Recognition primed decision making and what it means for change in records

I'm a huge fan of the work of Gary Klein.

Gary is the lead in what's being referred to as the "Naturalistic decision making" movement.

The area of research that Gary has focused on over his career (at least in my reading) has been around how we make decisions.

The traditional view for a long time was the so called "rational" model of decision making - one in which we mentally come up with several ways to achieve a task, evaluate them systematically, and then choose.

In his research, Gary went looking for that in practice - initially with firefighters - and instead found something complete different.

What he found, is now called "recognition-primed decision making."

What he found, is that there's no option evaluation.

What generally happens, is that we recognise a situation based on our prior experience and choose a course of action that has worked in the past - which is mostly a completely unconscious process. There is a "sanity check" - when we select the course of action, we perform a mental simulation (based on prior experience) just to reassure ourselves that it's going to work - if the simulation fails, we choose another course of action.

If you think about the decisions you make, particularly under time pressure, you'll probably find that there's a kernel there that resonates. For me, it's the mental simulation - I can think of many times when I've chosen to do something, and then had the moment where I realise that way won't work because of xyz, and we'd better do another way.

This is (for me) the core of what makes chance so hard.

When we have to change, recognition primed decision making means we have to do three things.

The first is that we have to learn to recognise new patterns. This is first, because there are things going on around us at the moment that we can't see at all - we just don't recognise that they're there because we don't have the expertise to do so. Think of it like a chiropractor - they can look at the way you're standing and tell you 15 things about your own spine that you don't even know AND IT'S YOUR SPINE - because they have the expertise to recognise the pattern. So learning to recognise new patterns comes first.

The second thing is that we have to learn to achieve success based on the new pattern - which means performing the new pattern, many times. This is why it's important to understand the mental simulation idea. The first time we do something new, how do we simulate it? The simulation should fail, because we've never done it before and we're not really capable of simulating something new.

The third - and least trivial - is that we have to forget our old pattern. At the neural level, we don't forget, we just create a new neural pathway based on the new pattern of action, and over time it gets stronger the more we use it. If we've got a neural pathway that's got 20 years of daily experience behind it, it's probably not going away - but we can strengthen a new pattern.

The thing I take away from this, is that learning has to be a team sport - more of an apprenticeship model than a "go and be brilliant all on our own model" because having someone who knows the new patterns, can detect them in an error free way and can show them to us is going to be faster than learning to see them on our own.

Simply - the best learning is the learning we do with someone who has expertise.

A simple example is disposition in databases.

"Not possible" is what most people say - and what I said for a long time.

Except that now I know someone that's doing it.

We've talked about it.

I know how.

How did she figure it out?

She knows a lot about records - and didn't know enough about databases.

So she went and sat with someone who knows a lot about databases and asked a lot of questions.

It turned out that the pattern she wanted was pretty routine for the DBA, they'd just never thought of it in records terms before - because they don't have that expertise.

Now, her mental simulation succeeds on the "disposition in a database" pattern.

And so does mine.