Home Research The Temporal Huddle
Optimizing Distributed Practice: Theoretical Analysis and Practical Implication

March 18 2009
Rankin Williams, a graduate student in the Gauthier lab at Vanderbilt, led a discussion with members of both the Gauthier and Palmeri labs (PEN) about an article by Cepeda and colleagues (IMS). Rankin contacted the authors with the following questions that emerged during the discussion (between Rankin Williams and Hal Pashler):

(Hal Pashler) First off, we have another, closely related, new paper that reflects a more comprehensive dataset, in certain respects, so you might want to look at this one, too:

(Rankin Williams) First, since performance in the final session in your study is actually pretty low (20% for objects in experiment 2), we were wondering what would happen if subjects were given a chance to re-learn the material at that point. Would savings in learning show the same dependence on gap as memory does?

Yes, I would guess that it would do so. The only dataset I know that was on point was collected by Tom Nelson, who unfortunately died a couple of years ago, without publishing these data. I even tried to get some of his powerpoints to post them on the web so his fascinating and hard to replace results would be available to interested researchers, but I wasn't able to make that happen.

We also discussed the differences in practice effects between memory tasks and perceptual learning tasks (where, if we remember well, the main story with gap has to do with improvements after sleep, not unlike your 1-day effect in Experiment 1).

Aah, but don't believe everything you read: :-)


Interestingly, perceptual expertise studies training people to individuate often use a task that involves memorizing the names of objects. In the end, we do not really care if they forget the names after the study- which is why we use post-tests including novel objects. We assume that the perceptual skills acquired outlive the memory of the name to object associations that we teach.

See also this: http://nkornell.bol.ucla.edu/Publications/Kornell.Bjork.2008a.pdf
(In a lab class, we just replicated the first study here but using 3 different categories rather than such a large number--figuring that this placed more emphasis on the perceptual learning and less emphasis on the name learning--and while we got a slight advantage for the "spaced" condition, it was far less than Kornell and Bjork got.)

Also note that K&B (understandably) confound temporal spacing with interleaving. It wouldn't surprise me at all if it is the interleaving that is making the difference here, rather than the temporal spacing.

Tim Curran and Chris D'Lauro's study on spacing during the acquisition of expertise suggests that spacing benefits the acquisition of perceptual skills too. Are there reasons to think that the two spacing effects are related (i.e. would optimizing distributed practice for retention of the associations and acquisition of perceptual skills necessarily yield the same optimal gap?)

Great question. Short answer: we don't know. These very long term studies take so much effort over such a long period of time that we are kind of burned out on them for a while. :-)

It is my sense that in most skill learning situations, forgetting rates themselves appear much more gradual, and in the context of slow forgetting, it is very hard to demonstrate powerful spacing effects--could be because the effects sometimes aren't present, or could be because spacing in general is retarding forgetting, so if forgetting is minimal, spacing effects are hard to detect even if very real. We are trying to explore the basic question of why there seem to be slower rates of forgetting in skill like tasks--something that Healy and others have worked on as well. I am hoping we may have a bit of new light to shed on that before long.
Rankin Williams - This e-mail address is being protected from spambots. You need JavaScript enabled to view it
Hal Pashler - This e-mail address is being protected from spambots. You need JavaScript enabled to view it