Elqayam & Evans (E&E) propose that the study of human mental processing is best served by keeping normative and descriptive accounts of reasoning competence separate. We enthusiastically endorse E&E's purely descriptivist approach, and we share their skepticism about normativism. However, we think that E&E have seriously underestimated how hard it is to undermine the view that a competence theory is also a normative theory. We begin by explaining why the link between competence and normativity is so deeply embedded in philosophical thinking. We then argue that dual-system theories of reasoning may provide a powerful new way of challenging this link.
Traditionally, the philosophical literature on rationality has focused on the question of when an inference is justified. In one of the most influential passages of twentieth-century philosophy, Nelson Goodman offered the following answer:
The validity of a deduction depends upon … conformity to valid rules …. But how is the validity of rules to be determined? … Principles of deductive inference are justified by their conformity with accepted deductive practice. Their validity depends upon accordance with the particular deductive inferences that we actually make and sanction. If a rule yields inacceptable inferences, we drop it as invalid. Justification of general rules thus derives from judgments rejecting or accepting particular deductive inferences.
This looks flagrantly circular …. But this circle is a virtuous one. The point is that rules and particular inferences alike are justified by being brought into agreement with each other. A rule is amended if it yields an inference we are unwilling to accept; an inference is rejected if it violates a rule we are unwilling to amend. The process of justification is the delicate one of making mutual adjustments between rules and accepted inferences; and in the agreement achieved lies the only justification needed for either. (Goodman Reference Goodman1965, pp. 66–67, emphasis in the original)
In A Theory of Justice, John Rawls endorses this “process of mutual adjustment of principles and considered judgments” as the primary method for justifying moral judgments and moral principles (Rawls Reference Rawls1971, p. 20). When this process is successful, we are in what Rawls labels “reflective equilibrium.” Though both the label and the description of this method are twentieth century products, a case can be made that the method itself was used by Aristotle, Plato, and many other important figures in the history of philosophy (see Rawls Reference Rawls1971, p. 51; cf. Stich Reference Stich2001).
Rawls famously goes on to note that there is an analogy between the method of reflective equilibrium used in moral philosophy and the Chomskian project of “describing the sense of grammaticalness that we have for the sentences of our native language” (1971, p. 47). For Chomsky and his followers, the project that Rawls has in mind goes by another name: describing a speaker's grammatical competence. A decade later, L. J. Cohen extended Rawls' analogy to the study of reasoning competence (Cohen Reference Cohen1981). To determine our reasoning competence, Cohen maintained, we must assemble lots of information about the inferences that we actually accept and then construct the simplest set of rules that captures most of those inferences. To construct a normative theory of reasoning of the sort that Goodman proposed, we would do exactly the same thing. Thus, Cohen argued, a descriptive account of our reasoning competence and a normative theory of reasoning that is supported by the reflective equilibrium method must coincide. Though people may make lots of performance errors in reasoning, their underlying reasoning competence must be normatively impeccable.
According to E&E, “inferring an ‘ought’ type conclusion from ‘is’ type premises is highly controversial, and considered by many authors to be a logical fallacy” (target article, sect. 5, para. 2). They are surely right that some inferences of that sort are very controversial in philosophy. However, the inference from the “is” type premise:
(i) This set of (moral or inferential) rules is in reflective equilibrium.
to the “ought” type conclusion:
(ii) This set of rules is justified.
is far less controversial. Indeed, many philosophers would insist that in both ethics and logic this is by far the best way of arguing that a set of rules is justified. But for E&E's critique of normativism to be persuasive, the inference from (i) to (ii) must be challenged.
One of the most effective challenges, we believe, is to emphasize the way in which such inferences can lead to radical relativism. In Stich (Reference Stich1981; Reference Stich1990) it was noted that it is logically possible that just about any set of inference rules could be in reflective equilibrium for a cognitive agent. So if the inference from (i) to (ii) is accepted, just about any set of inference rules could be justified. This strategy made little headway, however, since there was no evidence that it is psychologically possible for people to have significantly different reasoning competences. We are inclined to think that Stanovich's early work on individual differences goes a long way toward addressing that concern, though we do not endorse his claim that the inferential competence of people with high IQ should be considered normatively privileged (Stanovich Reference Stanovich1999).
Recent work on dual-system theories of reasoning poses another powerful challenge to the idea that reasoning competence must be normatively impeccable. If these dual-system theories are on the right track, then reasoning is subserved by two distinct systems, each with its own competence. Moreover, if – as E&E suggest – much of the content of System 2 is learned via explicit instruction or acquired from the surrounding culture, then it is to be expected that the inferential rules in System 2 will be different for different individuals and different cultures. If each person's reasoning is subserved by two quite different systems, and the second system varies significantly among individuals and cultures, the claim that an individual's reasoning competence must be normatively impeccable is very implausible indeed. As awareness of dual-system theories of reasoning becomes more widespread, it may finally begin to undermine the deeply entrenched philosophical view that a theory of reasoning competence has normative implications.
