Chapter 27: Reductionism vs Emergence

The most influential philosophical account of intertheoretic reduction was developed by Ernest Nagel in The Structure of Science (1961). On Nagel’s model, a theory T₂ (the reduced theory, e.g., classical genetics) is reduced to a theory T₁ (the reducing theory, e.g., molecular biology) when the laws of T₂ can be logically derived from the laws of T₁, supplemented by bridge laws that connect the vocabularies of the two theories.

Bridge laws take the form of biconditionals: “An entity has property P (in T₂) if and only if it has property Q (in T₁).” For example, a bridge law for reducing genetics to molecular biology might state: “Gene G is dominant if and only if the protein encoded by G is functional at half the normal concentration.”

The paradigm case of successful Nagelian reduction is the reduction of thermodynamics to statistical mechanics. The macroscopic concept of temperature is identified with (bridge-linked to) the mean kinetic energy of molecules. The gas laws (e.g., Boyle’s law) can then be derived from the statistical mechanics of molecular ensembles.

“Reduction... is the explanation of a theory or a set of experimental laws established in one area of inquiry, by a theory usually though not invariably formulated for some other domain.”— Ernest Nagel, The Structure of Science (1961)

But can this model be extended to biology? Can Mendelian genetics be reduced to molecular biology? Can neuroscience be reduced to molecular biology? Philip Kitcher’s landmark paper “1953 and All That: A Tale of Two Sciences” (1984) argued that the reduction of classical genetics to molecular biology fails because the bridge laws required are hopelessly complex and disjunctive. Mendel’s concept of a “gene” does not map neatly onto any single molecular structure; genes are realized by different DNA sequences in different organisms, and even within a single organism, the relationship between genes and DNA is many-to-many.

The most influential philosophical objection to reductionism is the argument from multiple realizability, developed by Hilary Putnam (1967) and Jerry Fodor (1974) in the context of the philosophy of mind, but with direct implications for biology.

The argument is straightforward. A higher-level property is multiply realizable if it can be implemented by different lower-level physical configurations. Pain, for example, might be realized by C-fiber firing in humans, by some other neural configuration in octopuses, and by silicon circuits in a hypothetical robot. If pain is multiply realizable, then it cannot be identical to any particular physical state, because it can be present without that physical state and (arguably) that physical state can be present without pain.

“The argument against reductionism is not that it is false but that it misses important generalizations. The special sciences capture regularities that would be invisible from the standpoint of fundamental physics.”— Jerry Fodor, “Special Sciences” (1974)

In biology, multiple realizability is pervasive. Consider the concept of fitness. An organism can be “fit” in an indefinite number of ways: by being faster, or better camouflaged, or more resistant to disease, or more attractive to mates. There is no single physical property that constitutes fitness. Similarly, the concept of a gene is multiply realized: genes are stretches of DNA, but different genes are constituted by entirely different nucleotide sequences. There is no single molecular structure that is “the gene.”

Fodor drew a powerful conclusion: if higher-level properties are multiply realizable, then higher-level sciences (biology, psychology, economics) capture genuine generalizations that would be invisible from the standpoint of physics. The generalization “organisms with higher fitness tend to leave more offspring” applies to all living things, regardless of their physical constitution. A physical description of each individual case would miss this regularity entirely.

Multiple realizability thus provides a principled reason for the autonomy of the special sciences: they are not merely convenient abbreviations for fundamental physics but capture real patterns in nature that physics cannot capture. This is the philosophical foundation for the claim that biology is an autonomous science.

The concept of supervenience provides a middle ground between reductionism and dualism. Biological properties supervene on physical properties in the following sense: there can be no difference in biological properties without a difference in physical properties. Two organisms that are physically identical must be biologically identical. But the converse does not hold: two organisms can be biologically similar (e.g., both “fit”) while being physically different.

Supervenience captures the intuition that biology is dependent on physics without beingreducible to physics. It is a relation of determination without identity. The physical facts fix the biological facts, but the biological facts are not identical to any particular physical facts because of multiple realizability.

