Cushman has argued that rationalization is a form of representational exchange that has evolved because it is adaptive. Rationalization enables humans to change the format of information not initially encoded in propositional form into reasons, in the form of familiar constructs of folk psychology, such as beliefs and desires, that can positively affect future action. Cushman's view, if correct, has implications for classic and ongoing debates in the philosophy of mind about the metaphysical status of folk psychological states. There are three main views about the nature of folk psychology. Fodor (Reference Fodor1980) has famously espoused a realist view of psychology, arguing that beliefs and desires (and hopes, fears, intentions, etc.) exist as propositionally structured entities in the head, and thus entail the existence of a language of thought. The representational and formal properties of these language-like states correspond to each other in such a way that the causally efficacious syntactic properties make possible reason-respecting behavior. Churchland's (Reference Churchland1981) eliminativism is diametrically opposed to this Fodorian view: Churchland has argued that folk psychology is a mistaken and bankrupt theory, and that a better understanding of the brain will reveal that constructs such as beliefs and desires do not constitute natural kinds, that the representational structures that the brain uses are not language-like in any sense, but are rather far more articulated and richly multidimensional than anything that language can capture. Churchland used examples from connectionist research to illustrate his eliminativist arguments, and one might think that recent developments in the connectionist tradition, such as deep learning, may serve to vindicate his early arguments. Somewhere between these views lies the instrumentalism of Dennett (Reference Dennett1987), whose position is that we can usefully characterize some systems as having beliefs and desires by taking a certain interpretive strategy, the intentional stance. To the degree that we can predict and explain a system's behavior by imputing folk psychological states to that system, so we are warranted in that imputation. For Dennett, even simple systems such as thermostats are legitimate targets of the intentional stance, even though no standard realist would support the view that thermostats have mental states. Dennett is a realist in that he claims there are real patterns in behavior that can be captured by the intentional strategy, but he remains agnostic as to whether intentional states map onto causally efficacious natural kind states that follow natural laws.
Although these classic debates in the metaphysics of mind seem far from the concerns of Cushman, these issues are deeply intertwined. Cushman argues that human rationalization is a process by which information from adaptive yet non-rational (in the sense of non-propositional) computational processes that drive behavior can be transformed in ways that make it accessible to rational processes. This species of informational exchange, moreover, encompasses a range of human cognitive functions: We call it rationalization when applied to our own behavior; we call it theory of mind when applied to the behavior of others.
If indeed rationalization – the process of metacognitively reframing observed behavior in terms of beliefs and desires – evolved because it was adaptive, then it must be the case that at least sometimes rationalization is causally efficacious, and the effects of rationalization have been overall beneficial. This seems to weigh heavily on the realist side of things. It does not entail that all behavior-causing states are folk-psychological states, but it does seem to entail that at least some of the states that cause behavior are precisely those emerging from the process of rationalization, which a fortiori are folk-psychological states. If so, it seems that a strong eliminativist position cannot be correct. It also seems that if eliminativism is correct – that is, if there are no classes of mental states that map cleanly to the propositional attitudes – then even if there is an important phenomenon of representational exchange, then whatever rationalizing amounts to, the propositional attitudes it yields cannot be the underlying causes of subsequent behavior. And if Cushman is correct, even the instrumentalist view seems overly quietist, for why favor a theory that refrains from speaking the truth?
While Cushman may be content to bite the bullet on metaphysical matters, there is nonetheless a tension in his argument. Cushman characterizes rationalization as (usually) a fiction, as it sees beliefs and desires as causes of our behavior where, by hypothesis, there are none. However, it is a useful fiction as it leads us to adjust our beliefs and desires to correspond to the fictional attributions, so that in subsequent action, we are driven by them. As a one-off account, this seems fine. But if rationalizing is as prevalent as Cushman indicates, and the propositional attitude-like outputs of the process affect behavior, then there is a constant looping: results of rationalization become causes of future actions, and rationalizations about these and subsequent actions do seem likely to identify the propositional attitudes that were causal, and thus they are not, by and large, fictions, but rather true descriptions.
If this tension makes Cushman's position seem unstable, one could find more plausible a slightly different story that nonetheless accepts the interesting parallelism between theory of mind and rationalizing. If social pressures were primary drivers of human evolution, as many have argued, one might think that mentalizing would be the primary adaptive phenomenon, and rationalizing may be the result of turning that adaptive tool upon oneself. On this version of the story, rationalizing could be a spandrel, and thus the story would not entail that rationalizing is rational or even beneficial (though it does not rule out the possibility). While the upshot for theorizing about representational exchange is unclear, on this alternative view, our ability to extract propositional attitude explanations from observing others’ behavior does not entail realism about propositional attitudes. Indeed, it may be most congenial to a Dennettian view, for unlike Cushman's story, this story highlights the value of the intentional stance, rather than the underlying causal machinery of propositionally structured reasons.
