BJKS Podcast
A podcast about neuroscience, psychology, and anything vaguely related. Long-form interviews with people whose work I find interesting.
BJKS Podcast
104. James Shine: Integrating neuroscience with fMRI, collaboration, and the importance of dumb questions
James (Mac) Shine is a PI and fellow at the University of Sydney. We talk about his background in sports, using fMRI to integrate various parts of neuroscience, collaboration, and much more.
BJKS Podcast is a podcast about neuroscience, psychology, and anything vaguely related, hosted by Benjamin James Kuper-Smith.
Support the show: https://geni.us/bjks-patreon
Timestamps
0:00:00: Mac's sporting background
0:07:46: Overview of Mac's review in Nature (w/ Emily Finn and Russell Poldrack)
0:14:03: The role of great editors in improving scientists and their work
0:32:53: Connecting different levels of description
0:40:07: Integration and specialisation
0:48:49: You can scan any animal with fMRI - but they're usually anaesthetised
0:54:13: The transfer from human fMRI to animal electrophysiology
1:01:53: N=1 studies and layer-fMRI in clinical neuroscience
1:16:17: Collaboration and building a multidisciplinary lab
1:26:52: The magic formula in science: annoyance, excitement, and a constructive mindset
1:34:51: Writing grants as a test to oneself, and the art of reframing
1:41:52: A book or paper more people should read
1:43:37: Something Mac wishes he'd learnt sooner
1:45:43: Advice for PhD students/postdocs
Podcast links
- Website: https://geni.us/bjks-pod
- Twitter: https://geni.us/bjks-pod-twt
Mac's links
- Website: https://geni.us/shine-web
- Google Scholar: https://geni.us/shine-scholar
- Twitter: https://geni.us/shine-twt
Ben's links
- Website: https://geni.us/bjks-web
- Google Scholar: https://geni.us/bjks-scholar
- Twitter: https://geni.us/bjks-twt
References and links
OHMB interview: https://www.youtube.com/watch?v=ucDj_94ovaU
Boyden, ... & Deisseroth (2005). Millisecond-timescale, genetically targeted optical control of neural activity. Nature Neuroscience.
Finn, Poldrack & Shine (2023). Functional neuroimaging as a catalyst for integrated neuroscience. Nature.
Friston, ... (2017). Active inference: a process theory. Neural Computation.
Munn, ... Larkum & Shine (2023). A thalamocortical substrate for integrated information via critical synchronous bursting. PNAS.
Newbold, ... & Dosenbach (2020). Plasticity and spontaneous activity pulses in disused human brain circuits. Neuron.
Pezzulo & Cisek (2016). Navigating the affordance landscape: feedback control as a process model of behavior and cognition. TiCS.
Poldrack, ... (2015). Long-term neural and physiological phenotyping of a single human. Nature Communications.
Rao & Ballard (1999). Predictive coding in the visual cortex: a functional interpretation of some extra-classical receptive-field effects. Nature Neuroscience.
Shine, ... (2011). Visual misperceptions and hallucinations in Parkinson's disease: dysfunction of attentional control networks?. Movement Disorders.
Shine, ... & Poldrack (2016). The dynamics of functional brain networks: integrated network states during cognitive task performance. Neuron.
Shine, ... & Poldrack (2016). Temporal metastates are associated with differential patterns of time-resolved connectivity, network topology, and attention. PNAS.
Shine & Poldrack (2018). Principles of dynamic network reconfiguration across diverse brain states. NeuroImage.
[This is an automated transcript that contains many errors]
Benjamin James Kuper-Smith: [00:00:00] You mentioned in I think the OHBM thing that you did a lot of, or you were, I think you were being modest. You said you were part of a successful sports team. I don't know whether you were being modest or you were actually just like the water boy or whatever the word is. Um, But I was curious, was that, I want to guess, was it basketball?
James Shine: I've played on a few basketball teams. It was actually, oddly enough, an American football team at the University of Sydney. And American football is quite a niche sport in Australia. So we didn't exactly have the world's greatest competition. But at one point my team was the winningest team in college football.
We had the best winning percentage. across all college football teams across the world. I think they ended up putting an asterisk in their almanac saying that we weren't actually part of division one football or anything like that. But yeah, man, that was a great time in my life.
No, I, so I wasn't exactly the water boy I played. I played on all the teams. I played a lot of different positions on the field and I ended up coaching in the end when my body started to fail a bit. And [00:01:00] yeah, it was a really great time of my life for sure.
Benjamin James Kuper-Smith: Nice. Yeah, I remember I I mean, American football is mainly a big sport in America from what I can say, maybe Canada, I don't know. It's the same in Europe. I met one, one guy who was like on the European team or something. I was like, Oh, you're the second person I met who ever played this sport.
James Shine: Yeah, Yeah,
Benjamin James Kuper-Smith: yeah, I always find it fascinating when people do kind of things that are unusual for the culture they're in or the place they're in.
How did that happen in your case?
James Shine: Uh, So my mom's from the United States. She was born well, she grew up in Pittsburgh in Pennsylvania and met my father in grad school. They moved to Australia. He's Australian. And they moved to Australia and had me and my brother and then we've lived here our whole lives. But I think when you're young and you have a sort of a feature of your, background that stands out like that you can either run away from it or you can kind of embrace it.
And for me, I sort of really enjoyed embracing that, you know, fell in love with basketball and American football and and then I'm getting an opportunity to [00:02:00] play through the University of Sydney. Actually, when I was still in high school, they had a team and a couple of friends of mine, we'd always be throwing a football around and we ended up going and trying out and just having a great time.
And, you know, American football is a funny sport. You have to, it's very technical and you need a lot of equipment, you know, the helmet, the pads, very specific plays, a lot of, a lot that goes into it, but there's something about it. It's in some ways it's kind of like the ultimate team sport where you have to just do your role at an incredibly high level.
and trust that the process will play out as a whole. And if any one player on the team doesn't pull their weight, you can quickly lose. And so we learned a lot of really important lessons about teamwork and commitment to a cause and fighting through adversity that I still use to this day and teaching my students and trying to fight my way through academia and all of its trials and tribulations.
Benjamin James Kuper-Smith: Yeah, what was your role in that specific thing? I mean, I know nothing about the
James Shine: Oh all right. So you're familiar with the [00:03:00] concept where there's a guy that gets a hand of the ball and his job is to throw it, you know, down the field. It looks like a little missile that he throws.
Benjamin James Kuper-Smith: that's the quarterback, right?
James Shine: right. Yeah. So I was I played wide receiver, which is the person that runs down and tries to catch the ball from the quarterback.
And also tries to block people out of the way of the other people with the ball running. So I was on the offensive side of the ball. Didn't play any real defense. I didn't do very much tackling lots of sort of strategic planning. You plan these very precise combinations of what are called routes.
So let's say the quarterback will say before, before he goes back, right. I want you to run up and. After you get 10 metres, turn sharply inside. And then he'll set, so the other guy who's maybe to my inside, run right past me and then cut to the left. And what we try to do is sort of coordinate it a little bit like a sort of gymnastics routine or something like that.
It can have the timing and the the sort of execution has to be just on point. And you try to do it such that you'll come away from your defender right at the right time. That the quarterback then has an angle to throw you the [00:04:00] ball. And so there's this real kind of sort of chess game that goes on the whole time where you're trying to work out, how can I scheme my way open?
How can I get a leg up on the defender? How can I make them think I'm going to go left? And then I go right. And all this stuff that I found just so fascinating. And it's still a major passion of mine to watch American football and watch the sports and hope like hell that my team wins, which they almost rarely do.
But.
Benjamin James Kuper-Smith: Yeah, I guess you probably chose a bad time zone then, I guess, to watch. I Yeah yeah, I was just, the reason I asked whether it was basketball is because your banner photo on Twitter is someone I think stopping another guy from
James Shine: Yeah, man.
Benjamin James Kuper-Smith: something like that. What is the significance of that photo?
James Shine: So, so my wife is from San Antonio in Texas and they have a great basketball team, the San Antonio Spurs. And one of their best players over the last 20 years was this Argentinian man, Manu Ginobili, who's [00:05:00] one of my favorite sports people of all time. He's just got heart and courage.
And every time the team needs him, he's there always doing the dirty work. In fact, he was an all star for a large part of his career. But he actually came off the bench because the coach convinced him it would make the team better. And he did it. He didn't care. He said, I want to do whatever's best for the team.
And so I really love Manu Ginobili. And late in his career, Manu even though he should have retired probably two or three years beforehand on his own admission, blocked probably the premier offensive player in the league at that time, James Harden. He kind of jumped around behind him and blocked him in this mo the way that I've never seen anyone block someone in basketball before in my life.
And so it was a real beautiful moment. And and then online I'd seen someone put that picture of him blocking on the classic picture of Michael Jordan doing his big free throw dunk from back in the 80s. I just thought it was a really funny combination. Anyway,
Benjamin James Kuper-Smith: Ah, I see.
James Shine: I'm a sports nerd. Obviously you're gathering this from the answers.
Benjamin James Kuper-Smith: Yeah. Yeah. Me too. Just I like football and tennis. So there's a complete non overlap there in terms of the actual sports.
James Shine: I've just been [00:06:00] playing FIFA 24 with my nine year old. He's obsessed at the moment.
Benjamin James Kuper-Smith: He's, you're probably never going to win a game then, are you?
James Shine: No, he's much, much better than me.
Benjamin James Kuper-Smith: not going to win anything against an obsessed you know. And so, yeah, it's funny. So I've never played basketball, but now I'm in Switzerland. They have this great university sports where if I sign up to the gym, there's also like a indoor Court, basically.
So now I, what's the word? Shoot hoops or whatever. I throw, trying to throw some baskets for 10 minutes. I just, I'm so bad because I've never played it right. I just miss all of them. It's so embarrassing, but it's a fun way to wind down. And I'm considering whether I'm making it a goal to learn to dunk.
We'll see.
James Shine: Oh yeah we should. I love basketball. I think it's, it's this crazy, it's so difficult and they make it look so simple. It's, so once you learn the skill, you can figure it out, but it's this incredible multi joint timing. You got to, you know, you got to, you know, snap your hips before your your shoulders go in and [00:07:00] your elbow and then your wrist and your fingertips and you have to hold the ball just so, and you have to really have your focus on the rim the entire time is, it's a real kind of meditative exercise.
Practice trying to shoot the ball properly. Um, I love
Benjamin James Kuper-Smith: Yeah, I was,
James Shine: hoop in my house.
Benjamin James Kuper-Smith: was really surprised just because I mean, I'd only played, you know, like a little bit as a kid, right? And then if you're a kid, the ball is so big and the hoop's so far away. And now yeah, exactly. So difficult. But now I'm, I mean, I also quite tall. So I'm like, This isn't that far away.
This is easy. And then you just miss every time it's so hard. Yeah. Anyway. I mean, I'm sure we'll get back to some of the team stuff. You mentioned later when we talk about lab coach and that kind of stuff. But yeah, we're not actually here to talk about sports.
Although it was fun. I mean, you, so your, you and Emily Finn, who's also been on the podcast a few episodes back and Russell Podrick wrote a review together in Nature uh, half a year ago, something like that. And kind of about using fMRI to integrate [00:08:00] various parts or approaches to neuroscience.
And I guess we'll kind of just I think use that as a framework and then kind of just dive into individual things kind of as we go through it. So could you just briefly kind of give a short abstract of kind of what the, yeah, just roughly what the paper's about. So we have something to kind of, yeah.
To have a rough idea for what's to come.
James Shine: Yeah, sure. So this was a really fun project to work on and quite a challenging one as well. So Ross had reached out to me and Emily about this sort of, special issue that was being put together in nature about human neuroscience and the focus on some neuroimaging. And one of the editors, Mary Elizabeth Sutherland had an idea of essentially trying to paint a picture of what fMRI, functional MRI, what role it could play in neuroscience modern neuroscience as we see it.
