[00:00:00] Christine Ko: Today is part 2 with Dr. Bruce Smoller! I did want to touch on his recent lecture at the ISDP and how it is relevant to diagnosis and, really, daily life. For anyone who wasn't able to listen to part 1, I will summarize part of his biography here. Dr. Bruce Smoller trained in Anatomic and Clinical Pathology at Harvard University, and he has made several moves from Stanford University, to the University of Arkansas, to becoming Executive Vice President of the United States and Canadian Academy of Pathology for a couple of years, all the way to University of Rochester, where he is currently now. Dr. Smoller is a prolific dermatopathologist and pathologist, former Editor-in-Chief of the Journal of Cutaneous Pathology, and he has served as the President of the American Society of Dermatopathology. He has received several different teaching awards, including the Nickel Award from the American Society of Dermatopathology, as well as a Lifetime Achievement Award from the College of American Pathologists. He has written many articles, over 300, and has been primarily involved in many books, just to name some of his academic accomplishments.
[00:01:17] At the March of 2024 International Society of Dermatopathology meeting, you were talking about visual recognition and diagnosis. Can you talk about visual diagnosis and how you avoid error or misperceiving something?
[00:01:33] Bruce Smoller: Sure. So this actually came up after my talk. Boris Bastian asked what I thought was a very interesting question. Bastian's question was, you seem to be saying that we all make our diagnoses based on blink, based on pattern recognition, instantaneous. How do you prevent making mistakes?
[00:01:52] So for those who didn't have the opportunity, or fortunately did not have the opportunity, to hear my lecture; in a nutshell, what I talked about was we all learn dermpath with these algorithms that can be quite elaborate. We have algorithms for all kinds of things.
[00:02:10] But early on, probably the greatest lecture I have ever been to in my entire life happened the fall semester of my junior year of college. A professor named Jernstedt, who was the learning theory expert where I went to college, came in one morning, this was a psychology class, and he said to all of us, at the end of this hour you're all going to be experts in Cubist art.
[00:02:34] Now I have to say, Philistine that I am, I didn't even know what Cubist art was, so it seemed hard to believe that in the end of an hour I was going to be an expert on that. Pretty much that's all he said. We had some sort of a primitive audience response system, and he told us to use it. And then he flashed up a picture, and it was some painting, and then it was followed by, okay, which artist painted this picture? And we had four choices. They were Leger, Braque Picasso and Gris; four artists. I had heard of Picasso. I had never heard of the other three, but we picked one, and then we immediately were told what the right answer was.
[00:03:14] He didn't say anything for the entire hour. At the beginning of the hour, the success rate was about 25 percent because I wasn't alone. None of us had heard of the other three. By the end of the hour, the success rate was over 90%. Now, clearly there were no algorithms because he didn't say a word. He just sat there and we all just pressed buttons, but somehow we all learned how to recognize a pattern.
[00:03:37] For instance, to me, Leger always has these cylindrical structures in his paintings, which stood out to me. Was there any algorithm that said if you have cylindrical things, think Leger? Nobody said that. I just figured that out on my own. For all I know, it's wrong, but I figured that out on my own, where to this day, when I walk into a, an art museum that has modern art, I point and say, Oh, that's Leger. Oh, that's Gris. And I'm all excited about the fact that's the only lecture I've ever had in my entire life about Cubist art. And it stuck with me. So it made me think, what are we doing with all our algorithms and how do we learn? Okay, that's how we teach, but how do the students learn? And I think what the students learn by or learn from is rehearsing.
[00:04:25] Over and over so that when they see a basal cell, after the 200th basal cell, you really don't need to talk about peripheral palisading and clefting and the N:C ratio and all. It's a basal cell. We all know it's a basal cell.
[00:04:39] Forget the algorithms. What are the algorithms doing? There were four artists. We could all get a handle on four artists and pick one out of four. But what are there, a thousand diagnoses in a dermpath textbook? How do you get your mind wrapped around a thousand entities? You can't. But if you break them down into algorithms, you narrow it down; you were down to four or five or six different entities as opposed to a thousand. So I think what the algorithms do is they make it into manageable bite-sized chunks for us.
[00:05:09] Okay, so how do you practice? This is where it gets a little bit complicated because yes, we look and say, Oh, that's a basal cell. Great. But what you need to make sure you understand is that while you instantly said that's a basal cell, did you look at the whole slide to make sure that there wasn't a little lentigo maligna lurking in the epidermis? So yes, we do the instant pattern recognition. But if we're smart, we move from Malcolm Gladwell's Blink to Kahneman's Thinking, Fast and Slow, because, okay, we got the blink part, but wait, relax. Just take a look for one second and make sure you're not missing something else. Did you think of Merkel cell? Did you think of small cell melanoma? Did you think of small cell squamous cell? Did you think of anything else that could give you small blue cells? Go through an elaborate thought process. Is that all it can be?
[00:06:07] Christine Ko: Yes, I love it.
