Attention 9/20/2012 8:12:00 PM
Attention
What is attention?
“Everyone knows what attention is. It is the taking possession by the mind,
in clear and vivid form, of one out of what seem several simultaneously
possible objects or trains of thought. Focalization, concentration of
consciousness are of its essence. It implies withdrawal from some things in
order to deal effectively with others” – William James
philosopher
Principles of Psychology (book)
Didn‟t use scientific method
Three different types of attention
Selective attention: choosing where we devote our resources
o Selectively focusing on competing stimuli
o Implies we have control over where we allocate our attention
Vigilance: have to maintain level of awareness and focus to be able
to respond to a rare event (post WW2 research focus)
o Deaths due to human error in war
o Attention was one of human weaknesses
Divided attention: multitasking, dividing attentional resources, cell
phones and driving
o Not good at dividing attention
o Sometimes must divide attention in terms of spatial field
What are your attentional limitations?
Varies within individuals Often better divided attention, harder to selectively attend or vice
versa
Selective Attention:
Most research on attention focuses on selective attention
Auditory
o Classic experiments done in auditory domain
Visual
o Most research, vision is dominant sense
o “rules” of attention often apply across senses
Attention Filtering / Selection
At any one time, you are being bombarded by sensory information
o Cannot process all of that information
o Information passes through filter
Before filter
o Unlimited capacity (auditory/visual system is capable of
handling all of the information in environment)
o All incoming info is processed
After filter
o Limited capacity (hence need for filter, to deal with all this
information)
o Only unfiltered info is processed
Inattentional blindness: failure to see caused by attention
o There is no conscious perception without attention but can
unconsciously detect pattern in the world even without
attention
Change blindness: observer‟s inability to detect changes in scenes
they are looking directly at
o If directly in scene, may need up to 12 observations to notice
o If in periphery, may been up 10 25 observations to notice
Where or when does the filter occur? – 2 major views
Early (Perceptual) Research
o 2 input channels, systems can handle it and detect it o system focuses on one channel over the other and blocks the
other channel
o after filter, can recognize and semantically understand
(extracting meaning)
can only extract meaning from one channel
Late (Response) Research
o Filter occurs later on
o Can detect and recognize and do some semantic analysis of
both inputs (can handle both at once)
o But, can only respond to one channel at a time
Only when you respond can you show your limits and
selectivity
Only attended input reaches consciousness
(Inattentional blindness)
Dichotic Listening Experiments (Tests of Hypotheses)
Stereo headphones
o Different speech message into each ear
Shadowing: asked to repeat speech presented to one of two ears
o Need to ask someone to do something with the information to
make inferences about what they are taking in
o Deliberately ask participant to be selective
Can practically ignore whatever is coming into your
other ear
Unattended channel and attended channel of sound How deeply is the unattended channel (ear) processed? (goal of
asking people to shadow)
o Do people notice the things we are inputting in other ear
o If they do, speaks to how deeply we process
*experiment can be done in visual field how many passes did red team
make? Did you see the dancing gorilla?
Fate of Unattended Stimuli?
If participants only notice physical changes (e.g. pitch, loudness,
basic changes) then only sensory analysis before selection (early)
o Changing from deep male voice to children‟s voice
o Can say if input was speaking, music or silence
o Low level processing
If participants notice changes based on semantics (e.g. language)
change then semantic analysis also before selection (late)
o Also notice high level changes
o To know one language from another is proof that there is high
level processing in other ear
1. Cherry (1953)
Participants noticed:
o Very little in the unattended channel
o Change in speaker (particularly gender)
o Change from speech to tones
Low level changes detected
Participants did not notice:
o Change in language
o Change from forward to backward speech
o Repetitions of the same message (up to 32 times)
High level changes not detected
Support for EARLY selection
2. Moray (1959)
“cocktail party” effect
trying to find evidence that there is, in some cases, interesting
things that are processes in unattended ear In unattended channel:
o “You may stop now” (6% detection)
do people stop when they hear that?
Are supposed to stop, experiment is done when they
hear this in attended ear
o “[participant‟s name], you may stop now” (33% detection)
more processing in unattended ear than Cherry had
stated
Support for LATE selection
3. Treisman (1960)
Instructions: shadow left ear
Capitals were not indicative of volume
Do people shadow the wrong thing?
o Started shadowing the wrong ear due to semantics
support for late selection?
o She said there was an early-late compromise
Early-Late Compromise
filter is located in early position
filter is not always perfect
o can‟t always ignore one channel over the other
o sometimes things leak through filter
o dangerous thing to completely ignore one channel in
environment o always a little bit of processing in case a relevant stimuli
comes in „ignored‟ channel, can change attention
o “cocktail party” effect
many other conversations in room but able to tune
them out, if conversation a few steps away mentions
your name or someone else‟s name you know, this
catches you attention
o can block undesirable stimuli and promote processing of
desired stimuli
o depends on nature of input of attended channel
simple, less analysis needed, more resources devoted
to unattended channel, perceived
complicated, more analysis, less resources devoted to
unattended channel, not perceived
Focused Attention:
We react faster to something that is the focus of our attention
Visual Spatial Attention
Able to allocate visual resources to certain important spaces
Overtly Attending: orient yourself, would to obvious to outside
observer where attention is
Covertly Attending: look out of periphery, not obvious where
attention is, high level sports, trying to hide intentions
Lab 2: Spatial Cueing
where things occur is important
given a cue about where something will happen in space
supposed to use covert attention o valid: cue is correct
o neutral: no cue
o invalid: cue is incorrect
reaction time differences is due to differences in attention
must give participant reason to trust the cues
o 80% valid cues
o 20% invalid cues
o if you didn‟t, participant would ignore the cue completely
because only true about half the time
neutral trials control trial, no bias of attention
valid trials lower response times, advantage
invalid trials higher response times, devote resources to cue and
must make up for the misguidance, disadvantage
Class Results distinct response time advantage in correct cue trials
notice neutral cue is not in middle in their graph
Was individual doing what you asked them to do?
How many trials did you do? (average)
Might not see effects in single individual
Aging and speed?
o Same for valid trials but slower for neutral and invalid trials?
Novelty of cue, we will respond to more quickly (orienting reflexes)
o Flashes in periphery of eye, danger
Stimulus Onset Asynchrony (SOA):
How to you prevent someone from pressing button when they know
button is going to occur, regardless of where
Stimulus 1: cue
Stimulus 2: target
One stimulus occurs before the other and can manipulate period of
time between cue and target
o How quickly can we move our attention around in space?
SOA = 0, simultaneous showing of cue and target
When SOA is short, there is no time for the cue to influence the
focus of attention before the target appears
o Must move attention over in direction of arrow
o Don‟t have enough time to use cue to your advantage
o Thus should be no reaction time advantage for valid trials at
short SOAs
foreperiod = SOA in milliseconds when SOA = 0, no response time differences
o doesn‟t matter what type of cue you had
at 200 milliseconds, maximizes advantage of reaction times
o no further advantage, all you need is 200 milliseconds to take
full advantage of cue
Divided Attention:
2 ways to divide attention
o Space (divide attention across all visual field)
o Tasks (multitasking, trying to do two things at once, what are
the consequences of doing this?)
Lab 2: Slow RTs on neutral trials show the costs of dividing
attention between locations (“advantage” for val
More
Less
