Hi!
in your example where 3.4 would be the corrected FPA. If you’d set 3.0, you will probably gain around 40ft/nm.
40ft high at the first DME check would entice you to set 3.4. This would lead to you paralleling the vertical profile, being 40ft high once again at the next DME check. Now you’d select 3.8, back on profile at the next DME check where you’d set 3.4 back (50% of crew would set 3.0 probably).
So you can see the need of anticipation, especially when ISA deviation is large.
note: the exact opposite corrections are required in hot conditions, especially with temp inversions etc.
Anyhow, the discussion is quite academic. The most important thing is that we understand how the FPA behaves in non-ISA conditions and make due allowance for this by understanding.