The Shallows

by Nicholas Carr

  Kansas State University scholars conducted a similarly realistic study. They had a group of college students watch a typical CNN broadcast in which an anchor reported four news stories while various info-graphics flashed on the screen and a textual news crawl ran along the bottom. They had a second group watch the same programming but with the graphics and the news crawl stripped out. Subsequent tests found that the students who had watched the multimedia version remembered significantly fewer facts from the stories than those who had watched the simpler version. “It appears,” wrote the researchers, “that this multimessage format exceeded viewers’ attentional capacity.”27

  Supplying information in more than one form doesn’t always take a toll on understanding. As we all know from reading illustrated textbooks and manuals, pictures can help clarify and reinforce written explanations. Education researchers have also found that carefully designed presentations that combine audio and visual explanations or instructions can enhance students’ learning. The reason, current theories suggest, is that our brains use different channels for processing what we see and what we hear. As Sweller explains, “Auditory and visual working memory are separate, at least to some extent, and because they are separate, effective working memory may be increased by using both processors rather than one.” As a result, in some cases “the negative effects of split attention might be ameliorated by using both auditory and visual modalities”—sounds and pictures, in other words.28 The Internet, however, wasn’t built by educators to optimize learning. It presents information not in a carefully balanced way but as a concentration-fragmenting mishmash.

  The Net is, by design, an interruption system, a machine geared for dividing attention. That’s not only a result of its ability to display many different kinds of media simultaneously. It’s also a result of the ease with which it can be programmed to send and receive messages. Most e-mail applications, to take an obvious example, are set up to check automatically for new messages every five or ten minutes, and people routinely click the “check for new mail” button even more frequently than that. Studies of office workers who use computers reveal that they constantly stop what they’re doing to read and respond to incoming e-mails. It’s not unusual for them to glance at their in-box thirty or forty times an hour (though when asked how frequently they look, they’ll often give a much lower figure).29 Since each glance represents a small interruption of thought, a momentary redeployment of mental resources, the cognitive cost can be high. Psychological research long ago proved what most of us know from experience: frequent interruptions scatter our thoughts, weaken our memory, and make us tense and anxious. The more complex the train of thought we’re involved in, the greater the impairment the distractions cause.30

  Beyond the influx of personal messages—not only e-mail but also instant messages and text messages—the Web increasingly supplies us with all manner of other automated notifications. Feed readers and news aggregators let us know whenever a new story appears at a favorite publication or blog. Social networks alert us to what our friends are doing, often moment by moment. Twitter and other microblogging services tell us whenever one of the people we “follow” broadcasts a new message. We can also set up alerts to monitor shifts in the value of our investments, news reports about particular people or events, updates to the software we use, new videos uploaded to YouTube, and so forth. Depending on how many information streams we subscribe to and the frequency with which they send out updates, we may field a dozen alerts an hour, and for the most connected among us, the number can be much higher. Each of them is a distraction, another intrusion on our thoughts, another bit of information that takes up precious space in our working memory.

  Navigating the Web requires a particularly intensive form of mental multitasking. In addition to flooding our working memory with information, the juggling imposes what brain scientists call “switching costs” on our cognition. Every time we shift our attention, our brain has to reorient itself, further taxing our mental resources. As Maggie Jackson explains in Distracted, her book on multitasking, “the brain takes time to change goals, remember the rules needed for the new task, and block out cognitive interference from the previous, still-vivid activity.”31 Many studies have shown that switching between just two tasks can add substantially to our cognitive load, impeding our thinking and increasing the likelihood that we’ll overlook or misinterpret important information. In one simple experiment, a group of adults was shown a series of colored shapes and asked to make predictions based on what they saw. They had to perform the task while wearing headphones that played a series of beeps. In one trial, they were told to ignore the beeps and just concentrate on the shapes. In a second trial, using a different set of visual cues, they were told to keep track of the number of beeps. After each go-through, they completed a test that required them to interpret what they had just done. In both trials, the subjects made predictions with equal success. But after the multitasking trial, they had a much harder time drawing conclusions about their experience. Switching between the two tasks short-circuited their understanding; they got the job done, but they lost its meaning. “Our results suggest that learning facts and concepts will be worse if you learn them while you’re distracted,” said the lead researcher, UCLA psychologist Russell Poldrack.32 On the Net, where we routinely juggle not just two but several mental tasks, the switching costs are all the higher.

