Commitments, Expectations, Affordances and Susceptibilities: Towards Positional Agent Programming
1. Commitments, Expectations,
Affordances and Susceptibilities
Giovanni Sileno (g.sileno@uva.nl),
Alexander Boer, Tom van Engers
Leibniz Center for Law
University of Amsterdam
29 October 2015 – PRIMA @ Bertinoro (FC)
Towards Positional Agent Programming
3. From institutional positions..
●
In a formal institution, each actor is bound to other
actors according to certain legal relationships,
associated to certain legal positions..
4. From institutional positions..
●
Positions have inter-dependencies, e.g.
– if a party has a duty to perform a certain action,
then there is another party that has a claim
towards the first
– if a party is in a certain position (e.g. duty to A),
this precludes the same party to be in another
position (e.g. no-duty to A).
5. From institutional positions..
●
Positions have inter-dependencies, e.g.
– if a party has a duty to perform a certain action,
then there is another party that has a claim
towards the first
– if a party is in a certain position (e.g. duty to A),
this precludes the same party to be in another
position (e.g. no-duty to A).
●
Hohfeld [1917] presented a visual organization of
fundamental legal positions
6. First Hohfeldian square
CLAIM
RIGHT
DUTY
correlative
opposite opposite
NO-CLAIM
NO-RIGHT
PRIVILEGE
LIBERTY
NO-DUTY
beneficiary perspective addressee perspective
W. N. Hohfeld. Fundamental legal conceptions as applied in
judicial reasoning. The Yale Law Journal, 1917.
9. Similar asymmetries
We found that these asymmetries can be related to
deontic logic:
G. Sileno, A. Boer, T. Van Engers. “On the Interactional
Meaning of Fundamental Legal Concepts", JURIX 2014
10. A viable laternative?
●
Blanché proposed in the 70s to extend the
Aristotelian square after considerations on natural
language.
11.
12. A E
I O
ALL NONE
SOME
NOT
SOME
Y
U
ALL or NONE
SOME and SOME NOT “≡ SOME”
13. A E
Y
forb A
+ -
obl A
“perm” A = faculty A
0
Deontic triangle of contrariety
positive polarity negative polarity
no polarity
14. A E
Y
forb A
+ -
obl A
“perm” A = faculty A
0
Deontic triangle of contrariety
positive polarity negative polarity
no polarity
Two negations operators:
polarity inversion and
nullification!
15. A E
Y
forb A
+ -
obl A
“perm” A = faculty A
0
Deontic triangle of contrariety
positive polarity negative polarity
no polarity
Two negations operators:
polarity inversion and
nullification!
20. A
E
Y
forb
obl
“perm” = faculty
DUTYCLAIM
NO-CLAIM
PRIVILEGE
beneficiary perspective addressee perspective
+
-
right to
protection against
right to
performance
First Hohfeldian Prism
G. Sileno, A. Boer, T. Van Engers. “On the Interactional
Meaning of Fundamental Legal Concepts", JURIX 2014
21. E
Y
“perm”, faculty to follow along
+
--
LIABILITYPOWER
DISABILITY
IMMUNITY
forb to follow along
obl to follow along
performer perspective recipient perspective
+A
Second Hohfeldian Prism
G. Sileno, A. Boer, T. Van Engers. “On the Interactional
Meaning of Fundamental Legal Concepts", JURIX 2014
(positive)
power
negative
power
23. ...to agential positions
●
intuition: correlativeness of legal relationships
holding between two parties can be put in relation
with the correlativeness of the agent with his
own environment.
24. ...to agential positions
●
intuition: correlativeness of legal relationships
holding between two parties can be put in relation
with the correlativeness of the agent with his
own environment.
●
the difference is one of extrinsic vs intrinsic
components:
– extrinsic, created by normative forces at social level
– intrinsic, by internal structuring processes at agent level
44. Computational grounding
●
We decided to ground our work on Petri nets:
– positions are local states, and events have impact
on local scale (mental causation)
– the two elements we consider are operations
(composition of transitions) for actions/events,
situations (composition of places) for conditions
and references.
47. From commitment to action
●
According to the prevent-acquire-cure-keep (PACK)
psychological framework [Ogilvie and Rose, 1995]
– the presence or absence
– of a positive (negative) condition
– guides the agent to select a certain behaviour,
– in order to promote (demote) such condition.
48. From commitment to action..
●
ACQUIRE:
– If you have a commitment towards a certain
target, not holding at the moment, and an
associated affordance is available, then use it.
49. From commitment to action..
●
PREVENT:
– If you have a negative commitment towards a
certain target, which is not holding at the
moment, and you have a negative affordance
towards it, then use such affordance.
– Inhibit affordances with undesired side-effects.
50. From commitment to monitoring
●
Not all that the agent may perceive or infer from his
knowledge is relevant to his commitments.
51. From commitment to monitoring
– identifying potential
situations enabling
changes to current
and potential PACK-
related positions
●
The relevance relation can be extracted from the
commitment specifications, considering two
directions:
c
52. From commitment to monitoring
●
The relevance relation can be extracted from the
commitment specifications, considering two
directions:
c
feedback processes
– identifying potential
situations enabling
changes to current
and potential PACK-
related positions
– circumscribing
success and failure
references
54. “I want to finish this article
before this weekend.”
55. “I want to finish this article
before this weekend.”
“I don't want to get to the
weekend without finishing this
article.”
They are similar, but not the same.
56. “I want to finish this article
before this weekend.”
“I don't want to get to the
weekend without finishing this
article.”
They activate different PACK rules
(ACQUIRE vs PREVENT) and therefore
different susceptibilities!
59. Conclusions
●
Most known agent languages/architectures today:
– conflates negative and null positions
– overlook the susceptibility components
(critical to design evidence gathering tasks)
60. Conclusions
●
Most known agent languages/architectures today:
– conflates negative and null positions
– overlook the susceptibility components
(critical to design evidence gathering tasks)
●
This contribution attempts to reduce this gap,
– grounding the associated practical reasoning in
cognitive research studies
61. Conclusions
●
Most known agent languages/architectures today:
– conflates negative and null positions
– overlook the susceptibility components
(critical to design evidence gathering tasks)
●
This contribution attempts to reduce this gap,
– grounding the associated practical reasoning in
cognitive research studies
– trying alternative solutions for known
computational/representational issues
62. Conclusions
●
Most known agent languages/architectures today:
– conflates negative and null positions
– overlook the susceptibility components
(critical to design evidence gathering tasks)
●
This contribution attempts to reduce this gap,
– grounding the associated practical reasoning in
cognitive research studies
– trying alternative solutions for known
computational/representational issues
●
Many extensions already prefigured!