Presumptions in Rational Inquiry

Without some recourse to presumption we simply could not manage to obtain the informative inputs indispensable for answering our questions about factual issues. Some degree of presumption (however small) incredibly attaches to our procedurally validated truths throughout the factual area where defeasibility is virtually inescapable. This becomes clear once one considers the issue of the observational inputs essential to inquiry in the factual domain. If their status always had to be that of established truths, we could never get off the ground, since observational experience just does not work like that. Thus if an “ordinary process of inferential reasoning” were at issue here, we would be in deep trouble as any such process can only extract truths from truths, and if the inputs themselves had to be established truths, the whole process would be vitiated at the outset. Happily this is not so. An inquiry procedure is clearly not an “ordinary process of inference”: it must afford an originative mechanism, capable of yielding an output of (putative) truths without demanding an initial input of previously established (putative) truths. Clearly if we hewed to the line that rationally discursive procedures can only extract truths from truths, we would be offered the unattractive choice between (1) accepting a “starter set” of nondiscursively self-evident or self-validating truths, or (2) a skepticism that admits defeat and gives up the whole project of a rational validation of truth-acceptance. This dilemma can be addressed effectively by resorting to presumptions.

Introduction

Theorists of knowledge have long recognized and emphasized that there are significant differences between theoretical and practical reason, between problem solving in purely theoretical matters (where nothing is at stake save the possibility of mistaken beliefs), and in practical matters (where actual harm of some sort might be incurred). There is, perhaps, no more striking illustration of this situation than the little heeded issue of the treatment of very remote possibilities – those whose probability is extremely small. Here the question before us is this: in regard to matters of practice, should a diminutive probability (one of an effectively infinitesimal magnitude ∈) be seen as being indistinguishable from zero and treated as having no magnitude at all? In deploying probabilities in expected-value comparisons that serve as guides for decision making, could and should we adopt theequation: ∈ = 0? Are we to dismiss those minute probabilities and presume them to be zero? Sometimes the answer is a decided affirmative. And this chapter accordingly addresses one very special case of presumption that answers in relation to our planning for the future, namely, the dismissal of diminutively far-fetched possibilities from our reckoning.

Effectively Zero Probabilities

Probability and cognitive presumption are closely interconnected. The step from probability to presumption is mediated by the consideration that when answers are needed the probability of the available alternatives can help to provide guidance.

Most philosophical theorists of knowledge – the classical rationalists (like Descartes and Spinoza) and empiricists (like Hume) as well as various later thinkers from Brentano to our contemporaries – have been concerned primarily with knowledge in the sense of what we know for certain. In consequence it has transpired that – apart from forays by mathematically oriented theorists such as F. P. Ramsey, J. M. Keynes, and R. Carnap into the domain of probability – contemporary epistemology has almost wholly neglected the range of conceptions in the region of uncertainty, the gray area of concepts that, like presumption and plausibility, have an indecisively tentative impetus toward truth.

The quest for certainty has continued to exercise a virtually hypnotic fascination on contemporary theorists of knowledge, who have, accordingly, tended to neglect cognitive claims that stop short of pretentions to definitive finality. But by rights epistemology should surely be seen as having a far broader range, concerned not just with knowledge as such but with a great variety of cognitive matters such as belief, conjecture, supposition, and much else. And it is just here – in the theory of cognition at large – that the concept of presumption comes into its own and provides a healthy antidote to the fundamentalist epistemologies whose inadequacy to provide an adequate theoretical basis for rational inquiry has become increasingly evident in recent years.

Default Inference

The topic of default reasoning also affords instructive insights into the nature of presumptions. A default in logic is a fall-back position in point of conclusion-drawing – one to which we can appropriately resort when circumstances prove uncooperative. But of course things ought not to go wrong in logic. So what is going on here?

Orthodox inferential reasoning proceeds via logically valid inference processes which do – and must – proceed from true premises to true conclusions. They review the inherent connections coming actually or assumptively accepted commitments. Logic functions within the limits of the given. By contrast, default reasoning – which involves an information gap between premises and conclusion – goes beyond this into uncharted territories. In consequence, plausible (though sometimes false) premises will lead to plausible (though possibly false) conclusions.

