Upload
navaneethans
View
86
Download
0
Tags:
Embed Size (px)
Citation preview
How Do I Get There?Navigating the Map in your Head
Navaneethan Santhanam
Johns Hopkins University
12th May 2014
Giving Directions
• Think of a time you gave someone directions– What did you include?
– What did you leave out?
– Why?
• Important for others to understand your directions– How do we know what
they might use?
Google Maps in the US
• Focus on
– Street names
– Directions
– Distances
• Why?
– Clear street signs
– Standardized
– Vernacular names = Official names
Google Maps in India
• Focus on
– Street names
– Distances
– LANDMARKS
• Why?
– Street signs hard to find
– Signs not standardized
– Vernacular names ≠ Official names
Individual Cue Preferences vary
Some people prefer using orientation
Others prefer using landmarks
“Head North and then East” “Turn right after the Transamerica Building”
Cognitive Maps & Navigation
• We all have different ‘cognitive maps’
– Internal map of the world around us
– Shaped by individual preferences
– Contains info about aspects of environment
• Google Maps in India
– Understood Indian users’ needs/preferences
– Provided info in straightforward manner
‘Our environments give us many options about what to use while navigating. With so many cues to choose from, which ones do we pick, and how do we use them?’
Roadmap
• Different Types of Cues
– What are Orientation Cues?
– Orientation Cues vs. Local Landmarks
– Why this distinction?
• How might Orientation Cues be useful?
– Intuition, Hypotheses & Predictions
– Experimental Findings
• What next?
Navigational Information Sources
Orientation Cues
Cues that provide info about orientation but not position
– Not informative about position within environment
Examples – constellations, mountains, city skyline
Operational Definition – does the cue provide position information or not?
Navigational Information Sources
Cues that provide position (and orientation) information
– Appearance depends on both position & heading
Examples – stores, street signs
Local Landmarks
Operational Definition – does the cue provide position information or not?
A bit more about Orientation Cues
Orientation Cues in Rodent Studies
• How have they been used?
– Rodents run in mazes with little or no cues inside
– Distinguishing extra-maze cues on walls of room
Orientation Cues in Rodent Studies
Long history in rodent maze studies
– Rats extremely sensitive to orientation cues
• Removing/rotating cues affected behavior
– Shown in various types of mazes
• T-Maze, Radial Arm Maze, Water Maze
Orientation Cues in Human Navigation?
• Do the rodent results translate to humans?
– Mazes set up to favor orientation cues
• Few (if any) landmarks
– If local landmark are present, rats do use them
Orientation Cues in Human Navigation?
• Do the rodent results translate to humans?
– Human environments have many local cues
– May not need orientation cues to navigate
Role of Orientation Cues in Navigation
• Help establish & maintain orientation
– May be disoriented without these cues
– Unable to reach destinations without them
• Provide sense of global structure
– Allow navigators to take direct paths
– Removal may result in reliance on the previously-seen familiar paths
Solution Index
LEARNING: Video tour with multiple chances to learn environment
NAVIGATION: Participants navigate to targets in environment
Dual Solution Paradigm (DSP)
Marchette, Bakker, & Shelton, 2011
LEARNING NAVIGATION
Familiar Path
Novel Shortcut
Familiar Paths Shortcuts
Video of environment
Experimental Design
• Orientation Cues
– 4 different mountains surrounding environment
– At ‘visual infinity’: only orientation info
• Learning
– Guided tour of environment
• Navigation– Cues Present: 50% of trials with mountains
– Cues Absent: 50% of trials mountains removed
Prediction: Do we need orientation cues?
PREDICTION: Orientation cues required for success Why?
• Navigators use mountains to remain oriented while navigating
• Removing mountains disorients navigators
• Unable to reach target
0
10
20
30
40
50
60
70
80
90
100
CUES PRESENT CUES ABSENT
Results: Do we need orientation cues?