You know, a little bit of a retrospective of, you know, where we've come from, what are some of the benefits and [00:09:00] costs of the field and the kinds of techniques. And then really, but more importantly, kind of prospectively, what could we do with it? What role could it play? And this is a very big. question to be asked because the field of human neuroimaging is vast with a lot of really amazing technically proficient and really conceptually challenging work to get your head around.
But rather than sort of staying embedded within that already kind of intimidating space we'd been, we're really motivated by this kind of broader concept of like, where is the field of neuroscience as a whole. You know, if we're trying to study the brain, what kinds of techniques can we bring to bear on it?
What kinds of questions could we ask of how a brain works or how a brain begins to fail? When you start to ask that question, you start to really have to grapple with the fact that a lot of the evidence that we have for how a brain works is really siloed away in multiple different subdomains within the field of neuroscience.
And to make this [00:10:00] really granular I had been at Society for Neuroscience. In fact, it was my first ever time at Society for Neuroscience the year before, when it was in San Diego. And I spent You know, a good proportion of my time there just wandering through the poster halls, interacting with different scientists that I hadn't met before.
Being from Australia, it's very hard to meet people face to face. And so I get to meet a lot of people whose work I loved and I interacted with before on Zoom. But also looking at all these posters and it really struck me after the second or third day that so many of the posters that I'd seen were just fantastic work.
But they were just completely different from one another. If you went from one poster on the neuroanatomy of the pontine nuclei down in the brainstem and their particular connections, let's say, in a rodent, you then walk down the corridor and now you're talking about computational models of criticality and criticality like processes that can come from manipulating a global gain parameter.
I mean, these are completely different modes of thinking, completely different types of background knowledge, and yet they're all studying the [00:11:00] same thing. They're all studying a brain. And it really struck me that I think if we really want to make progress on this problem of understanding how a brain works and how it fails, we really need to have better communication between these different areas.
And I think that's a very big problem. And that's a problem that the community as a whole, I think is going to have to incentivize in order for us to make it. The problem diminished, but what we tried to do with this paper was essentially make the argument that fMRI has a perhaps surprisingly important role to play in this process.
And the reason for that is that although fMRI has been maligned in the past for a lot of reasons, it's it's an, it's a non invasive technique, which is fantastic, but it's an indirect technique in that the signal that it's sensitive to is not actually the signal a neuron spiking. It's a measure related to blood flow or the ratio between how much deoxygenated versus oxygenated hemoglobin you have in a particular area [00:12:00] of the brain, which is then related itself to the amount of synaptic inputs that come into an area and how much it spikes.
But it's indirect. And so people in the past had often not looked to fMRI for playing this sort of important sort of endpoint role. But what's very clear when you look at the literature of fMRI is that it can be connected. to many of the different subdomains that are siloed. You can connect it to clinical imaging.
You can connect it to computational modeling, to a cognitive neuroscience experiment. You can connect it down to a systems level, micro level animal model. And the future that we envisaged was taking that connection very seriously and then asking, using fMRI as this Sort of bridge between these different areas.
Can we create a more synthesized neuroscience? Can we kind of bring together these different fields so that you know, a prediction that's made in a computational model can be tested in an animal model. We can then create a particular [00:13:00] signature of that prediction. We can then go and see if it looks the same in a cognitive neuroscience experiment, and then we can go test it in a clinical population.
And so we can imagine this kind of path now connecting all these different parts of neuroscience. You know, importantly, we don't think this is the only path. There are many different ways that you could navigate from these different domains and connect them together. But we were just trying to provide this kind of generative sort of exercise.
What would it look like if we did some of this work? Where could we see the big crux points going forward? And. And, you know, why we think there's reason for enthusiasm and excitement. If you're a young person entering into this field, here are all these cool opportunities. But we need help. And so that was really the kind of the Genesis behind the project.
Benjamin James Kuper-Smith: It's funny you mentioned Mary Elizabeth Sutherland, who I've also interviewed. So, and Russell Pordurak, we've spoken, but we haven't quite figured out to schedule it because he had time when I had, I didn't, and vice versa. So it's just [00:14:00] funny just seeing all of those names. I was like, okay.
Um, Also I didn't remember she's, she's more influential than I thought. I just saw like one talk that she did about I think when she just started at nature or something about getting behavioral sciences. I think she, she did one talk on YouTube. I found completely randomly. So I invited her and then it was like half of the, several people now have published papers with her as an editor since, since she's been on it.
Yeah, it's fascinating. I didn't expect that when I invited her based on this one talk I saw, but I guess
James Shine: She's just fantastic to talk to her. She has a real vision for, you know, where she could see the field. Sort of moving towards in terms of really beautiful outcomes that she can sort of identify threads that I think Can be really impactful for science and I've really come to learn over the years just how important of a role really kind of Far seeing editors play in helping shape the kind of trajectories of [00:15:00] scientific progress It's been a real eye opener for me as well
Benjamin James Kuper-Smith: Yeah, I thought I'd, if I'd asked this later, but I guess I might just take a, an even larger excursion now. Just because there was something you, so I listened, as I told you before I started recording, I listened or watched your video on the OHBM podcast, YouTube channel, whatever. And. You mentioned there, you talked about the Neuron paper you had with, I think, Russell Podrak, right?
And the, how that you initially submitted that to eLive and then Tim Behrens as the editor said, Yeah, you're not ready yet, basically. And I was really surprised by that, that I don't know, it seemed to me like he had a really, a much bigger role in I don't know, like the editors, I mean, I've had at least so far, I've always been just like, is it good enough or not?
Without saying I think you should do this next or this is the direction to go. So I'm just curious whether you could talk a little bit about that and kind of the, maybe the, yeah, let's say constructive roles that editors can have in like shaping a project.[00:16:00]
James Shine: Yeah um, yeah, I actually made a point to thank Tim, some years later, at a OHPM that I ran into him and he, Claimed as though he didn't remember that situation. So I don't know whether I was looking for mentorship that wasn't being delivered. Um, No, look, I think I jokingly made the point on that podcast that it was the best rejection I ever had because I think the, to set the scene a little bit more broadly for anyone who hadn't listened to the other podcast, I was working on a project in my postdoc with Russ Poldrack at Stanford in the US, and we found some interesting patterns in some neuroimaging data that we thought were really fun and interesting and kind of conceptually helpful for understanding how a brain could reconfigure to solve a cognitive task.
But we didn't really have a great explanation for how a brain could actually do that. We could, you know, we'd seen these patterns, but we didn't know what was underlying them. And we had prepared a [00:17:00] manuscript, which I thought was really great work. And I was really proud of it. I was proud enough to submit it and got rejected.
And was told essentially this isn't ready yet. You know, you need to go to the, get to the bottom of this. There's something interesting here, but it's not quite done. And I think you're right. I think up to that point as well, I was really of this mindset. Papers are kind of, kind of constrained object, right?
They're kind of this work that you put out. You put out a hypothesis, you have a set of methods and technologies and techniques, and then you assess the evidence and then you write conclusions and you move on. But I think there's this kind of broader meta level perspective, which is often much more aligned with why we all get into science in the first place, which I think is actually, separated from the process of academia, which I think actually ends up running a lot of our day to day lives.
And the scientist is not motivated, I don't think by an impact factor or motivated by a PDF, no matter how much it costs. [00:18:00] I think they're motivated by curiosity and they're motivated by I kind of drive to reduce your uncertainty about the world around you to kind of figure out something. And that was really a big moment for me because I think that kind of broke through the veneer a little bit.
He was some feedback that said this effectively passes the level required to have an academic outcome. But I don't think it requires it passes the level required to have a scientific outcome. Um, and I think it was really impactful for me because I got into into science and academia for for quite romantic reasons.
I was I just finished my medical training. I'd worked for a couple of years and I looked around the hospital and I looked around the kinds of options that were coming and the kinds of careers that I could have in that space. And they were all. really great careers, very meaningful work, you know, quite well compensated financially, [00:19:00] you know, quite a lot of job security.
You don't have to worry about whether or not you'd have a job next year or in 10 years or in 20 years. But I wasn't, I wasn't finding that it was kind of, resonating with that part of me that was just deeply curious about how things worked. Whereas science, I think, lets you do that. And I think if you're not careful in this game, you can kind of get caught in the trappings of academia and think, Oh, how do I get the tick?
How do I get the box? How do I get that next thing, that paper or something? And I think anytime you can come out of that world in and remind yourself that, you know, we're here to try to figure things out. And to help each other figure things out. I think it just, that little mindset change I don't know, it makes a big difference.
And so I, I try to talk about that a lot with my mainly, especially my postdocs. My, my grad students are more kind of just on in their projects. But with my postdocs, I try to really reinforce separating the things you have to do to remain in academia, separate those from the things that you have to do to be a really good scientist.
Thank you. [00:20:00] And I think it's on they're often overlapping. They don't have to be mutually exclusive. But I think sometimes they can lead you astray, and if you go too far to one side, you kind of miss it. You miss the kind of point that you could have if you stuck more with the science. So yeah, that was a really great moment for me, man.
And I tried to do the same thing now. So in terms of, you know, with the broader sort of editorial position, I think when you find a really good editor, they're going to make your paper better. They're going to make your science better. They're going to make your questions more refined. They're going to make them more nuanced.
They're going to make you more effective at communicating exactly your message and not some other. message that you may not have intended to communicate. And these are really difficult skills to refine as a scientist. In large part because when we get trained in science, we often focus so much on the technical capabilities.
We focus on whether or not we understand this method, this, how to get a good signal to noise out of my MR scanner, or how do I run this particular [00:21:00] mixed linear effects model or something to ensure that I get the right kind of conclusion that I can use to hypothesis test. But what really matters at the end of the day for communicating is B is making sure that you can say exactly what you intend to say in a way that's.
impactful for someone that they can they can reach in and understand. You haven't laden it with terminology that you didn't need. You haven't misled them with red herrings that actually don't end up having anything to do with the point you're trying to make. And we spend a lot of time in my group really trying to make sure that we understand the question.
and understand what we're trying to go after as our hypothesis and the clearest possible way to frame it. I'd say that 60 percent of the work is just really spending time there. And then once you have that lined up properly, the rest of the analysis and technology and techniques and things has a funny way of kind of getting out of the way and unraveling because you've got such a clear vision of what you're looking for.
[00:22:00] So this is a very long winded way of saying that. I think edit, editing and the process of being an editor is a deeply important part of being a really good scientist.
Benjamin James Kuper-Smith: Yeah, just from a very practical perspective, just because I don't have a lot of experience yet with peer review and that kind of stuff in general, publishing process. I mean, I've done some, but I wouldn't want to draw too many conclusions from that. But one thing I've just heard a few times is that, you know, you're not supposed to, as a reviewer, for example.
Supposed to suggest like new experiments or new things people can do or that kind of thing. Um, I don't know whether that's just something I heard incorrectly or whatever because it seems to me that you're very much in favor of it. Well, if it's well done, right? But saying you know, because in this sense, in this case, at least, it seems like you profited from someone saying yeah, you're not ready yet.
James Shine: No, I, in this case, in this case I absolutely did. And I made the point in the OHPM podcast [00:23:00] that question, that ref refining the extra work that we did in the point after that rejection. ended up finding new avid New Zealand roads that ended up being essentially the foundation for what is the sort of major thrust of my research group now that I run back at the University of Sydney.
So it for sure opened up a huge vista for me, but Let's not pretend for a second that I wasn't really cranky when I got the rejection and, you know, trying to drown my sorrows that evening. I think academia is really tricky in that regard. I think, so look, in, in terms of suggesting new experiments, the reason I smirked a bit is that in some fields that's absolutely what their goal is to provide a peer review service because that suggests all of the possible other conclusions that could have been made given your data and experiments and setup that you haven't yet ruled out.
And so if you go look in a field like physiology, let's say you, you're really good at patch clamping a neuron, which involves, you know, putting a little [00:24:00] pipette down and creating a suction on the edge of a neuron in a slicer. They can actually do this in awake animals. Now it's mind blowing to me. And they can actually measure the membrane voltage.
