In this essay, I will present and defend a version of modest foundationalism concerning epistemic justification. In order to defend it I will consider some possible objections coming from the competing positions of classical foundationalism and coherentism. However, as both of these approaches involve serious difficulties, I will counter these objections and show the advantages of modest foundationalism. At the end, it will, hopefully, have become clear that a modest foundationalism is able to integrate also coherentist as well as externalist intuitions to a certain degree, which enables it to be a far more plausible position than any of the extremes.
I have to note an important decision that precedes my critical examination of these opposing theories and my plea for modest foundationalism. This inquiry is based on the meta-epistemological point of view according to which the task of a theory of epistemic justification is a descriptive one, in a sense analogous to the distinction between descriptive and revisionary metaphysics drawn by Peter Strawson in the introduction of his Individuals1 Therefore, I take it to be one of the tasks of epistemology to explicate the standards underlying our ordinary use of the word ‘justified’. This preliminary decision concerning the goal and method of epistemology determines the evaluation both of classical foundationalism and coherentism in the following discussion.
2. The Way to Modest Foundationalism
2.1 Internalism, Externalism, and the Regress Problem
Before one can grasp the idea and the need for a modest kind of foundationalism as a somewhat modified version of it, one first needs to understand the genesis of a foundationalist approach to the problem of explaining justified beliefs. Thus, a sketch of the so-called regress problem will help us to gain insight into the field of possible solutions to the wider problem of epistemic justification.
The regress problem arises if one demands for a true belief B of an epistemic subject S to count as knowledge (in contrast to mere opinion) that S also have evidence which makes it likely for B to be true. For S to be able to adduce good reasons presupposes that S have reflective access to these grounds. This position is called in epistemology internalism (according to a kind of spatial metaphor that suggests that all it needs for a belief to be justified is to be looked for only within the reach of the subject’s awareness). The externalist counterpart does not make this requirement. For the externalist, all there needs to be is a relation of some sort (be it causal or reliable in a wider sense) between a fact F in the world and the belief B of a subject S about F that makes it likely that B is true. According to the externalist, one does not need to know how one knows or that the beliefforming process is reliable in order to be able to know something. We will put the externalist strategy aside for a moment to come back to the issue of how and why internalist theories are confronted with a regress problem.
If justification is a question of grounding, as the internalists suggest, this could possibly lead us into a situation similar to the Münchhausen trilemma. When asked for reasons for a belief B1 one may answer by stating other beliefs like B2 and B3. However, if we ask for reasons for B2 and B3, for the reasons of those reasons, and so on, like children, scientist or philosophers tend to do sometimes, then we fall into a regress. This regress can either be infinite or circular or end arbitrarily, according to this trilemma. Applied to the question of how a belief is justified this means that in the case of an infinite regress there would be no final basic belief transferring justification up in a chain of beliefs; in the case of circular justification the belief in question would have to justify itself (whether directly or indirectly); however, there are two possible cases for noncircular finite justificational chains: either the basic belief at the end of the chain which grants the following non-basic beliefs their justifiedness is itself justified or it is not. Distinguishing the last two may not be easy.
2.2 The Regress Argument
Out of this regress problem arises the so-called regress argument. It is an argument in favour of the position we will come to know as foundationalism that functions by elimination of three of the aforementioned options concerning the regress.
The first option we can dismiss is the infinite regress. This option seems to be psychologically impossible because a finite person just does not have infinite many beliefs. Moreover, always shifting the grounding even for simple empirical beliefs involves the risk of falling into a sceptical stance.
Secondly, the option of circular justification can be rejected insofar as one generally assumes a linear conception of justification. This view seems to be plausible regarding common sense. Therefore, it seems to be in line with the meta-epistemological commitments I outlined above. However, as we will see, coherentists challenge the linear model of justification.
The third option is easily rejected. It seems utterly implausible that an unjustified belief would confer justification on other beliefs. Even if the supposedly justifying basic belief were true, it could count as unjustified because of the way it came about (e.g., mere guessing). At this point, I think it is wise to incorporate some externalist intuitions into the foundationalist framework, i.e., acknowledging the significance of the reliability of belief-forming processes. The externalist does not exclude the possibility of us sometimes knowing how we formed our beliefs and whether we complied with our epistemic responsibilities; thus, in those cases, we can use this for the evaluation of the justificational status of our belief.
Although the regress argument only treats these three options in a negative way, it concludes that just one viable option is left. This option is foundationalism, i.e., the position that there are justified basic beliefs on which all justified non-basic beliefs depend.
