Figure 10.

(a) The success rate of the three learners and a non-learner (random choice) in a low-entropy (highly ordered) network, with rare predictors of food (see related TP matrix in the electronic supplementary material, table T10). (b), (c) The success rate of the three types of learners and a non-learner (random choice) in a low-entropy environment with rare predictors of food, but with a high frequency of ‘f’ (b, 25 times greater than standard runs; see related TP matrix in the electronic supplementary material, table T11) or with a long training set (c, 16 000 characters; see related TP matrix in the electronic supplementary material, table T10).