of rationality cannot apply to memory (or any other human domain, including economic behavior) without a framing of what the problem is. Conceivably, one might frame the problem of human memory differently and come up with different predictions. I would be less than honest if I did not admit that my knowledge of human memory influenced my framing of the basic memory problem. However, I think the framing is quite reasonable.
One of the reasons why I have stuck with ordinal predictions and refrained from mapping this analysis onto interval predictions about time and probability is that I cannot see an equally plausible way of further framing the problem to make that mapping. Thus, if we are going to take this rational analysis of memory beyond these ten first-order predictions, the major agenda item is study of the task facing human memory--that is, the task that the system has evolved to handle. Perhaps we can gather evidence for some way of further specifying the memory task.
Although the present results are preliminary, they do support the hypothesis that we can predict the phenomena of human memory from the assumption that it operates rationally. What are we to make of this result if it continues to hold up under further analyses? One might take the attitude that the experimental study of memory is unnecessary because its behavior can be deduced from a priori premises. I do not think this is the correct conclusion. It ignores the fact that it is not certain a priori how to frame the problem faced by human memory so that we can propose a rational solution. The experimental research provides guidance here. Moreover, the rationality hypothesis, even with a framing, is just a scientific claim and still requires experimental test. However, I do think that the hypothesis throws an amazing light on the years of experimental research into human memory. It seems that these experiments may have been telling us that human memory was designed rationally.
I should acknowledge Lynne Reder's invaluable contribution to this paper. If she had not got me reading and thinking about the work in information retrieval, I would never have discovered the framing of the memory problem that I have presented. Lynne also went through the manuscript with me to help assure that I got it right. I am also grateful for the comments of Bob Bjork, Gus Craik, and Roger Ratcliff. This research was supported by grant BNS 8705811 from the National Science Foundation.
Anderson J. R. ( 1978). "Arguments concerning representations for mental imagery". Psychological Review, 85, 249-277.
Anderson J. R. ( 1983). The architecture of cognition. Cambridge, MA: Harvard University Press.