective clothing, favoring stylish black coats instead, and instead of flashing lights their bikes carry things like tulips. Nor do the Dutch more regularly wear helmets than American cyclists; the reverse is actually true. Perhaps the Dutch just have better bike paths, or maybe the flat landscape makes it easier for drivers to spot cyclists. But the most compelling argument is that Dutch cyclists are safer simply because there are more of them, and thus Dutch drivers are more used to seeing them. Dutch culture may be quite different from American culture, but the “safety in numbers” theory also holds for comparisons within the United States—in Florida, for example, Gainesville, a college town with the highest cycling rate in the state, is in fact the safest place to be a cyclist. The lesson: When you see more of something, you’re more likely to see that thing.

In the gorilla experiment, an added condition made subjects less likely to see the gorilla: when their job got harder. Some subjects were asked to count not just passes but the types of passes—whether they were “bounce passes” or passes made in the air. “You’ve made the attention task that much harder, and used up more of your available resources,” Simons said. “You’re less likely to notice something unexpected.”

In driving, you might protest, we do not do such things as tally basketball passes. Still, there may have been times when you were concentrating so much on looking for a parking spot that you did not notice a stop sign; or you might have almost hit a cyclist because she was riding against traffic, violating your sense of what you expected to see. And there is another activity, one that we increasingly often indulge in while driving, that closely resembles that very specific act of counting basketball passes: talking on a cell phone.

Let me ask you two questions: What route did you take to get home today? And what was the color of your first car? What just happened? Chances are, your eyes drifted away from the page. Humans, perhaps to free up mental resources, tend to look away when asked to remember something. (Indeed, moving the eyes is thought to aid memory.) The more difficult the act of remembering, the longer the gaze away. Even if your eyes had remained on the page, you would have been momentarily sent away in a reverie of thought. Now picture driving down a street, talking to someone on a mobile phone, and they ask you to retrieve some relatively complicated bit of information: to give them directions or tell them where you left the spare keys. Your eyes may remain on the road, but would your mind?

Studies show that so-called visual-spatial tasks, such as rotating a letter or a shape in one’s mind, cause our eyes to fixate longer in one place than when we are asked to perform verbal tasks. The longer the fixation, the thinking goes, the more attention we are devoting to the task—and the less we’re giving to other things, like driving. The mere act of “switching” tasks—like moving from solely driving to talking on the phone while driving or, say, to changing whom we’re speaking to within the same cell phone call via call waiting—takes its toll on our mental workload. The fact that the audio information we are getting (the conversation) comes from a different direction than the visual information we are seeing (the road ahead) makes it harder for us to process things. Bad reception on the phone? Our struggle to listen more carefully consumes even more effort.

Now replace the gorilla of the basketball experiment with a car making an unexpected turn or a child on a bike standing near the side of the road. How many of us would see it? “Driving’s already attention-demanding enough—if you add in the cognitive demands of talking on a cell phone, you’re taking away whatever limited resources you had, and you’re that much less likely to notice something unexpected,” Simons said. “You might be able to stay on the road just fine, and you might be able to stay the same distance behind a car on the highway, but if something unexpected happens—a deer runs into the highway—you might not react as easily.”

The notion that we could miss unexpected things while talking on a cell phone is powerfully demonstrated by our seeming failure to notice the expected things. Two psychologists at the University of Utah found, after running a number of subjects through a simulator test, that drivers not talking on a cell phone were able to remember more objects during the course of the drive than those who were. The objects ranged in their “driving relevance” that is, the researchers ranked speed-limit signs and those warning about curves as more critical than Adopt-a-Highway signs. You might suspect that the cell phone drivers were just filtering out irrelevant information, but the study found no correlation between what was important and what was remembered. Most strikingly, the drivers using cell phones looked at the same number of objects as the drivers without cell phones—yet they still remembered fewer.

Drivers using a cell phone, as noted in the hundred-car study, tend to rigidly lock their eyes ahead, assuming a super-vigilant pose. But that stare may be surprisingly hollow. In a study with an admittedly small sample size, I took the wheel of a 1995 Saturn one day at the Human Performance Laboratory at the University of Massachusetts in Amherst, and got set for a virtual drive in the lab’s simulator. While I drove down a four-lane highway, a series of sentences was read to me via a hands-free cell phone. My task was to first judge whether the sentences made sense or not (e.g., “The cow jumped over the moon”) and then repeat (or “shadow,” as researchers call it) the last word in the sentence. As I did this, the direction of my gaze (among other things) was being monitored via an eye-tracking device mounted to a pair of Bono-style sunglasses.

