Skip to main content


Julien Céré

Julien Céré is a Team lead in data science.

Tell us a little bit about yourself and your career path. What do you do at Behaviour?

You might say it was the squirrels that brought me to Behaviour. My background is in biology and research, and my area of expertise is behavioural ecology. That’s the study of how animal behaviour is shaped by natural selection.  My master’s studies of squirrel conflicts over scarce resources ran into a brick wall because of a lack of data. At the beginning of my second summer of collecting data for my thesis, squirrels had become extremely scarce at my research site, so I quickly abandoned my initial thesis subject and migrated – successfully, this time – to a mathematical simulation.

Around the same time, a biologist friend (let’s call him Pierre-Olivier for now – we’ll come back to him later) and I became interested in video games as a research platform. The dynamics of many games are similar to what we observe in nature, and those behaviours are measurable thanks to the vast amount of data we can gather. No more problems with uncooperative squirrels!

So I began my Ph.D. and applied for an internship at Behaviour. Since then, I’ve worked in various positions related to video game data, and I’m now head of a small team of data scientists who develop predictions, carry out statistical analyses, conduct simulations, and design artificial intelligence, or AI, tools to assist the production teams. These AI projects can take the form of text mining tools, recommendation systems like the one used by Amazon, and predictions of when players will monetize or abandon our game.

What do you like best about data science? What are your biggest challenges?

It was curiosity that pushed me toward science. The first step of the scientific method is to ask the right question. So it’s as important for scientists to ask questions as it is for them to seek answers: it’s one of the few fields where curiosity is key! People come to data scientists with questions to answer or problems to solve. So you’ve got to ask the right questions and propose various hypotheses to come up with an appropriate solution to the original problem – whether that be an analysis or a tool. Finally, what motivates me about data science is to make the most of my curiosity and creativity to develop solutions that have a real impact on our products or the work of our collaborators. Data science is an exercise in collaboration: many projects require the synchronization of several disciplines, and that can be really challenging.

Tell us about your studies and why you decided to do a doctorate.

After finishing a contract in a lab where we were filming male crickets and wetas (giant crickets found in New Zealand) that were fighting over females, I found myself at a crossroads. As minuscule as it might be, the social impact of my work is a priority for me, so I opted to do a PhD in the hope that my (somewhat pioneering?) research would inspire other scientists. 

Dead by Daylight (DbD) is a game where a killer pursues and catches four survivors. So the parallel with predators and prey was obvious to me. Moreover, the behaviours of competition and cooperation (important mechanics in DbD) are also important subjects in the scientific literature in the field of ecology. In many respects, the evolution of cooperation between animals is still a mystery today. Animals want to optimize their survival and reproduction, so why would they help other animals at their own expense (loss of time or energy, risk of predation….) The answer to this question remains a mystery partly because it’s difficult, even impossible, to measure all the social interactions between animals in nature. Dead by Daylight, on the other hand, gives us reliable and complete measures of the interactions and behaviours surrounding altruism. So we can easily quantify the risks and benefits of helping other survivors.

So the goal of my doctorate is to study how cooperation behaviours can be compatible with natural selection. One of the hypotheses I’m testing is group selection: I take risks to help other members of my group since my group will then perform better than those that don’t cooperate. Another hypothesis is increasing the size of the group: I help others to increase or maintain the size of my group since there is strength in numbers, and I’ll be less likely to be captured by a predator. You can read my first article on the risks and benefits of playing with friends here.

What conclusions have you drawn from this study? Do you apply the results of your analyses in your work?

My research is more fundamental than applied, meaning that it’s exploratory, and you don’t always know where it will lead. At work, we’re more in the realm of applied research – trying to meet a real, tangible, sometimes urgent, need. But my results often contribute to discussions with the design teams by shedding light on the ways players interact and how they approach the gameplay. Generally speaking, it’s fairly common for the results of fundamental research to contribute to the development of applied solutions. With time, it’s possible that research such as mine will play a growing role in our decision-making processes. In fact, we’re working closely with the lab of Pierre-Olivier Montiglio (my above-mentioned friend!), who is now a professor and researcher at UQAM and devotes part of his group’s work to approaches using video games. And our own team of data scientists at Behaviour includes a research partner, Francesca Santostefano (post-doc in behavioural ecology), who works with us daily on questions of gameplay.  We’re also in communication with various communication and consumer behaviour research labs.

Do you have any advice to give us based on the results of this study?

People are often surprised to learn that I use video game data in my research on the evolution of animal behaviour. Since ecology is the study of interactions (as simple as that!), this was a no-brainer for me. The relationship between game design and ecology should be symbiotic (see what I did there?) since both fields have a lot to learn from each other.  Asymmetrical multi-player games are not easy to study: they are real ecosystems, and ecology suggests a series of approaches to address those challenges. Every scientific discipline aims to increase our knowledge of specific subjects using objective, proven techniques. So here’s my tip: when you’re facing a challenge or an uncertainty, it’s likely that one scientific discipline or another has already studied it, and the work of an entire community of researchers can help you make an informed decision!

Julien Céré
Team lead in data science