Robert Hanna’s new book is, like his previous work, ambitious, systematic and compelling. It is an eminently worthwhile read for anyone interested in a Kantian account of cognition or contemporary theories of representation and/or the a priori that will require but also reward close reading. Hanna’s target this time is the a priori, and he confronts this topic via his usual approach: his own Kant-inspired theory of non-conceptual representation. Early parts of the book review material familiar to anyone acquainted with Hanna’s past work and bring up to speed anyone who is not. From there, the remainder of the book consists of a deft, thorough and insightful application of this theory to all matters a priori, including but not limited to a ‘radically naïve realism’ in epistemology, a revival of the analytic–synthetic distinction, an account of the nature of logic, and three chapters detailing the solution to Hanna’s extended and generalized versions of Benacerraf’s dilemma regarding the impossibility of properly grounding a priori knowledge. Spoiler alert: in all of these cases, it turns out that non-conceptual content as Hanna understands it is the key to making sense of these previously intractable philosophical problems.
Given the space constraints of this review, I will let Hanna speak for himself on the content of the later chapters to whet the appetite of those curious about their main theses, before returning to the earlier more foundational ones for a bit of critique. On the analytic–synthetic distinction:
The distinction, I have argued, is based on the primitive, irreducible facts of conscious intentionality, mental content – including both conceptual content and autonomous essentially non-conceptual content – and human cognitive and practical rationality. (p. 228)
On the nature of logic:
The pure formal science of logic thus ultimately converges with Kantian ethics, as I understand it, and it thereby inherently expresses the morality of logic. Otherwise put, I am saying that the pure formal science of logic is at bottom a moral science, not a natural science. (p. 232)
And on the solution to the problem of grounding a priori knowledge:
What we require is a linguistically competent, healthy, developmentally normal, and (relatively) mature rational human animal, who can grasp both the autonomous essentially non-conceptual content of perception and also the conceptual and propositional content of statements or judgments, who has also learned the basics of basic arithmetic or Primitive Recursive Arithmetic, as well as the basics of basic pure general logic, and who is thus primed and ready for speaking her own natural language, and for non-conceptually and pre-reflectively or first-order consciously, but also conceptually and self-consciously, intaking her manifestly real world through direct, veridical sense perception. And that is all that is required. For she is thereby capable of performing High-Bar justified objectively necessarily true basic authoritative rational intuitions in mathematics and logic, and thus capable of achieving High-Bar objective a priori knowledge according to the highest and categorically normative principles of theoretical and practical rationality. (p. 440)
As Hanna points out, if his theory of non-conceptual content can achieve all this, then there is an inference to the best philosophical explanation the conclusion of which is that we ought to accept that theory. That inference will only be valid, though, if the theory itself is sufficiently articulated and prima facie tenable, and what I want to do now is pause to consider some reasons for thinking that it is not. We can begin with the nature of non-conceptual content itself. Hanna gives the following six(!)-part definition.Footnote 1
X is an essentially non-conceptual content of perceptions if and only if X is a mental content such that:
i. X is not a conceptual content, as defined by The Logical Cognitivist Theory of Concepts.
ii. X is included in a mental state, act or process that directly refers to some or another causally efficacious, actual individual macroscopic material being B in the local or distal natural environment of the minded animal subject of X.
iii. B can be identical to the minded animal whose mental state, act or process includes X – that is, X can directly self-refer.
iv. X thereby both uniquely (if not always perfectly accurately) tracks B’s thermodynamically asymmetric and temporally irreversible causal activities in time.
v. This location and tracking is undertaken in order to individuate, normatively guide, and informationally mediate the subject’s conscious intentional desire-driven body movements for the purposes of cognitive and practical intentional agency.
vi. X is an inherently context-sensitive, egocentric or first-person-perspectival, spatiotemporally structured content that is not ineffable, but instead is shareable or communicable in a special way. (pp. 69–70)
Given the importance of this thesis for everything that follows, and its vast and intricate complexity, it is surprising to find that Hanna offers a just a single positive argument in support of the entire edifice: a version of Kant’s argument concerning incongruent counterparts.Footnote 2 Hanna’s version of that argument is careful and thorough, but for present purposes, a quick summary will do.
