are based on predictions, Rouder says. "I had the insight that the all-or-none model makes strong predictions about the relationship between correct and error responses."
The all-or-nothing model "made precise predictions that were borne out in the group as a whole and for the bulk of subjects individually, despite some individual differences," says Cowan. "People seemed to remember objects, not partial information from the objects."
By showing that nature loves the simplicity of an all-or-nothing approach -- even with the complicated task of thinking and remembering -- Rouder and Cowan believe they've simplified the study of attention-deficit disorder and learning disabilities that compromise working memory.
The study wasenlightening on several other levels as well.
With fellow research team members Richard Morey, Christopher Zwilling, Candice Morey and Michael Pratte, Rouder experienced what he calls "the power of collaboration between different viewpoints." A skilled mathematical analyst, "but not an expert in memory," Rouder found in Cowan "a memory expert not trained in mathematical analysis."
At first, Cowan's model seemed "too simple to be true," Rouder says. "But I was wrong. Nelson Cowan's simple model does an admirable job of passing a rigorous test."
For Rouder and Cowan, the logical next step is to extend their findings into other areas of cognition. An obvious choice, they say, involves investigations into the nature of long-term memory.
If short-term memory stores information in bytes, memory of the long-term variety stores it in megabytes.In fact, says Rouder, long-term memory storage -- remembering, say, that "most police cars have red and blue lights" -- is fairly limitless."We know seemingly countless pieces of information, such as faces or images," he says.
From an evolutionary perspective, it may seem counterintuitive that less memory is actually more. But natural selection may have restricted working memory's short-term cache to increase efficiency and reduce clutter around what Rouder says is the "gateway to long-term memory."
"It is biologically expensive to use information, and the brain's solution is to use a little bit of it at a time," Cowan says. "A few theorists have suggested that groups of three to seven items are logically optimal. If much more information were in short-term memory at the same time, the pieces might interfere with one another."