Thank you, the naturalistic view of expertise Gary pioneered has vast and deep implications in many areas. I came upon shortly after Sources of Power was written and it has shaped my own consulting work radically for decades. I was vaguely aware of tools like cognitive task analysis but Gary's thinking helped me realize that seeing group decision making in terms of people's internal models of the situation was a very big deal for improving decision processes. There were popular ideas circulating like neurolinguistic programming that latched onto the cognitive modelling idea but they missed out on a key aspect that Gary had emphasized, the process of calibrating naturalistic expertise while we construct those models. I very much appreciate you helping to keep this valuable line of thinking alive and helping to introduce it to new audiences here. On the matter of biases, I do think they are ubiquitous and unavoidable and that good thinking requires us to learn compensations for them, which is one of the important things naturalistic expertise, well calibrated to situations, does for us.
Great article Jared - super interesting! Don't know if this has been said before, but this reminds me a lot of some ideas in heidegger: he essentially states that our skills result in the world 'disclosing itself' to us in certain ways, and that this disclosure then disposes us to act in certain ways. Seems like another subject that he was weirdly prescient on.
Gary cites Hubert Dreyfus' book "What computer's cant do" as one of his major inspirations, and Dreyfus himself is sometimes called Dreydegger because of his heavy emphasis on Heidegger. So you are not wrong to see the connection! But I am surprised you caught it.
I personally really liked the Dreyfus paper "Why Heideggerian AI failed and how fixing it would require making it more Heideggerian," and I suspect one of the reasons I like Gary's work is some sort of shared fundamental worldview that is somewhat Heideggerian
Not always. Experts often have lots of difficulty explaining how they do what they do because so much of their knowledge is tacit. They are pattern matching to what they have seen, felt, and heard in the past, and it can be difficult to explain how things differ from what you have sensed in the past (eg sometimes it is hard to explain why a CGI effect doesn't feel completely real). Their brain could be issuing thousands of predictions for what they should see, and what they actually see is subtly different from all of those predictions in subtle ways that is hard to explain.
In domains where everything is formalized and well ordered, like a clearly delineated physics problems, expert physicists can explain everything and use equations to solve the problem. But in naturalistic domains, some decision-making is based on gut instinct that something feels right, or feels off, and that is hard to explain.
Intuition is either merely feelings or expertise so practiced that it's second nature. But the fact that you may not actively participate in the decision when you're an expert means only that you didn't think it through explicitly, not that you can't explain why you did it after the fact if pressed.
Depends on what you mean by "if pressed." We often interview experts who tell us about incidents where they have no idea how they did what they did, or why they knew what they knew. Sometimes they go for years believing they have ESP simply because they have no other theories for how they did it. It is only by carefully going through an incident using Cognitive Task Analysis that we can help them to understand their own abilities.
If your definition of expertise requires an expert to be able to explicitly explain their decision-making process, then we are not talking about the same type of expertise.
A large part of expertise in domains like those NDM studies (military, law enforcement firefighting, aviation, athletics, chess) comes down to tacit knowledge, which by definition means not explicit. Experts who possess tacit knowledge often lack self-insight and will sometimes even invent wrong explanations to explain their decision-making process. This is why we use a systematic interviewing technique called Cognitive Task Analysis to elicit the processes underlying their expertise (as opposed to simply asking them how they did it).
I know Gary has done work with various intelligence agencies, but I am less familiar with that work and not sure where he has published about it. His paper on the Data-Frame Theory of Sensemaking gives some high level advice on intelligence analysis which might be worth checking out. I also have colleagues that are more involved in that space. I'll have to ask them what they would recommend
"Dad reflexes" might be another interesting thing to examine. I can recall one time while giving my son a bath, the tone of the water changed; I wasn't even looking at him at the moment, yet my hand immediately went to the diverter in time to smack it back down and stop the shower from spraying all over the floor.
Thank you, the naturalistic view of expertise Gary pioneered has vast and deep implications in many areas. I came upon shortly after Sources of Power was written and it has shaped my own consulting work radically for decades. I was vaguely aware of tools like cognitive task analysis but Gary's thinking helped me realize that seeing group decision making in terms of people's internal models of the situation was a very big deal for improving decision processes. There were popular ideas circulating like neurolinguistic programming that latched onto the cognitive modelling idea but they missed out on a key aspect that Gary had emphasized, the process of calibrating naturalistic expertise while we construct those models. I very much appreciate you helping to keep this valuable line of thinking alive and helping to introduce it to new audiences here. On the matter of biases, I do think they are ubiquitous and unavoidable and that good thinking requires us to learn compensations for them, which is one of the important things naturalistic expertise, well calibrated to situations, does for us.
