Julien RENOULT.

Tenure CNRS Research Scientist (CR1)
​PhD Evolutionary Biology & Ecology
Doctor in Veterinary Medicine
CNRS - CEFE UMR5175
1919 route de Mende
34393 Montpellier 5
​FRANCE
​julien[dot]renoult[at]cefe[dot]cnrs[dot]fr
+33 (0) 4 67 61 32 10
links: CEFE - ​Researchgate

Who am I.

I am an evolutionary biologist and a natural historian. As a scientist my main goal in to understand how do evolve complex and extravagant visual signals such as the eyespots of the peacock, the danse of the birds of paradise, and many human artworks. My research hypothesis is aesthetics. Using approaches of empirical psychology, computational neuroscience, artificial intelligence and philosophy, I am working at unraveling the biological bases of aesthetic experiences and at developing mathematical models of these experiences in humans. I then transfer these models to other animals to understand how aesthetics influence behaviors and, ultimately, the evolution of communication signals in nature.
I am also interested in methods to characterize visual phenotypes, with applications in community ecology, evolutionary biology and systematics. The methods I am developing are mostly inspired by the very active field of computer vision and include color spaces, machine learning and deep learning. 

Evolutionary Aesthetics.
"What is the role of beauty in the evolution of communication signals such as sexual displays?". Answering this question requires understanding the biological, i.e. neuro-physiolophical, bases of beauty and of the aesthetic experience. My approach is inspired by the fluency theory of aesthetics developed in human empirical psychology, which states that stimuli we find attractive are easy to process in the brain. In order to understand the role of aesthetics in natural communication systems, we thus need to model the "easy to process" part of the theory. With the collaboration of Dr Frederic Geniet and Dr François Molino from University of Montpellier, we are achieving this using artificial intelligence approaches, and more specifically Deep Convolutional Neural Networks (CNN), adapting metrics of information processing (sparseness, mutual information, information bottleneck theory) to estimate the efficiency of processing individual communication signals.

We are applying these models of efficient processing to different questions and communication systems:​
. "How much aesthetics has influenced the evolution of darter coloration?" Darters are colorful freshwater fishes inhabiting North American streams. With Pr Tamra Mendelson and Samuel Hulse, from University of Maryland, we are using comparative and behavioral approaches to study how features the environment determine the efficiency of processing male displays for females, and thus the perceived beauty of these displays.
. "Is sparseness a determinant of facial attractiveness?" A sparse stimulus is a stimulus that activates only a few neurons when processed by the brain. Sparseness is thus one measure of processing efficiency. With Dr Michel Raymond from the Institut des Sciences de l'Evolution de Montpellier (ISEM), using modeling approaches we are analyzing how the attractiveness of faces is influenced by the sparseness of their encoding in different brain areas.
. "Can CNN predict the beauty-driven benefits of being a prototype?" Prototypes are the most representative stimuli of they perceptual category; e.g., feminine women are the most representative of the "women" category as opposed to the "men" category. Prototypes are efficiently processed in the brain, and as predicted by the fluency theory of aesthetics, people like prototypes. A CNN trained to recognize stimuli also makes prototypes. Do CNN prototypes ressemble real-brain prototypes? Are stimuli resembling CNN prototypes perceived as beautiful? Using both modeling and behavioral experiments, with Bastien Guyl and Dr Marie Charpentier from the Institut des Sciences de l'Evolution de Montpellier (ISEM) we are investigating these questions in mandrills, analyzing how the femininity of females predicted by CNN explains socio-sexual benefits in this species.  

Computer Vision.
My second main research area is the development of methods of computer vision to study the ecology and evolution of visual communication. Examples of application are: 
. Quantifying the frequency of mimicry. With Thomas de Solan, Dr Pierre-André Crochet and Dr Patrice David, from the Centre d'Ecologie Fonctionnelle et Evolutive (CEFE) of Montpellier, we are using deep neural networks to measure the ressemblance between colubrine snakes and vipers of the Western Palearctic to test whether the former are Batesian mimic of the later, and to quantify the frequency of mimicry in this system.  
. Facial similarity as a cue for kin selection. With Dr Marie Charpentier from the Institut des Sciences de l'Evolution de Montpellier (ISEM)​, we are using deep neural networks trained for face recognition to measure facial similarity in mandrills (Mandrillus sphinx), and to study how this similarity explains behavioral biases toward kins in that species. 
. Identification of distinctive features for species recognition in the field. We use CNN and feature mapping techniques (e.g. Layer-wise Relevance Propagation) to identify morphological differences between species. We are currently applying these techniques to the problem of identifying African and American Royal Terns (Thalasseus maxium albididorsalis and T. m. maximus).

Natural History.
My research is inspired by Nature. From an evolutionary ecologist, this statement may sound trivial. Unfortunately, however, I am observing a growing gap between methodological and conceptual developments motivated by the quest for hyper-generalizable meta-global-results explaining why the natural world is in a bad state, and real natural facts. I am trying to fight this trend by going in the field, and by promoting natural history.

I am contributing to, and administrating the Fish Watch Forum - a database of Mediterranean and North-Eastern Atlantic fishes.

I am also curator of the numeric museum and citizen science database iNaturalist, and a dedicated contributor, having submitted photographic records of more than 5,000 species of plants, animals and fungi from over the world.
​