Bottleneck Theories of Attention
Bottleneck theories suggest that as attentional resources are limited, some filtering of information takes place; the issue is where in the system this occurs. Broadbent, investigating attention using auditory stimuli, suggests this happens early on, on the basis of physical characteristics, e.g. the location of sounds of their pitch. He presented three digits simultaneously to each ear, and participants were asked to report the digits either in pairs of by ears. The “pairs” condition, which involves constant switching from ear to ear, is more difficult. The easier “ears” condition reports from one location at a time, supporting the idea of selection on a physical basis.
This theory suggests that nothing about meaning is processed. However, Treismann found that attention is switched to unattended information if this is meaningful, so there is some semantic processing of unattended information. This is supported by the cocktail party effect, where if we are in a conversation with someone, attention is switched if we hear our own name in an unattended conversation. Deutsch & Deutsch challenge bottleneck theories, and suggests that there are no resource limitations on processing; selection takes place at the response stage. Whether selection is early or late appears to depend on the situation. Lavie suggested that when selection takes place is determined by perceptual overload, i.e. the amount of information available to the senses. Where this is high, selection is early, and where it is low, selection is late.