Professor of Electrical and Computer Engineering at Auburn University
“I was a lifelong chain smoker. Nothing in the world could stop me from smoking. Then, one day I had brain injury and my Insula was damaged. When I woke up, it felt like the urge to smoke had suddenly disappeared. It was as if a switch had been turned off. I could not believe what I was experiencing”
- By an anonymous ex chain smoker
Gopi Deshpande (GD): I am glad to be here talking to you about your work and your journey during your academic career. My first question to you is that if you are speaking to a non-scientist, how would you describe your research and what you do for a living?
Sarah Genon (SG): When I try to explain what I do to non-scientists, I explain that I use neuroimaging data to understand how the brain is organized, and then how this organization in turn relates to behavioral function in humans. And usually to make people aware that we need data, I also tell them that we work with big datasets that have been acquired all around the world, mainly in the US and in the UK.
GD: Do you feel that most non-scientists understand the complexity involved in this?
SG: Yes, I think most people are really aware of the complexity of the brain, how it is difficult to understand how the brain works, and how it relates to the complexity of human behavior. They usually find it fascinating
GD: Coming to your journey in this field, I saw that you started your undergrad as a marketing major. Then, you changed your major to psychological sciences, and did your masters and PhD in psychology.
Interestingly, you went back to school after becoming a PI to do a master's in computational statistics. You have had a very diverse background and different types of experiences. How do you think this has enriched your outlook towards functional neuroimaging?
SG: I think studying marketing made me aware of the importance of considering the global situation or the market. Really early in the first semester of marketing, I was convinced that this was not for me. In contrast, my statistical training directly contributed to my research, because it gave me a feeling of the structure of the data. I would say that my background in psychology gave me a feeling of the nature of the data that we work with, and my background in statistics helped me to investigate and understand the structure of the data that we work with.
GD: I think that is really important for a multidisciplinary field like this, right?
SG: Yes, yes. Especially when we are dealing with large data sets as we do.
GD: One of the research themes on which you have had a large impact is brain-behavior relationships, where you do a lot of churning of the data. What exactly triggered your interest in this particular area, as compared to many others that are out there?
SG: During my training in psychology, I went into research by looking at psychometric data in patients trying to understand how different behavioral aspects relate to each other. Later, during my PhD, I went into neuroimaging data. So it kind of followed naturally to understand the brain by relating neuroimaging and psychometric data.
GD: Moving forward, I feel that the challenge in uncovering strong neural predictors of behavior is probably also a problem of minimizing non-neural sources of variability in the data. I feel that people really don't appreciate this aspect fully. For example, hemodynamic variability is not accounted for in resting state connectivity analysis and a majority operate as if fMRI data is neural data. Therefore, I feel that it might still take some time to really address this challenge. I would like to know what you think is the timeline that would probably be involved in solving these challenges.
SG: I definitely agree with your point. Having worked with many different datasets of neuroimaging and behavioral data, my feeling is that we do not get as much relevant signal (in a statistical sense) as we hope for from the data for identifying robust neural predictors of behavior. In terms of timeline, I assume that we may still need a decade to address this challenge.
GD: Yeah. Which means that I think the field has a lot of potential going forward, right?
SG: Yes, there's still a lot of work to do.
GD: Another aspect of research that I found interesting was about your thoughts on the replicability crisis. I have read your thoughts in your paper on the analysis of a single neuroimaging data set by multiple teams. That paper (https://www.nature.com/articles/s41586-020-2314-9) really lays out what the problem is (which is the use of different analysis pipelines by different researchers), but what do you think are the barriers to standardization of analysis pipelines?
SG: I see the argument for a standard pipeline. However, from my experience with trying to use a standardized pipeline with different datasets is that a standard pipeline would be optimized for certain datasets, but not for other datasets. You would still have reproducibility issues. This is something we see a bit with the HCP data, which is high quality data and the pipeline that is suggested for this data works very well with these data, but they might not work very well with other datasets. Therefore, I think it is a complex issue.
GD: That point is well taken. The question then is how do you solve the replicability crisis. I think there have been two alternatives that have been proposed. First, Marek et al Nature paper (https://www.nature.com/articles/s41586-022-04492-9) that recently came out that suggests that we need sample sizes in the 1000s and 10,000s. The other alternative, which many people have been pushing for, including myself, is that we really need to reduce non-neural sources of variability in the data. I think both could help. What are your thoughts on this issue?
