Leaning from the steep slope: On coherence in response to Professor Jean d’Aspremont

1. Introduction

I am grateful for the opportunity to, as far as I am able, clarify certain issues and elaborate on others that arise from the response to my article published in this issue. ¹ In this response my article is characterized as being articulated around three moves, each of which comes in for criticism:

1. 1. A move towards meaning;

2. 2. A move towards context; and

3. 3. A move towards discipline.

In the following pages I would like to address a number of the issues raised and present my point of view. In doing so I will highlight where I believe my article has been misunderstood or misconstrued, and where I suspect we simply run up against principled disagreement.

2. A move towards meaning

The first main criticism, as I understand it, relates to my deployment of the concept at the heart of the theoretical framework put forward in the article, namely explanatory coherence. First, although set out explicitly in my original piece, it bears repeating that my central concern was to advance a coherence theory of epistemic justification. That is, the issue of when we are justified in taking a proposition to be true. In other words, I was concerned with the normative question of when, if ever, we can say that the Court’s factual determinations are justified, given the particular context in which it operates and the particular constraints to which it is subject. This is important. I (certainly) did not advocate a coherence theory of truth, nor did I set out to provide an intellectual history of the concept of coherence, nor to make any claims about coherence as a concept more generally beyond those plainly (and I hope clearly) laid out in my article.

2.1 On definition avoidance and the supposed conceptual emptiness of coherence

It is suggested that I engage in ‘sophisticated definition-avoidance’, and manage to ‘elude any definition of coherence’, ² and that the coherence principles set out in the article do not shed much light on how inference to the best explanation (IBE) would work if the framework were to be adopted. First, let me say a word on why I do not believe it to be the case that I avoid defining coherence.

At the heart of the framework, which draws primarily on the work of legal philosopher Amalia Amaya and epistemologist Paul Thagard, is the concept of explanatory coherence. ³ In accordance with this form of coherence, as stated on page 6 of my article, ⁴ something can be considered to be coherent when it maximizes the satisfaction of a set of positive and negative constraints. ⁵ This might not seem like the most intuitive conception of coherence, so it may be worth restating the fundamentals of the explanatory coherence framework.

Explanatory coherence is a tool that we can use to make sense of something when provided with a series of conflicting material, or as they are otherwise known in the literature, elements. ⁶ In an effort to find the best available explanation, these elements can be shown to either cohere or not cohere among themselves. If the elements cohere there is a positive constraint and if they do not cohere there is a negative constraint. To elaborate, a positive constraint, to return to the Pulp Mills example that I set out over the course of five pages in my article, is when two or more of the elements (claims of the parties) go together, such as the fact that there was discharge from a pulp mill constructed on the River Uruguay, and the claim that such discharge can cause eutrophication (excessive plant and algal growth). These two elements go together, and as such both should be accepted, or both should be rejected by the Court. When two elements are either accepted or rejected, they are considered ‘satisfied’.

In contrast, a negative constraint is when two elements (claims of the parties) do not go together, for example the claim that eutrophication happened because of human waste from a festival upstream, and the claim that eutrophication was caused by discharge from the pulp mill. In this case the two elements should not go together – if one is accepted then the other should be rejected. When one element is accepted and the other is rejected it is also then considered ‘satisfied’.

Through this process, elements (claims of the parties), are separated into those that are accepted and those that are rejected. Explanatory coherence, we will recall, is defined as a process of maximal fulfilment of a variety of constraints. In deciding which elements should be accepted and which should be rejected, the fact-finder is guided by a number of principles which are relevant to the particular type of coherence (in our case explanatory coherence). These principles are set out in my article, and while I do not have the space to run through each element in more detail here, one example may be helpful.

