The Evolution
19
A look at the above probabilities yields the following simple calcu-
lation expressing the ratio of the likelihood of "toast" to the likelihood
of "orange juice" if both profiles are treated as being equally important:
________P(toast∣Florida) + P( toast ∣ eggs)_________ _ . 0 + ∙ 5 _ 10
P(orange juice I Florida) + P(orange juice∣eggs) .01 + .04 1
This calculation shows clearly and simply that "orange juice," though
receiving support from both profiles, is still only one-tenth as likely
as "toast" and therefore would not be evaluated by memory as being
most likely.
This point is important because "orange juice," as the only
concept common to each list, is the intersection of all those concepts
having whole-to-whole links with "Florida," with all those concepts
similarly linked to "eggs." The calculations shown here demonstrate
that, given the above probabilities, an optimized memory which treated
"Florida" and "eggs" as two sources of statistical information would
find that the phrase Florida eggs suggested to it "toast," rather than
the intersected concept "orange juice.” In other words, if such a
memory were asked the cognitively simple question What is associated
with both Florida and eggs?, it would find "Florida" and "eggs"
suggesting to it "toast” rather than "orange juice,” which is to
say that the first concept that it thought of would not be the answer,
and memory would therefore be forced to employ some sort of trial-and-
error search strategy to arrive at it. This key point will be
explored later in detail.