The other day Luann and I were thinking of a long-ago friend and realized we’d forgotten the name of that friend’s daughter. Decades ago she was a spunky blonde blue-eyed little girl; we could still see her in our minds’ eyes, but her name was gone.
“Don’t worry,” I said confidently, “it’ll come back to one us.”
Sure enough, a few days later, on a bike ride, the name popped into my head. I’m sure you’ve had the same experience. This time around it prompted me to think about how that happens.
To me it feels like starting up a background search process that runs for however long it takes, then notifies me when the answer is ready. I know the brain isn’t a computer, and I know this kind of model is suspect, so I wonder what’s really going on.
– Why was I was so sure the name would surface?
– Does a retrieval effort kick off neurochemical change that elaborates over time?
– Before computers, what model did people use to explain this phenomenon?
So far I’ve only got one answer. That spunky little girl was Diana.
I’ve been fascinated by this kind of thing for a long time, and I think that a few things can be concluded. Some unconscious brain process has to be going on. We know that brain activity does consist of many more-or-less independent processes that are (sometimes) coordinated by some kind of “executive” function. So that fits right in. More interesting, perhaps, is how the process can continue overnight or for days without getting so distracted that it forgets what it is supposed to be doing. Our conscious selves may have forgotten long ago.
This phenomenon suggests to me that the older-data storage systems (from which the name “Diana” is finally retrieved) are different from the ones that store the readily-retrievable memories.
Maybe more interesting is how you can know that various proposed solutions are wrong even when you can’t remember the right one. You know for sure that the name isn’t “Mary” even though you can’t bring up the right name. How can that possibly work?
One possible answer is that you don’t recall the name alone, but a whole complex of context. A bare possible name like “Mary” doesn’t carry the context you expected so it can’t elicit the proper recognition. This leads to the question of how can such a context be constructed and maintained in our brains. I have no idea but I suspect that answers would lead to great advancement in the field of computer “intelligence” (and perhaps “common sense” as well).
What’s your guess as to when we’ll know?
I don’t have any idea; I wish I had some inkling. I have come to think that the people who try to build up a universe of relations (like an RDF graph) aren’t coming close to how things are stored in the brain. Neural nets probably come closer, but seem to be missing one or several key pieces. So we need to come up with some new ideas. Will they arise out of some detailed lab research on memory mechanisms in brains, or from some other kind of thinking?