The Role of Agency in Science

By Angus Menuge

1. Agents

Human beings habitually understand themselves as agents. An agent is an individual with reasons for its behavior. Agents have goals (things they desire), and produce behavior which they believe will achieve those goals. Thus Hans, a native of Wisconsin, positions his step-ladder under the roof because he believes his gutters are clogged and desires to save them from ice-damage. So much is part of the intuitive self-understanding called folk psychology.

2. Scientific materialism

The very idea of agency is problematic for scientific materialism. According to scientific materialism, everything that happens can be explained by the undirected behavior of matter. An event can occur as the result of a lawful regularity, or because of chance, or because of a combination of law and chance, but if matter is all there is and matter has no goals, then the appearance of goal-directed behavior is difficult for the scientific materialist to explain. Some materialists, like Paul and Patricia Churchland, think that agency is incompatible with materialism, and must be eliminated in favor of the materialistic categories of neurophysiology. This strategy is eliminative materialism. Other materialists, like Daniel Dennett, Fred Dretske, and Jerry Fodor, see agency as crucial to our self-understanding as rational beings, but hope to show that it can naturally arise from materialistic categories. This strategy is called naturalism of the mental.

3. Eliminative materialism

According to the Churchlands, folk psychology, with its talk of goals and purposes, beliefs and desires, is simply the last vestige of a pre-scientific worldview already largely displaced by the advances of scientific materialism. The laws of physics make no mention of the “goals” of heavenly bodies. Darwin, it is claimed, removed the need to speak of an intelligent designer of living organisms. The last frontier to be conquered is the human mind.

For the Churchlands, there are no such things as beliefs and desires. Beliefs and desires have to go because they have intentionality: Unlike material entities, beliefs and desires include thoughts about something with propositional content, such as the belief that gasoline prices have risen or the desire that gas prices will fall. Thoughts can be true or false, they may lack a real object (e.g. beliefs about leprechauns) and they may jointly provide a reason for an action, such as buying a hybrid automobile. By contrast, material events either happen or do not happen (but cannot be true or false); they can only stand in causal relations with other material events (so they cannot point to non-existent objects), and though they may cause a behavior, they do not give an agent’s reason for doing it.

The Churchlands maintain that the way forward is shown by advances in the brain sciences. Here, cognition is reduced to transformations of neural activation patterns. These patterns look nothing like propositional contents, nor is there anything analogous to an agent’s reason for acting. From this perspective, folk psychology appears redundant. However, there are a number of important reasons to reject eliminative materialism.

First, there is the abstraction problem. Neural activation patterns may explain specific bodily movements, but do not capture the abstract specification of human actions. Consider the action of greeting. One may greet someone with spoken words, a smile, a wave, a hand-shake, a hug and/or kiss, a card, an electronic message, an airborne banner, etc. Since each of these methods involves physically different behaviors, each requires a different set of neural activation patterns to explain it. But this fails to capture what all of the behaviors have in common, the fact that they are all actions of greeting. By contrast, folk psychology can appeal to a common desire or intention to greet. The categories of folk psychology are at the right level to account for actions, and not merely specific movements.

Second, there is the subjectivity problem. The transitions of neural activations are completely impersonal and in no way involve a point of view. But there is no doubt that there are subjects, individuals with distinct points of view. This has always been recognized by folk psychology since it seeks to provide personal reasons for an agent’s actions. Jack does not (ultimately) open the fridge because Jill believes it contains a beer. Note that it will not help the eliminativist to claim that points of view are illusory, since only something with a point of view can be subject to an illusion.

Third, there is the robustness problem. The eliminativist argues that folk psychology is silent about abnormal psychology, so it is probably a false theory, ripe for displacement. But when folk psychology goes, so will its ontology of beliefs and desires. However, as a normative discipline, folk-psychology is no more refuted by abnormal psychology than classical logic is refuted by fallacious reasoning. And, even if folk psychology were shown false, that would not discredit its ontology of beliefs and desires. At one time, folk physics supposed that weights were intrinsic, constant properties of bodies. The discovery that weight is a relation that varies with gravitation shows that the folk conception of weight needed reform. But no one concluded that weights do not exist. Given the entrenched status of the concepts of belief and desire in our self-understanding as rational beings, it seems much more likely that empirical failures in folk psychology would provoke a reformed conception, rather than elimination, of its ontology.