As the sort of fluctuations in this pipette they can then inject different kinds of chemicals, or they can inject current on this part of the neuron or something like that and understand it. A little bit like an electrician would understand a circuit. In that field, they have such precise control over what they're doing that if they've added in, let's say, a chemical, let's say they add in an NMDA receptor antagonist, But they haven't also first confirmed that there weren't other types of glutamatergic receptors in that particular area.
It might be that some of the conclusions they would make could have been explained by something else. And so they spend an inordinate amount of time agonizing over any other possible interpretation of their current results. What else could it be? And in this case, a peer reviewer plays a role. where they jump in and say, actually, have you thought of this?
I read this crazy experiment back in the [00:25:00] day where they did blah, blah, blah. They added some other thing. They added norepinephrine and it did this. Oh, well, I hadn't thought of that. They go and try it. And in a lot of cases, if you talk to physiologists, while they find this very frustrating sometimes because it can actually be used nefariously to essentially delay If you can come up with 100 experiments for someone to do, it takes them that much longer for them to publish their paper.
And if you're a competitor, you know, there's a lot of Machiavellian games going on. But let's say, let's assume for a second that it's well motivated. If you're a grad student, this is deeply frustrating because it takes you time. But if you're the PI or a senior postdoc and you really deeply care about what the answer is, because you want to know, This is worth its weight in gold.
You've just been given this gift by a reviewer because now you can be that much more precise. And if you speak to physiologists they really cash out a lot of their kind of scientific credibility and their ability to pick out those precise experiments you'd need to do. Now in, in, in the FMRI community, we don't share that same [00:26:00] level of nuanced feedback.
I think in part, because we have less control over Our experiments, we're doing more passive recordings of people in a particular context, like doing a. You know, this cognitive task or this clinical population, and we're then faced with this different problem, which is that we've got tons and tons of data, and it's very easy to fool oneself with statistics.
So most of the work that's done in peer review in fMRI comes down to the appropriate use of statistical techniques, the appropriate checks and balances, the right null model, the right kind of sort of, insurance that your. Not conducting or you're not making erroneous conclusions but that's a different kind of a process because if I went, let's say as a peer reviewer in fMRI, I went and said, well, actually your cognitive tasks that you just scanned 60 people on has this problem in it where now you have, you know, some issue with the kinds of timing of your tasks such that it's the subject expects there to be a reward at this time or something.
I can't expect you to go back and record all that again, it's going to cost you a few hundred thousand dollars [00:27:00] and you know, years of time. And that's just an impractical way to conduct ourselves as a community, but things are changing a little bit in the sense that there's a lot of open data. I think that we're starting to see more of a kind of proactive approach in the field where we can start to be more constructive with one another, which I think is great, but it really does come down to the kind of individual differences of the field that you're in.
For whether or not those things I think are helpful or not.
Benjamin James Kuper-Smith: Okay. Yeah. Okay. So as so often, it's about, yeah, norms in, in your particular speciality. Final question about this before we get to the thing we actually wanted to talk about. Um, why wasn't the paper then published in eLife later when you did the additional stuff?
Did you resubmit it there or because it sounds like that's, you know, you said, okay, you wanted us to do the thing. We did the thing. Here's our paper. But it ended up being published in Neuron. So how did that then happen?
James Shine: Yeah, that's a great question. So, You know, sometimes in the movies where there'll be like a [00:28:00] moment and then it'll sort of, there'll be like a montage. And then everything happens in a super intense way, and then you get to the end. That's kind of what happened for me after, after that rejection.
So once I kind of got over the hangover after you know, drinking myself into a stupor after rejection um, I got, yeah. No, I got to work really really, I went back and hit the books really hard. And I have a pretty broad interest, kind of, radius. I find lots of things really fascinating and I had actually a background in in clinical neuroscience in Parkinson's disease.
And so I had this sort of strong training in subcortical systems of the brain and neuromodulatory systems and things of that nature. And so I went back and read a lot of that literature and thought about the kinds of systems that could cause these kinds of network reconfigurations that we'd seen. And then I thought about brain stimulation literature, and I went and looked at the, you know, lesion literature, what happens if you have a stroke, what happens if you have, if you're anesthetized, went and looked at modeling, computational modeling, and tried my best to understand what you could [00:29:00] get from that.
And I ended up writing a review with Russ at that time for a special image in NeuroImage about a special edition of NeuroImage about kind of network reconfigurations, like what could cause them. And I think laid out a bunch of different potential hypotheses. But the one that, that really was attractive to us in the end that through a long series of conversations was this idea of the neuromodulatory system being involved in these reconfigurations.
And so we, we took a swing at testing a hypothesis using a dataset that we'd found from a collaborator. Where they had measured people's pupil dilation while they were in a scanner and and found a positive correlation between pupil dilation and this sort of network measure that we'd observed.
And it was in a very small sample only 14 people compared to the couple hundred that we've been using for the rest of the paper. But I think it was enough to suggest a new inroad in, in the sense of saying we've gone from just measuring this network phenomena. to, you know, this is a potential mechanism that could be causing this.
And I think [00:30:00] when you move into that realm of advancing the scientific discussion beyond just description towards actual kind of potentially testable hypotheses there's a chance to essentially, you know, know, kick it up an impact factor and test your luck. And so we, we took the the sort of challenge of writing it for a more general audience.
We're fortunate to get really nice reviews at Neuron and got published there. And as I say, you know, that was the start of everything else for me. I mean, afterwards, I started a collaboration with Michael Breakspear and learned about computational modeling, which now forms a huge part of my work.
I have hired two. Physics grads who we do lots of competition modeling work together. And I absolutely love the kinds of insights you can get from that. I did some pharmacological fMRI work with Sander Newenhouse and was able to test some of the predictions of this hypothetical framework with actual data and found really Lovely and confusing results there that led me down for future rabbit holes and so now, you know So much of my work is kind of you know [00:31:00] Changed through that Going down that path, but I wouldn't have it any other way.
It was a really great experience
Benjamin James Kuper-Smith: Okay, so you got rejected, then you submitted it to a better journal. Okay, that's nice. After maybe a bit of work,
James Shine: Well, yeah, that's what I
Benjamin James Kuper-Smith: ignore that for a second.
James Shine: for the not the montage of a year
Benjamin James Kuper-Smith: it must be one of the few times when when Yeah, usually I guess you submit it to the best journal you think and then it, you know, goes down x amount of steps after that in the hierarchy of journals.
I don't know if that's true,
James Shine: I don't know
Benjamin James Kuper-Smith: like in this case,
James Shine: I've heard that
Benjamin James Kuper-Smith: I mean, eLife is a great
James Shine: No, yeah, it's a great journal. And I, I don't know if I'd think about it that way. I've definitely heard of other stories that are far more impressive than going from a great journal. He left a gradual neuron. I've heard people that had a paper rejected one time, got really great feedback on it.
And once they incorporated the feedback, the paper was just so much more coherent, so much more close to what was originally intended by the [00:32:00] authors that they felt they could communicate it in a way that would actually resonate with a broad audience and then got into science. And I think this can happen, right?
I think we have this funny notion. That we throw the ball as high as we can and see how long it trickles down before it lands. As opposed to, this is why I say editing is so important, right? If you can really understand what you're asking and why someone else would care, and then actually have evidence that befits that.
you're actually in a position where you can communicate that to a broad audience. You can explain to someone who doesn't really know much about brains, why they should care about this particular nuance nerdy thing that you care about. That's, I don't know, that's like a bar to try to jump over. And I think if you can do that, then you should be proud of your work and submit it where it belongs.
Benjamin James Kuper-Smith: Okay. So your actual paper the thing because of which I invited you. Yeah, so I I wanted to, I should put it, not exactly questioned the assumption [00:33:00] behind it, but kind of maybe more set the boundaries for this kind of integrating across different fields because I think it's fairly uncontroversial to say that there's a limit to how, you know, far apart different kind of disciplines can be integrated.
You know, do we need to integrate some sort of transcription mechanism into social interactions in humans or something like that, like at some point it might just become a different kind of level of description and it's a different thing altogether in the sense that, you know, biology isn't applied physics or whatever.
It's not just quarks and then you go from there. So I was just curious kind of what you thought about the scope or scale here of like where it makes sense to use fMRI to integrate these different research areas.
James Shine: Yeah it's, I think you're right that depending on how you frame the question, you can kind of either think of it as a bit silly or maybe you can use it to get a bit of traction. So it's, you're right. It's silly to think we need to have a, or I would agree, I would argue that it's [00:34:00] probably not going to be the best use of our time to quarks and gluons and bosons are having an impact on whether or not, we're making decision A or B whether we're choosing the cupcake or the the apple there's going to be much more satisfying answers that come from a much higher level of description.
But I would argue that having conversations between specialists in different domains is a deeply effective means for both figuring out what you do and do not know, and what you do and do not understand, but also moving the needle to find out where the actual sort of truth out there is if there's such a thing.
Assuming that we aren't living in a simulation, that there is a reality out there which, you know, with some of this weird social media stuff, I know that it gets harder and harder every day. But assuming that's the case think that every perspective that's taken in science, and this isn't just in neuroscience, comes with a set of assumptions and and baggage that make it very effective at asking certain types of questions, but [00:35:00] really limit its ability to ask other types of questions.
And one of the things that's quite interesting is that you, when you start to think about this a bit. Is that oftentimes the kind of blind spot of one field is actually an area where the other field has really beautiful crystal clear 2020 vision. And as someone who's deeply interested in how the brain works and fails, if you want a satisfying answer, I don't think it's going to come.
And I might be wrong about this is my hedge, but I do not think it's going to come from one specific subdomain of a sort of segregated neuroscience that just has the right kind of measurement. It might be the case that happens, but what I think is more likely is that each of the different subdomains will be able to provide a piece of the puzzle and that connecting those bits together is going to lead to really satisfying answers.
In part, I think this is intuitively motivated by the kind of training that I got in medicine at the University of Sydney back in the day, where if you look in other organ systems in the [00:36:00] cardiovascular system, in the renal system, in the liver you know, in the lungs the explanations for how the system works, really can cross these different scales, these different resolutions from cells.
You know, take the heart as an example. You can describe it as a series of cells that are kind of loosely connected together, some of which like to pulse that send little electrical signals that pulse a little bit of different frequency and then pass it on to a little fibers that have this sort of interesting propagating electricity through them.
But it's only because they're connected together in such a fashion that they form a pump. That when the cells start to spike, they cause a traveling wave that then causes the contractile muscle cells to kind of shorten. That then causes them to contract. And then if there's blood inside those chambers, that blood gets spurted around at high pressure around your body and can feed organs that require oxygen.
that has been oxygenated by the lungs. So you can understand the system at these different scales from cells all the way up to the system. And then importantly, you can also turn around and say, well, what does it look like when the heart starts to [00:37:00] fail? If someone has a heart attack versus if someone has a mitral valve relapse, or if someone has you know, cardiomyopathy, you get these different types of failure modes that are related in different ways to the cells and the systems level description.
And they'll look like this when you put an EKG on someone's chest, or if you do an echocardiogram or you do an MRI. And. When you come to neuroscience with that lens it's sort of, it's just sort of fragmented, right? The cellular level descriptions that people are creating today in modern neuroscience are beautiful.
They're absolutely stunning. The kinds of techniques and tools they have where they can go in and selectively, they can put a virus into a very specific cell in a very specific area of an animal's brain. And then we, by flashing a light on and off, they can turn that cell group on and off. And what that means is that cell group can either spike or not spike.
And now they can then assay the behavior that is the consequence of that that lesion, that artificial lesion, which is temporary, mind you. And they can figure out what happens when I do that to an animal. Does it do [00:38:00] this function or that function? Does it, you know, wait longer for the reward? Or does it go exploring somewhere else, you know, out in the rest of the environment?
But importantly, you can then go take that vehicle and you can put it into an fMRI or an MRI scanner. You can calculate, you can get fMRI scans, and you can essentially say what happens when you take this precise control and you couple it to the kinds of systems level descriptions that we have. Because importantly, that systems level description is something I can get.