When I later watched a tape of my drive that plotted where my eyes had been looking, the pattern was striking. Under normal driving, my eyes danced around the screen, taking in signs, the speedometer, construction crews in a work zone, the video-game landscape. When I was on the phone, trying to discern whether the sentence made sense, my eyes seemed to train on a point very close to the front of the car—and they barely moved. Technically, I was looking ahead—my eyes were “on the road”—but they were gazing at a place that would not be useful in spotting any hazards coming from the side or even, say, determining whether the truck several hundred feet ahead might be stopping. Which is exactly why I smashed into its rear end. “You were driving like a sixteen-year-old” is how Jeffrey Muttart described it to me.

Our eyes and our attention are a slippery pair. They need each other’s help to function, but they do not always share the load equally. Sometimes we send our eyes somewhere and our attention follows; sometimes our attention is already there, waiting for the eyes to catch up. Sometimes our attention does not think that everything our eyes are seeing is worth its time and trouble, and sometimes our eyes rudely interrupt our attention just as it’s in the middle of something really interesting. Suffice it to say that what we see, or what we think we see, is not always what we get. “This is the reason the whole ‘keep your eyes on the road, your hands upon the wheel, use the hands-free handset’ idea is a silly thing,” Simons said. “Having your eyes on the road doesn’t do any good unless your attention is on the road too.”

As with the subjects in the counting test who did not see the gorilla, drivers (and particularly drivers talking on cell phones) would be shocked to learn, later, what they missed—precisely those things the in-car cameras are now revealing. “It is striking that people miss this stuff,” Simons said. “At some level it’s even more striking how wrong our intuitions are about it. Most people are firmly convinced they would notice if something unexpected happened, and that intuition is just completely wrong.”

Human attention, in the best of circumstances, is a fluid but fragile entity, prone to glaring gaps, subtle distortions, and unwelcome interruptions. Beyond a certain threshold, the more that is asked of it, the less well it performs. When this happens in a psychological experiment, it is interesting. When it happens in traffic, it can be fatal.

Objects in Traffic Are More Complicated Than They Appear:
How Our Driving Eyes Deceive Us

Try to picture, for a moment, the white stripes that divide the lanes on a major highway. How long would you guess they are? How much space would you say lies between each stripe? When first asked this question, I guessed about five feet, with maybe fifteen feet between the stripes. You might estimate six or even seven feet. While the exact length varies, the U.S. standard calls for ten feet, though depending on the speed limit of the road, the stripes may be as long as twelve or fourteen feet. Take a look at an overhead photo of a highway: In most cases, the stripe is as long as, or longer than, the cars themselves (the average passenger car is 12.8 feet). The spacing between the stripes is based on a standard three-to-one ratio; thus, for a twelve-foot stripe, there will be thirty-six feet between stripes.

I use this as a simple example of how what we see is not always what we get as we move in the unnaturally high speeds of traffic. You may be wondering how it is that humans can even do things like drive cars or fly planes, moving at speeds well beyond that ever experienced in our evolutionary history. As the naturalist Robert Winkler points out, creatures like hawks, whose eyes possess a much faster “flicker fusion rate” than humans’, can track small prey from high above as they dive at well over 100 miles per hour. The short answer is that we cheat. We make the driving environment as simple as possible, with smooth, wide roads marked by enormous signs and white lines that are purposely placed far apart to trick us into thinking we are not moving as fast as we are. It is a toddler’s view of the world, a landscape of outsized, brightly colored objects and flashing lights, with harnesses and safety barriers that protect us as we exceed our own underdeveloped capabilities.

What we see while driving is a visually impoverished view of the world. As Stephen Lea, a researcher at the University of Exeter, explains it, what matters is less the speed at which we or other things move than the rate at which images expand on our retinas. So in the same way that we easily observe a person 3 yards away jogging toward us at 6 miles per hour, we have little trouble tracking a car that is 30 yards away moving at 60 miles per hour. The “retinal speed” is the same.

