Henry E. Allison (1937–2023)

Henry E. Allison has been, with Graham Bird, the most important Kant scholar of the last hundred years. This is a bold statement that needs qualification and, most importantly, justification. The qualification is that I say ‘most important’ and not ‘most impactful’ because others may have been more cited. I have not checked, I am not interested, and I am ready to concede the point for the sake of the argument. But Allison, like Bird before him, has been ‘most important’ because he reconstructed a Kant that still today constitutes a distinct and compelling philosophical option irreducible to anything on the philosophical menu of the last and present century. And here comes the necessity for a justification. Allison made Kant a powerful alternative to the dominant trend of contemporary philosophy that is still arguably one or another version of naturalism, as opposed to something that can come to terms and be somehow made compatible with it. This is the thesis that I am going to defend, because I believe that this is the best way of honouring Allison, both as a scholar and as my mentor.

In the reading first presented in Kant’s Transcendental Idealism. An Interpretation and Defense, the book published in 1985 that gave Allison global recognition, the leading idea was a rather orthodox one: Kant’s Copernican revolution was the discovery that our knowledge is subject to certain conditions that enable it and yet at the same time ‘inform’ it with a mark of subjectivity (Allison 2004). The essence of transcendental idealism is the recognition of this subjective mark and the consequent restriction of our knowledge to objects ‘as they appear to us’. The thesis, however, goes hand in hand with empirical realism, that is, the thought that our knowledge is about objects that exist outside of our mind. And self-knowledge is no exception. We know ourselves as objects still ‘outside of the mind’, although obviously not in a spatial sense, but in the sense that our internal states are ‘turned into’ objects by our mind which as such remains distinct from the objects themselves. This idea was further developed with two sub-theses only one of which, it is my impression, has been fully grasped by other interpreters. The well-known sub-thesis is the two perspectives interpretation of the distinction between appearances and things in themselves. Because we can make experience of the world and of ourselves only through these conditions, objects will by necessity be as they appear to us. But once we realize that, it is equally necessary to think that the same objects we know through these conditions could be known either through different conditions by knowers with a cognitive apparatus different than ours or through no conditions whatsoever, as we could think God knows its own creation (and perhaps itself). Obviously these objects are ‘the same’ in a rather metaphorical sense because the idea is not that there is a tree as an appearance and ‘the same tree’ as a thing in itself, simply because we do not know how the world would ‘appear’ to these other knowers. They might have forms leading them to put together trees and rocks in a single concept, instead of differentiating between them. For all we know, these other knowers could not even ‘notice’ the object that we know as ‘tree’ because it is fully irrelevant in its distinctness. What is rather clear, however, is that, no matter how metaphorical, this sameness rules out two possibilities: (1) appearances and things in themselves should not, as most Kant interpreters before and after Bird and Allison thought, especially in the Anglo-American world, be considered, respectively, as mental entities and ‘real things existing outside of our mind’; this would make Kant a horrific combination of Berkeley and Descartes, with the famous synthesis of the rationalist and empiricist school brought about by the critical philosophy consisting in combining phenomenalism and a dogmatic affirmation of the existence of a set of things that ‘must be there’ even if we have no immediate access to them. In addition, (2) regardless of how much Kant’s practical philosophy needs a description of ourselves alternative to the one offered by science – otherwise, we could hardly consider ourselves free in the strong sense of ‘first causes’ of a chain of events, let alone in the even stronger sense of free as autonomous, required for moral responsibility – what is needed (and sufficient for the purposes of the theory) is only the possibility of an alternative description. What is needed is the possibility of conceiving of ourselves as free without contradiction, as opposed to the possibility of thinking that for each single one of us there is some sort of avatar in the noumenal world that, magically, does things that have repercussions on their phenomenal twins.

