83. Rachel Bedder: Rumination, teaching without grades, and managing yourself as a PhD student

Rachel Bedder is a postdoc with Yael Niv at Princeton. In this conversation, we talk about her research on rumination and repetitive negative thinking (in the context of a partially observable Markov decision process), her work as a curator, why she enjoys teaching without grades, how to manage yourself as a PhD student, 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: Teaching maths in prison
0:06:40: Teaching without grades
0:15:42: Working as a full-time research assistant (after BSc) and dealing with lots of rejections
0:25:51: How Rachel ended up doing a postdoc with Yael Niv
0:32:08: Discussing Rachel's conference proceedings 'Modelling Rumination as a State-Inference Process' (featuring partially observable Markov decision processes)
0:56:49: Rachel's background in art and curation
1:10:58: How to not turn hobbies into a stressful thing you need to get done
1:14:46: A book or paper more people should read
1:16:47: Something Rachel wishes she'd learnt sooner
1:19:05: Advice for PhD students/postdocs, with a twist: 5 tips for managing yourself during a PhD

Podcast links

• Website: https://geni.us/bjks-pod
• Twitter: https://geni.us/bjks-pod-twt

Rachel's links

• Website: https://geni.us/bedder-web
• Google Scholar: https://geni.us/bedder-scholar
• Twitter: https://geni.us/bedder-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

Episodes with Matthias Stangl and Toby Wise about postdoc jobs & fellowships:
https://geni.us/bjks-wise-postdoc
https://geni.us/bjks-postdoc-stangl

Episode with Paul Smaldino on modelling social behaviour, and with Eiko Fried on theories in psychology
https://geni.us/bjks-smaldino_2
https://geni.us/bjks-fried

POMDPs: https://en.wikipedia.org/wiki/Partially_observable_Markov_decision_process

Dear World Project: https://engagement.fil.ion.ucl.ac.uk/projects/dear-world-project/

5 tips for managing yourself during a PhD: https://www.rachelbedder.com/phdtips

Scientific virtues (including stupidity): https://slimemoldtimemold.com/2022/02/10/the-scientific-virtues/

Bedder, Pisupati & Niv (2023) Modelling Rumination as a State-Inference Process. https://doi.org/10.31234/osf.io/tfjqn
Burkeman (2021). Four thousand weeks: Time management for mortals.
McCullers (1940). The Heart is a Lonely Hunter.
Montague, Dolan, Friston & Dayan (2012). Computational psychiatry. Trends in cognitive sciences. 

Transcript

[This is an automated transcript that contains many errors]

Benjamin James Kuper-Smith: [00:00:00] I thought we could, uh, start with one of our commonalities. So when, you know, looking through CV and that kind of stuff, I realized actually we have quite a lot of things in common, um, to name just a few. Uh, we both worked as a research assistant before doing an MSc at UCL in neuroscience. Uh, we both ended a PhD in losses and gains in decision making. 
 
 

And we also both went to prison. So my question is what's your involvement with prison? 
 
 

Rachel Bedder: So I've been teaching in prison for one semester here, and I've been doing, um, tutoring for much longer. I'll say something that you won't hear from academics that often, which is I love teaching. Um, really feel particularly passionate about teaching foundational topics. Now, whether that be in computational modeling or even basic math, which is what I teach, um, in East Jersey. I think these kind of classes are where we really set our attitude to particular parts of education. So whether that be an introduction to [00:01:00] modelling, when we're first going to learn about the landscape of things and how you can engage with them. And mathematics is one that's particularly important because lots of people experience sort of math anxiety. And as someone who puts a lot of, effort into teaching, I think, and especially into learning better ways to improve myself as a teacher, you want to seek out experiences where you can have the most value as an educator. So one thing about working in prison is you're helping people to empower themselves, which is really a rewarding thing to do with your teaching, whether that be getting a qualification or just having a good experience engaging with learning and education generally. 
 
 

Benjamin James Kuper-Smith: How did, I mean, kind of, how did that start or why? I mean, you kind of answered it maybe a little bit, but why teach in prison? I mean, how did, how did you get to do this particular teaching gig, I guess? 
 
 

Rachel Bedder: So Princeton has a really great [00:02:00] program, which is integrated with this broader prison teaching in New Jersey. So they offered opportunities to go in and tutor initially, which I thought was a really exciting way to work on my teaching skills. Um, and the more, more I did this, the more. Um, the more you start feeling connected to the educational journey of these students and admiring what they're trying to do and, you know, in difficult circumstances, there's a lot of constraints, obviously, in teaching and learning in an incarcerated context. 
 
 

So it's just something I've been really interested in pushing forwards and working on. 
 
 

Benjamin James Kuper-Smith: It sounds like if you go to prison in Princeton, it might be good because you get good teachers then through the university, but, um, Okay. 
 
 

Rachel Bedder: providing a good teaching experience for sure. 
 
 

Benjamin James Kuper-Smith: But yeah, if you don't get into Princeton, there's always a way. Um, I mean, so what I'm curious, like, what kind of teaching is it? I mean, so you mentioned maths, but are we talking, going through stuff that many people learn in [00:03:00] school? 
 
 

Is it, is it like more like university level math? Like kind of, what kind of, 
 
 

Rachel Bedder: So I teach what's called pre college math. So it is a lot of things, foundational topics that you might do at the end of high school. It's to prepare people for taking the full math courses for an associate degree. So a lot of the stuff, you know, I haven't thought about for, you know, 20 years, 15 years almost. 
 
 

You know, how do I divide fractions? Like I had to re go back and be like, how do I do that? Um, cause we're so used to using a calculator, right? But they don't use calculators on these pre college style exams. And it's a really experience because I feel that well, I subscribe to this mantra that if you can't teach something to someone else, you don't fully understand it yourself. 
 
 

So going back to these foundations is quite fun and exciting for me as well to be like, why does that work like that? Do I know? you know, correcting yourself in that space. 
 
 

Benjamin James Kuper-Smith: tell me a little bit about the. pupils. Um, are they kind of what, what kind of [00:04:00] crimes did they commit? And I was, I'm curious, like, how are they like super motivated because it's like a way out or is it just something they have to do as part of the, I don't know, the part of their showing that they're trying to reintegrate into society or something like, yeah, I'm curious, kind of, I guess it kind of what I'm trying to get at is like what it's like teaching there, if there's any, if there's any differences or it's just the same thing with a bit more security around. 
 
 

Rachel Bedder: So primarily I think of them as students. They are my students and the fact that they're in a particular context for whatever reason is somewhat irrelevant to my, my interactions with them. So think about them in that way. The main way you are bumping up against the fact they're in a different context is just things like. Access, like for example, they can't send me an email midweek to say, I don't understand something on the assignment. So you are quite constrained to the actual time you have with them. And you've got to build everything in that time. And you know, [00:05:00] sometimes, uh, people are late for a particular reason. They couldn't get up from their, um, units in time. So the thing they always tell you is you have to be flexible. you have to just work with what you have sometimes. 
 
 

Benjamin James Kuper-Smith: Okay, yeah, it's just curious because I'm just kind of comparing this a little bit to, I mean, I, my PhD collected some data in prison, and in hindsight, I was kind of glad that I didn't know exactly what these people had done, because I didn't realize that like three quarters of them had killed someone, and that I was in the room with them by myself. 
 
 

And, uh, you know, obviously I knew like, There's very little chance of anything happening. Um, I'm also not exactly tiny, so I knew like there wasn't much of a chance, but I don't know. In hindsight I was like, some of these people did some really messed up stuff, and, but you can just kind of teach without thinking about, I mean, I guess you just get used to it also, if you go there repeatedly, right? 
 
 

I was just there for two, three days. 
 
 

Rachel Bedder: Yeah, you're, you're often going into parts of [00:06:00] the complex, which are educational specific. So, you know, people can't just wander in there randomly. They know exactly who's in there for what classes at what time. And you really can forget about it sometimes once you're in the classroom and you're just talking about exams and studying and all these things. 
 
 

And it really becomes an educational setting. Just like, um, working in a classroom at Princeton, for example. I, you know, they're, they're motivated to learn. They, often they're going for qualifications. Um, and you know, just want to, they just want to get the math done and learn. And you're like, great, I'm here to teach it. 
 
 

So that's really the whole context of being in the classroom with them. 
 
 

Benjamin James Kuper-Smith: Uh, you mentioned before we started recording, teaching without grades. Um, that's, you know, not exactly related to the prison stuff, but it's related to teaching in general. Uh, so what's, what's teaching without grades? Why do you like it? 
 
 

Rachel Bedder: Teaching without grades is a really exciting methodology that's becoming more and more popular, um, both at university settings and [00:07:00] also in, um, um, high school and things. So the, the idea is you try and motivate people to learn whatever is within your syllabus without using a grade as the ultimate reward. Obviously, sometimes you are constrained and you have to give people some kind of grade at the end. But what we've done, um, in a class I worked on with the ILN at Princeton called Animal Learning to Changing People's Minds, it's about reinforcement learning and other aspects of learning as well, had them set their own goals for the course. So these goals can be things like, I want to. program a computational model by myself. Now often someone who might set a goal like that will be someone who's never touched. before. And this is really exciting because they can look at the class and be like, what do I really want to get out of this? 
 
 

What will push me forward as a learner and as a student? And then they kind of assess themselves against that grade, um, or that goal they've made for themselves. And the idea is not that you will only be able to assess yourself as having an A, and [00:08:00] you assess them together. You talk about their ultimate grade together. achieving that, but it's like, how much effort did I put into that? How hard did I work at that? So you're really engaging people with the process of learning, which is really exciting to do. And I think a really refreshing for the students because, um, a lot of people there are, as understandably, very, very concerned with their grades things. 
 
 

And that can often get in the way of being the best possible learner and an experience. If you're thinking, how am I going to fit in this all for the exam and stuff? While you're listening to a lecture, you're not absorbing information in the same way. And the reason I'm really excited about it is, um, for two reasons, actually. 
 
 

One of them, again, it means as a teacher, I'm You can add so much value because you're not just always talking about like, how are we going to pass this test? What's going to be on the next test? You're like, well, what are you trying to understand? And how does that relate to your broader educational goals? secondly, I think it's such a exciting opportunity while we're. [00:09:00] You know, we're all scrambling to adjust our syllabuses based on large language models, right? Cause we know students can just do their homework a lot of the times with these. Um, and people are thinking of ways to do that. Oh, you have to, you can do it, but you have to submit your entire chat GPT history to check you didn't ask too specific a question. Um, or you have to sign a statement saying how you used it. But if you're using a methodology where someone is engaged with learning the topic. Um, then they're going to use these large language models in a way that is, um, honest to that and things. they might use it to look up an equation, but the only value in doing that is, is that helping me learn more rather than is that helping me write something down on a piece of coursework? So I think it's a really exciting opportunity broadly in university education to start embracing these methodologies in the current climate of these, you know, these powerful language models. 
 
