=Paper=
{{Paper
|id=Vol-2048/paper07
|storemode=property
|title=Argumentation Schemes for Critical Thinking on Current Events
|pdfUrl=https://ceur-ws.org/Vol-2048/paper07.pdf
|volume=Vol-2048
|authors=Nancy L. Green
|dblpUrl=https://dblp.org/rec/conf/icail/Green17a
}}
==Argumentation Schemes for Critical Thinking on Current Events==
https://ceur-ws.org/Vol-2048/paper07.pdf
Argumentation Schemes for Critical Thinking on Current Events
Nancy L. Green
University of North Carolina Greensboro, Greensboro, NC, USA
nlgreen@uncg.edu
Abstract recognition resemble aspects of six argumentation
schemes we have identified.
We have been examining a class of
arguments used in sophisticated analyses of In the field of argumentation studies, the most
current events, with the goal of developing a closely related schemes are Practical Reasoning and
visualization tool that would assist users in Argument from Positive/Negative Consequences, e.g.
understanding or producing those kinds of as described in [Walton et al., 2008]. The conclusions
arguments. While some current tools make of those schemes assert what an agent should do.
argumentation scheme sets available, they Thus, those schemes share the perspective of the field
do not describe an important class of of automated planning in artificial intelligence. In
arguments. This paper describes work in contrast, the arguments of interest here involve
progress to identify those argumentation reasoning about what an actor’s plan must be (or
schemes. must have been), i.e., the perspective of plan
recognition. In computational studies of
argumentation, Bex et al. [2009] present a scheme for
1 Introduction abductive practical reasoning, which can be used to
explain an agent’s motivation for taking an action.
Critical thinking on current events often involves
reasoning about the beliefs, goals, plans, and actions
of actors such as countries, governments, politicians, 3 Argumentation Schemes
etc. We have been examining a class of arguments The argumentation schemes were abstracted from
used in sophisticated analyses of current events, with arguments about plans of the Russian government (R)
the goal of developing a tool that would assist users in an article on Russia’s involvement in the Syrian
in understanding or producing those kinds of conflict [Weinberger, 2016] and in another article on
arguments. Such a tool could be of use in on-line Russia’s alleged attempt to influence the outcome of
environments for citizen engagement [Bex et al., the 2016 U.S. presidential election [Office of the
2013] or in educational settings [Scheuer et al., 2010; Director of National Intelligence, 2017]. The
Pinkwart and McLaren, 2012]. To scaffold the user’s following scheme was used, for example, to argue
task, our tool will support the visualization of that R acted to help the U.S. presidential candidate
arguments and provide a set of argumentation Donald Trump (T) to defeat Hillary Clinton (HC).
schemes, abstract descriptions of acceptable forms of The premise is that R’s actions (disclosing
argument. While some current tools make argument unfavorable information about HC through
scheme sets available, e.g., subsets of the schemes Wikileaks, promoting anti-HC propaganda, etc.) were
listed in [Walton et al., 2008], they do not describe an consistent with a plan to help T to defeat HC.
important class of arguments found in the analyses of
interest to us. This paper describes work in progress Argument-from-Inferred-Plan
to identify those argumentation schemes. Premises:
1. Actor is doing/did Act(s) consistent with a Plan
2 Background for achieving Goal(s). (Note: some acts of Plan
may not have been done yet.)
The arguments of interest in this paper are closely
Conclusion: Act(s) are/were part of Plan for
related to the field of research on plan recognition in
achieving Goal(s).
artificial intelligence and natural language processing
Critical Questions:
[Carberry, 1990]. The earliest work in that field used
a. In Actor’s view, is benefit/cost of Plan high
heuristic rules describing the relationships among an
enough to justify doing Acts of Plan? For
agent’s beliefs, goals, actions, and plans. Due to its
example, is it likely to Actor that this Plan will
computational complexity that approach to plan
have undesirable side effects?
recognition has been supplanted with probabilistic
b. In Actor’s view, is there a plausible alternative
approaches. However, the heuristic rules for plan
preferable plan for Goal(s)? For example, is the
38 18th Workshop on Computational Models of Natural Argument
Floris Bex, Floriana Grasso, Nancy Green (eds)
16th July 2017, London, UK
likelihood of the success of this Plan lower than Argument-from-Inferred-Appraisal-Based-
success of an alternative plan in the view of Actions
Actor? Premises:
1. In Actor’s view, Act(s) has/have likelihood of
The critical questions of the scheme are related to the Consequence(s).
