To me it all depends on what the receiving system wants and/or can handle.
For example that particular field can have multiple components. If in the detailed specification process it is determined the receiving system can only handle or only wants the first component then using component notation will assure that.
On the other hand, if the detailed specification process reveals the receiving system can handle all the components but the sending system is not currently sending them all I would use field notation so if in the future the sending system begins to send more components the receiving system will get them without change.
But the only hard and fast rule in my opinion is to do the detailed specification and then make a judgment call.
Unfortunately in too many cases the detailed specification process is not properly done – primarily for the purpose of hitting time lines.
If that is the case, then make your best stab and hope.
email: jim.kosloskey@jim-kosloskey.com 30+ years Cloverleaf, 60 years IT – old fart.