Comparability and Evaluation
Clarifying Criteria
| Authors: |
Stefan Rank & Paolo Petta
Austrian Research Institute for Artificial Intelligence |
|---|
| Date: |
2006-11-07 |
|---|
Presented at Humaine WP10WS • 2006-11-07. (Grey boxes like this one are notes or handout content.)
try to open the presentation in a new unadorned window,
then switch to fullscreen (F11 normally).
If the slide text doesn't fit in your browser window,
try decreasing the text size, try <Ctrl>+<->.
Use <space bar> to advance, <Page Up/Down> & <arrow keys> to navigate.
<T> (or clicking the empty-set-sign in the top or bottom right corner)
toggles between slideshow and outline.
Type a number followed by Return to switch to a particular slide.
Slide controls are hidden in the bottom right corner.
Hello my name is Stefan Rank. The title of my talk is ...
This presentation tries to clarify criteria for
designing and evaluating affective systems, first ask:
what are these systems, what do they have in common.
What Do We Have In Common?
Models of human emotion provide essential insight for the design and control of machines interacting with humans
- Emotional interaction to enhance usability
- Models of emotion to be used in intelligent agents
- BUT: widely different scenarios of use
- Motivation
- Purpose
- Deployment
Emotion is supposed to help for example in allocating and focusing mental resources when goals compete, or when the environment changes. The social or the physical environment.
Scenarios for Comparabilty
Detailed scenarios of use
- Help to compare systems targetting different functionalities
- Explicate the functional role of emotion
Emotion: What are situations / phenomena that we target?
- Psychologist: e.g. experimental data on human behaviour
- AI Researcher: e.g. multi-step decision making
- Engineer: e.g. believable real-time interaction
- Sociologist: e.g. data on strategies in social interaction
Basis for asking: What are the functionalities wanted?
Necessary to provide detailed scenarios of use. <s>
The scenarios of use contain the requirements and motivations for a specific agent or computational model and therefore allow to compare them.
The scenarios make explicit what kind of emotional functionality is wanted in an agent.
Scenario-based Comparisons
Range of emotional phenomena
Modelling fear vs. modelling fear, anger, and guilt
Interaction with humans
None vs. simulated vs. (restricted) dialogue
Interaction between agent and environment
Discrete simulation with infallible action vs. situatedness
Tasks and performance measures
Agents exploring an environment: efficiency vs. realism
Reportable Emotion Experience
range of emotional phenomena: critical property of the scenario.
What is a Scenario then?
A point in the niche space for affective agents
Possible purpose and environment of use
- Motivation and purpose
- Details of possible deployment
- Number and types of agents
- Interaction qualities (including user interface)
- Agent tasks
- Environment properties
- Possible (emotional) interactions, scripts
an idea of what a scenario consists of.
most importantly: what are the emotional episodes that can take place.
Scenario Descriptions
- Architectures have widely different
• Motivations • Purposes • Deployments
- We need to capture aspects relevant for comparison:
Emotional potential
- Computational psychology
| Motivation: | Test psychological theories |
|---|
| Purpose: | Create virtual subjects for experiments |
|---|
- Hormones in robots
| Motivation: | Stable, predictable behaviour for robots |
|---|
| Purpose: | Create robots for lab tests |
|---|
- Interactive narrative
| Motivation: | More engaging/varied stories in games |
|---|
| Purpose: | Games for children playing online |
|---|
examples for differing scenarios
Scenarios for Comparison
Motivation and purpose
Details of deployment
- User interface
- World
- Agents
- Agent-World Interaction:
- Positive and negative scripts
- Environment seen from the agents' viewpoint
- Ambiguity
- Social and cultural aspects
- Agent-Agent Interaction
Single agent → social emotions (shame, …) impossible
Conflicting tasks → prioritisation of control
World: (real/virtual/augmented reality, dynamics, regularities, discrete/continuous in time and space, time-stepped or continuous execution); Agents: number and types (including humans), their tasks; Environment seen from the agents viewpoint: sensing/actuating, sensorimotor coupling, tool use, time-stepped/continuous, asynchronous change; Agent-Agent Interaction: separate or world mediated, repertoire, unconditional channels, human as (another) agent
The End
Thank you for your attention!
Questions?
The following slides contain anticipated questions (AQ).
Disclaimer and Acknowledgments
- These notes reflect only the authors' views.
The European Community is not liable for any use that may be made of the information contained herein
- This work was funded by the EU FP6 Network of Excellence Humaine [IST-2002-2.3.1.6 507422]
- OFAI is supported by the Austrian Federal Ministry for Education, Science and Culture
and by the Austrian Federal Ministry for Transport, Innovation and Technology
Keyboard Controls
The following apply in any supporting browser besides Opera, which
uses the default Opera Show controls instead.
| Action |
Key(s) |
|---|
| Go to the next slide |
- [Space bar]
- [Return]
- [Enter]
- [Right arrow]
- [Down arrow]
- [Page down]
- Click the left mouse button outside of the control area,
Flash object, or movie
|
| Go to the previous slide |
- [Left arrow]
- [Up arrow]
- [Page up]
|
| Go to the title (first) slide |
[Home] |
| Go to the last slide |
[End] |
| Jump directly to a slide |
Type the slide number, then hit [Return] or [Enter] |
| Skip forward n slides |
Type the number of slides to skip, then hit any "go to next"
key (except [Return] or [Enter]) |
| Skip backward n slides |
Type the number of slides to skip, then hit any "go to
previous" key |
| Switch between slideshow and outline view |
|
| Show/hide slide controls |
- [C]
- Move the mouse pointer over the control area
|
Further details of the S5 user interface can be found at Eric
Meyer's S5 page and the s5project site .
References
| [CleeremansFrench1996] | Cleeremans A., French R.M.: From Chicken Squawking To Cognition: Levels of
Description and the Computational Approach in Psychology, Psychologica
Belgica, 36(1-2):5-29, 1996. |
