Wednesday, March 29, 2023

Bleg: Memory for Serial Order

I'm bumping this to the top of the queue because it seems at least obliquely relevant to my current thinking about what's going on in ChatGPT.
* * * * *

I’m looking for ethnographic evidence about how people recall events. I’ve got my own personal experience, which is just that, and I’ve got accounts in two rather old books: F. C. Bartlett, Remembering: A Study in Experimental and Social Psychology (Cambridge 1932); Heinz Werner, Comparative Psychology of Mental Development (International Universities Press, 1948). Does anyone have more recent, and more detailed, evidence?

While the question interests me in general, I also have a specific reason to be interested in it: It seems to me that the (culturally important) narratives of preliterate peoples are always told in event order – in particular, I’m thinking of the many examples Lévi-Strauss recounts in The Raw and the Cooked (and in subsequent volumes of Mythologies) and of the Winnebago Trickster stories as reported by Paul Radin; I have no reason to think that these reports are atypical, but they might be. They certainly don’t constitute a proper random sample.

Serial Memory: Bartlett and Werner

I’m interested in how we remember things that have an inherent temporal order, such as events in life history. One wakes up in the morning, performs one’s morning ritual (turn of the alarm, get out of bed, brews a cup of tea…), you go out the door, to the driveway, get in the car…and then a dog runs out into the street, WHAM!…and so on…“Honey, I’m home! the day sucked”…until you fall asleep. Similarly, pieces of music unfold in time.

How do you recall such sequences at a later date? In particular, how do you “get to” an event that happened in the middle of the sequence? It seems to me that that simplest scheme is simply to record the events in order as they happen. If that’s all you do, then it would seem that, if an when you want to recall the events, you must start at the beginning and replay them.

Unless, of course, you (that is, your mind) have indexed a sequence of events in some other way. But such indexing requires mental machinery above and beyond that required for “raw” recall. And many of us do have such machinery.

To Bartlett (Remembering pp. 264-265): He reported that native witnesses in colonial courts were often unable to recount the events of a day in a temporally selective fashion. When a witness was asked to testify about events that happened, say, in mid-afternoon they would start their testimony by recounting what happened once they awoke in the morning. If they were interrupted and asked to move directly to the events of the afternoon they would be unable to do so. Not only that, they would be unable to pick up their narrative from the point where it left off. Instead they would have to go back to the beginning of the day and start over.

Two things need to be explained:
1) Why these people couldn’t move directly to the events requested.
2) Why other people can.
I’m more interested in the first than in the second question.

The first question could be explained by postulating that they stored the day’s events as a dynamically evolving system in which each successive state of the system evolves from those that happened before. In such a system there recall is a matter of is re-evolving the system dynamics from the beginning.

If you wish to segment the memory stream so that you can enter it at various points, you need to be able to take "snapshots" of the system’s state at arbitrary points in its evolution. You can then re-start the evolution using any of those snapshots. Note, however, that this may involve some degradation of the evolution. To the extent that the snapshots are not accurate, re-evolutions started from one of them may be at variance from re-evolutions started from the beginning. Nor, for that matter, will re-evolutions started from different snapshots produce the same accounts of common events.

Heinz Werner reports (pp. 203-204):
The manner in which primitive peoples recite their songs affords us one example of the “all-or-nothing reaction.” Many aborigines are unable to begin their songs at any point in the text, but always have to commence anew at the very beginning or fail completely. The Papuans, for instance, cannot start a new song until they have danced and sung the very last verse of the one traditionally preceding it. Different strophes in a song cannot be interchanged. Similarly, no figure in the dances of these people may be omitted, or the whole intention of the dance is disrupted.
Such behavior is not confined to “primitive” peoples.

Reconstructing an Existing Musical Line

I notice that my memory for music displays a similar effect. Suppose I'm trying to play a tune that I've heard once or twice. I start right in and have no problem but then I make an obvious mistake, one that completely derails me. I now have lost the thread.

In this case I have no choice but to go back to the beginning and start over. It's generally not possible simply to go back to the beginning of the phrase where I made the mistake for I simply can't recall the beginning of the phrase. So, I have to go back to the beginning of the whole tune and start over.

This effect applies to chord changes as well – I’m an improvising jazz musician. In this case I'm improvising over a set of changes without any accompaniment to cue me to the changes. If I loose my sense of the unfolding progression I may have to go back to the beginning to reconstruct it.

Once I know a tune well I can start at major phrase boundaries.

Reconstructing a Musical Line I Invented

It often happens that, in the course of a practice session, I'll come up with a new musical line that I like, that I'm interested in developing into a well-formed tune. The first job is to be able to reproduce the line at will. And that is tricky.

More often than not I'll simply be unable to recall the line at a later practice session. I'll know that there was that line, and I man have some idea of its general characteristics. But I can't simply play it again. Sometimes it will come back spontaneously, and unexpectedly. And sometimes I can get it back by noodling around in the same musical "territory." That is, I usually have some idea of the general type of thing I was doing when the line first emerged--minor key, fast, slow, smooth, angular, etc. I may also have a sense of the general mood. Between psyching myself into the general mood by fantasy or whatever and noodling in the musical territory I can sometimes bring the line back. If I do this enough over a period of a day or two, I may reach a state where I can bring the line back at will.

Note that I have tried giving a name to the line when it first arises and then using that name to recall the line. This isn't very effective. I've also tried notating the line on the spot. The problem with this is that I'm not very facile at this. I have to sound the line note by note, phrase by phrase, and write each note or phrase down as I sound it out. This often so fragments the musical process that the line simply disintegrates.

Addendum, Department of D'oh!: It turns out that I've got a good source of information on this topic on my book shelf, David C. Rubin, Memory in Oral Traditions: The Cognitive Psychology of Epic, Ballads, and Counting-out Rhymes (Oxford 1995). Chapter 8, "A Theory of Remembering for Oral Traditions" deals with the topic in a satisfying way. And, yes, the phenomenon reported by Bartlett, Werner, and me is real.

No comments:

Post a Comment