Neo-Kantianism and the Roots of Anti-Psychologism

Click to increase image sizeClick to decrease image size

Notes

¹I thank Michael Bratman, Gary Hatfield, Nadeem Hussain, Elijah Millgram, Bernard Reginster, Alan Richardson, Tamar Schapiro, Wayne Waxman, and Allen Wood for comments and conversations about ideas in this paper. Earlier versions were presented at the Second International Conference on the History of Philosophy of Science (HOPOS 1998), the Pacific Division Meetings of the American Philosophical Association, and the William James discussion group of the Stanford Philosophy Department, where I benefited greatly from audience questions and comments.

²Kant's most famous defense of this claim appears in the Preface to the second edition of the Critique of Pure Reason, at B xxiv–xxxv. He writes, in part:

It is proved that space and time are only forms of intuition, and … further that we have no concepts of the understanding … for the cognition of things, except as … appearance; from which follows the limitation of all even possible speculative cognition of reason to mere objects of experience … I cannot even assume God, freedom, and immortality for the sake of the necessary practical use of my reason unless I simultaneously deprive speculative reason of its pretension to extravagant insights … Thus, I had to deny knowledge in order to make room for faith … Through [this] criticism alone can we sever the very root of materialism, fatalism, atheism, of freethinking unbelief, of enthusiasm and superstition.

[B xxv–xxxiv]

(Quotations from Kant follow the translations listed among the references, and cite the pagination of the standard Akademie edition, except for the Critique, where I use the standard A/B form to refer to pages of the first (1781 = A) and second (1787 = B) editions.)

³For Mach's statement of the early influence of Kant on his thinking (and an account of how he arrived at some of his central views by starting from Kant's standpoint and then rejecting certain key Kantian theses) see the long note at Mach 1910 [1886], 23–4. See also Avenarius 1888, xi–xiii, et passim.

⁴See Friedman 1999, 124–40, 152–62, and 2000, which highlight differences between the Marburg and Southwest neo-Kantians on the question of the distinction between being and validity, and related systematic issues. See also Richardson 1998, 116–38, which focuses largely on the Marburg school. Both authors treat aspects of neo-Kantian doctrine that were influential for the developing views of Carnap.

⁵In this connection, I have in mind especially the treatment of Köhnke 1986. See notes 23, 31 below.

⁶The point was well understood by the neo-Kantians under investigation here: in particular, see Windelband 1907 [1884], 282–4. For the purposes of this paper, I will remain content with the very broad and abstract characterization of norms as ‘rules of reasoning’, which is meant to cover both the cognitive and non-cognitive cases. There are of course interesting differences between the rules governing cognition and non-cognitive norms like those of morality, on the Kantian accounts of them. In the interest of space, I will not attempt to provide a detailed exploration of these matters here, though I offer a few remarks toward the end of section 4, below.

⁷Elijah Millgram pressed me to account for this possibility.

⁸See, for example, the discussions of the independence of normativity from the natural world in Brandom (1994, 156, 159–61, et passim). Similar ideas find expression in Korsgaard 1996.

⁹For general accounts that document the breadth of preoccupation with the controversy in turn-of-the-century German academic philosophy, see Martin Kusch (1995) and Matthias Rath (1994).

¹⁰Erdmann 1964 [1866], 671. (N.B.: This work was first published as the Appendix to Volume II of Erdmann 1878 [1866].) Rath (1994, 32) doubts that the expression is new with Erdmann, or that his presentation suggests so, but to my ear it has that ring: ‘this psychologism (as we would most like to call [Beneke's] doctrine) …’ (Erdmann 1964 [1866], 671). Rath follows Geldsetzer in positing an earlier occurrence of the term in A. Rosmini-Serbarti, but no citation is given. I have so far found no earlier attestation myself.

¹¹Windelband 1907 [1884], 318–54; (‘Critical or Genetic Method?’). Previously (1996 [1878–80], 386–98), Windelband had used the term in a more neutral descriptive sense like Erdmann's in a historical discussion of Fries and Beneke.

