Chapter 10 – Reductionism and Emergence

Chapter 10 Reductionism and Emergence


Reductionism is an analytical process, identifying the parts of something and examining their relationships to each other and to the whole.

But when someone puts forward an argument that sounds clear and logical, you may occasionally hear it dismissed with the words “that is just reductionist.” The word reductionist is used in such cases to imply that the argument is unduly simplified or distorts the issue. Reductionist thinking, it is implied, leaves out something essential, perhaps some romantic or supernatural element. And reduction, i.e., the use of reductionist thinking, is integral to science. So doubt is sometimes cast on science because it is reductionist.

Another criticism of reductionist science is that it is not holistic: it deals with individual aspects of the world but ignores the overall unity. This criticism often implies that science cannot see the world as God’s creation and it is accordingly inferior to the teachings of religion. I agree that reductionist science looks at individual parts of the world, and that it looks only at measurable evidence. But I think that, in doing so, science indeed deals with the whole, however large or small we might take the whole to be.


From time to time there is talk about the “conflict” or the “incompatibility” between science and religion. The differences between the two do not prevent scientists from having religious faith nor make non-scientists religious. But while an objective of science is to look for new discoveries that may improve or replace current accepted scientific “facts” and explanations, religion is belief in the truth a set of existing statements and explanations, which it tries to uphold in the face of contradicting scientific discoveries or explanations.

This difference in approach makes some people think that science is invalid or unreliable. They may invoke a supernatural entity as a solution to real or imagined scientific problems, describing “reductionist science” as atheistic, deterministic, or simplistic.

This chapter addresses such claims, discussing the concept of reductionism, its inherent limitations, and its often-misunderstood counterpart, the phenomenon of emergence.


Here are a couple of simple illustrations of reductionism, starting with one that does not directly relate to science.

Someone might have an old wind-up clock without knowing how it works. To discover how it works the person could examine the gears and the spring and the escapement, etc., and see how each of the parts interacts with the others to produce the integrated functioning timepiece

To give a scientific example, a molecule of water is commonly symbolised as H2O, which means that it is composed of two atoms of hydrogen and one atom of oxygen that have interacted to produce a new entity. Discovering the components of water and their characteristics enables an explanation to be made of many characteristics of water, some of which may be surprising when first encountered. But the examination of the parts is less direct than looking inside a clock.

Similar explanations apply for all chemical compounds. This may be expressed as “the whole is (the interaction of) the sum of its parts”, with the process of reduction being expressed as “reducing the whole to its parts”.

A reductionist argument might be that chemistry can be explained purely in terms of the physics of atoms. And the argument goes further. Biochemistry can be explained by chemistry. The operation of the cells of living organisms can be explained by their structure and their biochemistry, and similarly for the physiology of organisms. And the physiology explains the behaviour of the individual organisms, which explains how individuals operate as societies. But if it seems fanciful to explain sociology entirely in terms of physics, or even of mathematics, that is because it really is fanciful. And yet I once saw something very like that being quite seriously proposed.

The physical sciences – mathematics, physics and chemistry – are often referred to as the “hard sciences”, not in the sense of being difficult (which may people think they are) but as logical and rigorous. The biological and social sciences are then referred as the “soft sciences” implying that they are more intuitive and imprecise. This is undue stereotyping. Nevertheless, some scientists, or at least those associated with the physical sciences, feel this may be correct, and even that mathematics and physics are somehow “more true”, or at least more reliable, than the biological sciences, and even more so than the social sciences. But the formulas and the usually unseen atoms, molecules, etc., of the physical sciences are not obviously real to most people in the way that visible objects and organisms of the everyday world are. Some people regard human consciousness to be more real than any part of science.

This description of reductionism is crude, but it is not far removed from the general impression of what reductionism is. And it gives a hint as to why some people distrust any argument that is or sounds reductionist. It gives a picture of reductionism “starting at the bottom” and then building more and more layers of complexity. (Actually, most levels of complexity can include parts from any of the less complex levels. For example, the cell of an organism can be analysed to show that some of the parts are complex, like the different types of tissue, while other parts are much simpler, like individually acting molecules and ions.) But the very term reductionism implies starting and any level of complexity and “reducing” it to its parts, which seems to be the reverse of the “building-up” picture. But the picture can be turned upside down, so that it starts with something like: sociology can be explained in terms of psychology (and other factors); psychology can be explained by physiology and other factors); and so on to chemistry can be explained in terms of physics.