Great article Jared - super interesting! Don't know if this has been said before, but this reminds me a lot of some ideas in heidegger: he essentially states that our skills result in the world 'disclosing itself' to us in certain ways, and that this disclosure then disposes us to act in certain ways. Seems like another subject that he was weirdly prescient on.
Gary cites Hubert Dreyfus' book "What computer's cant do" as one of his major inspirations, and Dreyfus himself is sometimes called Dreydegger because of his heavy emphasis on Heidegger. So you are not wrong to see the connection! But I am surprised you caught it.
I personally really liked the Dreyfus paper "Why Heideggerian AI failed and how fixing it would require making it more Heideggerian," and I suspect one of the reasons I like Gary's work is some sort of shared fundamental worldview that is somewhat Heideggerian
When you have your priorities straight (formalize and specifically ordered), everything else is gravy.
Expertise means having mapped the boundaries of your field, and the capacity to explain it at any level of detail necessary.
Not always. Experts often have lots of difficulty explaining how they do what they do because so much of their knowledge is tacit. They are pattern matching to what they have seen, felt, and heard in the past, and it can be difficult to explain how things differ from what you have sensed in the past (eg sometimes it is hard to explain why a CGI effect doesn't feel completely real). Their brain could be issuing thousands of predictions for what they should see, and what they actually see is subtly different from all of those predictions in subtle ways that is hard to explain.
In domains where everything is formalized and well ordered, like a clearly delineated physics problems, expert physicists can explain everything and use equations to solve the problem. But in naturalistic domains, some decision-making is based on gut instinct that something feels right, or feels off, and that is hard to explain.
Intuition is either merely feelings or expertise so practiced that it's second nature. But the fact that you may not actively participate in the decision when you're an expert means only that you didn't think it through explicitly, not that you can't explain why you did it after the fact if pressed.
Depends on what you mean by "if pressed." We often interview experts who tell us about incidents where they have no idea how they did what they did, or why they knew what they knew. Sometimes they go for years believing they have ESP simply because they have no other theories for how they did it. It is only by carefully going through an incident using Cognitive Task Analysis that we can help them to understand their own abilities.
Anything actually understood can be explained in simple physical metaphors. If you don't actually understand it you're not actually an expert.
If your definition of expertise requires an expert to be able to explicitly explain their decision-making process, then we are not talking about the same type of expertise.
A large part of expertise in domains like those NDM studies (military, law enforcement firefighting, aviation, athletics, chess) comes down to tacit knowledge, which by definition means not explicit. Experts who possess tacit knowledge often lack self-insight and will sometimes even invent wrong explanations to explain their decision-making process. This is why we use a systematic interviewing technique called Cognitive Task Analysis to elicit the processes underlying their expertise (as opposed to simply asking them how they did it).
Merely well-practiced does not equate to expertise, which is explicitly a matter of knowledge, not merely capacity.
Great piece. Didn't they apply something similar in the field of intelligence analysis? If you have any more on this topic...
I know Gary has done work with various intelligence agencies, but I am less familiar with that work and not sure where he has published about it. His paper on the Data-Frame Theory of Sensemaking gives some high level advice on intelligence analysis which might be worth checking out. I also have colleagues that are more involved in that space. I'll have to ask them what they would recommend
Data Frame Theory of Sensemaking: https://www.taylorfrancis.com/chapters/edit/10.4324/9780203810088-13/data%E2%80%93frame-theory-sensemaking-gary-klein-jennifer-phillips-erica-rall-deborah-peluso
Data-Frame Theory of Sensemaking as a Best Model for Intelligence: https://www.jstor.org/stable/26201963
A Model of Reframing for Intelligence Analysis Teams: https://www.researchgate.net/publication/384069677_A_Model_of_Reframing_for_Intelligence_Analysis_Teams
Thank you. Appreciate it.
As a surgeon i was taught "good decisions come from experience. Experience comes from bad decisions."
That sixth sense development is an iterative process.
"Dad reflexes" might be another interesting thing to examine. I can recall one time while giving my son a bath, the tone of the water changed; I wasn't even looking at him at the moment, yet my hand immediately went to the diverter in time to smack it back down and stop the shower from spraying all over the floor.
Totally. Murphy's Law become an instinct with kids