[Comp-neuro] PhD student position in Decision Making, Sorbonne Universite, Paris (B. Girard, O. Davidenko)
benoit.girard at isir.upmc.fr
Tue Sep 18 17:40:17 CEST 2018
Fully funded PhD: Cognitive and computational neurosciences studies to
decipher how the social context affects individual food decisions
Institution: Sorbonne Université (Paris, France)
Disciplins: Computational Neurosciences, Behavioral Neuroscience, Brain
Hosting laboratory: ISIR, Institut des Systèmes Intelligents et de
Benoît Girard (ISIR, UPMC/CNRS) , supervisor
Olga Davidenko (INRA/AgroParisTech/UPSay PNCA), co-supervisor
Starting date: fall 2018
Key Words: Reinforcement learning, Human eating behavior, Brain imaging
Scientific environment of the project:
Excessive intake of sugars and salt are linked to deleterious
consequences on Human health. Current strategies to design and favor
healthier options have limited effects due to their poor acceptability.
SHIFT proposes a multidisciplinary approach, combining life, social and
computer sciences to understand the determinants, mechanisms, and levers
to modulate the acceptability of a meal option. This project will
address precise alimentary situations where margins of improvement with
respect to salt and sugar reduction are possible, ie. acceptability of
(i) water or low energy beverages in substitution of sodas and (ii)
sugar- or salt-reduced options at the end of main meals.
Interdisciplinary research activities will be conducted at the
populational, contextual and individual scales to decipher criteria
driving the acceptability foods. SHIFT gathers 5 academic partners: The
“Physiologie de la Nutrition et du Comportement Alimentaire” lab (UMR
PNCA; INRA-AgroParisTech-Université Paris-Saclay, project leader), the
“Institut des Systèmes Intelligents et de Robotique” (ISIR, Sorbonne
Université-CNRS-INSERM), the “Mathématiques et Informatique Appliquées”
lab (UMR MIA; INRA-AgroParisTech-Université Paris-Saclay), the
“Alimentation et Sciences Sociales” lab (ALISS; INRA), the School of
Psychology of the University of Birmingham and an industrial partner:
Danone Nutricia Research (Global Nutrition Department)
In the selection of foods as in many other decision processes,
individuals tend to conform to social norms (i.e. a
collectively-established acceptable behaviour). For instance, when
eating with others, if individuals are offered to choose between two
foods, they will take the decision that conforms to the choices made by
the others guests. In some cases, for instance when all guests need to
make a choice within a very short time lapse, there might be an
uncertainty on this social norm. Before making their decision,
individuals infer the appropriate choice from the available information
of the choices made by others and after taking their decision, subjects
judge the appropriateness of their own decision the basis of the overall
choices of commensals. The acceptability of a food option is modulated
by whether or not it is imitated by the commensals.
The aim of this thesis is to explore how the social context affects
individual decisions. The student will study the question from the point
of view of the individual: what are the contributions of various types
of social modulation (imitation, conformity to a social norm) to food
choices? Fundamental to every decision is the brain’s ability to
internally evaluate subjective values. The theoretical framework of this
thesis, the “value-based decision-making” (Pessiglione et al., 2006,
Johnson & Ratcliff, 2014), suggests that, although from an external
point of view, the outcome of a decision can be reduced to a binary
phenomenon (acceptance or rejection) the underlying mental computation
of the value is represented as a continuous function. Our working
hypotheses are that the aforementioned social modulations to food
choices can be integrated in an individual-centric modeling approach of
decision-making (as has been shown in simpler contexts by Burke et al.
(2010)), that the influences of these various types of modulations can
be disentangled, so as to help identifying the possibly specific neural
substrates. In this thesis, we will design a behavioural task dedicated
to investigate the different causes of social modulations, and we will
then explore which brain circuits are involved, using fMRI. We will use
the theoretical tools of value-based decision-making to model the
mechanisms by which social modulations affect decision. This will allow
us to derive time series of internal variables of the model as
regressors to refine the analysis of fMRI data.
The student will work to establish a fully specified model of a decision
task incorporating mathematical formalizations of the social
modulations. The model will be based on the so-called model-free
reinforcement learning algorithms (Sutton & Barto, 1998), to which
additional value update modalities will be added in order to take into
account the influence of commensals.
Burke, C.J., Tobler, P.N., Baddeley, M., and Schultz, W. (2010). Neural
mechanisms of observational learning. PNAS 201003111.
Johnson, E.J., and Ratcliff, R. (2014). Chapter 3 - Computational and
Process Models of Decision Making in Psychology and Behavioral
Economics. In Neuroeconomics (Second Edition), P.W. Glimcher, and E.
Fehr, eds. (San Diego: Academic Press), pp. 35–47.
Pessiglione, M., Seymour, B., Flandin, G., Dolan, R.J., and Frith, C.D.
(2006). Dopamine-dependent prediction errors underpin reward-seeking
behaviour in humans. Nature 442, 1042–1045.
Sutton, R. S., & Barto, A. G. (1998). Reinforcement learning: An
introduction. MIT press.
Applicants should send a CV, letter of motivation (max 2 pages), and
references via e-mail to benoit.girard at sorbonne-universite.fr and
olga.davidenko at agroparistech.fr, with [SHIFT] in the subject of the
mail. Review of applicants will begin immediately, and will continue
until the position is filled. The earliest start date is November 2018.
Benoît Girard, ISIR, UPMC/CNRS
Pyramide T55/65, CC 173
4 place Jussieu, 75252 Paris Cedex
Submit your replications to ReScience: https://rescience.github.io/
More information about the Comp-neuro