Reseacher, professor at the Free University
Annotation: In this article I will talk about Barry Stroud’s critique of transcendental arguments, which he understands as anti-sceptical reasoning, coined by Kant in “Critique of Pure Reason”. In the widespread reading, partially established by Stroud himself, we start from some important ways of thinking and experiencing the world and proceed to the conclusion, which, ideally, reflect necessary conditions of our thinking and experiencing the world in certain way. Thus we try to defend these conditions from the doubts of a sceptic. Stroud poses a famous trilemma for a TA proponent: we try to find some truth about the world, independent from a mind—option, which is impossible due to the restricted domain of TA; we can try to ‘bridge the gap’ between premises and a conclusion of TA by accepting transcendental idealism. The second option is ‘a verification principle’. Transcendental idealism is controversial, verification principle makes TA redundant. The third option is ‘modest TA’: we refuse to make conclusions about the world as it really is. Without denying that TA may face some serious problems, I will try to prove that Stroud’s critique is based on interpretative inaccuracies and methodological inconsistency. I will try to show that by highlighting Stroud’s conflation of two kinds of TA which he himself presents: ‘E-formulation’ and ‘doxastic formulation’. More importantly, I will criticize his understanding of the doctrine of transcendental idealism. In my opinion, Stroud faces a dilemma: either he bases his critique on an accurate interpretation of Kant—and in this case this critique may vanish, or—he refuses to tie TA to Kant’s conceptual framework. In the last part I will outline another problem with TA, formulated more broadly, then Stroud’s trilemma—‘the relevance problem’
Keywords: Kant, Critique of Pure Reason, Stroud, transcendental arguments, transcendental idealism, realism.
—
What are the transcendental arguments? They can be presented as having form of ‘If there is A, and if B is a necessary condition of possibility of A, then B’. This is one the possible definitions—or, at least, candidates for it. Why ‘candidates’? The reason why this supposed definition is not so good after all, lies in the context,—or, in other words, in ambiguity of understanding. Lets’ start with obvious. What a conclusion of these arguments (in this form) is supposed to mean? Should we accept it (on some basis, whether epistemic or practical)? Or it means that B is simply true? These are two different interpretations of such conclusion, or even two different conclusions. An acceptance of a conclusion can be based on it being true in non-epistemic readings of ‘truth’, since we can reasonably endorse the norm “You must believe what is true, if it’s known to be true or can justifiably claimed to be so”. Nonetheless, logical conclusion that P (in this case—B) is true is different from ‘therefore, you should believe P on a basis of X, be it ’P is true’ or something else. Another moment—B’s being a necessary condition of possibility of A. Not that it’s itself problematic, but that’s just not clear. Without this clause this formulation is narrowed down to modus ponens. Why introduce such a clause then? Another question—what an argument of such form gives us, what’s its use?
Maybe to respond to these questions (or to give a direction to these responses) we need to appeal to history. Transcendental arguments were popularized by Barry Stroud in 60s, especially by (and in) his work of the same name (“Transcendental arguments”, 1968 and “Kantian arguments, conceptual capacities and invulnerability”, 2006). He didn’t introduce them into philosophy, though. First (recorded) explicit use of the term ‘transcendental argument’ belongs to John Austin in his presentation “Are There A Priori Concepts” in 1939 (1961: 1-22). However, when Stroud talks about TA, he mostly means another philosopher—Peter Strawson and his work “Individuals” (alongside Sidney Shoemaker—“Self-Knowledge and Self-Identity”, 1963). According to Stroud, these arguments are ‘Kantian in nature’ and are dedicated to undermining or refuting scepticism. But what is exactly Kantian about it? Kant distinguished between ‘question of fact’ and ‘question of right’ (Kant, 1953, p. 84). First concern the acquiring of concepts we use, be it empirical or a priori concepts. Second brackets out such ‘genealogy’ and problematize exclusively whether we have a right to use them, or—to be a bit more precise—whether we are justified in applying these concepts. The question ‘how to apply these concepts correctly’ is a part of the ‘question of right’. Sceptic can be understood as someone (though this can be a completely virtual person) who questions or denies the applicability of certain concepts. In case of “Individuals”, for example, this would be a concept of a persisting particular. There could be another instances, of course—since a skeptical position described above can be formulated regarding many areas of knowledge and involve many concepts. But what differentiates TA from modus ponens? ‘A implies B’ literally implies, that B is a necessary condition of A. The difference lies in ‘necessary condition of possibility of A’. The question is quite complicated. However, some explanation can be given straight away. There is a reason why TA include this clause—their ‘Kantian’ origins. Kant tried to explain the possibility of cognition in general. He insisted that such task can be conducted only A Priori—without any appeal to empirical content of a cognition. Hence the clause we are talking about. Be it ‘A implies B’, it could be read as designating a logical connective between propositions which describe two particular facts or event, but such matters are not what Kant is interested in. ‘A’ specifically could designate particular event or fact. What Kant wants is universality. There could be several ways of achieving it (meaning: necessary steps, not sufficient), and one of them is substituting ‘P’ as standing for any proposition with ‘the possibility of A’. ‘A’ here represents a possible fact. Such possibility can be read in the different ways, but one emphasis is important: ‘possible’ means ‘not necessary actual’. In other words, there can be no instances of A, but this is irrelevant, since in Kantian programme we attempt to explain or justify against the skepticism something, which either cannot be instantiated, or doesn’t have to be instantiated.
