Iconic memory or icon? - 4
with the phenomenon of interest. In testing a hypothesis about a phenomenon, the more
similar the experimental task is to the phenomenon itself, the less discriminating it is
regarding the relative success of the rival hypotheses. This general argument can be
illustrated with the following trivial example (see Table 1).
Table 1. Schematic representation of the trivial example
Phenomenon:____ |
Wet highway____________ |
Wet highway____________ |
Explanation: |
It rained. |
Someone washed the |
Implications:______ |
(a) Wet highway.___________ |
(a) Wet highway.__________ |
(b) Wet side lanes |
(b) Wet side lanes | |
(c) Wet lower parts of lamp |
(c) Wet lower parts of lamp | |
(d) Wet leaves on treetops. |
(d) Dry leaves on treetops. |
Suppose that travelers A and B find the highway wet on their way back to their
hometown after a trip. Traveler A hypothesizes that it has rained, but B suggests that
someone has washed the highway. How can this dispute he settled? First, traveler A has
to work out what else must be true, other than the obvious implication indicated in (a), if
it has rained. Two such additional implications have been tabulated in the left column of
Table 1. Traveler B's hypothesis likewise implies events other than a wet highway (see
the right column).
Following Haber's metatheoretical assumption regarding ecological validity, successive
sections of the highway should he looked at in order to corroborate traveler A's
hypothesis. A less satisfactory piece of evidence to look for is whether side lanes in the
city are wet. These choices are made because the former is identical, and the latter is
similar, to the wet-highway datum. However, neither piece of evidence is effective in
discriminating between the hypotheses because either of them can readily be accounted
for by traveler B's hypothesis. Obviously, what is needed is a condition which will occur
if A's hypothesis is true, but not if B's hypothesis is true. As can be seen from Table 1, the
best data to look at are the leaves on top of a tall tree. Without doubt, leaves on tree tops
are very different from a highway. Yet, the discriminating test appeals to the leaves, not a
different section of the highway or the side lanes. In other words, the evidence used to
support the theory should not mimic the phenomenon to be explained by the theory under
investigation (a paint also made by Uttal, 1983).
A subject's response to a single-flash tachistoscopic presentation of stimuli is admittedly
very different from normal casual glances. It is also true that the iconic store is postulated
for visual information processing outside the laboratory However, the implication of our
trivial example is that the artificial nature of tachistoscopic presentation does not
diminish the ecological validity of iconic memory, the intuition for which comes from an