Using potential outcomes (or counterfactuals) to think about causality is quite useful, in my opinion. However, when talking about causes of a single event, confusion often arises because people (a) are used to thinking about events having a single cause and (b) mistake cause for blame. I will illustrate these ideas with a hypothetical example.
Causal definition
First, I will say that an action A caused an event B if the event B would not have occurred if action A had not been taken. That is, what actually happened was an action A was taken and at some time later an event B occurred. We will say that A caused B if, in that same world, you could go back in time and not take action A, then event B would not occur.
Example
Dan gets up in the morning, eats breakfast, and starts to drive to his father’s house. On his way there, a car goes through a red light and crashes into the side of Dan’s car.
Later that day Dan tells the story to his dad. His dad says, “Eating breakfast caused the accident.” Dan says “What? Some guy ran through a red light! That’s what caused the accident.”
Many causes, most blameless
Dan’s father is almost certainly correct. Had Dan decided to skip breakfast, he probably would not have been in that intersection when the other driver drove through the red light. Thus, action A (eating breakfast) caused the event B (that specific car accident).
For the same reason, it’s also true that the other driver caused the accident by not noticing the red light. There are many other causes. For example: Dan’s decision last year to move to the town where he currently resides; the driver who was in front of Dan who didn’t just sit at the green light, blocking Dan’s path to the intersection; Dan’s father, who called him the night before, asking if he’d come for a visit. Etc.
When someone says that your action caused something, it feels like they are blaming you (‘your action caused it’=’it’s your fault’). That might be what they are implying, but in general, your action causing something does not imply that you should have anticipated the effect. For the large majority of causes, the effect was not foreseeable.
When should cause imply blame?
To evaluate whether we should anticipate an event B caused from an action A, we need to think about population level causal effects. On average, if a large number of people (in a similar situation to your own) took action A, would the event B occur more often than if they did not take action A? If so, and if there was a way for you to know this, then it is fair to assign some blame to you (with the level of play proportional to the size of the causal effect).
For example, if a world where a large number of people eating breakfast results in many more car accidents than in the same world but where no one eats breakfast, and there was a way for Dan to know this, then it might be reasonable to put some blame on him for the accident. If that is not the case, then while his decision to eat breakfast might have caused the accident, there is no reason to blame him for it.
N=1 