File:Integrated-Information-Increases-with-Fitness-in-the-Evolution-of-Animats-pcbi.1002236.s005.ogv

English: This movie shows the trajectory of an evolved animat traveling through the maze after 2,000 generations of evolution in the top panel, and the inner workings of its Markov network brain in the lower panel. At this point in evolutionary history, the animat has learned to move forward whenever it stands in front of an opening, but otherwise performs a random walk. The fitness at this time point is of maximal. The animat in the maze is depicted with a triangle, and the trail it leaves reflects the activation pattern of its four internal nodes and its motor outputs, as described in Fig. 4 . The brain state (lower panel) shows all HMGs (U0–U10) and the probabilities in the state-transition tables as percentages (colored in shades of gray). Input bits (labeled iB) and output bits (labeled oB) are green if true and blue if false. The red element in each table indicates the element of the table selected at that time step based on the values of the input bits and the probabilities in that row. In other words, a table element turning red indicates which state of the HMG was selected as a function of the input. This is akin to a pattern of neuronal firings as a function of the inputs.