Chapter 6 - Concluding remarks
If it is the case that human reasoning is best evaluated within a context-sensitive consequentialist framework as I have argued in the previous chapter, what does this suggest for the study of human reasoning and rationality? I think there are implications for the way in which cognitive psychologists and philosophers conceive of some of the most important questions within the rationality debate. In particular, I would like to focus here on the main implications that the adoption of the proposed approach to rationality assessment would have in the analysis and the evaluation of subjects' final responses to reasoning tasks.
As seen in the previous chapters, the researchers involved in the rationality debate can be divided into two camps on the basis of their commitment to either a deontological or a consequentialist conception of rationality. It is by relying on either of these conceptions that subjects' performances on reasoning tasks have been assessed in terms of one's ability either to comply with a set of normative principles (in the case of deontological approaches) or to reach a specific range of goals efficiently (in the case of consequentialist approaches). In any event, whether we accept a deontological approach to rationality assessment, either standard or non-standard, or opt for a consequentialist approach, such as those based on evolutionary or pragmatist considerations, we are still a long way from being able to provide a complete assessment of reasoning performances. On the one side, deontological accounts appear to be unsatisfactory, because by defining good reasoning as reasoning that conforms to a set of normative principles, they do not explain the normative force of these principles. On the other side, while agreeing with the general idea underlying the consequentialist approach to rationality assessment, I have maintained that consequentialist approaches, whether based on evolutionary or pragmatist considerations, are flawed. With regard to consequentialist approaches of the evolutionary kind, I criticize the tendency of their supporters to justify the normative appropriateness of a reasoning performance by referring to the environment of evolutionary adaptation where its underlying cognitive mechanisms have evolved, rather than to its current context. Indeed, if reasoning is performed in a particular context and thereby is a situated event, it seems reasonable to think that it must be assessed with respect to that context, not with respect to the environment of evolutionary adaptation. As to the pragmatist kind of consequentialist approach to rationality assessment, by focusing exclusively on considerations concerning the value or significance of the goals pursued by the reasoners and the cognitive resources available to them for the attainment of their goals, pragmatists tend to neglect the distinction between the situational and cognitive constraints under which reasoners operate and consequently fail to distinguish the respective roles of those constraints in rationality assessment.
While the supporters of all of these four approaches have proposed different ways of characterizing what counts as rational behaviour, when it comes to the evaluation of reasoning performances they all display the same attitude, namely, they deal with the subjects' task interpretation as a flexible object to be adjusted depending on their own research targets. On the one hand, those who trace subjects' reasoning performances back to a predetermined normative standard assume that their subjects have understood the reasoning problem in the same way as they do and with the same background assumptions. On the other hand, other researchers hold that there is always an appropriate way to interpret the answers of the subjects as showing that they have understood the reasoning problem differently from how it was understood by the experimenter, so that their answers can be regarded as normatively appropriate to that which they respond to. In both these cases, the subject's task interpretation is not regarded as belonging to the reasoning performance and consequently as a part of what is subject to evaluation. So, researchers from both these trends leave it open to themselves to adapt their reading of the subjects' task interpretation to their theoretical target (that is, to supporting a certain hypothesis about human rationality). Thus, any response can be regarded as normatively appropriate or inappropriate directly following from the task interpretation that researchers impute to the subjects. In contrast to this widely accepted attitude, in the fifth chapter I have argued that subjects' task interpretations, as well as subjects' responses to tasks, require a normative background against which to be assessed as to appropriateness or legitimacy. To address this question, I have proposed a two-step normative framework for situationally establishing the legitimacy of subjects' task interpretations and the normative appropriateness of their responses. My approach specifies two kinds of constraints that operate within the experimental context. On the one hand, the subjects' understanding of the reasoning problem is always constrained by the information available in the problem context. On the other hand, as is held by consequentialists, the evaluation of success in reasoning cannot be separated from the evaluation of success in achieving goals. With regard to the first set of constraints, I have observed that participants and experimenters have many different interpretations of reasoning problems. Regardless of how experimenters interpret a reasoning problem, if the subjects' framing of the problem matches the contextual information available in the experimental situation, it should be regarded as a legitimate interpretation of that problem. On this view, the same problem may be framed in different ways on the condition that the resulting representation matches the information available in the problem context. As to the second set of constraints, depending on the goal-structure of framings, in order to assess reasoning performances we should first ask what purpose reasoner's answers are aimed at and then consider whether they are correct or incorrect with respect to it. Once a goal has been set, a reasoning performance is normatively appropriate depending on its effectiveness in achieving that goal. As a result, such an approach to rationality assessment holds that the information explicitly presented in the experimental context places constraints on the goals and frames that it will be appropriate for the reasoner to activate when dealing with the task, which in turn determine the types of reasoning strategy that will be effective and efficient.
It is clear that the subject's response does not explicitly suggest how she has arrived at that response. We can only make suppositions about what has happened. However, the fact that a subject does not respond to the reasoning problem as the experimenter expects does not suffice to regard her response as normatively inappropriate. Nor can we hold that, given the subject's task interpretation, her reasoning performance will be always normatively appropriate. If we assume that the context of evaluation is determined also by the reasoner's own interpretation of the problem context, the appropriateness of a given reasoning performance would demand of the reasoner too little: it would demand of her that she behaves in accordance with her own interpretation of the situation but not that her understanding of the task displays a correct grasp of the situation. By relying on the two-step normative framework, we take into account both the legitimacy of the subject's task interpretation and the normative appropriateness of her response. In conformity with such a framework, when analyzing the subjects' final responses to a reasoning task, we are faced with four different possible cases: (i) the subject's task interpretation is legitimate and her response is normatively appropriate with respect to that interpretation; (ii) the subject's task interpretation is legitimate but her response is normatively inappropriate with respect to that interpretation; (iii) the subject's task interpretation is illegitimate but her response is normatively appropriate to that interpretation; (iii) the subject's task interpretation is illegitimate and her response is normatively inappropriate anyway. Let us consider these four cases in greater detail.