Some Relevant Aspects of Science

Before exploring the issues further it would be helpful to first look at some relevant aspects of science.

Science is sometimes characterised as a combination of “stamp-collecting” and “storytelling”. The stamp-collecting consists of gathering and examining pieces of information about the world, and then classifying the pieces into groups according to their apparent relationships, differences and common characteristics. The storytelling then explains how the similarities and the differences might have occurred and how everything works together. And, of course, the storytelling is what is called scientific theory.

Science usually starts with the observable. When new things and new phenomena have been closely looked at, as distinct from someone just happening to notice them, theories are devised to explain how they operate and what has produced them. After noting what factors seem to be involved, a theorist begins the process of analysis (i.e., of reduction), which puts a degree of discipline into the act of observation. Then the possible ways in which the parts can interact are considered, to see what kinds of systems can be built from them and so explain the new phenomenon. So reduction is a part of the process of creating theories. Sometimes a theory in science appears to start with speculation, such as with Einstein’s theories of relativity. But even these were speculations about observed characteristics of light and mass.

Any analysis and explanation about how something works has to start with the whole thing rather than with the individual parts. While chemistry can be explained using physics, physicists could not have discovered much about how chemistry works without already knowing about chemical phenomena. And you would get no understanding about the workings of the clock by looking at the structure and other characteristics of the atoms of copper and the other elements the clock is made of. Similarly the characteristics of the atoms of carbon, hydrogen, oxygen and nitrogen in amino acids are not a good starting point in finding the workings of a specific protein.

When the clock has been analysed by looking at all the parts and their interactions, there would be no need to look for any extra element to see how it works. The whole clock is equal to the sum of its parts. It is a system, and the analysis enables the operation of the system to be explained.

But there is a difference between the assembled device and its unassembled parts. One consequence of this difference is that in many cases it can be very difficult to know precisely what will be produced by assembling the parts.

With a clock, each of the parts has its own specific role, so there will be only one or a few of some kinds of part. With a different kind of unassembled parts, as in a Lego set, it is possible to build many different things. With Lego, you might build a bridge or a tower, or a trolley, etc. You can decide what to build. But what does nature do, where the parts assemble themselves in a way that depends on their environment?


A group of amino acids can be built up into a huge range of different types of proteins. To be able to predict precisely which protein will be built on a particular situation you would need to know the particular gene sequence that is guiding the assembling. In other words, you need to know enough about the environment to be able to predict the outcome.

All this shows why sociology cannot be derived by starting from the physics of atoms and electrons. You cannot ever know enough about all the intervening levels of complexity by starting at even a few levels lower.


At a basic level (excluding the quantum behaviour of particles), the world operates consistently. No matter how many times something is correctly analysed, the same set of parts will always be found. So at each stage of complexity it should be possible to analyse something to find how the parts explain the whole. If different parts are sometimes found, it means that the thing analysed was slightly different from what it was thought to be. An example might be a molecule of carbon dioxide (CO2), which in most cases has carbon-12, which has six neutrons in its nucleus, but occasionally has the rare isotope carbon-14, which has eight neutrons. But it behaves chemically as carbon dioxide irrespective of which isotope it is because carbon atoms always have six protons. (There are slight physical differences because the mass of carbon-14 is about 17% greater than that of carbon-12.)

An important consideration with reductionism is the actual process of analysis. With the clock, the time-telling function is all that might be examined. But there are other aspects that could be relevant, such as the readability of the dial, the overall size, weight and colour, etc., which might be relevant to whoever uses the clock. The composition of the materials in each of the parts could be relevant to the robustness, accuracy and lifetime of operability. The process of taking it apart to see how it works would need to take account of how it was assembled: cutting it up with a hacksaw, for example, would not be a good way of analysing the processes of the mechanism.

