Cognitive Psyc. Chapter 3: Attention
The primary goal of attention research is to understand which
information is selected, how it is selected, and what happens to both
selected and unselected information.
External attention refers to attending to objects in the environment or
to specific features of those objects.
These objects can be of diff modalities, location, point in time, etc.
o All of these object characteristics can influence external
E.g. with Sperling’s partial report task… the tone cued
what row to attend to. They would only be able to
attend to that row if info still available in visual
Internal attention refers to regulating our internal mental life such as
planning what to eat for dinner.
It controls how information is managed in ST working memory and
Important distinction: EA = focus on perceptual objects as opposed to IA =
focus on trains of thought.
Milgram proposed that in midtown Manhattan you can spot 220,000 people
within a 10min radius of ones office…Compares to information overload … in
which perhaps one spends less time on each input, disregards low-priority
Bottleneck theory: that selection is necessary whenever too much
information reaches a bottleneck – a stage that cannot process all of it. In
other words.. how ppl select info when some info processing stage becomes
overloaded with too much info.
Concentration: investing mental effort in one or more tasks
Mental effort: The amount of mental capacity required to perform a task. Capacity theory: A theory proposing that we have a limited amount of
mental effort to distribute across tasks, so there are limitations on the
number of tasks we can perform at the same time.
Broadbent’s Filter Model
Study: Navy guys listened to 3 pairs of digits. E.g. 73-42-15
One digit of a pair was announced at one ear, while the other digit of a pair
would be at other ear. E.g. 741 to right ear… 325 to left ear.
Pairs separated by a 0.5 second interval
Subjects asked to report the digits in whatever order they choose
RESULTS: able to report 65% (remember 65 as in beatles as in SARGEANT
peppers lonely hearts club band) of the lists correctly.
Note: almost ALL the correct reports involved recalling all the digits
presented to one ear, FOLLOWED by all the digits presented to the
other ear. E.g. subject would say 741-325, or perhaps 325-741.
Instructed to recall the digits in order of arrival: the first pair of digits,
followed by the second pair, etc.
Pairs separated by a range of 0.5 to 2 second interval
RESULTS: Performance was better at the longer intervals… nevertheless it
was much worse THAN when subjects could recall the digits heard in
one ear and then the other ear.
Broadbent accounted for his findings by using the FILTER MODEL!
Filter model: The proposition that a bottleneck occurs at the pattern
recognition stage and that attention determines what info reaches the
pattern recognition stage.
The mechanical model consists of a Y-shaped tube and a set of
identifiable balls. The tube has a narrow stem that can accept only
a single ball at a time (the limited capacity perceptual channel), but upper branches (the sensory store) are wider and can accept
more than one ball at a time. At the junction of the stem and
branches is a hinged flap = the filter, which can swing back and
forth to allow balls from either branch of the Y to enter the stem!
Limited capacity perceptual channel: The pattern recognition stage of
Broadbent’s model, which is protected by the filter (attention) from
becoming overloaded with too much perceptual information.
The bottleneck relates to the navy guys experiment cuz if they were forced
to report the digits as they arrived, the flap would have to be shifted back
and forth to allow balls to enter the stem in the order in which they arrived.
The easiest case should be when the listener can report all the digits
entering one ear before reporting all the digits entering the other ear.
Therefore only a single shift of attention is required. BUT, the shift has to
occur BEFORE the information entering the unattended ear decays from the
Limitation of Filter Model: the sensory store would have to last fairly long
to operate as proposed; otherwise, the information would decay before it
could be recognized.
Treisman’s Attenuation Model
Shadowing: requires people to repeat the attended message out loud.
The initial findings of shadowing supported the filter model (ppl
almost complete unaware of the content of the msg played to
o However, later research showed they could report some info
on unattended channel on occasion (even their own names!)
Finding: The contextual effects of language can cause ppl to report words
on unattended channel and therefore shadow inappropriately. The two examples:
1) …I SAW THE GIRL / song was WISHING….
Me that bird/ JUMPING in the street…
2) …SITTING AT A MAHOGANY/ three POSSIBILITIES
… let us look at these/ TABLE with her head.
First line = msg asked to attend to
Second line = unattended msg
Words in Caps are the words actually spoken by the subjects.
Finding: The intrusions from the unattended channel fit the semantic context
better than the words on the attended channel.
