Se is all about inferring the match of an observed situation with a proto-type. Having seen the exact conditions before means that an inference to recognition of these conditions is warranted to them. This is how they are both decisive and unwilling to entertain non-standard alternatives.

Si is largely the same thing, however in Si's case there is no need for "inference", because the match is made on the level of observation itself rather than on the level of the situation that is removed from the observational layer of experience.