Throughout the history of science, indeed throughout the history of knowledge, unification has been touted as a central aim of intellectual inquiry. We’ve always wanted to discover not only numerous bare facts about the universe, but to show how such facts are linked and interrelated. Large amounts of time and effort have been spent trying to show diverse arrays of things can be seen as different manifestations of some common underlying entities or properties. Thales is said to have originated philosophy and science with his declaration that everything was, at base, a form of water. Plato’s theory of the forms was thought to be a magnificent accomplishment because it gave a unified solution to the separate problems of the relation between knowledge and belief, the grounding of objective values, and how continuity is possible amid change. Pasteur made numerous medical advancements possible by demonstrating the interconnection between microorganisms and human disease symptoms. Many technological advances were aided by Maxwell’s showing that light is a kind of electromagnetic radiation. The attempt to unify the various known forces is often referred to as “The Holy Grail” of physics. Some philosophers have even suggested that providing explanations is itself just a sort of unifying of our knowledge. But while unification (like simplicity) has often been hailed as a tremendous virtue in science, the meaning of the term is not altogether clear. Scientists often don’t specify what, precisely, they mean by unification. And in cases where what they mean is clear, different thinkers plainly mean different things by the term. What are the various senses of unification, and why has unification been such an important aim in the history of inquiry?
II. What is unification?
Looking at the tremendous variety of projects that have been termed “unifying,” it’s easy to become discouraged about trying to find a coherent unifying idea in the concept of unification. The unifications accomplished by Plato and Pasteur were certainly of a vastly different nature. The type of unification of Mendelian and Darwinian ideas by Fischer seems different from either of these. And Karl Popper’s arguments that the unifying principle underlying all of science was falsification seem to be discussing another kind of thing altogether. Nevertheless, it is possible to look at unification in a systematic way. At base, the concept of unification is about showing the connections between a set of different things. One might try to unify different sorts of epistemic goals, or different methodological approaches to the same epistemic goals. Unification of method is an issue that concerns scholars looking for what makes science, in general, an activity of special importance. (This was one of the concerns of the Logical Positivists, and of Popper). More often, however, it’s theories about what the world consists of that scholars try to unify.
Whatever is being unified, it should be noted at the outset that there are two distinct types of bringing-together that are each sometimes called unification. One family of bringing together might be called subtype and similarity (or SS) unification; the other might be called conjunction and coordination (or CC) unification. SS unification involves trying to show that things that seem different really share some or other dimension of similarity. CC unification involves trying to show some type of connection between things that may or may not be alike. An example of minimal CC unification would be describing a group of things that are connected by being next to each other in time or space by a single term – as when someone describes a large set of adjacent mountains as “the Appalachians” or a series of battles as “the Hundred Years War.” A more maximal CC unification is when a set of items comes to be recognized as having intertwining causal connections with other items such that, together, they make up a unified functioning system. We can now talk about the workings of the hypothalamus, pituitary, adrenal glands, and the hormones ACTH and cortisol in one breath by talking about an “endocrine alarm clock” that wakes us. Malinowski showed how the exchanges of ritual goods between 18 Melanesian islands created an elaborate social network he termed the Kula Ring. In between these are the sorts of unified pictures that come into being when it is discovered that previously-thought-to-be unrelated things have a cause and effect relationship – e.g., carbon dioxide emissions and the melting of polar ice caps. Since we have some type of unification whenever we uncover a relationship between entities or properties (including between dissimilar ones), then there is going to be an enormous variety of scientific discoveries that can be thought of as effecting a unification of a sort.
The other family of unification, the subtype and similarity family, involves showing that a group of seemingly different entities or properties belongs to a common general type. The most maximal thoroughgoing type of unification in this family is reductive identification. Maxwell’s work showing that light was not just related to electromagnetism, but actually was a form of electromagnetic radiation, is perhaps the best-known example of this type of unification by ontological simplification. A more minimal SS unification involves showing that different things are each members of a broader category of things sharing some properties. The recent claim that both the pattern of energy of atoms in gases at thermal equilibrium and the distribution of people’s income levels in developed countries follow an exponential distribution pattern (Hogan 2005) is an example of this more minimal type of SS unification. So is the claim that dolphins and pigs are both mammals. A number of scholars (e.g., Morrison 2000) have pointed out the importance of this type of unification, and how it’s different from the fully reductive kind. (Fewer have pointed out how this type is also different from CC unification). An even more minimal type of SS unification is claiming that different things are members of the same broader class -- not necessarily because each member shares a common set of properties with other members, but because each member is linked via some or other similarity to a central prototype. The class “fish” seems to unify things in this minimal way (see Gould 1983).