Before we turn to Stroud’s critique of transcendental arguments, we have to summarize the crux of them. These arguments are anti-skeptical, given the broad understanding of scepticism as a denial or doubting an applicability of certain concepts; even more broad understanding requires the notion of belief. A sceptic is someone who either denies that certain claim (X) has any epistemologically positive features, or doubts its’ having such features. ‘X’ here can stand for a claim about a concept in epistemological context (for example, ‘the concept N is meaningful’) or for a belief. What a proponent of a TA can offer to as an anti-skeptical aid? Remember, what we try to achieve by TA is, basically, either refuting or undermining a sceptic, or, to be more precise, to undermine or refute sceptical position regarding certain topic. To put differently, we try to prove or justify some Q. A proponent of TA starts with the premise, which either cannot be doubted by a sceptic, or is—rightly or not—accepted by him (this is an important nuance, but we will not touch it in this work). Then we move to the next premise: if P (stands for a possibility of some X), then Q is its necessary condition—or, in other words,—its condition of possibility. Then—the last step in a transcendental argument: Q holds1. A Q’s holding may be interpreted differently. First of all, ‘Q’ may represent not only a proposition in austere (metaphysical2) sense, but some another cognitive component. There is, however, a restriction—this component cannot be empirical. Taking this into account, we can say, that ‘Q’ is either a proposition (or something having propositional form), or a concept. In the latter case ‘justifying Q’ means justifying its meaningfulness or applicability. Another important point of diversion—the strength of TA. In this regard, we can give several interpretations. However, some interpretation may have some justified priority. According to Stroud himself, a proponent of TA seeks or at least should seek to achieve something salient, something, which cannot be properly doubted, denied or refuted by sceptic. This quite plausibly favours an attempt to find some truth in the strict sense—a true proposition or a fact. However, given the conceptual nature of TA’s first premise, its conclusion should be conceptual too.
But why exactly should it be conceptual? An attempt to answer this question will help us understand, what’s wrong with TA in Stroud’s opinion. Let’s look at his reasoning. According to Stroud, the transition from the first premise of TA on which we agree with a sceptic, is problematic. The source of its problematic nature lies in the different ontological and epistemological commitments in a first premise and in a conclusion. The former is—broadly—conceptual. It reflects or represents certain features of our thinking or experience of the world. However, a conclusion is understood in a realistic frame: it is supposed to reflect a relevant aspect of the world completely independently of our modes of thinking about and experiencing the world. Stroud bases such requirement on the text of ‘Individuals’ and Kant’s writing, especially ‘The Refutation of Idealism’—respectively (BXL). In this regard this claim is purely descriptive—it represents these authors’ intentions in the use of TA. However, there is—or can be—a good reason behind this requirement. Earlier we found out that Stroud understand scepticism quite broadly—it is any denial or non-acceptance of a dogmatic position regarding certain topic. To match such potential force of a sceptical position—if it is successful—we as anti-sceptics have to present an argument with an equal force. The statement which reaches out to reality (in a relevant aspect) as it is, is enough in the most extreme cases of scepticism. So, we have to state something about the world as it is when we make TA—at least in some cases. But it seems that such logical transition is incorrect. We cannot make a step from a proposition that reflects merely our ways of thinking and experiencing the world to a proposition about the world as it is independently of any thought and experience. This is a point which Stroud makes. However, he doesn’t stop here. Stroud actually proposes a choice for a TA proponent. According to him, we can—theoretically—‘bridge the gap of necessity’ between a first premise and a conclusion. The first option is a principle of verification of some sort (in order to believe S one must be able to verify S—e.g. to establish its truth-value). Why this principle? Well, sceptic could argue that in order to think about the world in a certain way or to have a certain experience of the world is sufficient to believe that world has some feature or is structured in a certain way. But that’s it—all we have at the end is merely a belief, which can be or would become false, whether its acceptance is universally justified or not. To fix this weakness we may endorse said verification principle. It’s just seems like a logical solution. There is a catch, however. If we accept this principle, we tacitly assume that somehow have an ability to judge about the world as it is (even if the world is understood in some odd idealistic frame as composed of mental entities only). But in this case the use of TA is rendered superfluous. The reason why we have to use TA in the first place is that using such strategy can undermine or refute scepticism indirectly. If we have a possibility of a direct refutation—why use TA? The second option Stroud proposes (mostly in “Kantian arguments…”) is transcendental idealism. In his opinion, we could try to ‘bridge the gap’ by stating that the world as it exist is not independent of our ways of thinking or experiencing it, quite the opposite—it’s constituted by these ways (2005, p. 159). The gap, according to such strategy, is not bridged by the revealing of correct meaning of the words ‘the world’ or ‘reality’. As far as I understand Stroud’s intention, he doesn’t try to show that any or at least some of TA’s conclusion are really about the world as it is, because ‘world as it is’ really means ‘the given manifold constituted by our (ways of) thinking and experiencing it’. Justifying such co-intensionality would render skeptical stance of ‘TA prove just a necessary way of thinking or experiencing of the world; or some necessary beliefs’ literally meaningless. But there is obvious price: the thesis that such co-intensionality holds is trivial—so, in the absence of any obvious reasons to think that it does hold its hard to prove it. It’s more reasonable to hold more modest thesis: ‘the world as it is’ and ‘the world as it is constituted by our thinking and experience’ do not mean the same, but the objects they refer to are the same object. Unlike the previous identifying move, this is not semantic, but metaphysical. We may meaningfully think of the world differently, but there is only one view, which is justified and which we should accept as a way of ‘bridging a necessity’ between the ways we experience the world and the way it is—transcendental idealism as it is described earlier. Stroud talks about transcendental idealism in this, metaphysical version. Despite accepting this option as meaningful and even prime facie plausible, Stroud denies it. The reason is quite simple. In his opinion, transcendental idealism cannot offer us much more than a scepticism does. It says that the world available for our cognition is confined to the makings of our own thoughts or experiences. In Stroud’s view, though not contradictory, transcendental idealism denies us an access to reality as it is (2005, p. 162). But if previous option, along with this one, are ultimately implausible for some reason, are we left with any other theoretical variants? According to Stroud, yes. He states that there is another option—modest TA (). Such transcendental arguments are restricted to our thinking and experience: we start from a premise reflecting our ways of thinking or experiencing the world and move to a conclusion, which identifies necessary conditions of it, which are themselves restricted to our thoughts and experiences. It’s important to note, however, that Stroud more often conceptualizes modest TA in doxastic terms: we go from beliefs about the world to another beliefs, necessary for the reasonably holding the former (). Since beliefs form a net (2005, p. 164), identifying their inner connections is an important task. Moreover, since we cannot escape having some of these beliefs, modest TA are promising.