 

Benjamin James Kuper-Smith: Okay, I never put it into context. It's funny, I guess I [00:10:00] have such a, I guess I never really cared about grades much. But like for me, I don't think I would have ever used them just because it's like, what the, I mean, why? I don't know. Um, but. Okay, so that is at university level also a massive problem because people are just trying to get the best grade and that's kind of the main thing people care about. 
 
 

Or 
 
 

Rachel Bedder: It's a, that's not to say everyone uses them to start with, but that's to say, I think they can be a problem the incentives are, you need to do the best you can on all these tests. Now, if that's the incentive, it incentivizes using these things. If the incentives are more about learning and growing as a student. Um, and you're embracing new topics, because you take a risk every time you take a class which is outside your skill set, right? You might learn a lot from it, you might ultimately get a lower grade. But if you're more engaging [00:11:00] people in this practice of what will I get out of this class, then you can take more risks. 
 
 

Benjamin James Kuper-Smith: I'm assuming this is more class sizes. I'm kind of curious on the practical side, how you coordinate a bunch of people who are all kind of trying to do different things within the same topic. Uh, yeah, how does that work? 
 
 

Rachel Bedder: Yeah, interesting. So I I did it with a class size of students, I think. And it is, it is time intensive from the teaching side. You have these short meetings with people often, or at least the way we did it, to check in with them, talk about their grades. Um, and it often relies on you giving a lot of feedback on assignments. So you don't return the assignment and say, well done, A. You say, this bit was really great. You could work on this bit a bit more. And the exciting thing is the students actually read the feedback. Because they're not just looking for the grade, they're looking for like, how can I improve and what did I do well? If one of their goals is, you know, specific to that assignment. people have done it, on larger scales. I'm not sure exactly of the [00:12:00] logistics of that. In all these cases, you have to think carefully about how the class size and the topic are going to interact. I think, I think it is scalable. Yeah. is more manageable with a smaller group, but it can scale. 
 
 

Benjamin James Kuper-Smith: Is it offered in many places? I don't know much about it, but for me, it's, it's interesting just because I, I've never cared about grades, whether I was in school or university or anything, like having to assign grades or just being involved in the whole system for me is a solid reason not to want to be a professor or something like that, because I just can't be bothered to deal with that kind of stuff. 
 
 

Uh, so I'm curious, like how common is it? Or, uh, I've certainly never heard of it really in the way you've described it before. 
 
 

Rachel Bedder: Yeah, so I think it's, um, certain professors and teachers, um, are starting to use it, but I would not say it's common on math at all. Um, you do have to really, you know, build your curriculum around it, which means shifting a lot of things in a class you might already have that's running really well and you don't want to change. [00:13:00] haven't seen any examples of institutions, whether that be schools or universities, trying to adopt it. Um, you know, that would be a huge paradigm shift in all of our educational systems, but I would say most teaching spaces, when I've been talking to other educators, they've at least heard of it now, even if they haven't used it themselves. 
 
 

Um, 
 
 

Benjamin James Kuper-Smith: saying like, hey, can I try this out and then university goes, uh, fine. And I 
 
 

Rachel Bedder: like, you know, it goes, um, it probably more like, no, please. All right. You can try it. And then, you know, you look at the course evals and the course evals, um, a great often students have a really good time in the class and they learn a lot. So, you know, the proof's in the pudding as it were. 
 
 

Benjamin James Kuper-Smith: would imagine that those are also quite popular because, you know, you can just actually focus on learning stuff rather than [00:14:00] try and learn something. But also, you know, you can just do the one thing rather than two things at the same time. 
 
 

Rachel Bedder: Yeah, I think so. It's the, the, when we've, we taught it, they're very popular courses and we've had to expand them based on demand. I will say that, um, the added factor is Yael is an amazing teacher in all contexts and people always want to sign up for her classes. So, yeah. 
 
 

Benjamin James Kuper-Smith: Okay. 
 
 

Rachel Bedder: but I definitely think students enjoyed it a lot based on my interactions with them. 
 
 

Benjamin James Kuper-Smith: And so it sounds like that's what you want to continue doing, or do you, are there benefits to a grading system, uh, in your experience? 
 
 

Rachel Bedder: Um, for me, it really depends on, on the setting. Some people, for example, in the, um, prison teaching context, you know, having a qualification is very, very valuable to people and they might be more grades orientated and the style of [00:15:00] the classes we're teaching, it can be useful to actually be more focused on how do we pass this exams and things, especially if someone might've had a, um, Not positive educational experiences in the past. 
 
 

So they, they might, um, served better by a traditional educational context. Um, so you have to kind of meet whatever class you have where they are and think of it works, but I would love to do more of it and things. And it's something you're doing when you're, um, mentoring, uh, undergrad students or master's students is you're not talking about. passing with their thesis all the time, they're talking about what exciting things can you learn and like what new skills do you want to develop and things. Yeah, so I would love to do more of it. 
 
 

Benjamin James Kuper-Smith: Yeah. So, I mean, as I mentioned, uh, right in the beginning, we both work as a research assistant. Mine was, I think, slightly more accidental than yours sounds like, or at least mine was shorter and scattered over a few places. You know, for many people, it's a question like, do you do a master's or a PhD or after your bachelor's or do you work as a research [00:16:00] assistant? 
 
 

So I'm curious, I'd like to talk for a few minutes at least about that kind of like what it's like, who it might be a good idea for. Um, so maybe, uh, why, maybe why did you start, uh, why did you do a research assistant after your bachelor's? 
 
 

Rachel Bedder: So the main reason I did a research assistant job, 
 
 

Benjamin James Kuper-Smith: Sorry, just to clarify, this also means, yeah, like, a full time job that's paid, we're not talking about, like, doing it, like, a few hours a week on the side or something, we're talking about, like, working as a full time job. 
 
 

Rachel Bedder: yeah, and in fact I started as an admin assistant and, um, was promoted to research assistant. So I can tell you loads of amazing things about that job and what I learned about it. But my number one reason for doing it was money So, um, so I'm from there, you can probably tell by my, my accent, or I hope you still can tell by my accent, is that I'm from the uk um, and master's degrees you have to pay for in the uk. 
 
 

And after I finished my undergraduate degree, I knew I wanted to do a PhD and I knew that I needed a Master's to [00:17:00] do that, I didn't have 10 grand to drop. I applied for this admin assistant role and then it became a research assistant role and I got, I got so much out of it. It was a study that was looking at quality of life in people with, um, moderate to severe dementia. 
 
 

So often people who are in care settings, nursing homes, um, care homes. And that's a huge organizational feat. To work in these, in these contexts, getting into the care home, interviewing people, um, interviewing people with dementia and cognitive impairments is a challenge in itself. Um, so I got a lot out of it in understanding how a research study of this size can work, some of the things that need to be put in place for it to be as smooth as possible for the researchers. I would recommend these kind of jobs and of course the recommendation depends on the, the country because we all have a, you know, undergrad to PhD system, which works slightly different. So I would recommend it for people [00:18:00] who might not be sure if research is for them or not. It's a good way to find out if you actually enjoy the nitty gritty day to day of running a study. Because science, you know, it's cool and fun and exciting to talk about hypotheses and all these things, but. Are you happy with filling in an Excel spreadsheet and debugging code for hours on end? Like this is the nuts and bolts of science, right? And it's good to know how these things work before you commit to a PhD of several years. I think of the, you might say unpleasant things about academia, people have different opinions on this, but, um, the, in the spirit of, you know, sharing the hidden curriculum. Being a known quantity in academia is extremely useful when you're doing things like applying for grad school and things, um, and taking an RA role can be another way to improve your network, be known by more people, be known at a different university. So I'm not endorsing that necessarily as a way we should be running academia, but I think it is something that people don't [00:19:00] realize often, um, when they're coming from their undergraduate degree. 
 
 

Benjamin James Kuper-Smith: Yeah, I mean, that's kind of, uh, it's kind of interesting that Even though we came at it from slightly different positions, many of the same things were true for me. And for me, it was more, I knew I was going to do my master's at UCL, and then I had like the sum and I was like, I don't, I should earn some money, otherwise I'll be very poor next year. 
 
 

So I just found this like one position I could apply for, like a summer, summer internship. And then as I was about to leave for my master's, I said, like, I know you're leaving, but if you're, if you weren't, there's a research assistant position, just open up and you could kind of. Have it or apply for it and you have very good chances. 
 
 

And for me, it was like, I was always like, well, if I'm applying for PhD, you know, I just don't have that great of a CV in a sense. It's like, I've done a little bit of things here and there. I've got like a little scholarship here and there maybe, but like nothing. That's really gonna go like, oh, that's amazing. 
 
 

You're like super competitive for like a amazing, you know, for competitive university. And so it was kind of cool that, you [00:20:00] know, I had. By the time I was in my master's, I, you know, presented at a conference, you know, and that was kind of cool. Uh, done a few different things that you can put on your CV, you, yeah, experienced the whole kind of setting of just doing research on a daily basis. 
 
 

And so in that sense, I think it can be very beneficial. Um, 
 
 

Rachel Bedder: So I will say one difference, important difference based on that story is I applied for, I think, 84 jobs before I got this 
 
 

Benjamin James Kuper-Smith: Okay. Yeah. 
 
 

Rachel Bedder: the job, I was taking the train back home afterwards. And when I got off the train, I walked to the opposite side of the carriage. And you know, in UK, these train carriages are really, really long. I got to the opposite side and some woman leant over and said, well done. So I must've been extremely loud and happy to have gotten that job this woman on the other end of the car too. And I was like, cool, thank you.[00:21:00]  
 
 

Benjamin James Kuper-Smith: Yeah. 
 
 

Rachel Bedder: I was over the moon. I was absolutely over the moon to have my first kind of official research job. So, 
 
 

Benjamin James Kuper-Smith: I mean, it was also, this was only like half a year and then I spent like two months somewhere else, randomly found a research position. That was two months. Uh, and then I, I'd already found like an internship for the summer at two months in the middle and just found like a research assistant position. 
 
 

That was two months for some random reason. So most, yeah, lots of coincidences, but your, your process sounds more like me trying to get a PhD, which was very, uh, involved in lots of. Lots of applications. I mean, how was that applying for all of those applications? I mean, was it the kind of thing where you, you know, just fire off the same application to everything? 
 
 

Did you really put lots of work into it? Um, kind of, what was that like? 
 
 

Rachel Bedder: every single one, now this was a long time ago now, but the way I remember it is, every single job I applied for, I [00:22:00] then spent a day and a half imagining being in that job, and where I would live, and where I would get coffee every morning, and what my apartment would look like. This, this is not good for your mental health, don't do this. 
 
 

You should, you should submit an application and forget about it. I think almost every job I applied for I would have taken. When I got to the interview stage for a few of them, was subsequently offered the job. I chose not to take them, based on the um, based on the interview and the measure you get of the place or the way you're spoken to by the interviewees. 
 
 

Um, you might notice I have what I hope is a minor lisp, but a bit of a lisp. Um, and 
 
 

Benjamin James Kuper-Smith: I haven't noticed, but okay. 
 