Between these extremes there are many ways scientists can accomplish an SS unification because there are many ways two things can be judged to be similar. The term “vertebrate” unifies all those animals that are thought to be similar by virtue of the internal property of having a backbone. The things unified by the concept “gene,” by contrast, need not share particular internal properties. Being a particular gene is defined largely in terms of having the external property of producing certain effects on developing organisms. Being a gene is one of many multiply realized properties, defined by similarities in their external functional role, rather than similarities in their internal structure. Another kind of similarity is having a shared holistic structure, rather than having shared particular internal features. The Hardy-Weinberg equation, for example, talks about overall population growth in highly diverse reproducing species. Things can also be classed as similar, not because they share particular properties, but because they share particular amounts or proportions of a property. So one might give a unifying description of a stage of growth in different plants by talking about a common reduced amount of chlorophyll at that stage. Things sharing large numbers of properties (but not necessarily one common one) can also be classed as a similar type. Marketing research might isolate groups of people who have a certain family resemblance about which one can make specific economic generalizations. Things could also be classed as similar because they lack certain things, even if they are quite varied in other ways. So a mentally retarded child and a PhD. chemist might both be described as autistics who lack an empathetic understanding of others. There are as many ways to unify as there are ways of finding similarity-based classes that one can make generalizations about. It is not surprising, then, that various scholars describe lots of different types of scientific achievements as accomplishing a unification.
It is not uncommon for scholars to be engaged in finding conjunction and coordination and similarity and subtype unification at the same time. The unifying term “Indo-European languages,” for example, is given to a family of languages that had both a causal effect on each other’s structure and a similarity to each other. And often, explaining things by connecting parts of various theories (CC unification) involves showing how something in one theory can be identified with something in another theory (SS unification). Why is the sky blue? We combine optics, chemistry, meterology, and biology when we say that, at certain times of the day, light goes through a certain amount of atmosphere, hitting small nitrogen and oxygen particles; the light bouncing off these particles has a wavelength of between .390-.492 microns, which is the blue and violet spectrum, and our eyes are especially sensitive to blue light. Describing how the various elements interact combines them into a CC unified story, while showing that the atmosphere is comprised of separated nitrogen and oxygen particles is an SS unification, as is identifying blueness with light that has a wavelength of between .455-.492 microns. Note also that understanding a concept that unifies various elements in a CC manner (sodium atoms can become linked to chlorine atoms through ionic bonding), also often involves making a unifying SS identification between the low level coordinating parts and the high level concept (sodium chloride=salt). There are not only different types of unification, then, but different types happen simultaneously. It’s no wonder that it is difficult to explain exactly what scientific unification is.
We see, then, why so many very different kinds of activities can all be thought us as providing us with a kind of unification. But why is coming up with such unifications thought to be such an important activity? One underlying feature that makes unification such an important virtue is surprisingly little discussed, but, nevertheless, is quite straightforward. Unification provides agents with a way of saving precious cognitive resources. It saves resources with regards to both the information possessable by individual agents, and the collective knowledge held by groups in libraries or computer banks. When various things are linked in a CC unification, the people who learn these unified theories come to have associative networks in their minds which provide efficient search engines for numerous facts. With a unified picture, the fact of what happens, say, after two carbon atoms are oxidized in the Krebs cycle can be located very quickly, without having to search blindly through a myriad of items in memory. Memory space is saved by CC unification as well. Rather than having to be stored, CC unification allows numerous facts to be inferred, based on perception and a knowledge of what causes what (e. g., seeing the positions of the sun and moon enables me to calculate that there was an eclipse in 23 days ago).