If I went, you and I go sit in the scanner tomorrow, I can scan you doing anything you want. You could listen back to a recording of this podcast for all the kind of awful descriptions I give. And um, and you could measure your response, right? Something super nebulous you could never do in an animal.
But then I could ask, as a scientist, is the kind of signature that I just saw in the fMRI data the same or different? And what happens when I take this animal and I stimulate and turn off this or that cell group? And what a beautiful way for us to essentially go from the sort of microscopic level of description to this macroscopic, [00:39:00] beautiful world of complex cognition and internal states and motivation and hopes and dreams and desires and all the amazing things that the human brain is capable of that are incredibly difficult to understand.
to precisely control in an animal. And now we can do those together in the same conversation. And to me, I think quite selfishly I want to be a part of a neuroscience that where that conversation can happen. Cause I just think that's what the answers, the really satisfying answers are going to look like.
Is there going to be this conversation across these different scales?
Benjamin James Kuper-Smith: I mean, it seems to me intuitively, that's also like where all the progress is usually made, right? And these kind of gaps between the different, I mean, not all of it, obviously you need specific people to figure out specific things, but a lot of the interesting thing, you know, I mean, the lab I worked in for my PhD did some clinical stuff and, you know, the interesting clinical stuff then came from developing tasks that had good computational modeling associated with them.
So you could then. Do them with patients. Right. Is I think a lot of these really interesting things [00:40:00] happen in this Yeah. In these bridges between different areas.
James Shine: But I would love to add to that because I think one of the really striking conclusions from writing this review with Em and Russ was that While we think that integration is going to be deeply important, if you went fully towards integration and let's say tomorrow, every single graduate student said, that's it.
All I'm doing is connecting across the different areas and you lost the specialists in each of the areas. You'd be arguably in a worse place than we are now, right? It's a risk because what you need is you need specialists in each of the areas to be able to conduct themselves in at such a high level that they can really trust each other.
The information that's coming from that area, you know, it's not, it's like the physiology example I gave, is it this receptor or that receptor, have we run all the checks and balances to make sure it's exactly that. And once we have that, then we can feed that to the model, then we can feed that to the.
person in the you know, the clinical population, what have you, to create their hypotheses. But without that specialization we run the risk of [00:41:00] essentially sort of turning everything into a little bit of a kind of homogenous mess. And so I think what really matters at the day is this kind of tension between specialization and integration on the two hands.
And then if you can find that tension, that's when the real real benefits start to come.
Benjamin James Kuper-Smith: From a career perspective, how would you advise to do that? Because I guess there's this, I put it? Is it, I think one common set of advice is to specialize first and then kind of branch out. Do you think that's sensible or I mean. Yeah, basically, or should there be people who from the beginning are these kind of integrators or do they just risk not really understanding anything or yeah, how do you, yeah, basically, I mean, you have students, right?
How would you advise them if they say I want to do this or that would you, It seems to me like it seems to be obvious what to do. If you want to be a specialist, just be a specialist. That's I think the easy
James Shine: doing the thing.
Benjamin James Kuper-Smith: [00:42:00] Exactly. But as soon as you want to be more of an integrated thing, do you think that's something you can start early on?
Or is it something that only comes once you've found some expertise in particular domains?
James Shine: No, it's a really great question and it is one we think about a lot and I don't think there's a kind of silver bullet answer for this. I think that if you have a predilection towards interdisciplinary answers, then it behooves you to know enough about the different fields that you're able to, you know, provide a You're able to inch forward the science in that field.
in a way that would be respected by the members of that field. So I think that's a, there's a sort of a challenge on your shoulders, which is that as a grad student, it can be deeply intimidating to just understand one subfield. I know how that feels. It's a really scary time as a grad student where you feel like, what am I, you get imposter syndrome and you think, Oh, what am I doing?
There's so much literature. It doesn't stop. And everyone's doing all this incredibly complicated stuff. I don't get it. And, [00:43:00] But you know, if you plug away at it, you kind of get some proficiency in an area enough that you understand it. I think the people that lend themselves towards being interdisciplinary enthusiasts, I think if you're interested in that, the best way to do it is by surrounding yourself with people that are interested in those other areas.
So if you're let's say you're someone that likes computational modeling and neuroimaging. And there's no group around you that does both. Let's say you're in a neuroimaging lab. What I would recommend is that you find out what kinds of opportunities there are around you to do any of that work.
And if not, if there's no one locally, then you find out online whether or not there's groups that are doing reading groups or groups that are you know, Neuromatch Academy is a great example. You can go and learn some of the skills and techniques that you'd need in order to conduct yourself at that level.
I think what you naturally find when you take this seriously is that the kinds of questions that show up to you, the kinds of adjacent possible for your next steps, tends to be a blend of the things that you've learned. You can't [00:44:00] help but notice that, oh man, this neuroimaging experiment would be so much better if we just combined it with this computational modeling approach.
And then you, then that's where you need real humility because what then comes from that is a need to collaborate. people that are experts in that area. And sometimes that comes from serendipity. You just might happen to meet another grad student who's in a lab like that at a conference. You go to all the posters at SFN.
Odds are you'll run into someone who's doing something cool that you like. Sometimes, and this has been the case for me, It's just being determined, done good about it And so I learned a long time ago that I had to check my pride whenever I wanted to work with a new specialist. And I just tell them, you know, I write emails you know, I'm interested in your work.
I think it's really impactful. I think we could do this together in this way and I'll, but I need to learn from you. And I think coming to people in science with a motivation that's about science and curiosity and enthusiasm. I think [00:45:00] when you find the right people, it really clicks. And, you know, five or six of my really great collaborators now have come from that route, you know.
So I'll give, as an example during the pandemic, I got really interested in reading a lot of cellular neuroscience literature that I had. Kind of put it off to the side for a long time because I just didn't have the time to read it, but Stuck alone at home with my wife and kids, you know, there was only so much Video games that I could play with my kids and TV shows to watch.
I started reading and one of the really key Results and groups that were doing just really cutting edge work There's a group in Berlin working on a particular type of cortical cell called a layer 5 pyramidal neuron. And I just thought these things were fascinating. They have this really beautiful interaction with the arousal system.
They relate to the thalamus in deep ways. And so a lot of my work ended up having, using their evidence as kind of, part of the scientific story that I was trying to create. And um, I just thought one day, you know what, rather than just reading [00:46:00] all these papers and having all these questions, because every time I'd read their papers, it's so foreign to me.
I'd have to underline things. What on earth is this? What on earth is that? I don't understand why they would do this. I ended up just emailing the senior author, Matthew Larkin, and he wrote back and actually has said that he'd read one of my papers which was kind of tangentially about this and he would love to talk.
And so we ended up hitting it off and I've had, you know, multiple You know, long conversations. Now we've I've actually seen Matthew play in you know, in a an orchestra in Berlin. He's a quite accomplished violinist. You know, had dinner with him multiple times. We've collaborated on multiple papers and And it's someone that I really love working with and I think that all came from just wanting to work with people that are doing, have a particular skill set, but just not having access to that locally and using the power of the internet and you know, getting over yourself.
Being okay sounding like an absolute moron. If you ask Matthew about it later, I'm sure he'll tell you that I said really stupid things to him and probably still do. Cause it's really hard in science to be around all the [00:47:00] details, right? But I think once you check your ego and just kind of get in, get into it, you can have these really great conversations with folks and then come up with what I think are just far more satisfying answers for the kinds of problems we're trying to solve.
Benjamin James Kuper-Smith: Yeah, Yeah, it's difficult. Um, it's just, it can often be difficult like, trying not to seem stupid, even though that is kind of the only way forward. Um,
James Shine: Do the, I go the other way. I lean into it.
Benjamin James Kuper-Smith: Yeah, but I was gonna say like, isn't that kind of the way you do it? Like, It's so much easier to go to someone and say like, Hey, I don't understand this thing.
Can you explain it to me? Rather than, you know, you know, kind of proactively saying what you don't know, rather than afterwards being like try, you know, when you have these things where you're, you kind of understand where they're going along. And you don't feel like you want to ask a question yet, but then it just keeps going.
And then you're like, now I've waited too long, now I can't ask anymore. So it seems to me that this kind of proactive approach of just, [00:48:00] yeah, going I mean it seems to me also just, I mean in a way you're also just actively signalling to everyone what you're trying to learn, right?
James Shine: I think that's a huge part of it and I don't know who said this, but someone's definition of an expert is someone who's made every mistake there is to make in a particular field. So I think in some ways we've got this obsession with kind of waking up one morning and just having this stroke of inspiration, like you understand it all.
And I just don't think that's how it works. I think you go out there with curiosity and enthusiasm and you put the question first and through a sort of long period of really being dumb, you start to learn the better way to think about it. And I think that's true for almost everything I've learned in science.
Benjamin James Kuper-Smith: Okay um, I had a question about the, so in the paper you mentioned your strengths and weaknesses of fMRI. I guess these are fairly general strengths and weaknesses, but also in [00:49:00] the particular context of this integration. I was just curious, it seemed to me a little bit as if two, one of the strengths and one of the weaknesses kind of cancel each other out.
So I was just curious whether you could comment on which was the I think one of the first strengths you mentioned is that, you know, you can kind of scan any species with fMRI, more or less. But then you also said at the end that the animals usually have to be anesthetized. So I was just curious, kind of, doesn't that largely mean that you can't really scan them the way you'd want to scan them?
And then also the transfer from the whatever imaging you do, or electrophysiology or whatever in the animal is then very difficult to do. to learn how that translates to fMRI if you can't do most of the tasks with fMRI with, let's say, a mouse or a rat.
James Shine: Yeah, no, you're spot on. And this, there's no free lunch here, right? It's, there's, it's not if we woke up tomorrow, we could just put all of the governments of the world's chips on this experiment. And this would be the perfect experiment to do. We'd have it, we'd have it [00:50:00] clocked. I think that what we're looking for integrated bridges that could be built across all of the different little jumps that have to be made in the hope that.
We could then reinforce those over time such that they'd become highways rather than little rickety bridges. And the problem of anesthesia definitely permeates animal fMRI work. There are a couple good things to say on the, in terms of, you know, going forward into the future. One is that people are starting to work out protocols where you can actually train animals to habituate to a scanner so that they don't.
become agitated and stressed and they can just lie without moving. This has actually been rapidly augmented by newer and different sequences that actually don't have the same awful sound that like a traditional fMRI sequence and epi sequence would. Some of them are actually. they use different gradients.
And so there's a lot of advances on that front. So the louder the sound, the more the animal stresses out, the more likely they're gonna move. And so if you can keep [00:51:00] it quiet, it's more still. The second thing is to say, we're developing quite rapidly a better and better appreciation of how to characterize the difference between the awake and the anesthetize state.
Which means that we can start to, I guess, Interpret whether or not elements that are happening within a particular recording are due to the baseline change in the brain state versus the added change on top. Now, that doesn't change the fact that any additional change you make to the brain in a particular state could be completely dependent on that state.
So for example. If I walk up to you and offer you 10 million if you raise your right hand when you're anesthetized versus when you're awake, you can make lots of money when you're awake, but you wouldn't when you're anesthetized, unless I was extremely unlucky and there was some kind of motor twitch.
So there's going to be certain things that when we conduct our experiment in the anesthetized brain, it's just fundamentally different. than when we're awake. And we have to take that we have to take experiments with a grain of salt and be very careful about the kinds of conclusions we make.
But things like computational [00:52:00] modeling allow us to make really precise statements about what it is we are doing to the brain and then what state we're recording it, such that we can actually develop predictions from anesthetized animal experiments with the brain stimulation that we can then test through a model to you know, conduct experiments.
So just as a constructive example I have a collaboration with Valerio Zerbi who's now at the University of Geneva and he does a lot of really great opto fMRI work where they'll put a virus into a, an animal's brain, usually a mouse or a rat. That virus will then express a particular channelrhodopsin in a population of cells.
