'Reasons' are features of our mental models of the world - they're not features of the real world.
Human minds build an internal simulation of the world around them, the mental model consists of 'objects' and 'agents', and 'arrangements' and 'relationships' by which they're organised, and 'rules' and 'laws' constraining how these entities change and transform. (I'm not saying this is a complete list.) We can start with an initial state of the world, and a set of causal laws and events, and use the model to predict the sequence of subsequent states. Or we can start with a final state, and reconstruct what events lead up to it. (Either to deduce what happened, or to try to figure out how to make it happen.)
'Reasons' are the chain of model manipulations we applied to get to a particular conclusion. They depend on the specifics of the mental model, and what we're trying to do with it.
So if we're using the model to predict the outcome state of a particular sequence of actions applied to an initial state, then the chain of actions and the relevant bits of the initial state are the reasons for the final state being what it is. The reason for C having the ball at the end is the initial state (A had it) and the two ball-passing actions. Other events not in this causal chain (like D having an apple and D eating it) are not reasons.
But if we instead were playing a game, and trying to figure out what chain of actions from the starting position we could take to win, then the 'reason' C has the ball is that C is closest to the goal, and A passing to B and B passing to C are consequences we introduce to make it happen. What is the 'reason' A passed the ball to B? A passed the ball to B, so that B could pass the ball to C, so that C could throw it in the goal and win the game. The 'x is the reason for y' relationship links them in reverse temporal order, because that's the way we're running the mental model.
Or we could start with both initial and final states, and be trying to figure out how we got from one to the other, maybe to figure out who to blame for the ball smashing the window. A must bave passed the ball to B because C was out of range, and there was nobody else on the basket court. Now the 'reasons' are in both past and future simultaneously.
'Reasons' are the steps in a proof. We start with axioms and assumptions, and proceed step by step to conclusions. We have a choice whether to consider only the immediately preceding steps as the 'reasons' for a particular conclusion, or the entire chain leading up to it. I don't think either choice is better or more widely accepted in everyday usage than the other. Again, it probably depends on what you are trying to do, and which is more convenient at the time.
Humans often get their mental model of the world around them and the world itself mixed up. We build a simplified mental picture of 'objects' and 'agents' and 'causes/effects' like toy building blocks we stick together in various ways, to try to understand and predict an infinitely complex world, and then we start questioning which of them are real. We invent 'paradoxes' where a particular kind of mental model we use to simplify the world doesn't quite fit, and suggest that it is reality that has somehow got it wrong, or is being especially mysterious in its inner workings. But we're wired that way. Building a mental model of it it how we perceive the world - to us, what we perceive is the world - and the abstraction step needed to separate them is not automatic.