The logical validity of inference rules in standard (truth-functional) logic is determined on an input-output basis, a valid rule being one that will invariably yield true outputs (conclusions) from true inputs (premises). All such inference rules will faithfully and unfailingly transmit the truth of premises to the conclusions. By contrast, the inference processes of default logic are such that the truth – real or suppositional – of the premises does not assure the truth of the conclusion but will at most establish that conclusion as contextually plausible. Such inferences are ampliative: the conclusion can go beyond what the premises guarantee.

Presumptions in Science

Philosophers of science often puzzle about the prominent but yet problematic role of such theoretical factors of inductive reasoning as regularity, continuity, simplicity, uniformity, conservation, and the like. It would clearly be problematic to regard such factors as representing fundamentally ontological facts about the world – an inclination on the part of nature itself (to put it somewhat anthropomorphically). It would, however, make perfectly good sense to view such factors as representing principles of procedural presumption – forming part of the regulative or methodological mechanism of our vehicles of explanation. One thereby avoids treating such features of systematicity as substantive findings about nature but rather as methodological, and procedural guideposts for our conduct of scientific inquiry – procedural principles of plausibility that constitute our evaluative standards for explanatory accounts.

In this way, a positive presumption of acceptability is taken to operate in favor of all the traditional parameters of systematization: consistency, uniformity, regularity (causality, rulishness, and lawfulness in all forms), simplicity, connectedness/coherence, unity/completeness, and the rest. These various defining facets of systematicity thus come to do double duty as elements of a criteriology for acceptability-as-true and as presumptive principles regulatively governing the conduct of inquiry. The resulting situation is set out in Display 9.1.

What Presumption Is All About

To presume in the presently relevant sense of the term is to accept something in the absence of the further relevant information that would ordinarily be deemed necessary to establish it. The term derives from the Latin praesumere: to take before or to take for granted.Presumption has figured in legal reasoning since classical antiquity. There is nothing modern or cutting-edge about it: it is one of the oldest tricks in the book.

Presumption found its first and still most prominent role in the context of the law, where a presumption mandates a trier to accept a certain fact once some other correlative fact has been established. The French Code civil defines “presumptions” as “Consequences drawn by the law or the magistrate from a known to an unknown fact.”Legal presumptions provide a way of filling in – at least pro tem – the gaps that obtain in conditions of incomplete information. (The “presumption of innocence” provides a paradigm example here.)

Such a legal presumption (praesumptio juris) is an inference from a fact that, by legal prescription, stands until refuted. Presumption of this sort is a gap-filling resource: it comes into operation only in the absence of relevant information or evidence, and it leaves the scene once suitably strong evidential indications come to view. One authority has elucidated the conception of presumption in the following terms:

The Purposiveness of Presumption

Whenever we act and actually do something on the basis of a presumption we take a step beyond it: we make the assumption that it is correct. This assumption may turn out to be incorrect. But even if so, this nevertheless does nothing to unravel the validity of that initial presumption in the circumstances under which it arose. Its erroneousness is a matter of hindsight unavailable before the fact and represents a misfortune rather than a mistake.

The appropriateness of presumptions is not to be assessed on an individualized retail basis; instead the matter is one of wholesale – of the statistically systemic appraisal of the principles of presumption that underlie the particular case. Presumption is a methodological and procedural resource and must be judged on that basis through its performance “on the whole.” Thus epistemic presumptions exist to enhance our access to usable information, and communicative presumptions exist to facilitate the transmission thereof. The ultimate test of the appropriateness and validity of our presumptive proceedings in their various domains will be a matter of their efficacy in the realization of the correlative purposes – an ultimately pragmatic standard.

In the late nineteenth century, C. S. Peirce in the United States and Ernest Mach in Europe suggested an evolutionary rationale for certain presumptions such as trust in the senses and in various analogies.

“Dr. Livingston, I presume” runs the famous exclamation with which H. M. Stanley greeted the long-lost explorer. And in saying this he as much as said that “this is what I am going to take to be the case unless and until further developments should show that it is not.” That is just exactly how presumption works.

The topic of presumption encompasses a wide range of practices within our quest for informative knowledge and practical decision. These particular resolutions, however, have a tentative quality in being taken to hold not with categorical assurance but rather provisionally and pro tem until such a point when (if ever) sufficiently strong counterindications come to light. Such presumptions carry a burden of proof that inclines upon anyone who is disinclined to accept them.