You have
Access
Cushman has argued that rationalization is a form of representational exchange that has evolved because it is adaptive. Rationalization enables humans to change the format of information not initially encoded in propositional form into reasons, in the form of familiar constructs of folk psychology, such as beliefs and desires, that can positively affect future action. Cushman's view, if correct, has implications for classic and ongoing debates in the philosophy of mind about the metaphysical status of folk psychological states. There are three main views about the nature of folk psychology. Fodor (Reference Fodor1980) has famously espoused a realist view of psychology, arguing that beliefs and desires (and hopes, fears, intentions, etc.) exist as propositionally structured entities in the head, and thus entail the existence of a language of thought. The representational and formal properties of these language-like states correspond to each other in such a way that the causally efficacious syntactic properties make possible reason-respecting behavior. Churchland's (Reference Churchland1981) eliminativism is diametrically opposed to this Fodorian view: Churchland has argued that folk psychology is a mistaken and bankrupt theory, and that a better understanding of the brain will reveal that constructs such as beliefs and desires do not constitute natural kinds, that the representational structures that the brain uses are not language-like in any sense, but are rather far more articulated and richly multidimensional than anything that language can capture. Churchland used examples from connectionist research to illustrate his eliminativist arguments, and one might think that recent developments in the connectionist tradition, such as deep learning, may serve to vindicate his early arguments. Somewhere between these views lies the instrumentalism of Dennett (Reference Dennett1987), whose position is that we can usefully characterize some systems as having beliefs and desires by taking a certain interpretive strategy, the intentional stance. To the degree that we can predict and explain a system's behavior by imputing folk psychological states to that system, so we are warranted in that imputation. For Dennett, even simple systems such as thermostats are legitimate targets of the intentional stance, even though no standard realist would support the view that thermostats have mental states. Dennett is a realist in that he claims there are real patterns in behavior that can be captured by the intentional strategy, but he remains agnostic as to whether intentional states map onto causally efficacious natural kind states that follow natural laws.
Although these classic debates in the metaphysics of mind seem far from the concerns of Cushman, these issues are deeply intertwined. Cushman argues that human rationalization is a process by which information from adaptive yet non-rational (in the sense of non-propositional) computational processes that drive behavior can be transformed in ways that make it accessible to rational processes. This species of informational exchange, moreover, encompasses a range of human cognitive functions: We call it rationalization when applied to our own behavior; we call it theory of mind when applied to the behavior of others.
If indeed rationalization – the process of metacognitively reframing observed behavior in terms of beliefs and desires – evolved because it was adaptive, then it must be the case that at least sometimes rationalization is causally efficacious, and the effects of rationalization have been overall beneficial. This seems to weigh heavily on the realist side of things. It does not entail that all behavior-causing states are folk-psychological states, but it does seem to entail that at least some of the states that cause behavior are precisely those emerging from the process of rationalization, which a fortiori are folk-psychological states. If so, it seems that a strong eliminativist position cannot be correct. It also seems that if eliminativism is correct – that is, if there are no classes of mental states that map cleanly to the propositional attitudes – then even if there is an important phenomenon of representational exchange, then whatever rationalizing amounts to, the propositional attitudes it yields cannot be the underlying causes of subsequent behavior. And if Cushman is correct, even the instrumentalist view seems overly quietist, for why favor a theory that refrains from speaking the truth?
While Cushman may be content to bite the bullet on metaphysical matters, there is nonetheless a tension in his argument. Cushman characterizes rationalization as (usually) a fiction, as it sees beliefs and desires as causes of our behavior where, by hypothesis, there are none. However, it is a useful fiction as it leads us to adjust our beliefs and desires to correspond to the fictional attributions, so that in subsequent action, we are driven by them. As a one-off account, this seems fine. But if rationalizing is as prevalent as Cushman indicates, and the propositional attitude-like outputs of the process affect behavior, then there is a constant looping: results of rationalization become causes of future actions, and rationalizations about these and subsequent actions do seem likely to identify the propositional attitudes that were causal, and thus they are not, by and large, fictions, but rather true descriptions.
If this tension makes Cushman's position seem unstable, one could find more plausible a slightly different story that nonetheless accepts the interesting parallelism between theory of mind and rationalizing. If social pressures were primary drivers of human evolution, as many have argued, one might think that mentalizing would be the primary adaptive phenomenon, and rationalizing may be the result of turning that adaptive tool upon oneself. On this version of the story, rationalizing could be a spandrel, and thus the story would not entail that rationalizing is rational or even beneficial (though it does not rule out the possibility). While the upshot for theorizing about representational exchange is unclear, on this alternative view, our ability to extract propositional attitude explanations from observing others’ behavior does not entail realism about propositional attitudes. Indeed, it may be most congenial to a Dennettian view, for unlike Cushman's story, this story highlights the value of the intentional stance, rather than the underlying causal machinery of propositionally structured reasons.
Acknowledgments
Thanks to Brian Fiala and Jonathan Phillips for helpful discussions.