Iconic memory or icon? - 2
Iconic memory store and Icon. The representation at the level of iconic memory cannot
be an icon for the following reasons: First, consider a brief nine-item display. Are there
nine icons, one for each itch? This possibility implies a unitization mechanism,
something beyond the capability of the retinoreceptors. Alternatively, there may be only
one icon within which the nine items are not differentiated. In this case, an icon is rich in
detail, but it does not have any storage capacity. However, Haber also identifies "iconic
memory" in terms of the observation that subjects' partial-report performance is superior
to their whole-report performance. This requires the assumption of a relatively large
storage capacity. Hence, the hypothetical mechanism studied with the partial-report task
cannot be the static icon (see also Massaro's 1983 commentary). Second, the icon
metaphor would commit one to the "integration" interpretation of backward masking.
This has been shown to be incorrect under some circumstances (Haber 1969; Turvey
1973).
Third, although the representation in the iconic store is "raw," "unprocessed," and
precategorical (Clark 1969; Sperling 1960; Turvey & Kravetz 1970; von Wright 1968),
this characterization is made with respect to a particular frame of reference - namely,
verbal coding. When information is not represented at the verbal level, it is said to be
"raw" or "unprocessed. " This does not mean a complete absence of any processing (Allik
& Bachmann 1983; Boynton 1983; Coltheart 1980; Hauske, Wolf & Deubel 1983; Julesz
1983; Klatzky 1983). A notable example of this type of thinking is the view that iconic
representation consists of visual features (Haber 1971; Haber & Hershenson 1980). That
is, iconic representation is the result of some processing beyond the retinal level, not
literally a retinotopic icon.
Theoretical notion, experimental task, and data. In their commentaries on Haber's target
article, some investigators (e.g., Breitmeyer 1983) have noted that visual persistence may
be found at many different levels of description. Coltheart (1980) has noted that there arc
qualitatively different kinds of persistence - namely, visual, neural, and informational.
What can be added here is that the various kinds of persistence are studied with radically
different experimental paradigms. For example, visual persistence was studied by Haber
and Standing (1969) with a task which relied on the subject's phenomenological report.
Neural persistence was studied by Eriksen and Collins (1967) with a two-part
presentation (or "integration") paradigm involving a forced-choice response.
Informational persistence was studied with the partial-report task by Sperling (1960).
As call be learned from Coltheart's (1980) review, conclusions obtained with a particular
paradigm (e.g., Eriksen & Collins's 1967 integration paradigm) arc not applicable to the
kind of persistence established with another paradigm (e.g., informational persistence as
studied with the partial-report task). This is because the visual system is under different
kinds of constraints in different paradigms. The bulk of the experimental data considered
by Haber deal primarily with visual and neural persistence. Although these data may be
damaging to the concept of a retinotopic icon, they are compatible with iconic memory.