The philosopher Jaegwon Kim has argued that supervenience without reduction is an unstable position. In a series of influential papers culminating in Mind in a Physical World (1998), Kim argued that if we accept both supervenience and multiple realizability, we face a dilemma: either higher-level properties are epiphenomenal (causally inert), or they are reducible after all (via “local reductions” to their particular physical realizers in each case).

Kim’s “causal exclusion argument” is that if the physical is causally closed — every physical event has a sufficient physical cause — then there is no causal work left for higher-level properties to do. Either biological properties are identical to physical properties (reductionism), or they are causally impotent (epiphenomenalism). This argument has generated an enormous literature, with responses ranging from nonreductive physicalism to emergentism to structural realism.

Alexander Rosenberg has been the most persistent and sophisticated defender of reductionism in the philosophy of biology. In Instrumental Biology, or The Disunity of Science (1994) and Darwinian Reductionism (2006), Rosenberg argues that molecular biology provides the correct level of explanation for all biological phenomena.

Rosenberg’s argument proceeds in two steps. First, he argues that biology has no laws of its own. Biological generalizations (“organisms with higher fitness tend to leave more offspring”) are too riddled with exceptions to count as laws. The only genuine laws in biology are the laws of chemistry and physics that govern molecular interactions. Second, he argues that molecular biology does provide adequate explanations of biological phenomena, even if these explanations are complex and context-dependent.

“The reason why there are no laws in biology is that biological systems are jerry-built by natural selection. Every apparent regularity has exceptions because natural selection operates on historically contingent variation.”— Alexander Rosenberg, Darwinian Reductionism (2006)

Critics respond that Rosenberg confuses ontological reduction (everything is made of physical stuff) with explanatory reduction (everything is best explained at the physical level). Even if organisms are nothing but molecules, it does not follow that molecular descriptions provide the best explanations of biological phenomena. The generalization that “predators tend to be less numerous than their prey” is a genuine ecological regularity that molecular biology cannot even state, let alone explain.

Emergence is the thesis that wholes can have properties that are not possessed by, and cannot be predicted from, the properties of their parts. The concept has a long history, going back to the British emergentists of the 1920s (C.D. Broad, Samuel Alexander, C. Lloyd Morgan), and has experienced a revival in recent philosophy of science and philosophy of mind.

It is crucial to distinguish two forms of emergence:

The philosopher Mark Bedau (1997) introduced the useful notion of “weak emergence” as a middle ground: weakly emergent properties are derivable from micro-level facts but only through simulation, not through analytic derivation. Complex adaptive systems (flocking behavior, market dynamics, cellular automata) exhibit this kind of emergence. The macro-level patterns are entirely determined by the micro-level rules, but they cannot be predicted without actually running the simulation.

In biology, the question is whether any biological phenomena are strongly emergent. Most philosophers of biology are skeptical of strong emergence in biology (as opposed to consciousness), but many argue that biological phenomena exhibit a form of emergence that is stronger than mere surprise but weaker than metaphysical irreducibility. The challenge is to articulate this intermediate position precisely.

Downward causation is the thesis that higher-level properties can causally influence lower-level processes. If true, it would be a powerful argument against reductionism: the behavior of parts would depend not only on lower-level interactions but also on the properties of the wholes they compose.

Examples of apparent downward causation in biology are abundant. A cell’s behavior depends on what tissue it belongs to (the same cell type behaves differently in different organs). An organism’s stress response — a whole-organism state — affects gene expression at the molecular level. Social structures (a dominance hierarchy in a primate troop) affect the hormonal states of individual organisms.

However, downward causation faces a serious philosophical challenge: Kim’s causal exclusion argument. If every molecular event has a sufficient molecular cause (the causal closure of the physical), then there is no causal work for higher-level properties to do. The apparent cases of downward causation are really cases where the physical microstate of the whole system causes changes in the microstates of its parts — it is micro-to-micro causation all the way down.