The practice of presumption arose initially in the law but subsequently became operative in virtually every area of rational endeavor, for presumption is a remarkably versatile and pervasively useful resource. Firmly grounded in the law of evidence from its origins in classical antiquity, it made its way in the days of medieval scholasticism into the theory and practice of disputation and debate. And it subsequently extended its reach to play an increasingly significant role in the philosophical theory of knowledge. It has thus come to represent a region where lawyers, debaters, and philosophers can all find some common ground.

Introduction

Presumption is closely bound to the idea of burden of proof (onus probandi), which is also at root a legal conception. It functions in the context of an adversarial proceeding in which one party is endeavoring to establish and another to rebut some charge before a neutral adjudicative tribunal. The very phrase (onus probandi) derives from classical Roman law where it affords one of the ground rules of probative procedure, governing the division of the labor of argumentation between plaintiff and defendant, which specifies tasks of marshaling evidence. Under the Roman system, nothing was conceded in legal actions as admitted: the plaintiff, as the initiating agent in laying a charge, had to make his case first (agenti incumbit probatio), then the defendant's countercase was argued on his exceptio, and thereafter the plaintiff's on his replication, and so on. The burden rested with the plaintiff in civil cases and with the state (as surrogate plaintiff) in criminal cases. Throughout, the “burden of proof” lay with the side active in making the allegations, subject to the fundamental rule that “the need for proof lies with him who affirms, not him who denies.”

The idea of burden of proof embodies a basic ground rule of the probative process. To say that the burden of proof rests with a certain side is to say that side must adduce the substantiation required to make its case.

Cognitive Fundamentals

Presumptions obtain principally with two ends in view. On the one hand there are the purely cognitive presumptions made for the sake of answering our questions and filling gaps in our information (as, for example, presumptions regarding the reliability of sources). On the other hand there are practical presumptions made for the sake of guiding our decisions regarding actions (as with legal presumptions that facilitate the resolution of cases; for example, that someone long missing is dead so that the person's estate can be distributed). At present, however, the cognitive sector of presumption will be at the focus of concern, with presumption as a source of (putative) information about the truth of things in the forefront.

A cognitive presumption stakes a claim that outruns the substance of actually available information; it is a proposition that, in suitably favorable circumstances, is accepted as true in the absence of any counterindications. This is a default position affording an answer to some question of ours that we adopt for lack of anything better and will keep in place until such time as something comes along to eject it from this position of favor. In this way, cognitive presumptions function as instrumentalities of rational economy. Thanks to them we need not remain bereft of answers to our questions until all the relevant returns are in – which is to say virtually never.

Specificity Prioritization

The story is told that Herbert Spencer said of Thomas Buckle (or was it the other way round? – as it could just as well have been) that his idea of a tragedy was a beautiful theory destroyed by a recalcitrant fact.A fundamental epistemic principle is at issue here, namely, that when the limited particularity of fact and the broad generality of theory come into conflict in the case of otherwise plausible propositions, then it is the former that will prevail. Facts, as the proverb has it, are stubborn things: in case of a clash, facts must prevail over theories, observations over speculations, concrete instances over abstract generalities, limited laws over broader theories. With factual issues specificity predominates generality when other things are anything like equal. And so a far-reaching Principle of Specificity Precedence comes into view with respect to rational inquiry.

The workings of such a Principle of Specificity Precedence can be illustrated from many different points of view. The practice of monitoring hypothetical theorizing by means of experimentation is characteristic of the scientific process, and the Principle of Specificity Precedence is fundamental here. Throughout, whenever speculation clashes with the phenomena, a conjectured hypotheses with the data at our disposal, or a theory with observation then it is generally – and almost automatically – the former that is made to give way. Presumption, that is to say, stands on the side of specificity throughout the realm of factual inquiry.

Communicative Presumption

In the absence of any evidence to the contrary, we proceed on the presumption that people do what they do on the basis of reasons, granting them the benefit of the doubt in point of rationality. We take the stance that there is an explanation of the basis of what they see as good reasons for proceeding as they do – that they act as they do under the impression that some good will come of it, some benefit for themselves or others. Accordingly, when asked to explain why someone did something, it suffices that we establish that this was the rationally appropriate thing for them to do – that this is what any sensible person would have done in the circumstances. We presume people act rationally. Only as a reluctant last resort do we judge that someone has proceeded on the basis of forces or motives outside the range of their rational control. Thus here, as elsewhere, a presumption of normalcy prevails. Realizing full well that people are not always rational, we find that they (fortunately) are so generally and ordinarily, and feel free to proceed on this basis. We grant people the benefit of any doubt and treat them as rational agents in the absence of convincing counterindications. The rational economy of process is clearly at work here.

