Expected decision accuracy as a function of the proportion of PLUS-animals among the decision-makers and as a function of the degree of conflict of preferences within a group. Note, that degree of conflict and the proportion of PLUS-animals are closely related to each other (both on the x-axes): the degree of conflict is maximal (conflict = 1) if the proportion of PLUS- and MINUS-animals is equal (i.e. at point 0.5 on the upper x-scales); the degree of conflict is minimal (conflict = 0) if the whole group consists either of MINUS-animals only (proportion = 0 on upper scale); or of PLUS-animals only (proportion = 1 on upper scale). The solid lines indicate decision accuracy of groups with conflict (i.e. groups consisting of PLUS- and MINUS-animals). For reasons of comparison, I also give the expected decision accuracies of Condorcet juries of similar size and expertise in the same environments as the conflict groups (dashed lines). Decision accuracies are given for the following parameter combinations of individuals' expertise, environments and number of decision-makers (i.e. group size). (i) Levels of individual expertise (low individual expertise: a, c, e, g, i, n = 3; medium individual expertise: b, d, f, h, j, n = 7; to save space, results for high individual expertise n = 25 are not shown but are qualitatively similar); (ii) environments are reflected by different distributions of the probability p(k) that option PLUS is the correct choice in the kth environmental situation (a, b: uniform distribution; c, d: symmetric unimodal distribution; e, f: asymmetric unimodal distribution; g, h: symmetric bimodal distribution; i, j: asymmetric bimodal distribution; details see table 2) and (iii) numbers of decision-makers (low number of decision-makers: black lines, m = 4; medium number of decision-makers: dark grey lines, m = 8; large number of decision-makers: light grey lines, m = 24). Upper scale: proportion of PLUS-animals among the decision-makers and lower scale: degree of conflict between decision-makers.