2.3 Towards a theory of parallel processing

When people are asked to guess about masked material, they are commonly able to provide some information, but it often lacks detail. For example, if participants in a Sperling-type experiment have recalled three letters, but are pressed for more, then they can often provide one or two. However, they generally do not know information such as whereabouts in the display the letters occurred, or what colour they were. These, of course, are exactly the kinds of detail that can be used to select items for report, and were believed to be usable in that role because they were characteristics which could be processed quickly and in parallel. The guessing results seem to turn the logic on its head, because the presumed complex information, such as letter identities, is discovered, while the simple colour and position information is unavailable. Coltheart (1980) offered an elegant solution to this problem, built around the semantic/episodic distinction used when describing memory. In the context of letters, semantic information would be the basic knowledge of letter identity. Episodic detail links the general identity to a specific occurrence: detail such as the fact that ‘N’ is in large, upper-case type, and is printed in red and at the start of the sign ‘NO SMOKING’. Coltheart proposed that items do not normally reach conscious awareness unless both the semantic and episodic detail are detected. So, for example, one would not expect to be having an ‘N-feeling’ (semantic) in the absence of a letter with some specific characteristics (size, colour, etc.) in the field of view!

It has become clear from electrophysiological studies that visual item identification occurs in a different region of the cortex from the areas which respond to colour or location. These different kinds of information have to be united, and this process, Coltheart (1980) suggests, takes time and attention. According to this account, Sperling's 12 letters, or even Evett and Humphrey's lion, are indeed processed in parallel to cause semantic activation, but the viewer will not become aware of this, unless able to assign the corresponding episodic details. Nevertheless, if pressed, the participant may sometimes admit to ‘having a feeling’ that an item might have been presented, although not know what it looked like

The important point to note in the above account is that attention is no longer being described as the process that selects material for complex serial processing (e.g. word identification). Instead, Coltheart suggests that attention is required to join the products of two parallel processes: the identification and the episodic characterisation. This idea that attention is concerned with uniting the components of a stimulus is not unlike a theory which Treisman has been developing (after her early auditory attention work, she now researches visual attentive processes). We shall consider Treisman's work (which does not involve backward masking), but first we should look a little further at what masking actually does to the processing of a stimulus.