Transcript Attention

Attention
• Determines which codes get processing
• Often associated with conscious awareness
• A continuum that varies with the amount of
conscious awareness and effort
conscious decision processes
(require attention and effort)
© 2001 Dr. Laura Snodgrass, Ph.D.
unconscious monitoring
(effortless)
1
Theories of Attention
• All theories agree on limited capacity and selectivity
• Disagree on location of limits and how selection is
done
• All share the basic information processing premises
– disagree on order of information flow
© 2001 Dr. Laura Snodgrass, Ph.D.
2
Basic Information Processing
Stimulus
Sensory Register
STM (working Memory)
LTM
Response
© 2001 Dr. Laura Snodgrass, Ph.D.
3
Dichotic Listening
• Common research technique to investigate selective
attention
• One message is presented to one ear and a different
message is presented to the other ear
• Subjects are instructed to shadow one ear
– shadow = repeat out loud what they hear
• What happens to your memory for the information?
© 2001 Dr. Laura Snodgrass, Ph.D.
4
Early Selection Model
Stimulus
Sensory filter theory (Broadbent)
Sensory Register
Sensory FILTER
STM limited capacity
LTM
Response
© 2001 Dr. Laura Snodgrass, Ph.D.
5
Early Selection
• All information enters the sensory register
• FILTER selects based on physical characteristics
– Only what passes through filter has access to LTM
– No selection based on meaning because meaning
is stored in LTM
© 2001 Dr. Laura Snodgrass, Ph.D.
6
Experiments that support early selection
• Cherry (53) - dichotic listening
– no memory for information in unattended ear
– not notice change from English to German
• Neisser & Becklen (75) - superimposed video images
– ballgame and handslap
– attend to one and not notice bizarre events in the other
© 2001 Dr. Laura Snodgrass, Ph.D.
7
Contradictory Results
• Cocktail Party Phenomenon
– own name is recognized in unattended ear
• Treisman
– subjects switch shadow to follow message when
message is switched to “unattended” ear
– only is message continues in other ear
• Both results indicate knowledge of meaning from the
unattended ear
© 2001 Dr. Laura Snodgrass, Ph.D.
8
Attenuator Model
• Treisman (64)
– How can subjects be influenced by the meaning in
the unattended ear?
– Unattended information is only dampened (attenuated)
not filtered completely
– Significant information gets through the filter
© 2001 Dr. Laura Snodgrass, Ph.D.
9
Late Selection Model
Norman (68)
Stimulus
Sensory Register
LTM
Selection based on pertinence or
saliency mechanism
STM
Response
© 2001 Dr. Laura Snodgrass, Ph.D.
10
Late selection
• All incoming information activates LTM
• Saliency
– expectancy or constant monitor for some stimuli
• Pertinence
– important to you
• Information with highest activation is selected for
response
© 2001 Dr. Laura Snodgrass, Ph.D.
11
Comparison
• Differences
– location of filter (before or after LTM)
– basis for selection
• Similarity
– no permanent memory for unattended (not selected)
information
© 2001 Dr. Laura Snodgrass, Ph.D.
12
Experiments supporting late selection
• Lewis (73)
– unattended information increases RT to attended IF
semantically related
• e.g. RT to animal names
• Mackay (73)
– unattended disambiguate the meaning of attended
• “they threw money towards the bank” in one ear
• “money” or “river” in unattended influences interpretation
• GSR study
– reaction to words associated with shock even if presented in
unattended ear
© 2001 Dr. Laura Snodgrass, Ph.D.
13
Capacity Theory
• Selective attention as the allocation of capacity, NOT
a block or filter
• Attend to more than one thing at a time
– divide up attention
• Shiffrin and Schneider (77)
– finite capacity to be divided
– attention = process of allocating resources to various
inputs
– divide attention among automatic and controlled
processes
© 2001 Dr. Laura Snodgrass, Ph.D.