In the case we worked on was in the locus coerulus, this really deep structure in the brainstem that provides all of the neurogenergic projections to the rest of the brain. And then even though the animals were lightly anesthetized, they were put in the scanner and a whole brain bold was recorded while they were scanned.
And essentially what he did was then go in and turn the light on and off in, in blocks of different frequencies. And then he could measure the whole brain bold signal. And through some [00:53:00]statistical modeling, you can then test hypotheses about whether or not, that particular brain structure when turned on or off compared to a sham.
When you turn on a different frequency of light that doesn't activate the channel whether or not that has the kinds of signatures that you would see if someone like you and me went and lay in the scanner and we did a typical analysis of the bolt data. And so these kinds of experiments are becoming more and more common.
And I think, you know, there's no perfect salve, it's not like we can, you know, we just can't do the same experiment in a mouse that we can in a human. You can't ask a mouse how to describe how it feels. You can't ask a mouse, you can't give a mouse a verbal instruction of a one sentence rule and then have it do a complex cognitive task.
It's just not something that is possible. And so in some ways we shouldn't be looking, I don't think, for that kind of an experiment. We should be instead asking ourselves, in what way can we create connections that can go all the way from one corner of this network to the other and back? And it will, [00:54:00] I don't think it'll be wormholes that jump right from one side to the other.
It's going to be more little paths that then can over time be coalesced to create a better picture of the object in the middle of all this, which is the brain.
Benjamin James Kuper-Smith: You mentioned one thing that I thought was quite interesting that is obviously also in the paper, which is the you hinted more at it here, that I think like in, in fMRI and that kind of stuff, I think, I guess we have a bit of a inferiority complex in that, you know, the animal people are so precise, etc.
In, in their methods and what they can measure and all this kind of things. I think at least many people I know in fMRI are kind of a bit frustrated with you know, how indirect and vague a lot of it is. But one thing I thought was kind of interesting is that, you know, as you mentioned, the transfer also should go the other way.
And one way in which human studies or fMRI can do this is by, you know, using the unique advantages that humans have, such as being able to understand instructions easily and quickly. Yeah, can you just expand a little bit on that and kind of, because I think that's a [00:55:00] really interesting and important point.
James Shine: Yeah. It's sort of, it's one of these sort of these points that once you sort of think about it, you can't unsee it in a way. That as much as people in kind of human imaging typically either love language experiments or hate them, language is this really kind of, human specific function that that really is sort of inextricably linked with our cognitive abilities.
And if you look at how animals are often trained for a particular behavioral experiment. The way that has to happen because they don't have language is this really exhaustive protocol of trial and error and reward. And they use things like operant conditioning and instrumental conditioning to have the animals learn through this trial and error how to do the particular function that they are interested in.
Whereas for a human, we just ask them, Hey, can you just go in and, you know, wiggle your right index finger, you know, two times a second or something for the next five minutes. Learning that. As [00:56:00] an animal, it's got to be incredibly difficult and you can even argue it as extreme that it might be that different kinds of brain circuits end up having to become involved in order for the animal to conduct itself in that way that are very different than the ones that a human does.
You know, if you got me to go into a dark room, I was really thirsty. You haven't given me any water. If you give me a dark room and there's a couple like little paddles in front of me that I can hit, I'm probably going to try to trash that room before I'm going to do your experiment, right?
Humans are not used to being trained in this way. We'd consider it some form of torture. And so we're so used to leveraging language and so used to using it as a shortcut to translate intention into action that it's just hidden in plain sight. For everything that we do in neuroscience, in human neuroscience.
So I think there's a really difficult let's just put to the side for the fact at the moment right now that mice brains and monkey brains and rabbit brains and cat brains and fly brains and coelacanth brains look completely [00:57:00] different to human brains. There are similarities, no doubt.
They've all got neurons in them, they're all organized in sort of similar kinds of ways, they have a similar kind of bowel plan, but the difference is really astonishing. But put that aside, we actually use them functionally in really different ways as well, and I think until we take that really seriously and think carefully about what we should expect the brain to look like in the case of overtraining an animal to perform a function versus asking a human to perform one we're going to be a little bit stuck.
There's a really big gulf there that's going to be very hard to overcome. But the hope is the benefit the sort of the silver lining in that process is that anyone that conducts animal experiments that sits there wondering to themselves after they've just spent six months training an animal to do a task, whether or not it actually is doing what they want.
We can actually devise experiments now that can test that, right? We can get a human to go in and we can say, I want you to do this task, but I want you to do it this way. I want you to do strategy A versus strategy B versus strategy C. [00:58:00] It turns out that, you know, they all look kind of the same, but A has this part of the brain more active, and B, this part lights up first before this other one, and C, it does this completely different thing.
And then you go look in your mouse. And you go see whether or not it does A, B, or C. And odds are, it'll probably do, you know, C star or something like that. It'll do something a little bit different. But at least you'll have inched forward trying to understand a little bit this black box that you can't look into.
But in, in a human, you can ask all kinds of questions. You can ask anything you want. So, I think this is one of the kind of points we try to bring out is that there are certain things that You could do in a human and experiments you could devise in a human that would just be extremely hard to do at a mouse in almost impossible, but could inform the kinds of conclusions and hypotheses you could test in that mouse data.
So we could envisage it going both directions in lots of different cases.
Benjamin James Kuper-Smith: Yeah. When you talked earlier about for example, let's say a mouse fMRI study and you want to, you know, connect that to human fMRI study to state the obvious, mouse brains are smaller and presumably [00:59:00] more or less, I mean, obviously there's like lots of species difference, blah, blah, blah, but presumably a hippocampus is a lot smaller than a hippocampus in a human brain.
Therefore my question is basically about resolution because if you use a, I'm assuming if you use a normal human fMRI scanner, you're not going to get that many pixels, voxels in a mouse brain. But I've also heard that I think they have really high tesla counts. Um, So I was just curious whether you could comment on that.
Is that is that a problem or does it actually even out pretty well?
James Shine: Yeah. It's a great question. And it, the answer you kind of answered for yourself. It does. It does even out as you start to go up magnetic strength, you can change the field of view as well. So you can get. Sort of, you know, approximately equivalent reductions in the size relative to the brain.
But there's actually a harder problem, which is that if I lined up, let's say I like took all the parts of a human brain and all the parts of a mouse brain, however we want to characterize them and sort of lay them up next to each other. There's not really a direct one to one [01:00:00] correspondence between lots of the parts.
They'll definitely be. equivalences. You could say something like, oh, your inferior olive looks a little bit like the mouse's inferior olive, and your medial dorsal thalamus looks a little bit like the mouse's medial dorsal thalamus. But there'll be certain areas, and in particular the kind of frontal parietal areas of the cortex, but also the anterior and posterior parts of the thalamus, parts of the striatum, parts of the cerebellum, parts of the brainstem, where you're going to find A lack of homology.
You're not going to really be able to find that same equivalent area. And so there's actually a lot of controversy and debate in the kind of sort of interspecies field about exactly how to do these kinds of comparisons the most effectively, what is the best way to parcellate the brain so that you have the best overlap between different species.
And there was actually a whole session at human brain mapping on just exactly this run by James Pang and Alex Finito down at Monash University in Melbourne. And it was a great session and there were more questions than [01:01:00] answers and it was really fun to, to kind of hear this debate ongoing.
Cause I think this is one of those areas that is just not going to lend itself towards. really simple golden bullet solutions. It's going to be, you know, you can do this and it requires this assumption and that lets you ask this kind of question, but be careful because you can easily fool yourself into thinking that this answer is actually a cool answer when really it's just an inevitability based on the difference in their brains.
So, interspecies neuroscience is just full of these kinds of trappings, but it's better than we're asking the questions I think they're not.
Benjamin James Kuper-Smith: In the paper, you also go a bit more systematically through four different well, the four main pillars or approaches or whatever you want to call it, or areas of neuroscience systems, cognitive, computational, clinical.
You mentioned before we started recording that Emily and Russell wrote the cognitive computational. Parts and you written all the systems and clinical one. Should we then maybe focus on the clinical one to get a [01:02:00] few maybe examples? I thought maybe one fun thing to do might be. So from what I understand Um, uh, so Russell Podrick analyzed his brain for or measured his brain for ages.
Um, And I think you analyzed some of that, right?
James Shine: did. I, I came along to
Benjamin James Kuper-Smith: probably one of the, one of the few cases
James Shine: Yeah.
Benjamin James Kuper-Smith: When a postdoc analyzed the supervisor's brain. Um, But you mentioned this as an example of end of one studies and in the clinical context. So maybe yeah, I'm gonna take this whatever direction you want to.
James Shine: Yeah, sure. So yeah, the, I think the paper itself was, it was deeply collaborative and I don't think it's fair. I found myself balking when when you described it as me writing a couple of the sections that I think it was really a group effort. It's more that those are the areas that I was a bit more familiar with I could speak to a little bit more closely.
But what we tried to do as well and. We debated this for a long time, was trying to kind of find the right resolution to describe neuroscience in a way that wouldn't be [01:03:00] particularly controversial, but still captured a lot of the kind of main players that we see in the field. And I don't think for a second that for communities is the right way to think about it.
Systems, clinical, computational and cognitive, but they were just ones that we had access to and knew a little bit about so we could speak to. You know, we, for example, in this review, we didn't talk a lot about the motor system or a lot of the really precise sensory systems that there's really beautiful work that's been done in those communities.
And a lot of times they don't use fMRI as much as they use things like electrophysiology. So it made it a bit easier. But I don't want to make it sound like these are the four, but so yeah, that's a caveat. But in terms of the clinical world, what we tried to do in each of the cases was to sort of imagine those different subdomains and then think, what kinds of problems are they facing?
And The clinical one was really interesting. I actually have a clinical background and when I first started and did my PhD in functional [01:04:00] neuroimaging, a lot of the questions I would get would be along the lines of quite practical, almost engineering type questions. You know, how are we going to use this scanner to help me diagnose this patient or that patient?
Or how can I make this worth my while for my patient? And I always felt really like I was sort of almost letting the patients down because fMRI really has not had. a tremendous amount of success in that area. There's definitely cases where it has. There's really the neurosurgeons have done a great job of extracting utility from imaging say pre op scanning to determine whether or not someone's language areas overlap into an area that they were planning to resect for a tumor or something like that, or finding an epi
Benjamin James Kuper-Smith: Or even in op now, sometimes
James Shine: exactly.
So there's a lot of great work being done. But so what we didn't want to necessarily cover that realm, it wasn't something that the three of us had worked on closely. What we wanted to point more towards was the problem that's facing the clinician that would might be interested in [01:05:00] fMRI. And the problem is that so many of our insights.
Because we're dealing with this large, noisy sample signal. So many of our insights require group level statistics. Yet clinical problems, by definition, are N of 1 problems. I've got someone in front of me, I have to make a choice. Do I give them this drug or not? Do I put them into this trial or not?
Are they in this group? Are they at risk of this or not? And, so it's a, the problem's almost framed in a different way. And one of the things that we are really encouraged by, and I think Russ's MyConnectome project is just a lovely example of someone kind of getting in there and, you know, rolling their sleeves up and just sort of being a change that they wanted to see.
So what Russ did was he said, we're doing too much scanning of one, one person at one time. Let's take a snapshot of this group at time point X. And we're not actually asking about X plus one, X plus two as it fluctuates over time. And there's a couple reasons that this is really important.
For one, a lot of people have symptoms that [01:06:00] fluctuate. On a K on a day to day and week to week and month to month basis and a lot of neurological disorders, which is the area where I did my training like Parkinson's disease and Alzheimer's disease, dementia with Lewy bodies are really characterized by these symptoms that fluctuate over the course of days and weeks and months.
And so if we don't know what a normal fluctuation looks like, we don't know how to tell the difference between what they're suffering and what someone else is suffering. But also importantly it also breaks us out of this. kind of realm of thinking that a snapshot is the system when really brains are inherently dynamic organs and are useful for coordinating behavior over multiple timescales.
