Loftus, Johnson, and Shimamura (1985)
Loftus et al. (1985) propose a method for investigating the "worth of an icon" by asking how long a stimulus with no iconic image must be presented to produce a level of performance equivalent to performance produced by a stimulus followed by an icon. That is, if d is duration for the stimulus that produces an icon, and d + a (a = additional exposure) is the duration of the stimulus that does not, then what level of a will produce equivalent levels of performance?
Stimulus 1: | + | . . . . . | (p = 100)
Stimulus 2: | + | (a) | (p = 100)
The value for a is estimates the worth of the icon in terms of additional physical exposure. Mathematically, the worth of an icon may be expressed as p(d, i) = p(d + a, 0 i). The implication of this relationship is that "information extracted from the icon might be parsimoniously characterized in terms of the information that could be extracted from a physical extension of the stimulus" (p. 2).
Subjects were shown pictures in brief displays, and then performed a recognition test in which the goal was to discriminate targets and distractors. The test phase followed presentation of all of the test stimuli. The following variables were manipulated: stimulus duration and noise mask (immediate, delayed, or none). The mask-delay intervals in the immediate and delayed mask conditions were 0 ms and 300 ms, respectively.
As expected, in all three mask-delay conditions, performance increased as exposure duration increased. Moreover, performance in the delayed-mask condition was higher than performance in the immediate mask condition. In addition, performance in the no-mask condition was slightly higher than performance in the delayed mask condition. This finding indicates that a mask disrupts performance, even at a 300 second delay. Loftus et al. conclude that because the duration of the iconic image is less than 300 ms, this effect must be attributable to something other than extraction of information from the icon.
What is the iconís worth? The principal question of the study concerned the difference between the immediate and delayed mask curves. Loftus et al. found that he curves were separated by approximately 100 ms at all stimulus duration levels. They concluded that "having an icon is equivalent to having approximately 100 ms of additional physical exposure duration" (p. 4). What are the implications of this finding?
A memory representation can be conceptualized as a set of information about the remembered stimulus. Based on this conceptualization, Loftus et al. offer two interpretations, one strong and one relatively weak, of their results. The first is that an icon is worth an additional 100 ms of physical duration in terms of a subset of the information that constitutes the memory representation. That is, a subset of information is extracted from the icon. For example, if a memory representation is based on spatial, color, and identity information, then the icon might only provide spatial information. The second is that the icon is identical to the set of information that makes up a memory representation. That is, "the memory representation that accrues from . . . an icon is identical in all respects to the memory representation that accrues . . . without an icon" (p. 4). With reference to the previous example, spatial, color, and identity information would be provided. The two alternatives are represented below:
Alternative 1: */*/*/*/*/*/* * * * * * * *
A subset of information persists
Alternative 2: */*/*/*/*/*/* */*/*/*/*/*/*
All information persists
The interpretations described above can be investigated by assessing memory for different types of information following stimuli presentation.
In Experiment 2, subjects were shown complex pictures (e.g., cityscapes) at various display rates, and with various mask intervals. Immediately after each presentation, they wrote down as many details as possible. Subjects scored their own data on the basis of detail inclusion. Loftus et al. found that subjects performed the response task with ease, and that responses were qualitatively similar. Moreover, the pattern of relations between stimulus duration, mask interval, and performance found in Experiment 1 was replicated. That is, there was a performance advantage for the delayed mask over the immediate mask, and the interval separating the curves (105 ms) was constant across levels of duration.
Experiment 3 was a replication of earlier experiments, with the exception that the dependent variable was a subjective rating of how well the pictures would be remembered. The same pattern of relations emerged: main effects of stimulus duration and mask condition, and no interaction.
The prior results suggest that the information extracted from an icon is the same as information extracted from the stimulus given 100 ms of additional exposure duration. Loftus et al. suggest that this finding might suggest that similar encoding processes are responsible for information extraction from stimulus and icon. To explore this possibility, stimulus luminance was manipulated. The prediction was that "The information extracted from the icon must still be equivalent to the information extracted from an additional 100 ms of the dim picture itself" (p. 8). In short, stimulus worth should be independent of luminance. The results supported Loftus et al.ís prediction: icon worth was 110 ms for bright and dim stimulus presentationsóthat is, an icon should be worth 110 additional seconds of physical exposure regardless of luminance.
The worth of an item in terms of additional exposure time is independent of stimulus duration. Loftus et al. explain, however, that this does not mean that the amount of information extracted from an icon (and the resultant increase in performance) is independent from stimulus duration. In fact, the amount of information extracted declines with stimulus duration. Thus, an icon makes a progressively smaller contribution to performance as a stimulus persists. But, the point is that an iconís worth remains constant regardless of its relevance to performance.
Loftus et al. also advance the claim that, while an immediate-mask and a delayed-mask stimulus can be distinguished perceptually, information extracted from stimuli presented under these conditions may be represented the same in memory. This claim is supported by the findings that 1) different techniques produced the same estimate of the iconís worth and 2) different memory assessment techniques produced same estimates.
Icon worth versus icon duration
It is important to distinguish between an iconís worth an its duration. According to Loftus et al., if an icon remained identical to the physical stimulus for the extent of its persistence and then disappeared abruptly, icon duration and worth would be the same. But, because the icon gradually fades, itís worth is shorter than its duration (usually estimated at about 250 ms.) That is, while the first 100 ms of the persistence are worth 100 additional ms of exposure time, the image during the remainder of persistence is degraded to such an extent that it has no worth in terms of its effect on performance.