In On Reflection, Hilary Kornblith offers a great analysis of the use of reflection in four central philosophical areas. According to him, many philosophers overestimate reflection because their views about what reflection is and what it is capable of achieving are “terribly inaccurate” (p. 1). Kornblith looks at the usage of reflection in philosophy as a mystified human capacity, and he wants to demystify reflection by offering a naturalist and realist understanding of its powers and capacities.
On Reflection is divided into two parts. The first part, the most extensive, explores four areas where reflection is central: knowledge, reasoning, freedom, and normativity. In the second part (the final chapter), Kornblith presents an outline of a realist understanding of the powers of reflection.
A recurring logical problem with the use of reflection and reflective scrutiny in these four areas is the infinite regress: we should not take for granted the reliability of reflection; we should place under scrutiny reflection itself. For example, in epistemology, philosophers argue that reflection is a necessary condition for having justified beliefs or knowledge. Second-order beliefs must evaluate first-order beliefs (unreflective beliefs). We consider our first-order beliefs unreliable and this is why we must evaluate them with second-order beliefs. But it appears we have no good reason to stop the regress here: why not also evaluate second-order beliefs? If we defend an internalist understanding of knowledge, where first-order beliefs must be evaluated reflectively by second-order beliefs, it seems we are engaged in an infinite regress since we do not have good reason to consider second-order beliefs as reliable if we consider first-order beliefs to be unreliable.
A similar infinite regress looms when we involve reasons to believe something in the process of reasoning. For example, if I think that p is a reason to believe q, I must have a reason to believe p, another reason to believe this reason, etc. According to Kornblith, this means that reflection cannot play the role we think it can: it cannot improve the reliability of our first-order beliefs and give us a regress stopping reason to believe something. If we have doubts about first-order beliefs, we should have the same doubts about second-orders beliefs and reflective scrutiny.
In the four areas examined by Kornblith, he identifies empirical problems with the powers and capacities of reflection traditionally taken for granted in philosophy. One of these problems is a distinction between two kinds of knowledge: animal knowledge as distinct from reflective knowledge, as proposed by Ernest Sosa. According to Kornblith, despite the fact that human adults have more developed cognitive capacities, we have no good reasons to adopt this kind of distinction. Also, a distinction between animal and reflective knowledge seems to be in conflict with our best theories in developmental psychology.
Kornblith devotes the first two chapters to knowledge and reason, where he presents his critique of internalism in epistemology in focusing on logical and empirical problems. In the third and the fourth chapters, he deals with views involving reflection in areas outside of epistemology. I think it is important to say something about these other analysis by Kornblith since his book is not limited to a critique of internalism in epistemology.
In the third chapter, Kornblith attacks the conception of free will defended by Harry Frankfurt, where the free-willed agent acts on the basis of second-order desires and beliefs (reflective states) and not on the basis of first-order beliefs and desires (unreflective states). According to this conception of free will, first-order desires fail to reflect one’s true self, and an action must be done on the basis of second-order desires to manifest free will. Kornblith has doubts about higher-order states playing this important role. If we consider that first-order desires fail to reflect one’s true self, why would not we think that second-order desires similarly fail? Second-order desires, like first-order desires, are not self-legitimating; according to Kornblith, they do not have extra power that first-order desires do not.
In the fourth chapter, Kornblith critiques the possibility that reflection and reflective scrutiny are the source of normativity, as Korsgaard and constructivists defend. For Korsgaard, an agent could have a reason to act only if this reason is grounded in a self-conception formed by reflection. Kornblith expresses many doubts about the possibility that obligations and reasons flow from a self-conception formed under reflection by an agent. Kornblith gives the example of a Ku Klux Klan member for whom the Klan’s membership is a fundamental part of his identity. Obviously, membership in the Klan provides bad reasons for action, and everyone (except, presumably, for Klan members themselves) will agree that the relation between a self-conception and the normativity of reasons cannot be so direct. On this point, Kornblith seems to underestimate Kantian constructivism: according to this theory (and its interpretation given by Korsgaard) only the self-conception that respects our humanity can give unconditional obligations. Obviously, membership in the Klan is ineligible as a self-conception that respects our humanity.
In the fifth and final chapter, Kornblith proposes his account and conception of the powers of reflection. He presents, very schematically, a more realist view of the powers of reflection. According to Kornblith, the principal error of the philosophical tradition about reflection comes from viewing unreflective processes from a third-person perspective, while viewing reflective processes from a first-person perspective. Once this change in shift is recognized, the unreflective and reflective processes do not look different in kind. The difference between these processes is only due to a shift in perspective; it is not a real difference but instead only a difference “created” by the philosophical tradition. For Kornblith, “if we are to understand reflection for what it is, we must examine it from the third-person perspective” (p. 163). To have a more precise, scientific, and naturalist conception of reflection, we must understand second-order mental processes in the same way we understand first-order and unreflective mental processes, i.e., by understanding the complex mental mechanisms involved. Only from a third-person perspective can we demystify reflection. The literature about System 1 and System 2 may serve as a good start for more realistic research on the powers of reflection.
Straightforward and very accessible, On Reflection is a must-have for anyone interested in epistemology, free will, or normativity.