SS unification also saves memory space. Classifying all variants of a certain arrangement of electrons, protons, and neutrons as belonging to a single unified kind – say “carbon atom”-- enables us to store information about carbon atoms in a single place in memory. This information can be continually referred back to, instead of having to have a space-hogging representation of a complex arrangement of electrons, protons, and neutrons for each place where carbon is present. SS unification can save time as well, for memory is far more efficiently searched if things are categorized as subtypes of subtypes of subtypes, rather than as independent facts (see Jones 2004 for a detailed discussion). The time and space resource-saving that unification provides, gives us access to far more information about the world than we could possess if we had to memorize facts about the world in a non-unified way. The more information we have access to, the better epistemic agents we are, and the easier we can meet our various goals. What’s more, the more unification we have, the fewer facts we must regard as brute unexplainable ones, derivable neither from being an instance of a more general fact, nor from a knowledge of causal antecedents. Presumably, we prefer that there be few facts we must regard as brute. Reducing their number is made possible by unification.
III. Unification and explanation.
Most scholars consider unification a highly desirable virtue in science. But there have also been scholars (notably Huxley, Freidman, and Kitcher) who hold that it is more than just a desirable virtue. For some, it is through unification that we really explain things with science. The most well developed version of this view has been put forward in a set of papers by Kitcher (1981, 1986, 1989, 1995). On Kitcher’s view, explanations are deductive derivations which conclude with a statement of the fact to be explained. But whether an account is really an explanation cannot be determined by looking at that account alone. To qualify as an explanation, an account first must be a particular instance of a general schematic "derivation pattern" whose concrete instances are used to generate lots of different conclusions. And that derivation pattern must itself belong to a particular set of derivation patterns that together constitute something called the "explanatory store." The explanatory store is composed of the smallest set of derivation patterns that together can be used to generate the largest amount of our total knowledge of the universe.(Kitcher's theory also has a requirement that the derivations be maximally stringent, preventing derivations from relying on overly vague terms.) By deriving conclusions from a sparse store of patterns, we show how numerous different facts about the world can be derived using the same patterns over and over again. We understand the world when we produce the most systematic, most unified representation of it that we can. We explain particular facts when we show how they fit into and can be derived from this best understanding of the world.
As one might expect, there have been numerous objections to the view that explanation is a type of unification. Among scholars most dubious about the unification theory are those who believe that the essence of explanation is revealing the underlying mechanisms (usually causal) that made an event happen. This view has been termed the ontic conception of explanation (see Coffa 1977, Railton 1980, Salmon 1989). The unification view, by contrast, is an example of an epistemic conception in which explanation is a matter of finding the generalizations that tell us that a certain type of event is the one we should expect. For enthusiasts of the ontic approach, epistemic approaches just do not capture what we ordinarily mean by explanation. This can be readily seen, according to enthusiasts of ontic approaches, by looking at the problems that things like asymmetry pose for epistemic conceptions. The most well-worked out epistemic approach to explanation has been Hempel’s Deductive-Nomological theory. Yet explanatory asymmetry poses severe problems for this view. Bromberger (1963) pointed out that describing the height of a tower and the angle of elevation of the sun together provide a D-N explanation of the length of the corresponding shadow; but a similar derivation using the length of the shadow and the angle of the sun to calculate the height of the tower does not intuitively explain the tower's height. Premises and conclusions can often be deductively derived from each other in either direction, but often only one direction of derivation is an explanation. Why? Theorists in the ontic tradition have thought that the tower and shadow case clearly illustrates that giving explanations means showing how information about effects derives from premises regarding underlying causal mechanisms.
Theorists holding the epistemic conception of explanation tend to be more skeptically inclined toward underlying mechanisms. They believe we must be cautious about asserting the existence of underlying mechanisms that are usually invisible and to which we rarely have any direct epistemic access. We hypothesize that certain invisible mechanisms or laws exist because we reason that if these existed then they could be responsible for our observations. But it is often the case that numerous different postulated combinations of mechanisms or laws could logically be responsible for our observations. Which ones best explain them? Ontic conceptions of explanation can’t really tell us, say epistemic conception proponents. The unification theory, on the other hand, counsels us to pick, out of the many possible derivations, the account that best helps systematize and unify our knowledge. Meanwhile, there are additional worries about how to find underlying causal mechanisms. Epistemic conception theories point to the fact that no one has yet given a fully satisfactory answer to Hume’s worries about showing what a causal connection is. How can locating causes be what explaining is all about, when we don’t really know what it is to locate a cause?