The second, less known yet fundamental sub-thesis is that if our knowledge is enabled (and informed) by certain epistemic conditions, then these cannot be considered as mere ‘general frameworks’ that can be overcome by new frameworks suggested by the progress of science. If Allison is right in interpreting Kant as warning us that our knowledge is irremediably linked to a set of epistemic conditions, then we need to take seriously the idea that no matter how more refined our knowledge will be it will always be ‘informed’ by some conditions. This means that we may change our (philosophical) theory about what these conditions are, by proposing conditions others than the ones Kant spells out in the Transcendental Aesthetic and Analytic, but we cannot expect science to ‘free’ us, even at the end of times, from the subjective mark arising from our contribution in the shaping of the objects of experience. The possibly evolving theory of what the ‘real’ conditions of our knowledge are is the specific field of philosophy, a field that may not be usurped even by the most developed science. The question related to the conditions of our knowledge is not a question that science will ever be equipped to answer. This is why, at least so interpreted, Kant is offering an alternative to naturalism. And in my opinion, this is why Allison is, with Bird, the most important interpreter of Kant of the last hundred years. He is so simply because he rendered for us a Kant that we can use in case we want to resist the dominant trend in contemporary philosophy.

This is, however, still a hypothesis more than a justification of our bold claim. Perhaps the significance of Allison’s work will be more visible if we ask two questions: (a) quite directly, how was Kant interpreted, especially in the Anglo-American world, before Allison’s Kant’s Transcendental Idealism?; (b) how did Kant contribute to the philosophy of the last century and would this contribution have been possible if the Anglo-American world had not ‘learned’ the Kant interpreted and defended by Allison, but some less ‘confrontational’ Kant, pruned from its ‘metaphysical excesses’ of transcendental idealism, from its stubborn attachment to an incompatibilist view regarding human freedom (Allison 1990, 1996) and to a view of morality as ultimately grounded in a form of freedom – autonomy – even higher than practical freedom, a faculty that by itself would already make a scientific description of human beings helplessly insufficient (Allison 2011, 2020)?

The answers to these two questions overlap to a certain extent but let us start from the second issue. So, how did Kant contribute to today’s epistemology? Answers legitimately vary if one looks at specific critical themes, such as the role of each cognitive faculty in the constitution of the object of experience; the role of reflective judgment in the sciences; the unity of science; the role of representation, consciousness, imagination; the account of causality; the role of constitutive and regulative principles; and so on. But if one were to identify the Kantian theme that has contributed most to the contemporary epistemological debate, it would be quite safe to say – and this could hardly be a surprise – that this is the possibility of synthetic a priori knowledge, with the strictly related theme of the possibility of a very special subset of synthetic a priori knowledge, which Kant labeled ‘transcendental’, that concerns itself ‘not so much with objects but rather with our mode of cognition of objects insofar as this is to be possible a priori’ (B25). ¹ The debate on synthetic a priori knowledge occupied, from logical empiricism on, much theoretical philosophy of the last century, laying the foundation not only of analytic philosophy but also of the tradition now perceived as essentially opposed to it – phenomenology (Hanna 2008). Most importantly, the stance philosophers take today on the viability and nature of this kind of knowledge (and of its transcendental subset) is going to affect decisively their position in regard to naturalism, arguably the philosophical orientation still dominant in our times. If one believes transcendental knowledge is possible, then philosophy has its own specific domain, and science, even the most advanced, will not be able to provide all theoretical answers, as methodological naturalism holds. If one denies the viability of transcendental philosophy, then the only alternative to naturalism seems to be some form of scepticism. Indeed, Kant’s epistemology is still here to remind us that there is an alternative, a third way between thinking that humans are denied access to objective, universally valid knowledge and the belief that in science, one must find the answers to all legitimate questions, including those of epistemology itself. Even a cursory look at the recent history of epistemology will suffice to prove that this is the case.

While the effort to harmonize Kant with the evolution of the exact sciences before Einstein’s relativity had been the central concern of Neo-Kantianism, the general epistemological orientation after relativity was largely inspired by and oriented toward a complete rejection of synthetic a priori knowledge or to a more or less profound reform of it. Logical empiricism was indeed grounded on the belief that all necessary a priori truth was analytic truth, while all ‘ampliative’, synthetic knowledge was necessarily a posteriori. The logical empiricists most respectful of the Kantian heritage, for example, Reichenbach in his early work The Theory of Relativity and A Priori Knowledge (Reichenbach 1920), attempted to reform rather than reject the critical philosophy by forging a conception of the ‘relativized a priori’ turning on the idea that constitutive scientific principles can be both a priori and revisable (de Boer 2011).

