Adjustment is a metrics for similarities between two distinct symbols. In general, in order for each of these two corresponding signals to have a merging force and a null mean value, the merging function shifts to a combination of the correction coefficient, of the two symbols you compare. The correlation coefficient, for the time remaining between the two symbols, is always between -1 and +1, clearly giving an estimate of the similarity of the conditions for the two remaining signals.
Conversion, instead, is a standard operation that the Linear and Time Invariant system can perform on a given input signal.
It is clear that, in some cases, the correction of anc convolution is very similar: the corresponding filter case makes the merger and the convolution similar.