¹²Lotze 1880 [1874], the so called big Logik. If the argument for Lotze's influence presented in the text below is correct, then he deserves a larger place in histories of the psychologism controversy than he receives, e.g. in Rath 1994. This is all the more true, since Lotze also himself made contributions in psychology, and was cited not only by anti-psychologistic thinkers like Windelband, but also by ‘pro-psychologistic’ figures in the controversy (see, for example, Stumpf 1910 [1907], 165). For the opposed, psychologistic view of the foundations of logic, consult Sigwart 1889 [1873], 22, et passim, and Wundt 1893–5 [1883].

¹³Such an interpretation of the ‘laws of thought’ was prominent among Lotze's contemporaries. For example, Sigwart insists that, despite the ‘normative character’ that is ‘essential’ to logic, nevertheless ‘we deny that these norms can be cognized otherwise than on the foundation of the study of the natural forces and functional forms which are supposed to be regulated by those norms’ (Sigwart 1889 [1873], 22).

¹⁴That is, logical laws are norms for belief because belief aims at the truth, and they are objective truths – indeed, the most basic and general truths we can attain. The question of normativity is given comparatively more emphasis in the Grundgesetze. There, for example, Frege writes that psychologism rests on a failure to appreciate

the double meaning of the word ‘law’. In one sense a law asserts what is; in the other it prescribes what ought to be. … The [laws of logic] have a special title to be called ‘laws of thought’ only if we mean to assert that they are the most general laws, which prescribe universally the way in which one ought to think … But the expression ‘law of thought’ seduces us into supposing that these laws govern thinking in the same way as the laws of nature govern events in the external world. In that case they can be nothing but laws of psychology: for thinking is a mental process.

(Frege 1964 [1893], 12–13)

But here, too, the decisive underlying point for Frege rests not on the distinction between natural and normative, but on that between the subjectivity of the subject matter of psychology and the objectivity of the subject matter of logic. For he goes on to complain that what is wrong with using such psychological laws as grounds for logic, is that they can at best be laws of someone's taking something to be true, rather than (objective) laws of truth:

being true is different from being taken to be true, whether by one or many or everybody, and in no case is to be reduced to it. … If being true is thus independent of being acknowledged by somebody or other, then the laws of truth are not psychological laws: they are boundary stones set in an eternal foundation.

(Frege 1964 [1893], 13)

¹⁵Hans Sluga (1980) deserves much credit for his perceptive account locating Frege's thought in the contemporary German philosophical context, and he argued that Lotze's statement of the problem crucially influenced Frege; Michael Dummett (1981) has insisted in reply that, while Frege knew (at least some of) Lotze's work, their disagreements are more important than their areas of agreement. Treatment of the prominent controversy between them is beyond my scope here.

¹⁶The text in question appears as ‘Seventeen Key Sentences on Logic’ in Frege 1979, 174–5.

¹⁷This is not to deny differences between Frege's and Lotze's conceptions of truth and objectivity. For example, Lotze claims that

it will be innocuous if we modify [the conception of truth] so that connections of ideas are true when they conform to the relations of the represented content which are the same for all representing consciousness, and not to the merely factual coincidence of impressions which take shape one way in this consciousness, and otherwise in another.

(Lotze, 1880 [1874], § III)

To a rigorous Fregean, perhaps this would already amount to psychologism. But Lotze's ‘identical for all consciousness’ does not simply mean the same as ‘identical for all (actual) consciousnesses’; Lotze's formulation is supposed to carry with it something like Kantian transcendental necessity, derived from the status of logic as a precondition for all possible inquiry. See, for example, Lotze 1880 [1874], § X, §§ 332–3. This stronger necessity allows the laws of logic to be distinguished as objectively valid, in contradistinction to contingencies of human psychology.

¹⁸For Kant, pure general logic is a ‘canon of the understanding’ (A 53/B 77), where a ‘canon’ provides ‘a priori principles of the correct use of certain cognitive faculties’ (A 796/B 824; my ital.). On this conception, logic is anti-psychologistic precisely because it focuses on the structure of the understanding's correct use (as to form), and thereby abstracts from what is proper to psychology – the particular ‘contingent conditions of the subject, which can hinder or promote this use’ (A 54/B 78–9). Those conditions are mere ‘causes from which certain cognitions arise’ and therefore ‘concern the understanding under certain circumstances of its application’, rather than the perfectly general norms to which its use is always subject, regardless of the circumstances (A 53/B 77). In the Logic, the centrality of normative considerations for Kant is even more apparent:

Some logicians, to be sure, do presuppose psychological principles in logic. But to bring such principles into logic is just as absurd as to derive morals from life. … In logic we do not want to know how the understanding is and does think,… but rather how it ought to proceed in thought. Logic is to teach us the correct use of the understanding.