The processes appropriate to the analysis of a manufactured device, a chemical, a rock, an organism, or an ecosystem are all different from each other. The value of any analytical process depends on the accuracy and completeness of the descriptions of everything in the analysis and on the validity of the reasoning employed. The process often leads to new information being discovered about either the parts or the whole or both. When new discoveries are made or new theories are devised, scientific theories are enlarged, modified or superseded

Sometimes no theory about a particular phenomenon can be devised that is completely consistent with the rest of science. More than one explanation may closely fit the evidence, each with different shortcomings. Then the theory that seems to be the best fit has to be tentatively accepted. Deciding which theory is the better fit is a continuing cause of controversy within science, which can be resolved only by re-examining previous findings, making new discoveries, or developing a better theory. But even when the evidence appears to support the observations or theories there may still be unknown shortcomings. This does not mean that science is inherently unsound, but that there is always the possibility of there being something missing. But the more completely and exactly the theory matches the evidence, the smaller the probability will be that there is.


In his book Why Us: How science discovered the mystery of ourselves James Le Fanu claims that the reason science has reached an impasse in the areas of fundamental physics, cosmology and an explanation of consciousness is because it is constrained by reductionism, which he says inherently excludes any consideration of some additional, non-physical, component that he presumably believes exists. To introduce such an entity as a component of a scientific explanation he would need to describe laws that govern that entity and show how they relate to the explanation. For the most part, I do not see any reason to think that such a non-physical material entity exists. I think that some non-material entity might exist, but I can go no further than suggest some possible characteristics that it might need in order to supplement an otherwise intrinsically impossible scientific explanation. (This is discussed in more detail in Chapter 4 The Nature of the Supernatural and Chapter9 The Hard Problem of Consciousness.)

In the course of trying to devise a reductionist explanation it often appears that there must be a missing, unidentified part. After working out what the missing part must be like, a search – or a race – is likely to be made to find it. The element helium and the planet Neptune are just two examples of missing parts of explanations, and their existence and characteristics were correctly predicted before they were discovered.

Dark matter and dark energy are examples of suggested missing parts in the explanation of aspects of cosmology. They have yet to be discovered. In this book I have suggested that the supernatural might be a missing part in the explanation of life, consciousness and the beginning of the material world. So, to answer Le Fanu, reductionism does not intrinsically exclude any non-physical entity, although I acknowledge that very few scientists would seriously suggest including it. But an example of where a supernatural entity might be invoked is in the explanation of consciousness.

Consciousness, as distinguished from intelligence, seems to be non-material and beyond scientific explanation. While it is possible to identify how the changes in the configurations of the networks of neurons in the brain produce intelligence, there seems to be no way the relevant configurations could produce consciousness. Neurons stimulate muscles to perform all their required functions in walking, talking, seeing, etc., but there no apparent equivalent organs to be stimulated to produce the feelings of pain, jealousy, pleasure, itchiness, etc. And, if such organs were discovered, I am unable to imagine how they could produce these sensations. So I would tentatively allow the possibility of a supernatural explanation. But this poses a problem in invoking a supernatural element in the explanation.

In any reductionist explanation, the contribution of every one of the parts, material or otherwise, must be an essential ingredient and not just an unnecessary extra. So, when a life force or a supernatural entity or anything else is suggested as a missing part of an explanation, its role cannot be acknowledged unless its essential contribution can be defined and its operation explained. Mere acknowledgement, however, is still not enough.

In the case of dark matter and dark energy, for example, which are claimed to contribute respectively an attractive force like gravitation and a repulsive force like that of electrons to other electrons, there is as yet no evidence that they actually exist. And so some theorists have proposed alternative answers to the phenomena they are purported to explain.


How then can reductionism be justified? There is certainly no reason to think that science explains everything, or that all the things that science explains are explained perfectly or absolutely correctly. Nevertheless, most of science seems to fit fairly well as a coherent body of knowledge and be a fairly good representation of what we see in the material world. But what needs to be justified is not the reliability of present-day science, but whether reductionism is an appropriate or valid method of scientific advancement. The arguments justifying reductionism depend on the increasing scope and reliability of science. The objections relate to the validity of the assumptions that scientific explanations rely on, the validity of the reasoning, and the accuracy of the observations.



