Partial Report Paradigm

In an attempt to determine whether or not subjects can see more than they can report, Sperling used a task with three stages. First, a fixation point was presented to the subject. Second, a 3x4 character array was briefly flashed on the screen for around 50 ms. Third, after the array was gone, the subject had to report the characters. Initially, Sperling had simply prompted the subjects to report as many characters as they could from the array (the full report condition). He found that subjects were typically able to accurately report only about four of the characters in the array (4.3 characters on average). It was clear from these results that there was a bottlenecksomewhere, which ultimately limited the number of objects that subjects could report. But where is the bottleneck located? Is the bottleneck is located in the information about the stimulus that is accessible by the subject or is it located in the subject’s ability to report the characters?

In order to decided between these two interpretations, Sperling changed the way that the subjects were instructed to report the characters. Instead of reporting all of the char- acters that they could, subjects were presented with an auditory tone that cued which row they should report from. There were three possible tones - a high tone, a middle tone, or a low tone - each of which corresponded to one of the three rows. For example, when a subject heard the high tone that was their cue to report as many of the characters as they could from the top row. The use of an auditory cue to solicit a report was a major im- provement over past implementations of the partial report paradigm, as it allowed Sperling to precisely control the time at which the cue was presented, which, in turn, allowed him to quantify the amount of information about the stimulus available over time. Using this task, Sperling found that, after being trained up on the task, the subjects in the partial report condition could accurately report most of the characters in a given row (3.1 characters out of 4 on average), no matter which row was cued. Since subjects were able to report roughly 76% of the cued characters in the partial report condition, compared to only 36% in the full report condition, Sperling concluded the bottleneck must be located in the subject’s ability

to report the characters in the array, suggesting that subjects have more accessible visual information than they can report, even in brief presentations.

This limited interpretation is generally accepted, even by the cognitive theorists. The results of partial report tasks indicate that there are two kinds of memory. The first is a transient, high capacity form of memory (i.e. ‘iconic’ memory) that contains all of the information about the stimulus that is accessible by the subject. The second is a more durable, low capacity form of memory (i.e. working memory) that contains the information the subject will report. Where opinions sharply divide is on the status of the information in the high capacity ‘iconic’ memory. Is this information conscious or unconscious? That is, does this information take the form of a visual experience as of the stimulus or is it merely unconscious visual information?

Sperling concluded that the information takes the form of a visual experience of the stimulus, saying that this kind of “short-term information storage has been tentatively iden- tified with the persistence of sensation that generally follows any brief, intense stimulation. In this case, the persistence is that of a rapidly fading, visual image of the stimulus” (p.26). Evidence in support of this hypothesis comes from two sources. First, Sperling found that when the subjects were presented with a bright white post-exposure field, instead of the usual black field, that the accuracy of both partial reports and full reports went down significantly. This kind of interference strongly suggests that the accessible information “depends on a persisting visual image of the stimulus” (p.27). Second, subjects in Sper- ling’s partial report task “report that the stimulus field appears to be still readable at the time a tone is heard” (p.20). This is true in all partial report tasks. Subjects generally report experiencing all the characters in the stimulus while it is present and for a short time after it has left the screen. This fits squarely with my own experience in partial report tasks and I predict you will also agree, if try the task for yourself.

If this is right, then subjects in partial report tasks experience nearly all of the charac- ters they are presented with and this experience persists in iconic memory allowing them

to give a partial report in response to a cue that arrives after the array has left the screen. The bottleneck only comes into play when the subjects have to encode and maintain infor- mation about the characters in working memory so that they can issue a report. Thus, the results of the partial report paradigm are often taken to support a mismatch between the capacity of visual awareness and the capacity of visual working memory. This interpreta- tion of the partial report paradigm has been widely accepted (Dretske 1981, 2006; Block 1995, 2007, 2011; Lamme 2003, 2006; Tye 2006, 2009; Prinz 2012). But, since there is significant variation in the views of those who accept this interpretation, I should be careful to note that I am merely claiming here that the subjects are nonpropositionally aware of the specific shapes of nearly all the characters in the array.2

It is important to see how limited this claim is. I am not claiming that the subjects are propositionally aware of the identities of all the characters or of any other facts about them. All I am claiming is that they are nonpropositionally aware of the specific shapes of the characters. If subjects are, in fact, nonpropositionally aware of the specific shapes of 9-12 of the characters in the array but they can only cognitively access and report about 4 of them, then there is a capacity mismatch between these two faculties. The 5-8 remaining characters are instances of nonpropositional awareness in the absence of cognitive access. Since similar conclusions have been reached using partial report tasks with different stimuli and different experimental setups (e.g., Landman et al. 2003; Sligte et al. 2008, 2009), I think that we can conclude that cognitive access isnotnecessary for external awareness.