In my opinion, Stroud’s reasoning has some serious flaws. Some of them are related to the explicit or implicit distinctions he makes. A good example is two different formulations of TA: regular, in which he uses the notion of experience (E-formulation) and another, in which he uses the notion of a belief (doxastic formulation). It seems like he uses these formulations interchangeably. But are they really interchangeable? In E-formulation Stroud talks about thoughts and experiences or, to be more precise, ways of thought or types of experiences. In other words, he talks about types of experiences and conceptual schemes. The conclusion of a transcendental argument should also be related to the same things (‘things’ in a wide sense, not ‘a particular’). Arguments, however, can’t be formulated without something, which has a propositional content. Thoughts are enlisted by Stroud himself, but it seems that he enlists thoughts and experiences on the same level: thoughts are, in this interpretation, either not ‘about’ experiences, or this ‘aboutness’ can be bracketed out, because thoughts are considered as a sub-category of experiences (I will mention and discuss the alternative option below). Thus, both premises and conclusion(s) of TA must be treated (as far as I can tell) as propositions regarding thoughts and/or experiences: from the P, which stand for ‘experience/thought X holds’ follows Q, which stands for ‘experience/thought Y holds’, where the latter is a necessary condition of (possibility) the former. Since ‘P’ and ‘Q’ designate some thoughts or experiences, we can bracket out the question whether P or Q are true or false (Q is implied by P, so it’s mostly about P). They simply represent certain thoughts or experiences taking place. This is not the case when we talk about beliefs though—or when we talk about thoughts in their epistemic or semantic (not ontological) aspect. Beliefs or thoughts (in aforementioned aspects) can be true or false, whereas experiences just happen in or to subject. This might look like a small difference, but it’s nonetheless important. When we start with experiences, the question ‘Are they true or false, justified or not’ is irrelevant or even meaningless. Consequently, there is no question whether sceptic agrees with us upon it. Experiences simply are or can be—there could be no point of disagreement. Unlike starting with experiences, accepting a false belief (be it about ‘the world as it is’ or about ‘the world as we experience it’) might be costly. Even if sceptic for some reason accepts some belief with us, this belief (let’s call it ‘A’) can be false or at least not justified (not ultimately justified—this would explain, why one can accept it in a first place). A belief B’s being a necessary condition of A doesn’t help here, because its being a necessary condition of a false belief doesn’t give us an important information. Yes, B can be true, but without the basis of true A we need another method of recognizing its truth-value. This fact makes the very use of TA redundant—were we had such method of establishing truth-value of B, why would we use indirect transcendental arguments in the first place? Theoretically, we can escape such uselessness without sacrificing TA—we just need to secure a first premise. It can be done by showing that A is true and/or justified. Roughly, we can say that A can count as knowledge (for a research purposes I bracket out the complication known as ‘Gettier problem’—see Gettier, 1963). There are at least two problems with this proposal. First of all, it lacks epistemic modesty. One of the reasons why we would try to use TA in the first place is their minimal commitment. Transcendental arguments in this variation lack this important feature—there is nothing modest in the claims of knowledge, justified or true belief (unless it’s very minimalist interpretation of justified belief, if we talk about justified-beliefs-claim). The second reason is related to pragmatic nature of TA. These arguments are ant-sceptical; Stroud understands skepticism quite broadly: it can include both metaphysical nihilism, eliminativism or a scepticism in a narrow epistemic sense: denying the possibility or attrubutability of knowledge or justified belief. Given this understanding, we cannot say that any particular sceptic will necessary deny any knowledge, but we cannot guarantee the opposite either. So, putting knowledge as a first premise requires proving that we do have this knowledge. But this substitutes ‘not asserting your own theses’ strategy, used by TA proponent (which is pyrrhonic in nature3) with a counteroffensive move. In such a way a possible pragmatic advantage of TA is lost—and, arguably, its core.
