College 1: Observations & Theory
A long history of m aking m achines after hum ans
Attem pts com e from a very rational approach to what a hum an is
M aking a robot after a hum an assum es we are not m uch m ore than
electronics and m echanics (18th century view)
Even adding intelligence (Turing) is a rational contribution
Yet, people interact with robots m ainly on a affective level
There are m any therapeutic advantages in applying social robots
People easily self-disclose to a robot without expecting to be judged by it,
that is strange:
W hat is a social robot, other than a sim ulation of hum an-like behavior by
com puter software, wrapped in electro-m echanical devices?
The sim ulated ‘agency’ does not even have to be fully autonom ous
A rem ote controlled robot will evoke the sam e results after a few m inutes
People tend to forget ‘the m an behind the curtains’ (W izard of Oz)
People focus on the robot as a com m unication partner on its own
People m ake plenty of attributions and project all kinds of qualities on to
robots, which are actually not there
Agency:
- Goals & Concerns
- Intentionally
- Artificial system
- Potentially capable of acting autonomously
Software agent:
Agency simulated by (semi) autonomous software system
Robot:
Software agent interacting through electro-mechanical devices
Social robot or android:
Robot specialized in humanoid situations
Care robot (or war robot, or tutor robot or...):
Social robot applied to care
,College 2: Observational studies
Interaction with a robot does not com e spontaneously
People are shy working with som ething they don’t know, or it sim ply does
not concern them (has no relevance to their life)
M ost people rely on others to know what they should do or think
M ake use of two types of people for the introduction of a robot: experts &
techno-forerunners that are m em bers of your target group
At first, m ake the robot ‘do som ething’ to draw attention
Does not have to be com plicated as long as it is not idle
Then, work on an application relevant to your audience
To stim ulate technology acceptance, the best way to get people experienced
with robots is to let them help m ake them (cf. Robopop)
Also prom ises the design of relevant applications
Creativity is encouraged by an open and diverse com m unity, unorganized
and focused on sharing, rather than defending intellectual property
The creative process roughly knows two stages:
- Open idea form ation
- M ore precise im plem entation
Before, during and after design, observation is prim e to understand
people’s unguided free interactions
Various form s to record observations:
- Field notes
- Audio/video
- Respondent journals
After sam pling, you m ake an observation coding schem e
Social robots encounter situations that have social, linguistic, as well as engineering
aspects. The field is young. There is no standard solution to any question. Hence,
multidisciplinary work is required.
Paro as cat was too familiar to be believable
People ignore robots (Paro), but for different reasons (Chang, Sabanovic & Huber, 2014)
Self-disclosure happens with a robot
CASA, but not entirely:
Paro is not judgmental, not going behind your back, a feeling of privacy
Moshkina, Trickett & Trafton, 2014:
Expect: The more human-like, the more social responses
,Experimental (observation) design:
Voice x Voice+Lips x Voice+Lips+Facial Expressions x Voice+Lips+Facial
Expressions+Gestures x Voice+Lips+Facial Expressions+Gestures+Gameplay
CASA prediction is to find an increasing trend (the more human-like, the more social
responses)
Result was parabolic growth: gameplay did not result in more engagement
CASA only goes so far: there is an optimum after which the value decreases
Possible reasons:
- If the robot confirms it social construct (for example functional) it is more
effective (functional robot is not playful)
- Novelty effect might wear out quickly
- People watch what others do before reacting themselves
- No relevance to the task at hand
In natural settings, people don’t come for your robot. Would random motion have done
the same?
Minimal social cues invoke more engagement than no social cues
More human-like behaviors lead to more social engagement
Gameplay did not add to the less social behaviors. CASA only goes so far (parabolic
growth).
There should be relevance to the task at hand
A robot is not a human and does not have to be, it occupies its own social niche
Some things machines do better than humans:
- People with dementia react to robot noise, but not to people.
- People self-disclose more to a robot.
