Cognitive Chunking

I'm looking at some sticks of wood, and a board, fashioned to fit together--pegs in holes, tabs in slots. Can you imagine what I'm looking at? Would it help if I described each piece in more detail?

Perhaps it would, but you would likely find it easier if you grouped some of these pieces into logical categories. "Four legs" is easier than four separate descriptions of similar, but slightly different nearly vertical sticks. Would it be even easier if I said I was looking at a wooden chair?

George Miller wrote a marvelous article¹ in 1956 that explored the capacity limits of "immediate memory." He had subjects identify tones that varied in pitch. When there were only a few tones to choose from, the subjects almost always got them right. When the number of tones grew, more identification mistakes were made, especially beyond 6 tones, or about 2.5 bits of information. If, however, the tones can vary in two dimensions, e.g., both pitch and loudness, the errors become more common at about 3.1 bits of information, or 8 tone/loudness combinations. The conclusion is that the memory span is the limiting factor, and the throughput can be enhanced by encoding multiple bits of information into each chunk of information.

This conclusion reminds me of Claude E. Shannon's work on information capacity of a channel related to the bandwidth and signal to noise ratio of the channel. When I worked in modem development, I learned that you could transmit more data if you encoded multiple bits into each transmitted signal, for it was the signaling rate that was limited, not the bit rate. The Bell 212A modem, for example, managed 1200 bits per second by encoding 2 bits per symbol at a 600 baud rate.

Similarly, humans recode multiple bits of information into a chunk, generally giving it a specific name and dealing with the chunk rather than it's component parts. We talk in terms of the chair instead of the sticks and boards that make it up.

We categorize our perceptions to make it easier to reason about them. It also makes it easier to share our perceptions with others. But the very act of categorization removes some of the information. Our categorization scheme predetermines what information we consider important and deletes the rest.²

As Barbara Tversky says, "Categories are so much easier than continua.³" That's why our cognitive processes use them. I find it worthwhile, though, to revisit the categories I use and think about what information I'm leaving out. Sometimes that information is important for some of the things I might be doing. Easier is not always better for the outcomes we want.

It is worth questioning, from time to time, our habits of chunking data and the categories that we use in order that we might notice ways in which it's working against us. That's not to say that we'll give up chunking and categorization, or even the patterns that we've been using. Instead it's to augment those patterns with new ones, ones that give us new possibilities. These possibilities may enable outcomes better than we've yet dreamed of having.

P.S. I'm looking for conversation on this topic. I would appreciate it if you could reply with a few words about your reaction to this newsletter issue. Think of it as extreme opt-in, non-persistent and totally voluntary tracking.

And if you'd like to talk further, you could schedule a Zoom Session with me to talk about it.

Schedule Zoom Session

Or you can also simply reply to this email or send an email to newsletter@idiacomputing.com to continue the conversation. There's a person, not a bot, on this end. I'd really love to hear from you.