The problem of rationality
Helm argues that evaluative judgments that together with felt evaluations constitute a person's rational pattern of import must themselves be sincere and practical. Sincerity is here simply a technical matter of occupying a transitive role in the subject's chain of reasoning. Practicality, in turn, requires that an evaluative judgment must consistently motivate the subject to act in accordance with itself. No doubt, sincerity and practicality are important criteria for the rationality of evaluative judgments. Yet they are clearly insufficient as they stand as a simple thought experiment shows.
Consider an already slender anorexic who makes an evaluative judgment that she is fat and must therefore put herself to a strict diet. The anorexic may embrace Helm's requirement of sincerity for she is both able to justify this particular judgment in light of her broader evaluative framework, her skewed evaluative beliefs about herself and excessive concern for her weight, willing to use this judgment to justify further evaluative judgments. Furthermore, the subject is consistently motivated to follow her evaluative judgment to action as she is anxious to engage in dieting. If, moreover, the anorexic person's felt evaluations about the necessity of dieting cohere with her evaluative judgments on the same topic, we should on Helm's account conclude that dieting is a matter of positive import for the subject. However, this is an absurd conclusion because fierce dieting cannot rationally be a focus of positive import for a slender and anorexic person.
The previous example indicates that we must find some other criteria besides sincerity and practicality for the rationality of evaluative judgments. To begin with, we can divide evaluative judgments into their constituent parts, that is, beliefs and values. An evaluative judgment, for instance, that the bear I come across in the forest is dangerous can be rewritten as a belief or thought that the bear is capable of inflicting severe or fatal damage to my physical constitution, thus affecting my health and capacity to survive. Now we have two items for evaluation in terms of rationality instead of one. Firstly, I can appraise whether in fact the bear is capable of inflicting severe or fatal damage to my physical constitution. This may depend, for instance, on its distance from me, on my available weaponry, or on my estimated possibilities to escape the scene which, in turn, depend on my physical condition, fitness, and so on. Secondly, I can also evaluate the status of health and survival in my set of values, which is obviously quite high. In sum, an evaluative judgment is rational if its constitutive beliefs and values are rational. But what does this mean?
I suggest that both values and beliefs are warranted by internal criteria of rationality. Here we can follow the guidelines of de Sousa who distinguishes between the compatibility and consistency of mental states as standards of coherence. To rephrase, two states are compatible with each other if they can both exist together whereas they are consistent if the states of affairs represented in the intentional contents of those states are able to coexist. As evaluative attitudes with the world-to-mind direction of fit, two or more values can be held irrespective of their content. We can value all sorts of things as long as we do not have to worry about their joint realization or coexistence. Even so, we can increase the compatibility of values by Helm's criteria of coherence. Here valuing is a form of reflexive caring about one's objects of care and it manifests as a coherent and projectable pattern of such reflexive emotions as shame, remorse, pride, or self-approbation about one's firstorder evaluations, both felt and cool, whose projectable patterns constitute objects of care. If an agent's evaluative perspective is divided into several coherent subpatterns of first-order evaluations, projectable patterns of reflexive emotions may separate his values from mere cares. Thus for instance a former drug addict may crave for a fix, feel satisfied when he gets it, and angry at those who try to interfere with this drug use, but if the same agent at other times sincerely judges that he should quit, feels ashamed of his relapses, and proud of his periods of abstinence, then abstinence rather than drug use has value for the agent.
For beliefs with the mind-to-world direction of fit, the situation is somewhat different. It is psychologically possible to have inconsistent beliefs such as p and q that cannot be true to together as long as their incompatibility goes undetected. This may happen easily as people rarely think through all logical consequences of their beliefs some of which may be incompatible. Explicit logical contradictions such as p and not-p are dissimilar since it is psychologically impossible to believe p and not-p in a single act of assent. Nor does it seem possible to understand believing logical contradictions in terms of failing to confront those beliefs with one another, as de Sousa (2007, p. 327) observes. The psychological question is still different from that about rationality: having inconsistent beliefs is subjectively irrational whether or not it is psychologically possible to have such beliefs. Furthermore, subjectively rational beliefs must be warranted on the basis of all relevant evidence conceivably available to the subject at the time. By the condition of conceivable availability we can rule out evidence that is beyond the subject's reach due to historical, cultural, social, or other limitations. Rational beliefs need not be warranted in an externalist sense either. The subject must have reasonable grounds to presume that his or her beliefs have been produced by reliable causal processes. But whether this in fact is the case, is of secondary importance for subjective rationality.
The main problem for subjective rationality is then the consistency of values. Evaluative attitudes are consistent if the states of affairs represented in their contents are able to coexist. If I for instance value both health and wealth but must work like crazy for the latter, these values are not consistent if the means of getting rich inevitably involve such workload that damages my health permanently. The example indicates how two or more evaluative attitudes can be consistent with each other: on the condition that realizing the state of affairs represented in the content of one attitude does not defeat the possibility of realising states of affairs represented in the contents other evaluative attitudes either simultaneously or at some later point of time. Georg Henrik von Wright (1989) called this kind of justification of subjective values "value rationality". The point is to qualify an agent's choices of intrinsically valued ends with knowledge of the causal consequences and prerequisites that are connected with attaining those ends. A practical problem for a rational agent is how to anticipate the causal prerequisites and consequences of realising different states of affairs as our knowledge of such causal connections is necessarily limited. Even so, such knowledge may be all we have if there is no other way of evaluating the consistency of valued ends before pursuing them.
The suggested criteria of subjective rationality for beliefs and values are capable of denying the rationality of the anorexic person's favorable evaluative judgment about dieting. The judgment fails to qualify as rational because the subject has disconfirming evidence to her belief about her fatness available any time she steps on a scale or sees her slender image in a mirror even though she does not want to consider this evidence. Accordingly, the rational pattern of evaluative judgments and felt evaluations that warrants the person's reckless dieting is founded on and maintained by a systematic restriction and distortion of the subject's available evidence. For the same reason, the subject's positive valuation of fierce dieting fails to qualify as justified as it does not rest on her well-founded and consistent beliefs.