Why does the 'Process Full' option fail to generate the usage-based aggregations, whereas the option Process 'Index' can?

Problem:

I have generated the usage-based aggregations. The 4 log properties have been set up. I have also fully processed the usage-based aggregation design. I can even see the new entries in the OlapQueryLog table every time when I run the query, although the StartTime is the UTC DateTime, instead of the OS DateTime. However, when I try to use SQL Server Profiler to watch if the usage-based aggregation is utilized in an MDX query, it appears that the SQL Server Profiler does not catch the 'Get Data from Aggregation'.

Solution:

Two sources (source 1 and source 2) indicate that the 'Process Index' should be used to populate the aggregations. After I do that, I can see the 'Get Data from Aggregation' event now.

Reason?

But I do not understand why 'Process Full' does not do that, whereas 'Process Index' does.