The capacity to integrate information is a prominent feature of biological brains and has been related to cognitive flexibility as well as consciousness (e.g., see Tononi et al., 2016). To investigate how environment complexity affects the capacity for information integration, we employed evolutionary simulations of artificial organisms (“animats”) controlled by small, adaptive neuron-like networks (“brains”). By applying measures of information integration, we showed that, under constraints on the number of available internal elements, animats evolve brains that are more integrated when more internal memory is required to solve a given task in their environment. This shows that, with regard to complex environments where there is a premium on context-sensitivity and memory, integrated brain architectures have an evolutionary advantage over modular ones. More specifically, integrated systems with greater internal complexity are capable of better capturing the causal structure of a rich environment, leading to better performance and greater fitness.
Directly below you will find a link to two simulators, each designed to better illustrate how animats evolve in an environment. The first simulation shows the evolution of an animat’s network and ability to integrate information (see ‘Phi’ graph) in a particular task environment. The second simulation shows one fully-evolved animat’s task performance while simultaneously showing the architecture and activation of that animat’s connectome.
Albantakis, Larissa, Christof Koch, Christopher Adami, and Giulio Tononi. “Evolution of integrated causal structures in animats exposed to environments of increasing complexity.” PLoS computational biology 10.12 (2014): e1003966.
Related work on animats
Albantakis, Larissa, and Giulio Tononi. “The intrinsic cause-effect power of discrete dynamical systems—from elementary cellular automata to adapting animats.” Entropy 17.8 (2015): 5472-5502.
Fischer, Dominik, Sanaz Mostaghim, and Larissa Albantakis. “How swarm size during evolution impacts the behavior, generalizability, and brain complexity of animats performing a spatial navigation task.” Proceedings of the Genetic and Evolutionary Computation Conference. 2018.
Albantakis, Larissa. “A tale of two animats: What does it take to have goals?” Wandering Towards a Goal. Springer, Cham, 2018. 5-15.
Juel, Bjorn, Renzo Comolatti, Giulio Tononi, and Larissa Albantakis (2019). When is an action caused from within? Quantifying the causal chain leading to actions in simulated agents. arXiv preprint arXiv:1904.02995.
Fischer, Dominik, Sanaz Mostaghim, and Larissa Albantakis. “How cognitive and environmental constraints influence the reliability of simulated animats in groups.” PLoS one 15.2 (2020): e0228879.
Albantakis, Larissa, Francesco Massari, Maggie Beheler-Amass, and Giulio Tononi. “A macro agent and its actions.” Top-Down Causation and Emergence. Springer, Cham, 2021. 135-155.
Albantakis, Larissa. “Quantifying the Autonomy of Structurally Diverse Automata: A Comparison of Candidate Measures.” Entropy 23.11 (2021): 1415.