EP 61 · 8:42

From correlation to causation

From Nobel Prize in Medicine 2026 Explained: Optogenetics

Episode
5/18
Watch Nobel Prize in Medicine 2026 Explained: Optogenetics
In this chapter

The hosts explain why 20th century neuroscience could only establish correlation, not causation, between neuron activity and behavior. Electrophysiology let researchers record action potentials and watch neurons like place cells fire, but correlating firing with behavior does not prove the neuron causes it. Electrical stimulation, pharmacology, and genetic knockouts each tried to test causation directly but fell short: electrical current spreads imprecisely through brain tissue, drugs diffuse too slowly and nonspecifically, and genetic knockouts are not reversible and allow the brain to compensate. This sets up three requirements for an ideal tool: genetic specificity to target exact cell types, millisecond-scale temporal precision, and full reversibility so a researcher can switch activity on and off and check whether behavior returns to baseline.

  1. 01

    The 2014 Nobel Prize in Physiology or Medicine recognized the discovery of place cells and grid cells in the hippocampus and entorhinal cortex.

  2. 02

    A study from Mayank Mehta's lab showed that hippocampal place cells can fire in response to visual cues alone, without movement, undercutting the idea that they only encode physical location.

  3. 03

    The Jennifer Aniston neuron, which fired selectively for images of the actress, is cited as a rare case of apparent single-cell specificity, though the on-screen note clarifies this was recorded in human patients and selective firing alone does not prove the cell causes recognition.

  4. 04

    The segment is dedicated to Shonali Dhingra, a co-author on the place cell paper and a mentor to one of the hosts, who recently passed away.

Transcript

1,826 words · auto-generated from the episode video

8:43synapses. >> So the question is if we want to study this thing, how do we establish causality? >> Yeah. Right. >> Right. Right. Right. >> Cuz in in for most of the 20th century, neuroscience was fundamentally an observational science. Okay. You observe what the neuron is doing and you try to correlate it to behavior. >> And researchers could record electrical activity of neurons. They could stimulate brain regions using electrodes. They could lesion particular structures. And this gives you a good sense of what is going on. But these methods have significant limitations. >> I can imagine because part of it's like you have to wait for the things to do what you're looking them to do and then hope and then track. And it's not quite

9:24>> it's it's not quite like a a two-way street of information. It's a one-way street of information. And it's done amazing things, right? Like consider electrophysiological recordings. So here you take an electrode. You can either put it in the vicinity of the cell or sometimes you can use a patch clamp and go inside the cell and see what the me the voltage is across the membrane from the inside to the outside. Neurons fundamentally use the transport of ions to create a kind of electricity that they use to talk and that's how you get these very very fast responses. Now with electrphysiology you can learn a lot about how action potentials propagate from one end of the neuron to the other. How neurons aggregate information like if a bunch of neurons are talking to the

10:05single neuron, how do you actually get this particular neuron to fire? Turns out there's ions that are coming in that change the membrane potential, right? Because if you dump a bunch of positive ions on the inside, then that's going to change the electrical environment. That electrical environment is then going to change the proteins that are nearby, which is going to let in more ions. That's going to let in more ions. You have this kind of runaway effect that creates action potentials. All of this was discovered using electrphysiology, meaning electrodes going into the physiology for us to measure precisely what the electrical environment is around a neuron or even inside of a neuron, right? Amazing stuff. Several Nobel prizes have happened because of

10:46this. Okay. But you reach a kind of ceiling on the kinds of science that you can do, >> right? Because suppose for example you're recording from a neuron in the hippocampus while the animal is learning a spatial task. This was very famously awarded the 2014 Nobel Prize in medicine and physiology for the discovery of place cells and grid cells in the hippocampus and in the in the entrinal cortex. And there on the lower right is my first paper that was in nature. I was a minor author in there. But the point of that paper was that you know originally it had been thought that these place cells in the hippocampus only respond to place. But the paper from my mantis lab that I was a part of

11:28um it showed that actually those same place cells can respond to visual cues only visual cues no behavior. Right? So now you have a relationship between place and this activity. Yep. >> The neurons activity. But now we've just shown that you know visual cues are enough to elicit the same cells to do the same kind of firing pattern that you see when when the animal is moving around. Right. >> I I remember we covered this briefly and it was very popular because people it was some it was the concept was like very surprising for people. >> Yeah. Yeah. Because they thought that you know play cells are place cells and that's it. They respond to place. Well, it turns out no, there's multimodal inputs that come in that are actually

12:10>> creating the activity of this particular neuron. And I just want to shout out um one author of that particular paper, Shaali Dingra. She was one of the first authors. She recently passed away. She was a huge mentor for me. She was a postoc when I was a graduate student. Um and a lot of the scientist that I became was because of her. So, you know, in memory of Chenali, um, I'd like to dedicate this whole episode because she she was a huge influence in my life as a scientist. The point of that paper was correlation doesn't establish causation, right? The the place does not establish why this particular place cell is firing and perhaps even the name place cell >> might not be that good of a name. It might be a multimodal