 

Rachel Bedder: very kind of you. Um, interview, Um, they, I think it was for something to do with asthma and they kept correcting how I say exasperated and they kept, every time I said it, they would get me to repeat it back correctly to them. 
 
 

And I was like, I don't want to work here. 
 
 

Benjamin James Kuper-Smith: Yeah. Yeah. 
 
 

Rachel Bedder: And I think it was the first job I was offered. And I was like, Oh, I'm going to [00:23:00] have to turn this down despite the fact I really need a job. Um, I was bartending at the time, so I was financially okay. So I was able to turn down that job. 
 
 

Benjamin James Kuper-Smith: Okay. So it also wasn't 83 rejections in a row. 
 
 

Rachel Bedder: No, I would 
 
 

Benjamin James Kuper-Smith: Cause that sounds rough. 
 
 

Rachel Bedder: 75. 
 
 

Benjamin James Kuper-Smith: Well, you got a lot better then. Um, 
 
 

Rachel Bedder: Yeah. I hope so. 
 
 

Benjamin James Kuper-Smith: But was this all for research jobs or was it? I'd be surprised if there even are like 75 to 80 research positions vaguely in your research area in a given few months. 
 
 

Rachel Bedder: Oh, I would have taken any research area, uh, which is probably the reason I was rejected from a lot of them. Uh, but also I applied for jobs in the art space. So my, my undergraduate degree is actually in art and psychology. So I was also looking at kind of, um, gallery assistants, um, and things like 
 
 

Benjamin James Kuper-Smith: Hmm, okay. Uh, Any advice for people applying for lots of stuff and getting lots of rejections? 
 
 

Rachel Bedder: [00:24:00] Um, any advice? I think. For every application. You do really need to treat it like it is a job that you want when you write the application. People can smell a generic cover letter from a mile off, right? as I said before, as soon as You submit that, just forget about it and get on to the next one, because it is brutal, um, applying for jobs at RA level, or applying for PhDs, and I guess, from my experience talking to other people, for faculty roles as well, it is brutal, so you, you know, you have to just keep plugging away, um, and it will work out for you eventually, one way or another, you'll find something. 
 
 

Benjamin James Kuper-Smith: Yeah, I mean, I agree with, uh, basically when I I mean, it's a slightly longer story, but I, first I had a PhD, then I stopped that, and then I applied for other PhD positions. Um, first time around was very smooth, and then it was very much not smooth after that, but [00:25:00] the, I had a kind of similar thing where I kept applying for things, and often I'd been in contact with a potential PhD supervisor before, sometimes even spoke to them, and they were all like very encouraging, and so I, you know, wrote my application, then I'd spent like, you know, days reading their papers, and sometimes I'd get invited to the interview then. 
 
 

That stings. Um, it's like, ah, I like, built up this like, fantasy of what I was gonna do and what we were gonna do in the PhD, and then it just falls apart, but, I mean, to some extent I guess it's normal and natural and part of it, but yeah, maybe don't lose yourself too much in the, the idea that you're gonna do something, like a particular position. 
 
 

Rachel Bedder: Yeah, for sure. It's rough out there. 
 
 

Benjamin James Kuper-Smith: Okay. Anyway, you eventually ended up getting positions, uh, more than one. Yeah. Maybe, I mean, so we want to, you know, uh, one of the reasons I invited you was because So we could talk about some of your recent work. And so let's, uh, maybe if you can, if you can [00:26:00] take basically a lot of the years now and summarize them kind of into how did you end up, uh, working in Princeton with, yeah, 
 
 

Rachel Bedder: So when I was thinking about applying for postdoc roles, deep, deep in the pandemic, I will add. So all that interviewing was done in my room. Something I realized was very important to me when doing my PhD is I am extremely motivated by the question above the methods. If something, if certain experiment or a certain research program is not asking questions I'm interested in, I cannot motivate myself to be working super, super hard on it. Um, and academia is a space where we want to be working hard, right? We really care about the work we do and the work we do is hard and tricky and full of setbacks. And if you're not motivated to do it, then that's a recipe for disaster. So. What we do in, um, the Niv lab and Yael's lab is we work on questions to do with mental health, mental wellbeing from [00:27:00] a computational lens. And that is to say people in the lab actually are often working on quite different questions about mental health, different, um, diagnoses, different symptoms, but it's all brought together by this general thread of, um, computational psychiatry. So that would just seem like such an exciting place to pursue independence as a postdoc. 
 
 

I can ask my own questions, but these questions are very similar to all the people around me. course I would be completely remiss of not saying that, you know, Yael is an incredible scientist. She has this technical brilliance and big picture ideas. The make, you know, someone, an amazing scientist, an amazing mentor to have those, and I've really benefited from the, um, her advice and mentorship. 
 
 

Benjamin James Kuper-Smith: we just talked about applications and that kind of stuff. So maybe how did you get your postdoc position? Was it the case that you contacted lots of people and then this happened to work out? Did you, did you meet before? Did you have a [00:28:00] position? Kind of what was the process for you during your PhD when you were applying for? 
 
 

Positions, or maybe just one position, I don't know. 
 
 

Rachel Bedder: So I applied, I applied for a few. I applied quite early. So this was what I thought would be my first round of, postdoc applications. So I only applied to places I would really, really want to go. I think one of them. Maybe two of them had a job advertised online, but two of them didn't. So I did four in my first round. that's quite, I think, standard for academia across the board. People don't wait, put a job application up traditionally and wait a long time. Often they have someone in mind, someone they've met with, maybe they met earlier in their PhD and they've had some peripheral conversations. So what I would say to anyone applying is don't. 
 
 

Um, I mean, if you find an advert, great, but don't wait for adverts. Start talking to people and connecting with people. Um, everyone I applied to work with was in network, so they knew [00:29:00] my PI to a certain extent, which meant I was a known quantity to them. I felt very comfortable cold emailing them, but you can really cold email anyone. Um, or often, you know, if you're going to the same conference as them, you can say, Hey, do you have 10 minutes to. Chat and things, and then you become a bit more known to them and stuff. Um, 
 
 

Benjamin James Kuper-Smith: just to interject there briefly, I mean, it could also be very useful to mail them before you go to the conference, and then it might be a bit easier than 
 
 

Rachel Bedder: Oh, sorry. 
 
 

Benjamin James Kuper-Smith: Oh, okay, yeah. 
 
 

Rachel Bedder: you know, you can during as well, especially if, uh, they gave a talk and you'd be like, Oh, I really loved your talk. Like, do you have five minutes to talk about this? Um, know, they know what's up sometimes when a PhD student is emailing to me to a conference. 
 
 

Um, 
 
 

Benjamin James Kuper-Smith: I mean, you could also be explicit about it, I mean, I think you might as well, right? I mean, why not? 
 
 

Rachel Bedder: Yeah, I would, I would advocate for that for sure. I think As a, as a British person, I find being direct very difficult [00:30:00] sometimes, but I think I remember being coached on one of my emails about You know, asking someone if I could visit or meet or something. And I had someone look at it and then I said, you haven't written anything about postdocs here. 
 
 

And I was like, Oh, I feel uncomfortable writing this. Like, no, you have to tell them what you want. 
 
 

Benjamin James Kuper-Smith: Yeah, yeah. 
 
 

Rachel Bedder: So yes, you can be explicit. Um, I should be more explicit for sure. 
 
 

Benjamin James Kuper-Smith: Ah, that's funny. As someone at the intersection of English and German culture, I very much understand the indirectness of English people, and some of the positive and negative consequences that can have, yes. Uh, so yeah, I know what you mean. Um, wait, what did I want to say? Oh yeah, so just a brief comment was that I, I did basically two episodes about postdoc positions. 
 
 

One was with Matthias Stranger, which was about getting a job, and one was with Toby Wise about getting a fellowship. So you can listen to those, I'll put them in the description. Um 
 
 

Rachel Bedder: And [00:31:00] Toby, I know very well, and he has been very successful in his fellowship. So people should take his advice. I will go back and listen to that and take his advice. 
 
 

Benjamin James Kuper-Smith: Yeah, it's funny. I should, I mean, I'm writing, I have a fellowship deadline in 10 days. Maybe I should listen to that episode. Yeah, it's been a while since we did that. So, um, sorry, but, so you, I guess, I mean, your, your PhD was also a bit on reinforcement learning, that kind of stuff, right? So you, I guess you, you knew the, that world that yeah, works and, and then it was just like, Oh, I think just really cool stuff on Conductor. 
 
 

And then you did kind of worked out and now you're there. Is that basically it? 
 
 

Rachel Bedder: Yeah, and I'm, I'm so happy, happy it worked out. The modeling stuff I do now is a little different to the stuff I did during my PhD. Um, it's more reinforcement learning involved now than it was before. Um, previously I was using more, um, models that you might found in neuro economics and things and models that track people's mood across a task, which are RL influenced. Um, but [00:32:00] now I'm much deeper in the reinforcement learning world. 
 
 

Benjamin James Kuper-Smith: Yeah, I mean, you're, uh, we'll talk a little bit about POMDPs, uh, later. I guess we can start talking about the paper itself. Uh, I don't know, was, is it a paper preprint conference proceedings? What, what exactly is the term? 
 
 

Rachel Bedder: It's a conference proceedings, um, from Cognitive Science Conference that was in May of this 
 
 

Benjamin James Kuper-Smith: Okay, yeah, called Modeling Rumination as a State Inference Process. And I thought maybe we could start, uh, fairly broad. What is computational psychiatry? What's, what's the purpose? How's it different from non computational psychiatry? 
 
 

Rachel Bedder: You'll get different answers if you ask different people this question, but the way I use computational psychiatry is think about a symptom of interest. In my case, I'm very interested in repetitive negative thinking, more specifically rumination. And I try and think, okay, what could be the underlying computations So you could say mathematics that might underlie a behavior or symptom. And can I write them down and [00:33:00] simulate them? So the computational part is the model and also the act of simulation as well. a really exciting thing to combine with any, any side of psychology, particularly psychiatry. Because when you write a computational model, there is nowhere to hide. And what I mean by that is every assumption you have about what's necessary for a process to occur, you have to write it into a formulation, which often reveals to you, you don't actually know enough about your own assumptions to write it as a model, which is really valuable as a scientist. 
 
 

It helps you move forward. And it also means, if I want to suggest something works in some way, I have to tell you exactly how it works. So, if I say, oh, mood affects how we value a choice, is it, is it an addition? Is it a multiplier? Like, what does that function look like? Now this can be perceived by people as reductionist in some ways, and essentially it is. But there's a lot of value in reductionism sometimes to move things [00:34:00] forward. to say it work like this? Oh, it doesn't. Great. Now we can try something else. So there's a lot of value in knowing what doesn't work as well as what does work. 
 
 

Benjamin James Kuper-Smith: Yeah, um, I have to again link to previous episodes. Uh, so just with Paul Smodino, he has a book out on computational modeling and mathematical modeling of social behavior. I did two episodes actually with him. Uh, one was one of the first, one was one of the last ones I did. Uh, and also one with Eichel Fried. 
 