There is another big problem with the way Stroud characterizes the distinction between E-formulation and doxastic formulation. It’s related to his interpretation of transcendental idealism. According to it, the world as it is experienced or thought by us is constituted by or dependent on the possibility of thought or experience (Stroud, 2005, p. 159). Accordingly, in E-formulation, any conclusion of TA represents reality exactly because it represents thoughts or experiences necessary for the thoughts or experience, from which we start such TA (to be a bit more picky—we start from a premise representing/substituting for such thoughts or experiences, but it’s more of a technical detail). But would this work for a doxastic formulation (if we assume that E-formulations or, in other words, E-transcendental arguments, may work from theoretical standpoint)? Given the assumption of transcendental idealism, a doxastic formulation of TA would go something like this: we go from A, which stands for a belief, to B, which is necessary for holding or having A; from the transcendental idealist’ standpoint B—and, possibly, A—somehow constitute the world as it is (at least, for us). Is this formulation really different from the previous one? From some perspective it may be not. When we say that a belief constitutes certain facts or features of reality (in the relevant aspect), we may mean three things: the fact that belief is held somehow changes or constitutes the reality (1); belief as a mental state changes or constitutes the reality (2); the conceptual structure of a belief changes or constitutes the reality (3). Whatever interpretation we choose, none of them is specific of a belief: we can come up with the same three readings while sticking to thoughts and/or experiences. Beliefs would be nothing more than a sub-category of a constitutive mental objects or states. To put it other way: the cognitive aspect of beliefs is either ignored, or bracketed out for some reason. What if we won’t ignore it? Well, we can still support transcendental idealism (theoretically). But even if we accept transcendental idealism, that wouldn’t mean that our beliefs—in their cognitive aspect—constitute the world in any way. Beliefs’ constituting the world would be possible, if they were cognitions, which Kant himself discussed: A Priori forms of understanding and sensibility (Kant, 1998, A19/B33). According to Kant, these forms do constitute the way in which the world is cognized by us. We could prima facie accept transcendental idealism in this original understanding—this view doesn’t seem to be contradictory (if we don’t lean towards Jacobi’s critique, or course—Jacobi, “Werke”, Vol. 2, pp. 291-310). However, there is at least one peculiar moment, which makes transcendental idealism prima facie acceptable. This view doesn’t state that these A Priori forms constitute the world in general or constitute its content. Transcendental idealism is concerned only with forms of cognition, which are in turn conceptualized in the certain way. But beliefs are not forms in the Kantian sense—they are neither concepts, nor intuitions. Moreover, beliefs can only be formed through concepts and—possibly intuition (if we stick to Kantian vocabulary). So, if we need some reason to think that beliefs are somehow metaphysically constitutive,—Kant doesn’t give us such reason. Perhaps, we can find another one? It’s dubious due to the very nature of beliefs. They are meant to represent reality or confirm to it (depending on your preferred vocabulary). In other words, one of the key features of beliefs is their potential factivity (which distinguishes them from knowledge, which is actually factive—Pritchard, 2012, p. 3—the definition of ‘factivity’). Beliefs in general can be true or false (necessary true or false beliefs being an exception). The concept of a belief itself doesn’t preclude transcendental idealism, materialism or any type of metaphysical position (except, perhaps those which preclude our possibility to form beliefs). This concept doesn’t imply any such view either (with the exception of a similar kind). When we say that a belief can be true or false, we doesn’t imply or implicate what kind of facts are, can or could be truth-makers of a belief. Our metaphysical views say exactly that, since they assert what kind(s) of facts reality consists of. We may—theoretically—accept transcendental idealism and use E-type of transcendental arguments, but it doesn’t mean that we are justified to use doxastic TA on the same grounds. Our believing taken in its cognitive aspect not only doesn’t constitute the reality in the way Kantian ‘cognitions’ do, but it also can be false even if transcendental idealism is true—simply because beliefs can be false by definition.
Transcendental idealism is one of the options Stroud thinks is possible for a TA proponent. But what he means by ‘transcendental idealism’ and is his understanding even correct? First of all, I need to clarify what ‘correctness’ means in this context. In order to be correct, interpretation must be based on a text (1) and be supported by sufficient reasons (2). Let’s give Stroud a benefit of a doubt and assume, that he based his understanding of transcendental idealism on the text of ‘Critique of a Pure Reason’. Is it supported by sufficient reasons? It’s hard to find any clear indication of that in the works about TA—article of the same name and ‘Kantian arguments’. Stroud characterizes transcendental idealism as a view, which postulates that ‘…the world (is)…transcendentally speaking, depends on or is ’constituted’ by the possibility of our thinking and experiencing things as we do’. This interpretation can barely be considered plainly incorrect. It’s in accordance with some Kantian fragments, such as this: ‘We have therefore wanted to say that all our intuition is nothing but the representation of appearance; that the things that we intuit are not in themselves what we intuit them to be, nor are their relations so constituted in themselves as they appear to us; and that if we remove our own subject or even only the subjective constitution of the senses in general, then all the constitution, all relations of objects in space and time, indeed space and time themselves would disappear, and as appearances they cannot exist in themselves, but only in us’ (Kant, 2013, A42). However, Stroud’s reading has some shortcomings. One of them is ambiguity. What this doctrine states, in Stroud’s view? Is the world as it exist in the most basic sense of the existence dependent on possibility of thinking and experiencing things as we do? We can quite safely say, that these are two different options (not necessarily mutually exclusive, however). If something constitutes the world, the world as a whole is somehow constituted by it. We are not obliged to say the same if speak about dependence. The world may be dependent on the possibility of our thinking and experiencing it in certain aspect and not be dependent in others. For example, the answer to the question ‘What exists?’ may depend on our accepted conceptual scheme. If we change the scheme, the answer changes too. However, this doesn’t mean or imply that our thoughts or perception at a time may change the facts of the world, which are no way related to our thinking or perceiving. An introduction of a new particle into our contemporary worldview might significantly change it, but my thought that it’s cold outside doesn’t change the weather in any way. Another good example of this ambiguity is Stroud’s notion of ‘possibility of thought and experience’. He tries to define transcendental idealism with this notion. But what does it mean? Two interpretations come to mind (they may be not exhaustive, but that’s not critical in this context, since I don’t try to make classification). First one—let’s call it substantial—states that Stroud means ‘possible thoughts and experiences’. Thus, the dependence or constitution is understood as a relation between certain object—thought or experience, to which certain modal status is ascribed—and the world. This reading quite legitimately leads to the question ‘What are these thoughts and experiences?’. If the whole world or even part of it is constituted by or dependent on certain domain of possible thoughts and experiences, then it’s important to know what thoughts we are talking about. According to the second, propositional interpretation, we talk about the possibility of thought and experience here. The world is crucially dependent on the fact that they are possible. This interpretation is starkly different from the previous one, since we don’t have to clarify or explain, how the whole world or its part is dependent on certain possible thoughts and experiences. When we postulate the possibility of thoughts and experiences, we don’t imply anything about the certain thoughts or experiences being possible. The thesis that the world is dependent on the possibility of thought and experience is, in my opinion, closer to Kant’s attitude. His ‘Copernicus’ turn’ is built upon the idea that the world is structured in such a way that makes its cognition possible, because the way the world exist or may exist corresponds to our possibility of cognize it; roughly put, thoughts and experiences are two basic forms (faculties) of cognition (). To be fair, one can say in Stroud’s defense, that Kant hasn’t clarified his notion of ‘possibility of experience’ (along with many others). This would be a weak line of defence. Not because Kant, unlike Stroud, is completely clear about this, but because he hasn’t defined transcendental idealism in terms of the world depending on or constituted by the possibility of thought. The very expression ‘the possibility of X’ is Kantian in character, but Kant is much more specific (which doesn’t necessarily translates into clarity) about the formulations of one of his central doctrines. And this brings us to another issue.
There is another problem with Stroud’s interpretation of transcendental idealism, or—to be more precise—two problems, which are related. One is more obvious: when Stroud talks about transcendental idealism, it seems that he is basing his account on the broad notion of ‘transcendental’. We can see it in his definition or quasi-definition of transcendental idealism, when he mentions that ‘the world is, transcendentally speaking…’ (my highlight). What he means by ‘transcendental’ here? We could present many guesses, but shouldn’t. Stroud don’t give us clear explanation of what he meant. In the first Critique the world ‘transcendental’ has several meanings. For Kant, one of the prominent and, probably, relevant use of ‘transcendental’ means ‘related to the conditions of possibility of experience’ (2013, A12/B26). The opposition of ‘transcendental’ is, from one side, ‘empirical’, and from another—‘transcendent’ (2013 A296; B353). The term ‘empirical’ is referred to any object, which is given in experience or—by the faculty of sensibility (2013, B353; A311; A320/B377; BXXVI). Another close variation of the meaning concerns any cognition—intuition or concept—which is based on experience or can be applied only to experience (2013, B353). The term ‘transcendent’ refers to the objects beyond the domain of possible experience. ‘Transcendental’ is, in certain sense, between empirical and transcendent. Maybe, Stroud does mean that when he uses this term. This charitable interpretation is acceptable or maybe even desirable. However, it gets quite confusing when we come to idealism. Kant doesn’t use the term ‘idealism’ randomly. It is a position, which postulates that some X is ideal, where ‘ideal’ is a metaphysical/epistemological status. The complete meaning of this notion is unclear; however, its opposition is ‘real’—and, accordingly, realism. We may presume that ideality and reality in the Kantian sense mean some kind of dependence and independence (but I won’t dive too deep into this issue4). What is clear though is that both idealism and realism are not positions about some topic. In ‘transcendental aesthetics’ Kant talks about space and time (). He tries to find out the answer to two questions: 1) is time an object, relation between objects or something else?; 2) do object exist as having spatial and temporal features independently from a cognizing subject or do the absence of a cognizing subject means that these features disappear too? Transcendental idealism is an attempt to formulate satisfactory and complex answer to these questions (to put it simply). Thus, this view—whatever interpretation we choose—is much more specific than the general thesis that the world is constituted by or dependent on the possibility of thought and experience.
