The Regress of Reasons
A problematizing approach to the solutions to the epistemic regress problem
Human beings are believed to be rational creatures. When reflecting on what makes humans unique, Arindam Chakrabarti wrote, “One important way in which the adjective “rational” has been understood in the West is as standing for the differentiating feature distinguishing human beings from other animals.” This assertion does not imply that humans always act rationally but have the power to be rational. Rational behavior is what is expected of us by virtue of our having the capacity to be so. Any deviation from this state warrants the label of irrational. The same does not apply to inanimate objects like rocks and tables or creatures like frogs and fishes because they are arational and have no choice to be otherwise. Why must one think of rationality as a power? One possible answer is that the conscious attainment of knowledge, as knowledge seems to be possible only if one has the capacity of rationality, which allows one to give reasons for the beliefs one holds. This comes from the definition of knowledge as a true, justified belief, widely accepted in the philosophical tradition. It seems irrational to hold a belief that one has if one cannot provide any justification or reason for holding it. We are rational because we act on the basis of our knowledge, knowledge that we have been able to justify through reason.
At a cursory glance, this belief might seem justified. However, a closer look at the nature of justification, formally known as an epistemic warrant, reveals that there has not yet been an unproblematic account explaining its practice. Take, for example, one’s knowledge that coffee tastes bitter. What reason does one have to believe that? Experience. What reason does one have to believe in the legitimacy of experience? Coherence and unity, perhaps. What makes those good indicators of truth? The questions are never-ending as one can question every reason provided and ask for its basis or warrant. This is known as an epistemic regress. The epistemic regress problem is the problem of justifying a belief with another belief that itself requires justification, ad infinitum. As can be observed from this example, the demand for reason implies that one must have a reason for each subsequent belief, because, in the absence of it, holding a belief to be knowledge is unjustified. The book Fading Foundations emphasizes that the point made by the regress problem is not a practical but a principled one. This blog aims to reflect on the nature of epistemic regress and its possible solutions, and see if there is a way to come to an agreement between our principles and beliefs pertaining to justification.
There is a divide in opinion as to what course of action one should take after realizing the regress problem. One might take a skeptical stance and claim that knowledge cannot be attained as we can never justify that our beliefs are indeed true. Another might claim that knowledge is possible because justification is possible. Most thinkers in the history of philosophy have opted for the latter stance. Even here, one has to choose from two seemingly undesirable options. For a justificatory chain in which Y is a knowledge claim and R1, R2, R3, and so on are reasons, one must either believe that the chain continues if one is to know the proposition Y:
Y←R1←R2←R3←R4…….(1.1)
One can, alternatively, hold that the chain terminates at a certain point, but this would call into question one’s claim of knowledge since Y would not be justified.
Y←R1←R2←R3←R4 (1.2)
These two choices lead to three broad possible positions. The first position is foundationalism (1.2), which posits the existence of certain foundational beliefs that are self-justifying or free of the requirement of justification. This position has been supported by philosophers like Aristotle, Rene Descartes and Bertrand Russell. The second position is coherentism (a modification of 1.2), as per which justification has a ‘holistic’ nature, looks like entire belief systems with beliefs that justify each other and is not a linear, one-dimensional chain of reasons. Some influential coherentists are Immanuel Kant, Johann Gottlieb Fichte and Georg Wilhelm Friedrich Hegel. The third position is infinitism (1.1), in which justification can take the form of an infinite chain of reasons. Historically, foundationalism and then coherentism have been popular positions supported by philosophers. Infinitism came to be considered as a serious possibility only recently with defenders like Peter Klein and Scott Aikin.
Aristotle, in his Posterior Analytics, seems to hold the belief that of the two stances — that one must give up on the possibility of acquiring knowledge, or that one must believe that obtaining a justification for every belief is possible, as the chain of justification terminates, the former is wrong, and the latter is correct. While one should not be a sceptic, one should also not expect to conceive of an infinite chain of justifications. The word that Aristotle uses for ‘justification’ is ‘demonstration’, for ‘knowledge’ is ‘understanding’, and for ‘principles’ is ‘primitive’. Following is the reasoning Aristotle gave in Posterior Analytics:
Now some think that because one must understand the primitives there is no understanding at all; others that there is, but that there are demonstrations of everything. Neither of these views is either true or necessary….But we say that neither is all understanding demonstrative, but in the case of the immediates, it is non-demonstrable — and that this is necessary is evident; for if it is necessary to understand the things which are prior and on which the demonstration depends, and it comes to a stop at some time, it is necessary for these immediates to be demonstrable. So as to that we argue thus; and we also say that there is not only understanding, but also some principle by which we become familiar with the definitions.
