Top Down and Bottom Up

In the early days of developing expert systems, there was a lot of talk about whether such systems should be built from the top down or the bottom up. The artifical intelligence focus has moved on to other perspectives, including the ever popular neural networks. In these discussions, we seem to forget an underlying principle that still has an impact on the difference between a novice and an expert.

Herbert Simon and Hubert Dreyfus represented the opposite ends of the pole in those early top-down bottom-up discussions. There were those, Simon included, who took the top-down approach. Sit a good stenographer in a room with an expert, ask questions, take copious notes, and feed the answers into computers to create expert systems. Hubert and Stuart Dreyfus, however, noticed something about expert behavior others hadn’t. The Dreyfus brothers noted that the mark of the novice was strict adherence to rules. The mark of an expert, on the other hand, was the ability to rise above the rules:

The capacity of experts to store in memory tens of thousands of typical situations and rapidly and effortlessly to see the present situation as similar to one of these, apparently without resorting to time-consuming feature detection and matching, suggests that the brain does not work like a heuristically programmed digital computer applying rules to bits of information (1).

Novices generally use rules very consciously. Experts, on the other hand, often use rules without even being aware that they are doing so. More importantly, experts make decisions by comparing past experiences, and these experiences are not reducible to rules. In fact, expert behavior is characterized by the ability to sort through and use a vast number of unique experiences which bear no names and defy complete verbal description.

A tragic consequence of all the top-down rulemaking is the complete lack of understanding of what constitutes expert behavior, and nowhere is that more evident than in the attempts to constrain what physicians do, how they practice, and how they are paid based upon rules.

Practicing physicians know that some characteristics of sick patients can’t be measured. We handle individual cases with specific nuances. Yet our practices are evaluated by algorithm-wielding nurses on case management phones and by…standards that attempt to generalize and quantify medicine. (2)

No one disagrees that there should be guidelines for patient care, but the deterioration of patient care begins when those guidelines are turned into rules which are misrepresented as real measures of physician expertise or the quality of the care they give. Quite simply, there is far more art to expert medicine than craft.  Are that comes with experience, not book learning, and most certainly not from cookbook medicine.

 Bottom-down Rules can be Deadly

One thing the federal government has never done well in spinning it’s regulations is to keep the big picture in mind as it gins out regulations that are difficult and at times downright dangerous to follow.  One case in point is the now popular regulation against side rails in nursing homes, and even in acute care beds. No one is pretending that restraining nursing home patients has at times in the past been abused. But overreacting to past errors by demanding that nursing home patients be restrained in no way, even by the presence of a side rail on a bed, simply indicates that the rule makers have likely spent little time caring for patients in a nursing home.

I’m sure there have been cases of patients getting their heads stuck in side rails, but 50 years ago when babies started getting their heads stuck in crib rails, crib manufacturers took it upon themselves to make the distance between the rails smaller.  With nursing homes, however, our government regulators did not require that side rails be redesigned to prevent the possibility of a patient’s head getting stuck.  Rather, the regulators simply banned the use of side rails from some misguided notion that all semblances of restraint are bad all of the time.

Anyone caring for the elderly knows that in nursing homes, a prime concern is to prevent patient falls.  Side rails helped prevent falls.  Now that nursing homes are no longer able to use side rails on patient beds, the number of patients falling and incurring head injuries or broken hips have increased tremendously.  But no one hears about this. As far as I know, no one is counting and making any correlation in the falls in nursing homes and the lack of bed rails. But physicians in the trenches can see it as plain as day.

Oh, the euphemisms abound. Quality improvement initiatives. Evidence-based medicine.  But make no mistake about it. The shifts coming in health care are still dangerously bound to rules, rules which plainly and simply have little to do with the most valuable care experienced physicians provide their patients—an intuition based upon that experience which defies rules.

References (obviously oldies but goodies):

1.  Dreyfus, H.L. and S. E. Dreyfus, “Why Skills Cannot be Represented by Rules,” In Advances in Cognitive Psychology, ed. N.E. Sharkey. Chichester, England: Elllis Horwood, 1986, p. 327.

2.  Charles Meyer, “A Crisis in Spirit, Minnesota Medicine 2, 8 (August 1989, p. 451).


Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Show Buttons
Hide Buttons