Unification theorists also believe that the asymmetry problem is not really a problem for their particular type of epistemic approach. While we can derive the height of a tower from information about the angle of the sun and the length of the shadow, the unification theory provides us with criteria for ruling out such derivations as non-explanatory ones. The explanatory derivations, according to the unification theory, are the ones that can be used to derive the largest set of facts about the world from the smallest set of derivation patterns. We can derive the dimensions of some objects, using a pattern that does so on the basis of their shadows. But we cannot derive the dimensions of transparent objects, luminescent objects, huge objects, or tiny objects, which do not cast shadows this way. We can derive the dimensions of almost any structure, on the other hand, using a general schema that might be called the Origin and Development derivation pattern. Since this pattern schema allows us to derive more facts than the shadow-based one, it is part of the preferred explanatory set. Deriving the height of the tower from knowing the intentions of the designer at the time it was built and any subsequent alterations made to the structure since that time is an instance of this schema. It is therefore the derivation of the tower’s height that should be deemed explanatory, not a shadow-based one. (Kitcher 1989: 485).
Unificationists have replied to other proposed asymmetry cases as well.
Eric Barnes (1992) discusses a case in which we can derive the fact that a dinosaur of a certain skeletal type existed, based on finding a fossil skeleton. But it might be the case that current paleontologists are in no position to tell us why skeletal structure S rather than others came to exist. Since there are no competing ways to derive the skeletal structure other than using an “evidentiary” derivation, says Barnes, the unification theory is forced to label this intuitively unexplanatory account as explanatory. Jones (1995) has responded that unificationists need not prefer the fossil-based derivation to other accounts. In our store of commonly used explanations of organism morphologies, is a pattern that can be termed the "Darwinian Evolution of Skeletal Structure Pattern." A Darwinian argument pattern would have us look at the predecessor skeletal forms and at the various selection pressures that could lead these forms to be modified. The problem with using this pattern is not that we could not fill in the pattern with detailed information about past conditions to generate the skeletal structure conclusion. Rather, the problem is that we really don't have enough access to the past to have complete confidence in the accuracy of the premises used in this derivation. But in this situation, we could still give a speculative account in which one generates the detailed conclusion using premises whose truth is, to varying degrees, less than certain. Alternatively, we could give a partial explanation using the Dawinian pattern, where we derive a less detailed conclusion using only premises that are well accepted. There is no reason that unification theory advocates would have to prefer, as more unifying, a more detail-yielding derivation based on a larger store of derivation patters to a partial or speculative derivation that comes from a smaller set that can generate a wider variety of conclusions. What’s more, Kitcher (1981) points out that among the ways to try to unify our knowledge, an important method is to try to use not only as few patterns as possible in generating conclusions, but also to use patterns that are similar to each other. In other words, our knowledge is unified by seeking to derive conclusions using patterns from the same general family. One can think of patterns of families as having a hierarchical tree structure, with the most general abstract pattern as the trunk, different more detailed instantiations of parts of it as limbs, and further more detailed instantiations of these limbs as branches. In the skeletal structure case, deriving the skeletal structure from an evidentiary rather than a Darwinian pattern, would not only force us to include an extra pattern in our knowledge systematization, but would also violate unificationism's edict to utilize the same family of patterns wherever possible. Unification theorists, then, believe that the asymmetry problem can be solved without having to reintroduce age-old problems regarding underdetermination and causation.