 

So I'll just link to those in the description if you want to have several hours on this topic. Um, 
 
 

Rachel Bedder: And the joke in the field as well, as always, um, it's the field that spawned a thousand perspective papers. So you can read, you know, many, many perspectives on what computational psychiatry means and what it should be doing, um, by lots of different groups. 
 
 

Benjamin James Kuper-Smith: I thought the, the rule is that you always have to have more reviews than Empirical papers, right? Is that the, 
 
 

Rachel Bedder: That is, that is the joke. 
 
 

Benjamin James Kuper-Smith: yeah. Yeah. 
 
 

Rachel Bedder: [00:35:00] it's poking fun at ourselves and things and yeah. 
 
 

Benjamin James Kuper-Smith: You already mentioned rumination and repetitive negative thinking. So maybe what, maybe we can just actually introduce the paper then. What did you do? And I guess maybe just first briefly, because I think we have to talk about some of the methods in more detail. Um, but yeah. 
 
 

Rachel Bedder: So this paper represents for us a kind of foundational Foundation will go, as it were, at seeing whether something like a POMDP, which is a partially observable Markov decision process, or a state inference model, is useful for thinking about rumination. So rumination, for um, the lucky listeners you might have who don't know what rumination is. It's a style of repetitive thinking, which is often prompted by thinking about a particular thing that happened to you or a feeling you're having. And some of the more insidious rumination is asking, why did this happen? What does it mean that happens? this could involve thinking about past memories [00:36:00] that are connected. 
 
 

So often the example I like to give is, oh, you're at a party and you have a slightly awkward interaction with someone. And then you're thinking, why did this happen? What does it mean? You know, thinking about this later when you're lying in bed. To do this, you might think about other examples of when you met people at parties, that person as well, other awkward interactions. So there's these like concrete memories we're coming up with, like sampling memories from our history. And that often is connected to some kind of abstract thought, like, I am socially awkward, or I am a terrible person, or something more useful like, Oh, that person was probably having a bad day. This is like an abstraction we're making from these series of memories. So rumination contains both these concrete samples and this abstraction. So for us, a state inference model was really exciting for that because it has a series of observations, and then it makes an inference from that. what we did in this paper is we simulated this model and tried to see if we could get it to, um, Um, or ruminate in a sense, sample for a really long [00:37:00] time, given particular conditions. 
 
 

Benjamin James Kuper-Smith: Oh yeah, so I guess, okay, so rumination is, yeah, it's interesting because earlier you mentioned negative repetitive, um, thinking and yeah, I wasn't sure what type, because you can have like catastrophizing or whatever. Um, I guess that's more about the outcomes of certain things, but you're more interested in the causes of the repetitive thinking about causes of events. 
 
 

So the kind of, yeah, I mean, inferring hidden causes, um, of the observations you have. 
 
 

Rachel Bedder: I'm fascinated by this idea of how people get to, I am a terrible person. Like what are the mechanisms that lead someone to think about their life and pull out that thought? Basically. I think that's, you know, fascinating to try and model in a sense. 
 
 

Benjamin James Kuper-Smith: Hmm. Yeah. So, um, we already mentioned, uh, POMDPs. Uh, maybe we can start with an MDP. What's the Markov Decision Process? And then we can work our way step by step to the more complex versions. 
 
 

Rachel Bedder: Sure. So, uh, Markov decision process or MDP [00:38:00] is something that's part of the, um, reinforcement learning universe. in an MDP model, you know, the kind of states of the world, and then based on those states, you can choose particular actions. So an example could be, if I'm in a bar. So that's a, or a fancy bar, let's say, that's a state and it has particular actions that are associated with it. 
 
 

So if I'm in a fancy bar, maybe I have to wait to be seated. So waiting would be an action with high value because it will lead me to a rewarding outcome, sitting down and having a nice drink. Um, another state might have different actions associated with it. So let's say if we're going into a local pub, the action of just going straight to the bar and sitting down has high value because that gets you to the reward quicker. But the value of standing and waiting at the front is less so because it's not going to get you anywhere. So an MDP assumes these kind of states of the world are known. I know if I'm in a fancy bar or I know if I'm in a local pub and it conditions what [00:39:00] actions you want to take based on that knowledge. So a POMDP, what it does instead is it assumes you're not 100 percent sure what state you're in. And it leverages that uncertainty against what actions I might choose. Um, so for example, if you walk into a place and you don't know if it's a fancy bar or a local pub, two actions, you know, going and taking a seat have different value in each state, but you might not be sure what state you're in. So what you can do is you can stand at the front and you can observe the environment, take in observations, and that will help you become more certain about which state you're in until you get to a certain amount of certainty, and then you can choose an action. the statistics of the environment, that will change how certain you want to be. for me, as a British person, going and sitting down and then being told I did the wrong action is extremely traumatizing and embarrassing. So I might stand at the front and wait and take more observations in. to make really [00:40:00] sure it is a local pub before I go and sit myself down. Whereas someone who's less concerned about that kind of social embarrassment might just take a little look and then just have a guess and sit down. 
 
 

And if they're wrong, they'll be like, okay, not the end of the world. Yeah. So this is, this is exciting because the observation idea, the idea that you take in observations over time, we thought of in the paper as sampling from memory. So how long do I need to think about something before I infer the state? 
 
 

Benjamin James Kuper-Smith: Yeah. Like do you, do you stand in front of the pub for five minutes and look inside? Or do you just wait there all evening until it's too late and the pub closes? That kind of thing. 
 
 

Rachel Bedder: Exactly, and also if you're in, if you're in the, the pub and you're standing there thinking about it, that might also be a negative action because everyone's bumping into you and they're like, what is this person doing? So waiting can also be something that doesn't have a value as well. So you have to leverage like, can I wait for ages to make all these observations before I choose an action? 
 
 

Or do I need to move quite quickly? And that's something we play with in the model as well. We look at the cost of taking observations and then we look at when you make this cost very low. So [00:41:00] sampling from memory, you can think of as a very uncostly thing. You can just sit there and do it, very little effort. When this cost is very low, does it mean you can take more samples? And indeed, as you might suspect, it means you can. 
 
 

Benjamin James Kuper-Smith: Uh, brief question about POMDPs. Um, I mean, so the, you know, the PO is partially observable. Is, is it just what state you're in? Is it also, I'm vaguely familiar with these things, probably should be more than I am, but You know you have states, actions, values, transition probabilities. Is it just the state that's, uh, not necessarily observed or can all of these be, uh, can you for example also know like I know what state I'm in, I know what I can do, I just don't know how good it is, you know, you don't know the value of the next state or that kind of thing is kind of what can be partially observable in this framework. 
 
 

Rachel Bedder: I see. Um, so I would not present myself as an absolute expert in POMDPs, but the way I use them, at least, and I think one of the most common ways to use them is indeed it's the state that's partially observable, and we often call these hidden states. to [00:42:00] distinguish from, like, a MDP style state. you're thinking about something like, uh, the reward function being less known, so you don't know exactly what reward you might get at the end, uh, we would tend to think about that in an MDP context, but of course you could have both. 
 
 

You could have a partially observable state and a unknown reward function, which requires a lot of learning to figure out. you build in all of these kind of uncertainties, it can make the model a little less, uh, wieldy to handle basically, because when you change one thing, it's obviously interacting with another thing, um, which is part of the, the fun and difficulty of computational modeling. 
 
 

And as one of the reasons in this model, at least we chose to keep it really, really simple because I think I mentioned briefly before is the value in using this model is to increase our understanding of a process. that means we have to have a really good grasp of what all the different working parts are doing, and it should also be accessible to people who don't have, you know, a really deep computational [00:43:00] knowledge because computational psychiatry as a field is really dependent on that crosstalk between computational people and people who are more engaged in psychiatry. So keeping the model of, you know, a nice level of understanding is very, very valuable. 
 
 

Benjamin James Kuper-Smith: Yeah, you don't want to frighten the clinicians. Um. 
 
 

Rachel Bedder: Yeah, it's not so much that I know you're being silly when you say that, but, uh, my concern is when we think about it that way, we're saying we're somehow saying computational stuff is more, uh, valuable or trickier or harder. I know someone who was often frightened by math. I appreciate that perspective, but I think it's more that, um, need to just be like talking in a way where we can both understand each other without having these whole histories. of knowledge. And that's, that's what you need for multidisciplinary practice or interdisciplinary, I should say, um, is you need to be able to have that cross talk, um, and you need to be mindful of that in everything you're doing. 
 
 

Benjamin James Kuper-Smith: Yeah, I mean, you can't require that clinicians [00:44:00] read about the entire history of computer science for them to understand your paper. Yeah, exactly. 
 
 

Rachel Bedder: It's up to you to make it clear and understandable. 
 
 

Benjamin James Kuper-Smith: Yeah, is, um, okay, so I mean, I'd like to use your, well, make the, how you use the polynomial be slightly more, uh, graspable and a bit more specific with some examples. Yeah, I mean you have states, uh, actions, rewards, that kind of stuff. How does this relate now a bit more specifically to rumination and what it does in this context? 
 
 

Rachel Bedder: Right, so one of the really exciting things about this model is take it as all the things, so the states, the actions, the rewards, are imagined by the agent. So this is something that's all going on in my mind, right? Whereas traditionally we might think about these models in terms of observing the external world, or sampling in an experiment, visual stimuli you might see. means we can think of every element of this model as a belief, a belief that person has about the world. How good or [00:45:00] bad do they think the outcomes of an action could be? Um, what possible states do they think exist or could be the candidate state to explain a particular event? So you can adjust all these different things see if that increases the amount of sampling someone would do. 
 
 

So the amount of observations or samples they would want to take from memory. So one of the things we did is we did perhaps the most obvious thing in first instance, is jack up the possible negative outcomes. So both these states have actions you could take. One action leads you to a positive outcome and one action leads you to a negative outcome. The other state is the opposite. The same action will take you to a negative outcome and then a positive outcome. So you need to know what state you're in. In order to choose an action at the end and finish the episode, finish the ruminative episode. So if you push up those losses really, really high, what you see is someone has to take more samples before they reach a conclusion [00:46:00] about what state they're in and can stop ruminating. And this makes sense, right? If there's some real big danger in the world that you believe exists, then sampling multiple times, being really, really sure You know, for want of a better word, it's rational. It's, it's the thing to be doing. You really need to figure these things out. So what we can do is we can use this model and say, rather than like, Or rather than pathologize the repetitive, repetitive nature of it, be like, Oh, they're just thinking for too long. 
 
 

Be what might people believe about the world, which leads them to think for too long, what beliefs might they hold about the things that will happen to them. And, you know, you can play with different bits of the model. You can change, uh, the noisiness of the samples. Do they think every sample they take from memory will be very helpful or will it be slightly more ambiguous? 
 
 

And that can affect how long you sample for. 
 
 

Benjamin James Kuper-Smith: And so, so state in this model in terms of rumination, let's say in a social context or something like that, is a state, uh, uh, uh, judgment about their own personality or [00:47:00] what exactly would that be? 
 