While driving, we get a gently undulating forward view. Things are far away or moving at similar speeds, so they grow slowly in our eyes, until that moment when the car in front suddenly and jarringly “looms” into view (and you notice their bumper sticker: IF YOU CAN READ THIS, YOU’RE TOO CLOSE). But now picture looking directly down at the road while you’re driving at a good speed. It is, of course, a blur. This is no less part of the actual environment in which we are driving, but we are physically unable to see it with any accuracy. Luckily, we do not usually need to see it to move safely—though, as we shall learn, there are other ways in which traffic puts our visual systems to severe tests.

Traffic illusions actually hit us before we even get in the car. You may have noticed how in movies or on television, the spokes on a car’s wheels sometimes seem to be moving “backward.” This so-called wagon-wheel effect happens in movies because they are composed of a flickering set of images (generally twenty-four frames per second), even though we perceive them to be smooth and uninterrupted. Like the dancers in a disco captured briefly by a strobe light, each frame of that movie captures an image of the spokes. If the frequency of the wheel’s rotation perfectly matched the flicker rate of the film, the wheel would appear not to be moving. (“I replaced the headlights in my car with strobe lights,” the comedian Steven Wright once joked, “so it looks like I’m the only one moving.”) As the wheel moves faster, though, each spoke is “captured” at a different place with each frame (e.g., we may see a spoke at the twelve o’clock position on one sweep, but at eleven forty-five on the next). So it seemingly begins to move backward.

As the cognitive psychologists Dale Purves and Tim Andrews note, however, the wagon-wheel effect can happen in real life as well, under full sunlight, when the “stroboscopic” effect of movies does not apply. The reason we still see the effect, they suggest, is that, as with movies, we perceive the world not as a continuous flow but in a series of discrete and sequential “frames.” At a certain point the rotation of the wheel begins to exceed the brain’s ability to process it, and as we struggle to catch up, we begin to confuse the current stimulus (i.e., the spoke) in real time with the stimulus in a previous frame. The car wheel is not spinning backward, any more than disco dancers are moving in slow motion. But this effect should provide an early, and cautionary, clue to some of the visual curiosities of the road.

“Motion parallax,” one of the most famous highway illusions, puzzled psychologists long before the car arrived. This phenomenon can be most easily glimpsed when you look out the side window of a moving car (though it can happen anywhere). The foreground whizzes past, while trees and other objects farther out seem to move by more slowly, and things far in the distance, like mountains, seem to move in the same direction as us. Obviously, we cannot make the mountains move, no matter how fast we may drive. What’s happening is that as we fixate on an object in that landscape, our eyes, to maintain their fixation, must move in a direction opposite to the way we’re going. Wherever we fixate in that view, the things we see before the point of fixation are moving quickly across our retina opposite to the direction we are moving in, while things past the point are moving slowly across our retina in the same direction as we’re traveling. (See the notes for a quick demonstration of motion parallax.)

All this eye movement and the relative motion of the objects we are seeing, as confusing as it seems, help us judge how far away things are from us. As Mark Nawrot, a psychologist at North Dakota State University and an expert in motion parallax, describes it, this is why film directors like Peter Jackson like to move the camera around a lot. Because we are sitting, stationary, in a theater, and thus cannot get the sort of depth cues our eyes give us when we move, Jackson moves the camera instead, to make the film appear more realistic. But the price we pay for the depth cues that motion parallax provides us is the occasional illusion that we may or may not consciously notice. In traffic, motion parallax may trick us into thinking that an object is far and stationary when, in reality, it is near and moving.

The mind can play tricks on what we see, but motion parallax reminds us that what we see while driving plays tricks on our minds. Sense and perception are connected by a quite busy two-way street. The white stripes on the highway and the distance between them are designed precisely as an illusion, to make these high speeds seem comfortable. If both the stripes and the distance between them were short, the experience might feel nauseating. In fact, in some places, engineers have tried to exploit this by employing “illusory pavement markings” to make drivers think they are going faster than they are. In one trial, a series of arrowlike chevrons were painted, ever closer together, on a highway exit ramp. The theory was that as the drivers began to pass more chevrons for each moment they drove, it would appear as if they were going faster than they really were, and would thus slow down. That study did find that drivers reduced their speed, but in other trials the results have been mixed. Drivers may slow once or twice simply because there are strange markings on the pavement, but they may also quickly acclimate to the markings.

