View
214
Download
0
Tags:
Embed Size (px)
Citation preview
International Commun tion Association
Johan F. Hoorn
Vrije UniversiteitFaculty of Sciences, Department of Computer Science
Section Information Management & Software EngineeringSubsection Human Computer Interaction, Multimedia & Culture
May 25, 2003San Diego, CA
Personification:Metaphor and Fictional Character
in CMC
www.cs.vu.nl/~jfhoorn
Personification,what is it?
Theory
Personification
Pierre Mignard (1694). Time Clipping Cupid’s Wings.
Fictional character(Time, Cupid)
used as a metaphor(Time is a man, Love is a boy)
for an abstraction(Time, Love)
Personification
Fictional character(Robby)
used as a metaphor(Human is machine)
for an abstraction(Help, Search, Navigate)
Bill Gates (1997). Robby the Robot.
Software agents can bepersonifications
No Personification
Fictional character(Builder)
used literally(Builder is a tutor)
for an abstraction(Help, Instruct, Create)
RealTimeAide (2003). Building tutor.http://www.realtimeaide.com/tutor/tutor.htm
For this agent,the metaphoric aspect
is missing
What’s the useof personification
in CMC?
Research question
Ease ofunderstanding Fun
Task relevanceUser support“Look and feel”Etc.
User effort Motivation
Literal icon/dialog
Metaphoric icon/dialog
Mediated person/Fictional character (FC)
Personification(FC plus metaphor)
Should we apply personifications?
Should we apply personifications?
User effort Motivation
Literal icon/dialog + (easy) - (no fun)
Metaphoric icon/dialog - (difficult) + (surprising)
Mediated person/ - (build a ++ (involve-
Fictional character (FC) relationship) ment)
Personification - - +++(FC plus metaphor) Personification is
more effort for more motivation?
Agents, what dothey communicate?
Theory
Agent-Mediated Communication
Sender Message Receiver
System’sstakeholder
(e.g., client,designer,manager)
Fictional character
End-user
Goals:- instruct- persuade- entertain
Goals:- be instructed- be persuaded- be entertained
+ metaphor
Match?
Agent-Mediated Communication
Sender Message Receiver’s perspective
Fictional character
End-user
+ metaphor
Humanprocessing
Goals:- instruct- persuade- entertain
System’sstakeholder
(e.g., client,designer,manager)
PEFiC
Metaphor process
Supportuser
goals?
noyes
Useagent
Don’tuse
agent
Goals:- instruct- persuade- entertain
Goals:- be instructed- be persuaded- be entertained
Match?
Receiver
System’sstakeholder
(e.g., client,designer,manager)
End-user
yes
no
Maintain agent
Alter agent
Message
Supportother goal?
no
yes
Agent-Mediated Communication
Sender’s perspective
http://www.csc.ncsu.edu/eos/users/l/lester/www/images/IPA/cosmo_ok.gif
Agent-Mediated Communication
Sender Message Receiver’s perspective
End-user
Humanprocessing
PEFiC
Perceiving and ExperiencingFictional Characters
For empirical evidence, see and hear:
Characters,how are they processed?
Results of other studies
Norm
Epistemics
Aesthetics
Ethicsgood
beautiful
realistic
bad
ugly
unrealistic
Involvement
Distance
Appreciation
dissimilar
irrelevant
negative valence
similar
relevant
positive valence
%
%
ENCODE COMPARE RESPONDFe
atur
es o
f situ
atio
nan
d Fi
ctio
nal C
hara
cter
Identification,empathy, sympathy,warm feelings, approach, etc.
Detachment,antipathy, cold feelings, avoidance,etc.
Appraisal domains
Mediators Fuzzyfeature sets
Subjective norm vs.group norm
PEFiC model
http://www.scpcug.com/wmwand12.html
Task-irrelevant features(goal ‘instruction’)
Relevant featuresif goal is ‘entertainment’
PeedyInvolvement
Distance
Example of PEFiC in action for factor Relevance to user goals
What is the roleof epistemics?