1. We can represent the difference between objects that are qualitatively identical.
2. Using only concepts, we could not represent the difference between objects that are qualitatively identical.
3. Therefore, we must represent the difference between objects that are qualitatively identical using something other than concepts. (1, 2)
As should be obvious, this argument certainly does not support everything in Hanna’s six-part content non-conceptualism. For example, the argument from incongruent counterparts says absolutely nothing about ‘desire-driven body movements’, ‘practical intentional agency’ or that which is ‘shareable or communicable’. Furthermore, the connections to what is, for example, causally efficacious, macroscopic and thermodynamically asymmetric are blanks left to be filled in more than they are clearly drawn from the argument itself. In fairness, Hanna does spend much of chapter 2 making his thinking about these further theses explicit, but even there the arguments are more suggestive than they are rigorous.
Allow me to take one example. Included in (ii) above is the condition that a non-conceptual mental representation ‘directly refers to some or another causally efficacious, actual individual, macroscopic material being’. The nit I want to pick here concerns Hanna’s claim that a non-conceptual representation will refer to something that is both causally efficacious and macroscopic. The reason that I want to pause over that conjunctive claim is that, while Hanna bills his version of non-conceptualism as having a distinctly Kantian provenance, this particular thesis is one that Kant explicitly eschews. That is, as I have argued elsewhere, Kant is a scientific realist who holds that it is not macroscopic or ‘manifest’ objects that are causally efficacious at all, but rather only those that are the object of our successful theoretical-explanatory posits (Landy 2015: ch. 5). Here is a quick and dirty argument for that conclusion:
1. That which is (as such) subject to causal laws is the real in appearance. (A205/B250)
2. Substance is the real in appearance. (A181/B255)
3. Substance can never be either created or destroyed. (A182/B224)
4. Macroscopic or manifest objects can be created or destroyed. (A185/B228)
5. Therefore, macroscopic or manifest objects are not substance. (3, 4)
6. Therefore, macroscopic or manifest objects are not the real in appearance. (2, 5)
7. Therefore, macroscopic or manifest objects are not (as such) subject to causal laws. (1,6)
If something like this is right, then much of the foundation of Hanna’s account of non-conceptual content slips out from under him. Hanna makes a very big deal of the fact that it is manifestly real objects that are the content of non-conceptual representations, and that this puts us essentially embodied human beings in what he calls The Grip of the Given, a state in which we track such objects with our ‘intentional desire-driven body movements for the purposes of cognitive and practical intentional agency’. That is, it is by tracking causally efficacious manifest objects that we act as agents in the world. So if what is causally efficacious are actually microscopic objects, then our intentional actions will end up directed at theoretical rather than manifest entities, which clearly cannot be right. We desire food qua food, not qua disturbances in the Higgs field.
The point here is not primarily that Hanna is wrong about what is causally efficacious, but rather that such an example – of the running together of two distinct theses, each of which requires its own defence and the conjunction of which requires a demonstration of compatibility – is representative of the kind of detail that is an unfortunate casualty of his (justified and exciting) grand ambitions in the book. As Hanna demonstrates, his theory of non-conceptual content can do a great deal of philosophical lifting. It is a powerful theory. But too often, at least for my tastes, I found the strides that Hanna takes in drawing conclusions from that theory, or applying it to this or that bit of long-standing dialectic, too broad. My guess is that, like myself, where readers agree with his conclusions, these holes will be filled in by the reader’s own arguments, but where readers disagree with those conclusions, these will appear to be frustrating omissions.