(Logic, Ak. 9: 14)

This argument of Kant is cited and found insufficient by Husserl (1970 [1900], 92; § 19).

¹⁹By contrast to the relative underappreciation of Lotze's importance for the psychologism debates, his influence on the Southwest school of neo-Kantianism, through its founder (and Lotze's student) Windelband, has been well recognized. See, for example, Wagner 1987, and Oakes 1988, 101, 102–3. To my mind, in fact, Wagner rather overplays Lotze as a source compared to Kant himself, whose relevance for Windelband's concerns was brought out strongly by the rising neo-Kantian line of Kant interpretation discussed in the text below.

²⁰This estimate is due to Hatfield (1990, 110).

²¹This characterization is due to Lange (1902 [1873–5], II, 30). Lange's entire chapter on Kant is of interest for understanding nineteenth-century versions of the psychological reading, which eventually provoked (in response) the orthodox neo-Kantian, purely epistemological approach to the theory of cognition. Many psychological readers wanted to replace Kant's transcendental account with an empirical theory of the mind. For example, Lange claims that Kant's great merit was to advance the project to ‘discover’ (Lange 1902 [1873–5], II, 29) the a priori components of experience, but that he had gone about the discovery in the wrong way:

The metaphysician must be able to distinguish the a priori concepts that are permanent and essentially connected with human nature, from those that are perishable, and correspond only to a certain stage of development. … But he cannot employ for this other a priori propositions… [or] so-called pure thought, because it is doubtful whether the principles of those have permanent value, or not. We are therefore confined to the usual means of science in the search for and examination of universal propositions that do not come from experience; we can advance only probable propositions about this.

(Lange 1902 [1873–5], II, 31)

²²Kitcher (1990, 84–6, 111–12, 135–6; 1995, 302, 306) and Brook (1994) offer recent accounts that compare Kant's ideas to results of contemporary experimental psychology. Their books also survey some of the textual evidence that make such comparisons illuminating for our attempts to understand Kant's theoretical philosophy.

²³See Cohen 1987 [1871], 1–2, 15, 38–46, 87–93, and 120–7. The underappreciation of Cohen's interpretation of apriority, and of the significance of this contrast with innateness in particular, is an unfortunate weakness of Köhnke's unsympathetic interpretation (1986, 273–301). Köhnke is determined to take Cohen's emphasis on apriority as a complete misreading of Kant, in the service of his larger program to attribute Cohen's position to political rather than internal philosophical motives. As a result, he misses Cohen's distinction between a priori forms of thought and innate psychological mechanisms, and thus the relevance of Cohen's new Kant interpretation for the question of psychologism.

On a more specific point, Köhnke charges that Cohen's reading underplays the role of the matter of sense in Kant's theory of cognition – and that may be true on balance – but Köhnke's own reaction ignores substantial evidence supporting Cohen, particularly the Critique's strong emphasis on the a priori basis of all knowledge (in so far as it depends on the categories and forms of intuition). Köhnke's extended (pp. 284–9) attack on Cohen's use of Kant's slogan that ‘we can cognize of things a priori only what we ourselves have put into them’ (B xviii), is especially implausible. Contra Köhnke, Kant's claim about a priori cognition at B xviii can not be dismissed as one restricted merely to the case of metaphysical judgments; rather, it is an application to those judgments of the Copernican revolution's ‘altered method of our way of thinking’ (B xviii) in general. Kant expressly applies the same ‘altered method’ to natural science as well, at B xii–xiv.

