Even assuming that the study's results are valid, that applies to those situations where breakthroughs would be happening anyway. Seldom do we have team-wide "aha!" moments from us all individually plowing through our separate workloads -- instead, we have those moments when we have occasion to meet (in whatever format) and discuss something in a group.
Where I work, upper management has taken the approach that each team should set its own minimum on-site hours, based on the needs of that team. For my own team, we are on-site only a day or two a week, and we coordinate to do our meetings on those days as much as possible. This seems to hit the sweet spot for us -- we are out of each other's hair and can focus on our respective workloads for most of the week, and we are all together in the office for part of the week to talk and catch up, share ideas, and do anything else that benefits from a group setting.
Subjectively, I agree with the premise that direct collaboration and brainstorming work better in person. That said, such activities are not the bulk of our day-to-day work, and I see little broad-based justification for the "everyone must be in the office, all day, every day, because we say so" data-lacking approach taken by companies like Amazon. On the contrary, such groundless diktats seem to do more to piss off employees than to engender team spirit.