On Chess…

A piece written for the May 06 issue of the “Chess Underground”.

At the chess board, I often analyze my thought process with greater intensity and devotion than the actual position. Aside from hurting my strength of play, it has so far proved to be a futile exercise. Reflective curiosity, as it relates to the neural basis of chess cognition, is not without purpose. A better understanding of how the human mind approaches chess could provide valuable insight into the most efficient route of improvement or, at the very least, it could lead to a wholesale acceptance of our deficiencies.

For centuries, the game of chess has enjoyed its reputation as the ultimate intellectual pastime. The association of chess and “intelligence” makes immediate intuitive sense, and for this exact reason, careful examination of the underlying implications are expectedly rare. The inspiration, then, for this article is simple. Does the common wisdom surrounding the relation of chess to intelligence line up with the relevant scientific literature?

Academic rigor compels us to move towards clarity, and posing the right set of questions is of key import to this aim. Are those who gravitate to the game more likely to display general forms of intellectual acumen? Is “intelligence” correlated with chess skill? If so, is the relationship one of causation? Does chess help nurture certain types of measurable intelligence? More immediate to my concerns, what are the underlying scientific themes? It would be lunacy to try to address all of these questions in the span of one essay, so we’ll take a more general approach.

The questions raised herein have been the subject of elaborate investigation by research scientists in the fields of cognitive psychology and neuroscience. In these disciplines, differences in chess skill are typically treated as the weighted construct of two factors, “intelligence” and “expertise” (5). Measures of intelligence emphasize the efficiency and speed of the mental processing apparatus, while measures of expertise underline the role of “domain-specific” knowledge. One of the fundamental studies that explored the roles of expertise and intelligence in chess was an experiment conducted by the Dutch chess master and psychologist Adriaan de Groot.

In the mid-40s, de Groot cleverly explored the ability of chess players to remember positions after brief exposure. The positions fell into one of two classes, those generated randomly and those extracted from actual tournament games. Master chess players performed supremely better than their novice counterparts when asked to reconstruct positions from real games (2). Surprisingly, this disparity in performance disappeared when randomized positions were used. These results underscore the role of domain-specific knowledge and experience as the primary indicators of skill- the general conclusion being that masters develop a very focused base of knowledge, of little statistical impact to mental tasks which fall outside their domain of expertise. More importantly, contrary to popular belief, there was no significant relationship between the depth and breadth of the search heuristic and chess skill (1).

The memory acquisition mechanism which seems to best explain expert pattern recognition is “chunking theory”. A “chunk” is a meaningful grouping of information, accessed as a single unit in long- term memory. According to this theory, stronger chess players have amassed a greater store of chess knowledge, in the form of such chunks (3, 4). Retrieval of these chunks is likely to occur when a player subconsciously notes a general resemblance in a given piece structure (8). The size and nature of these proposed chunks have also been explored. Recently published literature indicates that chess masters store chunks that contain the relative positions of up to 15 pieces (6). Furthermore, that grandmasters have a vocabulary of such patterns that contains anywhere from 10,000 to 30,000 chunks. These findings lend credence to “template theory”, which seeks to build on chunking theory and to alleviate minor technical concerns. It proposes that recursive “sub-chunks” are the building blocks of chunk hierarchy, and combine to form higher-order structures dubbed “templates” (6). This theory falls more closely in line with research that examines the nature and capacity of short-term memory during chunk acquisition.

Studies on blitz chess support the general ideas forwarded by these theories. For example, it’s been shown that 81% of variation in skill (as measured by chess rating) can be accounted for by the mental processes of chess players in the “fast mechanism” phase (7). “Fast mechanisms” involve recognition and recall, while “slow mechanisms” involve the serial search method used. Theoretically, the former provides the chunks or templates that guide your search tree. We’ve reviewed some of the major contributing factors to differences in individual chess performance. Any number of conclusions can be drawn from the research papers invoked in this brief tour, but they would be essentially incomplete. Establishing the interplay of these theories with those that address learning mechanisms would provide us with a more complete picture of the development of chess skill.

Sources:

1. Atherton, M. et al. A functional MRI study of high-level cognition. I. The game of chess. Cognitive Brain Research. published online (2002).
2. De Groot, A.D. Thought and choice in chess. The Hague, The Netherlands: Mouton. (1945, 1946).
3. Simon, H.A., & Chase, W.G. Skill at Chess. American Scientist, 61, 394-403 (1973)
4. Simon, H.A., & Chase, W.G. A simulation memory of chess positions. Cognitive Psychology, 5, 29-46 (1973)
5. Grabner RH et al. Superior performance and neural efficiency: the impact of intelligence and expertise. Brain Research Bulletin, 69, 422-439 (2006).
6. Gobet, F., & Clarkson, G. Chunks in expert memory: Evidence for the magical number four…or is it two? Memory, 12(6), 732-747 (2004).
7. Burns, Bruce D. The Effects of Speed on Skilled Chess Performance. Psychological Science, 15(7), 442-446 (2004)
8. Simon, H. A., & Gilmartin, K. J. A simulation of memory for chess positions. Cognitive Psychology, 5, 29±46 (1973).
9. Gobet, F., & Simon, H. A. Templates in chess memory: A mechanism for recalling several boards. Cognitive Psychology, 31, 1±40 (1996a).