critical questions of Practical Reasoning, except that 2. In Actor’s view, Consequence(s) is/are desirable,
instead of challenging a planner’s argument, they or is/are not desirable.
challenge a plan recognizer’s argument. (Critical Conclusion: Actor did (or intends to do) Act(s) to
questions are provided with the schemes to help users lead to positively appraised Consequences, or
understand/generate counter-arguments.) respectively, did not (or intends to not) do Act(s) to
The next scheme was used, for example, in an avoid negatively appraised Consequences.
argument for R’s attempt to influence the election Critical Questions:
based on R’s pattern of behavior of attempting to a. In actor’s view, is there a good way to do Act(s)
influence elections in other countries. To describe the while mitigating or avoiding negatively
actor’s behavior in terms of planning algorithms, it appraised consequences?
resembles creating a plan using cased-based- b. In actor’s view is the benefit/cost of doing Act(s)
reasoning (CBR) [Kolodner, 1993]. Thus, critical leading to positively appraised Consequences
question (a) involves the notion from CBR of worthwhile?
adapting old plans.
As far as we know, the field of AI planning does
Argument-from-Behavior-Pattern not address the creation of duplicitous plans.
Premises: However, there is a need for such a scheme in
1. Actor has/had Goal(s), which is/are similar to analyzing world events. The following scheme was
Past-goal(s). abstracted from an argument that R has built up its
2. Actor is doing/did Act(s), which is/are similar to military in Syria to limit U.S. operations in the
Past-act(s) that were part of a plan that was Middle East. The premises were that R built up its
successful in achieving Past-goal(s). military there for the alleged goal of fighting
Conclusion: Act(s) are/were part of Plan for terrorism, but the buildup was inconsistent with that
achieving Goal(s). goal. However, the buildup was consistent with the
Critical Questions: suspected true goal of limiting U.S. operations in the
a. In Actor’s view, can the old plan be successfully Middle East, a goal that the U.S. would oppose.
adapted?
b. In Actor’s view, is benefit/cost of Plan high Argument-from-Plan-Deception
enough to justify doing Acts of Plan? For Premises:
example, is it likely that this Plan will have 1. Actor did (or intends to do) Act(s) with Alleged-
undesirable side effects? goal(s).
c. In Actor’s view, is there a plausible alternative 2. Effect(s) of Act(s) is/are inconsistent with
preferable plan for Goal(s)? For example, is the Alleged-goal(s)
likelihood of the success of this Plan lower than 3. Effect(s) of Act(s) is/are consistent with
success of an alternative plan? suspected True-goal(s) of Actor.
4. Effect(s) of Act(s) is/are (or would be)
The next argument scheme is related to inferring negatively appraised and/or met with opposition
actions that might have resulted from an actor’s use by Protagonist
of Argument from Positive/Negative Consequences. Conclusion: Actor did (or intends to do) Act(s) as
For an example related to positive consequences, the (part of) a plan to achieve True-goal(s).
argument that R wanted to help T defeat HC in the Critical Questions:
election was supported by the premise that R a. In actor’s view, is benefit/cost of plan high
believed that President T would partner with R in enough to justify doing Act(s)?
counter-terrorism activities, a positive consequence b. Is it possible that actor does not realize that
in R’s view. effects of acts are inconsistent with alleged-
goals?
Some modern AI planning systems incorporate
affective reasoning into planning, e.g. [Gratch, 2000].
The next scheme involves not only reasoning about
an actor’s plan, but also the actor’s beliefs about the
18th Workshop on Computational Models of Natural Argument 39
Floris Bex, Floriana Grasso, Nancy Green (eds)
16th July 2017, London, UK
protagonist’s response to the plan. An example is the For an example of how the schemes may be
argument that since the U.S. has not resisted R’s combined, see Figures 1 and 2.
military buildup in the Middle East, R believes that
the U.S. will not intervene to prevent R’s future 4 Discussion
military buildup in the Far East. Arguments about the inferred plans of other actors
are important for critical thinking about world events,
Argument-of-Increasing-Boldness yet have not been recognized as an important class of
Premises: argumentation schemes. Real-world actors and their
1. Actor did Act(s) to achieve Goal(s). plans are more complex than the robot worlds
2. Act(s) was/were not resisted by Protagonist. modeled in the early days of artificial intelligence
3. Actor wants to do Similar-act(s) to achieve planning research. However, the early heuristics
Similar-Goal(s). proposed for plan recognition in AI can provide
Conclusion: Actor believes that Protagonist will not insight into the specification of argumentation
intervene to prevent Similar-Act(s). schemes for this class of arguments.