12:50sensory cell in the hippocampus. And so the idea being we've been doing this observational science which has given us a lot of advances um but we can't necessarily jump all the way to the conclusion that we can define causation >> just using that one-way street that we've been using thus far. >> Exactly. Yeah. And so you might say, okay, well let's try to establish causation by stimulating the neuron electrically. Right? You've got an electrode in there. Why don't I just pump electricity? The problem there is electrical stimulation is spatially imprecise. You've got an electrode, right? Unless it's targeting a single cell, then maybe you can you can make that cell fire, but then it's only going to be that cell, >> right? And how many cells are you going

13:32to try to elicit some kind of behavior? Often times behavior depends on networks of cells, not just a single cell. Although you can get lucky sometimes. There's the very famous Jennifer Aniston cell that responds only to when like the animal was shown a picture of Jennifer Aniston and it was literally called the Jennifer Aniston cell. But they got extremely lucky somehow, right? Where it's like they they showed the exactly the right actress and then they they elicited a response. So fine, there's causality, but then most of behavior is because of networks of cells, right? And sure, the electrical stimulation that you're putting an electrode in, that's going to cause some kind of activity,

14:13but it's going to be spatially imprecise. If it's outside the cell, that those electrons are going to leak everywhere because of the fluids in the brain. The Navier Stokes equation come back again and and so >> you're not you're not going to get um a nice one-way street. The resulting behavior could be caused by the neuron around there or it could be caused by single neurons somewhere else that are connected to some other part of the brain that is doing all of this nonsense. >> This this reminds me of the discussion around precision we talked about when we looked at hystotrist tripsy versus other uh ultrasound therapies in terms of using thermal ablasion which has more of a spread impact. >> Yeah. Because again the brain is conductive in it's also conductive in electricity

14:53>> versus having more mechanistic approach. The precision really matters here especially when you talk about the brain. >> Yes, exactly. Because single neurons can elicit crazy stuff, right? Um the other option you have is pharmarmacology. Pharmacology meaning like you you like inject some kind of receptor agonist or antagonist and then you you try to like block some type of receptor that blocks the neurons activity. But again, same thing there's a problem, right? Because the you inject something that's going to spread around because of diffusion. you're not sure if whether you targeted that specific circuit or not. The other problem with um pharmarmacology is the response time is quite slow, >> right? >> You're you're relying on diffusion. So you're injecting something, that thing

15:34has to spread, go to the neurons. >> Neurons are fast, >> right? >> So we've got to find a different way to do this, right? Now genetic approaches offer a way to get specific, >> right? because you can you can target specific neurons because you can be like, well, the parameal neurons in layer 3 have this specific type of genetic expression. And if I can hone in on that, then I can like turn it on or off. But that's not really reversible one thing, right? Once you turn it on, it's kind of hard to take out that kind of stuff. Um, and the brain also has time to compensate, reorganize, develop alternate pathways. Um, so

16:15again, not great. >> We're we're we're trying to get to a place where we can create a two-way street where there's a very well understood discrete control on our input into the system that does not have this like combinatorial effect in other ways that is. So we can isolate saying we only touched this thing and it only had this sort of next step outcome. And that way we can start to really probe uh with with an understanding that when we look for causation it we know that we didn't input more than we need. We didn't put too many ingredients into the recipe. >> Exactly. Exactly. Which we just changed one thing. Okay. So there are three things that we need for this perfect ideal technique. Okay. Here are the

16:56three things. First we need genetic specificity. Right. We need to be able to target specific types of neurons in a very specific location. Second, we need temporal precision, meaning timing needs to be extremely fast. There you see the electrical signature of an action potential recorded with electrophysiology. One of the great things with electrodes is that you can record like at 40 kHz. So 40,000 data points per second. So that's 40 data points in a single millisecond, right? In a thousandth of a second, you'll get 40 data points. So you can really look at the waveform and see what the waveform of an action potential looks like. You can see that the the width of an action potential is like a

17:37millisecond. It's it's right. So we need we need whatever tool we're developing to be extremely fast. And the second and the third actually I I couldn't find a visual for it so I just used the UNO reverse card. >> Uno reverse. >> But you want to you want it to be experimentally reversible, >> right? >> You want the ability to switch the neural activity on and off repeatedly, right? Because then you want that's how you really establish causation. You turn it on, something happens. Well, is that because you turn something else on? Or if you turn this off, do I recover the old behavior? >> Right. Do you go back to baseline? >> Exactly. Yeah. And one of the people who

From the episode
  1. EP 61

    Nobel Prize in Medicine 2026 Explained: Optogenetics

    How did proteins from algae give scientists control over brain cells? We explain the 2026 Medicine Nobel, optogenetics, memory experiments and early medical applications.

    Nobel Prize in Medicine 2026 Explained: Optogenetics

GeneticsNobel PrizeNeuroscience