[Martin Taylor 2010.03.23.10.45]
[From Bruce Gregory (2010.03.23.0955 EDT)]
[From Rick Marken (2010.03.22.1940)]
Bruce Gregory (2010.03.22.1428 EDT)
Rick Marken (2010.03.22.0920)--
Could you explain how pattern recognition prunes the (position-move, I
presume) possibilities?
BG: When you recognize a pattern you have seen before in a game, you are likely to
recall the successful (or unsuccessful) moves made in that context. This allows you
to focus your attention on those moves and ignore other possible moves.
As I recall that's basically the way Chase and Simon interpreted their
results, which is what led me to see their explanation of chess skill
as S-R. ...
I'm reading this sequence of interchanges with some bemusement, because I see the chess question from a quite different angle than either Rick or Bruce. To me it looks like a perfectly ordinary case of controlling in imagination, the same as controlling for whether to take an umbrella when one is dry inside the house but can see that it is raining outside.
In the umbrella case, one is controlling for perceiving oneself to be dry when outside. One is not outside, and there is no error in one's perception of wetness-dryness, so a simple real-time control system would not act to get an umbrella. However, one can imagine being outside, and in that imagined state, one is wet (error) without an umbrella but dry (no error) with an umbrella. So one picks up the umbrella before going out.
If you can imagine a desert dweller of a child who had no prior experience of going out into rain, would that person, seeing the rain through the window, imagine either that they would get wet when they went outside or that taking the umbrella would keep them dry? I think that person would have to have great analytic skills to be able to have those possibilities in imagination. In most of us, they come from memories of what rain does, memories of the way the world works (putting something over your head prevents the rain from hitting your head), and memories of howan umbrella functions.
Likewise in the chess game, although a fast computer might be able to analyse the possibilities from moves far into the future, yet even Deep Blue needed to be fed with a lot of "memories" of what happened from different positions in past games. Human grandmasters don't have Deep Blue's analytic capacity, but they have an advantage in being able to develop perceptual categories from positions that have something they see as being in common. Beginning players are taught about categories such as "fork", which are properties of many positions. They don't need to memorise every position that contains a fork, but learn a perceptual function that creates a perception of fork from most positions that contain one. If one is faced with a fork, that's usually a less desirable position than when one isn't. So, when controlling in imagination, seeing the positions that would result from various possible moves, one might imagine positions that led to an opposing fork to be positions with greater error in the "I want to perceive myself as winning" control system than positions not leading to an opposing fork.
Grandmaster chess players presumably have perceptual functions that produce categories far less concrete than "fork". "Controlling the centre" might be an category perception of intermediate complexity. Show the grandmaster a board position, and he might perceive the degree to which the position belongs to that category for each player, and perceive it as directly as a skilled car driver perceives a developing dangerous road position.
The memory that led to Bruce's initial set of questions is in the connections of these category perceptual functions and their weights. This is precisely the domain of reorganization, and of the "pruning" described by Bill P.
To deflect a possible critique of this view, I should point out that even though perceived categories be associated with discrete labels, nevertheless patterns in the world often have a less-than-perfect membership in the category. We have a category perception "fine day", and "nasty day", but there are days with some clouds and a sprinkle of rain that are somewhat fine and somewhat nasty (there's a whole domain called "fuzzy logic" to deal with this very normal situation). So, in the chess case, some position might be perceived as a better "control the centre" position than another would be.
I realize this doesn't address many of the issues explicitly raised by Rick and Bruce, but it's the way I see chess play, and most other behaviour that we might call "planning", such as taking the umbrella when one sees that it is raining outside. That kind of planning could, presumably, be done by logical analysis, but it is much more likely to be done by controlling in imagination using memories of related prior situations that have created categorical perceptual functions.
Martin