Defenders of downward causation respond in several ways. Some deny causal closure: perhaps the physical is not causally complete, and higher-level properties inject novel causal powers into the world. Others adopt a constitutive rather than causal model: higher-level properties do not cause changes in lower-level properties but rather constrain or structure them. The new mechanistic philosophy, discussed below, offers a framework in which interlevel relations can be understood without the problematic notion of literal top-down causation.

Since the early 2000s, the new mechanistic philosophy has emerged as a powerful alternative to both classical reductionism and emergentism. Developed by Peter Machamer, Lindley Darden, and Carl Craver in their landmark paper “Thinking About Mechanisms” (2000), the mechanistic approach characterizes scientific explanation in biology in terms of mechanisms — organized systems of entities and activities that produce regular changes from start to finish.

“Mechanisms are entities and activities organized such that they are productive of regular changes from start or set-up to finish or termination conditions.”— Machamer, Darden, & Craver, “Thinking About Mechanisms” (2000)

The mechanistic approach has several advantages:

• •Multilevel explanation: Mechanisms are inherently multilevel. A mechanism has components (lower-level entities) that are organized to produce a higher-level phenomenon. Mechanistic explanation naturally integrates different levels without requiring either reduction to the lowest level or appeal to mysterious emergent forces.
• •Organization matters: The same components arranged differently produce different phenomena. This captures the anti-reductionist intuition that wholes are more than the sum of their parts, while explaining why they are more: the organization of the parts makes a difference.
• •Faithful to practice: The mechanistic model accurately describes what biologists actually do. Molecular biologists discover mechanisms (of DNA replication, protein synthesis, signal transduction); neuroscientists discover mechanisms (of synaptic transmission, memory consolidation); ecologists discover mechanisms (of population regulation, community assembly).

Craver and Bechtel (2007) developed the concept of “mechanistic levels” as a replacement for the traditional layer-cake picture of levels (physics, chemistry, biology, psychology). On their view, levels are local, defined relative to particular mechanisms, not global. The level of the synapse is defined relative to the mechanism of neurotransmission; the level of the gene is defined relative to the mechanism of protein synthesis. This eliminates the question of whether “all of biology” reduces to “all of chemistry” and replaces it with specific questions about specific mechanisms.

Ernst Mayr argued throughout his long career that biology is fundamentally different from physics and cannot be reduced to it. In Toward a New Philosophy of Biology (1988), Mayr identified several respects in which biology is autonomous:

• •Historical explanation: Biology relies essentially on historical (narrative) explanation. The question “Why do birds have feathers?” requires an evolutionary narrative, not a deduction from physical laws.
• •Teleological explanation: Biology employs teleological explanation (function, purpose, design), which has no counterpart in physics.
• •Population thinking: Biology studies populations of unique individuals, not interchangeable tokens of types.
• •Dual causation: Biological organisms are subject to both proximate (mechanistic) and ultimate (evolutionary) causation. Physics has only proximate causes.

“The philosophy of biology is not simply a philosophy of physics applied to organisms. Biology has its own concepts, its own explanatory strategies, and its own standards of evidence.”— Ernst Mayr, Toward a New Philosophy of Biology (1988)

The autonomy thesis does not require any exotic metaphysics. It is compatible with physicalism (everything is made of physical stuff) and even with supervenience (biological properties depend on physical properties). What it denies is that this dependence implies explanatory reducibility. Biology is autonomous not because it describes a different realm of reality but because it captures patterns, regularities, and explanations that are invisible from the standpoint of fundamental physics.

• •Nagel, E. (1961). The Structure of Science, Chapter 11.
• •Putnam, H. (1967). “Psychological Predicates,” in Art, Mind, and Religion.
• •Fodor, J. (1974). “Special Sciences (Or: The Disunity of Science as a Working Hypothesis),” Synthese 28.
• •Kitcher, P. (1984). “1953 and All That: A Tale of Two Sciences,” Philosophical Review 93(3).
• •Rosenberg, A. (2006). Darwinian Reductionism, Chapters 1–4.
• •Machamer, P., Darden, L., & Craver, C. (2000). “Thinking About Mechanisms,” Philosophy of Science 67(1).
• •Kim, J. (1998). Mind in a Physical World, Chapters 1–3.