Optimality Theory (OT) is a linguistic framework that uses constraints and rankings rather than rules to map inputs to outputs in language. It generates all possible candidate outputs and then evaluates them based on how well they satisfy ranked constraints to determine the optimal output. The document provides examples of how OT can be applied to robot ethics scenarios by ranking constraints like "do not injure humans" and production of language by ranking needs like taste, ease, and cost in choosing a type of coffee. Tableaus are used to evaluate candidates against constraints.

1. How Does OT Work?
Presenter: Aicha ADOUI

2. How Does OT Work?
In the most basic way:
“[…] Instead of rules to figure out what is and is
not ‘allowed’ in a language OT uses
constraints and structures […] as systems that
map from the input to the output. The input is
referred to as the underlying form whereas the
output is the surface realization[…]”
(Optimality Theory 101: Constraints > Rules, by Gretchen McCulloch, 2014)
Presenter: Aicha ADOUI

3. 1.1 OT in General: eg. Robot Ethics

4. 1.1 OT in General: eg. Robot Ethics
Isaac Asimovs ethical rules for the behaviour of
robots: the “three laws of robotics”

5. 1.1 OT in General: eg. Robot Ethics
Isaac Asimovs ethical rules for the behaviour of
robots: the “three laws of robotics”
a. Robot Ethics and Potential Conflicts
1. A robot may not injure a human being or,
through inaction, allow a human being to come
to harm.

6. 2. A robot must obey the orders given to it by
human beings, except where such orders
would conflict with the First Law.
3. A robot must protect its own existence, as
long as such protection does not conflict with
the First or Second Law.

7. b. Robot Ethics in OT

8. b. Robot Ethics in OT
*INJURE HUMAN : A robot may not injure a
human being or, through inaction, allow a
human being to come to harm.
OBEY ORDER: A robot must obey the orders of
human beings.
PROTECT EXISTENCE: A robot must protect its
own existence.

9. c. Ranking:
*INJURE HUMAN OBEY ORDER 
PROTECT EXISTENCE

10. D. Story time ^^

11. D. Story time ^^
Human says to Robot: Kill my friend!
1. R kills H’s friend
2. R kills H (who gave him the order)
3. R doesn’t kill anyone
4. R kills himself

12. Evaluating Possible Outcomes
1. Setting up a tableau
2. Assigning violation marks
3. Eliminating suboptimal candidates

13. Step 1: Setting up a Tableau
Jochen Trommer jtrommer@uni-leipzig.de A Crash Course in Optimality Theory

14. Step 2: Assigning Violation Marks
Jochen Trommer jtrommer@uni-leipzig.de A Crash Course in Optimality Theory

15. Step 3: Eliminating Suboptimal
Candidates
Jochen Trommer jtrommer@uni-leipzig.de A Crash Course in Optimality Theory

16. Jochen Trommer jtrommer@uni-leipzig.de A Crash Course in Optimality Theory

17. 1.2 OT in General: e.g I need a Coffee!

18. 1.2 OT in General: e.g « I need a Coffee! »
Input: how to get a coffee?
GEN (options/candidates):
1- Don’t bother at all
2- Make terrible instant coffee
3- Brew your own really good coffee from scratch
4- Get a tasteless cup at the nearby corner store
5- Get a really good coffee from slightly-further-
away Starbucks or
6- Get a really good but expensive coffee from an
indie coffee-shop at quite a distance away

19. 1.2 OT in General: e.g « I need a Coffee! »
CON (you have 4 needs to meet!):
1- You want caffeine
2- You want it to be easy
3- Taste good
4- You don’t want it to be expensive.

20. N.B: We put the candidates and the
constraints in a tableau, with the constraints
ranked in their importance to you from left to
right, we can figure out where you should get
your coffee (this whole step is known as
evaluation or EVAL).

21. Tableau
#Evaluation Time

22. edited

23. OT in SUM
 Take the input and generate (GEN) an infinite
number of possible outputs (add elements,
delete them, modify them, anything goes)
 Evaluate (EVAL) them to see how well they
follow or violate the constraints and rankings of
the language.
 The output is the candidate that is optimal
because it violates the fewest or lowest ranked
constraints