Home > Determinism, Philosophy, Philosophy of Mind, Philosophy of Science > Nonreductive Agent Causation Part II: Four Points of Analysis

Nonreductive Agent Causation Part II: Four Points of Analysis

February 20, 2012 Leave a comment Go to comments
Reduction -  Photo courtesy of Smittenkitchen.com

"But if nonreduction is true then we would never have Maple Syrup." Photo courtesy of Smittenkitchen.com.

In Part I of this two-part post I introduced an extended dialogue between Timothy O’Connor and Derk Pereboom that spans physicalism, reductionism, agency theory, and quantum physics. O’Connor posits a purely physicalist theory of agency based on the formation of macroproperties which instantiate in sets of microproperties which reach a certain threshold level of complexity. Once this level is reached, an emergent macroproperty, constituted as an agent causal power, can then enact downward causal influence over its microproperties without being subject to upward causation or determination from its constituent microproperties. Pereboom takes O’Connor to task for failing to account for the influence of distal causes, which nevertheless determine the behavior of the agent causal power, but to counter the invocation of an emergent property, Pereboom alleges that even in a statistical model rather than a deterministic one, we are still left with distal causes as the ultimate originator of action. In the comment section of a previous post, Aaron Kenna rightly makes mention of this, viz. that statistical, indeterministic, and deterministic worldviews all fail to provide the freedom required by agency theories/moral responsibility. In a future post I shall discuss this point further, using Strawson’s “basic argument” as an example. But for now, let’s turn to four points of analysis on the conversation between O’Connor and Pereboom to see what we can make of it.

 Four Points of Analysis

I have four points of analysis I would like to center on with regards to the conversation between O’Connor and Pereboom. They are: (I) emergent agent-causal libertarianism is a much simpler theory than others because it does not suffer from the problem of interaction. (II) Regardless of its parsimony in avoiding substance dualism, this agent-causal emergentism necessarily suffers from the issue of the plausibility of an undetermined, emergent macroproperty. (III) O’Connor claims that his thin account of agent causality only waits on developments in neuropsychology and other sciences for further development, but I am not convinced that this is so, nor do I believe many of his opponents should be. (IV) Finally, throughout the conversation it appears as though Pereboom and O’Connor come to a disagreement regarding reductionism that appears to end in stalemate, one which may be insurmountable.

I. One of the greatest strengths I see in O’Connor’s theory is that it successfully avoids the problem of interaction in a relatively novel way. By resisting the common libertarian move of slipping into dualism, as seen in Richard Taylor’s Metaphysics, O’Connor successfully and consistently classifies the human agent as, “a wholly biological organism, whose macroproperties are either constituted by or dependent on the properties of certain elementary physical particles, organized into complex subsystems at a number of levels.”[1] By classifying the agent as purely biological and his causal power as a macroproperty derived from microstructures, his theory is not subject to the question, “Yes…but if not physical then what exactly is the mind/self/soul/agent?” While it avoids this issue, it does face a similar question regarding what exactly constitutes the macroproperty, when it emerges, etc. which I discuss in (III), although this is not nearly as spiny an issue as the Mind-Body problem.

While strict substance dualism may not have many remaining champions in academic halls these days, soul/body dualism has given way to mind/body dualism such that one need not posit the existence of an atomic self (i.e. soul, atman, etc.) over and above an immaterial mind. By couching agent causal powers in terms of physical properties, O’Connor aligns himself on a naturalist-friendly line. And, as mentioned, he believes further scientific advances will only confirm and strengthen his theory. This is head and shoulders better than other theories of mind which may not answer naturalist inquiries as smoothly or consistently. I say this because it seems increasingly more difficult to ignore scientific advances and their possible implications on philosophical theories, from the recent surge in neuroscience studies focusing on free will and action theory, to ongoing discussions of biological reduction, etc. Despite this naturalist-friendly approach to agency theory, emergent agent causation as a theory relies upon the irreducibility of the emergent macroproperty by way of a supervenient relationship.

 II. Despite being slightly more parsimonious than some of its brethren, O’Connor’s theory seems to encounter trouble with the attempt to argue that the macroproperty derives from determined microsystems yet can act as it wills without being bound by similar determinism. As O’Connor describes it, this agent is a “mover not wholly moved.”[2] Though central to his account of agent-causal action, O’Connor spends a puzzlingly small amount of time describing the emergent macroproperty, and simply states that it is a macroproperty emerging from microstructures whose causal role, with respect to the object it arises within, is not reducible to the causes of the microstructure.[3] While this is perhaps a consistent account, it does raise the question of just how this is possible. For this reason, Pereboom’s criticism (represented by (1) in O’Connors reiteration) of this point is warranted. Intuitively it seems far more plausible that, because microsystems can be explained by laws, so too should more complex systems. Indeed, O’Connor bears the burden of proof that such emergent macroproperties can even make sense given that he accepts the determinacy of microstructures[4]. More explanation in way of how exactly these macroproperties sit outside of determinate laws or predictions based on indeterminate probabilities is needed before it becomes plausible.