- Robots do not deceive: what you see is what you get
Unified Theory of Accaptance and Use of Technology (UTAUT):
Experience is prime for acceptance, so we should start young
UTAUT identifies four key factors (performance expectancy, effort expectancy, social
influence, and facilitating conditions) and four moderators (age, gender, experience, and
voluntariness) related to predicting behavioral intention to use a technology and actual
technology use primarily in organizational contexts
Position of the robot is key. The robot needs to be the lesser
Order is less informational than chaos or mess
Creativity thrives on a mess
4 creative techniques:
1. Comparison to other domains
2. Provide question instead of solution
3. Demonstrate failing prototype, then go to step 2
4. Cooperation & sharing knowledge
, College 4: H um an relations with m achines
People love and hate different things, dolls and objects included
It is too easy to dism iss them as crazy, because 1) they are not delusional
and 2) what decides what m akes other people happy?
Everybody has the capacity to love anim ated objects, som e m ore than
others
M edia Equation predicts that people will autom atically respond naturally
and socially to m edia such as robots
W hat seem s real is m ore im portant than what is real
The M edia Equation holds that people react to what is there; they prefer
easy and sim ple interactions
That is why they rely on hum an ways of dealing with robots
From this, CASA evolved
CASA says that the m ore a com puter or robot com plies with hum an
(conversational) rules such as reciprocity, the higher the perform ance (e.g.
self-disclosure)
CASA predicts that people apply hum an conventions m indlessly to robots.
They do not think, but just act
This only goes so far!
There is evidence that people som etim es do think, perceive differences and
respond differently to robots than to hum ans
Affective Com puting attem pts the creation of an em otional responsive
robot/agent system
This brings artificial system s closer to the M edia Equation and CASA
Apart from sim ulating em otions, Affective Com puting whishes to recognize
em otions of users
This is a heroic attem pt, because in spite of all the physiological
m easurem ents, em otional experience cannot be read from em otional
expression (Kuleshov effect)
Self-report is hard too, particularly afterwards: em otions subside once you
reflect on them in a survey
M edia Equation & CASA observed m any phenom ena of people loving and
hating their m achines and suggested an evolutionary and neurological
explanation
Theory of Affective Bonding goes further in trying the integration of
experiential m odels of behavior, with faster and slower pathways of the
brain
A long history of m aking m achines after hum ans
Attem pts com e from a very rational approach to what a hum an is
M aking a robot after a hum an assum es we are not m uch m ore than
electronics and m echanics (18th century view)
Even adding intelligence (Turing) is a rational contribution
Yet, people interact with robots m ainly on a affective level
There are m any therapeutic advantages in applying social robots
People easily self-disclose to a robot without expecting to be judged by it,
that is strange:
W hat is a social robot, other than a sim ulation of hum an-like behavior by
com puter software, wrapped in electro-m echanical devices?
The sim ulated ‘agency’ does not even have to be fully autonom ous
A rem ote controlled robot will evoke the sam e results after a few m inutes
People tend to forget ‘the m an behind the curtains’ (W izard of Oz)
People focus on the robot as a com m unication partner on its own
People m ake plenty of attributions and project all kinds of qualities on to
robots, which are actually not there
Agency:
- Goals & Concerns
- Intentionally
- Artificial system
- Potentially capable of acting autonomously
Software agent:
Agency simulated by (semi) autonomous software system
Robot:
Software agent interacting through electro-mechanical devices
Social robot or android:
Robot specialized in humanoid situations
Care robot (or war robot, or tutor robot or...):
Social robot applied to care
,College 2: Observational studies
Interaction with a robot does not com e spontaneously
People are shy working with som ething they don’t know, or it sim ply does
not concern them (has no relevance to their life)
M ost people rely on others to know what they should do or think
M ake use of two types of people for the introduction of a robot: experts &
techno-forerunners that are m em bers of your target group
At first, m ake the robot ‘do som ething’ to draw attention
Does not have to be com plicated as long as it is not idle
Then, work on an application relevant to your audience
To stim ulate technology acceptance, the best way to get people experienced
with robots is to let them help m ake them (cf. Robopop)
Also prom ises the design of relevant applications
Creativity is encouraged by an open and diverse com m unity, unorganized
and focused on sharing, rather than defending intellectual property
The creative process roughly knows two stages:
- Open idea form ation
- M ore precise im plem entation
Before, during and after design, observation is prim e to understand
people’s unguided free interactions
Various form s to record observations:
- Field notes
- Audio/video
- Respondent journals
After sam pling, you m ake an observation coding schem e
Social robots encounter situations that have social, linguistic, as well as engineering
aspects. The field is young. There is no standard solution to any question. Hence,
multidisciplinary work is required.
