If you want to poke holes in someone’s ethical system, you can come up with a bizarre, highly implausible scenario and show that the ethical system recommends an apparently perverse or reprehensible response.
I’ve seen this quite a lot in philosophical and religious discussions. Indeed, I’m pretty sure I’ve used it myself. (This despite the fact that, at the end of the day, I don’t think we should rely too heavily on our moral intuitions, since there is no guarantee they are optimal on any particular metric).
But, assuming that moral intuition is useful to some extent, I’d like to propose an alternative to the one-off implausible scenario test.
For starters, implausibility should be explicitly taken into account. It seems to me that a moral system has much bigger problems if it produces unreasonable solutions to common, everyday problems than to rare, highly contrived ones.
So let’s imagine applying more than one moral scenario to the ethical system under question. Let’s put a list of scenarios together, some of which are common (should I cut in line at the deli if I’m in a hurry?) and some of which are rare (if I’m forced to choose between killing one person or five, what should I do?).
For each scenario in the list, the ethical system will produce a prescribed course of action that either conforms well to our moral intuition (it “sounds right”) or does not (“sounds wrong”).
We therefore have two scales. We have the frequency of the moral scenario (how commonly it occurs) and the degree of conformity its ethical solution has to our moral intuition (how “right” it sounds).
How do we combine these scales to get a reasonable idea of the ethical system’s performance?
I’ll present one possible solution here, with the use of an example. Let’s say that our list of moral scenarios contains ten items ordered by frequency: The first scenario in the list is the most common, while the last item is the most rare. We then label them A through J. Here they are:
(Exactly how frequent each scenario is, is not important, so I’ve not included labels on the vertical axis.)
Now let’s apply these scenarios to our (hypothetical) ethical system and see how intuitive the proposed solutions are:
It turns out that, in this particular case, the intuitiveness of each moral scenario decreases from left to right, just like the frequency data: The most commonly occurring moral scenario (Scenario A) also produces the most intuitive moral prescription, while the rarest scenario (J) produces the least intuitive moral prescription.
This sounds like the sort of situation we could live with because, at the very least, an ethical system should get the most commonly occurring moral problems “right”. And if it gets some problems “wrong”, it’s better that these be very rare.
We can get a more specific measure of this combination of frequency and intuitiveness by multiplying the above two graphs together. Let’s multiply scenario A’s frequency by its intuitiveness. This will give me a “reasonableness” score for scenario A. Then I’ll do the same for scenarios B, C, D, etc. Here’s what I get:
Moral scenario A is the most “reasonable” because not only does it produce an intuitive result when applied to our ethical system, but it’s also quite a common scenario, which means that a lot of intuitive moral decision-making will be going on.
Scenario J, on the other hand, gets a “reasonableness” score of almost zero because it’s associated with a counter-intuitive moral solution. It’s impact is also reduced thanks to its rarity.
One way to make a single numerical assessment of these findings is to add up the reasonableness scores of all ten scenarios. With the numbers I’ve used to generate the above graphs, I get a total of 220 points.
Now let’s see what would happen if the trend of intuitiveness was reversed:
Now our ethical system is giving us a highly counter-intuitive solution to Scenario A. This is unfortunate, because Scenario A is still the most commonly occurring one. This means that a whole lot of counter-intuitive moral decision-making will be going on.
At the other end of our list, Scenario J produces a nice intuitive result. But this is not much consolation because Scenario J is so rare.
Once again, let’s multiple the frequency of each scenario by its intuitiveness. Here’s what we get:
Overall, the reasonableness scores are lower than before. Scenario A now has a low score because it has a counter-intuitive moral solution. Scenario J (like before) has a low score because it’s rare. When I add up the reasonableness scores, I get a mere 32 points.
Ideally, we want an ethical system that will produce an intuitive solution for all scenarios, regardless of how common they are. If this were the case – if scenarios A through J all had the maximum intuitiveness value seen on the blue graphs above – then the resulting reasonableness scores would add up to a whopping 1100 points.
It’s fun playing with these numbers, but the central message is that there are more sophisticated ways of assessing a moral system than simply throwing one highly implausible scenario at it, and judging it on the basis of that scenario alone.
Besides, no ethical system I’ve come across can escape at least one or two counter-intuitive results. Unless we’re prepared to throw all ethical systems out, we need to find a better way of assessing them.