The second problem can be formulated as a question—why appeal to transcendental idealism at all? This view postulates the dependence of certain kind. The specific content of TI (transcendental idealism) is unclear, though. Particular set of theses included in this doctrine may change depending on the interpretation. The relevant set may include three claims: space and time are neither objects, nor a relation between objects—they are forms of our sensible intuitions of objects (1); these forms of sensible intuitions of objects are not empirical, they are A Priori (2); objects which we cognize in space in time, are not ‘things in themselves’—objects, which exist independently of our forms of sensible intuition (3); we can cognize only the objects of our intuition—thus, we cannot cognize things in themselves (4); we can think about things in themselves using the categories (5); we cognize objects in space and time because things in themselves affect us by activating our sensibility (6)5. If we base our understanding of TI on A-edition exclusively, we won’t include the central thesis of ‘The Refutation of Idealism’ in B-edition, which states that the objects of the external world exist independently of our cognition. Another point of divergence—the scope of objects, which are—broadly speaking—cognizable according to TI. For example, Lisa Shabel in the third chapter of “Cambridge Companion to Kant and Modern Philosophy” writes that ‘the pure intuitions of space and time and the pure concepts of the understanding apply to all—but only—appearances, and not to things as they are in themselves’ (2006, p.114) when she attempts to describe the essence of transcendental idealism. This understanding doesn’t seem to get along with the aforementioned reading, according to which we at least can think about the objects using the categories of understanding. Another issue occurring in the discussion dedicated to TI is related to the question ‘what objects can affect us and activate our sensibility?’. Karin de Boer, for example, criticizes Henry Allison for thinking (Allison, 2004, pp. 64-65) that by ‘object, which affects us’ Kant means either phenomena or things in themselves, while ignoring other options (2020, Chapter 4, Part 2, pp. 103-105). These interpretations are in friction, but they are not the most extreme. Robert Hanna proposes in his book “Kant and the Foundations of Analytic philosophy” a different interpretation, which can be—in principle—called revisionistic. This interpretation is built on the Kant’s intention in the first Critique. Hanna carefully uses Kant’s documentation, especially letters (to Garve and Herz), and comes to quite peculiar conclusion—Kantian project is not primarily metaphysical. The answer to the question “What is?” or even ‘What is possible?” is not Kant’s primary concern. The goal of CPR is to answer a different question, which we might call ’the question of intentionality’ or ‘question of representation’. Since the first Critique is dedicated to giving a response to this question, Hanna calls Kantian project ‘the cognitive semantics’ (2001, p. 19). The question Kant seeks answer to may be formulated as ‘How is possible to our mind to represent (zu darstell) things?’. This would be a correct, not precise enough formulation. According to Hanna, Kant is more concerned with the representational capacity of (purely) A Priori concepts: “As to how my understanding may form for itself concepts of things completely a priori, with which concepts the things must necessarily agree . . . this question of how the faculty of the understanding achieves this conformity with the things themselves, is still left in obscurity (PC Ak. x. 131)”. Accepting this frame of reasoning, Hanna formulates his understanding of transcendental idealism as a conjunction of two theses—representational transcendentalism and cognitive idealism. The first doctrine states (in this interpretation), that “all representational contents, and thereby the contents of all cognitions, are strictly determined in their underlying structure by certain universal, innate, a priori human mental capacities—in a word, transcendental capacities—that make experience itself possible” (p. 20). In support of his interpretation Hanna brings up A11/B25 in CPR. He defines the second thesis as a claim that “all the proper objects of human cognition are nothing but objects of sensory experience (which Kant calls ‘appearances’ or ’phenomena’17).”.
I did not bring Hanna’s interpretation up simply to inform a potential reader of its peculiarity. By showing it I tried to tell how drastically different the doctrine of TI can be interpreted. For some the accuracy of their reading is simply a question of adapting a proper interpretation of a Transcendental Aesthetics (mostly), but for others—Hanna being an example here—it is a question of following Kant’s methodological intention. Thus I want to illustrate a simple point. The doctrine transcendental idealism is hard to interpret and—consequently—hard to apply. If there is an opportunity to avoid appeal to it—perhaps it would be better to avoid it.
But here the problem comes. Stroud proposes two options for a TA proponent (hence the name ‘Stroud’s dilemma’), if he wants to ‘bridge the gap of necessity’ from premise of TA to its conclusion: accepting some kind of ‘verification principle’, which would make the very use of TA obsolete (1); accepting transcendental idealism—the claim that the world is dependent on the possibility of thought and experience (2). He proposes another, third option, which doesn’t presuppose an attempt of ‘bridging the gap’—the modest strategy, which consists in identifying the necessary beliefs which lay foundation for other beliefs. What if Stroud avoided the second option? It would be simply logically inconsistent. There are—theoretically—two way of connecting beliefs about the reality with reality (or conceptual schemes about reality and reality itself): either we claim that the reality as it exist independently of our mind is accessible to us (1) or we claim that we have such access, because reality as it exist is not (metaphysically) independent of our mind (2). The first variant is embodied by ‘the principle of verification’, the second—by transcendental idealism (as Stroud understands it). It’s important to note: I don’t mean that these two options enlisted by Stroud correctly reflect aforementioned choice. I simply state that in the Stroud’s argument there are no other suitable options. However, as I think about it, Stroud cannot state that his second option available for TA proponent is transcendental idealism without significant cost. A quick reminder: Stroud claims, that TA are Kantian in nature. What does it mean exactly? We can hold one of the two hypotheses here. The first is weaker: it states that Kant uses this type of arguments or that the only type of the arguments Kant uses is this. The first claim is useless—both from historical-philosophical and theoretical standpoint. The fact that Kant used certain arguments doesn’t tell us anything important about these arguments’ use, functionality or structure. The fact that Kant uses only this type of arguments is richer in this regard. Let’s assume it’s true (it doesn’t look like this—see Paul Guyer, “Kant and the Claims of Knowledge”, 1987; Karl Ameriks, “Kant’s Transcendental Deduction as a Regressive Argument”, 1978, pp.273-87). This (assumed) fact may reveal to us some important information about Kant’s method. But only to a certain degree. If Kant use only TA, the term ‘TA’ itself becomes a synonym to ‘the arguments that are used by Kant’. This is not a contentful description. To understand what TA are we still need to interpret the first Critique (and perhaps not only the first). The second interpretation is stronger. It says that TA are connected to crucial features of Kantian project, be it methodological or doctrinal. This thesis doesn’t mean that we cannot formulate the Kantian doctrines without