Prediction Result
0
10
20
30
40
50
60
70
80
90
100
CUES PRESENT CUES ABSENT
% O
F TA
RG
ETS
FOU
ND
0
10
20
30
40
50
60
70
80
90
100
CUES PRESENT CUES ABSENT
% O
F TA
RG
ETS
FOU
ND
Removing Orientation Cues does not affect navigational success
Prediction: Do orientation cues help us navigate more directly?
PREDICTION: Fewer shortcuts when mountains are removed Why?
• No mountains reduced global structure info
• Navigators use mountains to orient towards target
• Without mountains, navigators rely on past experience0
10
20
30
40
50
60
70
CUES PRESENT CUES ABSENT
% T
AR
GET
S FO
UN
D U
SIN
G S
HO
RTC
UTS
MoreFamiliar Paths
More Shortcuts
Results: Do orientation cues help us navigate more directly?
Prediction Results
0
10
20
30
40
50
60
CUES PRESENT CUES ABSENT
% T
AR
GET
S FO
UN
D U
SIN
G S
HO
RTC
UTS
MoreFamiliar Paths
More Shortcuts
0
10
20
30
40
50
60
CUES PRESENT CUES ABSENT
% T
AR
GET
S FO
UN
D U
SIN
G S
HO
RTC
UTS
Removing Orientation Cues does not affect shortcut usage
Interim Conclusions: Experiment 1
• Orientation Cues don’t see to matter
– Removing cues didn’t affect success
– Removing cues didn’t affect solution used
• Why might this be happening?
– Navigation may not depend on orientation cues
• Landmarks may compensate during navigation
– Orientation cues may serve different function
• Measures not sensitive to what they actually do
Establishing Orientation
• Use Orientation Cues before navigating
– Coming out of a subway and using skyline to orient
• Use Local Landmarks to guide navigation
– Use storefronts, billboards, street signs, etc to guide movements
New Hypothesis: Initial Orientation
• Help establish orientation
– Help navigators orient when initially in environment
• Does initial orientation differ between Present & Absent trials?
– Need task that focuses on initial orientation
– Task should also reduce influence of local landmarks
& maintain & maintain
‘Initial Orientation’ Task
• Use Dual Solution environment
– Measure of cue effects on establishing orientation
– Try to keep task as similar as possible
• Judge target orientation
– Participant indicates target direction from start location
– No translation allowed, only rotation
• Restrict ability to use landmarks
Task Structure
Orientation CuesSame as previous study
LearningGuided tour of environment
Task Structure
TestProcedure
– Participant dropped at start location & provided target
– Participant turns & orients to face goal location
Conditions
– Present condition: 50% of trials with mountains
– Absent condition: 50% of trials mountains removed
Angular Error (θerr)Definition – diff between participant’s final direction & actual direction from starting point
θerr
Participant’s pointing direction
Actual object direction
0
10
20
30
40
50
60
Cues Present Cues Absent
An
gula
r Er
ror
Predicted Angular Error
Results: Angular Error
Prediction Result
0
10
20
30
40
50
60
70
CUES PRESENT CUES ABSENT
AN
GU
LAR
ER
RO
R (
de
gre
es)
0
10
20
30
40
50
60
CUES PRESENT CUES ABSENT
AN
GU
LAR
ER
RO
R (
de
gre
es)
Having Orientation Cues does not improve accuracy
Interim Conclusions: Experiment 2
• Had no impact on judgments– No improvement in angular error on
Present vs. Absent trials
– Also looked at: rotation time, initiation time
• HOWEVER: Are Present & Absent trials treated the same way?– Don’t know if same strategy used for both
– May use different approach if orientation cue removal is noticed
Open Questions
• Shift to landmarks if cue removal noticed?
– Orientation Cues used when available
– If Orientation Cues absent, landmarks used
– Cue absence causes a strategy switch
• Do participants use available Orientation Cues?
