Sep 15, 2026 · 1 hr 27 min · 10 segments
Neuroscientist Dr. Thad Polk (University of Michigan) joins Dr. Mark Guadagnoli to explain why aging isn't all decline, it's a process we have real control over. They discuss why brains…
Thad PolkGuest
Mark GuadagnoliHost
It's really interesting to me because in a way you kind of started with modeling the mind and then you started realizing, oh, the brain's something different.
Like, how would you describe that, the difference between modeling the mind and what the brain is?
Um, what Newell wanted to do was to, in some sense, build a programming language that has the constraints of the human brain, that uses the same sort of memory access mechanisms and the same learning mechanisms and things like that, um, that humans use, and then see if you can simulate human thought using that same kind of constrained, um, architecture.
Um, the problem, though, is that then you're always trying to test that just using behavior, and what you really want, I think, is, um, some way to, um, actually test the underlying assumptions of the architecture itself.
And that, to me, is what, um, cognitive s- neuroscience, uh, does with different techniques, including, uh, functional MRI.
You know, it's interesting, um, uh, w- we haven't talked about this before, but when I was in graduate school, um, the g- the thing that you and I had in common was the sort of cognitive psychology area, and I was on, you know, the human performance side as well, and, [clears throat] uh, you were o- over on the neuroscience side.
So, so a lot of the work that we did was around, uh, you may remember, uh, McClelland and Rumelhart's parallel distributed processing models.
And, and what you're making me think about now is the-- It's not, it's not a layer, it's not a single line, but it's multiple lines and, and to be able to figure out the computation that you're talking about i- is just an enormous, enormous project all the way through.
It's interesting to me because AI has come so far s- it w- seemingly so fast.
I mean, you know, we're talking now about, uh, you know, 40 plus years, but it really sort of, you know, burst onto the scenes ... seen just recently.
But you have an interesting story about AI, and, uh, [laughs] and your own experience with it as far as who you are.
It's really interesting to me because in a way you kind of started with modeling the mind and then you started realizing, oh, the brain's something different.
Like, how would you describe that, the difference between modeling the mind and what the brain is?
Um, what Newell wanted to do was to, in some sense, build a programming language that has the constraints of the human brain, that uses the same sort of memory access mechanisms and the same learning mechanisms and things like that, um, that humans use, and then see if you can simulate human thought using that same kind of constrained, um, architecture.
Um, the problem, though, is that then you're always trying to test that just using behavior, and what you really want, I think, is, um, some way to, um, actually test the underlying assumptions of the architecture itself.
And that, to me, is what, um, cognitive s- neuroscience, uh, does with different techniques, including, uh, functional MRI.
You know, it's interesting, um, uh, w- we haven't talked about this before, but when I was in graduate school, um, the g- the thing that you and I had in common was the sort of cognitive psychology area, and I was on, you know, the human performance side as well, and, [clears throat] uh, you were o- over on the neuroscience side.
So, so a lot of the work that we did was around, uh, you may remember, uh, McClelland and Rumelhart's parallel distributed processing models.
And, and what you're making me think about now is the-- It's not, it's not a layer, it's not a single line, but it's multiple lines and, and to be able to figure out the computation that you're talking about i- is just an enormous, enormous project all the way through.
It's interesting to me because AI has come so far s- it w- seemingly so fast.
I mean, you know, we're talking now about, uh, you know, 40 plus years, but it really sort of, you know, burst onto the scenes ... seen just recently.
But you have an interesting story about AI, and, uh, [laughs] and your own experience with it as far as who you are.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.