²⁴In this respect, Cohen (1987 [1871]) is in accord with the most influential readings of Kant's theoretical philosophy today. See, for example, emphasis on the epistemological (non-psychological) character of Kant's view in Guyer (1987, 232, 241–6, 303–5, 315–16, et passim). As an indication of the breadth of consensus on this particular approach to Kant, note that it unites otherwise widely diverse readings, such as Strawson 1966, Henrich 1976, Allison 1983, and Longuenesse 1998, as well as Guyer. Kant himself suggests an anti-psychologistic understanding of his notion of the form of experience at Prol. 4: 304, and A 85–7/B 177–19.

²⁵Cohen's endorsement of the Kantian position actually represents a conversion from earlier, psychologistic views, expressed in papers from 1867 and 1869 (reprinted in Cohen 1928, I, 30–140). Köhnke's (1986, 273 ff.) hypothesis that Cohen changed under pressure of considerations that arose in his attempt to overcome Trendelenburg's objection to Kant's theory of the apriority of space and time strikes me as plausible, though full discussion is beyond the scope of this paper.

²⁶Riehl (1925 [1879]) presents an orthodox version of the Kant revival program, but Cohen's account of Kant even became influential in works of this sort by psychological readers. Lange, for example, writes that Cohen's book forced him to embark on a ‘renewed review of the entire Kantian system’ (Lange, 1902 [1876], II, 115) for the second edition of Geschichte des Materialismus, at the end of which he found that he ‘must concur on most points with Dr. Cohen's interpretation’ (Lange, 1902 [1876], II, 130). On the key point at issue here, however, Lange was not won over. He continued to believe that Kantian ‘conditions of the possibility of experience’ should be understood as rooted in our ‘organization’ as physio-psychical beings. This was the position criticized by Cohen in the passage quoted in the text above.

²⁷See Riehl 1925 [1879], II, 8–16. This text raises some doubts about Rath's suggestion that Riehl's logic, epistemology, and theory of science are supposed to depend on ‘psychological means’ (1994, 76) in an unqualified way.

²⁸Köhnke (1986, 58–105, cf. also 23–57) traces the emergence of Erkenntnistheorie as a philosophical discipline back to the early nineteenth century; figures such as Schleiermacher, Ernst Reinhold, F. E. Beneke, and Cohen's teacher Trendelenburg, play key roles in the story.

²⁹Windelband 1907 [1884], 318–54. I cite the papers from Präludien in the pagination of the more widely accessible third edition of 1907, for which many of the papers were revised, but I have checked all claims about Windelband's 1884 views against the first edition.

³⁰‘The Doctrine of Cognition from the Ethno-psychological Viewpoint,’ (Windelband 1875).

³¹In considering reasons for Windelband's conversion, it is worth noting that the same number of the Zeitschrift containing his 1875 paper carried a positive review of Cohen 1871, emphasizing the question of apriority, and also that Windelband's view was criticized in a note by the editor of the journal (Steinthal 1875). While sympathetic to the idea that achievements like the discovery of logical laws have cultural preconditions, and can be subjects for völkerpsychologishe investigation, Steinthal (citing Lotze) insists that such investigations must refrain from making logic itself psychological. This aspect of Steinthal's position is overlooked by Köhnke (1986, 365), who rejects any possible explanation of Windelband's change on internal philosophical grounds, being determined to trace it to Bismarkian political motives, tied to the anti-socialist hysteria that followed assassination attempts against the Kaiser in 1878.