2.3 Modest Foundationalism
In contrast to the common way of presentation I will introduce modest foundationalism before classical foundationalism. I do that out of the conviction that it can be formulated independently of classical foundationalism and that it approximates our everyday standards for the use of the word ‘justified’ way better wherefore I take it to have priority over classical foundationalism. Moreover, I am quite sceptical about the fact that classical foundationalism, as the name suggests, is supposed to precede modest foundationalist ideas in the history of philosophy (at least if one assumes that classical foundationalism commences with Descartes). However, the latter would have to be examined more indepth in a historical inquiry which is beyond the scope of the present essay.
Modest foundationalism, now, is a position that combines the two basic claims of foundationalism with two more specific theses. To remind us, foundationalism in general claims (1) that there are non-inferentially justified basic beliefs and (2) that the justification of all non-basic justified beliefs depends on those basic beliefs. The two additional theses of modest foundationalism state that (3) for a belief to be justified it is not necessary for it to have come about infallibly (i.e., to be an instance of knowledge) and (4) that justification can be transferred to non-basic beliefs in a variety of ways (deduction, induction, inference to the best explanation, etc.). Provided that one has a rich concept of experience—like McDowell, according to whom conceptual capacities are always involved in our so-called inner and outer experiences —, one can understand how it can serve as a source of justification. Furthermore, it seems to be a reliable source in many circumstances and this is often all that we need. This shows that externalist intuitions can be integrated into the theoretic framework of modest foundationalism in a fruitful way.
In the next sections, I will clarify the advantages of this position against the backdrop of the dispute between modest foundationalism, classical foundationalism, and coherentism.
3. Objections to Modest Foundationalism
3.1 Modest Foundationalism vs. Classical Foundationalism
Classical foundationalism would object to modest foundationalism that the criteria it sets for beliefs to be justified are too weak. According to classical foundationalism, a basic belief need be acquired in an infallible fashion in order for it to count as justified. That means that a belief would only be justified if one cannot be mistaken about the proposition in question.
In light of the meta-epistemological framework I outlined earlier, classical foundationalism seems to set too high standards for justification which results in a counterintuitive restriction of the set of possible justified basic beliefs (and their sources). Simple perceptual beliefs and memory beliefs would be excluded from this set because they cannot be acquired with absolute certainty, as Descartes showed in his Meditations by his methodological scepticism (e.g., the evil deceiving demon as the worst case). As I think, similar to KANT in the introduction to his Critique of Pure Reason, that our knowledge of the world begins with these perceptual beliefs, and certainly not with indubitable a priori beliefs, it would be quite a revisionary view on the justificational status of our beliefs that even achieves the opposite of what it set out to do, namely, playing into the hands of the sceptic by demanding too much of finite epistemic subjects who usually do not go through the whole reflection we find in Descartes.
A second objection the classical foundationalist could raise is that the modest foundationalist is too liberal concerning the transfer of justification from basic beliefs to non-basic beliefs. The classical foundationalist would insist that the only sure way to confer justification is deduction. However, this again seems too restrictive because many beliefs we usually count as justified (especially empirical beliefs) either came about via inductive or abductive forms of reasoning or by no inference at all. Insofar as epistemic justification is an evaluative notion, and considering the fact that we strive to ground our actions on beliefs which attained a positive epistemic evaluation in this sense, it seems reasonable to assume that the prevailing positive evaluation of non-deductively acquired beliefs indicates their significance for our everyday life. A theory of justification which does not live up to this practice is unrealistic. Deduction may be one way to draw conclusions but it is and it should not be the only legitimate form of justification. Again, the strict account of justification presented by classical foundationalism opens the door for the sceptic who could put into question whether there are any beliefs about the world at all that are justified if deduction is the only way to go.
In the next section, we will discuss objections from the other side of the theoretical spectrum. As we will see, coherentism offers some critical insights which concern both classical and modest foundationalism.
3.2 Modest Foundationalism vs. Coherentism
Coherentists base their position on an observation that they make about our practice of justifying. According to them, we justify our beliefs by citing other beliefs. They draw the radical conclusion from this fact that only beliefs can justify beliefs. Therefore, the mutual support of beliefs ensures their justifiedness.
