Titin.net

These two methods of perceiving information work in parallel. The slow, sequential method is the most obvious and is easy to describe. This slow method is what we use to read and understand.  The other way of perceiving information is fast. Even more, we perceive more then one thing at once (in parallel). It is unfortunate that it is believed that this faster way of perception is uncontrollable. You cannot but perceive the color of the text of this blog post. You also perceive the location of this post on the screen without exerting effort. We must design electronic medical records interfaces that allow us to tap into this fast information processing power. But, how?

I must draw a picture. You are sitting in your office and someone brings you Mr. Smith’s medical chart. It is huge. You open the chart and the ink used to write the last note is fading. Automatically, you make the assumption that Mr. Smith’s health condition is complex and that he has not been to the clinic for a while. You start skimming Mr. Smith’s chart. You occasionally stop to read a sentence or a paragraph. In going through Mr. Smith’s chart you have used the two information processing methods:

In above described events, ‘reading’ was the only act where slow information processing was used. Automatic processing was used to make the assumption that Mr. Smith health condition was complex, he had not been to the clinic for a while and probably skimming. The slow information processing (e.g. reading) is called controlled information processing.

Automatic processing does not impose cognitive load, we can process more than one thing at a time. Automatic processing is also robust under stress. We cannot use automatic processing for reading and understanding. However, we can use automatic processing for locating data and picking up cues. Using automatic processing leaves our mental capacity free to deal with the task at hand as treating Mr. Smith.

How to develop these automatic processes?

To develop automatic processing for something we need to train our brains. Although, automatic effects are appreciated after few trials, full automaticity requires hundreds of trails to develop. Therefore, training for locating data and cues needs consistency (consistent patterns and consistent cues). Yet, the main risk of automatic mapping is it being uncontrollable, once developed, we will always make the associations from the trained inputs. For automatic on screen search it took participants three times as more time to unlearn and relearn a new search pattern than time took for the initial search training. (Schneider and Chein 2003)

Nygren(1991) interviewed ICU physicians and proved that they have developed automatic processing for locating data and placing this data in preliminary context. In her experiment, she presented paper lab results to physicians after replacing all characters with the letter X, but maintaining the location. (see figure) Physicians could make conclusions as those lines are liver enzymes results. On computer screens if we present these results without some sort of consistent pattern we will not tap the benefit of automatic processing and clinicians will exert mental effort to perform previously effortless tasks. In an experiment, physicians needed more time and effort to interpret computer reports which happen to be ‘pattern-dead’. Even though these physicians were used to these computer reports! (Nyrgen, Waytt and Wright 1998) see figure.

Figure: Left: laboratory report with all letters and figures replaced with X, but relative poisions maintained. Right: “pattern-dead” computer-generated laboratory report. (from Nygren, Waytt and Wright 1998)

References:

Nygren, E., Wyatt, J.C. & Wright, P., 1998. Helping clinicians to find data and avoid delays. Lancet, 352(9138), 1462-6.

Nygren, E., 1991.  Reading documents in intensive care. I. Pattern recognition and encoding of characteristics of the information media. Report 21/91 from Center of Human Computer Studies, Uppsala university Uppsala. Quoted in

Nygren, E., Johnson, L., and Sandblad B, 1992. The art of the obvious. Automatically processed components of he task of reading frequently Implications for task analysis and interface design. Center of Human Computer Studies, Uppsala university, Sweden.

Schneider, W. & Chein, J., 2003. Controlled & automatic processing: behavior, theory, and biological mechanisms. Cognitive Science, 27(3), 559, 525