– Keep cues throughout Navigation phase
–Rotate cues in 50% of trials
– Compare navigation for Original vs. Rotated
Experiment: Rotated Orientation Cues
Use Dual Solution navigation task
Orientation Cues Same as previous study
LearningGuided tour of environment (same as previous studies)
Navigation
Present condition: 50% of trials with mountains
Rotated condition: 50% of trials mountains rotated
Rotated Cues: Schematic
0°
Rotated Cues: Schematic
180°
Rotated Cues: Measures
• Effects in overall navigation performance
– Still look at Success Rate & Solution Index
– May not see effect in such ‘global’ measures
• Local Landmarks may take over nav
• Rotated cues could disorient at start
– Compare Time to Initiate Movement for Original & Rotated cue trials
Predictions: Success Rate
Prediction: Rotating cues reduces success rate Why?
• Navigators disoriented by rotation of cues
• May find it difficult to use cues to navigate successfully
• Unlikely since we haven’t seen such effects in previous experiment0
20
40
60
80
100
ORIGINAL CUES ROTATED CUES
Succ
ess
Rat
e
Preliminary Results: Success Rate
Prediction ResultsPRELIMINARY n = 11
0.0
20.0
40.0
60.0
80.0
100.0
ORIGINAL CUES ROTATED CUES
% o
f TA
RG
ETS
FOU
ND
0
10
20
30
40
50
60
70
80
90
100
ORIGINAL CUES ROTATED CUES
% o
f TA
RG
ETS
FOU
ND
Rotating Orientation Cues does not affect navigational success
Predictions: Solution Index
Predictions: Rotating cues reduces shortcut use Why?
• Rotating cues disorients navigators
• May make them rely on previously-seen paths
• Unlikely, since we don’t see this effect in previous study
MoreFamiliar
Paths
More Shortcuts
0
10
20
30
40
50
60
70
ORIGINAL CUES ROTATED CUES
% T
AR
GET
S FO
UN
D U
SIN
G S
HO
RTC
UTS
Results: Solution Index (Preliminary)
Prediction Results
MoreFamiliar Paths
More Shortcuts
PRELIMINARY n = 11
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
ORIGINAL CUES ROTATED CUES
% T
AR
GET
S FO
UN
D U
SIN
G S
HO
RTC
UTS
0
10
20
30
40
50
60
70
ORIGINAL CUES ROTATED CUES
% T
AR
GET
S FO
UN
D U
SIN
G S
HO
RTC
UTS
Rotating Orientation Cues does not lead to reduced shortcut usage
Prediction: Time to Initiate Movement
Rotating orientation cues increases time to start moving Why?
• Rotating cues may stump participants
• Not sure which way to go initially
• Most likely effect if cues help in establishing orientation0
0.5
1
1.5
2
2.5
3
3.5
4
ORIGINAL CUES ROTATED CUES
Tim
e t
o In
itia
te M
ove
me
nt
(s)
Prelim Results: Time to Initiate Movement
Prediction ResultsPRELIMINARY n = 11
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
ORIGINAL CUES ROTATED CUES
TIM
E TO
INIT
IATE
MO
VEM
ENT
(s)
0
0.5
1
1.5
2
2.5
3
3.5
4
ORIGINAL CUES ROTATED CUES
TIM
E TO
INIT
IATE
MO
VEM
ENT
(s)
Rotating Orientation Cues does not lead to longer ‘Time to Initiate’
Conclusions so far…
• What do we know about orientation cues?– Don’t seem to be critical for navigation/orientation
in these environments
– Typical human environments may not require such cues• Our environments are less ‘locally-rich’ than real world
• When (if ever) are they used?– If orientation cues perfectly predict navigational goal
– If local landmarks are uninformative/confusing• Need to test by reducing local landmarks during navigation
Future Directions
• Eliminate local landmarks
– What exactly is a local landmark?
– Lots of redundant sources of nav info
– What kind of info do landmarks give us?
• Remove alcoves (where objects are located)
– Might find it impossible to navigate
• Remove objects
– Might change representation of environment
Relevance
• Understanding how we navigate is important
– Understand individual preferences for cues
– Understanding how we use them
• How does this help us?
– Customize navigational experience for each person
– Highlight cues specific to each individual
– Important for tools such as Google Maps & Google Glass
Acknowledgements
Thank you to everyone in the Shelton Lab!
Dr. Amy Shelton
Dr. Amy Stephens
Ben Nelligan
Dr. Steven Marchette
Andrew Furman