Work is a factor, this isn't even disputable. What is disputable is how much work is needed. Now naturally if you increase the load and do the same number of reps you increase work but in this case you increase the load and do not do the same # reps overall. What can be done, well ...... add a set? Cluster until you hit the same number of reps? or drop the load and do another set? Any or all three work.
Here's an example I've used several times and some are probably getting sick of it
Say you are using 100 Lbs and you are doing your 5's
1st set you hit 5 or a work total of 500
2nd set you hit 3 or 300
3rd set you hit 3 again or 300
Total of 1100
With dropping the load you get
1st set 5 @ 100 or 500
2nd set 5 @ 90 for 450
3rd set 5@ 80 for 400
Total = 1350(Note: granted this is a simplistic view on a complicated mathematical process but without knowing the actual displacement of each myosin head and how many are attached at any given time or the actual force they are generating for every muscle that is active during the movement, it's good enough)
Now since in either case you are still working to the same relative failure point (assumed) I.E. 1 or 2 reps from absolute MMF the recruitment and rate coding should be about equal but TUT would be actually higher with the drop system since your are increasing the total time that max tension is applied.
Anecdotally, reverse pyramids have been used for years and years, IE running the rack. Something old schoolers did by using the heaviest DB on the rack then each set step down to the next DB in weight.
Emperically, a vast percentage of the studies that denote sizable hypertrophic gains use this same concept. IE matching or maintaining work via reduction in load for subsequant sets while still perfroming the same number of overall reps. (See the article on work at my site).