Ayoob Shahmoradi

Can a being think without ever having sensed anything? I argue that it cannot. Thinking requires mental representation, and representation requires some mechanism that generates it through contact with a domain — what I call a sensory system, broadly construed. Inferential transitions can transmit and transform representational content, but they cannot create it from scratch. This asymmetry entails that pure thinkers — beings with thought but no sensory capacities, as David Chalmers imagines — are impossible. The argument also constrains debates about AI: No amount of inferential sophistication can replace the sensory grounding that makes representation possible.

What does the most fundamental type of perceptual (or mental) reference look like? One view suggests that perceptual reference requires representation-as, while another holds that it requires discriminating the perceived item. I distinguish five types of discrimination, three of which are personal-level and distinctively visual, and explain their implications and interrelations. Next, I argue that the plausibility of the claim that perceiving something requires discriminating it—rather than simply attributing properties to it—depends on the type of discrimination at issue. A weak form of discrimination trivializes the debate, while stronger types of discrimination cannot be fully understood without incorporating representation-as. Thus, representation-as appears to form the more fundamental level of personal-level representation.

A theory of reference attributed to Frege, Russell, and others holds that referring to an object requires the ability to uniquely individuate it. According to a famous story told around campfires on winter nights, a group of young revolutionaries, led by Kripke and Donnellan, was destined to tear down the Frege–Russell edifice of reference—and indeed, they did. Reflecting the spirit of the 60s and 70s, this makes for a compelling tale. However, the truth is that what these young revolutionaries actually did was to merely suggest a type of reference largely ignored by others. The norms of success for this type of reference, constrained by its specific profile, cannot be generalized to all reference. Moreover, it presupposes a more fundamental kind of reference that lacks this distinctive profile.

I develop a theory of (mental) reference according to which there are two ways of referring to an object. One can refer to an object by relying on a previous instance of successful reference to that (or some other) object. I call this type of reference 'dependent reference'. Alternatively, one can refer to an object 'independently'. A referential attempt is independent if and only if it is not dependent. I argue that these two types of reference differ as they adhere to different norms. While it may be possible to think about an object without the ability to uniquely individuate it if one thinks about it dependently, the same may not hold true in the case of independent thought. At least, the classic arguments that attempt to demonstrate the possibility of thinking about something without the ability to uniquely individuate it are irrelevant to independent reference. Furthermore, not all reference could be dependent. Along the way, I do a few other things as well.

What does the most fundamental type of perceptual (or mental) reference look like? One view suggests that perceptual reference requires representation-as, while another holds that it requires discriminating the perceived item. I distinguish five types of discrimination, three of which are personal-level and distinctively visual, and explain their implications and interrelations. Next, I argue that the plausibility of the claim that perceiving something requires discriminating it—rather than simply attributing properties to it—depends on the type of discrimination at issue. A weak form of discrimination trivializes the debate, while stronger types of discrimination cannot be fully understood without incorporating representation-as. Thus, representation-as appears to form the more fundamental level of personal-level representation.

My general worry is that Schellenberg’s arguments against naive realism, generalism, and Russellian representationalism do not seem to be successful. Thus her attempt at ruling these views out fails. Her main arguments rely on a shared premise whose plausibility, in the absence of an appropriate theory of particulars, is hard to assess (§2.1). Apart from that, these arguments rely on an under-specified notion of constitution; there seems to be no sense of the term that makes all the premises of her major arguments true without trivializing their conclusions (§2.2). It also seems to me that the challenges that Schellenberg raises for her Russellian opponents, or similar problems, arise for her own view (§3). Finally, Schellenberg believes that an advantage of her view is that it entails the possibility of seeing an object without seeing it as being a certain way. In §4, I argue that it is hard to imagine how one could see without seeing as.

There are passages in Kant’s writings according to which empirical intuitions have to be (a) singular, (b) object-dependent, and (c) immediate. It has also been argued that empirical intuitions (d) are not truth-apt, and (e) need to provide the subject with a proof of the possibility of the cognized object. Having relied on one or another of the a-e constraints, the naïve realist readers of Kant have argued that it is not possible for empirical intuitions to be representations. Instead they have argued for a relationalist reading of empirical intuitions in terms of an acquaintance relation between the subject and the intuited object. For the sake of argument, I will grant the naïve realist reader of Kant that empirical intuitions should satisfy all the a-e constraints. Nevertheless, by incorporating these constraints, one by one, into a representationalist theory of empirical intuitions, I will show that not only doesn’t a naïve realist reading of empirical intuitions follow, but also that the naïve realist has hastily overlooked a range of perfectly representationalist readings of intuitions available to Kant and his representationalist allies. On the positive side, I will argue that there is an extra constraint on intuitions—i.e., that givenness does not require presence to consciousness—that directly goes against any naïve realist account of intuitions.