Therefore our analysis reveals that sheep decision-making is not based on a single mimetic effect but rather on a rule which balances two mimetic opposite effects: follow the departed individuals but remain with the non-departed individuals.
Then, from this departure until the next, the two still non-departed naïves witness a new group state which consists of the two departed individuals (the trained and the first follower) and one non-departed individual, and so on.
Moreover, for a given number of departed individuals, the departure rate decreased with the number of non-departed individuals (Fig. 2B).
These departed individuals are treated as living on in a social sense, reincarnated in the ancestral world where they continue to guard the living.
In fact, the experimental departure rate also increases with an increasing number of departed individuals (Fig. 2A), and increases sharply when the number of departed individuals is greater than half of the group, whatever the group size (Fig. 4).
For all group sizes, the departure rate increased sharply with the number of departed individuals: individuals were increasingly stimulated to depart as the number of departed animals increased (Fig. 2A).
Model 3 with both factors (departed and non-departed) is orders of magnitude (1400 times) more likely to be the best explanation for following rates compared to model 2 with departed individuals only.
For a group of fixed size N, it is advantageous to remain while the number of departed
To verify this hypothesis, we tested the relevance of this model against more parsimonious models: responding only to the trained sheep departure (initiation), or responding only to the departed individuals.
Model 2 allows a distribution of the fitted rates that is closer to the observed responses since they are allowed to reflect the stimulating effect of the number of departed individuals (Fig. 3D, α = 0.09, β = 1.4, r2 = 0.80).
A key feature was the use of groups of different sizes (N = 2, 4, 6, 8) so that sheep were exposed to various combinations of two factors: the number of departed individuals (including those sheep departing in response to the trained leader) and the number of non-departed individuals.
