Eye Movements

From Chessprogramming wiki
Jump to: navigation, search

Home * Knowledge * Cognition * Eye Movements

Eye movements of a chess champion [1]

Eye Movements,
are voluntary or involuntary movements of the eyes helping in visual perception to fixate and track visual stimuli, to acquire information applied to memorization and learning in heuristic search and pattern recognition.

Of special interest in chess and cognition are eye movements of chess players at the chess board, where cognitive experiments and computer chess simulations were conducted by various researchers to understand how chess masters grasp the important features of a chess position, giving hints to the search concerning move selectivity and ordering. In computer chess simulations, eye movement is related to attack and move generation, either one-many of one piece from its origin square to multiple target squares, or many-one, that is one particular target square attacked/defended by multiple pieces as in Square Attacked By with bitboards .


Perceiver & MAPP

The experimental program Perceiver by Michael Barenfeld and Herbert Simon was able to duplicate the eye movements of a master by adhering to the simple relations of attack and defense [2], further extended by Kevin J. Gilmartin and Simon into a system called MAPP (Memory-aided Pattern Perceiver) which uses the learning mechanism of EPAM [3].


CHUMP, by Fernand Gobet and Peter Jansen [4], is an application of the Chunk Hierarchy and REtrieval STructures (CHREST), a pattern learning approach where an eye movement simulator is the only part of the system where the rules of the game, that is how pieces attack and move, influence the learning process.

Skill in Chess

Excerpt from Eye movements at the Chess Board in Skill in Chess, 1973 by Herbert Simon and William Chase [5]:

Serial or Parallel?

Explanations in terms of heuristic search postulate that problem solving, and cognition generally, is a serial, one-thing-at-a-time process. (We are oversimplifying matters to make the issue clear, but the oversimplification will suffice for the present.) Many psychologists have found this postulate implausible and have sought for evidence that the human organism engages in extensive parallel processing [6]. 
The intuitive feeling that much information can be "acquired at a glance" argues for a parallel processor. Of course, the correctness of the intuition depends both on the amount of information that can actually be acquired and upon what is meant by a "glance." If a glance means a single eye fixation (lasting anywhere from a fifth of a second to a half-second or longer), then we know that there are high-speed serial processes (e.g. short-term memory search, visual scanning) that operate within this time range [7]. Thus, it is certainly interesting and relevant to find out how the human eye extracts information from a complex visual display like a chess position and to see whether this extraction process is compatible with the assumptions of the heuristic search theories. 

Tichomirov and Poznyanskaya

A pair of Russian psychologists, Tichomirov and Poznyanskaya, placed an expert before a chess position with instructions to find the best move, and they observed his eye movements during the first 5 seconds of the task [8]. The eye movements were inconsistent with the hypothesis that the subject, during these 5 seconds, was searching through a tree of possible moves and their replies. 
To describe further what Tichomirov and Poznyanskaya found, we must say a word about how the eye operates. The eye has a central region of high resolution, the fovea (about 1° in radius), surrounded by a periphery of decreasingly lower resolution. Most information about visual patterns is acquired while the fovea is fixated on them; and the eye moves abruptly, in so-called saccadic movements, from one point of fixation to the next. There are at most about four or five saccadic movements per second.
In Tichomirov and Poznyanskaya's record of the first 5 seconds of their subject's eye movements, there were about 20 fixations. Most of these centered on squares of the board occupied by pieces that any chess player would consider to be of importance to the position. There were few fixations at the edges or corners of the board or on empty squares. Moreover, a large number of the saccades moved from one piece to another, where the former piece stood in a "chess" relation — that is, an attack or defense relation—to the latter. For example, the eye would move frequently from a pawn to a Knight that attacked it, or to a Knight that defended it, or from a Queen to a pawn it attacked.
It is important to note that the saccadic movements were not random — therefore, that some information must have been acquired peripherally about the target square before the saccade began. From other evidenge, we know that a strong chess player can recognize a piece within a radius of 5° to 7° from his point of fixation; for eye-movement studies show that he can frequently replace such a piece correctly on a board when he has had no closer point of fixation to it [9]. 

See also


1960 ...

1970 ...

1980 ...

1990 ...

2000 ...

External Links





feat. Avery Sharpe and Louis Hayes


  1. Hans-Werner Hunziker (2006). Im Auge des Lesers: vom Buchstabieren zur Lesefreude. (In the eye of the reader: from letter recognition to the joy of reading) Transmedia Verlag, Zurich, ISBN 978-3-7266-0068-6, Peripheral vision from Wikipedia
  2. Herbert Simon, Michael Barenfeld (1969). Information-processing analysis of perceptual processes in problem solving. Psychological Review, Vol. 76, No. 5
  3. Herbert Simon, Kevin J. Gilmartin (1973). A Simulation of Memory for Chess Positions. Cognitive Psychology, Vol. 5, pp. 29-46.
  4. Fernand Gobet, Peter Jansen (1994). Towards a Chess Program Based on a Model of Human Memory. Advances in Computer Chess 7
  5. Herbert Simon, William Chase (1973). Skill in Chess. American Scientist, Vol. 61, No. 4
  6. Ulric Neisser (1963). The Imitation of Man by Machine. Science, Vol. 139
  7. Saul Sternberg (1969). Memory-scanning: Mental Processes revealed by Reaction-time Experiments. American Scientist, Vol. 57, No. 4, pdf
  8. Oleg K. Tichomirov, E. D. Poznyanskaya (1966). An Investigation of Visual Search as a Means of Analyzing Heuristics. Soviet Psychology, Winter 1966-67 (from Voprosy Psikhologii, 1966, 2, 4)
  9. P. C. Noordzij (1966). Het registreren van oogbewegingen bij schakers. Psychology Laboratory of the University of Amsterdam
  10. Fovea centralis from Wikipedia

Up one Level