Cognitive convenience tells the whole story here.

INTRODUCTION

No scientific theory has stimulated more intense debate over conventionalism than the theory of relativity. From early on, scientifically minded philosophers, such as Schlick, Reichenbach, and Carnap, and philosophically minded physicists, such as Eddington and Weyl, scrutinized the new theory, hoping to uncover within it an epistemological revolution parallel to that which had been wrought in physics. Conventionalism was central to many of these early explorations, and remained so throughout the twentieth century. The debate centered around the status of geometry. As we saw, Poincaré distinguished the conventionality of geometry from other manifestations of conventionality in science. In the case of geometry, he believed, there was a particularly powerful argument for the empirical equivalence of various physical geometries, and thus for the underdetermination of geometry by experience. On Poincaré's view, to prefer a particular (physical) geometry is basically to prefer a particular formulation of the facts of physics, facts independent of any such formulation, hence the conventionality of geometry. Subsequent discussion of conventionalism in the context of the theory of relativity continued to center on the conventionality of geometry, which is, accordingly, the subject of this chapter.

Whereas Poincaré had to invent fictitious worlds and fictitious theories to make his point, the advent of the theory of relativity, which breaks with both Newtonian mechanics and the theory of space and time in which it is embedded (Newtonian dynamics and Newtonian kinematics), obviated the need for fiction. Nevertheless, the theory that emerged was quite different from those Poincaré had envisioned.

INTRODUCTION

It is widely agreed that the principle of tolerance, which upholds “complete liberty with regard to the forms of language,” epitomizes Carnap's philosophical outlook. Reflecting on this principle, Carnap notes that a more adequate designation would have been “the principle of conventionality” (1942, p. 247), or “the principle of the conventionality of language forms” (1963, p. 55). It should be remembered that insofar as conventionalism is considered a philosophy, Carnap would have been reluctant to characterize himself as a conventionalist: “I want to emphasize that we are not a philosophical school and that we put forward no philosophical theses whatsoever” ([1932] 1934, p. 21, emphasis in original). Accordingly, “between our view and any … traditional view there cannot be identity – but at most agreement with the logical components. For we pursue logical analysis, but no philosophy” (p. 29, emphasis in original). And yet, conventionality is at the heart of Carnap's thinking; so much so that, to a considerable extent, his work can be seen as a series of attempts to uncover the conventional aspects of knowledge and thereby bring to light the connection between the classic philosophical conundrums, such as the nature of a priori knowledge and the controversy over realism, and the conflation of truth and convention. It is telling that, while the principle of tolerance maintains that there are no rights and wrongs, no “morals,” in the conventional choice of the appropriate language for a given task, the language Carnap uses to express the implications of his principle is emotionally charged and has pronounced moral undertones.

INTRODUCTION

While it is generally agreed that Henri Poincaré was the first to expressly articulate a conventionalist position, there is much less agreement as to what exactly his position was. As well as a considerable number of interpretations, Poincaré's work has inspired a broad spectrum of responses, from attempts to substantiate and extend conventionalism, to purported refutations. Some have gone so far as to challenge the characterization of Poincaré as a conventionalist, implying that he misrepresented, if not misunderstood, his own position. Pierre Duhem is widely considered the cofounder of conventionalism. Although his advocacy of conventionalism is less explicit than Poincaré's, his work nonetheless advances the case for conventionalism considerably. The central pillars of Duhem's philosophy of science are a holistic conception of scientific theories, and the ensuing critique of the feasibility of crucial experiments conclusively verifying or refuting individual hypotheses. If, despite the indecisive nature of observation, scientists come to prefer one theory to another, they must be invoking considerations other than mere compatibility with experience. Hence, conventionalism. Nevertheless, The Aim and Structure of Physical Theory, Duhem's major philosophical work, has also been described as an attempted synthesis between conventionalism and realism (McMullin 1990), and even as a treatise against conventionalism (Maiocchi 1990).

Recall the two forms of conventionalism distinguished in chapter 1, the underdetermination of scientific theory by observation and the conventionalist account of necessary truth.