 

Rachel Bedder: So I think about them as. inferences, like the world is like this or the world is like this, or this person is, like this or like that. So one I often use is my boss is mad at me or my boss was having a bad day. you can think of these as states. They're very abstract, right? Because the. You know, you can't see that. 
 
 

I can't see, draw a picture of my boss having a bad day. It doesn't exist in a particular place in space. often how we think about states. But I think thinking these things are states to me, like their beliefs about the world, which change what actions I should do. So they're states at a very high level of abstraction, but they're still states in themselves. 
 
 

Benjamin James Kuper-Smith: Yeah, I mean, I don't know. I might check this out if, if you basically already answered this question. Um, we'll see. Uh, the question is basically like, what does the POMDP allow you to do that you can't do without it or with different methods? Um, yeah, I don't know whether you already kind of answered this in a way, but. 
 
 

Rachel Bedder: So really [00:48:00] cool thing about the POMDP or using a model grounded in reinforcement learning generally. this kind of normative framing. What should someone be doing? So given the rewards and the uncertainty in the environment, we can actually calculate best action in any given situation. Now, initially that can seem pathologizing, right? 
 
 

Because we're saying this behavior is good. You're doing the right thing. And this behavior is bad. You're doing the wrong thing. Um, but you can flip it on its head and say, let's assume the behavior someone You know, under these reductionist, um, principles we've made, uh, let's assume the behavior they're performing is the right thing to do. What would they need to believe about the world for that to be the case? So you can talk about everything within this like incentive framework and this rewards framework, is not to say like I fully subscribe to a view of that's how the whole. It's a really valuable way to ask questions, I think, because it allows us to say, why would someone do [00:49:00] this rather than just describe it? 
 
 

We can say why, why this behavior, rather than just what does this behavior look like? 
 
 

Benjamin James Kuper-Smith: Uh, one question I had whilst you were talking was, uh, kind of what does this model tell you beyond the rumination context? I mean, is this also a model that tells you something about reinforcement learning in general that kind of, um, I'm not, I can't quite tell right now whether it's using kind of fairly standard POMDP and in this context and elucidating the context or whether you're also advancing the model itself to then tell you something a bit more about reinforcement learning itself that could be used in various different contexts. 
 
 

Rachel Bedder: So am I advancing the world of POMDPs? Absolutely not. This is, this is extremely vanilla POMDP. Um, I think the advance I'm trying to make scientifically is that application of it to something like repetitive negative thinking. 
 
 

Benjamin James Kuper-Smith: Yep, okay. Yeah, you mentioned already a little bit about the loss gain asymmetry. Yeah, could you elaborate, just because I'm interested in loss gain asymmetries, a little bit more about, yeah, the asymmetries in this [00:50:00] particular context. 
 
 

Rachel Bedder: so something we only partially show in this paper, but we've subsequently, um, simulated more is one thing we see is when you increase the losses, so the possible negative consequences you think might be in the world, it increases the sampling, but we actually see the opposite for gains. When you put the gains up really high, so there might be really great things in the world. 
 
 

You actually reduce the amount of samples you take as the gains get higher. Now this, this to us immediately seemed very counterintuitive, but I think that's because we assume decisions are between equal magnitudes. So there's really good things, there's also really bad things. But we show in this particular circumstance that if you just think there's really good stuff, you know, you can just take a few samples, you know, and just be like, okay, I'll just take an action because it's so likely to be good and great. That we don't need to waste a load of time thinking about it, basically. And in this model, thinking is costly. You lose or you take a penalty, a small penalty each time you take another observation or another sample from memory. [00:51:00] Um, 
 
 

Benjamin James Kuper-Smith: So, but isn't that more about the distribution of outcomes? Basically, you, let's say you have ten outcomes that are all negative. If all of them are really negative, then it also doesn't really matter what you do. Yeah, 
 
 

Rachel Bedder: be some good thing that you can do eventually. If all things are bad, actually the model says don't bother ruminating, just do something. So there always needs to be this possible good things and these possible bad things. under this framework because, you know, why, if everything's bad in the world, none of these behaviors to be useful for me. 
 
 

Now, depending on the distribution of those negative things, there might be some very small negative things and some very big negative things, then you would see, I think, the increase when you increase the asymmetry between both of those. But I'm actually not completely sure. So I think that I'm so fascinated with generally is how Context changes the relationship between losses and gains, [00:52:00] it's very tempting to want to make. Very general statements about if you increase losses, then you'll sample more than what you increase gains. But this isn't a very particular context. It's one where we've dropped the cost of sampling to be very, very low, because we're thinking about it as thinking. So sitting and taking more samples is, you know, it has a very small penalty. But actually what we see is when we increase that penalty to make thinking more costly, you see the function for gains changing. So when it's very costly, you actually take more samples. The gains. Now, this is a smaller effect, but it is there and it shows you the importance of context on when you think about leveraging losses against gains. 
 
 

And, you know, one reason that's really interesting is because when anyone is doing any one of our experiments, which looks at gains and losses, they're bringing all this baggage. from their real life about their beliefs about and negative things. There are negative things more likely than positive things. 
 
 

If I take a risky decision and something's bad, I'm, you know, am I [00:53:00] just unlucky and I always get the bad thing? And that's why I believe about the world. So when you can build a model which adds more and more of this context in, um, you can have a slightly better grasp of what context see these particular asymmetries in. Um, and I think that's really exciting to understand. 
 
 

Benjamin James Kuper-Smith: so, I mean, this was a simulation. Did you also Or, I mean, did you? Not on that paper, but are you collecting any data, seeing whether this is, I don't know, people who have social anxiety or something like that actually do this, or? 
 
 

Rachel Bedder: So that's a great question. And one that think is plaguing us a lot, which is how to test the model, right? So the model makes a bunch of predictions, um, can we test these? That's one of the really difficult things about working in the space of repetitive, negative thinking is is no behavior to. directly observe in a kind of time series. you can't get someone to come in the lab and just do their [00:54:00] natural rumination process. So in, in terms of matching it to rumination, this is, this is not something I think we will be directly pursuing, but we'll be thinking about how we could test this model in other forms, some of the predictions it makes. the next thing I think we will test is just some of the basic statistics of the question. So given if we make the penalty of taking a sample very low in a very simple, um, experimental task where you're sampling visual stimuli. Can we see this relationship between loss and gains in certain conditions that matches some of the relationships we see in the model? Um, so we're taking it a little bit away from the ruminative space to test some of the assumptions in a much drier context. you know, studying it in that space, I think, will allow us to then build on the results of that and think about the model a little bit more. So there is going to be a feedback process between theory. And data. But they won't be quite as clean as just fitting the model to, you know, someone's personal [00:55:00] private thoughts. 
 
 

Benjamin James Kuper-Smith: Yeah. 
 
 

Rachel Bedder: But you know, it's an aspiration. It's 
 
 

Benjamin James Kuper-Smith: Yeah. I'm just curious, is that something that, I'm wondering whether this is strength or weakness of it, or it's just independent, like it's assigning a good or bad outcome to, um, The difficulty of testing this model is just a bit silly. Uh, I'm just curious because as when initially you said it, I thought like, ah, do you wish you'd maybe done simulations that you can test more easily because then you kind of have like one project that really, uh, you can explore the one a bit more fully maybe, or do you think like, no, this is cool, like we did something that we, you know, can't do with behavioral testing in the model and so they're more complimentary. 
 
 

I'm just curious how you think about that. 
 
 

Rachel Bedder: Yeah. So I think the, one of the most important. we have from doing this work is just asking if this is generally a useful framework for understanding rumination. Does the model work in ways that we would expect it to if it was mirroring this behavior [00:56:00] slightly? Um, so things like if you increase the losses, does it increase the samples? 
 
 

Yes, it does. Um, which is consistent, you know, with, um, belief someone who might be diagnosed with a mood disorder might have. Like, I believe the negative things in my environment are worse. So we've kind of established that, and there will be a longer paper, a longer theoretical paper that goes alongside this at some point. we can start thinking about other ways to think about state inference. In a reinforcement learning context that might allow us to come up with more, um, models that can be more testable and things, but the value here is really in first instance, does it even make sense as a framework to use? And I think we've established that it does. 
 
 

So I'm excited to see directions. We could go with it, even if they're not directly testing this model. Exactly. 
 
 

Benjamin James Kuper-Smith: Okay, great. Um, Should we talk curation? Slight topic change. Uh, yeah, I mean, uh, you mentioned earlier you did art and psychology, so maybe just going back briefly to that, um, [00:57:00] why? Why did you study art and psychology? 
 
 

Rachel Bedder: So the reason I applied to do that, uh, so in the UK you apply straight up for the topics you want to do. There's no moving around in the middle and things and changing your major or minor. Um, so I applied for that because I thought, you know, just as simply as I like art, I like psychology. What's the job that puts them together? 
 
 

Art therapist. Okay, so I'll go and do this and then maybe I'll do a master's degree in art therapy. So I think that was my thinking. As I progress throughout. undergraduate degree. I realized my interest was really more in research. of the ways this, uh, uh, this became apparent to me is I would talk to people about my artistic practice and they would say, what do you mean variables? 
 
 

Like, what do you mean changing the methods? And I would talk about my whole artistic practice in the context of, um, the scientific method. In a way, it was so coming back and ingraining into my art that, you know, data and all these things. Um, but I was like, this is really where my. [00:58:00] my passion is and things. I wonder, was that, was that a smart decision to do as a degree, if you want to be a researcher and maybe we can talk a bit more about that later in terms of like things we wish, regrets we have and things we wish we could have done differently. But it's led me to, you know, do all these really interesting curation projects, which have been, you know, really exciting to do alongside a research career. 
 
 

Benjamin James Kuper-Smith: Yeah, I mean, it's funny, like, if you do art It's part of your degree. It doesn't sound like it's gonna, it's not the obvious direct path to increasing your employability or anything like that. But wait, didn't Matthew Botvinick study art history or something like that? I think 
 
 

Rachel Bedder: something like 
 
 

Benjamin James Kuper-Smith: one of the, like, I think it was him, right? 
 
 

Rachel Bedder: I think so. I don't want to spread fake news, but I 
 
 

Benjamin James Kuper-Smith: Yeah, yeah, yeah. Um, so if you, if you want to do something with reinforcement learning, study art. Um, yeah, so, sorry, I don't know whether you want to, 
 
 

Rachel Bedder: Yeah, I was just, just a [00:59:00] funny anecdote, I guess, in that space. Or, so, I have friends from both fields now, or both professional worlds. And to all my Scientist academic friends, they'll always ask me for like random art things or design things and stuff and then all my artistic friends Uh, they ask me like random science things. 
 
 

It's funny because i'm in one space I'm like the artsy one and the other space i'm like the 
 
 

Benjamin James Kuper-Smith: yeah. 
 
 

Rachel Bedder: Um, which can be quite funny 
 
 

Benjamin James Kuper-Smith: Yeah, that's basically my life. Again, being German and English, you're always the other, uh, you're always the thing that makes, you're always the thing that makes you different in some sense. Uh, sometimes it's good, sometimes not. 
 
