Begging the Question in Arguments Based
on Testimony
DOUGLAS WALTON
Department of Philosophy
University of Winnipeg
515 Portage Avenue
Manitoba
Winnipeg
Canada R3B 2E9
E-mail: d.walton@uwinnipeg.ca
ABSTRACT: This paper studies some classic cases of the fallacy of begging the question based
on appeals to testimony containing circular reasoning. For example, suppose agents a, b and c
vouch for d’s credentials, and agents b, d, and e vouch for a’s credentials. Such a sequence of
reasoning is circular because a is offering testimony for d but d is offering testimony for a. The
paper formulates and evaluates restrictions on the use of testimonial evidence that might be
used to deal with such problematic arguments. One is called the Non-repeater Rule: in an
extended sequence of argumentation based on testimony, once a source x has been appealed to
at any given point in the sequence, that same source x must never be appealed to again at any
next point in the same sequence.
KEY WORDS: argumentation schemes, begging the question, circular reasoning, evidence
law, fallacies, multi-agent systems, testimony, reputation management, source trustworthiness
INTRODUCTION
This investigation begins with some classic cases of circular reasoning in
arguments based on testimony that have appeared in logic textbooks as
examples of the fallacy of begging the question. Circular argumentation
occurs in cases of arguments based on testimony where a source presents
original testimony, and a secondary appeal to testimony is then made
appealing to the same source to back up the evidential worth of the first
appeal. The study of the problems posed by such cases is shown to lead to
concerns about the evidential worth of arguments based on testimony in
law. It also reveals connections to deeper problems of circular reasoning and
of the fallacy of begging the question. Solutions, it is argued, require taking
account of how an argument was used in an attempt to prove a conclusion
that was doubted by one participant in a dialogue or conversational setting.1
These problems are shown to represent fundamental issues not only for
argumentation theory and logic, but for agent communication systems
currently being developed in distributed computing. Trust and testimony as
Argumentation (2005) 19:85–113
Springer 2005
DOI 10.1007/s10503-004-2071-1

factors in evaluating argumentation are important concerns for artificial
intelligence in multi-agent technology for Internet communication
(Wooldridge, 2000). These systems are based on arguments that assume,
subject to default, that testimony can be seen as provisionally acceptable,
unless there is evidence to the contrary. Thus the problems posed are hardly
trivial, even though the classic cases used to illustrate them in the logic
textbooks don’t initially seem very serious.
The classic cases that are used over and over again in so many textbooks
appearing in the ‘standard treatment’ (Hamblin, 1970) are shown to be
deeply problematic. These problems, it is shown, resurface even after the
literature on a fallacy has become fairly sophisticated. So much for the
standard treatment, many might say – it was always pretty superficial
anyhow. There is something to this dismissal. It would be better to use
examples of arguments that are obviously very important in matters like
science, law or public affairs, in order to convince readers of the seriousness
of our subject. Even so, some of these classic examples have a certain charm,
and do present serious problems that are worth trying to solve. Their
apparent simplicity can be an advantage compared to the complexity of
many cases that occur in specialized contexts like artificial intelligence and
legal argumentation.
TWO TYPES OF CIRCULARITY
Before approaching cases of circular testimony, it may be helpful to re-
view the literature briefly on the fallacy of begging the question, and on
the kind of circular argumentation that gives rise to charges of having
committed this fallacy. Two types of circularity in arguments are recog-
nized by Woods and Walton (1975), one is called equivalency circularity
and the other dependency circularity. The equivalency type of circularity
is the simpler to explain, and also the easier to identify in a given case. In
the equivalency type of circularity, one of the premises of an argument is
the same proposition as the conclusion to be proved. Either the two are
stated using different tokens of the same sentence, or the two differently
worded sentences by which they are stated express the same proposition.
In the very simplest type of equivalency circularity, the premise and the
conclusion are expressed by the very same sentence. This is a type of cir-
cularity that would not be likely to deceive a respondent who is serious and
attentive, but it is worth illustrating, to make some general points about
circular reasoning. Consider the following example.
The Auckland case
Jenna asks Sean to prove that Auckland is in New Zealand, and he replies,
‘‘Auckland is in New Zealand, therefore Auckland is in New Zealand.’’
86 DOUGLAS WALTON