(Go back to Thinking and cognition, or to home page.)

The "Size" Limits of Short-Term Memory

Using an overhead projector, show students a long series of digits very briefly--at most just one or two seconds. Ask them to write down as many of the digits as they can remember. Take a poll of students' success rate (how many remembered 100%, how many remembered 80%, 60%, etc.). Discuss why it is hard to remember the digits from only a brief exposure, and what strategies they could use to improve their recall (e.g. repetition, looking for patterns in the digits, etc.). After identifying a few memory strategies, show them another series of digits briefly. Did students' average performance improve? Discuss why or why not.

Tip: if you use digits that only seem random but really contain a subtle pattern, this activity can also illustrate the effect of "executive" or organizational processes. Flashing the series 1 2 4 9 1 6 2 5 3 6, for example, may produce poor recall at first because it seems random. Pointing out that the digits are equivalent to the predictable series 1² 2² 3² 4² 5² 6² , however, usually improves recall dramatically. The memory benefit comes from the fact that long-term knowledge (simple counting) and a single learning strategy (applying a single rule over and over) can be substituted for "brute memory" of the digits individually.

- Seifert Seifert Nov 8, 2010

Are Short-Term Memories Processed Sequentially or in Parallel?

The computer metaphor at the base of information processing can mislead some students into thinking that human thought is necessarily sequential--one bit happening only after the previous bit has finished. In fact there is evidence that some forms of thinking happen in parallel--one bit of thought process happening more-or-less at the same time as another.The latter, parallel form of processing is roughly akin to "multi-tasking," and almost always involves a noticeable decrement in performance for each of the parallel thought processes. Here is a classroom activity that illustrates these ideas:

Materials Needed:
  • Two overhead transparencies--one that contains many "x's" and the other that contains only a few "x's". (Follow the links to see scanned examples of these transparencies). You will need to show these for controlled amounts of time to the class as a whole, so you also need either:
  • An overhead project or a computer projector capable of displaying the transparencies. (Personally, I found the old-fashioned overhead projector more convenient for this purpose.)

Procedure--Phase 1: Counting Only One Letter in a Display
  1. Begin with the transparency with lots of x's on it. Announce to students that you are about to show them this transparency for a limited period of time, and that they are to search for and count the number of x's on the transparency.
  2. Tell students that when they have finished counting, they should raise their hand and leave it up.
  3. When about 2/3's of the class have raised their hands, stop the counting task. This usually happens after a couple of minutes.
  4. Important: note how much time has gone by when you stop the counting task. Record this amount.
  5. Check with students to get an informal consensus on how many x's they spotted in the transparency. The exact number found is not important, but it should be in the neighborhood of about 25 x's or so.

Phase 2: Counting Five Letters in a Display
  1. Using the second transparency (the one that contains few x's), now ask the students to count and find the grand total of the following five letters: x, r, s, v, and w. There is no need to keep a count of the letters individually--just find the grand total of all five.
  2. As before, tell students to raise their hand and leave it up when they have finished counting. Stop the process when 2/3s of the class is finished.
  3. Note how much time has bone by when you stopped the counting task this time. Record the amount.
  4. As before check with students about how many letters they found. It should be similar to the number in Phase 1--i.e. about 25 letters altogether.

Phase 3: Counting the Five Vowels in a Display
  1. Using the same transparency as in Phase 2 (the one that contains few x's), ask the students to count and find the grand total of the five vowels in the display: a, e, i, o, and u. As before there is no need to count vowels individually--just find the grand total of all five.
  2. As before, tell students to raise their hand and leave it up when they have finished counting. Stop the process when 2/3s of the class is finished.
  3. Note how much time has bone by when you stopped the counting task this time. Record the amount.

Discussion:
  • If thinking happens sequentially, then Phases 2 and 3 should take five times as long as Phase 1. Chances are, though, that they did not; more likely they took longer, but not five times as longer.
  • This finding suggests that the mind is "cutting corners" when it has to track multiple letters simultaneously--i.e. in some way it is processing multiple letters at the same time, rather than sequentially.
  • The mind is NOT able to process five letters in a way that is fully parallel, however, because counting five letters does in fact take somewhat longer than counting just one letter.
  • As with other information processing tasks, a familiar organizing idea helps in keeping track of new information. In this task, for example, looking for vowels--a familiar grouping--probably went a bit faster than looking for an unfamiliar grouping of the five consonant letters.
  • Bottom line:What do these trends suggest for how students think about new academic information as they learn it? What do they suggest about how teachers might organize it and/or encourage students to organize it?
  • Here is a helpful start on thinking about human multitasking--a form of parallel processing.

- Seifert Seifert Dec 8, 2010