Link Search Menu Expand Document
COSMOS

Group projects

COSMOS 2025 Group projects

You can find a more detailed overview about each project here

P1: Wisdom or madness of crowds? (based on Toyokawa, Whalen & Laland, 2019)

Investigate human choice behaviour in a three-armed bandit task to see when social learning is helpful and when it isn’t

  • Beginner friendly
  • Model code written in R
  • Data and code are here

P2: Social learning in correlated environments (based on Witt, Toyokawa, Lala, Gaissmaier & Wu, 2024)

Investigate how humans integrate social information when it is positively correlated in a multi-armed bandit task

  • Advanced
  • Model code written in Python
  • Data and code are here

P3: Do collectives appear to have memory? (partly inspired by homing pigeons, e.g. Collet et al. 2021, but also other problems)

Develop a navigation simulation, compare individual vs collective runs and quantify the difference

  • Advanced
  • Early simulation script in R
  • See Damien Farine for the data

P4: Evolution of Social Strategies in a Lattice-Structured Population (based on Nakamaru, M, Matsuda, H, & Iwasa, Y, 1997)

Replicate the original study (Python code available) with varying group size (who each player plays the repeated Prisoner’s Dilemma game with, whose strategy each player imitates), updating rules, etc.

  • Beginner friendly
  • Python code available
  • See Mayuko Nakamaru for the data

P5: Teaching Multiple Agents via State Intervention

Develop your own web experiments where real-time multiple RL agents learn from interaction with the environment and physical intervention

  • Intermediate
  • Web experiment code provided in Javascript/Typescript; Simulation code provided in Python
  • Data and code are here

P6: Information integration on social network (inspired by Jiang, Mi, Zhu, 2023)

Investigate how network structures affect individual learning and decision-making

  • Intermediate
  • Model code written in Matlab
  • Data and code is here

P7: Conformist transmission in multilevel societies (based on Aplin et al., 2015 and Cantor, Chimento, Smeele et al., 2021)

Investigate the impact of network structures on transmission

  • Beginner friendly
  • Model code written in Python and R
  • Data and code are here)

P8: Observational learning (based on Morishita, Yadav, Murawski, and Suzuki, 2025)

Investigate which strategy provides the best behaviors prediction to account for learning from the experience of others

  • Beginner friendly
  • Model code written in Python
  • Data and code are here

P9: Inferring and predicting emotions (based on Houlihan et al., 2023)

Predict how people attribute emotions based on contextual information on how they play a real Prisoner’s Dilemma game, using information (images, information about their actions) from a live game show

  • Advanced
  • Model code written in Python
  • Data and code are here