Out of sight, not out of mind : visual processing and immediate memory

The following thesis details 15 experiments using computer generated stimuli and extending the known aspects of perception to the early stages of visual information processing. A standard procedure in information processing is to use a backward masking paradigm to limit the amount of time for which the stimulus is available. Initial experiments established that a secondary non-visual task which required active attention increased the time required for correct discriminations while secondary unattended stimuli did not. These divergent findings were evaluated in conjunction with other theoretical considerations and re-analyses of existing data. It was concluded that, contrary to previous assumptions, the backward masking procedure was primarily reflecting factors such as strategy use rather than speed-of-processing per se. An alternative paradigm for investigating the nature of visual information processing is to present a series of stimuli which must be reported. Previous attempts to infer the structure and nature of the processes involved have typically used interference from secondary tasks which require conscious effort or attention. The use of any dual-task paradigm is confounded by difficulty in separating out the involvement of general-purpose resources. A,major concern of the subsequent experiments was to investigate whether it was possible to selectively impair performance by irrelevant peripheral stimuli. Initial experiments established that this could be done and that the presence or absence of interference was altered by changes in the target or surround in a fashion not predictable from their overt similarities. Rather, it is argued, they are explicable only in terms of whether or not common visual subsystems were involved. These conclusions were supported by experiments which yielded a successful double dissociation between the location and object characteristics (both colour and pattern) of the task. In conclusion it is argued that a sound appreciation of the underlying visual mechanisms is essential to the understanding of the early stages of the processing of visual information. In the earlier stage this approach has led to a re-evaluation of a widely accepted paradigm, in the latter portion it has led to support for the notion that spatial and object characteristics are coded and stored separately. Implications of this approach in other areas are discussed.