The Shallows

by Nicholas Carr

  But the extensive activity in the brains of surfers also points to why deep reading and other acts of sustained concentration become so difficult online. The need to evaluate links and make related navigational choices, while also processing a multiplicity of fleeting sensory stimuli, requires constant mental coordination and decision making, distracting the brain from the work of interpreting text or other information. Whenever we, as readers, come upon a link, we have to pause, for at least a split second, to allow our prefrontal cortex to evaluate whether or not we should click on it. The redirection of our mental resources, from reading words to making judgments, may be imperceptible to us—our brains are quick—but it’s been shown to impede comprehension and retention, particularly when it’s repeated frequently. As the executive functions of the prefrontal cortex kick in, our brains become not only exercised but overtaxed. In a very real way, the Web returns us to the time of scriptura continua, when reading was a cognitively strenuous act. In reading online, Maryanne Wolf says, we sacrifice the facility that makes deep reading possible. We revert to being “mere decoders of information.”10 Our ability to make the rich mental connections that form when we read deeply and without distraction remains largely disengaged.

  Steven Johnson, in his 2005 book Everything Bad Is Good for You, contrasted the widespread, teeming neural activity seen in the brains of computer users with the much more muted activity evident in the brains of book readers. The comparison led him to suggest that computer use provides more intense mental stimulation than does book reading. The neural evidence could even, he wrote, lead a person to conclude that “reading books chronically understimulates the senses.”11 But while Johnson’s diagnosis is correct, his interpretation of the differing patterns of brain activity is misleading. It is the very fact that book reading “understimulates the senses” that makes the activity so intellectually rewarding. By allowing us to filter out distractions, to quiet the problem-solving functions of the frontal lobes, deep reading becomes a form of deep thinking. The mind of the experienced book reader is a calm mind, not a buzzing one. When it comes to the firing of our neurons, it’s a mistake to assume that more is better.

  John Sweller, an Australian educational psychologist, has spent three decades studying how our minds process information and, in particular, how we learn. His work illuminates how the Net and other media influence the style and the depth of our thinking. Our brains, he explains, incorporate two very different kinds of memory: short-term and long-term. We hold our immediate impressions, sensations, and thoughts as short-term memories, which tend to last only a matter of seconds. All the things we’ve learned about the world, whether consciously or unconsciously, are stored as long-term memories, which can remain in our brains for a few days, a few years, or even a lifetime. One particular type of short-term memory, called working memory, plays an instrumental role in the transfer of information into long-term memory and hence in the creation of our personal store of knowledge. Working memory forms, in a very real sense, the contents of our consciousness at any given moment. “We are conscious of what is in working memory and not conscious of anything else,” says Sweller.12

  If working memory is the mind’s scratch pad, then long-term memory is its filing system. The contents of our long-term memory lie mainly outside of our consciousness. In order for us to think about something we’ve previously learned or experienced, our brain has to transfer the memory from long-term memory back into working memory. “We are only aware that something was stored in long-term memory when it is brought down into working memory,” explains Sweller.13 It was once assumed that long-term memory served merely as a big warehouse of facts, impressions, and events, that it “played little part in complex cognitive processes such as thinking and problem-solving.”14 But brain scientists have come to realize that long-term memory is actually the seat of understanding. It stores not just facts but complex concepts, or “schemas.” By organizing scattered bits of information into patterns of knowledge, schemas give depth and richness to our thinking. “Our intellectual prowess is derived largely from the schemas we have acquired over long periods of time,” says Sweller. “We are able to understand concepts in our areas of expertise because we have schemas associated with those concepts.”15

  The depth of our intelligence hinges on our ability to transfer information from working memory to long-term memory and weave it into conceptual schemas. But the passage from working memory to long-term memory also forms the major bottleneck in our brain. Unlike long-term memory, which has a vast capacity, working memory is able to hold only a very small amount of information. In a renowned 1956 paper, “The Magical Number Seven, Plus or Minus Two,” Princeton psychologist George Miller observed that working memory could typically hold just seven pieces, or “elements,” of information. Even that is now considered an overstatement. According to Sweller, current evidence suggests that “we can process no more than about two to four elements at any given time with the actual number probably being at the lower [rather] than the higher end of this scale.” Those elements that we are able to hold in working memory will, moreover, quickly vanish “unless we are able to refresh them by rehearsal.”16

  Imagine filling a bathtub with a thimble; that’s the challenge involved in transferring information from working memory into long-term memory. By regulating the velocity and intensity of information flow, media exert a strong influence on this process. When we read a book, the information faucet provides a steady drip, which we can control by the pace of our reading. Through our single-minded concentration on the text, we can transfer all or most of the information, thimbleful by thimbleful, into long-term memory and forge the rich associations essential to the creation of schemas. With the Net, we face many information faucets, all going full blast. Our little thimble overflows as we rush from one faucet to the next. We’re able to transfer only a small portion of the information to long-term memory, and what we do transfer is a jumble of drops from different faucets, not a continuous, coherent stream from one source.