It is somewhat hard to understand what the connection is supposed to be between the present issues and the political concerns emphasized by Köhnke. But we need not pursue the matter in detail in order to reject Köhnke's interpretation, which is definitively vitiated by its unconvincing treatment of ‘Über den Einfluss des Willen auf das Denken’ (1878; reprinted in Präludien as ‘Über Denken und Nachdenken’; Windelband 1907 [1884], 243–77). Köhnke alleges that Windelband's (more or less) bad faith political motives are revealed by the reprint's deletion of a sentence from the original article maintaining that beliefs in the moral/practical realm should not be treated as ‘certain’, but merely as objects of our ‘lively hope’ (Windelband 1878, 288). The passage is supposed to conflict with his Präludien-era view that there are objective and a priori norms (like the goodness of Bismarkian conservatism?) accessible with certainty by immediate intuition; hence the deletion. But, first, the deletion does not change the sense of the surrounding paragraph at all; if anything, its main point (about the problematic but intimate connection between knowledge and non-cognitive interests) is clarified and strengthened – an obvious reason for the edit. What is more important, the overall thesis of the article, while not definitively anti-psychologistic like the paper on critical method, nonetheless fits quite well with the general outlook of the value theory defended in Präludien. In its ringing closing passages, Windelband insists that the effects of the will (and its interests) on thinking must be constrained by a scientific commitment to truth, which first gives our thinking value, and which commits us to thinking ‘as an ethical duty,’ governed by the (fixed) ‘rules of correct thinking’ (1878, 296–7 = 1907 [1884], 276–7; cf. 274–7). Already in this paper, Windelband's conversion to a strict form of anti-psychologism is clearly underway, and his later interest in the dominion of values over thinking is already present (in an essay first delivered in 1877). Thus, it predates the assassination attempts, thereby refuting, not supporting, Köhnke's political explanation of the change in Windelband's views.

³²The quotation is from Kant's Groundwork, where he articulates this assumption for the case of principles of morality, which

must have their source entirely and completely a priori, and, at the same time, must have their commanding authority from this … . Hence everything empirical … is highly prejudicial to the purity of morals, where the proper worth of an absolutely good will … consists just in the principle of action being free from all influences of contingent grounds.

(Kant 1997 [1785], Ak. p. 426, my emphasis)

³³Kant identifies necessity as a definitive mark of apriority from the very beginning of the first Critique; see, for example, B 4, A 1–2.

³⁴Fred Dretske (2000) defends an approach to cognitive norms which treats them as dependent in a similar way on contingent facts about us. On his view, truth is not by itself a normative notion at all, but a purely descriptive relation of isomorphism between representations and the objects they are about. Insofar as there are cognitive norms, they arise only because we happen to prefer representations and beliefs of a certain sort (the true ones), and therefore pursue those. This desire, or end, on our part is a completely contingent fact about us.

³⁵For a complicated, interesting (and largely orthodox Kantian) argument that the binding force of hypothetical imperatives does require something more than simply the meeting of the antecedent condition, see Korsgaard 1997, and note 37, below. Wood (2001) expresses a similar intuition.

³⁶For a compelling discussion of the self-defeating structure of values Nietzsche has in mind, see Reginster 1997. For my point here, of course, Nietzsche need not be right that this analysis captures Christian values. All I require is that there could be such self-defeating patterns of values, and that it may be best for some people to adopt them, despite the resulting practical irrationality.

³⁷Thus, in (1996) Korsgaard argues that, in seeking to ground my reasons for action in a conception of my practical identity, there is an irresistible regress through all possible contingent identities I might have, so that I reach a genuine ground for my reasons only in my non-contingent practical identity as a human being as such. Korsgaard seems to believe in addition (in her 1997) that even the force of the ‘ought’ in a hypothetical imperative, by itself, cannot be explained unless the imperative makes itself necessary for the agent, in the right sort of way. But ‘the right sort of way’ here takes ‘necessary’ to mean not alethic necessity, which is Kant's mark of apriority, but something like ‘rationally necessary’ (Korsgaard 1997, 221), or ‘necessary if I am to be a rational agent’. Korsgaard could still infer the apriority of a norm from such necessity, if in fact my identity as a rational agent is a strictly necessary identity for me, but here we return to the considerations made prominent in her ‘regress of conditions’ arguments (from 1996), which to my mind are the fundamental ones here.

³⁸On this interpretation, in effect, all the imperative force is contained in the consequent, and so it binds (or not) as a simple function of the truth of the antecedent and the validity of the hypothetical imperative as a whole. I admit that in this argument the validity of the hypothetical imperative is assumed; precisely that is the home of the normative force, whose applicability then depends on the (factual, contingent) truth of the antecedent. But again, that assumed normative force gets ‘engaged’ and binds me only given some additional contingent fact (the truth of the antecedent), and this point suggests that some serious argument would be needed to show that its validity must depend on a priori grounds. The considerations in the text are meant to indicate that Korsgaard type regress considerations do not yet do this work.