SG: Yeah, I see the argument for a larger data set in the sense that we have really small effect size at the moment, to be honest, across many different behavioral aspects and neuroimaging modalities. So in that sense, to see robust effects, we need really large data sets. However, from a broader perspective, what we need is also more relevant signal in the brain data and in the behavioral data. So there I agree with your point that we need to find ways to extract the relevant information or the relevant features from the data, or to find different acquisition methods that will give us more relevant signal to correlate.
GD: From the technical aspects, let us move on to your keynote talk during the annual OHBM meeting. Congratulations on the honor. Could you give a brief preview of how you got started with OHBM, and your journey so far towards becoming a keynote speaker.
SG: I very well remember my first OHBM meeting in Barcelona in 2010. I had started my PhD at that time. I was studying memory processes with neuroimaging techniques in Alzheimer's disease. I saw an inspiring talk by Eleanor McGuire. I was fascinated by the plasticity, the vulnerability, and the behavioral functions associated with the hippocampus. Now a decade later, I am going to talk about the complexity of the hippocampal organization and the challenges in brain-behavior mapping as a keynote speaker.
It is interesting from this historical point of view, how 10 years ago, I was listening to Eleanor McGuire’s talk, and I find myself in that position today. I also remember that during my first OHBM meeting, I made a lot of friends, and many of us actually happen to still be around at all OHBM meetings. Therefore, I think OHBM meetings are really exciting, not only from the scientific side, but also from the social and the personal side.
GD: I agree with you and I am glad that there is an in-person option this time to enable that social interaction.
SG: Yes, because social interaction is important to meet new people and to discuss science in a relatively casual way.
GD: I completely agree with that. Staying with the topic of your journey, you have published in many high impact journals relatively early in your career. So who are the people who trained you and inspired you, the people behind your success? And does your journey hold lessons for the young trainees and the next generation who might be at the OHBM meeting this year, listening to your talk, and aspiring to be in your position a decade later?
SG: I had the chance to do a postdoc in Simon Eickhoff’s group. As a young, dynamic, open minded and available leader, he was really a great mentor for me. This postdoc position was also in Katrin Amunts’ institute. She has a very large overview of brain research and I immensely benefited from that. In addition, I had the chance to collaborate with very kind and thoughtful senior researchers such as Thomas Yeo, and that experience gave me the confidence in developing my own views and opinions. For the younger generation, I would say that it is important to look beyond your own view. Your viewpoint might appeal to you now, but it could appear to be completely different when seen from a different standpoint. It is a question of broadening one’s view and OHBM meetings help in that process.
GD: You are a psychologist, and you might appreciate how our biases box us into thinking in one way.
SG: If you go to a psychology meeting in your specific field, people will have their model and then you will just see this view of each person advocating his or her own model. But at OHBM, it is more of a broader community, which is not focused on a specific psychological process, but has a broader neuroscientific view.
GD: Could you share some of the challenges that you have experienced in your research career?
SG: My challenges have been similar to what others have experienced in their research careers. The first one is the difficulty in finishing your PhD. It is like a cycle in which you acquire data, and then you do your analysis. Then you do more analysis, which reveals you need more data, and the cycle seems to continue endlessly.
Generally, in research, you need a lot of resilience against rejection. Rejection of paper, grant proposal, application for postdoc position and so on. Nevertheless, as we say, what doesn't kill us makes us stronger.
GD: My final question to you. You are a mother of a one year old, and we all know that this can be very demanding in terms of time. So how do you balance work and life- related commitments? What is your advice to mothers out there with young kids wanting to be successful researchers?
SG: That is a very good question. To be honest, I think there is no perfect life-work balance, it will always be a challenge. Only thing that one can do is to be aware and be prepared to accept this challenge when it comes. The first advice that comes to my mind is to care about one’s mental health. I am a clinical psychologist by training, and I have worked with anxiety and depression therapeutic groups with a lot of women. I have read a lot of blogs and books written by mothers, and it is striking to me the pressure in which women are locked, mothers in particular. In a research career, there are a lot of pressures, of course, but a mother role also comes with a lot of pressures and what we really need is to find a way to keep a clear mind and face this pressure. I think one advice for that would be to identify and define, in advance, a set of top priorities on both sides - on the personal and professional side - and to stick to those priorities. This might mean that you might have to decline some commitments in order to avoid being overwhelmed, but to keep the focus on what really matters. However, one needs to keep in mind that one can create greater opportunities in the future when the right time comes.
GD: I think that is very good advice. I hope a lot of mothers who are interested in pursuing research careers would take heart from that advice. Thank you very much for sharing your thoughts with us and it was a pleasure chatting with you. I am really looking forward to your keynote address during the OHBM meeting in June. See you in Glasgow!
SG: Thank you and see you soon!