Selective Attention & Multitasking

• Our senses are constantly stimulated by many information sources – visible objects, sounds, surfaces in contact with the skin, smells, proprioceptive and other sensations from within the body
• At any given moment, we seem able to process in detail the information from only one or two such sources, though we may be aware of the presence of many others in the background
• In this lecture: 'I ask three linked questions: When and why is it hard to divide attention between information sources? When we focus attention on one source, to what level are unattended sources processed? What is attentional selection for?'
• Early research on attention

  1950s-1970s
• Early research on attention

  Used listening tasks with two streams of spoken words, often one played to each ear via earphones (dichotic listening). Subjects tried either to shadow (repeat) one while ignoring the other (focused attention), or to divide attention between them
• Main question in early research on attention was how early in processing attentional selection took place
• More recent work on attention

  1980s on
• More recent work on attention


  Mostly focused on visual attention, often using reaction time methods. Converges on the conclusion that attentional selection is multilevel, flexible, and responsive to processing load. Focus now on functions of attentional selection such as feature-binding and selection-for-action
• When one tries to divide attention between simultaneous speech messages, comprehension is poor
• We are good at focusing on one speech message and ignoring others provided that the messages differ in physical characteristics, not just lexical or semantic properties
• Broadbent's model (1958)


  Explained attention to spoken sources in terms of a switchable "filter" located relatively early in the processing stream
• Unattended (ignored) speech could (at least partially) activate meanings ("breakthrough" phenomena)
• Treisman's modification (1969)


  Proposed early attentional filtering that attenuates, not completely blocks, information from ignored sources. Selection also happens later in processing
• In vision, attention is usually coupled to gaze direction
• We look at the object to which we are attending to bring the retinal image onto the fovea, where acuity is maximal and colour discrimination good
• Experiments suggest little semantic analysis of unattended streams
• Attention in vision


  • Usually coupled to gaze direction
  • Bringing the retinal image onto the fovea for maximal acuity and good colour discrimination
  • Can attend to parts of the visual field away from fixation (covert attention)
  • Attention can be "covert" or "overt"
  • Attention can be attracted by sudden onsets, changes, or movements in the visual field
  • Exogenous shifting of attention is more rapid and automatic than deliberate "top-down" controlled or "endogenous" shifts of attention
  • The "spotlight" of visual attention is flexible, it can be broad or focused
• Posner and colleagues cueing paradigms

  Measure the deliberate orienting of covert attention away from fixation to a visual location in anticipation of a stimulus occurring there, without moving the eyes, via the processing advantage conferred by attention
• ERPs and fMRI show enhanced activation in areas of extra-striate and striate visual cortex


  Corresponding to a deliberately attended region of the visual field away from fixation
• Perceptual load theory postulated by Lavie's hypothesis
• Early researchers like Broadbent tended to think of selective attention as defensive
• Focal attention is needed not only to allow the object's features to access categorical representations
• Focal attention is the mechanism that integrates or binds features at the same locus in the visual field to form stable object representations
• Combining features involves a basic bottleneck between sensory processes and the representation of objects for action and memory