Critical Questions:
a. In actor’s view, is benefit/cost high enough to Acknowledgments
justify doing Similar-act(s)?
b. In actor’s view, is it likely that Protagonist will We appreciate the assistance of graduate student
not resist Similar-act(s)? Michael Branon. This material is based upon work
supported in whole or in part with funding from the
The following two schemes were used together. Laboratory for Analytic Sciences (LAS). Any
The first was used in an argument that R is opinions, findings, conclusions, or recommendations
attempting to coerce the U.S. into not resisting R expressed in this material are those of the author(s)
expansion by the threat of conventional war or a and do not necessarily reflect the views of the LAS
nuclear response. The second was used to argue for and/or any agency or entity of the United States
resisting the attempted coercion by providing Government.
evidence that R would be incapable of acting on
those threats. References
[Bex et al., 2009] Bex, F., Bench-Capon, T., and
Argument-of-Coercion Atkinson, K. Did he jump or was he pushed?
Premises: Abductive Practical Reasoning. Artificial Intelligence
1. Actor threatens doing Act(s) that Actor believes and Law 17, 2009, p. 79-99.
are negatively appraised by Protagonist.
2. Actor suggests that Actor will not do Act(s) if [Bex et al., 2013] Bex, F., Lawrence, J., Snaith, M. &
Protagonist does Coerced-act(s). Reed, C. Implementing the Argument Web.
3. Coerced-act(s) are consistent with Actor’s Communications of the ACM. 56(10), 2013, p. 66-73
Goal(s).
Conclusion: Actor is attempting to coerce [Carberry, 1990] Carberry, S. Plan Recognition in
Protagonist to do Coerced-act(s). Natural Language Dialogue. MIT Press, 1990.
Critical Questions:
a. In actor’s view, is protagonist likely to believe [Gratch, 2000]. Gratch, J. Emile: Marshalling
that actor could or would carry out threats? Passions in Training and Education. In Proceedings
of 4th International Conference on Autonomous
Argument-for-Resisting-Coercion Agents, June 2000.
Premises:
1. Actor is attempting to coerce Protagonist to do [Kolodner, 1993] Kolodner, J.. Case-Based
Coerced-act(s), via threat of doing Act(s) that Reasoning. San Mateo: Morgan Kaufmann, 1993.
Actor believes are negatively appraised by
Protagonist. [Office of the Director of National Intelligence,
2. In actuality, Actor is incapable of doing the 2017] Office of the Director of National Intelligence.
Act(s). Assessing Russian Activities and Intentions in Recent
3. If Protagonist does Coerced-Act(s) it may have US Elections. National Intelligence Council, 2017,
negative consequences for Protagonist. ICA 2017-01D.
Conclusion: Protagonist need not do Coerced-
Act(s).
40 18th Workshop on Computational Models of Natural Argument
Floris Bex, Floriana Grasso, Nancy Green (eds)
16th July 2017, London, UK
[Pinkwart and McLaren, 2012] Pinkwart, N. and [Walton et al., 2008] D. Walton, C. Reed, and F.
McLaren, B.M. (Eds.) (2012). Educational Macagno. Argumentation Schemes. Cambridge
Technologies for Teaching Argumentation Skills. University Press, 2008.
Sharjah: Bentham Science Publishers.
[Weinberger, 2016] Weinberger, K. Putin sets the
[Scheuer et al., 2010] Scheuer, O., Loll, F., Pinkwart, stage for the incoming U.S. administration. Institute
N., and McLaren, B.M. Computer-Supported for the Study of War, 2016. (Downloaded from
Argumentation: A Review of the State of the Art. )
Computer-Supported Collaborative Learning 2010,
5(1): 43-102.
Figure 1. Example of argumentation schemes in combination
.
18th Workshop on Computational Models of Natural Argument 41
Floris Bex, Floriana Grasso, Nancy Green (eds)
16th July 2017, London, UK
Figure 2. Example of argumentation schemes in combination
.
42 18th Workshop on Computational Models of Natural Argument
Floris Bex, Floriana Grasso, Nancy Green (eds)
16th July 2017, London, UK