And here we are pretending like this one little seven minute window we got was the brain of that patient. So I think what Russ did was really great and and really kind of opened the door for a lot of really awesome stuff that's come along afterwards. In fact, multiple groups have made
Benjamin James Kuper-Smith: so just for context, he, how much did he scan them? What exactly did he do? I can't actually
James Shine: Oh, yes. So, so Russ was
Benjamin James Kuper-Smith: to provide a rough like
James Shine: Oh, yes. [01:07:00] My apologies. So, so Russ conducted what was called the MyConnectome project. And so what Russ did was for it was about twice a week on but sometimes he could only scan himself once over the course of about a year and a half. He went into an MR scanner and had structural scans and functional scans, both resting state as well as task collected over the course of this period.
He also kept a diary, got blood draws, kept a food diary. He wrote down, you know, how he was feeling every day, whether he was in a cranky mood or a good mood and a variety of other different types of tests and microbiology tests and things of that nature. And the idea was to create this almost sort of snapshot of Russ as an entity at a particular point in time, and then drag that forward and see what the fluctuations look like.
And his data set's open. He's, he made a really cool website where you can go and poke around and look at some of the measures and how they co vary with one another. It turns out, long story short, that there's a lot of really, what we call low [01:08:00] dimensional signals that constrain the major sources of variation.
So you do see a lot of variability, but the variability kind of tends to sort of fall into one or two or three categories, depending on how you carve things up. And a few groups have analyzed the data and essentially replicated the same kind of phenomena. But importantly, then it lets you ask, well, would that kind of co variation be the same kind of thing we'd expect if someone was going through a major depressive episode, or if someone was on the way to decline into Alzheimer's or experiencing raging mania, like Donald Trump appears to be at the moment.
So we can ask these kinds of questions and have a baseline behind them, and I think that's I think a really important element for our field to take seriously. So yeah, so there's, nowadays there's heaps of groups that do really cool work on this. One of my favorite examples is Dylan Newball, who worked with Nico Dosenbach at WashU.
Was part of a, one of these dense sampling projects. Sort of, suites and then halfway through one of the scanning sort of, protocols, he broke his arm and had to get a cast on [01:09:00] his arm. And so then they kept scanning him throughout that period. And then when he got his cast off and then they could actually look and see in his resting state networks, what was happening, how are they reconfiguring as he kept his arm in this cast off.
Over time and how do they re, re reconfigure back afterwards? And it's, I think this is one of these things that is just another beautiful example of what FMRI can be used for. 'cause he's this, you know, serendipitous situation where you can have a postdoc have a broken arm. You're not supposed to go out and break their arm.
He his arm had to break serendipitously,
Benjamin James Kuper-Smith: that's a good statement to make as a PI that's our indefinite decision.
James Shine: Well, I'm not, I'm also not scanning my postdocs, but
Benjamin James Kuper-Smith: Okay.
James Shine: but yeah, in this case, we were able to learn something about the kind of functional organization of the brain through this, you know, one off event that ended up being really interesting and informative for understanding organization of the motor system and the sensory system.
And I think that's, that's really great kind of dense sampling outcome. Oh
Benjamin James Kuper-Smith: Yeah, the what I find interesting is also that I, so I talked to [01:10:00]Peter Bandettini a while ago and I, we did a bit of a like history present and future of fMRI and as the future he mentioned well, at least two but the two I can remember right now off the top of my head that he mentioned as like exciting new avenues are this kind of deep scanning or whatever you call it.
I can't remember, but there's, you know, lots and lots of scanning of the same people. And so that's an example of this. And the other one you mentioned, which you also allude or which also discuss in the clinical neuroscience part is the layer specific or layer fMRI. So. Yeah, that was interesting to me because I did a bit of predictive coding in my master's or my master's project wasn't predictive coding.
So, yeah, what's the link between hallucinations and layer fMRI?
James Shine: Yeah. So. A bunch of my grad student work was actually trying to understand why people with Parkinson's hallucinate. And so I've had this as a kind of long standing interest. A lot of people wouldn't know that people with Parkinson's had hallucinations. They're [01:11:00] kind of classically known to have this tremor, and they're quite stiff, and they're slow, and they often have a very mask like expression, but they also have actually a large proportion of people with Parkinson's, particularly in the later stages of the disease, have a lot of symptoms that don't fall into those categories.
They're they have really terrible sleep, they have trouble with their memory and their cognition. and a substantial proportion, in fact, over half in less instances hallucinate. And they see images a little bit, the way I would describe it is it's a little bit like they're cloud gazing, but not wanting to.
They kind of over interpret subtleties in the world around them visually and they'll see objects or items that aren't there. And this can be quite harrowing for people, particularly as they lose insight into the fact that these are kind of, sporadic hallucinations. So. Theoretically, there's a lot of really cool work computational work thinking about the organization of the cerebral cortex in particular, the outer layer of the brain, and how different streams of information can [01:12:00] be processing.
You can imagine evidence coming in from the sensory periphery, and what you expect to see, your prediction or your prior, can have a huge impact on how the actual evidence that you receive actually is interpreted. And so one prediction or one hypothesis about how hallucinations occur is that you have this kind of overvalued prior and undervalued evidence.
You kind of go into the world expecting to see something and then the tiniest little bit of evidence comes in and you jump it at the shadows. And the reason this is important for the layer fMRI work is that There's these different streams of information are thought to propagate in different parts of the cerebral cortex.
There's a theory called predictive coding was popularized by Rowan Ballard. And then sorry was suggested by Rowan Ballard. And then I think popularized quite a lot by Carl Fristen. In his sort of active inference framework that suggests that predictions and priors ought to exist to sort of either sort of super granular and then infragranular layers, whereas the evidence comes via the granular layers, the middle layers of the cortex.
And whenever you get a mismatch between them, you could imagine [01:13:00] A hallucinatory episode occurring and so now you can go in with advents in recording. Now we can actually really zero in on these different layers of the cerebral cortex. And the state of the art, you know, maybe sort of 10 or 15 years ago, it was really to sort of just get a cortical area.
You'd have a kind of bold response in, you know, let's say V1, but nowadays with particular types of sequences and really cool different types of statistical modeling, you can actually now refine out the different layers of the cerebral cortex to an approximation, but now make predictions about if someone's in a scanner and is having something that we would equate to a hallucinatory episode.
which is another very difficult problem to solve, which we can go to if you'd like. But if you get them in that scenario, you can make a prediction about it and you can do hypothesis driven science about particular layers having blood flow in certain situations where you see a hallucination.
And I think this is really exciting for clinical neuroscience because if we start to develop really sophisticated models of how this system actually works as a [01:14:00] whole and when it goes awry in these kinds of disorders like hallucinations. We can then start to connect them up with the animal work that can make that particular signature more or less likely.
Let's imagine that we have a prediction that a particular neuromodulator like dopamine is involved in hallucinations. There's been some recent work linking these things together. Now we can come into an animal model with optogenetics, turn on and off the levels of dopamine, let's say with a viral expression of a channelrhodopsin in something like the ventral tachymental area.
We can then turn that on and off. We could then say, does that make the signature that we saw go up or down? And then do we see that same thing in the human that is experiencing hallucinations? And I think this is a really exciting time. If I was going to hedge my bets, I would bet that the answers will end up being extremely complicated.
And I, in that I think there's multiple different ways for people with disorders of the brain and mind to hallucinate. And some of them will be related to dopamine, some probably to acetylcholine, some to noradrenaline [01:15:00] serotonin, others to EI imbalance, others to structural abnormalities, others to other kinds of complex systems level disruption.
So I don't think there'll be a kind of one size fits all answer for why people hallucinate, but I think we'll be able to come up with testable predictions that can actually move the needle forward. And that's the hope.
Benjamin James Kuper-Smith: As a way to ask better questions,
James Shine: Yeah, exactly. That's really the, I think the undercurrent of this whole paper is that neuroscience is really hard and we need to help each other to ask better questions if we're going to make our way out of this siloed kind of space we're in now. And so let's try to come up with ways that we can bridge those gaps.
And we think fMRI is really well placed, but there are many others. Electric physiology is going to be extremely helpful for a lot of these questions. We're really excited about focus ultrasound in our group. And things like Meg, like the optically pumped Meg is really exciting. There's so many opportunities for neuroimaging now that aren't just linked into bold.
[01:16:00] And what we want to see is more integration rather than just the use of one particular technique.
Benjamin James Kuper-Smith: also have an episode with Gareth Barnes about the
James Shine: Oh. Oh, great.
Benjamin James Kuper-Smith: MVGs, yeah. I will link all of that in the papers we discussed in the description, as always. Maybe I should have mentioned that earlier, anyway.
Yes, as we talked a bit in the beginning, and I'd like to expand a little bit, this is the kind of collaborative aspects of your work, your lab. Yeah, as I you know, Listen to this other interview you did.
And you mentioned there that you basically, I think you mentioned it today also that you kind of hired people with very different backgrounds from your own to learn from them. And I think in the interview you said it was like having a second postdoc phase where you could learn from your postdocs. Yeah.
Can you, I found that very interesting. Um, because for example, just as a little background, like I want to multiple, probably two longer postdocs, just because I want to have this prolonged period of just learning how [01:17:00] to get really good at what I'm interested in rather than having to do teaching or admin or any of that kind of stuff.
So it was very interesting to me to see that maybe you found some sort of way around that, getting the professor salary while still doing a postdoc um, or not. But yeah, I was just curious about that approach. Maybe can you talk a little bit about the kind of. Maybe when you started your lab what kind of what was your approach there to hiring and running the lab here?
James Shine: So when I finished my postdoc with Russ, I was part of, I was on a fellowship through the Australian government, the National Health and Medical Research Council, which gave me a couple of years of salary when I came back from the states to essentially kind of start to build my own group. And I had to think very carefully about what I wanted to achieve in that period.
And around that time You know, I had a series of really, I thought, fascinating questions to ask about how the organization of the arousal [01:18:00] system could impact whole brain dynamics and reconfigurations and kind of network like level signatures. That was a really quite, I think, novel space to be asking questions.
Up until that point, most of the work that was done on the arousal system was really down at the kind of physiological level where you would, you know, really control. this animal's system, or you know, look at this cell type under a microscope, or the network stuff was done at a really kind of abstract mathematical level.
People weren't really asking questions about exactly how the brain was embedded embedding a network. They were asking more about what the network looked like. And that's not to say there wasn't work in that space. It just wasn't the kind of main, the main thing. And I wanted to integrate those things together.
And so I had this real challenge. How do I. test these hypotheses. And and I recently started collaborating with Michael Brakespear, who's a computational neuroscientist from Australia. That was on on the advice of Russ who showed, I think, quite a lot of wisdom in, in sort of putting us in a room together.
And I had a really great experience with it. I was really terrified [01:19:00] and nothing at all about how to do computational modeling. I had no mathematics in my background at all. You know, I did, you know, first year undergraduate maths and sort of got a bit freaked out by trying to calculate determinants on matrices and kind of ran screaming towards chemistry and biology and and psychology.
But so, so, but I knew that there was a power in it. I knew that if you could understand how to do this properly, you can make explicit your hypotheses, you can make explicit your assumptions, and you can essentially then test this hypothesis in front of you, given all your assumptions, and know exactly what you put into it, and see exactly what comes out.
It was this real control that I didn't see in a lot of the other experiments that I was conducting which were much more descriptive, you know, put someone in a scanner, we'll get some imaging data, ask this question, you know, what does it look like if I threshold the network like that or something? It was very descriptive.
But I also knew I couldn't do it on my own. And so when I got my first grant, I really pitched it around trying to develop our [01:20:00] understanding of how disorders like dementia arise. Systems in the brainstem that are becoming impaired had this impact on the whole brain level. How could we know?
Well, one way we could know is we could do imaging, but another way we could know is we could do modeling. And I knew I needed to hire someone. So I looked around my network and asked around for people that had a background in In modeling that had done work in you know, mathematics or physics or engineering or something like that, that had this kind of comfort with that kind of mathematics.
