Quantifying the Limits of Human Information Processing
Early discoveries related to the information processing capacity of human's were welcomed by Applied Cognitive Psychologists, because for the first time they could provide their engineering colleagues with precise numbers. For example, the Hick-Hyman Law and Fitts' Law allowed quantitative estimates of the bandwidth of the human information processing channel (approximately 7-10 bits/sec); and G.A. Miller's famous paper provided a quantitative estimate of the capacity of working memory of from 5 to 9 chunks.
Thus, when the engineers asked how much information to put into a display - the psychologists could provide a number - probably not more than 7 or 9 chunks. However, the smart engineers (and the smart psychologists) were not very satisfied with this estimate. They realized that the numbers were meaningless unless it was possible to specify what constituted a chunk with respect to the domain being represented.
The Power of Chunking
If you read G.A. Miller's paper thoroughly, you will discover that the ultimate conclusion is that, due to the power of chunking, there seems to be no practical limit to the capacity of working memory. Miller describes how a colleague was capable of remembering long strings of binary digits, by using various strategies for recoding them into chunks.
It is a bit dramatic to watch a person get 40 binary digits in a row and then repeat them back without error. However, if you think of this merely as a mnemonic trick for extending the memory span, you will miss the more important point that is implicit in nearly all such mnemonic devices. The point is that recoding is an extremely powerful weapon for increasing the amount of information that we can deal with. In one form or another we use recoding constantly in our daily behavior. (Miller, 1956, p. 95
In this example, the stimulus (strings of binary digits) had no intrinsic structure - so the chunking strategies used were essentially mnemonic tricks (e.g., using an octal coding). That is, the chunking structure is imposed by the observer as an alternative internal representation.
Building on de Groot's observations of chess, Chase and Simon illustrated the power of chunking in relation to expertise. Their research showed that Expert Chess players had superior ability to recall positions after a very brief exposure to a chess game, than more junior players. While the recall of junior players seemed to be in the range of 7 or so pieces, experts could often reproduce the entire game. However, these differences in recall between expert and junior players were essentially eliminated when the recall task involved pieces randomly positioned on the chess board.
It appears that the 'chunking' ability that allowed the superior recall of the experts was dependent on preserving the structure of the game of chess. When the constraints of the game were eliminated, the recall advantage disappeared. This suggests that chunking structure was not an arbitrary mnemonic structure, but rather it was dependent on the intrinsic structure of chess (e.g., the rules of the game, the intentions of the players, the strength or weaknesses of positions relative to winning the game).
Chunking, Attunement, & Coordinative Structure
Perhaps, chunking facilitates memory and problem solving in a fashion analogously to how coordinative structures facilitate motor control. The superior memory capability and the ability of expert chess players to see a good option quickly suggest that they are tuned to the functional constraints of the game of chess, in the same way that a specific coordinative structure might be tuned to accomplish a specific motor function (e.g., see discussion of golf shots in previous post on requisite variety).
With regards to the discussion of requisite variety in an earlier post, the tuning to the functional constraints of the game would tend to make the signals more salient (e.g., the strengths/weaknesses of various positions) and reduce the noise (i.e., possibilities inconsistent with the constraints of the game). When the functional constraints of the game of chess are removed - the advantages of this tuning disappear.
In a similar way, Gibson's ecological optics and the related notion of optical invariance, can also be seen as an ecological basis for chunking. In other words, the optical structure provides a means for tuning into the natural constraints (or natural dimensionality) associated with control of locomotion - making the relative signals salient (allowing direct perception).
This also has clear implications for designing graphical interfaces - as emphasized in Ecological Interface Design. The key is to design representations (e.g., analogs or metaphors) that make the structure or constraints of a process salient. Thus, helping people to tune into meaningful chunks or dimensions with respect to the process control problem. The key challenge then, is to discover the natural structure intrinsic to the process being controlled (e.g., the constraints, laws or invariants). This is the ultimate goal of work analysis (e.g, Vicente (1999).
Chunking to highlight the deep structure of a problem
A key point here is to get beyond the numbers (7 + or - 2) and to get beyond the idea that chunking is a simple mnemonic trick in order to appreciate that it is possible to parse problems into functionally meaningful chunks. This is illustrated by Wertheimer's (1959) concept of productive thinking. He shows that productive thinking depends on representations that parse a problem in terms of its deep structure. The key point is that the practical power of chunking comes when an observer is tuned to and uses the natural structure of the problem (e.g., constraints, patterns, invariants, categories) in productive ways.