He believed that those who conceive of the infinite regress as a legitimate problem in acquiring knowledge misconstrue that the starting point for justification is justification. As per what he wrote in his Metaphysics, it is not. He claims that the starting point of knowledge is ‘primitives’ or principles, which are ‘prior’ to justification, and that justification depends on them. This position is known as foundationalism, as it involves being faithful to a foundation composed of certain basic beliefs, propositions or experiences that are beyond doubt. It is on the basis of these foundational beliefs that other beliefs come to form.
Another prominent foundationalist is Descartes. He replaced Aristotle’s demonstrative method with his intuitive one. As per him, Aristotle was wrong in believing that foundational principles could be arrived at through a process of demonstration, which is deductive, as deduction in itself requires a few premises whose truth value is given to start with. Such a ‘givenness’ is not permitted within the Cartesian system of doubt. This is why Descartes proposed that foundational principles gained through self-evident intuition necessarily have a univalent truth value, making any belief system that is built upon them have a strong and indubitable foundation. But what does it mean for certain beliefs to be beyond doubt or ‘self-evident’? Could such a claim be rationally justified? What if there is disagreement amongst people as to which principles are beyond doubt and which are not? For example, theist philosophers like Descartes would include the belief that God exists as a ‘necessary’ being in such a self-evident pool of beliefs, but an agnostic or atheist thinker may disagree. As may a theist thinker who doubts the legitimacy of foundationalism. In such cases, the claim of legitimacy of such a system weakens.
Despite its shortcomings, foundationalism has been a dominant position in epistemology. Even Immanuel Kant seems to understand the prevalence of such a system, though his support for the same is ambiguous. An analysis of The Critique of Pure Reason by the authors of Fading Foundations discusses this:
In his Critique of Pure Reason, he emphasizes that from the fact that every event has a cause, it does not follow that there is a cause for everything. Similarly, from the fact that every proposition has a reason, it does not follow that there is a reason for the entire justificatory chain. Yet, says Kant, humans have a natural inclination to posit such a foundational cause or reason, and Kant’s text does not always make it very clear whether this inclination should be resisted or put to practical use.
As dissatisfaction with the claim of self-evident nature of foundational beliefs arose, coherentism came to be a rising position for addressing the regress problem. Even so, foundationalism was still the position held by most major philosophers, like Rudolf Carnap. Carnap’s foundationalist mission was to construct a logical worldview that relies on certain ‘protocol sentences’, which is another term for foundational beliefs. Coherentism arose because of the lack of desirability of the idea that there exist certain foundational beliefs justify themselves. Coherentist Otto Neurath wrote, “No sentence enjoys the noli me tangere which Carnap ordains for the protocol sentences”. Instead, coherentists are of the opinion that beliefs are always justified in comparison to other beliefs and not restricted to certain foundational principles that have a ‘sovereign standing’. As per this position, the justification for a belief Y might look like:
R2←Y←R1←R2 (1.3)
At a cursory glance, it might appear that the coherentist position is to promote a belief that justifications take the form of one-dimensional loops. However, their system is considerably more intricate. Justification for them is a much more holistic process which involves a complete belief system and not just a few elements of those systems. Coherence of a system exists in degrees as opposed to the absolute nature of a foundationalist basis. One might ask what makes a system more or less coherent. Several coherentists have come up with different criteria to determine the coherence of systems. They lack a consensus as to which criterion is more legitimate. Even so, all of these criteria are called into question by the same criticism which Bertrand Russell put forward:
Whatever the standards of coherence may be, it seems likely that alternative sets of propositions will meet them….although the highly respectable Bishop Stubbs died in his bed, the proposition “Bishop Stubbs was hanged for murder” can readily be conjoined with a whole group of others to form a set which passes any plausible coherence test; and indeed, the same can be said of the propositions that make up any good work of realistic fiction.
This is strong objection which suggests that a coherent system is not necessarily truth-conducive and that even an entirely fictional system with the highest level of coherence could be constructed.
The third approach to the regress problem, infinitism, is significantly less favored by thinkers as compared to foundationalism and coherentism. Infinitists believe that it is not absurd for the justificatory chain to be endless. One could keep on providing reasons for a particular belief they hold. Until the beginning of the 21st century, this position was not talked about or was, at best, mentioned to be dismissed as being unfeasible. The system of epistemology was largely influenced by the foundationalism put forward by Aristotle and Descartes. The dominant foundationalist tradition used the counter-intuitiveness of infinitism as requiring one to ‘go through infinitely many things’ in order to justify something against it. Atkinson and Peijnenburg in their work Fading Foundations, while admitting the counter-intuitiveness of this position, point out factors that work in its favor:
It cannot be denied that infinitism sits well with some modern ideas about the nature of knowledge, such as that knowledge is essentially fallible and that the human search for it is, indeed, without end. Despite many attempts to show the contrary, it is not at all clear how these ideas, which so many of us endorse, can be smoothly combined with foundationalism or even coherentism.