These experiments have been focused on exit ramps because they are a statistically dangerous part of the highway. One crucial reason involves a particular illusion we face in traffic: “speed adaptation.” Have you ever noticed, when driving from a rural highway onto a village road with a lower speed limit, how absolutely slow it feels? When you again leave that town to rejoin the rural highway and its higher speed, does the disparity seem as noticeable? The longer we drive at high speeds, the harder it is for us to slow down. Studies have shown that drivers who drove for at least a few minutes at 70 miles per hour drove up to 15 miles per hour faster when they hit a 30-miles-per-hour zone than drivers who had not previously been traveling at the higher speed.

The reason, as Robert Gray, a cognitive psychologist at the University of Arizona, explained to me, is something that might be called the “treadmill effect.” After running on a treadmill for a while, you may have noticed that the moment you stop you may briefly experience the sensation of moving backward. As Gray describes it with driving, neurons in the brain that track forward movement begin to become fatigued as a person looking ahead drives at the same speed for a time. The fatigued neurons begin to produce, in essence, a negative “output.” When a person stops (or slows), the neurons that track backward motion are still effectively dormant, but the negative output of the forward neurons fools you into thinking you’re moving backward—or, if you’re changing from high speeds to lower speeds, it can fool you into thinking you have slowed more than you actually have. The illusion cuts both ways, studies suggest: We underestimate our speed when asked to slow down and overestimate our speed when asked to speed up. This helps explain why we often go too fast coming off a highway (and hence the chevron patterns); it might also explain why drivers entering a highway frequently fail to reach the speed of traffic by the time they’re merging (frustrating those in the right-hand lane who are forced to slow).

We misjudge speed in all kinds of ways. Our general perception of how fast and in what direction we are moving—indeed that we are moving at all—comes largely, it is thought, from what has been called “global optical flow.” When we drive (or walk), we orient ourselves via a fixed point on the horizon, our “target.” As we move, we try to align that target so that it is always the so-called focus of expansion, the nonmoving point from which the visual scenes seem to flow, approaching us in a kind of radial pattern—think of the moment in Star Wars when the Millennium Falcon goes into warp speed and the stars blur into a set of lines streaming away from the center of the ship’s trajectory. The “locomotor flow line”—or what you and I would call the road—is the most crucial part of the optic field in driving, and the “textural density” of what passes by us influences our sense of speed. Things like roadside trees or walls affect the texture as well, which is why drivers overestimate their speed on tree-lined roads, and why traffic tends to slow between noise-barrier “tunnels” on the highway. The finer the texture, the faster your speed will seem.

The fineness of the road texture is itself affected by the height at which it is viewed. We sense more of the road’s optical flow the closer we are to it. When the Boeing 747 was first introduced, as the psychologist Christopher Wickens has noted, pilots seemed to be taxiing too fast, on several occasions even damaging the landing gear. Why? The new cockpit was twice as high as the old one, meaning that the pilots were getting half the optical flow at the same speed. They were going faster than they thought they were. This phenomenon occurs on the road as well. Studies have shown that drivers seated at higher eye heights but not shown a speedometer will drive faster than those at lower heights. Drivers in SUVs and pickups, already at a higher risk for rollovers, may put themselves at further risk by going faster than they intend to. Studies have shown, perhaps not surprisingly, that SUV and pickup drivers speed more than others.

The reason we have speedometers, and why you should pay attention to yours, is that drivers often do not have a clue about how fast they’re actually traveling—even when they think they do. A study in New Zealand measured the speed of drivers as they passed children playing with a ball and waiting to cross the street. When questioned, drivers thought they were going at least 20 kilometers per hour (or about 12 miles per hour) more slowly than they really were (i.e., they thought they were going 18 to 25 miles per hour when they were really doing 31 to 37). Sometimes it seems as if we need someone standing on the side of the road, actually reminding us how fast we are really going. This is why we see “speed trailers,” those electronic signs posted by the road that flash your speed. These plaintive appeals to conscience are usually effective, at least in the immediate vicinity, at getting drivers to slow down slightly—but whether drivers want to keep slowing down, day after day, is another issue. The speed trailers work, when they do, because they give us crucial feedback—which, as mentioned in the previous chapter, we so often lack on the road. Some highway agencies, responding to rising numbers of often-fatal rear-end crashes, have tried to put feedback of sorts right on the road, in the form of painted dots that inform drivers of the proper following distances (in one case, someone responded by painting a dot-eating Pac-Man on the highway). Drivers’ following distances have tended to increase after dots are put down. Noise also gives feedback: We know we are going faster when the amount of road and wind noise picks up. The faster we go, the louder it gets. But have you ever found yourself listening to the radio at a high volume and then suddenly noticed you were speeding? A variety of studies have shown that when drivers lose auditory cues, they lose track of how fast they’re going.

