Hi, I was curious to watch the lecture and hear the favored explanation by Robert J. Nemiroff.
I think he chose the explanation proposed by Drezet - that those photons that reach the detectors do not have formed interference. This is however wrong, and is not consistent with the solution of Fresnell diffraction integrals for the double slit - lens system. Check this article, where for the first time are shown the calculated probability distributions for Afshar setup in coherent (both slits open) and incoherent setups (one slit open at a time, or both open but different polarization filters are placed on the pinholes).
D. Georgiev. (2012). Quantum Histories and Quantum Complementarity. ISRN Mathematical Physics, Article ID 327278, 37 pages
http://www.isrn.com/journals/mp/2012/327278/
There is much more in the article, just check it out. I favor the application of Feynman's sum-over-histories, which makes things quite transparent. In the book "QED: The Strange Theory of Light and Matter" Feynman made very nice footnote remark about the principle of complementarity, which is that once you sum-over-histories you do not need to rely on the principle of complementarity. You just add the corresponding quantum amplitudes contributed by different histories and you can "see" what cancels with what. Complementarity is correct, but not in the way all advocates of the Copenhagen Interpretation believe. In coherent Afshar setup there is no which way information at the images to begin with, so placing wire grid is irrelevant. This was independently proved by Tabish Qureshi in 2007, associate professor at JMI, India, who posted his work in ArXiV just a couple of months later after my first pre-print was released in PhilSci in 2006.