Cognitive diversity

Cognitive diversity is the presence of different perspectives, judgments, and privately held information within a group. In The Wisdom of Crowds, it is not a decorative preference for variety but a practical condition of sound collective judgment: difference gives a group material to test, combine, and correct. When members share the same assumptions or follow one another, discussion can magnify error rather than resolve it; when independent views remain available and can be brought together, the group has a better chance of seeing what any one person misses.

Meaning and Function

The book distinguishes cognitive diversity from sociological diversity. The relevant question is not simply whether group members look different or come from different demographic backgrounds, but whether they approach a problem with divergent opinions and ways of interpreting evidence. Diversity matters because it makes discussion consequential: a group can reconsider its judgments only if it encounters ideas or information that do not merely repeat what it already believes.

That principle is connected to decentralization and local knowledge. People close to a particular problem or job may possess experience that cannot readily be summarized or transmitted elsewhere. Decentralization encourages specialization and independence, increasing the information available across a system. But it has a corresponding weakness: information discovered in one part of the system may never reach the others. A capable collective therefore needs both local autonomy and a means of combining the knowledge that autonomy produces.

The book applies this balance to markets and organizations as well as discussion groups. Google, for example, relies on the local knowledge of millions of Web-page operators, using their activity to make searches more useful. The broader organizational challenge is to make individually gathered knowledge collectively usable without erasing the local conditions that made it valuable. Cognitive diversity is not achieved by gathering people in one room alone; the group also needs ways to hear, retain, and aggregate their different judgments.

NASA and the Cost of Suppressed Dissent

The January 24 meeting of NASA’s Mission Management Team illustrates what happens when a group has too little cognitive diversity in practice. The engineers of the Debris Assessment Team had serious concerns about the foam strike and maintained that they needed images of the Orbiter’s wing to make an informed analysis. Their reservations did not receive a serious hearing. During the meeting, Linda Ham summarized the conclusions briskly, and the discussion contained no meaningful examination of minority opinions or of the uncertainty in the analysis. No member asked to see the Debris Assessment Team’s study or asked the team to quantify the uncertainties.

The failure was structural as well as interpersonal. NASA might be imagined as meritocratic and bottom-up, but the book describes its culture as deeply hierarchical. The Mission Management Team’s discussions were at once overly directed and insufficiently organized: Ham would pose a question and someone would answer, but the team did not systematically invite other members to address specific issues. The result was a narrow supply of analysis, despite the existence of relevant expertise elsewhere in the organization.

The team also began with a conclusion and interpreted new information to fit it. Research by Garold Stasser helps explain the danger. In an experiment, groups assessing psychology students relied overwhelmingly on the two pieces of information shared by all members, even though particular participants had additional relevant facts. Unshared information was often discounted as unreliable or unimportant. Free-flowing discussion could compound the problem: members tended to talk most about what everyone already knew, while novel messages could be reshaped to fit existing beliefs. The NASA example thus shows how cognitive diversity can be present in an organization’s distributed expertise yet fail to influence its decision.

Dissent as a Test of Judgment

Research by Chandra Nemeth on mock juries shows that even a minority view can improve a group’s reasoning. The dissenting position need not ultimately be correct. Its presence makes the majority interrogate its own assumptions more rigorously, producing more nuanced decisions and a more careful process. The value of dissent is therefore not limited to a lone holdout persuading everyone else to reverse course. A different opinion can improve deliberation simply by making easy agreement harder.

The book applies that finding directly to NASA. A devil’s advocate willing to argue that the foam strike might have seriously damaged the wing could have forced the Mission Management Team to examine its conclusion more critically. In the absence of such opposition, the team’s meetings may instead have worsened its judgment through group polarization. Discussion is not automatically a moderating force: under some conditions, people who talk together move toward more extreme versions of a position they already share. A group can deliberate at length and become less, not more, open to correction.

For this reason, the book treats cognitive diversity as a safeguard against consensus formed too quickly, not as a guarantee that every meeting will be wise. It is the presence of a meaningful alternative—and a process that lets it be heard—that forces the group to distinguish tested judgment from an assumption merely repeated by its members.

Corporations, Hierarchy, and Local Knowledge

The same problem appears in corporate management. The book describes large American companies with rigid hierarchies as systems that obstruct information as it travels from workers to executives and back again. In one company, a new product idea had to pass through 223 different bureaucratic paths before acceptance. At a GM plant in Van Nuys, a worker had spent two years welding a bracket without knowing that it held the car’s sunshade in place. The episode makes the cost of informational separation concrete: workers can be kept from understanding the product they help build, while managers remain distant from the work and its consequences.