And it just so happened quite serendipitously that a PhD student had just, was just about to graduate from a really excellent group at the university of Sydney. So, Eli Moola and I met, we had coffee and within about 10 minutes, we just knew that we had hit it off and it was just going to be great.
We. have very similar kinds of tastes in science and the kind of scientific experience. We're both very creative and curious scientists and really working together collaboratively. And it was a little bit like he had the kind of skills that I needed and I had a bunch of [01:21:00] neuroscience and imaging skills that he was interested in and we were able to kind of click really well.
And you know, at the time it seemed like the only thing I could have done in retrospect, I've talked to a few mentors and they were, they said it would have been a much easier option if you just, you know, went out and found someone that did what you did and then just train them up. to do what you had trained to do.
And I think there's an element of that, of me that wonders what had happened if I had done that. But I think I would have been in a far worse place. Academically, I would have had far less interesting questions. And I think I really put a lot of my success and success of my group down to just how lucky I've been to hire people like Eli.
And then subsequently Brandon Munn is another physics postdoc. And just the kinds of, you know, questions you can ask when you really give yourself into this ignorance that comes with working with experts in other areas. I just, I find it, it's almost like a an addiction. How I just, I love that feeling of being the dumbest person in the room and surrounding myself with these [01:22:00] brilliant, hardworking, smart people that can teach me something to help me understand better this problem in front of us.
And I think that mindset, I've been very fortunate to find other people that share something similar. And now the students that we have attracted to the group have, they're just, they blow me away. I mean, I feel like I'm doing an honors project with some of my grad students at the moment with how fast they're making progress.
And it's really, I think it's really about finding that right mindset. I think that really opens up opportunities for really rapid progress in a kind of complicated space like this.
Benjamin James Kuper-Smith: It just occurred to me that is actually going back to what we said earlier about being fine with, you know, being stupid or seeming stupid or not knowing questions and answers and asking people questions and that kind of stuff. I guess it's also a lot easier if you're just not from the same background as them.
James Shine: yeah.
Benjamin James Kuper-Smith: If someone is from exactly your background, then you're like, shit, I should have
It's like a physicist, you're like, [01:23:00] yeah, I mean, you should know stuff that I don't
James Shine: Yeah.
Benjamin James Kuper-Smith: If not, then what did you do?
But how do you, what, okay. Just a very practical question. How do you find the right people for this?
Because I mean, I have, I once had a supervisor who wanted to, what was. I think he wanted to do, I mean, this was like before the big AI hype, but he wanted to do a lot of more like classification kind of stuff with brain data. And wanted to get someone from the kind of computer science AI kind of world roughly.
And so again, this was before the big hype and still he was like, you know, the great people in this field are in the field. You know, they're not necessarily gonna come to you know, some random neuroscience lab, well, not random, but you know, to, to a very good neuroscience lab because they, you know, they're great at the thing because they like doing the thing.
So they continue doing the thing. So I was just curious how you and I don't know whether being in Australia here, it makes it better or worse in the [01:24:00] sense that you know, I guess in Europe it's maybe a bit more flexible to move from one country to another. But yeah, so I was just curious kind of how do you.
basically get the people that you want from different disciplines to work in your lab or work on the projects you're interested in, that kind of stuff.
James Shine: Yeah, it's a great question. It's sort of an ever evolving question because, you know, we're you know, relatively young lab been around for five or six years and just sort of figuring it out as we go. We've just actually gone through a recent hiring process. We got some great funding with some collaborators at the University of Sydney through an Australian government grant this last cycle.
And I've actually got two new postdocs that are arriving in the next couple of weeks, which we're really excited about. They were actually both international candidates so one was from the UK and the other from Canada and we'd met one of them from beforehand through human brain mapping and then the other candidate we kind of knew through the circles.
We know the group that he worked in and some [01:25:00] of the other people. So there's a little bit of an element of, kind of knowing people in the network and having the work that you do be valued at a high level such that if you put out a call, then the PIs in that group or the grad students in that group, the postdocs are aware of your work and can say, Oh yeah, I know what they do.
And I really would like to do that too. So I think there's an element of the kind of network helping you. In our case, we had lots of really great candidates and we had a really tough decision on our hands in order to to whittle that, that list down. I think we ended up with two really brilliant young scientists, and we're really excited to have them join.
But in that case it kind of took care of itself in some sense because we've been kind of actively engaged in the network. I think in terms of the broader kind of question about how do you go about creating these links and opportunities, I think you kind of said, I'll have to paraphrase, but you said something before that if you kind of put your intentions out, into the world.
It has this funny way of coming back to you. I don't mean that in a kind of hippie dippy way. I mean it in a sort of. If [01:26:00] you're talking to other people in your network about what you care about, and you're saying, man, I really want to do this kind of experiment, but I just don't have the expertise, it'll kind of hopefully trickle around to the point where some bright young student who's also really interested in neuroscience, but hasn't known where to find the application.
It was like, oftentimes what I find is people are sitting there going, man, I didn't get the you know, outcome I was looking for. Maybe I'll go work in industry, or maybe I'll go do this other thing. And if you hit them at the right time, you can, you know, you know, find a way to kind of save them from that that outcome of lovely salary and, you know, career stability and drag them into the murky depths of academia.
But, you know, so far that's worked out really well for us, and I've been really fortunate to have great collaborators that have great suggestions for people that can work with us. But I don't think there's like a magic answer either. It's really hard.
Benjamin James Kuper-Smith: Yeah, and I guess there's a lot of, I mean, yeah, I guess in a way what you want is someone who's really great in a topic but slightly disgruntled with it.
James Shine: Ha! Yeah, Yeah.
Benjamin James Kuper-Smith: You know, what's, [01:27:00] what's the change? Um, uh, Yeah.
James Shine: the scientific world go around. I think that's maybe an underappreciated point. We talk about this all the time in the lab that if you're trying to find a problem to work on, the one you'll be the most motivated to solve is the one that is annoying you the most. You know, I can't believe people do this.
Why would they ever do that to their experiment? And I say to my students, well, why don't you just go do the other thing? You think it works like that? Let's go see. Let's figure it out. If you get it right and you're correct and you find evidence for it, then you've just found your next question and it's sort of done for you.
And you also get to work in a field then that you really like and the questions are being asked in a way that you think is a more accurate reflection of reality. And I think that's a really nice way to give young scientists agency in this really complicated space.
Benjamin James Kuper-Smith: by them finding the thing that they are annoyed [01:28:00] slash interested
James Shine: resolve it. You know, don't just don't let anyone can point out something that's wrong. Okay. It takes a different type of person to say, that's wrong and I think this is the right way. And now I'm going to go about trying to collect evidence for it. And then to me there's a real, there's a real allure to conducting kind of idiosyncratic science in this way where you say, I.
I think you really have to do it in a really respectful way as well. What you don't want to do, and I think that you see too much of, particularly in the past, I don't see this as much in the communities I hang out in, but it's definitely there in some scientific communities. You see the established scientist just pooh poohing everyone's stuff.
Ah, that's crap. It doesn't work. I tried that. One time it didn't work. It's wrong. You know? And just sort of throwing down from on high all this kind of negative feedback. I don't like that. I, what I like is. What I try to encourage my [01:29:00] students to take the mindset of is, find the thing that you think is the problem, but then go away and we try to apply Dan Dennett's rules.
I don't know if you've heard these before, but he says that the rules of polite discourse in science, and in his case, philosophy, was that if you wanted to argue with someone, your job was first to actually describe the problem in such a beautiful way that they wish they thought of it themselves. Then you tell them all the things that you learned from your common perspective.
And only then are you allowed to provide any form of criticism. And I think this is actually, you know, a really great mindset to get in, right? What we're trying to do together is figure things out. And you've been trying this thing, and that works really well in some cases, but I can see this little thing that might be a problem.
I reckon if we did it this other way, we'd get somewhere a little bit further. And here's exactly how I can respectfully test that. And here's where the evidence will follow the way. And you try to follow that evidence, And see if you can come back and provide a synthesis that makes the whole field better.
I think [01:30:00] if you can come in with that constructive mindset, there's no way we're not going to make progress. That's a that's a, to me, it's not a guarantee, but it's the right way to be conducting ourselves. And it comes from a dissatisfaction, right? But it just, but it can't, it comes from dissatisfaction, but it can't be mediated through.
I think a negative confrontational approach. That's my take. Maybe people disagree with me, but I prefer to be
Benjamin James Kuper-Smith: No, I mean, I think, yeah no, I mean, I agree. I think the, that's also the whole point of I mean, it's almost by definition the case, right? It seems to me that, you know, progress only works by being able to do what the other people were able to do and then add something to it. You're obviously not going to get like a complete, you're not going to, you know, completely necessarily be able to do exactly what they did and then more.
But yeah, if you just have something that like. You know, I can do this thing that you couldn't do, but like 90 percent of what you were able to do, I can't do, then my model is just not that good, probably, I don't know, depends, maybe there's specific cases, but like [01:31:00] the general case is to have something that takes what exists and then goes beyond it, rather than just,
James Shine: Yeah. Constructively.
Benjamin James Kuper-Smith: yeah, you also mentioned somewhere I think you were quoting somewhere else, and I'm misquoting them because it was from memory.
But, there's a, there's an idea, at least, let's put it that way. Someone said just worry that your students are excited.
James Shine: Yeah. So this was, this was during the pandemic and and we would just, you know, everyone just kind of beating our heads against the computer screen on zoom all the time. And I think, you know, people were a bit stir crazy and unsure what was going on. And I was talking to a mentor Ethan Scott who kind of gave me, I think what was really great advice, which was don't worry so much about whether or not their projects every week you know, they've got the paper written this many percent, or they've got this much of the data analyzed or something.
Just try to figure out if they're motivated or not, try to figure out whether or not they're enjoying what they're doing. They feel like they're on the right course that they kind of know what the goals are and whether [01:32:00] they're heading in the right direction. and I think that it ref that reflect, they kind of lit some, a little fire under me in a way, because that's kind of what we used to do as on my sporting teams that I played on for so long.
We made sure that we all, you know, wanted to win the game that was in, you know, a foregone conclusion, but that we wanted to win in the right kind of a way. We wanted to play in a sportsman like fashion. Fashion. We wanted to get everyone involved. We wanted to play really beautiful football and things like that.
And. So you can kind of immediately then take all these lessons you learned from the past and just apply them without really much tweaking to this problem in front of you of motivating brilliant young people. And honestly it's like a magic trick. It works so well to focus more on what is driving people and what they can, what, to meet them where they are, you know?
You know, I'm having a bad time right now. My relationship's breaking up. You know, I've got this debt that I haven't been able to pay. I'm feeling really burnt out. [01:33:00] All these kinds of things just pop out. And if you didn't know about those, you'd be sitting there as a manager or a PI or something going, Man, how come that bloody paper hasn't come across my desk yet?
Why are they so slow on this side and the other? But there's a moment you ask them that question, you go, Oh, cool, they're feeling stressed right now. They should have a break. They should go off and relax. And, you know, during the pandemic, this was obviously much harder, but nowadays it's much easier to do this.
And now we're really fluid and flexible with everyone's work schedule to the point where, you know, we have bigger problems with people probably pushing a little too hard and probably working more than they should. Because they're really motivated to solve the problems because they're really kind of itchy, frustrating problems that you guys want to know what the answer is.
And so you have to almost pull back on the Accelerator a little bit and then start to ask, how do we buy time for some of those more peripheral interests? How do we ensure that people are, you know, taking time to balance work and life that they're, you know, getting the chance to, you know, explore and not just [01:34:00] stay locked into the question that is right in front of them.
Because I think those are really deeply important parts of having a really satisfying scientific career. So, but, you know, I'll take those problems any day over the, Problems of getting annoyed at people because they haven't got some outcome done by time X. I think, I just think that old kind of corporate language doesn't really mesh with really kind of curiosity driven modern science.
And the motivational technique that Ethan suggested really just to me does. And I, so that's really a core part of our philosophy in my group.
