There is simply no answer to this question. Even if your table structure doesn't change, queries can get different execution paths over time depending on the amount of data, the indexes, the constraints, because of bind variable peeking, and scads of other factors. Entire books have been written on the subject.
cagcowboy's answer is incorrect. Oracle will rewrite your query to provide what it thinks is the best execution plan. Queries like the one you describe are often transformed by subquery unnesting. My guess is that 9 times out of 10, queries similar to the ones you describe will have the same execution plan.
In my opinion, start with what is most readable, what will make it clearest to someone else reading your code (or yourself, reading it six months from now), what your intent is. If your query runs unacceptably slow, only then try to optimize it.
As Branko Dimitrijevic points out, two queries you think are the same often aren't equivalent. In your two examples, if salesman_name is not unique, your first query will throw a ORA-01427: single-row subquery returns more than one row
exception, but your second example will work fine.