The robot car Junior, as you will recall, did not need to be able to “see” brake lights because he knew exactly how far the car ahead of him was, to within a few meters. For humans, however, distance, like speed, is something we often judge rather imperfectly (hence the Pac-Man dots). Unfortunately for us, driving is really all about distance and speed. Consider a common and hazardous maneuver in driving: overtaking a car on a two-lane road as another approaches in the oncoming lane. When objects like cars are within twenty or thirty feet, we’re good at estimating how far away they are, thanks to our binocular vision (and the brain’s ability to construct a single 3-D image from the differing 2-D views each eye provides). Beyond that distance, both eyes are seeing the same view in parallel, and so things get a bit hazy. The farther out we go, the worse it gets: For a car that is twenty feet away, we might be accurate to within a few feet, but when it is three hundred yards away, we might be off by a hundred yards. Considering that it takes about 279 feet for a car traveling at 55 miles per hour to stop (assuming an ideal average reaction time of 1.5 seconds), you can appreciate the problem of overestimating how far away an approaching car is—especially when they’re approaching you at 55 miles per hour.

Since we cannot tell exactly how far away the approaching car might be, we guess using spatial cues, like its position relative to a roadside building or the car in front of us. We can also use the size of the oncoming car itself as a guide. We know it is approaching because its size is expanding, or “looming,” on our retina.

But there are a few problems with this. The first is that viewing objects straight-on, as with an approaching car, does not provide us with a lot of information. Think of an outfielder catching a fly ball—a seemingly simple act, but one whose exact mechanics still elude scientists (and the occasional outfielder). One thing that’s generally agreed upon, as University of Missouri psychology professor Mike Stadler notes, is that balls are harder to catch when they are hit directly at a fielder. Fielders often have trouble gauging distance and trajectory, and they find they need to move back or forth a bit to get a better picture; studies have shown that fielders have a harder time judging which balls can or cannot be caught when they are asked to stand still. Viewing a car head-on or directly from behind, as we almost universally do, is like viewing a baseball hit right at you: It doesn’t give us a lot to go on.

Another problem is that the image of that car, when it does begin to expand in our eyes, does not do so in a linear, or continuous, way. The book Forensic Aspects of Driver Perception and Response gives this example: A parked car that an approaching driver sees 1,000 feet away will double on the retina by the time the driver is 500 feet away. Sounds about right, no? But it will double again in the next 250 feet, and again in the last 250 feet. It is nonlinear. To put it another way, we can tell the car is getting closer—although this itself may take as much as several seconds—but we have no idea of the rate at which it is getting closer. This difficulty in judging closing distance also makes passing the lead car a problem; studies have shown that it is struck in about 10 percent of overtaking crashes. Another way to think about this is to imagine what happens to skydivers. For much of their fall, they have little sense, looking downward, of how fast they are falling—or even that they’re falling at all. But suddenly, as the distance to the ground begins to come within the limits of human perception, they experience what is called “ground rush,” with the terrain suddenly exploding into their range of view.

If all this was not enough to worry about, there’s also the problem of the oncoming car’s speed. A car in the distance approaching at 20 miles per hour makes passing easy, but what if it is doing 80 miles per hour? The problem is this: We cannot really tell the difference. Until, that is, the car gets much closer—by which time it might be too late to act on the information. One study that looked at how and when cars decided to pass other cars on two-lane highways found that they were as likely to attempt a pass when an oncoming car was approaching at 60 miles per hour as when it was coming at 30 miles per hour. Why? Because when the passing maneuver began, the cars were about 1,000 feet apart—too far to tell the speed of the opposing car. At those distances we are not even really sure if the car is coming toward us or not; the fact that it’s in the opposite lane, or that we can see its headlights, might be the only giveaway.

