Embedded Files
This website contains real-time videos of the natural and lab environments used in our simulations and is meant as supplementary material that can help reviewers visualise the behaviours discussed in the main paper.
Experiments
Experiments
A) Observing populations in their natural environment
Video 1: Snapshot of 1000 timesteps of simulation in the 2 seeds ( Resources are green and agents are black, we changed the color palette compared to the paper for readability)
Video 1.a) Seed 1 timesteps 600 000 to 601 000
Video 1.a) Seed 1 timesteps 600 000 to 601 000
Video of 500 000 first steps : https://www.youtube.com/watch?v=LiTdUp8rOic
Video 1.b) Seed 2 timesteps 600 000 to 601 000
Video 1.b) Seed 2 timesteps 600 000 to 601 000
B) Observing populations in lab environments
1) Do agents adapt their behaviour based on the density of resources?
We introduce a single agent and deactivate reproduction to ensure there are no effects from other agents. We also deactivate resource regrowth to better control the amount of resources.
We evaluate in three types of environments: the low-resources environment has 10 resources, the medium-resources environment has 20 resources and the high-resources environment has 60 resources.
Agent with ID 758
Agent with ID 762
Low resources
High resources
2) Do agents adapt their behaviour due to peer pressure?
We keep resource regrowth deactivated but allow the agent to reproduce, so that every approximately 20 time steps a new agent may appear.
We evaluate in the same three environments, now interested in whether the presence of others leads to more consumption compared to the previous case. We look at the same two agents to facilitate comparison.
Agent with ID 758
Agent with ID 762
Low resources
High resources
Long video of 500 000 timesteps of simulation
Long video of 500 000 timesteps of simulation
Video of 500 000 steps of seed 1 (starting from random weights)
Video of 500 000 steps of seed 1 (starting from random weights)
Page updated
Google Sites
Report abuse