In the second half of the century, at least in the Anglo-American world, the attempt to isolate in Kant a more general philosophical core no longer dependent on the truth of the mathematics and physics of his time continued, followed by a large debate on the nature and viability of transcendental arguments (Stroud 1968). In the hugely influential The Bounds of Sense (1966), Strawson suggested to search for a doctrine of the ‘necessary conditions of the possibility of experience’ more general than a theory about the necessary presuppositions of Euclidian geometry and Newtonian physics and independent of transcendental idealism, considered helplessly dogmatic and, as such, unworthy of being salvaged. The resulting ‘metaphysics of experience’ would thus identify necessary features of anything that we can intelligibly consider experience, such as the reliance of our self-knowledge on some (relatively) permanent objects in the external world, a commitment to the unity of the spatiotemporal world in which any experience has to find its place and to certain very abstract (yet not merely logical) principles such as permanence and causality. Strawson’s ‘descriptive metaphysics’, however, had no room for genuine synthetic a priori knowledge, let alone for transcendental knowledge as Kant intended it. The necessary features of our experience are not conceived as elements the knowing subject ‘lends’ to the objects of experience. Certainly, these features are not to be found in sensations; hence, they are a crucial component of experience no empiricist could account for. And yet they are not ‘given’ by the knowing subject either, as the transcendental idealist would say. They are just components of a conceptual scheme without which experience would not be intelligible for us. The task of philosophy is to identify these features and articulate the necessary connections between them (Strawson 1966: 22–3). Moreover, echoing Reichenbach’s notion of a relativized and historicized a priori (Reichenbach 1920), Strawson thinks that these conceptual features may vary as science proceeds: ‘They are not static schemes, but allow of that indefinite refinement, correction, and extension that accompany the advance of science and the development of social forms’ (Strawson 1966: 44). By so doing, Strawson was opening to a form of soft methodological naturalism (to be distinguished from the hard type championed by Quine) in which the only thing unchangeable in our knowledge, and as such not influenced by scientific progress, were logical necessary truths.

Michael Friedman, in this vein, thinks that the ultimate point of the Critique is to explain how Euclidean geometry and Newtonian physics are possible (Friedman and Bird 1998: 122). Thus, ‘Kant articulated and contextualized the Newtonian framework’ and ‘the logical empiricists did the same for the Einsteinian’ (Friedman and Bird 1998: 129). From this perspective, vindicating a central role for Kant in contemporary epistemology is rather hard, because much of Kant’s philosophy would be inseparable from a scientific worldview by now surpassed. Similarly, Kant’s anti-naturalistic resources are weakened, if not lost completely. Philosophy is no longer distinguished from science in the sense that the latter is about objects while the former is about ‘our mode of cognition of objects’ (B25) but merely in the sense that they operate at two different levels of generality about the same set of objects. Philosophy articulates and contextualizes the paradigms or constitutive frameworks that science uses and takes for granted during its ‘normal’ (in the Kuhnian sense) periods. Hence, its task today is evidently that of articulating and letting emerge the specific framework contemporary science employs or, if one considers the status of ‘permanent revolution’ physics offers, the numerous frameworks all competing revolutionary theories assume. Friedman thinks that this does not amount to a complete surrender to naturalism because the philosophical articulation is not, as Quine would have it, a particular application of science to the problem of identifying its general framework, and its result would not be just one extra piece to be added to the web of beliefs that holistically constitutes knowledge. It would still be ‘transcendental’ philosophy because we aim for the identification of the conceptual necessary conditions of (contemporary) science.