Elqayam & Evans (E&E) propose that the study of human mental processing is best served by keeping normative and descriptive accounts of reasoning competence separate. We enthusiastically endorse E&E's purely descriptivist approach, and we share their skepticism about normativism. However, we think that E&E have seriously underestimated how hard it is to undermine the view that a competence theory is also a normative theory. We begin by explaining why the link between competence and normativity is so deeply embedded in philosophical thinking. We then argue that dual-system theories of reasoning may provide a powerful new way of challenging this link.
Traditionally, the philosophical literature on rationality has focused on the question of when an inference is justified. In one of the most influential passages of twentieth-century philosophy, Nelson Goodman offered the following answer:
The validity of a deduction depends upon … conformity to valid rules …. But how is the validity of rules to be determined? … Principles of deductive inference are justified by their conformity with accepted deductive practice. Their validity depends upon accordance with the particular deductive inferences that we actually make and sanction. If a rule yields inacceptable inferences, we drop it as invalid. Justification of general rules thus derives from judgments rejecting or accepting particular deductive inferences.
This looks flagrantly circular …. But this circle is a virtuous one. The point is that rules and particular inferences alike are justified by being brought into agreement with each other. A rule is amended if it yields an inference we are unwilling to accept; an inference is rejected if it violates a rule we are unwilling to amend. The process of justification is the delicate one of making mutual adjustments between rules and accepted inferences; and in the agreement achieved lies the only justification needed for either. (Goodman Reference Goodman1965, pp. 66–67, emphasis in the original)
In A Theory of Justice, John Rawls endorses this “process of mutual adjustment of principles and considered judgments” as the primary method for justifying moral judgments and moral principles (Rawls Reference Rawls1971, p. 20). When this process is successful, we are in what Rawls labels “reflective equilibrium.” Though both the label and the description of this method are twentieth century products, a case can be made that the method itself was used by Aristotle, Plato, and many other important figures in the history of philosophy (see Rawls Reference Rawls1971, p. 51; cf. Stich Reference Stich2001).
Rawls famously goes on to note that there is an analogy between the method of reflective equilibrium used in moral philosophy and the Chomskian project of “describing the sense of grammaticalness that we have for the sentences of our native language” (1971, p. 47). For Chomsky and his followers, the project that Rawls has in mind goes by another name: describing a speaker's grammatical competence. A decade later, L. J. Cohen extended Rawls' analogy to the study of reasoning competence (Cohen Reference Cohen1981). To determine our reasoning competence, Cohen maintained, we must assemble lots of information about the inferences that we actually accept and then construct the simplest set of rules that captures most of those inferences. To construct a normative theory of reasoning of the sort that Goodman proposed, we would do exactly the same thing. Thus, Cohen argued, a descriptive account of our reasoning competence and a normative theory of reasoning that is supported by the reflective equilibrium method must coincide. Though people may make lots of performance errors in reasoning, their underlying reasoning competence must be normatively impeccable.
According to E&E, “inferring an ‘ought’ type conclusion from ‘is’ type premises is highly controversial, and considered by many authors to be a logical fallacy” (target article, sect. 5, para. 2). They are surely right that some inferences of that sort are very controversial in philosophy. However, the inference from the “is” type premise:
(i) This set of (moral or inferential) rules is in reflective equilibrium.
to the “ought” type conclusion:
(ii) This set of rules is justified.
is far less controversial. Indeed, many philosophers would insist that in both ethics and logic this is by far the best way of arguing that a set of rules is justified. But for E&E's critique of normativism to be persuasive, the inference from (i) to (ii) must be challenged.
One of the most effective challenges, we believe, is to emphasize the way in which such inferences can lead to radical relativism. In Stich (Reference Stich1981; Reference Stich1990) it was noted that it is logically possible that just about any set of inference rules could be in reflective equilibrium for a cognitive agent. So if the inference from (i) to (ii) is accepted, just about any set of inference rules could be justified. This strategy made little headway, however, since there was no evidence that it is psychologically possible for people to have significantly different reasoning competences. We are inclined to think that Stanovich's early work on individual differences goes a long way toward addressing that concern, though we do not endorse his claim that the inferential competence of people with high IQ should be considered normatively privileged (Stanovich Reference Stanovich1999).
Recent work on dual-system theories of reasoning poses another powerful challenge to the idea that reasoning competence must be normatively impeccable. If these dual-system theories are on the right track, then reasoning is subserved by two distinct systems, each with its own competence. Moreover, if – as E&E suggest – much of the content of System 2 is learned via explicit instruction or acquired from the surrounding culture, then it is to be expected that the inferential rules in System 2 will be different for different individuals and different cultures. If each person's reasoning is subserved by two quite different systems, and the second system varies significantly among individuals and cultures, the claim that an individual's reasoning competence must be normatively impeccable is very implausible indeed. As awareness of dual-system theories of reasoning becomes more widespread, it may finally begin to undermine the deeply entrenched philosophical view that a theory of reasoning competence has normative implications.