The previous paragraph referred to gaps in the explanations of certain phenomena. That is, some missing part still needs to be found, which might, for example, be the agency referred to as dark energy. But there is another kind of missing part; the one implied in the saying the whole is greater than the sum of its parts, which applies to the vast majority of cases.

A complete clock, assembled with knowledge and skill, can do more than the sum of what the unassembled parts can do. The difference arises from the applied knowledge, skill and effort, and the use of appropriate tools required to assemble the clock. Is reductionism blind to this fact?

Similarly, water is very different from hydrogen and oxygen in the form of gases. As an example, water can exist in the form of vapour, of liquid, of formless ice and of crystals. The crystal form occurs as snowflakes, which can assume an almost limitless number of shapes. Yet in each case it is still an assembly of molecules of H2O. Particularly in the form of snowflakes, it could look as if water can become more than just the sum of its parts. At every level of complexity in the natural world, people will point to the differences between the whole and its parts, and ask where the new characteristics came from. Is reductionism blind to this fact?

One answer is that in nature the new characteristics were provided by some entity outside the material world of science, something supernatural. This can be a comforting answer, because reductionism implies a mechanical determinism, which may feel OK for basic physical and chemical processes, but uncomfortable when reductionism is applied to living things, including us. Also, it seems reasonable. In nature there is no noticeable assembly plant with a staff of experts and their tools, so there must be an unseen skilled agency.

There is indeed an agency, and its process is known as emergence. But it is not supernatural. Emergence is the counterpart to reduction: reduction is analysis and emergence is synthesis. Emergence pervades the natural world. It is an outcome of the operation of the laws of science. By the laws of science I mean our descriptions of the forces of nature that we observe, that is, the gravitational, electromagnetic and nuclear forces.

Nature’s parts assemble themselves, unlike those of manufactured goods. The parts in a clock do not spontaneously join together to produce a connected system. And there are not many different ways – other than swapping one screw or cog for another, etc. – in which they can be usefully assembled. But electrons, protons, atoms and molecules intrinsically interact, and they connect in ways that can become extremely complex as the assembly grows in size. Also, the number of possible arrangements increases enormously as the number of components or arrangements increases. In Nature, each particular arrangement is merely one of the myriad possibilities, each with its own characteristics – its solidity or fragility, its reactivity or inertness, etc.


Each of the many forms of water is produced automatically by well-known forces acting in an environment where temperature, pressure and other conditions determine the specific outcome. With snow, the precise details of each of these influences, including the movements of tiny specks of dust in the cloud where the snow is formed, act as a system and have a critical effect on the form of each individual flake. So although something produced by the unplanned processes of nature can be analysed to find how it works, it may be impossible to predict many of the possible ways a particular group of its parts could be assembled, or how the assembly would behave. It may also be impossible to predict the effect of an alteration or disturbance.

To illustrate how something like this might happen, think of very flexible buoyant sticks with fishhooks at both ends. One in isolation may seem very simple. But toss a pile of them into a bucket and shake them vigorously and one or more tangled masses will be produced. Throw several such tangles onto a pool of water in which there are a lot of fluffy tennis balls and stir them around. Repeat this several times with the products of each round being added to the previous result. A range of small rafts would probably be produced, with size and shape depending on the number of balls, the sizes of the tangles of sticks, and apparently haphazard things like the manner of stirring.

If the sticks and balls were strongly interactive and there was a larger range of interacting objects, and the conditions of mixing were different every time the experiment was tried, then it would become increasingly difficult to predict what would be formed. But it would be difficult to argue that the whole conglomeration was more than the sum of the parts, particularly when the mechanical actions of shaking and stirring, or in other words, the process, were recognised as parts.

Nature is much more complex than this example. It is more diverse in type and more reactive in many ways. Among the types of structures that can emerge naturally are systems of interacting parts whose operation changes in response to changes in their environment. For example, they may swell or contract or change shape depending on whether they are wetter or dryer, or hotter or cooler, or change on contact with a specific chemical or in an electrical field.