One may obviously have all sorts of doubts about this attempt to reconfigure transcendental philosophy. If its role is that of articulating the necessary conditions of contemporary science, transcendental philosophy is rendered fully dependent on what science does and where it is going. It is, notwithstanding all efforts by Friedman, nothing but the science of the main tacit assumptions of today’s science. The whole project of reforming metaphysics to set it on a secure path is gone, and the very expression ‘metaphysics of experience’ has acquired a distinctly un-Kantian undertone. Other scholars are more optimistic as to the possibility of separating Kant from any specific scientific worldview. Graham Bird, for example, thinks, correctly in my opinion, that Kant’s transcendental philosophy defines the general conditions of experience and only derivatively explains how those conditions can make sense of a specific science, which of course Kant considered certain, definitive, and immune from future falsification or reform as his transcendental philosophy (Friedman and Bird 1998: 131–51). The Metaphysical Foundations of Natural Science may be about the general articulation Friedman thinks is all one can hope from philosophy, but the Critique of Pure Reason operates at a different, more general level. Even in the Analytic of Principles, where Kant seems to be closest to a concern with the justification of the basic principles of Newtonian mechanics, he makes clear that his focus is rather on more general principles that enable the scientific (Newtonian) principles: ‘We treat only of the principles of pure understanding in their relation to inner sense … It is through these principles of the pure understanding that the special principles of mathematics and physical dynamics become possible’ (B201-2). Similarly, Kant establishes a fundamental difference between the metaphysical exposition of the concept of space (and time) and the transcendental exposition of the same concept(s). The former is fully independent of Euclidean geometry; the latter, which as Kant says is about the ‘explanation of a concept as a principle from which insight into the possibility of other synthetic a priori cognitions can be gained’ (B40), is about how one can make sense of the synthetic a priori character of that specific geometry. Perhaps even more forcefully, in the B Introduction, Kant is quite explicit that the question ‘How are synthetic judgments a priori possible?’ is more general and more fundamental than the questions about the a priori status of the sciences. As Kant puts it, ‘[i]n the solution of the above problem there is at the same time contained … the solution to the questions: “How is pure mathematics possible? How is pure natural science possible?”’ (B20).

Obviously, even if a part of Kant’s theory of knowledge can be separated from the sciences of his time, this does not mean that the redeemed part is compatible with the sciences of our time. Certainly, a rather meagre consolation would be to say that the critical philosophy is merely about the manifest aspect of reality, that is, about the way we perceive the world in our everyday experience and has no ambition to account also for the way in which science tells us the world is. This solution reduces the critical philosophy to the analysis of the conditions of the possibility not of experience but of something very close to what Kant calls illusion (Schein) or, at best, to secondary qualities.

The very project of a critical philosophy, and this was the point dearest to Allison, stands or falls with the possibility of identifying in the objects of experience, all of them, including those described and analysed by the most advanced sciences, very general features that can only be accounted for by taking into consideration the presence and influence on the final cognitive product of a finite knowing subject. Kant had his own set of transcendental conditions, and he may have been wrong about them. Nothing major would happen to the critical philosophy if we were to discover, through the progress of science or thanks to some philosophical argument, that these conditions are different, slightly or significantly, from those Kant identified. But a Kantian should not expect that these conditions are merely conceptual hidden assumptions embedded and fully reducible to the scientific picture of contemporary science (Friedman’s quasi-naturalistic suggestion). By presenting us with new objects of experience, scientific progress may show the necessity to revise what we thought were the true conditions. But the intellectual tension that guides the search for what Allison called epistemic conditions (Allison 2004), that is, conditions of experience in general, not of the most recent science, should never be abandoned if one wants to remain faithful to the spirit of the critical philosophy. This would be disproved in its very essence, by scientific progress or by some other source, if there were compelling reasons to think that it is open to human cognizers to know the world (cognizers included) as if they had no influence whatsoever on the cognitive product – as if, to use the common metaphor, we could experience the world with a view from nowhere. This would evidently be the triumph of transcendental realism and of naturalism, which is nothing but the former’s contemporary face. We go back to the point from which we started. Probably the most significant contribution the Kant reconstructed by Allison had to offer to contemporary theory of knowledge is a clear alternative to naturalism. This contribution does come at a price that many (Kantians included) find too high to pay: the acceptance of transcendental idealism, at least of the very minimal quota of it I just sketched.

Henry Allison interpreted and defended a Kant that remains a champion of the Enlightenment, someone who thought that our dignity and value ultimately depends on our possessing a rational faculty that allow us to find lawlike regularities in the world in which we live, regularities that have objective validity even if they are always about the world as it appears to us, but even more depends on reason’s special capacity to let us decipher (in most cases at least) what we ought to do and act accordingly in all possible circumstances. But the Kant one learns through Allison is one that remains epistemologically modest and prudent, having discovered how our natural drive toward closure and objectivity can become a serious metaphysical mistake, either in the form of a transcendental illusion before the critical cure, or in the form of a refusal to acknowledge the intrinsic limitation of our knowledge, that is, transcendental realism. A mistake that as a karst river re-emerges in new guises in the history of philosophy and occupies, as we saw, such a large area of contemporary philosophy. Allison is the most important interpreter of Kant in the last hundred years because he explained and defended, as no one else did with a comparable level of detail, knowledge and intellectual precision, a Kant that may and should even today keep on performing its critical function.