transcendental arguments—it might be true, but too specific for our purposes. This thesis also doesn’t mean that everyone who uses TA instantly becomes Kantian simply due to the fact the she uses arguments of this kind. To say that we need to identify Kantian doctrine. It is not a simple task, to say the least. The suitable interpretation of the strong thesis is, in my opinion, this: we cannot formulate and use TA without some appeal to Kant’s conceptual framework, method or doctrine. If this is what Stroud meant when he said that TA are ‘Kantian in nature’, then his critique of TA faces a dilemma. On the first horn of it we formulate the choice for TA proponent by appealing to distinctly Kantian conceptual framework—as Stroud does or tries to do. On the second hor—the denial of this approach. Necessary appeal to Kantian conceptual framework creates a problem. Either one should clarify what transcendental idealism is or justify why there is nothing to clarify. I will not try to decide what of these options is preferable, but I will say that both tasks are hard in execution. Besides, in both cases one will have to prove that a particular interpretation of transcendental idealism is true and, more saliently, that the second option is indeed a transcendental idealism after all. The second point needs justification, since Kant himself gives us two distinct grounds for distinguishing realism and idealism—‘transcendental’ and ‘empirical’(B44/A28). To put it simply, ‘transcendental’ ground is related to the ontological status of the conditions of possibility of experience (which in Kant’s view are restricted to space and time as the conditions of sensibility and categories as the conditions of understanding). On the other hand, ‘empirical’ is related to the ontological status of phenomena—objects, given in experience. Theoretically, one can accept that empirical objects exists without a subject’s cognizing them (or: a subject’ cognizing them) without accepting that space and time are themselves objects, independent of cognition—and visa versa. So, Stroud is faced with the task: to prove that the second option for TA proponent is transcendental idealism and not its empirical counterpart. I wouldn’t say it is impossible, but I would say that it requires some serious amendments to Stroud’s initial formulations, since his understanding of transcendental idealism (the world is constituted by or dependent on the possibility of thought and experience) is so ambiguous, that it can include both transcendental and empirical idealism. Stroud’s proponent could, of course, attempt to prove that these positions are equivalent, though they don’t seem to. But this move would still require clarification of the essence of Kant’s doctrine. Let’s turn to the alternative now. What if Stroud had chosen the other option? Since TA, in his interpretation, are ‘Kantian in Nature’, abandoning distinctive Kantian framework means that we do not talk about TA anymore, though these arguments are similar in some aspect to TA. These arguments can be fruitful, but since they lack distinctive ‘Kantian’ features, they are not ‘transcendental arguments’. Moreover, in theory, we can formulate the same trilemma for this type of arguments, but the available options won’t include transcendental idealism (though it may include ‘verification principle’ in some version—see below)
How can we solve this dilemma, which arises for a Stroud’s proponent? Is there a way out of the choice between being forced to clarify what transcendental idealism is and prove that one of the option in trilemma is indeed a transcendental idealism and talking about something which cannot be strictly speaking characterized as ‘transcendental arguments’. One of the strategies is to demonstrate that the choice is itself either nonsensical, or built upon a false dilemma. I could go in this direction. But in this case some fruitful proposals may be lost (the other reason is being charitable to myself). So, my proposal leans towards the revision of the very notion of ‘transcendental philosophy’ and ‘Kantian philosophy’, as far as they are interchangeable (and I don’t think they are completely interchangeable). Of course, I won’t revise these notions completely, only in the aspect relevant to the formulation of Stroud’s trilemma. Stroud doesn’t propose any unambiguous and clear idea of what he calls ‘transcendental’ or ‘Kantian’. So, in order to revise his ideas I should have reconstruct them. Not only is it complicated—in my opinion, it’s unnecessary complication. We don’t need to form a complete account of what ‘transcendental philosophy’ or ‘Kantianism’ is—only an outline. My offer is to distinguish between Kant’s philosophy and Kantian philosophy; accordingly, we can use adjectives ‘Kant’s’ and ‘Kantian’ while referring to certain aspects of an account, theory, a particular thesis or hypothesis. The expression ‘Kant’s arguments’ refers to the arguments used in his works (not only in CPR). This expression is not meant to reflect any important aspect of Kant’s argument—only the fact that Kant used it. The arguments used by Kant may be united in some peculiar way or they may not—it’s the task of a future inquiry to find out. Unlike this expression, ‘Kantian arguments’ designate any arguments, not necessarily used only by Kant, which in some way are essentially tied to some components or aspects of Kant’s method or doctrine or borrow elements from said method or doctrine. This notion is purposefully loose, since I see no reason to state in advance what elements of method and doctrine are essential to Kant’s philosophy and what—aren’t. This looseness can—in principle—allow to formulate Stroud’s trilemma for TA without being strictly Kantian in the conceptual framework. Stroud himself gives us a hint how it’s possible. He states that “There is first the question of the security or invulnerability of certain fundamental ingredients of our thought against sceptical attack. And there is the further question of whether that security is to be established by arriving at positive metaphysical results—by proving the truth of the propositions our acceptance of which the sceptic would question” (2005, p. 165). The first question he raises is the question of the force of TA, meaning—the anti-sceptical force. We don’t need to reinterpret anything in this fragment. The second question, however, needs such reformulation. Stroud asks whether the anti-sceptical force can be achieved by getting some positive metaphysical results. ‘getting some positive metaphysical results’ can be interpreted differently, but one of the aspects Stroud considers important is an ability to reach the world understood as independent of the possibility of thought and experience in conclusion. Given the conceptual or doxastic character of a first premise of TA, the insistence on realistic understanding of a conclusion (thought as a necessary condition of TA’s success) gives us a problem of relevance: we apparently have to think of TA as having mutually conceptually inconsistent premises and conclusion. This inconsistency comes from the modesty of TA’s premises in combination with the requirements, which are expressed by accepting a realist framework. To put it simply: this is a problem coming from anti-realist premises and realist conclusions, where we cannot solve it simply by accepting ‘anti-realism’ or ‘realism’ as strategies without significant sacrifices. If we accept realist premises—by accepting some form of verification principle—we will thus make TA redundant, since we at least can make meaningful statements about the world as it is—so, we use indirect arguments? If both our premises and conclusion are framed as realist, then the problem of relevance obviously disappears, but the problem of redundancy arises. If both our premises and conclusion are framed as anti-realist, then we don’t face the problem of relevance, but the force of TA becomes questionable—shall we strictly follow Stroud in this regard.