  It’s important to emphasize that the Net’s ability to monitor events and automatically send out messages and notifications is one of its great strengths as a communication technology. We rely on that capability to personalize the workings of the system, to program the vast database to respond to our particular needs, interests, and desires. We want to be interrupted, because each interruption brings us a valuable piece of information. To turn off these alerts is to risk feeling out of touch, or even socially isolated. The near-continuous stream of new information pumped out by the Web also plays to our natural tendency to “vastly overvalue what happens to us right now,” as Union College psychologist Christopher Chabris explains. We crave the new even when we know that “the new is more often trivial than essential.”33

  And so we ask the Internet to keep interrupting us, in ever more and different ways. We willingly accept the loss of concentration and focus, the division of our attention and the fragmentation of our thoughts, in return for the wealth of compelling or at least diverting information we receive. Tuning out is not an option many of us would consider.

  IN 1879, A French ophthalmologist named Louis Émile Javal discovered that when people read, their eyes don’t sweep across the words in a perfectly fluid way. Their visual focus advances in little jumps, called saccades, pausing briefly at different points along each line. One of Javal’s colleagues at the University of Paris soon made another discovery: that the pattern of pauses, or “eye fixations,” can vary greatly depending on what’s being read and who’s doing the reading. In the wake of these discoveries, brain researchers began to use eye-tracking experiments to learn more about how we read and how our minds work. Such studies have also proven valuable in providing further insights into the Net’s effects on attention and cognition.

  In 2006, Jakob Nielsen, a longtime consultant on the design of Web pages who has been studying online reading since the 1990s, conducted an eye-tracking study of Web users. He had 232 people wear a small camera that tracked their eye movements as they read pages of text and browsed other content. Nielsen found that hardly any of the participants read online text in a methodical, line-by-line way, as they’d typically read a page of text in a book. The vast majority skimmed the text quickly, their eyes skipping down the page in a pattern that resembled, roughly, the letter F. They’d start by glancing all the way across the first two or three lines of text. Then their eyes would drop down a bit, and they’d scan about halfway across a few more lines. Finally, they’d let their eyes cursorily drift a little farther down the left-hand side of the page. This pattern of online reading was
confirmed by a subsequent eye-tracking study carried out at the Software Usability Research Laboratory at Wichita State University.34

  “F,” wrote Nielsen, in summing up the findings for his clients, is “for fast. That’s how users read your precious content. In a few seconds, their eyes move at amazing speeds across your website’s words in a pattern that’s very different from what you learned in school.”35 As a complement to his eye-tracking study, Nielsen analyzed an extensive database on the behavior of Web users that had been compiled by a team of German researchers. They had monitored the computers of twenty-five people for an average of about a hundred days each, tracking the time the subjects spent looking at some fifty thousand Web pages. Parsing the data, Nielsen found that as the number of words on a page increases, the time a visitor spends looking at the page goes up, but only slightly. For every hundred additional words, the average viewer will spend just 4.4 more seconds perusing the page. Since even the most accomplished reader can read only about eighteen words in 4.4 seconds, Nielsen told his clients, “when you add verbiage to a page, you can assume that customers will read 18% of it.” And that, he cautioned, is almost certainly an overstatement. It’s unlikely that the people in the study were spending all their time reading; they were also probably glancing at pictures, videos, advertisements, and other types of content.36

  Nielsen’s analysis backed up the conclusions of the German researchers themselves. They had reported that most Web pages are viewed for ten seconds or less. Fewer than one in ten page views extend beyond two minutes, and a significant portion of those seem to involve “unattended browser windows…left open in the background of the desktop.” The researchers observed that “even new pages with plentiful information and many links are regularly viewed only for a brief period.” The results, they said, “confirm that browsing is a rapidly interactive activity.”37 The results also reinforce something that Nielsen wrote in 1997 after his first study of online reading. “How do users read on the web?” he asked then. His succinct answer: “They don’t.”38

  Web sites routinely collect detailed data on visitor behavior, and those statistics underscore just how quickly we leap between pages when we’re online. Over a period of two months in 2008, an Israeli company named ClickTale, which supplies software for analyzing how people use corporate Web pages, collected data on the behavior of a million visitors to sites maintained by its clients around the world. It found that in most countries people spend, on average, between nineteen and twenty-seven seconds looking at a page before moving on to the next one, including the time required for the page to load into their browser’s window. German and Canadian surfers spend about twenty seconds on each page, U.S. and U.K. surfers spend about twenty-one seconds, Indians and Australians spend about twenty-four seconds, and the French spend about twenty-five seconds.39 On the Web, there is no such thing as leisurely browsing. We want to gather as much information as quickly as our eyes and fingers can move.