Combinations can be produced in which their characteristics become “non-linear”, that is, small changes can produce effects that are disproportionately large, or large changes produce effects that are disproportionately small, or sometimes no change at all. Without knowing the precise conditions it becomes impossible to predict the outcome: but once there is an outcome, it can be analysed into its component parts, with the processes of its assembly explained. The outcome may, and generally does, have properties that seem to be greater than the sum of the parts. These are emergent properties. And this applies not only to inanimate material but also to living tissue and organisms.

It is not only tightly connected systems that produce emergent properties. Flocks of birds, shoals of fish, colonies of insects and microorganisms act as organised systems to find food, protect their members and arrange their environment. It has been found that the individual members of such groups act in accordance with a few simple rules that apply to particular aspects of their environment, which includes members of their own and other groups. (While these animals are the “parts”, they are all complex within themselves.)


When looked at individually, the members or parts of any kind of system display only a part of their characteristics. But when they interact with something, additional characteristics are revealed. An atom of hydrogen by itself is just a proton and an electron held in partnership by electrical attraction and sub-atomic forces. When brought close to another atom it is seen to be chemically active. Its reaction with certain types of molecules produces an acid, and when the acid contacts water the hydrogen atom dissociates from the main body of the acid, leaving its electron behind and becoming a positively charged hydrogen ion, which then loosely associates with a molecule of water.

A virus by itself is just a deeply textured bag made of proteins and containing some genetic material. But when it comes in contact with a living cell whose coating has a texture that it can latch on to – like the two textures of velcro or a burr hooking onto clothing – the genetic material of the virus can pour into the cell and the cell will assemble copies of the virus. If the cell is already infected with a virus, some of the genetic material of the two viruses may be combined, which will sometimes result in new strains of one or both viruses. (Some swapping of virus DNA or RNA with the DNA or RNA of the host cell may also occur.)

And human beings in isolation do not reveal the all characteristics that determine how they will behave when interacting with one other person, or in a small group, or in a mob.

So unexpected inherent characteristics of something may reveal themselves when it interacts with various parts of their environment. These “hidden” characteristics must be included when one adds up the parts to equal the whole, as they contribute to the emergent properties of the whole.


But is reduction really sufficient to show how every aspect of every kind of observed phenomenon occurs? Reductionism often becomes less reliable as things get more and more complex and generally more poorly understood, as in the workings of brains or ecosystems or societies. And when something that is already very complex is subject to imprecisely known influences, then analysis and prediction become even more uncertain. A reductionist analysis can go only as far as the available information and analytical skills allow. Further resolution may, of course, be found by further research.


Actually, it is never possible to know whether everything has been discovered about something. While this does not mean that there must necessarily be some external non-material entity whenever there is an unexplained detail, some people hope it might. So whether an explanation is sufficiently complete could be one of opinion – or is it just one of emotion?

It might be possible to answer this question by looking at some assumptions that underlie science. Two principles that are intrinsic to scientific reasoning are causality and consistency. Causality means that everything that happens is the result of some process. Consistency means that each (precisely defined) type of process always operates in the same way. (These principles might appear to be contradicted by probabilistic quantum phenomena at the scale of atoms and smaller particles, but they are supported by observations at larger scales.)

Causality and consistency lead to the principle that all scientific theories must be compatible. Reductionism, as a part of scientific reasoning, includes the assumption that these principles are valid. This assumption is supported by the high (but not perfect) level of consistency between scientific theories. With the exception of unresolved issues, all scientific theories remain in accordance with observed phenomena.

If causality and/or consistency were untrue, might an apparently emergent characteristic be attributable to the action of some non-material entity? It might be argued that these principles appear to be true only because some non-material entity makes them appear so. But invoking such a “helpful” entity would not affect the status of science as a reliable explanation or as a foundation for research and invention.

