January 21, 2014
Psyc 330 Lecture 3: Attention and Speech
*IN LATE 1800’s PEOPLE TRIED TO DO PSYC EXPERIMENTS, BUT IT WASN’T UNTIL
AFTER BEHAVIOURIST AREA THAT ATTENTION EXPERIMENTS BEGAN TO BE
CONDUCTED PROPERLY (IN THE 1950’s)
*Look at outline map on Canvas, pretty much sums up the important stuff
1. Part 1: first attempts to study the limits of attention to speech
a. Happened in 1950’s
2. Part 2: Broadbent and the filter metaphor
a. Broadbent studied how well you can listen to two people at one time.
3. Part 3: early selection model OR late selection model
a. Late selection proposed after Broadbent
b. Which is better for understanding attention and speech?
4. Part 4: intermediate selection model
5. Part 5: when in doubt, study the brain
a. EERP study, PET, FMRI data
In world war two, people wanted to know how you can listen to two people at the same time on
How many auditory inputs can be processed at the same time?
• Cocktail party: lot’s of conversations, attempting to listen to your conversational partner,
but can you tune into other conversations around you? Somehow brain allows you to
block out other conversations around you, and you can focus on the conversation you
are currently having.
• Anything salient in another conversation aside from the one you are currently listening
too, it can capture your attention
• In 1950’s wanted to know how our brain operates so that you can block out everything
and only focus on person you are listening too?
How does the brain focus on the only relevant input while at the same time monitoring other
irrelevant inputs. • What happens to speech we block out?
o Clearly not blocking out completely as we are still able to pick up on salient
words which help to capture our attention.
Studying attention to speech
• Dichotic listening: two messages presented at the same time, can listen to one or both of
• Selective listening: ignore one side of conversation, but listen to the other one.
• Shadowing: asked to repeat what they are listening too just to see what participant is
attending too. So listening to two conversations and repeat the message that they pay
attention too they won’t be addled to listen to second message.
o Phrase shadowing: listening to three or four words and then repeating them
back. Listen to phrase and repeat it back.
o Phonemic shadowing: Phonemes = basic sounds you make. Need to repeat
each sound as they hear it. Repeating what you are hearing almost
simultaneously to the person speaking.
Researcher can be even more confident that they are paying attention to
the correct message they should be listening too.
Studying Attention to Speech
What aspects of unattended speech do we notice?
• Can tell experimenter sensory or physical properties of the unattended message.
o Ie. Know that the speaker is a woman vs man, or loud or soft voice, high pitch or
low pitch voice.
• People do not remember meaning of the unattended message
Auditory selection is not observable.
• Mental processes are not observable.
Does presenting speech to the left ear or the right ear make any difference?
• Sensory inputs are contralateral to our brain hemispheres.
o Right ear goes straight to the left hemisphere (language is in left hemisphere.
Broca and wernickes hemisphere).
o Left ear goes to right hemisphere.
Right hemisphere passes over the corpus callosum Part 2: Broadbent and the filter metaphor
Proposed the first attention model
• Broadbent: filter that blocks out other conversations so you can focus on others you
want to listen too
o First attention model.
o Divided attention experiment: subjects listened to digits that were presented to
both ears at the same time.
Wanted to know the order people recalled the digits in.
• Recalling in the order they were presented, or ear by ear recall
(grouped. All digits in left and then all digits in right).
o Found that they were recalled ear by ear (grouped).
o Split/span experiemnt
Early proponent of information theory
• Y tube grating mechanism
o Pipelines, info from right ear and left ear goes to central junction so it looks like a
o Information from the left channel passes through the bottleneck, hinge flap
swings to left side and then information from the right channel passes through
the bottleneck after information from left channel has passed.
Early Selection Model
Behaviourism….the mind is a “black box”
• Can see what goes into the black box and can see what is coming out of the black box
but can’t see what really goes on IN the black box.
• Not into looking at the mental processes
• Stimulus can see what is going on in black box Response
• Wanted to break box down into smaller boxes (functional analysis)
Broadbent’s early-selection model (cognitive approach)
• All speech streams go into a sensory buffer where they are stored and then gets filtered
through the attentional selection stage. o Attentional selection stage filters out the irrelevant speech streams we are not
• Conversational inputs (there can be many different conversational inputs) Sensory
buffer attentional selection semantic analysis longer-term memory attended
• Semantic analysis: speech that makes it through the attentional selection is analyzed for
• Sensory buffer was a novel concept therefore it excited other researchers in box
modelling to be inspired and do more research and fix what is almost a “good” model.
• Critique: broadbents theory is too strict and does not account for the salience of certain
words (ie. Hearing your name) that may capture your attention.