I would even be willing to concede for purposes of this argument that it might be true that any psychological fact capable of satisfying the antecedent would have to be something like ‘having such-and-such end (for a reason)’, and not merely ‘having such-and-such desire’. (The thought would be that even the hypothetical ‘ought’ always raises the question whether I have reason to satisfy that desire; see Wood (2001) for discussion.) Still, the concession does not yet settle the question of apriority, since it could be a contingent fact about me that I have a reason, for example, to adopt the pursuit of a rich life as an end. The apriority or non-contingency of the basis for every norm follows only if ‘having an (objective) reason’ must be an a priori or non-contingent fact about me. But this is just to assume already the very Kantian intuition in question, that reasons, or norms, must have an a-priori basis.

³⁹There are far-reaching and subtle issues connected with this picture of the way the categories and forms of intuition contribute to an explanation of the normative force of cognition, which I lack the space to pursue in detail here. I provided a sketch of some of the relevant issues and Kant's approach to them in Anderson 2001.

⁴⁰Thanks to an anonymous reviewer for this journal for pressing me to become clearer on the issues raised in the last several paragraphs. For interested readers, further discussion of points concerning the Kantian account of cognitive normativity may be found in Anderson 2001.

⁴¹This space ‘outside’ of nature might take very different forms. Kant himself of course depends on the distinction of the appearances of nature from things in themselves; Korsgaard (1996, 1997, esp. 322–3, 344–54), by contrast, focuses on the difference between third-person, and first-person, or agent-centered, conceptions of ourselves to do much of the same work.

⁴²My thoughts in section 4 are especially indebted to conversations with Nadeem Hussain and Tamar Schapiro, though of course neither can be held responsible for my persisting in some of the views I express here, which they will probably still take to be mistaken.

⁴³Windelband explicitly complains of Kant's ‘dualism, whose overcoming ought to become the most important task of his followers’ (Windelband 1907 [1884], 316).

⁴⁴Allen Wood has suggested (in conversation) that the problematic epistemology pertaining to transcendental (or otherwise radically objective) norms can be seen as one strong motivation for non-cognitivism about value. Since it is so difficult to give an epistemology describing our cognitive access to objective values, it is tempting to conclude that our access to value is not cognitive at all. Windelband's emphasis on the ‘immediate evidence’ of a priori valid values is especially unhelpful from a Kantian point of view, since, qua immediate, such evidence would have to be intuitive in character (see A 19/B 33). But it is a basic starting point of the critical philosophy that we have no intellectual intuition of the sort that could provide access to non-sensible, a priori values (see B 68, B 71–2, A 51/B 75, B 135, B 138–9, and B 145). I reconstruct Kant's argument for the impossibility of intellectual intuition in a different context in Anderson 1998.

⁴⁵See Wood 2001 for a clear current discussion that echoes this strategy.

⁴⁶See Kant's Prolegomena, Ak. 4: 280–5; cf. the Critique, at A 24, B 40–1.

⁴⁷The first two examples figure famously in the strategy of the Critique of Pure Reason; the latter is deployed in similar fashion by Wood 2001.

⁴⁸One way of thinking about the lack of power in Rickert's system of values is to note that his classification of different types of value operates by dividing the class of values into exclusive and exhaustive subclasses, according to the method of traditional logical division. As I have argued elsewhere (Anderson 2004, and forthcoming), such a method produces all and only analytic truths. Kant's own uses of the transcendental method of argument arrive at more powerful, synthetic results precisely because they begin from a more substantively conceived regress base, including ultimately, the possibility of experience.

In addition, it is worth noting that in any merely analytic system of the Rickertian sort, the members of a division must exclude one another. For that reason, once Rickert distinguishes different sorts of value in his system (the primary division separates cognitive, ethical, aesthetic, and personal value) he renders these realms of value completely incommensurable. It is unclear how such a scheme could underwrite the objectivity or rational force of complex normative judgments in which we balance competing values against one another.

⁴⁹The limits of judgments based on this sort of concept containment were identified already by Kant. See Cassirer 1953 [1910], 3–9, and for discussion, Anderson 2004, and forthcoming.

⁵⁰For discussion, see Friedman 2000; 1999, 152–62. See also Richardson 1998, 134–8.