It could also be argued that, since we cannot know that these principles are true in every possible case, inconsistently anomalous phenomena might sometimes occur as the cause of a scientific mystery yet to be solved. There could, for example, be truly random phenomena that break the rules of causality and consistency. An example might be the quantum process of decay of individual atoms of a radioactive element, in which the time intervals between successive instances of decay are haphazard enough to be used to provide pseudorandom sequences of numbers. But at larger scales of size, radioactive elements each have their own specific period of time that it takes for half of any particular piece to decay.

Also, there are consistent but inexplicable phenomena. Examples are the accelerated expansion of the universe, and galaxies that spin but unexpectedly avoid flying apart. These have been attributed to “dark energy” and “dark matter”, neither of which has yet been detected or explained. Since a reductionist explanation requires that all the interacting agencies that are involved in a process need to be identified and their characteristics examined, both dark matter and dark energy and dark matter must be regarded to be hypothetical.

A reductionist examination considers the thing in isolation or with its near environment. This leads to one of the arguments against reductionism.. The issue can be illustrated by thinking about the workings of a motorcar. When analysing a car in order to understand all of its properties and functions, is it sufficient to consider the car in isolation, or should the driver and the road be included in the analysis? If the answer to this question is ‘yes’, should other drivers in their cars, and more roads, and other kinds of things also be included? And if so, where do we stop?

This harks back to the claim that “failure to be holistic” makes reductionism incomplete and unreliable. I think that this unreliability is not unlike that of induction, in which “enough” examples of a particular phenomenon are sufficient to justify claiming the phenomenon always happens in a certain way, and, if the physical explanation is accurate, logical and complete, there is no need to introduce some extraneous additional entity. In reductionism, if the thing being examined plus its immediate environment is sufficient to explain what needs to be explained, there is no point in looking more widely. If it is not sufficient, a wider analysis is made. Neither induction nor reduction is completely infallible, because it is impossible to include all possible occurrences of a phenomenon or to include all of the ever-expanding environment of whatever is being analysed. But they are the best tools that are available, and in practice account is taken of apparently inconsistent phenomena and cases of external influence. And often, for some purposes, it is enough to examine only the particular thing in question.

When the whole of science is considered, in contrast to one particular phenomenon, reductionism makes induction more reliable. Reductionism, through its analysis of the processes that produce phenomena, provides explanations based on already-established theories and explanations. It may be argued, of course, that these underlying theories are also based on induction, and this argument can be extended to every part of science The requirement that all scientific explanations have to be logical, based on (inductive) evidence and be consistent with each other greatly strengthens science’s claim to reliability although it does not absolutely confirm the validity of science.


If any unknown entity, either natural or supernatural, is proposed in order to account for an otherwise unexplained emergent property, then some description of that entity is needed. For a natural entity to be credible its description should include how the entity might produce the emergent property and the explanation should be capable of being tested. This would not be possible if the proposed entity were supernatural, because we can only conjecture what characteristics a supernatural entity might have. And this means that the justification for any such entity must always be tentative.



So, does reductionism make science essentially deterministic and atheistic? Does it make science narrow-minded or can reductionist science actually be holistic? Does it make science simplistic or unreliable?

If every theory in science is logically consistent with every other (or when theories are not compatible the differences are acknowledged as problems to be resolved), if the idea of consistent cause-and-effect is true, and if logic is entirely valid, then science must be entirely deterministic. That does not mean that everything in the future could then be predicted. To do that would require knowledge of everything about the present, which is intrinsically impossible. Even then, chaos theory would require absolute precision and quantum uncertainty blurs precision.

Chaos can amplify uncertainties whenever a process is repeated. This makes prediction increasingly imprecise as it moves further into the future. The same problem also affects reconstruction of the past. The past, of course, has left many relics, which can provide checks, but we can never be sure we have reconstructed completely accurately. The inherent imprecision of quantum theory might be explained as merely the granularity or “graininess” of nature, or as the result of our present lack of knowledge of some principle underlying quantum theory, or as room for some supernatural entity to slightly influence the material world. Nevertheless, reductionism does imply that the whole world is just a very complex machine running within clearly defined, although sometimes imprecise limits.

