Intelligent systems research seminar (58307111) Spring 2007
Instructor: Tei Laine, PhD
(tei.laine at cs dot helsinki dot fi).
Office hours Mon, Wed 12-12:30 A213.
Time and location: Periods III and IV, Thu 10-12 B119
Artificial intelligence (AI) has been inspired by what many consider the highest level of intelligence, namely the human intelligence. Research in AI has been driven by an attempt to reproduce human thinking, reasoning and language capabilities. The actual achievements have been varying, best characterized as many small instances of progress, and a few major setbacks.
Perhaps the whole approach of antropomorphizing computers is not correct. Still relatively little is known about the mechanisms and processes behind human cognition, so it may turn out futile to base machine intelligence on principles not perfectly well understood.
Animals are not just four-legged furry humans with smaller brains but they represent a whole range of intelligent behaviors in their natural environments. For instance, it is not known why some species cheat and others do not (cheating has been traditionally considered very human trait since it requires high level cognitive capabilities, such as intentionality, understanding of consequences etc.). In order to understand and properly implement artificial intelligence it may be beneficial to study simpler mechanisms that are necessary or sufficient to these advanced behaviors.
But a whole lot more can be done to understand animal intelligence than just to observe individuals in their own habitat. Human cognitive capabilities can be either directly measured (memory span) or assessed indirectly from their performance (rationality) in the task that requires reasoning, problem solving or language processing. Unfortunately, with animals we do not have this luxury, since they do not reason or use language unless we teach them to. Therefore, learning plays a crucial role in the study of animal intelligence.
Unlike human beings, who have practically no limits (individual differences notwithstanding) in what they can learn, different animal species vary dramatically in their ability to learn specific tasks. This raises the obvious question: what is the difference between cognitive systems of seemingly equal species that makes it relatively easy for one of them to learn a task, and quite impossible for another.
The goal of this seminar is to study a spectrum of intelligent behavior exhibited by different animal species and relate these forms of behavior to intelligence implemented in artificial systems in order to gain understanding in the basis of intelligence and learning.
1. Meetings
The seminar meets every week to discuss various topics on animal and artificial intelligence. For each class each participant writes and submits a short essay discussing relevant issues regarding the topic; the essay is not supposed to be a review of the readings but rather a critical commentary on them, or alternatively discussion of questions raised by the readings.At each meeting one of the participants is nominated as a chair: the chair's task is to introduce the topic and lead the discussion.
In the end of the semester each participant turns in a term paper on the topic of one's own choice. Preliminary outlines of the term paper will be submitted by the end of the third period.
Here is a tentative list of paper topics.
The seminar language is English. However, Finnish natives may choose to submit the essays and the term paper in Finnish.
2. Prerequisites
Open and curious mind, and interest in artificial as well as natural intelligence are the only requirements. Also some basic understanding of cognitive systems, and/or current status of the artificial intelligence research and its toughest challenges may be helpful. Central concepts of the field will be covered in class.Grades
The grade is based on the class room participation, essays and the term paper.
Grades have been posted in Intranet.
3. Schedule and topics
| Week | Meeting | Topic |
| 1. | Jan 18 | Introduction (Readings) |
| 2. | Jan 25 | Anthropomorphism and Comparative Analysis (Readings) |
| 3. | Feb 1 | Brains and Neurophysiology (Readings) |
| 4. | Feb 8 | Learning and Adaptive Specialization (Readings) |
| 5. | Feb 15 | Memory and Time Binding (Readings) |
| 6. | Feb 22 | Imitation and Social Learning (Readings) |
| 7. | Mar 15 | Consciousness and Theory of Mind (Readings) |
| 8. | Mar 22 | Deception (Readings) |
| 9. | Mar 29 | Language and Logic (Readings) |
| 10. | Apr 12 | Perception, Sensation and Face Recognition (Readings) |
| 11. | Apr 19 | Mental Tool Kits (Readings) |
| 12. | Apr 26 | Paper reports |
4. Material
The seminar material consists of journal articles made available as the semester progresses, and chapters from the book: Marc D. Hauser (2000), Wild Minds: What Animals Really Think. Henry Holt and Company, New York.Articles listed in readings that do not have a link attached, can be found in the course folder in the room C127.
Here are some interesting articles I found after the seminar that may interest some of the participants. Students' term papers are also posted behind this link (in Intranet).
