Iconic memory or icon? - 3
Ecological validity of Iconic memory. 1n emphasizing visual and neural persistence in
his target article, Haber has ignored the capacity of the iconic store, thereby overlooking
its ecological validity The ecological significance of the large capacity of iconic memory
was recognized by some of the commentators on Haber's target article (e.g., Loftus I983).
More specifically, the presence of a large-capacity buffer renders selective attention
possible (Coren 1933). The joint presence of' informational persistence and of a large
storage capacity makes it possible to integrate information obtained from functionally
discrete inputs into the visual system (Adelson 1983; Coltheart, 1983; Hauske et al. 1983;
Uttal, 1983).
To Haber, the most serious problem with the iconic store is that its theoretical properties
were established on the basis of an artificial task (see also Haber, 1983r). It can be
argued, however, that there has been a confusion between the source of a theoretical
insight and the converging operations used to study a hypothetical mechanism.
Admittedly, much of what we know about information persistence comes from studies
using single-flash tachistoscope presentation of stimuli, an artificial type of stimulation.
However these studies are not the reasons for postulating a large-capacity temporary
storage in the first place. Sperling (1960) noted that his subjects felt subjectively that they
had seen more than the number of items they could recall when given the typical
attention-span task. This observation is an analog of our common anecdotal experience
that we have seen more than we can describe. So long as the theoretical properties of the
iconic store arc consistent with the real-life phenomenon for which the iconic store is
proposed, the theory of the iconic store is ecologically valid.
Haber, however, found fault with the artificiality of the experimental tasks used to test the
theoretical statements about iconic memory. His metatheoretical assumption in the target
article is that in order to study a phenomenon the task used must mimic the phenomenon.
The more similar the task is to the phenomenon, the more valid are the data.
Loftus (1983) has noted that Haber's metatheoretical assumption is not the one adopted
by most scientists. Loftus's example is that a near-vacuum is an artificial condition. Yet,
much of what we know about gravity is learned from studying falling objects in a
near-vacuum. This example can be amplified by showing that the logic of scientific
investigation requires that the experimental task used to study a phenomenon be an
atheoretical one (Feigl, 1970). As can be seen, an atheoretical task should not mimic the
phenomenon of interest.
To begin with, identifying some event as an instance of a phenomenon requires an
interpretation, a theory-dependent activity. The dilemma confronting a scientist is that,
given any event, it can be interpreted in many different ways. In other words, the same
event may be an instance of diverse phenomena, depending on one's theoretical
assumptions.
At the same time, a hypothesis proposed to account for a phenomenon is of course
consistent with the phenomenon. By the same token, a rival hypothesis is also consistent