  The information flowing into our working memory at any given moment is called our “cognitive load.” When the load exceeds our mind’s ability to store and process the information—when the water overflows the thimble—we’re unable to retain the information or to draw connections with the information already stored in our long-term memory. We can’t translate the new information into schemas. Our ability to learn suffers, and our understanding remains shallow. Because our ability to maintain our attention also depends on our working memory—“we have to remember what it is we are to concentrate on,” as Torkel Klingberg says—a high cognitive load amplifies the distractedness we experience. When our brain is overtaxed, we find “distractions more distracting.”17 (Some studies link attention deficit disorder, or ADD, to the overloading of working memory.) Experiments indicate that as we reach the limits of our working memory, it becomes harder to distinguish relevant information from irrelevant information, signal from noise. We become mindless consumers of data.

  Difficulties in developing an understanding of a subject or a concept appear to be “heavily determined by working memory load,” writes Sweller, and the more complex the material we’re trying to learn, the greater the penalty exacted by an overloaded mind.18 There are many possible sources of cognitive overload, but two of the most important, according to Sweller, are “extraneous problem-solving” and “divided attention.” Those also happen to be two of the central features of the Net as an informational medium. Using the Net may, as Gary Small suggests, exercise the brain the way solving crossword puzzles does. But such intensive exercise, when it becomes our primary mode of thought, can impede deep learning and thinking. Try reading a book while doing a crossword puzzle; that’s the intellectual environment of the Internet.

  BACK IN THE 1980s, when schools began investing heavily in computers, there was much enthusiasm about the apparent advantages of digital documents over paper ones. Many educators were convinced that introducing hyper
links into text displayed on computer screens would be a boon to learning. Hypertext would, they argued, strengthen students’ critical thinking by enabling them to switch easily between different viewpoints. Freed from the lockstep reading demanded by printed pages, readers would make all sorts of new intellectual connections among diverse texts. The academic enthusiasm for hypertext was further kindled by the belief, in line with the fashionable postmodern theories of the day, that hypertext would overthrow the patriarchal authority of the author and shift power to the reader. It would be a technology of liberation. Hypertext, wrote the literary theorists George Landow and Paul Delany, can “provide a revelation” by freeing readers from the “stubborn materiality” of printed text. By “moving away from the constrictions of page-bound technology,” it “provides a better model for the mind’s ability to reorder the elements of experience by changing the links of association or determination between them.”19

  By the end of the decade, the enthusiasm had begun to subside. Research was painting a fuller, and very different, picture of the cognitive effects of hypertext. Evaluating links and navigating a path through them, it turned out, involves mentally demanding problem-solving tasks that are extraneous to the act of reading itself. Deciphering hypertext substantially increases readers’ cognitive load and hence weakens their ability to comprehend and retain what they’re reading. A 1989 study showed that readers of hypertext often ended up clicking distractedly “through pages instead of reading them carefully.” A 1990 experiment revealed that hypertext readers often “could not remember what they had and had not read.” In another study that same year, researchers had two groups of people answer a series of questions by searching through a set of documents. One group searched through electronic hypertext documents, while the other searched through traditional paper documents. The group that used the paper documents outperformed the hypertext group in completing the assignment. In reviewing the results of these and other experiments, the editors of a 1996 book on hypertext and cognition wrote that, since hypertext “imposes a higher cognitive load on the reader,” it’s no surprise “that empirical comparisons between paper presentation (a familiar situation) and hypertext (a new, cognitively demanding situation) do not always favor hypertext.” But they predicted that, as readers gained greater “hypertext literacy,” the cognition problems would likely diminish.20

  That hasn’t happened. Even though the World Wide Web has made hypertext commonplace, indeed ubiquitous, research continues to show that people who read linear text comprehend more, remember more, and learn more than those who read text peppered with links. In a 2001 study, two Canadian scholars asked seventy people to read “The Demon Lover,” a short story by the modernist writer Elizabeth Bowen. One group read the story in a traditional linear-text format; a second group read a version with links, as you’d find on a Web page. The hypertext readers took longer to read the story, yet in subsequent interviews they also reported more confusion and uncertainty about what they had read. Three-quarters of them said that they had difficulty following the text, while only one in ten of the linear-text readers reported such problems. One hypertext reader complained, “The story was very jumpy. I don’t know if that was caused by the hypertext, but I made choices and all of a sudden it wasn’t flowing properly, it just kind of jumped to a new idea I didn’t really follow.”

