The Evolution
39
That this order effect should manifest itself again makes sense, as
the restricting adjectives used (see Freedman & Loftus, Table 1) could in
general be expected to operate by part-whole links (e.g. white, yellow,
liquid, green, enormous, sma11, long, etc.), and the noun categories used
in general could be expected to operate by whole-to-whole links (e.g. the
concept "food" probably has whole-to-whoIe links with the various types
of food; and the concept "president" probably has whole-to-whole links
with all of those diverse people who have been president, and so forth).
The order effect probably would have been larger if care had been
taken to choose word pairs that must be treated as Type II intersections
(as, for example, Africa striped intersects the concept "Zebra"). Thus
some of their word pairs appear to allow Type I intersection. For instance
it is conceivable that memory treats bird followed by yellow as a Type II
intersection, and yellow followed by bird as a Type I intersection, thereby
evading an order effect. This is a possibility because the concept "bird"
probably has whole-to-whole links with the different forms of "bird"
(e.g. "sparrow," "ostrich," "hummingbird," "vulture," etc.), as well as
part-whole links with all those forms of "bird" that are prototypically
"bird-like" (e.g. "sparrow," "robin," and "canary," which have in common
pretty much the same shape).
There is every reason to expect that memory makes liberal use of such
redundancies in storing information, as there is no rule forbidding a com-
plex concept such as "bird" from having whole-to-whole links with the
different forms of "bird," as well as an additional set of part-whole links
with those forms of "bird" involving the same shape and design. It would