That said, one of the places that I found myself balking is at the issue of what precisely the content of a non-conceptual representation is and how the content of such a representation interacts with other kinds of content, most importantly the conceptual content that constitutes our beliefs, desires, etc. Of course, the answers to these questions are crucially important to completing Hanna’s system, but I found his treatment of them ultimately unsatisfactory. Hanna declares his main theses in a myriad of ways throughout the book. Here is one statement of his take on the content of a non-conceptual representation:
the sensed properties and relations of individual causally efficacious macroscopic material beings in the local or distal natural spatio-temporal environments of conscious animals partially constitute veridical perceptual acts or states, by partially constituting the subjective experiential mental content of those perceptual acts or states. (p. 116)
So Hanna’s theory is that a non-conceptual representation has as its content the properties and relations of that object the properties and relations of which cause the representation to have the content that it has.Footnote 3 This relation goes by various monikers throughout the book depending on what Hanna’s focus is. For example, in chapter 2, where the non-conceptuality of such representations is the focus, it is The Grip of the Given; in chapter 3, where the epistemological role of such representations is at the fore, it becomes The Veridicality Relation.Footnote 4 In any case, this relation is the core of Hanna’s theoretical apparatus, and as such there are a number of questions that one might have about it. For example, how do such representations normatively interact with the conceptual representations that Hanna agrees are at play in belief, inference, etc.? Are they the mere causes of such representations? Are they their matter? Can they stand as justifiers to them? Hanna’s answer to this last question is that non-conceptual representations do stand in justificatory relations to conceptual ones:
autonomous-essentially-non-conceptually-accessed reasons can justify beliefs, choices, or actions directly or non-inferentially when a manifest fact is simply delivered, demonstrated, or presented to the relevant essentially embodied capacities of a rational agent . . . More precisely, autonomous essentially non-conceptual content is presupposed by all rational conceptual/propositional content whatsoever, and thus it is inherently proto-rational. (p. 108)
It is not at all clear (at least to me) how this answer is supposed to go. For all its complexity, for example, the first sentence simply states that non-conceptual representations justify conceptual representation non-inferentially. That much simply has to be true. The question is how that is possible. What precisely is the relation of a non-conceptual representation to a conceptual one such that the former in some way justifies the latter? The second sentence appears to answer that question, and that answer appears to depend in large part on the meaning of ‘proto-rational’.Footnote 5 As Hanna sees it, a non-conceptual representation is proto-rational insofar as it provides what Hanna calls non-conceptual, low-bar, context-sensitive, causally reliable knowledge (pp. 22–3). He further holds that this this is genuine knowledge because it is essentially reliable, it exercises certain epistemic capabilities, and its phenomenology is maximally evidential (p. 21). It is the second and third conditions here that appear to make non-conceptual representation knowledge as opposed to mere causal reliability, and that is due to their inclusion of these ‘epistemic capabilities’ and this ‘evidential’ relation. So, once again, in order to assess Hanna’s claim, we need to know what Hanna means by those phrases. The epistemic capability he references here is that of the capacity for animals and humans alike to have knowledge by acquaintance, or to be capable of being in certain states that are the product of a context-sensitive causally reliable cognitive mechanism (p. 21). The evidential relation is that between a piece of knowledge by acquaintance and its object. Thus, as far as I can tell, the proto-rationality of non-conceptual content amounts to a non-conceptual representation’s being the product of a cognitive process that is reliable at tracking certain worldly objects. As Hanna notes, this puts such representations in ‘the space of reasons’ only in an extended sense. My concern is that for all that Hanna has written I do not understand what this extended sense is.Footnote 6
Ultimately, I agree with Hanna that a properly Kantian account of cognition makes it a necessary condition of our conceptual representation of the world that we deploy a set of mental proxies the features of which are caused by and stand in for the features of the objects of such conceptual representations (cf. Landy 2015: ch. 3). What I think Hanna’s account is missing, however, is a solution to the (long-standing and difficult) puzzle of how such states can be both non-conceptual and normatively governed or justificatory. Without that solution, however, it is not clear that Hanna’s otherwise powerful, wide-ranging and systematic programme gets off the ground.
