Completed Scrabble (left) and crossword grids (image from Toma et al 2014).
Every year, hundreds of word lovers arrive from across the US to compete in the American Crossword Puzzle tournament. They solve clues (e.g. “caught some Z’s”) and place the answers (e.g. “sleep”) in a grid. Meanwhile, a separate group of wordsmiths gather regularly to compete at Scrabble, the game that involves forming words out of letter tiles and finding a suitable place for them on the board.
Both sets of players have exceptional abilities, but how exactly do they differ from each other and from non-players of matched academic ability? Some answers are provided by Michael Toma and his colleagues, who have performed the first detailed comparison of the mental skills of the most elite crossword and Scrabble players in the US. Previous studies on gaming expertise have mostly involved chess players, so this is a refreshing new research angle.
Toma’s team recruited 26 elite Scrabble players (in the top two per cent of competitive players, on average; 20 men) and 31 elite crossword players (in the top 7 per cent, on average; 22 men) to complete several tests of working memory – the kind of memory that we use to juggle and use information over short time-scales.
For example, there was a visuospatial task that involved judging whether images were symmetrical, while also remembering highlighted locations in a series of grids that always appeared after each symmetry image. Another challenge was the reading span task (a test of verbal working memory), which involved judging the grammatical sense of sentences, while also remembering the order of individual letters that were flashed on-screen after each grammatical challenge.
The researchers anticipated that the Scrabble players would outperform the crossworders on visuospatial working memory, whereas they thought the crossword players might be superior on verbal working memory. These predictions were based on the contrasting skills demanded by the two games. Scrabble players often spend hours learning lists of words that are legal in the game, but unlike crossword players, they don’t need to know their meaning. In fact many Scrabble players admit to not knowing the meaning of many of the words they play. On the other hand, Scrabble players need skills to rearrange letters and to find a place for their words on the board (a visuospatial skill), whereas crossword players do not need these skills so much because the grid is prearranged for them.
The researchers actually uncovered no group differences on any of the measures of visuospatial and verbal working memory. However, in line with predictions, the crossword competitors outperformed the Scrabble players on an analogies-based word task – identifying a pair of words that have the same relation to each other as a target pair – and the crossworders also had higher (self-reported) verbal SAT scores than the Scrabble players (SAT is a standardised school test used in the US). The two groups also differed drastically in the most important strategies they said they used during game play – for instance, mental flexibility was far more important for crossworders, whereas anagramming was important for Scrabble players but not mentioned by crossworders.
Both expert groups far outperformed a control group of high-achieving students on all measures of verbal and visuospatial working memory. This was despite the fact the students had similar verbal SAT levels to the expert players. So it seems the elite players of both games have highly superior working memory relative to controls, but this enhancement is not tailored to their different games.
Toma and his team said that by looking beyond chess and studying experts in cognitively demanding verbal games, their research “helps to build a more general understanding of the cognitive mechanisms that underlie elite performance.” From a theoretical perspective, their finding of no working memory differences between Scrabble and crossword competitors is supportive of a domain general account of working memory – the idea that there exists a single mechanism that supports the processing of verbal and visuospatial information.