Perhaps the most significant proponent of infinitism is Peter Klein. The reason he gives for supporting this position over the other two is that it is the only one that provides ‘an acceptable account of rational beliefs’. A good justificatory chain is necessarily non-repeating and infinite. He identifies what he calls the ‘two intuitively plausible constraints on good reasoning’, the Principle of Avoiding Circularity (PAC), as per which justificatory chains cannot reasonably be circular, and the Principle of Avoiding Arbitrariness (PAA), as per which justificatory chains cannot reasonably do away with the requirement of reason at any step. Klein, in his work Human Knowledge and the Infinite Regress of Reasons, describes the two principles as follows:
PAC: For all x, if a person, S, has a justification for x, then for all y, if y is in the evidential ancestry of x for S, then x is not in the evidential ancestry of y for S.
PAA: For all x, if a person, S, has a justification for x, then there is some reason, R1, available to S for x; and there is some reason, R2, available to S for R1; etc.
Coherentism violates the former, as it appears to be ‘a thinly disguised form of begging the question’, says Klein, whereas foundationalism violates the latter, as it demands one to accept an arbitrary ‘base’ belief or many such beliefs. Infinitism adheres to both. One might question whether the imposition of such two principles is legitimate. Still, it seems to derive from a generally accepted understanding of knowledge as true justified belief, where justification makes all the difference in the ability of an agent to make a knowledge claim.
Infinitism has its fair share of objections. One of the strongest objections is the finite mind objection, which stems from the intuition that because the human mind is finite, it cannot house an infinite chain of reasons justifying a belief, let alone multiple such chains for multiple beliefs. Klein attempts to address this objection by pointing out that what exactly is understood to be or qualifies as a ‘belief’ is ambiguous. Two interpretations that he believes are acceptable are that it refers to a repository of already formed beliefs, or that it signifies a human’s ‘disposition to forms dispositions’. An infinitist, he believes, takes on the second meaning because that would allow for there to be an infinite possibility of beliefs. A second objection is that infinitism is akin to skepticism as it fails to ‘settle’ an issue. This, Klein points out, stems from a Cartesian foundationalist system of epistemology. It posits the existence of certain a-priori principles that need to be foundationally indubitable. Klein conceded that even though infinitism does not provide a ‘final guarantee’ of the truth of a belief or a system, foundationalism seems to do it only superficially and not genuinely as it claims. Descartes believes that the foundational ‘first principle’ should be accepted because they are ‘clear and distinction’, but what makes ‘clear and distinct’ an acceptable criterion? Without a reason, this criterion lacks legitimacy. Klein’s replies to these two objections seem convincing, even if they cause a sense of unease at losing the level of certainty we associate with knowledge, the level that foundationalism and coherentism claim to provide. One might question the motivation for coming up with a foundationalist or a coherentist system. Is it because it helps one ally one’s ‘principles’ and practices? Is that a good enough motivation? Would an introduction of a system to serve as a resolution of a problem and not on its own accord diminish its legitimacy? In other words, it seems that the construction of a system to improve its explanatory power or coherence by fitting the existing beliefs into it seems like a weak ground for creation. Rather, if one looks at the infinitist point of view, it is not doing any such thing. Infinitists do not seem to modify how we reason. Their counterintuitive belief that justification chains can be infinite seems like the only position that seems to do justice to how we generally understand knowledge — as a belief whose truth is proved through justification.
Sources
Aristotle. Posterior Analytics. Oxford University Press, USA, 1994.
Aristotle. The Metaphysics. Penguin UK, 2004.
Atkinson, Doug, and Jeanne Peijnenburg. Fading Foundations. Springer eBooks, 2017. https://doi.org/10.1007/978-3-319-58295-5.
Carnap, Rudolf. The Logical Structure of the World and Pseudoproblems in Philosophy, 1967. https://ci.nii.ac.jp/ncid/BA65389346.
Descartes, René. Descartes: Meditations on First Philosophy: With Selections from the Objections and Replies. Cambridge University Press, 1996.
Deutsch, Eliot, and Ronald Bontekoe. A Companion to World Philosophies. Wiley Blackwell, 1999.
Kant, Immanuel. Critique of Pure Reason. Cambridge University Press, 1999.
Klein, Peter. “Human Knowledge and the Infinite Regress of Reasons.” Noûs 33, no. s13 (October 1, 1999): 297–325. https://doi.org/10.1111/0029-4624.33.s13.14.
Nottelmann, Nikolaj. “Otto Neurath on the Structure of Protocol Sentences; a New Approach to an Interpretative Puzzle.” Journal for General Philosophy of Science 37, no. 1 (March 1, 2006): 165–86. https://doi.org/10.1007/s10838-006-0484-2.
Plato. Theaetetus. Hackett Publishing, 2012.
Russell, Bertrand. “II. — On the Nature Of Truth.” Proceedings of the Aristotelian Society for the Systematic Study of Philosophy 7, no. 1 (June 1, 1907): 28–49. https://doi.org/10.1093/aristotelian/7.1.28
