It is difficult to think of human intelligence as simply one mechanism that that can accurately distinguish about seven inputs. It is also very difficult to think that the only way one can get better is by nesting these mechanisms. I’m not saying that Miller’s magic number has no ground, but I do wonder if he simply discovered a mechanism humans use when they first come in contact with a new problem. When the novices and experts were mentioned I started to think about what makes a computer program more efficient when compared to another. Given a task, one computer program is usually deemed superior if the time for it to execute is faster, if it can keep this speed with a bigger data set, and if it consistently produces accurate results. I would argue that the same criteria can be applied to determine the difference between an expert and a novice. Experts solve problems faster, usually can handle greater data sets, and are rather consistent in terms of their out put (they don’t make many mistakes). In computer programs you generally see a performance boost when a faster algorithm is applied. A substantially modified bubble sort (the worst kind of sorting algorithm save randomly sorting) does not perform much better than a regular bubble sort. Similarly I would like to think that experts develop a structure that is more sophisticated than the 7 input algorithm that Miller proposes we use.
I like the idea of using heuristics to find an optimal solution to a problem. When you have many ways to tackle a problem finding the method that is more useful will definitely increase your performance on a task. In terms of intelligence, this approach also fits with the idea that intelligence is partially defined by how well we can adapt. The problem I see is the generation of these heuristics. Where do they come from? How do we create them? If intelligence is how well we adapt to a new obstacle then surely the generation of these heuristics must be an important component. Even with Miller’s idea of chunking we have to wonder how we create the definition of a chunk. Yes chunking allows us to process more information faster, but at some point biologically, or by some other means, we must identify patterns that we can then use as chunking criteria. For some reason I don’t see the addition of dimensions to be a sufficient answer.