With regard to case (i), two sub-cases may be distinguished. In the former, the subject's response corresponds to what the experimenter regards as normatively appropriate under her interpretation of the task. So, also the subject's task interpretation is assumed to correspond to the experimenter's. Such a condition resembles that envisaged by evolutionary psychologists. According to them, every time we devise a reasoning problem in which the information available in the problem context is presented in an appropriate form and the problem itself recalls a problem from the environment of evolutionary adaptation, we prompt the subjects to frame the problem accordingly, which leads them to a normatively appropriate response. In such a case, on the basis of evolutionary considerations, we assume that both the subject's task interpretation and her response match those assumed by the experimenter. In the latter sub-case of (i), the subject's response diverges from that expected by the experimenter, but can nevertheless be regarded as normatively appropriate with respect to a legitimate task interpretation (different, in turn, from that assumed by the experimenter). Such a way of dealing with empirical data is open to two of the approaches to rationality assessment we have considered: the conversational approach and the consequentialist approach inspired by pragmatism. By adjusting the subject's task interpretation or, respectively, the goal with respect to which the reasoning performance is expected to be effective, these approaches are able to re-interpret any reasoning performance as rational. Indeed, we have seen that the former places too weak constraints on the subject' task interpretation, while the latter leads directly to relativism. If we rely on the notion of objective context at both steps of assessment, as the two-step normative framework proposes, several suggestions coming from these two trends of research can be accepted while avoiding their undesirable consequences.
As to case (ii), the experimenter assumes that the subjects have understood the problem presentation in the same way as she does and with the same background assumptions, so that she can assume that they have framed the reasoning problem in a legitimate way, but with respect to this interpretation of the problem, the experimenter assesses the subjects' responses as normatively inappropriate. Such a case corresponds to the one in which an experimenter compares the subjects' reasoning performance with a predetermined normative standard without taking into consideration what the subjects are really doing. Such a way of assessing the subjects' responses to reasoning tasks resembles the attitude taken by the supporters of the Irrationality Thesis. On their view, people understand correctly the reasoning problem but, in solving it, rely on heuristics which do not lead to good outcomes reliably enough.
Turning to (iii), the case in which the subjects' framing of the reasoning problem does not match the information available in the experimental context, but they may be held to use a rational reasoning strategy nevertheless, it is Oaksford and Chater who can be described as construing the experimental results in this way. In their view, contrary to the judgement of most psychologists, the most common response to the standard version of the selection task should be regarded as rational. In arguing for this claim, they have to assume that the subjects consider the selection task as a task of inductive hypothesis-testing. If it were so, however, the subjects would misinterpret the reasoning problem. Indeed, they have explicitly been told that the conditional rule applies only to the four cards in front of them, as opposed to some wider class of which these cards might be a sample. Oaksford and Chater hold that if we accept this task interpretation, we can admit that the subjects are using a rational reasoning strategy which leads most of the time to successful outcomes in ordinary life. In a case such that the one envisaged by Oaksford and Chater, I hold that the difficulty for subjects is not in their reasoning badly, but in a misperception of what the situational context is. In other words, they are using a rational reasoning strategy in the wrong situation.
Finally, there is case (iv). In such a case, both the subject's response and her interpretation of the task are inappropriate given the information available in the situational context. Even if the experimenter hypothetically assumes as appropriate the subject's personalized representation of the problem, she will conclude that the reasoning strategy the subject has used is inappropriate with respect to that interpretation of the problem. This case is, however, a mere combinatorial possibility. It seems to be very difficult to establish from the subject's response, assuming that it is inappropriate to her task interpretation, what her task interpretation was like and whether it was itself inappropriate. Ideally, such a position may be assumed to hold by the supporters of a radical Irrationality thesis. They might argue that people have to change their cognitive strategies if they want to achieve their goals. These changes should be concerned with both people's capacity to understand and frame reasoning problems and their ability to reason appropriately.
Coming to a conclusion, it seems to me that the supporters of the different approaches to rationality assessment that I have examined in the previous chapters have each a partial view of what could happen in a reasoning task. Although, as just seen, when assessing a reasoning performance, one is faced in principle with four different cases, they tend to consider only the one which fits best with their theoretical targets. They often do not care about the other possible cases. But, it seems to me, in assessing reasoning performances we are not allowed to reject any of the four cases a priori. As a consequence, the two-step normative framework presented here can be seen as more satisfactory, in that it enables us to conceive of all the four cases. Thus, the two-step normative framework of the context-sensitive consequentialist approach appears to have a potential for use in combination with traditional approaches such as the evolutionary and pragmatist ones in particular, towards an overarching framework for the assessment of human rationality. Understanding human reasoning and rationality depends upon how we conceive of the reasoning performance and its relation not only to the cognitive context, but also to the situational context wherein it occurs. A satisfactory analysis of this relationship provides reasoning studies with a fundamental challenge that could require extensive research and development for some years to come.