 

Rachel Bedder: Yeah, it can be tricky sometimes because you feel like you're not fully Respected is not the right word, but fully endorsed as being part of that space, but I think you just have to lean into it because you know, not to sound too cheesy, but what makes us unique is ultimately defines us as, as researchers as well. 
 
 

What unique perspective are [01:00:00] you bringing to this question? 
 
 

Benjamin James Kuper-Smith: Yeah, I mean, I think whenever you're like two different things, people would often just assume you're from the other also. I mean, I also had it with friend groups. I wasn't someone who was always like in one like group of like a few people. So I think a lot of people just assumed I was like part of that friend group because I hung out with them also. 
 
 

But it was just like a bit everywhere. Um, and the same for like 
 
 

Rachel Bedder: Have you heard of this idea of, um, humor A when you're with friend group B? Have you heard this? 
 
 

Benjamin James Kuper-Smith: No, I think I can imagine it, but 
 
 

Rachel Bedder: like in sometimes different social spaces or friendship spaces and professional spaces and cultural as well, we have different humors we 
 
 

Benjamin James Kuper-Smith: Mm. Yeah. 
 
 

Rachel Bedder: situation and use the other humor. 
 
 

And then if I was like, what are you talking about? And you're like, Oh, 
 
 

Benjamin James Kuper-Smith: Yeah. 
 
 

Rachel Bedder: realize this joke would not land 
 
 

Benjamin James Kuper-Smith: Yeah, yeah, yeah. 
 
 

Rachel Bedder: Um, which I think was one of the biggest, uh, cultural shifts I had moving from the US. 
 
 

Benjamin James Kuper-Smith: Hmm. 
 
 

Rachel Bedder: UK to the U. S., [01:01:00] sorry, is oh, they don't find that mean joke funny. 
 
 

Benjamin James Kuper-Smith: Yeah. They just think you're mean. 
 
 

Rachel Bedder: They think I'm just being mean, where all I'm trying to say is, I want to be friends with you. That's why I'm being so mean to you. Is 
 
 

Benjamin James Kuper-Smith: Yes, exactly. Why isn't my abusive behavior? It's a sign of me liking you. Yeah. No. 
 
 

Rachel Bedder: think, like, how did it come across to the other person? 
 
 

Benjamin James Kuper-Smith: I mean, I had it in the UK, like very obviously that Americans would often also make the comments and then say, I'm just kidding. Which for me would ruin the joke, but I guess it's just different norms and what people expect it means. 
 
 

Rachel Bedder: sly wink, you know. The little sly wink that says I'm joking. If 
 
 

Benjamin James Kuper-Smith: Yeah, I'll just, 
 
 

Rachel Bedder: joking, it's like 
 
 

Benjamin James Kuper-Smith: yeah, yeah, it's not funny anymore. But yeah, the, going back to the directness, I've definitely had some moments where I think I was a bit too German in my English interactions, but because I sound like this, people thought I was just being really rude. Um, um, anyway, uh, art and creation again.[01:02:00]  
 
 

Uh, I mean, so just briefly, you also mentioned, I mean, you mentioned you did apply for some, like, art kind of jobs, right? So it's, it doesn't sound like you were, like, completely bent on, or was it just because you wanted to earn money? And the art world was your way of making a lucrative career to finance your psychology masters? 
 
 

Rachel Bedder: Yeah, when you say it like that, it doesn't sound so smart, right? Like, the lucrative career of working in the art world. I think I really would have just taken anything at that 
 
 

Benjamin James Kuper-Smith: Okay. 
 
 

Rachel Bedder: think if I had started working in the arts, I might have been persuaded away. From these things, because it's so much fun working with artists and in the art space. 
 
 

You know, it's, it can be exciting like every single day. So maybe that would have tempted me away. 
 
 

Benjamin James Kuper-Smith: I thought I understand you have done some of that work whilst being a researcher, um, so maybe, I don't know whether you want to talk about the AX& S collective or something else that makes sense. I'm just curious that kind of what, uh, what did and do you [01:03:00] do in that context? 
 
 

Rachel Bedder: So X and S collective, which is for art times neuroscience was a group of me and three other women who all worked within the science and art spaces. And we would look to put on shows or commission artwork or run talks that. Help people connect ideas between art and science or use the arts to help people understand particular scientific things So I did that before my before I started my PhD I did a lot of it between my undergrad My PhD that as we discussed before it was quite a while because I was earning that and in that cash But a really fun project I did during my PhD which was so exciting because I've never done our art science project Which spoke to the research I was doing at the time is um, we ran this project As part of the Max Planck UCL Center, and also the Wellcome Trust Center for Neuroimaging, which are both parts of UCL. 
 
 

I'm sure you've heard of them, but, 
 
 

Benjamin James Kuper-Smith: I've been at them, [01:04:00] yeah. 
 
 

Rachel Bedder: psychology buildings at UCL that I wondered if we were overlapping or not. 
 
 

Benjamin James Kuper-Smith: Yeah, no, not the Max Planck, but the welcome, yeah. 
 
 

Rachel Bedder: hmm, yeah. So, Um, this project, what we had people do, I'll tell you what we had them do. And then I'll tell you, tell you why on earth we had them do that. Um, as we have people write on these postcards, um, thoughts or feelings that have been preoccupying them recently. And then they put them in this post box. So they like sent the thought away. So I had this nice kind of fun aspect to it. And then what we did with those postcards is you would then become the kind of postmaster afterwards. 
 
 

And you would see postcards that other people had written. Um, and these are anonymous, of course. you would have to put stamps on them and these stamps said things like mood, effort, confidence. So your idea is you had to read what that person wrote and say what kind of psychological elements or cognitive factors might this be related to. and then as people did that, we hung them up and then we had [01:05:00] discussions with them about labeling things. And we were like, okay, we've used things like confidence, effort here, but also people can get diagnostic labels. What do you think is more useful in terms of understanding someone's symptomology or how these, um, connecting this postcard with five other postcards that say mood and effort? Is that useful? Does it help us understand something? Um, and talk to people a bit about this idea and computational psychology of thinking of transdiagnostic symptoms. So things that cut across many, many different diagnoses that might connect people so we didn't have a specific message or thing we wanted them to come away with, but we wanted them to think a bit more about diagnostic systems and how we think about, um, symptomology as well. And then we had, you know, from that we had about, I think just under a thousand postcards people had written. We did it at festivals. Um, we did it at university open days and things. And what we did is we them all up, or I would say a paid, a paid intern typed them all up. Um, paying your interns is very important. And [01:06:00] then we typed up the content. We had the stamps people have put on it. I think there was six or seven different categories. So mood, effort, appetite, some other things, confidence. And then we used a very, very basic machine learning algorithm For each card, we had different weights of how much these different factors that grouped together loaded on it. 
 
 

So how much is this card similar to this other card in three different dimensions? Then we had an artist make those into graphics, and then we put them all on this big wall at an exhibition. you could look at the graphics and make connections and make patterns and then turn them over and read the content from the postcard and see, Oh, based on. 
 
 

this connection I've made, do these people seem more similar or different to another group of patterns I've made? So we're talking about like homogeneity and heterogeneity in, in disorders and things. Um, really fun project because you could have, you know, amazing conversations with people. Um, we sometimes call it, um, [01:07:00] Like object oriented conversations, so people find talking to a scientist directly, if they're not used to doing that, quite intimidating. Your, your listeners can hear, you know, I sound obviously incredibly intimidating. Um, but, you know, as a, they've never engaged with 
 
 

Benjamin James Kuper-Smith: I'm trembling. 
 
 

Rachel Bedder: Yeah, exactly. But if you talk via an object, so if you have something in your hands, you have something you're making together, people are much more confident, and sometimes they literally talk to the object. But they're talking to you, but they look at the object and actually helps you have really interesting conversations. And especially when you're using content that someone might have provided. So they might've written on a postcard or stamp the postcard. They're kind of the expert because you're asking them, why did you do this? 
 
 

So what does this mean? And, um, they're getting to feed back into science and scientific ideas in that respect. so it was a really great project to be involved in. 
 
 

Benjamin James Kuper-Smith: And this was students, general population, or what kind of people filled in the cards, labelled them, that kind of stuff. 
 
 

Rachel Bedder: So we [01:08:00] did a couple at the university, so they were students, but we also did it at kind of local family orientated events and Latitude Festival as well, which is a big music festival that happens in the UK, which is also very family orientated. we designed it so it would work for all ages. So if you're a very young child, you can just write on a postcard and scribble, and then you can stick some stamps and things. 
 
 

And if, and if like a small child did it, we put that to the side so it didn't go in the big. Data stuff. And if anyone wrote anything, you know, deeply personal or that could be, you know, unpleasant for a small child to read, we hung those ones up higher so they just couldn't see them. but, but the children did write some very sweet things on it. 
 
 

So the prompt on the postcard is, uh, what is something that's been preoccupying your mind recently? And you'd be at a festival and they'd be like, I'm very excited to get ice cream later. so they had fun with it too, but it was general population mostly. 
 
 

Benjamin James Kuper-Smith: I'm also excited for ice cream. Um. 
 
 

Rachel Bedder: Always, right? 
 
 

Benjamin James Kuper-Smith: Uh, yeah. [01:09:00] Uh, are you, I'm curious, I mean, are you still, I mean, yeah, are you, basically, I'm curious, like, how you, or whether you combine this now still with your, not necessarily with your work, uh, actually, maybe that's the question, um, do you, is there now, for example, in some of the stuff you do, is there some direct relation to your scientific work, or do you treat them as just completely separately? 
 
 

And maybe, what, what are you, are you doing right now? Are you just painting, taking photos, or are there other projects that I'm not aware of? 
 
 

Rachel Bedder: So right now they are very separate for me. They fluctuate in their connection. Um, and right now this is a moment when they're separate. This is for, for two, two broad reasons I would say. So, my art practice right now is mostly painting, in the more traditional sense. I'm very lucky to have a small studio in my apartment where I can work on these things. for me having something outside of academia that I can work at, I feel good at, I'm learning more things doing, is extremely valuable to my, health in that space, and my mental well being. [01:10:00] Secondly, so that's keeping them separate, you know, means I can throw all the science out the window when I walk in the studio, um, well, ideally, obviously you can never completely do that as a scientist, but. 
 
 

Benjamin James Kuper-Smith: heh heh. 
 
 

Rachel Bedder: Um, this is why I study repetitive negative thinking. 
 
 

Benjamin James Kuper-Smith: Yeah. 
 
 

Rachel Bedder: Um, I'm only half joking there. But, the other reason is, these art science projects are reliant on a lot of structural things. So, institutional support, if you want to do it in a university context. Uh, you need to know the grant landscape for Art projects that relate these things. 
 
 

Um, and also you need a community and these are, um, some things now I've moved to a different system. I'm not saying they don't exist, but I don't have those things anymore. I don't understand them in the same way. So, um, I hope in a few years to be able to scale back up to doing these kinds of projects where I have more knowledge of the landscape and also more, more time as well. 
 
