Table 1.

Pay-off matrices for our experimental game. (A) Receiver's pay-offs: the receiver always does best by matching its behaviour to the state of the environment. The receiver obtains 1 unit of food (three pellets in the experiment) if it chooses the accept action in the true state, or chooses the reject action in the false state; otherwise it obtains nothing. (B) Signaller's pay-offs: if the state is true, the signaller obtains 1 unit of food (three pellets) when the receiver chooses accept. If the state is false however, the variables a and b determine whether the signaller's incentives are aligned with the receiver's. When a > b, the signaller does best when the receiver ‘accepts’ regardless of which state applies. When a > b, therefore, the signaller and receiver incentives are opposed when the state is false. In the reverse situation, when b > a, the signaller and receiver incentives are always aligned. In our incentives-aligned condition a = 0.33 and b = 1.0, while in our incentives opposed a = 1.0 and b = 0.33. (C) Signalling costs: if the signaller chooses to signal, it must pay a cost (c) determined by the current treatment (where c = 0, 1, 4 or 7 shuttle flights). Models predict stable honest signalling when (a − b) < c. In order to make correct comparisons, we assume that the parameters a, b and c are normalized to the value of a single unit of food.

		state
		true	false
(A) receiver's pay-offs
receiver action	accept	1	0
	reject	0	1
(B) signaller's pay-offs
receiver action	accept	1	a
	reject	0	b
(C) signalling costs
signal emitted	signal	c	c
	no signal	0	0