I think this idea is based on an insufficient analysis of our justificational practices and their underlying assumptions. For, often we stop the regress of giving reasons by stating that we have had a perception or memory concerning the fact in question, or are acquainted with it through reliable testimony. Of course, in principle we could state some further meta-beliefs that tell more about how we evaluated the reliability of the belief-forming processes involved, but in practice that is rarely done. Nevertheless, we usually seem to assume that our beliefs are justified in virtue of the reliability of our cognitive capacities and that of others. Thus, according to our practice of giving and asking for reasons, this is sufficient for the justification of many beliefs (one could count as basic). This shows that we do not assume that our beliefs are justified only by other beliefs but also by the operations of our cognitive capacities that manage to ground these beliefs in facts. Thus, our practice implicitly involves externalist and foundationalist assumptions.
Furthermore, in contrast to foundationalism, coherentism seems to lack grip to the world. McDowell coined a famous phrase for this feature of coherentism: “Frictionless spinning in a void”. It is counterintuitive that two sets of beliefs, A and B, which exhibit the same degree of coherence but differ insofar as A is additionally supported by experience, should count as equally justified. Modest foundationalism, in contrast, can account for this difference which we usually see in those cases. Yet, modest foundationalism does not exclude consideration of coherence. On the contrary, it accepts coherence as an important factor for justification. However, modest foundationalism insists that there are some beliefs which are justified at least to some degree independently of the support they get from other beliefs.
I set out to present and defend a version of modest foundationalism in this paper. As we have seen, it has several advantages over classical foundationalism and coherentism and is able to stand their objections. Moreover, I made clear how modest foundationalism can be construed in a way that allows us to extend the traditional internalist and foundationalist framework in order to integrate an externalist as well as a coherentist dimension into it. Thereby, we attain a powerful theory which conforms to our linguistic and justificational practices and, accordingly, our meta-epistemological commitments.
Bonjour, Laurence; Sosa, Ernest. 2003. Epistemic Justification: Internalism vs. Externalism, Foundations vs. Virtues. Malden, MA: Blackwell.
Elgin, Catherine. 2014. “Non-foundationalist Epistemology: Holism, Coherence, and Tenability”. In: M. Steup, J. Turri, E. Sosa (Eds.): Contemporary Debates in Epistemology (Second Edition). Malden, MA: Blackwell, 244-255.
Goldman, Alvin. 1979. “What is Justified Belief?”. In: G. Pappas (Ed.): Justification and Knowledge. Dordrecht: Reidel, 1-23.
Klein, Peter. 2014. “Infinitism is the Solution to the Regress Problem”. In: M. Steup, J. Turri, E. Sosa (Eds.): Contemporary Debates in Epistemology (Second Edition). Malden, MA: Blackwell, 274-282.
Lemos, Noah. 2007. An Introduction to the Theory of Knowledge. Cambridge: Cambridge University Press.
McDowell, John. 1996. Mind and World. Cambridge, MA: Harvard University Press.
Nagel, Jennifer. 2014. Knowledge: A Very Short Introduction. Oxford: Oxford University Press.
Strawson, Peter. 1959. Individuals: An Essay in Descriptive Metaphysics. London: Methuen.
 Cf. Strawson 1959, 9-11.
 Cf. Goldman 1979.
 In the presentation of internalism and externalism I followed roughly Nagel 2014, 61.
 Peter Klein defends such an infinitism (Klein 2014).
 Cf. Ch. 1.2 in Bonjour/Sosa 2003.
 The reconstruction of the regress argument was mainly inspired by Lemos 2007, 47-49.
 I think it is wrong to ascribe (infallibility to beliefs (or even knowledge) as Lemos and others do. In contrast to these uses I take it to be a property of persons (that includes certain facts about their capacities).
 He developed this account famously in Mind and World (McDowell 1996).
 However, I assume that we are also aware of its limits (although that is a vague matter).
 Cf. Lemos 2007, 51.
 For us finite beings that is often enough to get by.
 Cf. Lemos 2007, 55.
 Although there is no universally accepted criterion of coherence, at least the following is required: the components must be consistent, cotenable and supportive. Since the last two are matters of degree, there remains the question of how coherent an account must be in order to be justified. Coherence conduces to epistemic acceptability only when the best explanation of the coherence of a constellation of claims is that they are (at least roughly) true. For a defence of this account see Elgin 2014.
 As Lemos notes, coherentists fail to argue for the thesis that our activity of justifying beliefs determines what it means for a belief to count as justified. (Lemos 2007, 74) Modest foundationalism, on the contrary, allows us to draw this distinction.
 Always requiring such meta-beliefs can also lead into an infinite regress and, consequently, to scepticism. (Cf. Lemos 2007, 76-78)
 Cf. Lemos 2007, 80,
 McDowell 1996, 11.
 Cf. Lemos 2007, 80-81.