 

Benjamin James Kuper-Smith: One thing that [01:11:00] I, so I often do lots of stuff on the side. like this, but I don't, well, I don't really consider this creative. But, um, one thing I often find difficult is that I tend to then turn it into another big project that can be stressful. I'm curious, is it, uh, I don't know, you go to your atelier, and then you just paint and forget about everything? 
 
 

Or is it also like, Oh, I got to make this painting and, you know, the first day is great. But then once you work on it, and you want to, I don't know, I've never really painted much, but I'm curious, like, is this also something where it can be a really nice complementary activity that takes you out of it, or it can also be something that becomes stressful in itself. 
 
 

I'm curious whether what that's like for you. 
 
 

Rachel Bedder: Yeah, so I try to be very mindful of that, of not making another thing that I need to succeed at in some way. So for that reason I have a lot of three quarters finished paintings, I think, because I just let myself, [01:12:00] if I'm not enjoying working on that particular bit at the moment, I'll just come back to it later. 
 
 

There's no external influence being like, you need to get this done, and things, and you have to try very hard not to internalize that. But I think the fact that it can be hard work is nice because it's hard work in such a different way. Um, you know, there's many different ways you could try and express something with paint. Um, you're not just trying to, uh, make something that represents something possibly in the external world, but you communicating things with the material and stuff. So there isn't really what. You might call it a right and wrong way to do it. So you can really embrace sometimes, it sounds so cheesy, but where the painting takes you. Um, you know, I sometimes think, well, what does the painting want me to do And because it has this intellectual side to it and this hard work side to it, technically it can take you away from those other Hard work things that, um, might have more stresses associated with them because it's your [01:13:00] literal job, um, and you need to succeed in that to, financially support yourself. 
 
 

Whereas the stresses that come with art are just about the art. That's so contained to this tiny, tiny world. Um, that can be very satisfying. 
 
 

Benjamin James Kuper-Smith: Yeah, yeah, the taking, say, not trying to paint something, but using the painting as a way of exploring certain things definitely sounds like a good way of not making it become a thing that you need to be perfect at. 
 
 

Rachel Bedder: I do also engage, have a few different arts things I do. So watercolor and, um, analog photography that I'm completely aware. I am not. I just do them for pure joy and fun. And those things would be really great as well. Um, there's this author, um, called Oliver Berkman. He wrote this book called 4, 000 weeks. it? You're nodding like, 
 
 

Benjamin James Kuper-Smith: yeah, I've read it. 
 
 

Rachel Bedder: okay, I'm evangelical about this book. Um, but he says in it, he says you should have a hobby [01:14:00] you enjoy doing. And it's even better if it's something you're not that good at. it's this idea that you can't make every aspect of your life productive. You should do some things just for the pure joy of doing them, and not be thinking like, oh, how can I improve? 
 
 

How can I monetize this? So I think of my watercolors that way. People often say, oh no, you're, you're a great painter. I'm sure your watercolors are amazing. They're not. They're awful. I have so much fun doing them. So it's a nice thing to embrace as well. 
 
 

Benjamin James Kuper-Smith: Yeah, it's also nice that people do say that they're amazing. It's, you still don't want people to go like, oh god, what's the, yeah. But, I agree. Embracing being almost intentionally bad at something. Whilst actually trying to be good at it. I guess it's a bit of a, 
 
 

Rachel Bedder: It's so freeing. It's really nice. 
 
 

Benjamin James Kuper-Smith: As always, Well, for the past 10, 15 episodes or so. Uh, recurring questions at the end. Um, do you have a book or paper that you think more people should read? Can be famous, non famous, old, new, whatever. 
 
 

Rachel Bedder: So I knew this [01:15:00] question was coming, and I really, I struggled over it. Not because I couldn't think what to recommend. It's because there's so many things I wanted to recommend. And the thing I converged on is this wonderful, um, novel The Heart is a Lonely Hunter by an McCullers. It was written in the, um, I think the early 20th century. And the reason I recommend this novel for someone who studies psychology and thinks a lot about people's beliefs in their internal worlds is this book is from the perspective of multiple different characters who interact. And it really is about loneliness as the human condition and what we perceive from other people's actions. You know, other characters who are also lonely that we don't. Know about and how this can isolate us in our individual worlds, and it's such a beautiful study that feeling and the way we think [01:16:00] about each other and treat each other You know my interest in psychology, I think You know, and there's always a danger of creating myth and legend about ourselves, right, when we're invited to. 
 
 

But I think it really comes from a love of reading and literature and trying to understand these rich and alive people have, or why people might do things, and thinking more about that. And I think this book was really, you know, it had a profound effect on me terms of my thinking. 
 
 

Benjamin James Kuper-Smith: Great, yeah, that's, it's one of those books I've known the title of for ages, because it's one of those titles you don't forget, but I haven't read it yet. So I guess maybe this is enough of a, of a push to actually read it. 
 
 

Rachel Bedder: Yeah, it's a, it's such a beautiful title, isn't it? It's, it's wonderful. 
 
 

Benjamin James Kuper-Smith: Yeah, second question, something, you already hinted at this a little bit, something you wish you'd learned sooner. Yeah, 
 
 

Rachel Bedder: my answer to this question has evolved so much over the years. Um, if you'd asked me maybe five years ago, [01:17:00] in the middle of my PhD, I would have said, I wish, I wish, I wish I hadn't done an art degree. I wish I'd done maths or neuroscience or something more scientifically technical, because I was feeling sort of behind in my knowledge compared to my peers, but as I've, you know, advanced in my career or move forward my career, I've realized that. Yeah. As I mentioned earlier, it's the idiosyncrasies we have that lead us to asking particular questions and hypotheses and studying things in particular ways. So now I really try not to of any of my experiences I've had as somehow subpar to a different experience I had, because it would have led me to a different research space or a different way of asking questions. And that's not to say that would be better or worse, but, you know, embracing the one that you're working on. often feel this insecurity more recently thinking about, you know, um, you, you asked earlier about if this is moving forward, POMDPs. It, [01:18:00] it is not, and often I feel insecurity, like, well, maybe someone who is a POMDP expert should be doing this work and not me, you know, someone who lives, drinks and breathes POMDPs. 
 
 

But they wouldn't ask the question in the same way. They wouldn't have, you know, put these links together. So I think that's a really nice, uh, Breakthrough to have as a scientist to think about, you know, your unique experience makes you ask questions in a particular way And we need lots of different unique experiences to tackle these problems from the different angles to make progress 
 
 

Benjamin James Kuper-Smith: yes, yeah, you can. Sometimes it's not quite clear how something becomes useful or relevant, but it's actually I mean, there are some things maybe that, uh, can be a bit of a waste of time, but it's actually, if you actually think about it and you want to use it, it's actually kind of hard to do something useless. 
 
 

Rachel Bedder: Mm hmm 
 
 

Benjamin James Kuper-Smith: Um, there's always some sort of way it comes back. Uh, it doesn't mean maybe I shouldn't be a bit more effective sometimes, but, a bit more straightforward in the approach, but yeah, I know what you mean. [01:19:00] Yeah, I mean, it just gives you a completely different perspective, right? 
 
 

Rachel Bedder: Yeah, 
 
 

Benjamin James Kuper-Smith: Final question. Uh, well, this one's a little different. So the question is usually advised for PhD students, postdocs, people on that kind of border like me. But I thought today, uh, we could do it slightly differently because you had a blog post called five tips for managing yourself during a PhD. Uh, so I thought we could go through that a little bit and yeah, basically use, use that kind of as the framework for your answer. 
 
 

Uh, so you're, you had five points or five tips. Uh, the first one was ask questions and don't be afraid to be wrong. So, 
 
 

Rachel Bedder: yeah 
 
 

Benjamin James Kuper-Smith: what does that mean? 
 
 

Rachel Bedder: So lots of these, now I look back on this blog post, which is a few years old, lots of them are very much inspired by things I wish I had done, or things I, my peers didn't do. Um, so ask questions is always, you know, often, and many people are plagued by this idea that I don't want to ask a question in a talk because it might seem stupid or too basic in some way, and you're, you [01:20:00] know, you might be newly in a particular lab or department, and you're very intimidated by the knowledge of people around you, um, and it stops you asking questions. But often I've noticed is the people who ask the most questions, even when they're really simple clarification questions, um, you said this. Did you mean this or did you mean this? Or can you go over that part again? They're the people who really deeply understand things in the end. And it can seem quite intimidating to ask those questions, but they're so valuable in terms of like getting everything you can from a talk. Um, and also as someone who gives more talks now than I used to, there's nothing I love more than like a nice, simple question, which shows me that people are listening to what I'm saying and they want to understand it. That's, that's wonderful as a speaker. I often think. For myself, one of the criterias I hold for like, did that talk go well or not, was did a junior person ask a question? If no junior people asked a question, then I wasn't clear, and I didn't create an environment where they [01:21:00] felt they could. I think that's a really important standard to think about. 
 
 

Benjamin James Kuper-Smith: Yeah, that's a good heuristic for giving talks. Okay, so you don't, so you don't prefer the this is more of a comment than a question kind of questions. 
 
 

Rachel Bedder: not. Um, does anyone? Um, 
 
 

Benjamin James Kuper-Smith: well, I guess people with comments to make. 
 
 

Rachel Bedder: yes, of 
 
 

Benjamin James Kuper-Smith: Uh, but yeah, um, but yeah, it's really, yeah, it's really interesting you mentioned, um, I mean, The whole, you know, part is about not being afraid to ask questions. One of the. Um, when I interviewed Adam Mastroianni, he had a, uh, the book or paper he recommended was a blog post called Scientific Virtues by Sly Maud Ty Maud. 
 
 

And that's basically a, um, I've read it now. Uh, that's a post that basically advocates for, if you want to be a scientist, you should have these scientific virtues that are often the opposite of what people think [01:22:00] you should be as a scientist, and one of the, I agree with the post at some point, so I wouldn't wholeheartedly agree, but it's definitely interesting to think about because it gives you lots of counterexamples from the past. 
 
 

Scientists who've said certain things that really are quite different than what you expect and one of them is be stupid or dumb or something like that. And they use examples from like people like Niels Bohr who everyone said just asked the dumbest questions you could imagine and you always felt like he just didn't get anything but seems to have gotten something. 
 
 

Um, so, uh, I'll link to that also. Lots of links this episode. I'll link to that one also because that I guess relates very closely to that. Point two, talk to people about your science while you're doing it, not just when you finish it, 
 
 

Rachel Bedder: Yeah, so this is, I think, actually important at all stages of your career, but especially when you're more junior, because I think this is a time you're often least likely to do it. So when you present your work to a small group, you might want to think, this needs to be completely finished and completely polished, because I need to show everyone what a, what a great [01:23:00] scientist I'm becoming. 
 
 

Um, but actually often that means lost out on some really valuable advice some way through sometimes how to do something a lot more efficiently or easily would have been really great rather than suffering through this, uh, convoluted version of the analysis you might've come up with. And also these complicated things you might come up with someone else's feedback can be really, really useful. Um, and the reason I, I put it as a, a tip is because I feel like at least I didn't do enough of this. I felt so. Um, I'm much like if I showed that I didn't understand things, I was showing that I don't understand anything. I thought if they can see, if they can see weakness here, that they're going to smell blood and think, well, she doesn't understand, you know, how to do a dynamic programming in the first year of PhD. 
 
