How to drastically increase the sample-size of human-based datasets ? Feedbacks from a video-game - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2019

How to drastically increase the sample-size of human-based datasets ? Feedbacks from a video-game

Résumé

When scientists need to record signals from human participants such as their eye-movements or their brain activity, they need to ask them to physically come to a research facility. This process is costly and time consuming. As a consequence, the human sample size in a typical neuroscience paper is below 100, and the cohort often consists in students from the local university. This highly biased population, ubiquitous in the literature, has been named the WEIRD population : Western, Educated, Industrialized, Rich, and Democratic. It is one of the causes of the replication crisis that stroke the neuroscience community (see Power failure: why small sample size undermines the reliability of neuroscience. Button et al., Nature Reviews Neuroscience 2013). One way out of this crisis is to drastically increase the sample size and diversity of human-based experiments. Here we argue that video games can be a useful and cost-effective solution, as they qualify everyone with a computer, table or a smartphone (in 2017, 32% of the world population owned a smartphone). In this project, we developed a video game to build a planetary-wide dataset of how people navigate. Sea Hero Quest (www.seaheroquest.com) is a mobile gaming app that records users' sense of direction to determine how navigational skills decline with age. The game has so far been downloaded 3.7 million times, has been played in every country in the world and is the biggest dementia study in history. Through its innovative approach, it has collected a huge amount of data that would have taken more than 9,000 years to acquire in a traditional lab setting. We will use this data to create the world's largest benchmark of how humans navigate, which will then go on to become a critical diagnostic tool for dementia in the future.
Fichier non déposé

Dates et versions

hal-02348482 , version 1 (05-11-2019)

Identifiants

  • HAL Id : hal-02348482 , version 1

Citer

Antoine Coutrot. How to drastically increase the sample-size of human-based datasets ? Feedbacks from a video-game. Folle Journée de l’Imagerie Nantaise, Feb 2019, Nantes, France. ⟨hal-02348482⟩
72 Consultations
0 Téléchargements

Partager

Gmail Facebook X LinkedIn More