The principle of competition: this is where two elements (claims of the parties) independently explain the evidence, but compete with one another. For instance, the claim that the eutrophication at issue in the Pulp Mills case was caused by discharge from the pulp mill, and the claim that it was caused by the festival, both independently explain the evidence (the eutrophication). However, for the purposes of the adjudicative process they cannot both be said to explain the evidence, and so they can be said to represent a negative constraint. This means that one of them must be rejected, i.e., the Court must choose which of the two claims it believes best explains the evidence, taking into account the particular institutional setting in which adjudication is taking place (including the issue of which party bears the burden of proof, and the applicable standard of proof).

Once the elements have been satisfied, in other words once they have been separated into accepted and rejected elements, an overall assessment of the satisfied elements is made. Through this assessment we can consider a particular claim to be the most coherent, provided there is no other way of dividing the elements to produce an alternative set of accepted elements that has greater weight.

I hope that this brief restatement of the framework serves to illustrate what I believe to be its contribution, namely a theoretical account of something that international courts must do in every case involving contested facts. That is, weigh the competing claims of the parties and justify why they prefer particular claims over others. I also hope that this example shows how the framework for explanatory coherence would work in practice and brings out more clearly the role of explanation, and begins to address the alleged self-referentiality of the notion of coherence. I would like to add a few words on the latter point.

2.2 On the supposedly circular nature of explanatory coherence

The claim of vicious circularity is one that coherence theories have long had to contend with. ⁷ In relation to my article in particular, it is suggested that explanatory coherence is circular because the eventual selection as justified of the most coherent explanation is a process which involves reference to various constraints and coherence principles. I would suggest that such an argument only holds any water if one envisages a linear process of justification whereby justification moves along a sort of ‘pipeline’ from A to B, with more foundational beliefs supporting and justifying other beliefs further down the line. ⁸

However, this is not the way coherence relations (and here I am speaking about the notion of coherence more broadly, not just about explanatory coherence) work. Rather, an assessment of coherence in its various forms is made in a holistic manner, ‘so that justification emerges from mutual support among the whole body of beliefs’. ⁹ As such, the fact that an assessment of coherence is made by reference to other constraints and coherence principles is exactly the point – it is a feature not a bug. That is not to say that this ‘web of belief’ approach to justification somehow solves the regress problem (to which we will return in a moment), but it does highlight the assumption of linear justification that underlies such critiques of coherence.

All of this having been said, I can to a certain extent see the concern relating to vagueness of coherence as a notion. This is particularly so if care is not taken, and the sets and elements are not fleshed out in sufficient detail. As Amaya readily admits, ‘[c]oherence-based reasoning boils down to intuitionism unless we have a clear account of the inference processes that are at work in coherence judgments’. ¹⁰ This to me highlights the importance of providing a clear and detailed account of how any coherence framework would operate when applied to a particular problem. And indeed, this is one of the motivations for the project; to do exactly that.

2.3 On deferred meaning

Finally on the group of criticisms relating to the move towards meaning, Professor d’Aspremont asks: ‘… what is a judgement of the Court if not an elegant textual assemblage that constantly pushes back the meaning of a word or a text to another word or text [sic][?]’. ¹¹ I would like to say a brief word on this provocation. ¹² The central idea here, drawing on the work of Derrida and other postmodernist theorists of that ilk, ¹³ is that meaning is never settled but rather constantly deferred to other terms which are themselves in need of definition (and so we carry on eternally until one is left at the bottom of the swimming pool with all the water out of it). ¹⁴ Accordingly, modernity has failed us, and any attempt to achieve a fixed definition of coherence is doomed to fail.

In response, I would point out that I never suggested that there was such a thing as an objective, transcendental concept of explanatory coherence. Rather, my conception of coherence is one that is arrived at through a process within a particular institutional setting (international adjudication). I made no attempt to link my specific notion of explanatory coherence relating to the fact-finding process of the ICJ to coherence as a notion more generally.