14
Automatic and Controlled Processes
• Automatic
• Controlled
– not use limited capacity
mechanism
– not use working memory
– not interfere with other auto or
control processes
– occur in parallel
– effortless
– not interruptible- once initiated
continue to completion without
control (e.g. Stroop effect)
– does NOT lead to learning (no
LTM)
© 2001 Dr. Laura Snodgrass, Ph.D.
– use limited capacity
mechanism
– limited # can operate at once
– usually sequential
– necessary for learning (LTM)
– two types
• conscious and
accessible
• veiled, unconscious, e.g.
a memory search
15
Research on the capacity theory
• Distinguishing automatic and controlled processes
– dual task - if performance decreases then both must
be controlled; if no decrease then one or both
automatic (or not exceed capacity)
© 2001 Dr. Laura Snodgrass, Ph.D.
16
Research on the capacity theory
• Overlearning - many controlled processes can
become automatic with “overlearning”
– motor skills
– components of reading (Stroop)
– letter search task
– practice to avoid panic
• emergency response becomes automatic
– concurrent tasks e.g. reading and taking dictation
– retrieval of test items
© 2001 Dr. Laura Snodgrass, Ph.D.
17
Research on the capacity theory
• Investigations of
– allocation
– vigilance
– switching
– selection
– flexibility
• focused vs. divided attention
• e.g. jet fighter pilots and bus drivers
© 2001 Dr. Laura Snodgrass, Ph.D.
18
Research on the capacity theory
• Performance under two types of constraints
– data limitations
• not enough data, no matter how much cognitive
resources are applied
• e.g. poor quality copies, asked to land a jet
– resource limitations
• can do better if pay more attention
• e.g. listen to lecture and take notes
• Performance is limited by the demands that the task
places on the cognitive system
© 2001 Dr. Laura Snodgrass, Ph.D.
19
Research on the capacity theory
• Measure capacity demands of a given task by amount
of interference
• Secondary Task technique and cognitive effort
– the harder the primary task the poorer the performance
on the secondary task
– star tracing experiment
© 2001 Dr. Laura Snodgrass, Ph.D.
20
Secondary Task Technique
• Ellis
– memory increases with increasing cognitive effort
– measure effort with secondary task
– primary task is hard or easy anagrams
• ootdcr vs. tordoc
– secondary task is tone detection (probe)
– found - memory and RT to probe increases for
harder anagrams
– measure difference in capacity demand
© 2001 Dr. Laura Snodgrass, Ph.D.
21
Secondary Task Technique
• Posner and Boies
– letter matching and tone
detection
600
500
– look at RT to tone during
different points in the task
shows that cognitive load varies
during the task
400
– notice RT during rehearsal and
during decision
100
300
RT
200
0
warning
© 2001 Dr. Laura Snodgrass, Ph.D.
first
second
22
Secondary Task Technique
• Keele - investigate cognitive effort of different tasks
• Assumption - each task has two stages
– receive sensory input and activate memory
– mental operations based on memory activation
• mental operations such as:
• recognition and search, compare and match,
rehearsal, response initiation, movement correction,
counting (beyond subitization)
• Stage one takes little (or no) attention (effort) but stage
two does
© 2001 Dr. Laura Snodgrass, Ph.D.
23
Other theories of attention
• Logogen model - levels of activation in memory
– as concept reaches a threshold of activation you
become consciously aware of it (pay attention)
– preset criterion level of activation
• ignore = set high
• expectation = set low
• redundancy and context = low
• spatial location and physical characteristics can be
used to set high or low
© 2001 Dr. Laura Snodgrass, Ph.D.
24
Other theories of attention
• Posner and Keele
– attention = mechanism that coordinates the different
codes
– optional filtering of input based on physical
characteristics
– cost and benefits of presetting attention
– individual differences in flexibility
• focused and divided attention
• Treisman and Gelade
– attention glues the codes back together
© 2001 Dr. Laura Snodgrass, Ph.D.
25