Training - How do I decide between equally attractive moves?

I believe this is summed up with the term 'Accuracy'? Essentially some of the training problems have robust positions with no obvious weakness, and present equally attractive moves - at least to me.

As an example, here is one I'm currently looking at:
en.lichess.org/training/62077

Here are the moves I'm considering: postimg.org/image/3w11uhs59/1f64d484/ for those that think that link might be scary looking, it is just an image - but the candidate moves are:

Option 1: a3
Option 2: Bd2
Option 3: O-O

I've not made my move yet, so I will tell you it turns out in a moment. I'm thinking Option 3 however, and this is my line of reasoning:

Black last moved b6, I am thinking to provide a door for Bc8. I don't think c5 is a threat since that will lose him the Bb4 after a3, Ba5, b4 (assuming Nc3 reaches safety first). So it has to be that he is making a door for his Bc8. Bc8 only has two squares to go to and I can't see any major advantage of either of them, both of them means he will be biting on a pawn wall. So I don't think Black's move is a threat, or a potential threat. If anything he has essentially trapped Bb4 at least for the immediate future.

That leads me to ask about if it is ready for capture - Nc3 is under threat and my thinking is if I put pressure on Bb4 it will go for the exchange for the reasons mentioned above about it being trapped. I prefer pieces for free, so I want to free up Nc3 to move so I can then pawn-swarm Bb4. To do that I would need to castle, Option 3 is castling, I'm going for that.

Option 3 was not correct. Puzzle failed. Neither were Options 1 or 2, and this is where I am at the moment. My Training rating just yo-yo's between 1750 and 1850 which is where these kind of exercises appear, and I get as many correct as I do incorrect.

For those that don't want to know the solution skip this bit - (I don't know if the forum suppors spoiler tags, so this is my attempt to obfuscate the answer so someone doesn't read it accidentally - it was #*!f@#*#@3*!#).

When I look at the solution, I think, well, I wouldn't choose that. Doesn't it weaken my King-side? I haven't completed development yet? Nc3 is still pinned. What am I missing?

Well, the problem with this position is the tatctical part. As you can expect from a tactic trainer, a given task involves a tactical blow. The solution is just so much better because it wins a piece. Your question seems quite obvious to answer as you have an engine sitting there. Look at its lines and you will understand that b6 opens up the h1-a8 diagonal. After the first move f4 Black has to remove his Knight on e5(or just sac it right away), which is btw actually nice to achieve, and then you can create a double attack by opening up the diagonal with 2.e5, threatening both the Ra8 and the Nf6, winning material.
So if there are equally attractive moves, you must be sure to not overlook a tactical solution. If there is no such forced motiv then its basically not that great of a difference.
To your point of view: a3 as well as Bd2 look like they're not necessary yet. But castling would be a move you will play anyway so it will most of the time be better. Also Bd2 looks a little bit passive, so I wouldn't even really consider it. Otherwise you got a decent look at the position, except for your tactical vision.
Just practice that and look for forced lines. Then you should be able to find those moves easier after some training.

it is a tactic not a positional question f4 knight moves e5 double attack

Thanks @Eldrail I think that its a tactics trainer should push me towards looking at those sort of solutions more so than just regular moves which are good. I actually didn't notice unti I moved through with the analysis engine that e5 threatens exf6 as well as Bxa8. OK got it.