So at the crucial distance where one must make a decision, the driver has no idea of a key variable: the “closing rate” of the other car. This is why you may have been forced to rather suddenly abandon your attempted passing and make either a voluntary or a forced return to your own lane. We “cheat” like this regularly, relying on a car’s perceived distance without taking into account its speed. One study, looking at drivers’ left turns across oncoming traffic, found that when the speed of approaching cars was doubled, drivers’ estimates of the safe “gap” in which they could cross, which you would guess should have also doubled, went up by only 30 percent. These small discrepancies are the stuff of crashes.

Evidence suggests that we are sometimes fooled into thinking things are not as far away as they appear (and not only the approaching objects in our mirrors!). Studies have shown that people think small cars are farther away than they really are, either because we maintain a mental image of a larger car or because there is less of the car to actually see. Large objects, though, also create problems. Researchers have long been puzzled about the relatively high number of drivers killed while crossing railroad tracks—often when visibility was clear and warning signals were in place. It raises an obvious question: How could a driver not see something as large (and as loud) as a train? One answer is that a driver may have crossed the same set of tracks three hundred times in the last year without ever seeing a train, even when the signals were flashing. Did they simply not expect it on the 301st trip across the tracks? Did they “look but not see”? The influential psychologist and vision expert H. W. Leibowitz, in what has become known as the “Leibowitz hypothesis,” offered another possible explanation: biases in the drivers’ perceptual systems.

Large objects often seem to move more slowly than small objects. At airports, small private jets seem to go faster than Boeing 767s, even when they are moving at the same speed. Even experienced pilots who are aware of the actual velocities fall for this illusion. The reason, Leibowitz argued, is that there are two different subsystems that influence the ways our eyes move. One system is “reflexive”—we do it without conscious thought—and is triggered by seeing contours. This system helps us continually see things while we ourselves are moving.

We also use, more actively, “pursuit” eye movements. This is how we view moving objects when we are stationary. We can tell how fast something is moving, Leibowitz said, by how much effort it takes this “pursuit” system to see it, and by how much object there is to see. The larger the object, the less our voluntary systems have to work, and the slower the object seems.

How much slower? Judging by a test of the Leibowitz hypothesis done by researchers at the University of California at Berkeley, a lot slower. Subjects looking at a computer screen were asked to estimate the speed of a series of large and small spheres that moved toward them. Despite the presence of stationary posts and lines on the ground that subjects could use as helpful cues to judge speed, the study found that most people still thought a smaller sphere was moving faster—even when a larger sphere was moving 20 miles per hour faster. It was not until a large sphere was moving twice as fast as a smaller one that subjects were no longer convinced that the latter was moving faster.

The problem with visual illusions—and it has been argued that all human vision is an illusion—is that we fall for them even when we know they are illusions. Imagine that you are not even aware of your visual shortcomings. This is what happens when we drive at night. We think we can see better than we actually can—and we drive accordingly. We “overdrive” our headlights, moving at speeds that would not allow us to stop in time for something we saw in the range of our lights. Why do we do this? Leibowitz’s theory was that when the ambient light goes down, we lose the use of certain eye functions more than we lose others, in a process he called “selective degradation.” Our “ambient vision,” which happens mostly on the peripheral retina, helps us with things like walking down the sidewalk or staying on the road; this degrades less at night. Because of this, and because the roadside and the center lines are brightly illuminated by our headlights (studies show that we look at these lines much more at night), we essentially think we are seeing all there is to see.

But another element of our vision performs much worse at night, Leibowitz argued: the focal vision of the central retina. This is what we use to identify things, and it is the more conscious part of our vision. Most of the time, there is nothing to see on the road at night except the red taillights of cars, road signs (which we see and remember more at night), the brightly reflective pavement markings, and the section of road just in front of the car that is bathed in the full glow of our headlights.

Yet when a nonilluminated object enters the road—an animal, a stalled car, a piece of debris, or a pedestrian—we cannot see it as well as we might have thought we would based on how well we seem to be seeing everything else. We are blind to our blindness. Remember this the next time you are out walking. Studies have shown that pedestrians think drivers can see them up to twice as far away as drivers actually do. According to one expert, if we were to drive at night in a way that ensured we could see every potential hazard in time to stop—what is legally called the “assured clear distance”—we would have to drive 20 miles per hour.

