Lethe

Yesterday I walked into the kitchen pantry while my mind wandered as it often does. I suddenly realized I had forgotten for what I had gone in there. I’m pretty sure this wasn’t a senior moment (something one always questions after a certain point) since I commonly did this very thing as a teenager – in fact more frequently then. In those days my mom (overestimating my academic achievement) called me the “absent minded professor.” Then, as now, my mind tended to wander onto events of the day, future plans, recent conversations, and various other matters so that I would forget the task at hand. I remembered what I was after yesterday immediately after exiting the pantry by the way: it was toothpaste, which in fairness is not something one usually retrieves from a pantry. Earlier in the morning I had noticed the tube in the bathroom was empty.
 
This sort of forgetfulness has been understood in broad outline for decades. Short-term and long-term memory storage are separate mechanisms. The former, mediated by the hippocampus, normally is good for up to 30 seconds – short enough to forget why one walked into a pantry. In order to kick a memory into long-term storage in the cortex you either have to deliberately concentrate on it or by happenstance associate it with some stimulus such as pain, fright, pleasure, intrigue, and so on. This is why soon after an hour’s commute we can’t remember most of it – only the parts where we had to concentrate because of some stressful road conditions. By the next day we might forget even these. In essence, the default setting of the brain is to forget; to remember requires effort, whether evoked spontaneously or by intent.
 

forget-me-nots
(Myosotis sylvatica)

Storing something into long-term memory changes the wiring and chemistry of neurons in the cortex so that some connections carry charges easier. These engrams are our memories, but even these frequently fade. (They are also subject to manipulation, which is why “recovered memories” in particular are notoriously unreliable.) A memory can be made not just long-term but permanent either by repeatedly calling up the memory (thereby reinforcing the wiring changes) or by associating it with some major stimulus. Traumatic stimuli are especially effective at locking in a memory, but more pleasant stimuli work, too. A very imperfect but useful analogy is a flexible green tree branch. If you flex it a little and quickly let it go, it will snap back to its original shape. It will, in essence forget the flex: an analog of forgetting a short-term memory. If you flex it repeatedly, however, or give it a sharp extreme bend, it will not snap back all the way, but will retain a partial “memory” of the flex.
 
Recent studies indicate that forgetting is an essential part of the process of making new memories long-term. Encoding all the details of an experience would clutter our minds and overload our capacity to speedily get any meaningful use from the memory. It’s important to remember at what corner to make a turn to get to a destination; it is not important to remember the rust patterns on every storm drain we pass on the way to the corner, so we don’t. It’s important to remember that a dog growled before it bit, not that there was a grass stain on its back left leg. Our brains tend to filter out the unimportant data (though odd details sometimes stick), so we are very good at remembering the gist of an event (the Darwinian advantages of which are obvious) but not the surrounding fluff. This is why eyewitness testimony is so often poor. Was the getaway car green or black? Were there two shots or three? Eyewitnesses often (in fact, usually) conflict with each other about the details and they are very open to influence; for example, if told a car was green they are likely to remember it as green whether it was or not.
 
Recalling just the gist aids survival because it helps us generalize in useful ways. In the dog example above, it’s good to generalize that a growling dog might bite, not that a growling dog with a grass stain on its back left leg might bite. That is too much detail; we could underestimate the risk of a stain-free growling dog. This is a commonplace issue in artificial intelligence as well. The problem is called “overfit,” which is when a self-learning AI attaches too much significance to random coincidences, thereby diminishing its ability to generalize and make useful predictions. An effective AI needs to be able to “forget” the unimportant stuff.
 
There are people with extraordinary long-term memories. The condition is called hyperthymesia, but it is a relative thing. People with hyperthymesia forget, too; they just forget less than the rest of us. This sounds like a great advantage, and in some ways it can be. The actress Marilu Henner famously has this type of memory; it doesn’t seem to cause her any trouble and it certainly has helped her remember her lines. Overall though, people with hyperthymesia do not do any better in life than other people; on the contrary they are at higher risk of PTSD and other obsessive disorders.
 
Experiments with rats have demonstrated various ways that neurotransmitters (dopamine in particular) affect memory retention. Understanding the process better might help with the treatment of PTSD (not enough forgetting) on one end and Alzheimer’s (too much forgetting) on the other. The scifi trope of excising or implanting specific memories does not appear to be in the cards for the foreseeable future, however, which is probably just as well.
 
I’ll end here since there is something I need to retrieve from the pantry. There was something else I wanted to say, but I forget what it was.
 

Beverly Bremers – Don't Say You Don't Remember