 

How can we believe anything she ever said? She obviously doesn't understand anything, but this isn't true, right? When we see people asking good questions and wanting to understand things and showing when they're not sure, being really [01:24:00] clear about I don't know the answer to that or I need to work more on that actually I think helps you trust them as scientists a lot more rather than someone who is trying to blag it a little bit and show this really polished version of something. say very frankly, I think there's also a gendered aspect to this, um, possibly a racial one as well, or any minority group is that sometimes you feel you're not given the benefit of the doubt. If you show some, um, that you're still learning in some space and things, you often don't want to show that because then you'll think, you know, you're not, you're not given the benefit of the doubt sometimes that you might just not understand this one thing. So it can be, I think, very difficult for minorities and women in science at a junior part to really embrace. That idea of getting feedback and talking to people as they go. 
 
 

Benjamin James Kuper-Smith: but you still advocate for it even if you, I mean, are people actually not giving the benefit of the doubt, or do you think that's more like something that people think will happen, but it won't? Uh, 
 
 

Rachel Bedder: the idea that, even if it's an idea [01:25:00] you hold, it's not coming from nowhere, right? It's coming from experience and talking to other people who've had that experience. So you might not be correct in every instance, but, think as a generalization, there's something to it. I would still advocate for doing it because it does hold you back as a scientist if you're not Doing these things and it's very sad that some people may feel the need to hold themselves back because of these things, but you can seek out communities and a peer group and you know, your own personal network where you're comfortable asking these questions and doing these things. 
 
 

Benjamin James Kuper-Smith: third point is, yes, you can cold email people. Okay. 
 
 

Rachel Bedder: who just doesn't have a huge amount of respect for nonsensical implicit rules about how you would engage with people in an academic hierarchy. I'm like, why can't I just email this person? Like, why not? [01:26:00] not to say you can email senior professors absolute nonsense with, like, long lists of questions and Asking them to do things, but people always appreciate an email, you know, saying you enjoyed their paper and you have a question about it. 
 
 

Um, can you connect me with someone in your lab who could talk more to me about this? Or do you have any jobs available? We talked about that earlier, being implicit about that. Explicit, sorry, about that. Um, and I realized so many people won't email people outside their network because uh, that's not the done thing. And, you know, I always say what's the biggest harm? Like, they'll ignore your, ignore your email. Um, actually I did see on Twitter recently someone reply, or someone post a reply, a professor saying like, uh, something like, Why do you think I would have time to answer emails from someone I don't know? Which, that's the worst scenario we're all imagining, right? Is that email back and be like, how dare you? But you've gained information in that sense, as you do not [01:27:00] want to work with that person or go anywhere near them ever again. So you can put that on its head that you've found out that's not a person you want to necessarily try and engage with in your research career. would like nothing more if people cold emailed me being like, cool paper, Rachel. So if anyone wants to, but feel safe and 
 
 

Benjamin James Kuper-Smith: Yeah. Yeah. 
 
 

Yeah, uh, you can put that on your website as like the first thing, just please tell me you like my work. Something like that. Um, yeah. I mean, I, I, I completely agree, um, also with the caveat that like, you know, be respectful. Um, and don't just, I've also seen people like anonymously post things where people like, yo, and then, you know, to one of their professors or whatever, and it's just like, yeah, I mean, be respectful and professional, but, uh, yeah, I know it's even, 
 
 

Rachel Bedder: But so to me, that's the kind of respect [01:28:00] that goes both ways, right? So I was never advocating it. We shouldn't be respectful to people at all levels of the hierarchy, but being like overly having an additional level of things which stops you from doing anything. Like, for example, I would not email a professor I don't know saying, yo, what's up? 
 
 

But at the same time, I would not appreciate an email from a professor to me saying, yo, what's up? If they, if they don't know me. So that I kind of respect as a bi 
 
 

Benjamin James Kuper-Smith: yeah, 
 
 

Rachel Bedder: right? 
 
 

Benjamin James Kuper-Smith: yeah, and I guess maybe also as part of this from my own experience, like it's also completely normal to be completely nervous 
 
 

Rachel Bedder: Mm 
 
 

Benjamin James Kuper-Smith: whilst you're doing this, I guess by now I've kind of gotten used to it, but, I mean, even like, I mean, before the podcast, anyway, podcast makes it a bit easier because I can invite people to a thing and potentially that gives them some benefits. 
 
 

And, but even there in the beginning, it just, you know, spending so much time with these emails and it was obviously even more before that. So I guess it's also just normal part of the process to be intimidated by people who you think do really cool stuff. Uh, so yeah, mail them anyway. 
 
 

Rachel Bedder: Yeah. And may I add [01:29:00] one thing to that? If you don't get a reply, it's okay. Take it. Millions of emails a day. You can read nothing into the fact you didn't get a reply. Yeah, so don't know, if it's really important, like, do you have any jobs? Send another email. But if it's not so much, you can, you know, maybe email someone in their lab that you might have a more specific question for. 
 
 

Benjamin James Kuper-Smith: Yeah, definitely. Uh, point four, set clear and consistent boundaries about your availability. 
 
 

Rachel Bedder: I think sometimes we think of boundaries as walls. Like, big cement brick walls. they aren't. Boundaries are like bricks with no cement. You constantly have to be rebuilding them when they get pushed out a bit and rebuilding them again. So this is an ongoing that will happen, uh, throughout your career. 
 
 

And I recommend you start early, which is letting people know through your behavior, what is a boundary for you? So for example, [01:30:00] prefer not to receive emails asking me to do something outside of work hours. That's not to say, you know, you can't ask me to do something tomorrow, but I'm not doing anything at 7pm. And the way I enforce that boundary is by not replying to that email, right? I've shown someone how to treat me. So, if you're replying to emails at 7pm, 3am, doesn't teach anyone about your boundary, and they will continue to send you emails at that time because you have not expressed this boundary to them. So early on, be I mean, of course, always be respectful with these things, but up how you want to be treated can be really important, um, what timelines want for things. For example, if you're someone who will If someone emails, and this admin, so I learned this from my research assistant job. If you're someone who can be emailed and be like, I need this in one hour, and you know that person knew [01:31:00] about this thing for like weeks and just didn't tell you, um, and you get it done in that hour, what you've, uh, think about it from a reinforcement learning context again, you've incentivized this behavior of I can wait till the last minute. So showing people, you know, I can't do it in an hour, but I can do it at this time. Tells people they need to give you more notice in future. you have to do this again and again and again, you have to be consistent, that's what I meant by the boundary is really just a wall that is crumbling and you have to put the pieces back up all the time, which is exhausting, because you're advocating for yourself, which can be scary, it does come with a lot of benefits, and I recommend people start early on, because once you've shown that you don't have these boundaries, it's very hard to build up the walls afterwards. 
 
 

Benjamin James Kuper-Smith: Yeah, definitely. Yeah. So it's more like a sand wall when the like waves wash against it and 
 
 

Rachel Bedder: Yeah, 
 
 

Benjamin James Kuper-Smith: to like restock it everywhere 
 
 

Rachel Bedder: A 
 
 

Benjamin James Kuper-Smith: once in a while. 
 
 

Rachel Bedder: so, uh, I don't know, the image is more Like it's [01:32:00] never going to work. It's 
 
 

Benjamin James Kuper-Smith: Yeah. That's too crumbly. 
 
 

Rachel Bedder: I like 
 
 

Benjamin James Kuper-Smith: Yeah, exactly. 
 
 

Rachel Bedder: cement. Cause you're like, sometimes it will be, look like a great wall, but then there'll be a gust of wind and it'll make a little hole and you put the hole up again, um, but you know, you can use whatever visual analogy you like, as long as you're putting up those boundaries. 
 
 

Benjamin James Kuper-Smith: Okay. Well, at least as long as you're aware that doing so is going to be beneficial to you. Um, point five and the final one have shorter time. Have short time goals outside your research projects and use them to build confidence. 
 
 

Rachel Bedder: So we talked a bit about this earlier in terms of these, um, the painting I do and also the artistic projects I do that, you know, aren't so successful. So science can be. for two broad reasons, right? One of them is the act of doing science is hard. It's messy. It's stochastic. Things don't always work out. 
 
 

We're dealing with ungraspable things sometimes, especially in psychology constructs that we have to find a way to measure. It's also hard because the [01:33:00] incentives of academia are to manage sometimes. So it can be very hard to live. In a healthy way, when you put your entire self concept around these two things are very to deal with and messy, like, you're gonna have setbacks in academia, in science, all the time, you have to be used to these things. But if your whole self concept is built around that, that can be very hard to emotionally take. So having these other things in life that you get joy from, that you can feel confident in can be such a good buffer against these times when, know, when science is hard for one of those reasons. think it really helps you stay in the field if you've got this buffer, this emotional buffer against these setbacks, it can help you stay in science for longer, um, in a healthy, in a [01:34:00] healthy way. 
 
 

Benjamin James Kuper-Smith: Yeah, I mean, one thing I also find interesting about the, um, the short time goals, um, about that aspect about it is that I hadn't planned for this to happen, but my podcast has really, like, allowed me in a way, um, could also be negative, but let's, framing it positively, it's kind of allowed me to have, be a bit more kind of about my scientific work, because I guess when I started my PhD, I really wanted to like get papers out, you know, like, I want to have something to show and publish stuff and that kind of stuff. 
 
 

And then since I've been doing the podcast, which is, you know, like, I mean, at times I had like no episode for two months or something. But it has been pretty much like an episode every two weeks or so, let's say, I kind of have gotten rid of that, like, I'm, I'm putting stuff out there regularly. So now, like, I don't feel that need for my Research where I'm fine now saying like, okay, I know we could finish it now, but like, let's just take another two weeks to do something. 
 
 

Now, maybe I take that a bit too far sometimes, but, um, [01:35:00] yeah, I think having, especially goals that are on the scale of at most weeks rather than months, uh, is really beneficial because yeah, research projects are just going to take at least a year, basically. I mean, you can have short ones, especially if you collaborate a lot, but it's going to take at least months. 
 
 

Ideally. Yeah, 
 
 

Rachel Bedder: wrote at least a year, knocking on two years now, but, um, having these like short rewards in your life can just be so beneficial. And like I said, if the goal is to stay in academia, you need to find a way to do it in a sustainable way where you can have a healthy relationship with your work. And ideally this makes your work even better, right? That's the, that's the hope anyway. Mm hmm. 
 
 

Benjamin James Kuper-Smith: exactly, we'll see. Okay, anything else you want to add? 
 
 

Rachel Bedder: Um, no. 
 
 

Benjamin James Kuper-Smith: Okay, perfect. Well, [01:36:00] then, thank you very much. 
 
 

Rachel Bedder: Thank you very much. Thank you for having me. It's been a really fun conversation.