Within this specific institutional setting, I maintain that intersubjective meaning is possible, in other words that we can talk of a factual claim being more or less coherent than another without falling into complete meaninglessness. ¹⁵ And in fact I believe that the practice of counsel and judges who participate in this process in this particular institutional setting serves to show that we are comfortable, at least to a certain extent, in engaging with concepts that are not completely determinate. Some concepts are more determinate than others, and this is most often the result of their contestation at a particular point in time. ¹⁶

That may mean that, say, the meaning of ‘armed attack’ under Article 51 of the UN Charter, a notion highly contested by states and other actors due to its political importance, may be less determinate than, say, some technical concept in the law of the sea. But the choice is not between complete determinacy (which is impossible to achieve) or complete indeterminacy. It is possible to say that something is more or less coherent than something else. This of course leaves room for contestation, but it is certainly not arbitrary or meaningless. ¹⁷

Finally, it is perhaps interesting to consider the implications of deconstruction for the establishment of the facts within the adjudicative system. ¹⁸ Although those engaging in deconstruction by definition do not attempt to offer any practical proposals, it is tempting to consider what this approach would mean for the fact-finding process. Consider, for the sake of argument, the position taken regarding the eternal deferral of meaning. What would this mean for the actors involved in the process such as the international judge or the parties themselves? What would the written and oral submissions of the parties look like regarding the facts contested by the parties? Perhaps the parties would not feel able to contest the facts at all, were they to accept that all meaning is eternally deferred. And how would the international judge be able to carry out their duty of making factual determinations to facilitate the legal syllogism? Such questions, I would suggest, illustrate that the deconstructive project is not a very fruitful way of thinking about the object of my inquiry. ¹⁹ This leads me on to the second main second criticism, relating to the move to context.

3. A move to context

Professor d’Aspremont criticizes the use of context in my article for three reasons: (i) that my article exhibits the ‘the elementary belief that context exists out there and can be ascertained independently from the text or practice whose interpretation it supports or guides’; (ii) that sensitivity to context exhibits ‘some very crude empiricism’, i.e., the idea that there is a ‘reality-in-itself’ which is ‘accessible and cognizable provided that the right methods are used’; and (iii) the role context plays in the framework is a further facilitator of the deferral of meaning. Since the last point was already addressed in the previous section, I will not say anything more on that for the moment.

I would like to say a little bit about why I believe the suggestion the coherence framework is related to a ‘never defined context’ is wide of the mark. Now, it is entirely fair to say that context plays a pivotal role in my article – the aim of which is to explore the question of epistemic justification. The approach to epistemic justification adopted in the article, and perhaps this could have been brought out a little more clearly, is one with context as its fulcrum. Let me explain what I mean.

The traditional approaches to epistemic justification, i.e., to accounts of the justificatory structure of an individual’s beliefs, are foundationalism and coherentism. ²⁰ Coherence theories of epistemic justification have been advanced in response to perceived weaknesses in the traditional foundationalist account in accordance with which a belief in the ‘superstructure’ is justified by a belief in the ‘foundation’ – these so-called foundational beliefs ‘possess a degree of non-inferential justification in virtue of which they constitute the ultimate foundation of our knowledge’. ²¹ Coherentists argue that beliefs are justified ‘in virtue of their relations with other beliefs, specifically, in virtue of belonging to a coherent system of beliefs’. ²² However, coherence theories themselves suffer from their own weaknesses, including vagueness, circularity, conservatism, and so on. ²³ As such, both the foundationalist and coherentist approaches to epistemic justification are beset by significant flaws.

