Skip Navigation

Science secret of grand masters revealed

August 6, 2004 By Mark Peplow This article courtesy of Nature News.

Chess experts gain the edge over opponents by falsifying their own ideas.

For all you budding Kasparovs out there, a team of cognitive scientists has worked out how to think like a chess grand master. As those attending this week's Cognitive Science Society meeting in Chicago, Illinois, were told, the secret is to try to knock down your pet theory rather than finding ways to support it - exactly as scientists are supposed to do.

"This is a new result in the psychology of chess, as far as I know," says Mark Orr, a chess enthusiast and Ireland's first international master. The research could help developing chess players to hone their skills, he adds.

In deciding which move to make, chess players mentally map out the future consequences of each possible move, often looking about eight moves ahead. So Michelle Cowley, a cognitive scientist and keen chess player from Trinity College Dublin in Ireland, decided to study how different chess players decide whether their move strategies will be winners or losers.

Along with her colleague Ruth Byrne, she recruited 20 chess players, ranging from regular tournament players to a grand master. She presented each participant with six different chessboard positions from halfway through a game, where black and white had equal chances of winning and there was no immediately obvious next move.

Each player had to speak their thoughts aloud as they decided what move to make. Cowley scored the quality of the move sequences by comparing them with Fritz 8, one of the most powerful chess computer programs available.

She found that novices were more likely to convince themselves that bad moves would work out in their favour, because they focused more on the countermoves that would benefit their strategy while ignoring those that led to the downfall of their cherished hypotheses.

Conversely, masters tended to correctly predict when the eventual outcome of a move would weaken their position. "Grand masters think about what their opponents will do much more," says Byrne. "They tend to falsify their own hypotheses."

"We probably all intuitively know this is true," says Orr. "But it's never a bad thing to prove it like this."

Strategic thinking

The philosopher Karl Popper called this process of hypothesis testing 'falsification', and thought that it was the best way to describe how science constantly questions and refines itself. It is often held up as the principle that separates scientific and non-scientific thinking, and the best way to test a hypothesis.

But cognitive research has shown that, in reality, many people find falsification difficult. Until the latest study, scientists were the only group of experts that had been shown to use falsification. And sociological studies of scientists in action have revealed that even they spend a great deal of their time searching for results that would bolster their theories1. Some philosophers of science have suggested that since there is so much rivalry within science, individuals often rely on their peers to falsify their theories for them.

Byrne speculates that the behaviour may actually be widespread, but that it could be limited to those who are expert in their field. She thinks the ability to falsify is somehow linked to the vast database of knowledge that experts such as grand masters - or scientists - accumulate. "People who know their area are more likely to look for ways that things can go wrong for them," she says.

Byrne and Cowley now hope to study developing chess players to find out how and when they develop falsification strategies. They also want to test chess masters in other activities that involve testing hypotheses - such as logic problems - to discover if their falsification skill is transferable. On this point Orr is more sceptical: "I've never felt that chess skills cross over like that, it's a very specific skill."


  1. Latour B., . Laboratory Life: The Construction of Scientific Facts, (2001).


Need Assistance?

If you need help or have a question please use the links below to help resolve your problem.