Attention

Psy315 - Class 10

(This page last updated 12 October, 2006.)

 What do we mean by attention?

• Selection
  • To cope with limited processing capacity
  • To choose which inputs will govern response
• Differences in task demands
  • Attention focused narrowly
  • Attention spread widely

Most early work in attention was done in the auditory domain.

Shadowing Task

• Dichotic Presentation
  • A different speech stream in each ear.
• Shadowing
  • Repeat one stream.

Early Results from Shadowing Experiments

• Good performance at repeating shadowed stream.
• Could detect some changes in unattended stream
  • Switch from voice to tone
  • Switch from male to female voice
• Could not detect other changes
  • Switch to new language
  • Forwards vs. backwards speech

Later Studies using Subject's Name

• Subject's name included in unattended stream.
• Sometimes (but not always) detected.
  • "You may stop now" detected 6%
  • "John Smith, you may stop now" detected 33%
• Name must be processed fully enough to be identified, at least part of the time.

Alternative Explanations for Shadowing Results

• Attentional filter lets info from unattended stream slip through occasionally.
• Attentional filter attentuates unattended stream rather than blocking it completely.
  • Detectors with low thresholds, such as that for your own name, can still be triggered.
• No filtering: Many words identified but soon forgotten.

Conclusions from Auditory Experiments

• Cannot conclude that early auditory filtering is perfect.
• Also difficult to conclude that unattended stream is processed as fully as attended stream.
• Probably some degree of attenuation of unattended stream based on simple properties such as location and pitch.

More recently, much of the research effort has been devoted to visual attention.

Inattentional Blindness

• Experiment by Mack and Rock
• Basic task: Is cross wider than it is tall?
• Cross only visible for 200 msec
• Task requires focused attention.

  Regular Trial Display

  

• On critical trial, square appears with cross.
• 25% of subjects do not see it.

  Critical Trial Display

  

Change Blindness

Surprising that some changes in visual images so hard to detect.

• They are usually fairly large changes.
• They are often in central positions where viewers will be fixating.

Why are these changes so hard to detect?

• They do not change the overall meaning of the scene.
• Both versions are consistent with the "gist".

The changes are only detected when you are already attending to the relevant part when the change occurs.

• Attention is necessary for a full and complete representation.
• Attention is limited to one part of the scene at any one time.
• Is it possible to have a full and complete representation of the entire scene in memory all at once?
• We remember the overall meaning of the scene (the "gist"), but not the details.
  • But we have the impression that we have a rich and complete perception of each visual scene.
  • These results suggest that at any one time, we "see" less than we think we do.

next class: Attention to Locations and Objects

More demonstrations of change blindness and inattentional blindness are available at the Change Detection Database webite.

Psych 315: Cognitive Psychology Kyle Cave Psychology Dept. U. Mass.