This understanding of TA and, consequently, the problem of relevance, creates some salient questions. What are these ‘anti-realism’ and ‘realism’, besides vague ‘independence of a mind’—statements? What is mind-independence? Is this distinction categorical and—therefore—is the problem of relevance inevitable? Maybe, it occurs in some domains of knowledge, but not in others? Maybe, it doesn’t occur at all and thus comes from misunderstanding? These are questions for another work.
1. Austin, J. L., 1961. “Are there a priori concepts?” In his Philosophical papers, Oxford: Oxford University Press.
2. Strawson, P. F., 1959. Individuals: An essay in descriptive metaphysics, London: Methuen.
3. Shoemaker, S., 1963. Self-knowledge and self-identity, Ithaca, NY: Cornell University Press.
4. Stroud, B., 1968. “Transcendental arguments,” Journal of Philosophy, 65: 241–56; reprinted in Stroud 2000b: 9–25. [All references are to the latter version.]
—, 1994. “Kantian arguments, conceptual capacities, and invulnerability,” in P. Parrini (ed.), Kant and contemporary epistemology, Dordrecht: Kluwer; reprinted in Stroud 2000b: 155–76. [All references are to the latter version.]
—, 1999. “The goal of transcendental arguments,” in R. Stern (ed.), Transcendental arguments: problems and prospects, Oxford: Oxford University Press; reprinted in Stroud 2000b: 203–23.
—, 2000b. Understanding human knowledge: philosophical essays, Oxford: Oxford University Press.
5. Kant, I., 1781/1787. Critique of pure reason, P. Guyer and A. Wood (eds. and trans.), Cambridge and New York: Cambridge University Press, 1998. [References are in the standard pagination of the 1st (A) and 2nd (B) editions. A reference to only one edition means that the passage appeared only in that edition.]
6. Vahid, Kamid, 2011, “Skepticism and varieties of transcendental argument”, LOGOS & EPISTEME, II, 3: 395–411.
7. Callanan, J.J., 2006 “Kant’s transcendental strategy”, The Philosophical Quarterly, 56: 360–81.
8. Cassam, Q., 1987. “Transcendental arguments, transcendental synthesis, and transcendental idealism,” The Philosophical Quarterly, 37: 355–78.
9. Russell, Murray Adam & Tillman Chris (eds.), 2022, The Routledge Handbook of Propositions, NY: Routledge.
10. Frege Gottlob, 1960, “On Sense and Reference” in In Darragh Byrne & Max Kölbel (eds.), Arguing About Language. Routledge. pp. 36–56.
11. Gettier, Edmund, 1963, “Is Knowledge a Justified True Belief”, ANALYSIS: 23.6.
12. Thorsrud, Harald, 2009. Ancient Scepticism, Acumen.
13. Bett, Richard (eds.), 2010. The Cambridge Companion to Ancient Philosophy, Cambridge: Cambridge University Press.
14. Coliva, Annalisa & Pritchard, Duncan. 2022. Skepticism, London: Routledge.
15. Jacobi, F.H., 1812, Werke, edited by F. Roth and F. Köppen, 6 vols, Leipzig: Fleischer.
16. Pritchard, Duncan, 2012. Disjunctivism, Oxford: Oxford University Press.
17. Shabel, Lisa, 2006, “Kant’s philosophy of mathematics” in Guyer Paul (eds.), Cambridge Companion to Kant and Modern Philosophy. Cambridge: Cambridge University Press: 94–128.
18. Allison, Henry, 2004, Kant’s Transcendental Idealism, New Haven: Yale University Press. Revised and Enlarged Edition.
19. Allison, Henry, 2006, “Kant on Freedom of the Will” in Paul Guyer (eds.) Cambridge Companion to Kant and Modern Philosophy. Cambridge: Cambridge University Press: 381–415.
20. Hanna, Robert, 2001. Kant and the Foundations of Analytic Philosophy, Oxford: Clarendon Press.
21. Guyer, Paul, 1987. Kant and the Claims of Knowledge, NY: Cambridge University Press.
22. Ameriks, Karl, 1978, “Kant’s Transcendental Deduction as a Regressive Argument”, Kant Studien: 69 (1–4): 273–287.
DOI: 10.55167/610b227d6b58