  A second test by the same researchers, using a shorter and more simply written story, Sean O’Faolain’s “The Trout,” produced the same results. Hypertext readers again reported greater confusion following the text, and their comments about the story’s plot and imagery were less detailed and less precise than those of the linear-text readers. With hypertext, the researchers concluded, “the absorbed and personal mode of reading seems to be discouraged.” The readers’ attention “was directed toward the machinery of the hypertext and its functions rather than to the experience offered by the story.”21 The medium used to present the words obscured the meaning of the words.

  In another experiment, researchers had people sit at computers and review two online articles describing opposing theories of learning. One article laid out an argument that “knowledge is objective” the other made the case that “knowledge is relative.” Each article was set up in the same way, with similar headings, and each had links to the other article, allowing a reader to jump quickly between the two to compare the theories. The researchers hypothesized that people who used the links would gain a richer understanding of the two theories and their differences than would people who read the pages sequentially, completing one before going on to the other. They were wrong. The test subjects who read the pages linearly actually scored considerably higher on a subsequent comprehension test than those who clicked back and forth between the pages. The links got in the way of learning, the researchers concluded.22

  Another researcher, Erping Zhu, conducted a different kind of experiment that was also aimed at discerning the influence of hypertext on comprehension. She had groups of people read the same piece of online writing, but she varied the number of links included in the passage. She then tested the readers’ comprehension by asking them to write a summary of what they had read and complete a multiple-choice test. She found that comprehension declined as the number of links increased. Readers were forced to devote more and more of their attention and brain power to evaluating the links and deciding whether to click on them. That left less attention and fewer cognitive resources to devote to understanding what they were reading. The experiment suggested a strong correlation “between the number of links and disorientation or cognitive overload,” wrote Zhu. “Reading and comprehension require establishing relationships between concepts, drawing inferences, activating prior knowledge, and synthesizing main ideas. Disorientation or cognitive overload may thus interfere with cognitive activities of reading and comprehension.”23

  In 2005, Diana DeStefano and Jo-Anne LeFevre, psychologists with the Centre for Applied Cognitive Research at Canada’s Carleton University, undertook a comprehensive review of thirty-eight past experiments involving the reading of hypertext. Although not all the studies showed that hypertext diminished comprehension, they found “very little support” for the once-popular theory “that hypertext will lead to an enriched experience of the text.” To the contrary, the preponderance of evidence indicated that “the increased demands of decision-making and visual processing in hypertext impaired reading performance,” particularly when compared to “traditional linear presentation.” They concluded that “many features of hypertext resulted in increased cognitive load and thus may have required working memory capacity that exceeded readers’ capabilities.”24

  THE WEB COMBINES the technology of hypertext with the technology of multimedia to deliver what’s called “hypermedia.” It’s not just words that are served up and electronically linked, but also images, sounds, and moving pictures. Just as the pioneers of hypertext once believed that links would provide a richer learning experience for readers, many educators also assumed that multimedia, or “rich media,” as it’s sometimes called, would deepen comprehension and strengthen learning. The more inputs, the better. But this assumption, long accepted without much evidence, has also been contradicted by research. The division of attention demanded by multimedia further strains our cognitive abilities, diminishing our learning and weakening our understanding. When it comes to supplying the mind with the stuff of thought, more can be less.

  In a study published in the journal Media Psychology in 2007, researchers recruited more than a hundred volunteers to watch a presentation about the country of Mali played through a Web browser on a computer. Some of the subjects watched a version of the presentation that included only a series of text pages. Another group watched a version that included, along with the pages of text, a window in which an audiovisual presentation of related material was streamed. The test subjects were able to stop and start the stream as they wished.

  After viewing the presentation, the subjects took
a ten-question quiz on the material. The text-only viewers answered an average of 7.04 of the questions correctly, while the multimedia viewers answered just 5.98 correctly—a significant difference, according to the researchers. The subjects were also asked a series of questions about their perceptions of the presentation. The text-only readers found it to be more interesting, more educational, more understandable, and more enjoyable than did the multimedia viewers, and the multimedia viewers were much more likely to agree with the statement “I did not learn anything from this presentation” than were the text-only readers. The multimedia technologies so common to the Web, the researchers concluded, “would seem to limit, rather than enhance, information acquisition.”25

  In another experiment, a pair of Cornell researchers divided a class of students into two groups. One group was allowed to surf the Web while listening to a lecture. A log of their activity showed that they looked at sites related to the lecture’s content but also visited unrelated sites, checked their e-mail, went shopping, watched videos, and did all the other things that people do online. The second group heard the identical lecture but had to keep their laptops shut. Immediately afterward, both groups took a test measuring how well they could recall the information from the lecture. The surfers, the researchers report, “performed significantly poorer on immediate measures of memory for the to-be-learned content.” It didn’t matter, moreover, whether they surfed information related to the lecture or completely unrelated content—they all performed poorly. When the researchers repeated the experiment with another class, the results were the same.26