Last, there is the coherence problem. Paul Churchland has never provided a coherent explanation of what it would mean to abandon folk psychology. He cannot describe it as rejecting folk psychology and accepting a neural replacement, because the whole idea of rational rejection and acceptance is part of folk psychology. More fundamentally, eliminativism undermines the rationality of science because it removes anything that could count as reasons for the scientist’s actions. If there are no intentional states, then scientists do not literally design their experiments, analyze their data, infer conclusions and devise alternative theories. But then there is no reason to say that scientific practice is rational. Although scientific materialists claim that science provides the reasons to be a materialist, materialism undermines the very idea of having a reason for anything. Any worldview capable of defending the rationality of science must be one that allows scientists to have identifiable reasons. Since it allows intelligent and goal-directed causes as part of nature, Intelligent Design is in the right position to do this.

4. Naturalism of the Mental

Even among materialists, eliminativism is unpopular. Most materialists believe in naturalism of the mental: Intentionality and other problematic mental categories are real, but can be shown to be compatible with materialism. There are two main problems for this project: (1) There is good reason to think agency cannot be understood in purely materialistic terms; (2) If human agency is simply declared part of nature, then the naturalist has abandoned materialism by allowing the existence of directed causation, and in that case, can no longer exclude the possibility of non-human (and possibly divine) agency active in nature.

4.1. The failure of naturalism

Early on, naturalists proposed the Identity Theory, according to which types of mental states could be identified with types of brain states. Thus, perhaps all pain states are C-fibers firing. The theory fell, because empirical study of the brain revealed important structural differences between the brains of creatures who feel pain. There is no interesting neural similarity between all and only creatures who feel pain. However, there are characteristic physical causes of pain, and characteristic responses to pain, among diverse creatures. So functionalism suggested that pain, and all other mental states, are states with the functional role of mediating characteristic causes and effects. Since the same causal role can be physically realized in different ways (just as a mousetrap can be built in many ways), the objection to identity theory was avoided.

But functionalism has numerous troubles of its own. Most fundamental is the problem of giving an illuminating materialistic account of functions. Functions can be understood either non-teleologically, as formal mappings between elements, or teleologically, as contributing to a system’s goals.

In the non-teleological case, crucial features of the mental are not explained. Impersonal mappings give no account of the subjective experience of being in a mental state such as pain. The functionalist theory could hold perfectly for some creature that had the right behavior but did not feel pain. Functionalism might be satisfied in a world of zombies. Further, the mappings between physical elements do not exhibit intentionality (they are not about anything). For example, a computer could map questions about baseball to their correct answers using a large database. It would then fulfill a functionalist account of “understanding” baseball, even though the computer has no baseball concepts and outputs the answers merely by matching the form of uninterpreted elements. And finally, impersonal mappings do not explain an agent’s personal reasons for action.

On the other hand, if functions are viewed teleologically, then naturalism may account for the mind, but only by allowing that directed causation is part of nature. Yet this directly contradicts materialism. Thus either functionalism is unable to naturalize important features of the mental, or it does so by abandoning materialism.

As an ingenious alternative, Dennett proposes that intentionality can be fully explained by “Mother Nature,” or natural selection. Dennett thinks that Mother Nature has a low-level “intentionality” in that it makes “choices” that better serve organisms. But there is no good reason to accept this. Natural selection has no reasons for its “choices,” and it has no goals, only selecting on the basis of past performance. As I have argued previously elsewhere, Dennett equivocates between an authentically materialist Mother Nature, which offers no explanation of intentionality, and a mythological Mother Nature, who possesses intelligent characteristics incompatible with materialism.

4.2. The legitimacy of ID as science

The second problem for naturalism is that if there is even one case (the human one) in which the evidence shows that agency is an irreducible feature of reality, then materialism is false and agency is a legitimate causal category for scientific explanation. But then the question of whether agency is evident anywhere else in nature can only be settled empirically. Granted the reality of irreducible agency and the existence of rigorous criteria for detecting its effects, only empirical data can decide whether design is manifest in biology, cosmology, and elsewhere. Thus Intelligent Design is manifestly a legitimate scientific research program.


Agency is the Achilles’ heel of scientific materialism. If the materialist eliminates agency, he undermines the rationality of science. But agency also fails to reduce to materialistic categories. So, if we want to preserve the rationality of science and follow the evidence wherever it leads, we must conclude that agency is an irreducible causal category. And that is precisely the claim of Intelligent Design.


Published March 30, 2016