What does this say about the idea that reductionist science “cannot see the wood for the trees”? Science certainly sees the trees. It sees the parts of the trees – leaves, bark, branches, sap, etc. It sees the veins and the cells and other parts of the leaves, and it sees the molecules they are composed of. And it also sees the wood, and the landscape that the wood is in, and the continent the landscape is in, and the planet, and so on, even, possibly, to a multiverse. But each tree is a whole in itself, as is each leaf and cell and molecule. And so is the ecology of the forest, and the continent, and the solar system, etc. Science looks at each as a part but also as a whole in its own right.


The unity of science leads to the view that everything is an interconnected whole. Since science deals only with the natural, i.e., observable, world, the supernatural is, by definition, excluded from it. This implies that whatever can be observed, measured and analysed, and which displays some consistent underlying relationship between observation and cause, must implicitly be material not supernatural.

Acceptance of science as the current explanation that best fits what is observed about the world does not necessarily imply atheism. That depends on how you define atheism, and on your view of what relationship a supernatural agency might have with the material world. People who believe there exists a non-material entity (and this includes some scientists) may think that science is narrow-minded – although they might also accept its findings.

Reduction would make science narrow-minded if it opposed “looking at the bigger picture”. But it doesn’t. Some scientists, however, can become so wedded to a particular scientific principle that they will not entertain any argument of evidence that contradicts or modifies it. (See Chapter 13 The Will to Believe – and to Disbelieve.) Anyone who is analysing a situation, scientific or otherwise, can choose any starting point among all conceivable levels of size or complexity.

Reductionism does not explicitly exclude the supernatural, but scientists would need to have good reason to prefer a supernatural to a material explanation. The only justification that I can envisage for invoking a possible supernatural agency is in cases where there is some reason for thinking that science is intrinsically unable to provide an explanation. As already stated, I do not think there is reason to do this in the theory of quantum mechanics or in cosmological issues such as the nature of dark energy or dark matter.

Science rests on the validity and completeness of its observations, which are continually published and checked, and on the validity of the assumptions it is based on, which are continually examined and argued. Despite the care with which science has been put together, there is always the possibility that any observation it rests on has been misinterpreted, or that phenomena yet to be observed will contradict present theories, or that the reasoning that explains the observations contains unnoticed incorrect assumptions. This does not mean that any particular part of science must be wrong. Most of it matches very closely the world as we see it. Scientists continually strive to widen its perspective, modify its theories accordingly, and refine its accuracy.

There will always be statements made in the name of science but based on inadequate information and/or poor reasoning. Usually they are challenged, but many are accepted or defended on faith or tradition. So science, while always being intended to be rigorous, is sometimes justly criticised. But criticising it merely because it is reductionist is unjustifiable. Properly used, reductionism, as one of its key tools, by no means makes science simplistic or unreliable.


In the opening section of this chapter I mentioned that notionally, although not practically, the workings of human society might be considered to be explainable in terms of fundamental physics. Now I am going to suggest that also, and perhaps a little more practically, the laws of fundamental physics might be thought to be explainable in terms of the characteristics of human society. Further, I would suggest that every step in the hierarchy of reductionist explanation could be thought of as a particular, but incomplete, view of reality. Reductionism suggests that the concept of “a thing” or of “reality” must always mean many things, not just one – at least to the limited understanding of human beings.

Reductionism suggests that the concept of “a thing” or of “reality” must always mean many things, not just one – at least to the limited understanding of human beings. More than two thousand years ago the philosopher Plato wrote a parable about prisoners watching shadows that flickered on the wall of a cave, thinking they were seeing reality but on being released saw the outside world. But Plato was implying that what we see as reality is still only the shadows of some unknown ideal forms. The eighteenth century philosopher Immanuel Kant looked for the “thing in itself” (Ding an sich) as distinct from what we think we see. The twentieth century philosopher Ludwig Wittgenstein wanted a language that had real meaning, i.e., that faithfully described tangible things or processes. Perhaps reductionism, in all its levels, provides the closest that humanity, limited in its ability to sense the world and to interpret what it senses, will ever get to satisfying the hopes of Plato, Kant, Wittgenstein and all others who have strained to find true reality.

But how could we ever know how close to reality any item of science, or anything else, is?