Hierarchy also shaped what people were willing to report. Studies cited in the book found that young executives who were more anxious to advance communicated problem-related information less accurately, while upward mobility among middle managers was associated with withholding news about failures such as fights and budget problems. When a career depends on pleasing superiors, independent analysis and unwelcome information become risks. A corporation may have many employees and still lack cognitive diversity at the point where decisions are made if its incentives punish disagreement and its managers are insulated from experience beyond the executive suite.

The book links this insulation to weak competition and the failure to adapt. Ford’s investment of hundreds of millions of dollars in the Edsel, for which there was no consumer market, is offered as one example. In the 1970s and 1980s, few American management or product innovations were pioneered by large domestic companies. Japanese and West German manufacturers began offering better products more quickly and paying closer attention to consumers. The Pinto, introduced by Ford in 1971, became emblematic of the era; over the decade, American corporate profits, market share, and productivity growth fell sharply, and Chrysler and Lockheed required government bailouts. The book’s argument is not that diversity alone would have prevented each failure, but that insulated leadership loses access to the perspectives and information needed to recognize trouble and develop alternatives.

Building Organizations That Can Hear Difference

The proposed alternative is not simply to remove managers or transfer every decision to a committee. Decisions about local problems should, as far as possible, be made by people close to them, whose experience provides tacit knowledge that cannot be fully passed up and down a hierarchy. The book notes that fewer than 2 percent of American companies made substantial use of what were called high-performance work systems, despite evidence supporting employee independence. In one experiment, people given a button that could turn off loud, random background noise solved five times as many puzzles and made fewer proofreading errors than those without the button—even though no one pressed it. The mere knowledge that they could affect their working conditions mattered.

Decentralization can also improve coordination. When workers can make decisions about how work is done, companies need fewer orders and threats, reduce supervision and transaction costs, and free managers to attend to other tasks. Toyota’s production system illustrates the approach: frontline workers are trained in a broad range of skills and are expected to understand the production process from the bottom up. Such practices give workers a more complete view of how their decisions relate to the work around them.

Alfred Sloan’s approach at General Motors valued argument over status: ideas were to be sold, not simply imposed by order, and proposals had to withstand informed criticism. Jack Welch’s effort to make General Electric a “boundaryless corporation” likewise sought to break down boundaries between divisions, reduce layers of management, and make the relationship between bosses and subordinates less rigid through Work-Out sessions in which managers faced public criticism from employees. The book notes that Welch did not always heed opposition, particularly in some major acquisition decisions, but identifies boundarylessness as one reason GE flourished while many older American industrial corporations struggled.

Decision markets offer another way to bring independent judgments together. At InnoCentive, employees traded on profiles and experimental data for six drug candidates, three known to be likely to receive FDA approval and three likely to be rejected. The diverse employee market quickly sent the likely winners’ prices up and the likely losers’ prices down. The book argues that such markets can bypass political infighting, sycophancy, and the confusion of status with knowledge by giving people anonymity and incentives to find and act on useful information. But the markets are meant to inform major corporate decisions, not make them: their purpose is to add a collective source of judgment, not to replace all other forms of decision-making.

Diversity Beyond the Meeting Room

Cognitive diversity also explains why short selling can improve the stock market’s judgments. If investors can only bet that a stock will rise, market prices draw disproportionately on optimistic judgments. The book notes that only about 2 percent of shares on the New York Stock Exchange are shorted in a typical year. Short sellers are not necessarily more insightful, and their skepticism can be wrong. Their value lies in introducing judgments that may differ from those of investors who expect prices to rise. Like dissent in a group, opposing positions increase the chances that biases will offset rather than reinforce one another.

The same logic appears in the example of betting on an NFL point spread. If bettors could wager on only one of the two teams, their forecasts would still have to be accurate for them to make money, but they would be more likely to share the same biases. When they were wrong, they could be very wrong. Restricting short sales similarly increases the chance that prices will become substantially distorted; the book points to Internet stocks, which were difficult to short, as one setting in which prices rose to extraordinary levels. A market is not made wiser merely by having many participants. It needs room for judgments that challenge the dominant direction.

The book’s recurring distinction is between a group’s size and its intellectual range. When investors or decision-makers are too alike, information and expectations converge. During a bubble, the media may amplify that feedback loop: dissenting warnings become scarce during the boom, and during a downturn it becomes difficult to find voices cautioning against panic. The media does not create the bubble, but it can intensify the convergence that turns a market into a mob. Cognitive diversity matters, then, not because disagreement is always correct, but because a collective judgment becomes fragile when its members’ expectations and sources of information all point in the same direction.