From character to metaphor
Agent-Mediated Communication
Message Receiver’s perspective
End-user
Humanprocessing
RMP
Race model ofMetaphor Processing
For empirical evidence, see:
Part of Epistemics
descriptive figurative descriptive figurative
realistic
descriptive figurative descriptive figurative
literal metaphor
unrealistic
ASSOCIATION COMMUNICATION FORM EPISTEMICS
drooling feet constrained
suit
‘tutor is a human’
‘product presenteris a dog’’
‘human is a machine’
‘conversation partner is a human’
Metaphor is part of Epistemics
http://www.ics.uci.edu/~kobsa/courses/ICS104/course-notes/metaphors.ht; http://www.techfak.uni-bielefeld.de/ags/wbski/lehre/digiSA/Methoden_der_KI/WS0102/methki15.pdf
literal metaphor
drooling (too enthusiastic)(saliva)
Metaphors,how are they processed?
Results of other studies
Categorymatch?
Sufficient descriptive ANDdescriptive/figurative
intersection?
Sufficientdescriptive/figurative
intersection?
Calculatedescriptive/figurative
intersection
Activatedescriptive and
figurativefeatures
Activatedescriptive and
figurativefeatures
‘Anomaly’
‘Metaphor’‘Literal’
no
no yes
Calculatedescriptiveintersection
Racemodel ofMetaphor
Processing
human machine
yes no
EEG: N400 atfrontal cortex feet constrained
Cosmo
How come metaphorsare harder to get but
do not take more time?
Discussion
Errors are the answer
Sufficientdescriptive/figurative
intersection?‘Metaphor’‘Literal’
no yes
Response times for literal and metaphor are about equal.No way telling whether these two information sourcesare serial or parallel
Problem:
Calculatedescriptive/figurative
intersection
Calculatedescriptiveintersection
If serial (1 before 2), applying metaphor is more time consumingand probably, more difficult to understand
If parallel, metaphor can be applied without losing time-efficiencyand trouble of understanding
(1) (2)
Investigate Lateralized Readiness Potential (LRP) in response topartial error pattern (after Coles et al., 1995)
Solution:
Calculatedescriptive/figurative
intersection
Calculatedescriptiveintersection
(1) (2)
‘Literal’ ‘Metaphor’
Few errorsfor ‘Metaphor’
Many errors for ‘Literal’ invisible in behavioral measures(e.g., RT) because they arecorrected before response execution visible in EEG
Thus, speed isnot the difficulty
in metaphorbut
accuracy is
For full argumentation, see:
‘Metaphor’ ‘Literal’
Partial error ‘Literal’
LRP low
LRP highCorrect ‘Metaphor’
Predictions for contralateral effects of finger movementduring metaphor processing
(fictitious data)
stimulus
responsebuttons
motor cortex
stimulusonset
stimulusonset
Shall we apply personifications, then?
User effort Motivation
Literal icon/dialog + (easy) - (no fun)
Metaphoric icon/dialog - (difficult) + (surprising)
Mediated person/ - (do I like the ++ (personal
Fictional character (FC) character?) -ized)
Personification - - +++(FC plus metaphor)
PEFiC RMP
Appreciation (Fun)Task relevanceValence (User support)Aesthetics (“Look and feel”)Ethics (Good bot vs. bad bot)Epistemics (Graphic rendering)Similarity (cf. Avatars)Involvement-distance
N400 (surprise)Two information sources:- descriptive- descriptive/figurativeTime efficiency
Category mismatchError prone (LRP)
high
high
Personification ismore effort for more motivation
Future work
We developed a software package for testing existing and newly created agents:Stimulus and trial production, RTs, and in the future, questionnaires and EEG extensions.
Downloads: http://www.antbed.tk/
What is it?
What can youdo with it?
Create environments in PowerPointand let the agent do its actions
Actionpreview
Personification:Metaphor and Fictional Character
in CMCTHE END
Wanna know more? Visit www.cs.vu.nl/~jfhoorn