Visual Selection Attention (Lecture 4) Flashcards
What is attention?
Sternberg 1999: “Attention acts as a means of focusing limited mental resources on the information and cognitive processes that are most salient at a given moment.”
Selective attention: selecting from visual (and auditory) inputs
Attention is limited in capacity.
Vision: we can typically see more than one thing in the visual field at once, but how do we select what is important and disregard the rest?- Posner’s (1980) spotlight ‘or torch beam’ model/theory of attention?
Some selection mechanism is necessary to process the info of interest.
What’s the problem?
- attention largely determines what you perceive.
- efficient/reliable selective attention is critical given the cluttered nature of the environment.
- complexity and info overload characterise most environments.
How does visual attention work?- what exactly is selected?
Historically, some debate.
Two camps:
- those that propose visual attention selected from regions of space.
- those that propose visual attention selects instead on the basis of objects.
The space-based view: visual attention is directed toward and selects on the basis of regions of space within a visual scene.
The object-based view: attention directed towards objects rather than potentially empty regions of space.
The space-based view
Visual attention is directed to and selects on the basis of regions of space in a visual scene.
Many analogies have been proposed over the years:
- spotlight (Posner, 1980).
- zoom-lens (Eriksen & St James, 86).
- multiple spotlights (Awh & Pashler, 2000).
All share underlying view that objects fall within ‘beam’ of attention, and are then processed further as a matter of priority.
Object-based view
Attention selects from objects themselves, rather than potentially empty regions of space.
Duncan (1984): suggests that objects, or groups of objects are parsed in accordance with Gestalt laws and are then subjected to further processing (committed to memory).
Space-based view: covert vs. overt attention
Overt = you look at what you are attending to. Covert = attention moves independently of eyes (a mental shift)
Space based view: the cueing paradigm (Posner, 1980)
Ppts have to respond (button press) to the onset of a target on a screen (light or other simple visual stimulus).
Target is preceded by cue whose function is to draw attention to the occurrence of a target in space (comes in different forms- arrows, flashing shapes etc).
As a rule, target detection is fastest when it is presented at the cued location.
Two conditions: valid cues (indicated location of targets appearance); invalid cues (incorrect location).
Typical finding: target performance is faster at the cued location compared to when the target appears in the un-cued location.
Makes use of covert attention (at least when invalid trial).
On 80% of trials the arrow pointed to the direction where the target would appear (valid) and 20% of trials the cue was invalid. Neutral trials presented a double-headed arrow with no cue.
Spatial cueing paradigm (Posner, 80): results
When attention was shifted to the correct area (valid= 80% of trials) RT to target was faster.
When cue was invalid (20%) RT were slower (than valid and neutral conditions).
The slowing of responses on invalid trials was attributed to a three part process:
1) disengaging attention.
2) moving attention to the true location.
3) engaging attention at the new location.
Interpretation: moving attention is a cognitive phenomenon not tied to physical eye movements but to an internal mechanism instead.
Space-based selection: zoom lens model (Eriksen & St. James, 1986)
They argued that the window of attention can be increased or decreased according to task demands.
- eg. when driving a car, you attend to as much as possible, but if something runs out in front of you, your attention focuses on that event.
Spotlight vs zoom lens models study (LaBerge, 83)
Ps saw series of 5 letter words. Occasionally, a probe requiring a rapid response was presented instead of, or after, the word. The probe could appear in any locations of the 5 letters.
Two conditions:
- focused attention: ps asked to categorise middle letter of the word (focus on a narrow space).
- unfocussed condition: ps asked to categorise the whole word (spreading attention wider).
Findings:
- in focused condition, detecting the probe was fastest when it appeared at the central letter (the one Ps were told to focus on).
- in unfocussed condition, detecting the probe was equal for all 5 letter locations.
- shows attentional ‘spotlight’ does appear to be able to zoom in and out.
Selective-visual attention: multiple spotlights model (Awh & Pashler, 2000)
Propose that we may use multiple spotlights of attention.
Could conserver cognitive resources by avoiding attending to irrelevant regions of visual space that fall between relevant areas.
Selective visual attention: space-based view- summary
The multiple spotlight model is the most recent.
Whichever view is the ‘right’ one, attention can select from regions of space.
The object-based view: typical manipulation
If attention can select objects, then overlapping objects should be selected independently of one another.
That is, there should be difficulty in attending to two objects at the same location.
If attention selects space, then both objects should be process at the same time, as both fall with the ‘beam’ of the spotlight.
OBS study by Duncan (1984)
Found: ps slower to report two properties that belonged to different objects, compared to two properties of same object.
Two-object cost:
- attention selects objects.
- if attention selects space (rather than object), the two-object cost would not arise as both objects would be selected (as they appear at same location).
Summary
Visual selective attention: selection process allowing task-relevant info to be extracted from the scene.
Historical debate between object-based and spaced-based theorists:
- space-based view attention selects from regions of space (what falls inside is processed).
Likened to: spotlight; zoom-lens; multiple spotlights.
- object-based view: attention can select from objects, even if they overlap at the same location.