Another kind of illusion bedevils us in fog. When fog rolls in on a highway, the result is often a huge, multicar chain-reaction crash. An incident that occurred in 1998 near Padua, Italy, involving more than 250 cars (and the death of four people), is an extreme example of a rather common condition. These sorts of events must be due to poor visibility, no? Obviously, it is harder to see in a fog. But the real problem may be that it is even more difficult to see than we think it is. The reason is that our perception of speed is affected by contrast. The psychologist Stuart Anstis has a clever demonstration of this; he shows that when a pair of boxes—one colored light, the other dark—are moved across a background of black-and-white stripes, the dark box seems to move faster when it crosses the white sections, while the light-colored box appears to go faster as it crosses the black sections. The higher the contrast, the faster the apparent motion, so even though the two boxes are moving at the exact same speed, they look as if they are taking alternating “steps” as they shuffle across the stripes.

In fog, the contrast of cars, not to mention the surrounding landscape, is reduced. Everything around us appears to be moving more slowly than it is, and we seem to be moving more slowly through the landscape. The idea that we are not aware of this discrepancy is suggested in studies showing that while drivers tend to slightly reduce their speed in foggy conditions, they do not do so by enough to ensure a safe margin—even when special temporary warning signs have been set up. Ironically, drivers may feel more comfortable staying closer to the vehicle ahead of them—so that they do not “lose” them in the fog—but given the perceptual confusion, this is exactly the wrong move. Similar things happen in the whiteout conditions of snow, in which it is not uncommon for drivers to crash into the back of orange-colored snowplow trucks with flashing lights. The culprit is not a slippery roadway but low contrast. Drivers may see the back of the truck “in time,” but as they think it is going faster than it actually is they may not brake accordingly.

A simple object, present on every car, is a symbol of the complex interplay of what we see and what we think we see on the road: the side rearview mirror. This itself is a curious, and rather overlooked, device. We might think of it as an essential safety feature, but it is unclear to what extent, if any, it has actually reduced the number of crashes. Moreover, studies show that many drivers do not use it during lane changes, the time when it would be most helpful, relying instead on glances over the shoulder. Then there is the issue of exactly what we are seeing when we look in that mirror. Depending on where you are in the world, either both side mirrors or just the passenger-side one will be convex, or curved outward. Because of the natural blind spots that exist beyond the edges of any car mirror, the decision was made, beginning in the 1980s, to reveal more of the scene at the expense of the driver’s ability to correctly judge distance. Better to see a car improperly than to not see it at all. This is why convex mirrors come with a familiar warning: “Objects in mirror are closer than they appear.”

But Michael Flannagan, a researcher at the University of Michigan’s Transportation Research Institute, has argued that something very strange is going on when we look in that mirror. Mirrors of any stripe tend to puzzle us. As a simple experiment, trace the outline of your head in a foggy bathroom mirror. People tend to think they are tracing the actual size, whereas actually it is half. The convex side-view mirror presents a particularly distorted and what he calls “impoverished” visual scene, with many of the typical visual cues we use to judge the world rendered more or less invisible. The only thing that reliably indicates distance, Flannagan says, is the retinal size of the image of the car we see. But the size of the car, like the entire “world” depicted, has been shrunk by the convex mirror. The curvature of the mirror means that everything is in essence being drawn closer to the viewer, which is why it is puzzling that things actually look farther away.

But it gets trickier still. Researchers can predict, by measuring the viewing angles and the geometry of the mirror, how much the mirror is distorting the image. (This distortion is greater when a driver looks over to the passenger-side mirror than when he looks at his own, closer mirror; thus, Flannagan notes, it’s a bit of a mystery why in the United States we do not allow driver’s-side convex mirrors.) In a number of studies, however, Flannagan and his colleagues have found that people’s estimates of the distance of objects is not as far off as the models predict they should be. “The vehicle behind you looks less far away than it ought to based on the smallness of the image size, as if people were somehow correcting a bit,” he says. “They’re not going on just this retinal size; they know something is making them less susceptible to the distortion on paper than they ought to be.”

These puzzles led Flannagan and his fellow researchers to a conclusion that might serve as a better warning label for side-view mirrors: “Objects in mirror are more complicated than they appear.” The same could be said of driving, as well as our ability to drive, and probably us too. It is all more complicated than it appears. We would do well to drive accordingly.