  That’s true even when it comes to academic research. As part of a five-year study that ended in early 2008, a group from University College London examined computer logs documenting the behavior of visitors to two popular research sites, one operated by the British Library and one by a U.K. educational consortium. Both sites provided users with access to journal articles, e-books, and other sources of written information. The scholars found that people using the sites exhibited a distinctive “form of skimming activity” in which they’d hop quickly from one source to another, rarely returning to any source they had already visited. They’d typically read, at most, one or two pages of an article or book before “bouncing out” to another site. “It is clear that users are not reading online in the traditional sense,” the authors of the study reported; “indeed there are signs that new forms of ‘reading’ are emerging as users ‘power browse’ horizontally through titles, contents pages and abstracts going for quick wins. It almost seems that they go online to avoid reading in the traditional sense.” 40

  The shift in our approach to reading and research seems to be an inevitable consequence of our reliance on the technology of the Net, argues Merzenich, and it bespeaks a deeper change in our thinking. “There is absolutely no question that modern search engines and cross-referenced websites have powerfully enabled research and communication efficiencies,” he says. “There is also absolutely no question that our brains are engaged less directly and more shallowly in the synthesis of information when we use research strategies that are all about ‘efficiency,’ ‘secondary (and out-of-context) referencing,’ and ‘once over, lightly.’”41

  The switch from reading to power-browsing is happening very quickly. Already, reports Ziming Liu, a library science professor at San José State University, “the advent of digital media and the growing collection of digital documents have had a profound impact on reading.” In 2003, Liu surveyed 113 well-educated people—engineers, scientists, accountants, teachers, business managers, and graduate students, mainly between thirty and forty-five years old—to gauge how their reading habits had changed over the preceding ten years. Nearly eighty-five percent of the people reported that they were spending more time reading electronic documents. When asked to characterize how their reading practices have changed, eighty-one percent said that they were spending more time “browsing and scanning,” and eighty-two percent reported that they were doing more “non-linear reading.” Only twenty-seven percent said that the time they devoted to “in-depth reading” was on the rise, while forty-five percent said it was declining. Just sixteen percent said they were giving more “sustained attention” to reading; fifty percent said they were giving it less “sustained attention.”

  The findings, said Liu, indicate that “the digital environment tends to encourage people to explore many topics extensively, but at a more superficial level,” and that “hyperlinks distract people from reading and thinking deeply.” One of the participants in the study told Liu, “I find that my patience with reading long documents is decreasing. I want to skip ahead to the end of long articles.” Another said, “I skim much more [when reading] html pages than I do with printed materials.” It’s quite clear, Liu concluded, that with the flood of digital text pouring through our computers and phones, “people are spending more time on reading” than they used to. But it’s equally clear that it’s a very different kind of reading. A “screen-based reading behavior is emerging,” he wrote, which is characterized by “browsing and scanning, keyword spotting, one-time reading, [and] non-linear reading.” The time “spent on in-depth reading and concentrated reading” is, on the other hand, falling steadily.42

  There’s nothing wrong with browsing and scanning, or even power-browsing and power-scanning. We’ve always skimmed newspapers more than we’ve read them, and we routinely run our eyes over books and magazines in order to get the gist of a piece of writing and decide whether it warrants more thorough reading. The ability to skim text is every bit as important as the ability to read deeply. What is different, and troubling, is that skimming is becoming our dominant mode of reading. Once a means to an end, a way to identify information for deeper study, scanning is becoming an end in itself—our preferred way of gathering and making sense of information of all sorts. We’ve reached the point where a Rhodes Scholar like Florida State’s Joe O’Shea—a philosophy major, no less—is comfortable admitting not only that he doesn’t read books but that he doesn’t see any particular need to read them. Why bother, when you can Google the bits and pieces you need in a fraction of a second? What we’re experiencing is, in a metaphorical sense, a reversal of the early trajectory of civilization: we are evolving from being cultivators of personal knowledge to being hunters and gatherers in the electronic data forest.

  THERE ARE COMPENSATIONS. Research shows that certain cognitive skills are strengthened, sometimes substantially, by our use of computers and the Net. These tend to involve lower-level, or more primitive, mental functions such as hand-eye coordination, reflex response, and the processi
ng of visual cues. One much-cited study of video gaming, published in Nature in 2003, revealed that after just ten days of playing action games on computers, a group of young people had significantly increased the speed with which they could shift their visual focus among different images and tasks. Veteran game players were also found to be able to identify more items in their visual field than novices could. The authors of the study concluded that “although video-game playing may seem to be rather mindless, it is capable of radically altering visual attentional processing.”43

  While experimental evidence is sparse, it seems only logical that Web searching and browsing would also strengthen brain functions related to certain kinds of fast-paced problem solving, particularly those involving the recognition of patterns in a welter of data. Through the repetitive evaluation of links, headlines, text snippets, and images, we should become more adept at quickly distinguishing among competing informational cues, analyzing their salient characteristics, and judging whether they’ll have practical benefit for whatever task we’re engaged in or goal we’re pursuing. One British study of the way women search for medical information online indicated that the speed with which they were able to assess the probable value of a Web page increased as they gained familiarity with the Net.44 It took an experienced browser only a few seconds to make an accurate judgment about whether a page was likely to have trustworthy information.