Evidence from change-blindness experiments · Change Blindness is a measure of attention We notice...
44
Conspicuity of motorcycles in traffic: Evidence from change-blindness experiments 1
Transcript of Evidence from change-blindness experiments · Change Blindness is a measure of attention We notice...
Conspicuity of motorcycles in traffic:
Evidence from change-blindness experiments
1
2
3
Introduction
1981*
*Hurt, Ouellet, & Thom (1981)4
35%
65%
Motorcycle Collisions: Multiple Vehicles
Other
Right-of-WayViolation
26%
74%
Motorcycle Collision Types
Single VehicleCollision
Multiple VehicleCollision
Typical Right-of-Way Violation
5
Introduction
Common belief: Collisions are due to conspicuityMotorcycles are difficult to detect because they are small
6
Countermeasures:Daytime running lights
Introduction
7
ThenNow
Introduction
Countermeasures:Headlight modulators
8
ThenNow
Introduction
Countermeasures:Fluorescent jackets
9
ThenNow
Does enhanced conspicuity reduce collisions?
10
Does enhanced conspicuity reduce collisions?
*ACEM (2009)
11
35%
65%
1981
28%
72%
2009*
Other
Failure to Seethe Motorcycle
20%
80%
2009*
SingleVehicleCollision
MultipleVehicleCollision
26%
74%
1981
Motorcycle Collision TypesMotorcycle Collisions: Multiple Vehicles
Introduction
If motorcycle collisions are due to poor conspicuity,
And if we have improved motorcycle conspicuity,
Then why have failure-to-see collisions increased?
Are these collisions really due to poor conspicuity?
Are motorcycles even inconspicuous?
12
Change-Blindness
Are motorcycles less conspicuous than cars?
Demo
13
Are motorcycles less conspicuous than cars?
Change Blindness is a measure of attention
We notice changes for attended objects
Change-Blindness
16
Static Change-Blindness
17
Static Change-Blindness
1895% Confidence Interval
Are motorcycles detected less frequently than cars?
Sager et al. (In Review)
Static Change-Blindness
Detection times
1995% Confidence Interval
1
3
5
7
9
11
Car Motorcycle
Re
acti
on
Tim
e (
s)
Static Change-Blindness: Discussion
Motorcycles are visible:
Higher detection rates than cars
Similar detection times to cars
20
Contrast
Saliency (Sensory Conspicuity)
21
Orientation
Saliency (sensory conspicuity)
22
Spatial Frequency
Saliency (sensory conspicuity)
23
Colour
Saliency (sensory conspicuity)
24
Static Change-Blindness
Saliency maps and Gaze Maps
Perception is More than Sensation
Context
Intention
Memory
26
Static Change-Blindness
Car and Motorcycle Gaze Maps
Static Change-Blindness: Discussion
Motorcycles are not invisible:Higher detection ratesSimilar detection timesSimilar gaze maps
Saliency maps do not predict gaze maps
Solving motorcycle collisions through conspicuityIs solving a problem that does not existAnd it is solving it the wrong way
28
Dynamic Change-Blindness
But these images were static
What happens when people actually drive?
29
Dynamic Change-Blindness
30
Dynamic Change-Blindness 1
Ss Drove down a straight road
Screens flickered onceA parked vehicle was removed
Ss indicate change detection
31
Dynamic Change-Blindness 1
32
0
10
20
30
40
50
60
70
80
90
100
Motorcycle Car
Pe
rcen
t d
ete
ctio
nDetection Rate
t58 = 2.03p = .047
Motorcycles are detected more frequently than cars
95% Confidence Interval
Ss Drove toward intersection
Screens flickered onceA vehicle was removed on
half the trials(entering or exiting)(car or motorcycle)
Ss indicate change detection
Dynamic Change-Blindness 2
SDT: Was ist das?
34
Target present Target absent
Target reported
Hit False Alarm
Target not reported
MissCorrect
Rejection
Dynamic Change-Blindness 2
35
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
Motorcycle Car
d'
Sensitivity
0
1
2
3
4
5
6
7
8
Motorcycle Car
β
Bias
t40 = 3.48p = .001
t40 = .24p = .81
95% Confidence Interval
Dynamic Change-Blindness: Discussion
Results replicate findings from static change blindness experiments
Motorcycles are not invisible:Higher sensitivity for motorcycles than for cars
36
Conclusion
Motorcycles are (very) visible.
Why?
Sensory conspicuity is not the issue.
Efforts should be directed at education
Because the problem is likely a judgement issue
37
Acknowledgments
Daniel M. BernsteinFarhad D. DasturDavid J. Froc
RAsDawn-Leah McDonaldJohn Dema-alaKevin SmithCarley WoodAndrew LoweryAmaris TokJackie KingAaron Richardson
38
Questions?
39
Additional Slides
Braking Behaviour
41
Additional Results
42
0
2
4
6
8
Car Motorcycle
Me
an F
ixat
ion
Co
un
t
Fixation Count
95% Confidence Interval
Additional Results
43
0
1
1
2
2
3
Car Motorcycle
Tim
e t
o F
irst
Fix
atio
n (
s)
Time to First Fixation
95% Confidence Interval
Additional Results
44
0
200
400
600
800
1,000
1,200
Car MotorcycleMe
an T
ime
on
Tar
get
(ms)
Time On Target
95% Confidence Interval
