Notes on Pushing Five Buttons

By what principle might we decide the number of buttons to press in Caspar Hare’s Two Parallel Tracks – Six Buttons case?1 Here are some options:

  1. Expected Value Principle: You ought to maximize the expected number of lives saved. (consequentialist)
  2. Killing vs. Allowing Principle: You ought not kill one person, in order to save five. (non-consequentialist)
  3. Go for Glory Principle: You ought to choose that which maximizes the chance of the best possible outcome.

In Two Parallel Tracks - Six Buttons, these three principles all recommend mutually exclusive choices. The Expected Value Principle says you ought to press six buttons. The Killing vs. Allowing Principle says you ought to press zero buttons. And the Go for Glory Principle says you ought to press five buttons. Since there is no way to obey two principles at once, our decision will depend on which principle we think is morally best.

My contention is that the Go for Glory Principle leads one to absurdities. If I am correct, and the Go for Glory Principle is absurd, then barring the suggestion of a plausible new principle, one ought to act according to either the Expected Value (consequentialist) Principle or the Killing vs. Allowing (non-consequentialist) Principle. Since each of these principles recommend a unique choice in Two Parallel Tracks - Six Buttons (six buttons and no buttons, respectively), this means one ought to either press six buttons, or no buttons, depending on their principle of choice. In other words, one ought not press five buttons.

My argument is straightforward. Consider first a practical case. Imagine you are allowed to place a bet on one of two options:

a. You have an 99% chance to win $10,000, and a 1% chance to win $0 

b. You have a 1% chance to win $10,001, and a 99% chance to win $0

It is quite clear that option (a) is far superior to option (b). But in this scenario, the best possible outcome is winning $10,001. Therefore, according to the Go for Glory Principle, you ought to choose (b), since this maximizes your chance at achieving the best possible outcome. But we have already said that option (a) is a far better choice. So following Go for Glory leads to a terrible bet.

Consider now an ethical case:2

c. You have a 99% chance to save 1,000 people, and a 1% chance to save 0.

d. You have a 1% chance to save 1,001 people, and a 99% chance to save 0.

Go For Glory recommends taking option (d). Again, it is clear that acting according to the Go for Glory Principle is absurd. Therefore, Go for Glory cannot be a principle upon which one acts rationally or morally. As for Two Parallel Tracks - Six Buttons, the other two principles in consideration strictly recommend pressing no buttons, or six buttons. So, under the principles in consideration, there is no rationale for choosing five buttons.

In order to say otherwise, one must suggest a principle that achieves two things. The principle must, first, uniquely recommend pressing 5 buttons over any of the other options; second, it must be an intuitive rational/moral guide in other cases. This second condition ensures proposed principles have some reasonable general motivation.

My strong feeling is that any suggested principle which recommends pressing five buttons will always fail to satisfy at least one of these two criteria.


  1. Hare, Caspar (2016). Should We Wish Well to All? Philosophical Review 125 (4):451-472.↩︎

  2. Assume in this case that for either (c) or (d) you are not killing anyone.↩︎