But even if one is not committed to an ontic conception of explanation, there are other problems that unificationists must overcome if they are to convince people that an explanation is the account that is the most unifying. Chief among these is the fact that unificationists have never spelled out in any detail how to choose between accounts that are unifying in different ways. One way that we could better unify our knowledge is by accounting for far more facts, even if that mean increasing the number of patterns of derivation we must use. Another way is to use far fewer patterns to account for a high number of facts derived, while perhaps being able to derive fewer facts. A third way is to could increase derived facts and reduce the number of patterns by playing with the stringency requirement. Most likely, we could increase our “unification score” best by doing some combination of the three. There are, however, a theoretically infinite number of ways that one could add scores on the number of conclusions, paucity of patterns, and stringency to get a higher unification score than the best previous systematization (e.g., 5 + 5 + 5 = 15, so does 5 + 6 + 4, so does 5 + 5.0001 + 4.999…). Current formulations of the unification account say little about how to choose between perhaps radically different systematizations that are “tied” with respect to their unifying power.
This isn’t automatically a problem. Scientists often give quite different explanations of the same phenomena. The idea that different accounts can be unifying in different ways, may be why they do so. Indeed, the fact that the unification theory allows different types of accounts to be thought of as explanatory might be thought of as a special virtue not a liability. Ultimately, however, it can only be a virtue if different ways of unifying do not also end up counting various intuitively unexplanatory derivations as unifying explanations. At this point, when a scholar proposes that an intuitively unexplanatory account can count as part of the best systematization of our knowledge, unificationists respond by showing there is a more unifying systematization that produces an intuitively explanatory account . Kitcher has expressed optimism that in actual scientific practice (as opposed to the world of logical possibility), we do not find intuitively unexplanatory accounts that stem from systematizations that are more unifying than others(1989 1995). Neither the optimism of unification proponents, nor a few case by case demonstrations, are enough to convince skeptics that there are no unexplanatory systematizations that unify our knowledge as well as the systematizations that produce intuitively explanatory accounts. To satisfy their critics, unificationists need to find ways of showing that no intuitively unexplanatory accounts could be part of our most unifying knowledge systematizations. They might do this through a) explicating additional principles that further limit which unifying knowledge systematizations are more unifying than others, and/or b) additional arguments showing why current principles, or augmented ones, will generally rule out systematizations that yield intuitively unexplanatory derivations. Without these, discussions of how celebrated scientific explanations have unified our knowledge cannot convince those who doubt that explanation isa form of unification.
In summary, unification is undeniably important in science. There appear to be many different types of unification. There also appear to be important links between the different types. Whether unification is at the heart of science, enabling us to give and identify explanations, remains an important issue for debate.
Barnes, E. (1992) “Explanatory Unification and the Problem of Asymmetry,” Philosophy of Science 59, 558-571.
Bromberger, S. (1963) “A Theory about the Theory of Theory and about the Theory of Theories,” in W. Reese (ed.) Philosophy of Science, The Delaware Seminar, John Wiley, New York.
Coffa, J. A. (1977) “Probabilities: Reasonable or True?” Philosophy of Science 44,
Hogan, J (2005) “Why It Is Hard To Share Wealth.” New Scientist 2490 pp. 6
Jones, T. (2004) "Reduction and Anti-reduction: Rights and Wrongs." Metaphilosophy
25:5 pp. 614-647.
Jones, T. (1995) "How the Unification Theory of Explanation Escapes Asymmetry."
Erkenntnis, 43 pp. 229-240.
Kitcher, P. (1981) “Explanatory Unification, “ Philosophy of Science, 48, 505-
Kitcher, P. (1986) “Projecting the Order of Nature, “ in R. Butts (ed.), Kant's Philosophy of Physical Science, D. Reidel, Dordrecht.
Kitcher, P. (1989) “Explanatory Unification and the Causal Structure of the World,” in W. Salmon and P. Kitcher (eds.) Minnesota Studies in the Philosophy of Science, Volume XIII, University of Minnesota Press, Minneapolis.
Kitcher, P. (1995) “Explanation, Causation, and Unification” Jamieson Lecture, University of London.
Morrison, M. (2000) Unifying Scientific Theories. Oxford University Press, Oxford.
Railton, P. (1980) Explaining Probability. Ph.D. dissertation, Princeton University.
Salmon, W. (1989) “Four Decades of Scientific Explanation,” in W. Salmon and P. Kitcher (eds.) Minnesota Studies in the Philosophy of Science, Volume XIII, University of Minnesota Press, Minneapolis.