Since 'computational' as used in the context of Betterology is not in common usage, perhaps it would be best to define it through a commonly understood behavioral pattern.
'Computational' is used in this context as different from a more traditional or knowledge based understanding of the a set of options.
Let's take the set of options associated with mapping a route through London during a typical rush hour compute.
"Google Maps takes a computational approach to providing driving instructions."
A traditional view of this set of options would be to value the knowledge of a seasoned London cab driver. With years or even decades of experience under his belt, he might know the best options for getting through.
A computational view of this same set of options would be a route defined by Google Maps. Computed and even re-computed in real time from thousands of incoming data points, it can take into account variables that would not be knowable to the cabbie in real time.
If you've ever given Google Maps a series of tests on a longer urban route that you knew very well indeed, you already know how difficult it is to beat or even match a comutational approach of this type.
It goes without saying that your best bet would be a London cabbie using Google Maps - the best of both worlds! So 'computational' does not have to conflict with a traditional view of options, it can be an excellent complement.
Driving routes are but one of hundreds of option sets that we had always taken for granted as 'knowledge' based, yet could be 'computational' if given enough live data and computing power.
As humans, we cannot be expected to see these opportunities in normal circumstances. When the first telephone was invented, Western Union declined to purchase the invention because it owned the telegraph exchange, and it just "knew" that the phone could never compete with the telegraph.