In flocking, a large number of individuals move cohesively in a common direction. Many examples can be found in nature: from simple organisms such as crickets and locusts to more complex ones such as birds, fish and quadrupeds. In this paper, we study the flocking behavior of a swarm of robots where information about two distinct goal directions is present in the swarm. In general, we can identify three different macroscopic objectives that we might want to attain: (a) a swarm that moves to the average direction among the two (for example to avoid the obstacle) without splitting; (b) a swarm that selects the most important of the two directions (for example the direction to avoid danger) and follows it without splitting; (c) a swarm that splits in a controlled fashion in the two directions (for example, in the parallel task execution case). This paper proposes a solution for the first objective: a method for moving the swarm along the average between the two conflicting goal directions. We show that this objective can be attained by simply using a similar methodology as the one proposed in earlier work. We execute systematic experiments using a realistic robotics simulator. In the experiments, a small proportion of robots is informed about one goal direction, another small proportion about the other goal direction, and the rest of the swarm is non-informed. We study the effect of what we believe are the critical parameters: the overall proportion of informed robots, the difference between the size of the two groups of informed robots and the difference between the two goal direction. We show that, using the proposed method, the system is always able to follow the average direction between the two.