Working memory and mental workload

Abstract

It is remarkable how little educators seem to know about how the mind works. I suppose one could advance the argument that they need not know; the inner processes of the mind have little bearing on ordinary curriculum decision-making. But I beg to differ. One parsimonious definition of learning is that it amounts to changes in long-term memory—no more and no less. If that is the case, then it would seem fairly important to find out how one goes about changing long-term memory. Before we get into the substance of this editorial, then, a brief primer on how the mind works, beginning with profound apologies to those who may well view this as simplistic. We begin with sense organs, of which the most obvious in an educational setting are eyes and ears (though touch really matters in the operating room). These sensory inputs communicate with long-term memory through ‘‘working memory’’. (It used to be called short term memory, but that is too static a term). Working memory in turn is viewed as having several components—a ‘‘phonological loop’’ to process auditory information, a ‘‘visual spatial sketchpad,’’ an ‘‘episodic buffer’’ and a ‘‘central executive’’ (Baddeley 2006). While hypothetical constructs (they’ve never been seen on MRI) they are based on decades of experimentation. It is here that conscious memory operates—from the deliberate rehearsal of a telephone number to the mental calculations of a multiplication. Although there are 4 components, there is an overall severe limit on capacity, amounting to 7 ± 2 ‘‘chunks. Working memory then communicates with long-term memory, both in retrieving information from memory and in inputting new information to memory. The operation of LTM is quite remarkable. Somehow, we are able to access exactly the information we need in very short time intervals, despite the fact that our brains are operating at the snail—like pace of a few hundred milliseconds per operation. As I wrote in an earlier editorial (Norman 2011), the Jeopardy game of 2010 in which ‘‘Watson’’, the super-duper IBM computer, beat Ken was remarkable, not because Watson won, but because Ken came so close to Watson. To beat Ken, Watson ripped through the equivalent of a million pages of text a second. Ken surely could not compete at that level. How does Ken do it? Because his memory is ‘‘associative’’—each piece of information is accessed through predetermined