Experimental Psychology
PSY 433
Chapter 8
Attention and Reaction Time
Where’s Waldo?

http://www.youtube.com/watch?v=EvWh6PMi9Ek&feature=player_embedded#
Says Wikipedia: Werner Herzog is
often considered one of the greatest
figures of the New German Cinema.
His films often feature heroes with
impossible dreams, people with
unique talents in obscure fields, or
individuals who find themselves in
conflict with nature. French filmmaker
François Truffaut once called Herzog
"the most important film director alive”.
This is actually him narrating Where’s
Waldo.
Two Aspects of Attention
 Divided attention – what happens when we
try to engage in two cognitive processes at
once
 Selective attention – how we switch mental
resources from one cognitive task to another.
People don’t multitask well.
Visible Bottleneck Task
 http://opl.apa.org/contributions/Pashler/prp.html
 http://dualtask.org – visible bottleneck I and FAQ
 This task illustrates how difficult it is to pay
attention to two things at the same time.


Both tasks require a choice of response and
the same cognitive resource cannot be
devoted to both tasks at the same time.
The competition goes away when one task
does not involve a choice (e.g., press any
button when you hear a tone).
Donders A, B & C Revisited
 A is simple RT – see a stimulus and press a
key
 B is stimulus choice RT – see one of two
stimuli and decide whether to press a key or
not
 C is stimulus and response choice RT – see
one of two stimuli and decide which of two
keys to press
 B-C gives response selection time
 C-A gives ID time.
Donders Tasks
 S1  R1
Donders A
 S1  R1
Donders B
S2  R2
 S1  R1
S2
Donders C
Donders A -- Simple
A Reaction Time
C Reaction Time
C Minus A
Baseline
Identification
Time
Selection Time
Donders B -- Choice
B Reaction
Time
C Reaction
Time
B Minus C
Testing for Modularity
 Donders A, B & C implies that the parts of the
task are modules.
 How can component modules be identified?

When one component module can be
changed without changing the others, it is
independent.
 If Donders was correct then the three parts
should be separately modifiable.

Pure insertion – addition of a module without
affecting the duration of the other modules.
Response Force
 Donders pure insertion could not be tested so
an additional variable was added – response
force.
 Response force – the amount of pressure
exerted on a response key.

Force increases with stimulus intensity.
 For Ulrich et al. (1999), response force was
the same for a Donders A and B comparison,
even though RTs were different.
RT and Integrated Force (Fig 8.3)
This result is consistent with pure insertion.
B and C Reactions (Fig 8.4)
The B and C RT’s should differ but they do not.
A and C Reactions (Fig 8.5)
Now RT’s differ as they should but force is not consistent with pure insertion.
Confounding Stimulus Intensity
 Confound – when one or more independent
variable is simultaneously varied so we can’t
tell which is responsible for an effect.
 In Ulrich et al.’s experiment (green LED to left
or right), two things were varied:


Mapping of stimulus to response (one hand vs
two)
Apparent brightness (focusing on a single light
instead of both lights)
 Replaced by letters X and S not lights.
The Same Experiment using
Letters not LED Lights (Fig 8.6)
Now RT differs as it should and Force supports pure insertion.
Results using Letters
 When X vs S is used, the confound of
stimulus intensity is eliminated (controlled)
and the results support pure insertion.
 Part of the problem is that all three Donders
experiments were not presented:

Authors wished to avoid transfer effects that
occur in within-subject experiments.
 To avoid this, present all three Donders
conditions or do the experiment betweensubject.
Speed Accuracy Tradeoffs
 RT cannot be used as the only dependent
variable because subjects change accuracy
to maximize speed:


Speed & accuracy are sometimes inversely
related.
RT & Accuracy must be jointly considered.
 Examining more than one dependent variable
may be crucial to understanding the
processes involved in a task.
Theios (1973)
 Subjects had to name a digit presented
visually.
 Probability of the digit was varied from 0.2 to
0.8.
 Reaction time was unaffected by probability
of the digit, but accuracy was greatly affected.

Highest error rate with lowest probability.
 To increase accuracy (keep error rate
constant), RTs would have to increase.
RT and Error Rate as a Function of
Stimulus Probability
RT stays constant but
as stimulus probability
increases, errors
decrease.
Dual Task Processing
Speed-Accuracy Tradeoff
Psychological
refractory
period here
SOA (S1-S2 Interval in ms)
A Central Bottleneck
 We can only process one thing at a time
within a single modality (vision, hearing).
 Central cognition may be the most important
bottleneck – the central bottleneck.
 Whether two tasks can be done at once
depends on whether they compete for the
same resources.

Schumacher dual-task experiment.
Stimulus Onset Asynchrony (SOA)
 Pashler presented a modified Donders B task
in which S1 and S2 were not presented
simultaneously.


The interval between them is called stimulus
onset asynchrony (SOA)
The shorter the SOA, the greater the effect on
RT and errors.
 The period where one task interferes with the
other is called psychological refractory period.
Dual Task Processing
Pashler’s Paradigm
 Task 1 – hear a tone and press a key with the
left hand.
 Task 2 – vocally call out the name of the
highest digit in a display of eight digits.
 When subjects are not required to respond
quickly to Task 2, accuracy is not affected.

It only occurs with a requirement to make a
speedy response.
Capacity Sharing Explanation
Response selection
S1
Response selection
S2
The resource is shared during the
time when the tasks overlap.
Explanations
 Central bottleneck models – some common
internal processing stage is required by both
tasks, creating a bottleneck for resources.
 Central capacity sharing models – a resource
called capacity must be split across the two
tasks, reducing the capacity available to
either task and reducing efficiency.
 Both models predict the same observed
results – the author prefers capacity models.
Stress and Cognitive Control
 Two possible effects of stress on cognition:


Stress requires attention (capacity) thus
decreasing performance.
People adapt to stress by finding more
efficient ways of doing tasks increasing
performance (strategies change).
 Steinhauser’s experiment:


Long/short interval cues to respond to digit or
letter – 6M
Is letter a consonant/vowel, is digit odd/even?
Results
 For the low stress condition there was an
interaction between stimulus interval and task
repetition (same task next or different next).

No interaction in the high stress condition.
 Under low stress there was a relatively higher
cost to changing tasks quickly.
 Under high stress the cost was the same.
 This is consistent with the idea that under
stress cognitive strategies change.
Change Blindness Demo
 http://viscog.beckman.illinois.edu/flashmovie/23.php