Paro as cat was too familiar to be believable
People ignore robots (Paro), but for different reasons (Chang, Sabanovic & Huber, 2014)
Self-disclosure happens with a robot
CASA, but not entirely:
Paro is not judgmental, not going behind your back, a feeling of privacy
Moshkina, Trickett & Trafton, 2014:
Expect: The more human-like, the more social responses
,Experimental (observation) design:
Voice x Voice+Lips x Voice+Lips+Facial Expressions x Voice+Lips+Facial
Expressions+Gestures x Voice+Lips+Facial Expressions+Gestures+Gameplay
CASA prediction is to find an increasing trend (the more human-like, the more social
responses)
Result was parabolic growth: gameplay did not result in more engagement
CASA only goes so far: there is an optimum after which the value decreases
Possible reasons:
- If the robot confirms it social construct (for example functional) it is more
effective (functional robot is not playful)
- Novelty effect might wear out quickly
- People watch what others do before reacting themselves
- No relevance to the task at hand
In natural settings, people don’t come for your robot. Would random motion have done
the same?
Minimal social cues invoke more engagement than no social cues
More human-like behaviors lead to more social engagement
Gameplay did not add to the less social behaviors. CASA only goes so far (parabolic
growth).
There should be relevance to the task at hand
A robot is not a human and does not have to be, it occupies its own social niche
Some things machines do better than humans:
- People with dementia react to robot noise, but not to people.
- People self-disclose more to a robot.
- Robots do not deceive: what you see is what you get
Unified Theory of Accaptance and Use of Technology (UTAUT):
Experience is prime for acceptance, so we should start young
UTAUT identifies four key factors (performance expectancy, effort expectancy, social
influence, and facilitating conditions) and four moderators (age, gender, experience, and
voluntariness) related to predicting behavioral intention to use a technology and actual
technology use primarily in organizational contexts
Position of the robot is key. The robot needs to be the lesser
Order is less informational than chaos or mess
Creativity thrives on a mess
4 creative techniques:
1. Comparison to other domains
2. Provide question instead of solution
3. Demonstrate failing prototype, then go to step 2
4. Cooperation & sharing knowledge
, College 4: H um an relations with m achines
People love and hate different things, dolls and objects included
It is too easy to dism iss them as crazy, because 1) they are not delusional
and 2) what decides what m akes other people happy?
Everybody has the capacity to love anim ated objects, som e m ore than
others
M edia Equation predicts that people will autom atically respond naturally
and socially to m edia such as robots
W hat seem s real is m ore im portant than what is real
The M edia Equation holds that people react to what is there; they prefer
easy and sim ple interactions
That is why they rely on hum an ways of dealing with robots
From this, CASA evolved
CASA says that the m ore a com puter or robot com plies with hum an
(conversational) rules such as reciprocity, the higher the perform ance (e.g.
self-disclosure)
CASA predicts that people apply hum an conventions m indlessly to robots.
They do not think, but just act
This only goes so far!
There is evidence that people som etim es do think, perceive differences and
respond differently to robots than to hum ans
Affective Com puting attem pts the creation of an em otional responsive
robot/agent system
This brings artificial system s closer to the M edia Equation and CASA
Apart from sim ulating em otions, Affective Com puting whishes to recognize
em otions of users
This is a heroic attem pt, because in spite of all the physiological
m easurem ents, em otional experience cannot be read from em otional
expression (Kuleshov effect)
Self-report is hard too, particularly afterwards: em otions subside once you
reflect on them in a survey
M edia Equation & CASA observed m any phenom ena of people loving and
hating their m achines and suggested an evolutionary and neurological
explanation
Theory of Affective Bonding goes further in trying the integration of
experiential m odels of behavior, with faster and slower pathways of the
brain
