It would be very interesting to see exactly what type of imaging they performed on him. Bear in mind, an abnormality in the hippocampus or other memory related structures could be sub-mm in size and result in a significant impairment.
If, for example, he did not have an MRI with diffusion weighted imaging with a sufficiently high B value (gradient value) they may be missing/have missed a very small stroke (this test is also time sensitive -only positive within the first few days of insult).
If they did not perform susceptibility weighted imaging, they may have missed a tiny micro-haemorrhage. Both of these scenarios can result in memory loss if critical structures are involved.
There are also several technical issues that may have resulted in a sub-optimal scan, precluding finding a tiny abnormality. There's also the question of whether the person reading the scan got it right or whether the strength of the magnet/technical parameters of the study were sufficient to resolve a tiny abnormality.
Either of these scenarios could have been a consequence of reduced blood flow (hypoperfusion), a small clot (embolus) or a tiny blood vessel injury (dissection) that could have occurred during the procedure - possibly related to the anaesthetic but not definitively so.
I suspect there is a definitive abnormality that the clinicians that saw him were unable to pick up. This seams a lot more likely than the idea that we need to revise our entire conceptualisation of memory formation to suit a psychological theory.