Back in 2002, I published an article about how people learn “tacit knowledge,” using as an example how students of veterinary medicine learn a diagnostic category.

I’ve gotten a lot of mileage from that example over the years (in speeches and such), and I’d like to update the article, because there were some aspects I missed back then.

The topic

Veterinarians like to classify animals along three axes as a high-level overview. These axes are:

• Is the animal bright or dull?
• Is it alert or depressed?
• Is it responsive or unresponsive?

Veterinary students are trained to accurately answer those questions. That is, for a given cow, a veterinarian trained at (say) Illinois will give the same answers as one trained at Ft. Collins.

How is that consistency achieved? The following is the training procedure as described to me by my wife, Dawn, who did such training for many years.

(Note: the description below has minor procedural differences from how things are done nowadays, but I don’t think any of them matter for this topic.)

Background

At the University of Illinois College of Veterinary Medicine, students spend their last year “rotating” through the various clinical services the teaching hospital offers. A rotation is two weeks long. That means that, every other Monday, all the students currently “on clinics” in the Food Animal section move to some other section (like Anesthesia or Exotic Animal Medicine), and a new batch of somewhere around six to ten students arrives.

Those students are there to soak up hands-on training. The existing cases “in the ward” are divided up among the new arrivals, and each new patient that arrives is also assigned to a student.

A good deal of the hands-on training is done by two or so interns or residents. These are people who’ve completed their Doctorate in Veterinary Medicine but are there to gain more experience (both hands-on and academic). They will demonstrate procedures to the students (on actual patients) and watch as the students try their hands at medicine. (“You’re going to have to push that syringe harder; remember, you’re punching through leather.")

There are, at a given time, one or two “clinicians” responsible for all these people. They are faculty members whose job includes spending some fraction of their time on clinics. Dawn was one of those people, and spent 50% or more of her time on clinics.

Clinicians, day to day, do the same sort of demonstrating and watching as residents and interns do. The difference is that clinicians have greater knowledge, experience, and responsibility if something goes wrong.

The setting for all this training is what looks like a large warehouse room: high ceilings, cinder block walls, various bulky machinery like headgates. The picture below shows the ambiance.

Training

The ward has a day-to-day routine.

Students arrive first. They look at their cases. For each case, they write up a SOAP (Subjective/Objective observations, Analysis, and Plan). As the name implies, the SOAP consists of observations of the patient’s state, the student’s interpretation of what that means disease-wise, and the student’s plan for what to do that day.

One thing that will be noted down in the SOAP’s subjective observations is whether the patient is bright or dull.

The attending clinician (Dawn) will wander onto the ward at some later, leisurely hour (0800, for example) and lead everyone in morning rounds. That means that everyone gathers around a stall to hear the student present the case.

For some reason, students tend to err on the side of judging a dull animal to be bright. Say the first student (nowadays most likely a “she”) does that. Dawn would correct her, saying (perhaps) “No, that cow’s dull. See: she’s not cleaning her nose.“ Cows clean their noses with their tongues. It’s weirdly charming. Here’s a video.

You can imagine the students furiously memorizing a rule:

NOT cow-cleaning-nose => dull

Then the group moves to the next stall. Let’s say this cow is bright, but her student has marked her as dull in the SOAP. So:

Dawn: No, this cow is bright.
Student: But she’s not cleaning her nose! You just said…
Dawn: But her ears are perky.

You can imagine students creating a new ruleset:

NOT cow-cleaning-nose AND NOT ears-perky => dull
cow-cleaning-nose => bright
ears-perky => bright

This process continues with other animals and during other rotations. Eventually, the students get good enough at evaluating bright vs. dull. What’s interesting is that, by then, they’ve lost the rules. What was once effortful and algorithmic is now perceptual and automatic. They can see things you and I do not.

My reflections and conclusions

• I think it likely that this ability to perceive dullness is built on something innate. I bet most parents know the experience of looking at a child and thinking “She’s getting sick.” In my recollection, it’s something about the eyes. I suppose that innate ability is repurposed in veterinarians.

  However, Dawn points out that students have trouble identifying dull cows until they’ve seen enough healthy (that is, bright) cows, which argues (I think) against my claim.

• One might believe that the students have accumulated a good suite of rules and simply apply them unconsciously. That is, “losing the rules” means losing awareness of the rules, not the rules themselves. But that’s an assumption – a belief system, really – not an evidence-based conclusion. I choose to credit the self-reporting of experts.

• I don’t imagine that Dawn, when challenged about why a cow is bright, accesses a rule like perky-ears => bright. Rather, her brain is scrambling to invent an answer, latches onto something that seems salient, and presents that as the reason. But it’s an output of the decision mechanism, not a part of it.

• I further suspect that perky-ears => bright is not presented so much an explanation as a way to get the “why? why? why?” portion of the student’s brain to shut up while some other brain mechanism does the actual learning.

  That is, the lesson of “Look at her ears. They’re perky” is more about making the ears salient for perception than it is about building mental categories of ear-types.