Memory

On a recent trip to visit my former home of Cambridge, I found myself playing piano and singing with friends on several occasions. The last night I was there, I opened a Broadway songbook to a tune by a man whom one friend dubbed the "Kenny G of musical theatre"-- namely, Andrew Lloyd Weber. The tune was from Cats:

Daylight
I must wait for the sunrise
I must think of a new life
And I mustn't give in
When the dawn comes
Tonight will be a memory too
And a new day will begin.

Jibing aside, ALW has some out-of-the-ballpark hits, and although I feel most of my friends cringe every time I want to play the song, I can't shake the notion that "Memory" is one of them. If you took the song out of the context of the musical Cats and handed it to a soulful songstress to sing, I firmly believe that we'd think of it a little differently. But more to the point, Weber uses the term "memory" quite aptly in his music, showing that the here and now will one day become a memory too. And maybe the memories of multiple nights spent walking down memory lane will be consolidated into a more abstract memory.

Memory has been on my mind a lot recently, and not just memories of old friends in old towns. In a seminar class in the linguistics department, we are reading papers devoted language and what psychologists call "working memory" -- the memory for stuff we're doing in the here and now, and theorized to include some set of representations that occupy a more privileged status than other stuff. In my own research, I think more about that other stuff, i.e., "long-term memory" -- the imprint we have of all our experiences, plus a whole helluvalotta hypothesized structure which leaves us with many theorized types of memories. "Procedural memory" refers to learned behaviors, mostly typified as motor skills or procedures. Examples might be playing a musical instrument, riding a bike, lots of what goes into driving a car, and, some would say, even certain aspects of language processing. "Declarative memory" refers to memories for facts about the world and the stuff in it ("semantic memory") and to memories for particular events, or episodes, that we have personally encountered (so-called "episodic memory"). For my own part, I'm very interested in understanding how this last breakdown (between memory for stuff we know, or at least think we know, and memory for stuff we've directly experienced) should be thought about. Surely the stuff we directly experience can become part and parcel of the stuff of "facts" that we know -- we see a lot of red apples, and so we come to believe that apples are (at least often) red. But what's the internal organization of all that, and what stuff do we actually access when we're talking about apples?

There are plenty of questions to ask with regard to language and long-term memory. Whether understanding someone who is talking about apples requires someone to engage in processing all of the perceptual information about apples (such as their color) is a question that is part of a hot topic in cognitive science currently. But in my own research, I hope to answer a slightly broader question, which gets at some of the same nitty-gritty questions regarding which features of a concept are accessed during its processing--does the amount of information about (or experience with) a concept directly affect how easily it is processed during language comprehension? The structure of that information will almost certainly have consequences for processing, as well.  For instance, the information might be neatly bundled together (let's say we are apple experts and know that really red apples are also shiny and large, encompassing information about perceptual aspects of apples) or more sparse (perhaps we don't really know much about red apples other than we've seen them grow on trees in Washington, and our parents used to pack them in our lunches when we were kids, facts that are not as neatly related).

In my research, I'm starting out by investigating these kinds of questions using short texts (jargon-y term here would be "discourse") where I manipulate the amount of information that is given about various people who are described in the text.  But what I really want to know is how understanding language about stuff we really know a ton about (people who are very close to us; topics we've spent years studying; our favorite hobby, whether it be cooking or painting or singing; cities we know well) might differ from understanding language about stuff that's more vague to us (sort of like when I hear my friends talking about baseball in front of me; or like when I talk about sight-reading choral music to the people talking to me about baseball!).

I'm thinking our rich memories may help us understand each other at times and perhaps hinder our communication at others.

"Invidious, insidious, a pretender, an impostor, a quack"

No, those aren't adjectives describing my perpetual impostor syndrome, nor do those words come from a line from another Dorothy Parker poem -- they are, rather, some words on how a philosopher quoted in a 1993 Psych Review paper on concepts described the notion of similarity.  I tend to disagree -- I am quite happy with considering similarity to be a useful notion in thinking about categories.  When we are trying to overcome a tip-of-the-tongue phenomenon (something that happens to me quite frequently, especially when I've been talking a lot), we often try to "grab" words from our "mental hat" (at least, this is how I've conceptualized it) and sometimes miss and grab ones which are similar in some dimension.  At least this is my anecdotal folk theory... though I'm fairly certain that science partially supports my understanding.  For instance -- maybe I'm trying to remember that the capital of New York is Albany, and I'll be reminded of Anaheim, another "A" city which is not the most striking city in a state with a very striking city (NYC, L.A.).  There's something complicated about how that might all break down, but similarity computation seems to be a parsimonious explanation for at least part of that strange cognitive journey.

There are theories in sentence processing which posit that similarity-based interference between structures sharing features (semantic, syntactic, other?) can cause some problems in sentence processing.  Recently I've been thinking more about how similarity might be _helpful_ for sentence processing.  Maybe in some cases, it might be hard to keep track of things someone's talking about because they're all so similar, but in other cases, it might be great that items in long-term memory are stored in such a way that similar things can interfere -- because those other similar things might just be relevant in the near future.

All of these mumblings are mostly due to the "core" seminar class that first-years here are meant to take.  I happen to find the course really well done.  Somehow it is a seminar course which manages really to function as a seminar course, with each of us taking turns leading the discussion (today it was my turn), and everyone reading papers thoughtfully and responding in kind.  I really like the chance to take overview courses (or T.A. them) as it always reminds me about parts of cognition I rarely think about, and how that relates to questions I'm really interested in.  Now off to think some more about feature relevance, event structure, and specificity -- after of course posting one of the cutest little vignettes I have ever come across in psycholinguistics (taken from Nieuwland and Van Berkum, 2007, and originally in Dutch):

"A woman saw a dancing peanut who had a big smile on his face.  The peanut was singing about a girl he had just met.  And judging from the song, the peanut was totally crazy about her.  The woman thought it was really cute to see the peanut singing and dancing like that.  The peanut was {salted / in love}, and by the sound of it, this was definitely mutual.  He was seeing a little almond."

(Guess which continuation was easier for participants?)