Benjamin James Kuper-Smith: Sounds like a pretty
James Shine: We have a good time. We get, we're a bit silly, but that's okay. We don't take ourselves very seriously. Okay.
Benjamin James Kuper-Smith: Yeah, it's the Shine Lab. It's a little bit silly, but you know, having fun. Um, uh,
Yeah. I mean, so we're talking about being excited. Uh, One thing that I find sometimes a little bit difficult in science is all the stuff you have to do that is not science that helps you. that allows you to do the [01:35:00] science.
Right now I'm writing a, not grant proposal, but fellowship proposal, I guess is the correct term. So it's just my salary basically. Yeah, I mean, this is going to be very vague questions in this sense. But I remember you also mentioning that. And so I was just curious whether we could talk with that for a few minutes and Yeah maybe as a broad question, kind of what's your approach to writing a basically begging for money in various forms.
Um, I mean, that's what it is, right? But um,
James Shine: way to think about it.
Benjamin James Kuper-Smith: Yeah.
James Shine: I use the I use grants and fellowships as a way to help figure out if I know what I'm trying to do. I use it as a kind of test for myself. And what I say to myself is the extent to which I really can communicate why I'm doing what I'm doing. Why it's important, what the question is, what the gap is, why my approach needs to happen, why me, why now, all these why [01:36:00] questions.
If I can put them down on paper in a way that's still interesting for someone to read, then maybe that I'm someone that could deserve this money this time. And there are times when we all sit down to write grants and we'll realize that the why just isn't there. And we just sort of stop. We just say, no, it's not really the time for this.
That, that idea is not quite ready, or we really need this other technological advance for that to happen, or man, why would we not just go get this data set? Why do we need to bother getting extra money for this thing? It's already out there. We just go ask the question. And I think it's important to do that if you can.
Now, oftentimes in academia, it's not. We're not that fortunate, right? We need to sing for our supper, as you're saying, and essentially get funding in order to stick around a bit longer. So then in that case, I would say, try to see if you can really nail down why it is what you're doing, what you're doing and what you would, you know, what you would just desperately love to do if someone would just [01:37:00] let you.
And I think if you can, this, you know, this is why it comes back to the kind of motivation being the key thing. If you figure out why you're doing what you're doing and you can really clearly state that I think it shows up on a page, man. I think when you read a grant where someone's just like, dude, this is awesome.
Let me do it. Okay. I cannot wait to answer this question. It's like a scientist. We can kind of resonate with that. Honestly, I don't know the first thing about, you know, you know, retinal surgery or renal grafts or, you know, some kind of geothermal event or something. All these scientific grants I have to read and assess.
But you see the story behind the detail. You're looking for that, you know, right place, right time, right question. Kind of a constellation. So, you know, treat that as a challenge, right? And then, you know, there's side effects too. If you get that right, then you've now gotten a bit better at communicating your work to a broad audience.
You've gotten a bit better at synthesizing a complex problem down into a set of paragraphs that constantly you know, really get to the point quickly. [01:38:00] And these are the kinds of things that you see in, you know, high impact neuroscience papers in big journals. They're the kinds of skills you need to conduct yourself in academia.
So I think I view them as sort of soft skills that you can practice and learn while also learning something more about who you are and what you're trying to do. That's in the best case scenario.
Benjamin James Kuper-Smith: I mean, I think this reframing of these kinds of things, I think is something that I need to work on or that would help me because I think I, you know, I generally see it as just you know, Not begging for money, but there's this kind of like justifying yourself that I just really don't, I just don't have that.
And I just don't care. Like just, can I just please do it? I don't know. So
James Shine: you ever read Kurt Vonnegut?
Benjamin James Kuper-Smith: I've read something. Yeah. What did I read? Cat's cradle, I
James Shine: Cast
Benjamin James Kuper-Smith: I don't think I'm sure I finished it.
James Shine: So Kurt Vonnegut had this really great way, I think, of playing with framing where he could take the most awful, horrible stuff that happens in the world and find cute ways to flip it around so that it could be something different. [01:39:00] And I think if you read a lot of his work, you can kind of learn that trick that it's up to you to determine how you handle what's the card that's, the card that'll be dealt to you.
Yeah. Very easy to say as a kind of, you know, privileged second generation academic white male sitting in a, you know, developed country. And so I'm not pretending that other people have got things that are just frameshifts away from being okay. There are awful things that happen in the world. But to the extent that you can perform one of those frameshifts and laugh at yourself a little bit and laugh at the situation, it can help to kind of, you know, Diffuse, acknowledge the tension, right?
Like you're right. It sucks that we have to sing for our supper. In some cultures in the world, once you become a scientist and get a PhD, you just get a salary for the rest of your career. And it's, you then have to every now and again kind of demonstrate that what you're doing is, you know, consistent with the kinds of scientific advances.
But that's different than saying, please can I stay? And we've got this weird thing happening in the Western world now where we're [01:40:00] kind of treating science like it's like a, you know, like where people are hedge fund managers or something and they're trying to say, Oh, is this the thing that'll hit it big?
I just don't think that's how science works. I think science is a, an iterative process of curiosity driven kind of reduction in uncertainty. And who knows where the cool answers will come from. And you heard me talking, you know, in this podcast about optogenetics a number of times. It's a really amazing tool.
If you listen to Carl Dyseroth, who's one of the founders of optogenetics, The only reason that it exists is that some nerd back in the 70s thought it was really cool that these little things fluoresced, and he wanted to understand how they worked. And because he figured out the mechanism by which they fluoresced, later on, they could use that technology to make the tools that you could turn neurons on and off with light.
And no way that you could have gone back to the 70s and had that scientist understanding the fluorescing in these little these little creatures. There's no way that you could have ever guessed that would let you control neurons. The person would be laughed out of the gym if you did that.
And so I think we're misguiding [01:41:00] ourselves if we think that science has to have this clear out, you know, path to answers. It's not like that. It's a process that we're all engaging in together. Anyway, acknowledging it's good. Laugh at yourself and find a way to reframe it so that it can be fun.
Benjamin James Kuper-Smith: Yeah, and your reframe is then to use it as an opportunity to clarify why you're doing what you're doing, whether you should even be doing it almost,
James Shine: know yourself better, know why you're doing what you're doing, and then all that motivation is just sitting there for you to tap into whenever you get, have a bad day, or get a rejection, or you know, get frustrated with the weight of the world, and you just go, you know what, at least I get to do this, and I know exactly what I'm trying to do, and I need this person to help me do it, and I'm going to go find that person, and they'll help me be better.
Benjamin James Kuper-Smith: Okay. Um, yeah, so at the end of each episode, I ask my guests the same three questions. The first one is, what's a book or paper you think more people should read? This could be old, new, famous, unknown. Just anything you, you would recommend people to read.
James Shine: [01:42:00] Yeah, I was actually asked this question recently by a journalist at the Transmitter, the Simons Foundation's online magazine, and I thought about this a lot, and I ended up sort of settling on a really beautiful paper that was written by Paul Czizek and Giovanni Pizzullo um, which is, let me see if I can find the exact title for you, hold on.
Benjamin James Kuper-Smith: I'll also, again, put a link in the description.
James Shine: so it's Navigating the Affordance Landscape, Feedback Control as a Process Model of Behavior and Cognition. I'll send you a link. So it's a really beautiful piece of work where Paul and Giovanni were trying to think about the organization of a brain as a whole and the kinds of problems that our brain's trying to solve and what that architecture ought to look like.
And Paul is a a movement specialist works on motor problems in macaques, but also has a deep passion for evolutionary neuroscience. And Giovanni was always a really brilliant [01:43:00] computational neuroscientist and one of these Kind of people that's fluent in seven or eight different scientific fields and so just write some most beautiful papers.
And this paper I read it when I was a kind of late grad student, early postdoc, and it really just opened my eyes to the kinds of shapes that I think the answers of how a brain works and fails will look. It's a really beautiful paper. I would strongly recommend it.
Benjamin James Kuper-Smith: Okay, cool. And as always, it follows the logic of either something I haven't heard about at all, or something I've heard about a million times, and and this is one of the ways I haven't heard about
James Shine: Oh, great. You're in for a treat.
Benjamin James Kuper-Smith: Okay. Um, Second question is something you wish you'd learned sooner.
This can be from your work life, from your private life, whatever you want. Just something that you think if you'd learned that sooner that would have helped you out, and maybe also How you learnt it, or kind of what you did about it.
James Shine: Yeah. I think early on in my academic career when I was a grad student. I felt this [01:44:00] weird, I felt the typical imposter syndrome that everyone feels, and you know, I still have that same tickle in the back of my mind, but I've sort of tried to reframe that as a growth mindset, you know, it's anytime you feel that it's just a chance for you to learn something new.
But I think the one that, that I really, The most disappointed in myself for letting Ferment was this notion that if you are different, you don't have anything to add. And I felt as a, as someone with a clinical background, every time I'd go to a conference, I felt like I was, it's almost unworthy to talk to people because I couldn't talk about neuroimaging or I couldn't talk about computational science, or I couldn't talk about disorders of this, that and the other.
And I just, I don't know. It was awful. It kind of ruined a couple conferences for me. I came away from them feeling. Really small and insignificant and just like I wasn't part of something. And the further I've gone on in my career, the more I've realized that the individual differences that make you unique are awesome and something you should absolutely foster and hold onto and bring out.
Because [01:45:00] now when I come to a problem in computational modeling, but I, I say, Oh man, that parameter actually reminds me a bit of the thalamus. People go, what's that? You know, they hadn't really thought of it because they've been so busy dealing with the really complicated mathematics. But now my unique perspective can add a whole new line of questions that can then bring up interesting new avenues and inroads.
And so I think your differences make are actually the thing that gives you a really satisfying unique thing to say, and you shouldn't be afraid to bring them out. That's what I wish I learned earlier.
Benjamin James Kuper-Smith: Yeah, I mean, I guess that's how unique contributions come about, right? By being unique and a bit different.
James Shine: Yeah.
Benjamin James Kuper-Smith: Uh, Final question. Yeah, I started my postdoc two months ago, roughly. Any advice for people in this kind of phase? So like late PhD, early postdoc phase?
James Shine: Yeah. Kurt Vonnegut, who we brought up before, he had an uncle that uh, that used to say to him, whenever you find [01:46:00] yourself in a position, that's just fantastic and amazing, you need to say, you need to stop everything and say, if this isn't nice, I don't know what is. And I think take advantage of the time you're in as a postdoc.
It's an absolutely amazing time in the life of a scientist. It's not without its stresses. It's. It's really fraught with opportunities for feeling low and small and irrelevant and dumb and all these other things. But if you can appreciate it for the fact that you have the most amount of academic freedom you will have for probably your entire career, and that you could literally go anywhere you wanted intellectually, so long as you can justify it, you've got the reins and it's your chance to drive the car wherever you want to go.
That's not to say you should go driving it off a cliff. You should try to drive it to a really great location, but this is where real progress comes from. People that are trained in how to think like a scientist have interests and dissatisfactions with the way [01:47:00] that the world works right now, but have the capability to find out how to actually resolve it.
And you're in a position where you're not tied down. You're not stuck doing, you know, Writing this course or teaching this, you know, semester, and you're not, you're also not stuck down intellectually in the sense that if you choose to embrace a new field, that's on you. It's on you to actually do that. So You know, to me, a post, the postdoc is the time, that's when you get to be the most pure version of science, scientists that we get to be in the modern version of academia, unless you can do well enough that you get fellowships and then you can start to buy back some of that freedom.
And so, yeah, enjoy it while it lasts and yeah, and notice when you're enjoying what you're doing.
Benjamin James Kuper-Smith: Okay, yeah, so drive to the cliff and enjoy the nice view, but don't
James Shine: I drive off it? Yeah. Unless you've got a
Benjamin James Kuper-Smith: yeah.
James Shine: Yeah
Benjamin James Kuper-Smith: um,
yeah, [01:48:00] I think that was it for me, so thank you very much.
James Shine: great. Thanks for having me.