It is as a result of these flaws that context-sensitive theories of epistemic justification have been advanced. ²⁴ As illustrated throughout the article, but particularly at Section 4.3., I believe that a form of coherence theory of epistemic justification that is sensitive to context is the most appropriate for application to the adjudicative process. Contextualism adopts a third approach to the regress problem, namely that there exist certain beliefs that are not in need of justification in certain contexts – i.e., contextually basic beliefs. ²⁵ As Annis has stated, ‘a belief is contextually basic if given an issue-context the appropriate objector-group does not require the person to have reasons for the belief in order to be in a position to have knowledge’. ²⁶ This type of epistemic justification works well for the adjudicative context, where parties to a contentious case are not required to prove facts ‘all the way down’ – but rather can point to certain accepted beliefs, such as logic and mathematics and so on, and expect that the judge can consider the stronger of the factual claims to be justified against that background. ²⁷

This is a recognition of the fact that factual inquiries are carried out for different purposes, instrumentally. In the context of the adjudicative process factual inquiries are carried out to establish the operative facts from which to draw normative conclusions. As I stated in my article, the justificatory standard within the adjudicative context is a reflection of the fact that the establishment of the facts is just one of a number of goals that the process strives to achieve in a contentious case, including the peaceful settlement of the dispute.

And this is a point that I would like to reiterate. I have never claimed that there is one transcendental ‘context’ that we can discover provided we just use the correct method. All I mean to say is that the method for arriving at factual determination is different for scientists in a laboratory in, say, the Charité – Universitätsmedizin Berlin, than it is for ICJ judges sitting in a court in The Hague. I would suggest that these specific concerns about context do not apply to the argument made in my article, which merely points out that there is no global standard of justification – that justification is sensitive to the context in which fact-finding is carried out.

4. A move to discipline

Thirdly and finally, I would like to address the criticisms of my article related to epistemic responsibility. As set out in the article itself, explanatory coherence in itself depends to a large extent on the responsible behaviour of those involved in the process. Professor d’Aspremont calls into question the idea of epistemic duties, and further contends that I have created a creed for international judges and what they should believe, and as a result end up ‘performing the role of a priest who dictates to judges what they should believe’ as well as desiring ‘to be among those who compose the very prescriptions constitutive of the international legal discipline’. I would like to reassure the reader that neither suggestion could be further from the truth. ²⁸

Epistemic duties are important for my coherence framework due to the fact that, as I make clear in my article, coherence in itself does not provide a route to justification in a way that is indubitably truth-conducive. In fact, it remains an issue of philosophical controversy whether the sceptical challenge can ever be fully met in relation to the relationship between belief and truth. ²⁹ That said, the role that epistemic responsibility plays in Amaya’s framework, and in my own, is providing a route to making factual determinations that are true or approximately true.

Relatedly, it is crucial to acknowledge that, in accordance with my framework, the justification of a factual determination that comes from an inference to the best coherent explanation is very much dependent on the behaviour of the actors involved in the process. In contrast with formal, theoretical types of reasoning such as deduction, ‘the correctness of a piece of coherence-based reasoning … depends upon subjective features of the subjects who perform the reasoning’. ³⁰ And again this mirrors the adjudicative process, where sub-optimal factual determinations are sometimes made because crucial evidence is not put before the court, or because a lawyer makes a hash of arguing their case, or the judge makes an incompetent or unfair assessment of the facts. How seriously a judge takes their responsibilities as a judge has a material effect on the quality of their factual determinations, and in the same way the behaviour of the subject in the process of epistemic justification has the power to make an impact.

This is where epistemic responsibility comes in. Rather than telling judges what to think, surely the definition of a futile endeavour, what I mean to say is exactly what I said in my article, namely that there exist certain duties which are related to the various roles that we all perform in our lives. Just as we ascribe certain duties of parents towards their children or teachers towards their students, we expect certain things of international judges qua their role as judges. And I do not believe those duties to be particularly controversial. Take for instance the duty that we ascribe to judges to favour claims about facts that are supported by evidence over those which are not. How could it possibly be any other way? Could one seriously contest that it is the role-related duty of the judge to, taking another example, revise their beliefs in light of new evidence?

5. Conclusion

Much more could be said about all the issues raised in response to my article. Nevertheless, as I said at the beginning, I am grateful to all involved for the chance to clarify certain things and elaborate others. It is likely that a layer of principled disagreement will remain, but hopefully readers will now be better equipped to decide which account they find to be the most coherent and which they find to be most quixotic.