There are also problems with the the physicalists model of Alzheimers / Dementia because long-term memory tends to survive brain damage much better than short term memory. Arguably short term memory of what occurred 2 minutes ago, is bigger (or at least as important) evolutionary advantage than remembering the day 70 years ago when our Grandma bought us a ice cream at the seaside.
To get around this the physicalist will suggest it is not the memory which is necessarily damaged but the programming or processing of memory.
Karl Lashley's 30 year experiments (removing various parts of animal brains - yuk!
) to see how memory affected ability to do learned tasks (memory tended to survive better than expected). He gave up assuming memory was in any particular region but never quite took the leap of assuming memory might not be fully in the brain. If memory survives brain damage (regardless of how it processes or filters memory) it is a step nearer the case for consciousness and identity might survive brain death?