In case you haven't yet encountered it, Sudoku is the latest rage among middle-school nerds and others looking for a "different" sort of mental challenge. It is a mathematical puzzle that involves filling in a grid of 81 squares with varying combinations of the numbers one to nine in which the same numeral cannot appear more than once in any given row across or down, or in any group of nine adjacent squares.

Sounds simple? Well, while some mathematicians over the centuries have suggested that God can be found in numbers, but you'll have a Devil of a time solving a Sudoku puzzle - and with a good reason. Doing Sudoku involves neural pathways that even the most powerful computers can't replicate. Computer scientists now say that studying the way in which humans solve Sudoku puzzles may lead to the development of more intelligent computers.

According to a paper published by John Hopfield of Princeton University, humans brains use a unique set of neural pathways known as "associative memory." This is the process by which we discover patters by studying a partial clue, or as Sherlock Holmes might say, "from one, deduce the whole." Computers can store and process huge amounts of data at blistering speeds, but associative memory is beyond their current capabilities.

In his paper, Hopfield provides an algorithm of associative memory. He believes this algorithm could be implemented in silicon chips. "[Associative memory] may account for our strong psychological feeling of 'right' or 'wrong' when we retrieve a memory from a minimal clue," says Hopfield. "This fact may account for our strong psychological feeling of 'right' or 'wrong' when we retrieve a memory from a minimal clue."

An Australian colleague, computer scientist Andrew Paplinski believes that Hopfield's model could lead to more accurate facial recognition computer technology.