The contextual cues were not sufficient to cause subjects to change
permanently to the unattended message in order to follow the
meaning of the passage, BUT the results did raise some questions
for the filter theory.
o E.g. if the filter blocks unattended stuff, how could they recall
stuff like their names on unattended channel?
Due to the inconsistencies of filter theory to account for this, Treisman came
up with a model that has 2 parts: a selective filter and a “dictionary”.
The filter in Treisman’s model does not completely block out the unattended
msg as filter theory does. Instead, the filter attenuates the info, making it
less likely to be heard.
Recognizing a word occurs in the dictionary if the intensity or loudness of the
word exceeds its threshold (the min. intensity needed for recognition).
Two important things about thresholds:
1) They vary across words (some have lower like ur name, or higher like
Thresholds relate to Treisman’s model cuz words on the unattended channel
were sometimes incorrectly shadowed if they better fit the context of the
msg on the attended channel. The Deutsch-Norman Memory Selection Model
Recall: the models proposed by Broadbent (filter theory) and Treisman
(attenuation theory) propose that the bottleneck occurs at the pattern
However, the Deutsch bros. and Norman propose that the
bottleneck occurs after pattern recognition.
Therefore the problem is not one of perception, but one of
SELECTION into memory after perception occurs.
o Because selection occurs later, these models are often
referred to as late-selection models.
Late-selection model: Proposal that the bottleneck occurs when
information is selected for memory.
The model assumes that words in both conversations (in a
shadowing experiment) are recognized but are quickly forgotten
UNLESS they are important.
Words on the attended channel are important cuz ppl have to
shadow them. Words on the unattended channel are usually
unimportant cuz listener is asked to attend to other channel.
Although recognized, they are quickly forgotten unless they are
important – a persons own name for e.g.
Broadbent’s model: 2 most important stages: filter and sensory store
Attention rep’d by the filter that determines what info is recognized.
An unattended msg can be recognized only if attention switches to
that msg before it decays from the sensory store.
Treisman’s model: 2 most important stages: filter and pattern
The filter attenuates the unattended msg, with the idea that very
few words are recognized on the unattended channel, unless their
thresholds for pattern recog. Are LOW enough to be exceeded by
(go beyond) the attenuated msg.
Deutsch-Norman model: 2 most important stages: pattern recognition
and selection. Both msgs are recognized, but only words selected into MEMORY
can be later recalled.
Treisman said that ppl should do equally well in tapping to target words on
the attended and unattended ears if the bottleneck occurred at the
She proposed that ppl should do much better in tapping to the target words
in the attended ear if the bottleneck occurred at the PERCEPTION STAGE.
The participants detected the target words 87% of the time that it occurred
in the attended ear and only 8% of the time that it occurred in the
unattended ear, supporting the hypothesis that the bottleneck occurred at
the PERCEPTION STAGE.
However Deutsch’s didn’t accept these results as evidence against
their theory. They said that the shadowed words on the attended
msg are more important cuz they are shadowed, and this added
importance increased the probability that they would elicit the
It is harder to select information based on meaning as opposed to pitch or
location; more mental effort (capacity) is required for late selection after
pattern recognition than for early selection before pattern recognition.
Example of a Capacity Model
Capacity theories are concerned with amt of mental effort required to
perform a task.
Kahneman’s “Attention and Effort” (1973) helped to shift the
emphasis from bottleneck theories to capacity theories.
o Kahneman: a capacity theory assumes there is a general limit
on a person’s capacity to perform mental work. His model
was designed to supplement, not replace, the bottleneck
models. A bottleneck theory says interference occurs cuz the same mechanism,
such as speech recognition, is required to carry out two incompatible
operations at the same time.
A capacity model proposes that interference occurs when the demands of 2
activities exceed available capacity.
Allocation of capacity: when a limited amount of capacity is distributed to
When the supply of attention does not meet the demands, the level
of performance declines.
K’s model says that the amt of capacity available depends on the level of
arousal: more capacity is available when arousal is moderately high than
when it is low!
BUT, very high levels of arousal can interfere with performance.
Yerkes and Dodson’s 1908 law: performance is best at intermediate
levels of arousal.
The level of arousal can be controlled by feedback (evaluation) from the
attempt to meet the demands of ongoing activities, as long as the total
demands don’t exceed the capacity limit.
Choice of which activities to support is influenced by both enduring
dispositions and momentary intentions.
o Enduring dispositions: reflect the rules of involuntary
o Momentary intentions: reflect our specific go