[00:06:09] Bruce Smoller: I think what the residents when they sign out with me, especially when I was chair, they really wanted to impress me because I was the chair of the department. So the game became, who can bark out the diagnosis fastest. And I would always say, don't do that. Don't do that. Yes, it's a basal cell. We all know it's a basal cell. But before you say as quickly as you possibly can, it's a basal cell. Did you think Merkel cell? Did you think small cell squamous cell? Did you think melanoma?
[00:06:36] Christine Ko: Yes. I loved your lecture. You encapsulated it very well just now. It was really only once I read Daniel Kahneman's book, Thinking, Fast and Slow, that I realized that this was relevant to how I could make a diagnosis and, like you said, prevent error. He emphasizes that quick thinking, that blink, that System 1, can be absolutely correct, but it can also be completely wrong, and you don't realize unless you question yourself. And then also System 2 actually is not the way that we normally think. The way we normally go through life is System 1, because otherwise it would just take too much time to take three steps if we're constantly thinking, do I take this step? Do I not? What's the worst that can happen? It took me a couple of years, but after reading that book, I started teaching that way, the way that you're saying. I'd say there's System 1. I think it's this, but what's the worst thing? Could there be also a second diagnosis, a melanoma hiding in there? And so I try to do that now, System 1, System 2, and just check myself, like, what's my algorithm? What's the worst thing it can be? Is there something else? Is there a third thing? Did I really get it right? May I ask, how do you continuously improve?
[00:07:48] Bruce Smoller: Well, there's an assumption there that I do.
[00:07:50] Christine Ko: I think it's a safe assumption based on your CV.
[00:07:53] Bruce Smoller: How do I improve as a diagnostician or how do I improve as a teacher? As a diagnostician, I still read the journals that come along. I also, interestingly, I listen to my trainees. For instance, I don't actually know all of the immunostains that are used in surgical pathology to distinguish ovarian cancer from colon cancer. I don't know. I don't care. It doesn't affect me on a day to day basis, but the once a year that we have a metastasis to the skin and I want to know, okay what antibodies do we work this up with? And the trainees say, Oh, this is what they do in SurgPath. Thank you. Let's do it. So it's a matter of not thinking you know everything but being open to, huh, I never heard of that. That's way cool. I never thought of that I never heard of that. Let's do that.
[00:08:45] So reading and listening to students is how I become better at what I do. How do I become better at teaching? If you watch your trainees, when you're pontificating around the microscope, you can see the things that register with them and the things that don't. And it's the same thing as when you're giving a talk like a grand rounds. Maybe the first couple of times you're so scared, you're hoping you don't pee in your pants kind of thing. You're not really attuned to the audience very well. By the time you're my age and you've given however many hundred grand round presentations, I know what I'm doing up here, I've done it a million times. What's working, what's not? And you start to actually be responsive to the audience based upon facial expressions, etc. You pay strict attention to that. So I do that around the scope every day. And monitor, am I losing them? Are they getting this?
[00:09:40] Yeah, feedback.
[00:09:43] Christine Ko: Yes. Feedback. So in addition to you talking about System 1 and System 2, you've now touched on, when you describe the best lecture you ever went to, and just getting immediately the answer, whether you were right or wrong, after you chose something, Picasso, or whatever. That's feedback. And to keep doing that same thing. Obviously, like you said, it's just four choices, but to have that sort of discrete goal. That's deliberate practice. Another concept in cognitive psychology, probably the best way to improve. That kind of deliberate practice. You set a goal. I'm going to learn about Cubist art. I look at these four, individuals and one painting and choose. Then you replicated that with your own study of psoriasis, lichen planus, I forget the two others.
[00:10:25] Bruce Smoller: Yeah I called Professor Jernstedt, and we went out to dinner. I drove up to college and we went out to dinner, and I asked him if he would mind if I stole that. And I told him what I was going to do. And he was very flattered. Obviously, we're all flattered when somebody thinks something we've done is good. And so I did replicate the study with second year medical students who knew none of the diagnostic criteria of lichen planus, lupus, erythema multiforme, and psoriasis. And I put together hundreds and hundreds of slides and just asked them to do the same thing, but this time it was all computerized. I replicated exactly what Professor Jernstedt had done. So yeah, instant feedback.
[00:11:05] Christine Ko: Do you have any final thoughts?
[00:11:07] Bruce Smoller: My final thoughts are, as I'm going to the back end of my career, that I've been incredibly lucky. I've had a 35 year run of working in good places with good people. I'm learning lots of stuff all the way, being paid to think, which is a nice thing. I get paid basically to think, and this discussion was really all about thinking, which to me is great fun. I look forward to continuing to think, but just using a different substrate moving forward. I don't know when I'm going to call it quits, but it's not too far into the future. So I want to thank you for one, the complimentary things you said about me, and also for the opportunity to chat about all of this. Thank you.
[00:11:53] Christine Ko: Thank you. Thank you.