Disclaimer. Don't rely on these old notes in lieu of reading the literature, but they can jog your memory. As a grad student long ago, my peers and I collaborated to write and exchange summaries of political science research. I posted them to a wiki-style website. "Wikisum" is now dead but archived here. I cannot vouch for these notes' accuracy, nor can I even say who wrote them. If you have more recent summaries to add to this collection, send them my way I guess. Sorry for the ads; they cover the costs of keeping this online.
Prins. 1996. Ecology and behavior of the African buffalo. London: Chapman & Hall.
These notes cover only chapter 8, "Selecting Grazing Grounds: A Case of Voting."
Each evening, African buffalo arise from rest and move to a feeding ground. The decision as to where to feed cannot be made lightly; buffalo must know where they have fed lately, how quickly the grass is likely to grow back in each area, which area has the best grass, and where competitors (elephants, etc) are likely to have fed. This is a multidimensional decision, which Arrovians would expect to be subject to cycling.
From observing the Manyara herd, Prins concludes that a voting mechanism exists by which the herd, consisting of almost 1000 buffalo, makes its decision.
When it comes to to feed, the herd arises as one and immediately moves uniformly toward its chosen feeding ground for the evening. Apparently, by this time, a decision has already been reached about where to feed. But how is this decision made?
African buffalo rest during the hottest afternoon hours. From around 1700 to 1800--the hour before feeding begins--individual buffalo will stand for a minute or so, gaze in one direction, then lay back down. Only a handful of buffalo will be doing this at any one time. Not all buffalo participate--mainly the permanent (i.e. female) adult members of the herd, almost all of which participate. This suggests that there is some sort of voting process going on.
Prins recorded the direction in which each buffalo looked, and calculated a mean direction. As it turns out, when the herd arises to feed, it invariably moves in the mean direction of all these "votes." What's more, the votes seem to reflect a measure of consensus. Prins records the "vector length" for each evening's vote. If every adult cow looked in the exact same direction, the vector length would be 1; if cows looked in all sorts of direction, the length would be close to zero. Thus, by examining both the vector length and the mean direction, we get a feel for the degree of consensus in voting. Generally, the vector length was quite large, implying quite a bit of consensus.
Prins calls this behavior "voting." I suspect that "voting" is the wrong analogy, however. If this behavior were analogous to what we call voting, we might expect to see conflicting factions, or at least a lot less consensus in the initial votes. Instead, this behavior is more analogous to what we call "coordination games." The buffalo are indifferent as to which feeding ground they go to, as long as it's a good one and everybody else goes there too. The buffalo are merely coordinating on a direction to travel. Perhaps this is more similar to Congressional voting (where party members must coordinate on a procedural bill) than to popular voting (where two or more factions compete).
Prins seems to recognize this distinction. He sees the herd acting as an "information center"; by averaging across individual cows' preferences, the herd comes to an optimal decision. As Prins points out, this may be why younger buffalo and non-permament members of the herd (bulls) don't participate; they defer to the more experienced adult cows. I would point out that in human politics, this is analogous to a voter using a public opinion poll, or the advice of an expert panel, in forming her opinion about an issue. So the buffalo aren't necessarily "voting" per se, but they certainly seem to be pooling their information into a collective decision. This isn't unlike Galton's experience with a group of people estimating an ox's weight--see discussion of this in Conradt and Roper (2005).
Research on similar subjects