Practical reasoning is a goal-directed sequence of linked practical inferencesthat seeks out a prudent line of conduct for an agent in a set of particular circumstances known by the agent. Where a is an agent, A is an action, and G a goal, thetwo basic types of practical inferences are respectively, the necessary conditionscheme and the sufficient condition scheme (Walton, Pract. Reas., 1 990; see alsoSchellens, 1 987).G is a goal for aDoing A is necessary for a to carry out GTherefore, a ought to do AG is a goal for aDoing A is sufficient for a to carry out GTherefore, a ought to do A

A person who puts forward an argumentation anticipates criticism, and bychoosing a particular type of argumentation, using the one argumentationscheme rather than the other, he implies that he thinks he knows which routewill lead to the justification of his standpoint. At any rate, whether he reallythinks this or not, if he is to be taken seriously by the other party, he may beheld to be committed to deal with the critical questions which pertain to ajustification via the argumentation scheme that is inherent in his argumentation.In relying on a certain argumentation scheme, the arguer invokes a particular testing method in a dialectical procedure, in which certain criticalreactions are relevant, and others not. Each argumentation scheme calls, as itwere, for its own set of critical reactions. In conjunction with each other,these reactions constitute a well-rounded test for checking the soundness of anargumentation of the type concerned. (p. 98)This way of describing argumentation schemes suggests that they are normativelybinding, in the following sense. If the hearer accepts the premises of the speaker'sargument, and the argument is an instance of a genuine and appropriate argumentation scheme (for the type of dialogue they are engaged in), then the hearer mustor should (in some binding way) accept the conclusion. This does not appear to be"validity" in the same sense in which the word is familiarly used in deductive (orperhaps even inductive) logic. But it does appear to express a normative or broadlylogical sense of validity, bindingness, conditional acceptability, or whatever youwant to call it

Arguments like the one in Case 1 . 1 lie on a razor's edge: They are somewhatweak and unreliable, and apt to fail occasionally, but they are not so bad or inherently erroneous that they should be called "fallacious" in all instances. On theother hand, they can tum out to be fallacious, in some cases. And, in particular,they run the risk of committing the secundum quid fallacy as quite a general sort offailure they are prone to. If this is right, a new approach to fallacies is called foran approach that takes more care in assessing the particulars of a given case.In arguments like the one in Case 1 . 1 , the premise, if true in a given case,does give a reason for accepting the conclusion. But it is not a conclusive reason,and it is subject to default relative to what is known (or becomes known) of thefurther circumstances of the case. The problem then is to find the underlying structure of inference in such a case that enables one to identify and test the correctness(or incorrectness) of the argument as an instance of an argumentation scheme