In other publications (not mentioned in Part I because they were not strictly part of the dialogue itself) O’Connor relies upon a supervenient relationship to get him the irreducibility his theory requires, as well as the downward causal influence that makes the emergent agent causal power worth considering. For a few reasons I am not able to detail my entire argument here as to why the use of a supervenient relationship does not do O’Connor’s theory any favors, but it should suffice to say that a slight modification of Jaegwon Kim’s exclusion argument allows us to demonstrate that such an emergent property as O’Connor posits would result in overdetermination and thus there is no reason to posit the existence of an emergent property over and above the microproperties, as it then has no novelty to offer.

III. The above (II) is not the only aspect of O’Connor’s theory that begs for more explanation. Presenting the idea that some critics may find his representation ‘thin’, O’Connor says:

Taking the agency theory seriously within a basically materialist framework brings for a whole host of theoretical problems and issues such as the following: When does a physical system qualify as an “agent”?…how do event- and agent-causal processes interact? These, however, are obviously empirical matters, requiring extensive advancements within neurobiological science…The answers to such questions will not be shown by philosophical work in action theory.[5]

In the above quotation I have omitted some of the issues he lists, but the two I have left do not seem to be issues that will easily be resolved by vast scientific advances. The question of when a physical system would qualify as an “agent” seems as though it would be plagued by issues similar to deciding how many grains of sand constitute a pile. I do not see how any advance in science would allow us to point to a specific moment when the macroproperty emerges and constitutes an “agent”. Similarly I fail to see how science could explain the interaction between event-causes and O’Connor’s irreducible agent-causes, especially when he does not give any qualification of how event causes and determined microstructures should or would weigh upon or inform an agent’s causal powers. In a previous post I have detailed how recent advances in neuroscience call into question the hope O’Connor has for future advances affirming rather than damaging his theory.

IV. Finally, the question of reductionism looms large throughout this exchange between Pereboom and O’Connor, and in the end it appears as though they are on opposite sides of the fence. O’Connor does not see any issue with the emergent macroproperty being non reducible down to microstructures and microelements. However, Pereboom does not see any issue with the statement that any law governing or explaining microstructures must necessarily govern or explain the macrostructures that build upon them. Similarly, when O’Connor responds to Pereboom’s book he condemns him for falling in line with the common trend of assuming reductionism.[6] While the entire discussion does not turn completely on this one point, large swaths of it do. O’Connor’s theory requires that the macroproperties be irreducible, otherwise they are determined. Pereboom’s criticism requires that the macroproperties be reducible in order to show that they are determined. I do not see either party giving ground, and so it appears as though this aspect of the discussion must end in a stalemate with no clear victor.

 Knowing why reductionism is important for these theories is different from knowing how it is important. The purpose of reductionism, as I understand it, is to reach the lowest common denominator so what we can form a more parsimonious theory that does not posit more than it needs to. Similarly, as can be inferred from both parts of this post, reductionism usually robs agency theory of freedom and moral responsibility. Is reductive analysis beneficial? Well, biologically it would appear to be so. Reduction allows us to strengthen the case for certain mental states being reducible to chemical imbalances, or brain states more generally.  This then allows us to alter the state of the brain via surgery, medication, or other forms of physical therapy, rather than working from a top-down model such as psychology, etc. I am inclined to lean toward reductive explanations, but the divide between the two positions seems to boil down to a disagreement. I would be interested in hearing from readers what they believe might be a possible advance that could provide more definitive evidence for one position over the other.


In the end, the debate over reductionism is not easily settled, but I do believe that Pereboom has effectively pointed to several weaknesses in O’Connor’s theory. By proposing a purely biologically constituted agent O’Connor successfully avoids the traditional issue of interaction. However, he does create and leave unanswered the question of how exactly a macroproperty can emerge from but not be irreducible to the microsystems that give rise to it. Surely this is an issue surrounding reductionism and supervenience, but intuitively I side with Pereboom that this does not seem coherent with regard to the laws of physics or quantum mechanics as we know them. Ultimately, O’Connor’s theory does not seem nearly as problematic as a dualistic theory of agency but, as my analysis shows, O’Connor’s account could stand to be thickened on multiple levels.

[1] O’Connor, Review, 309.

[2] O’Connor, 258.

[3] O’Connor, 263.

[4] Indeed, O’Connor seems to merely say “why not?” when approached with this question: “But given that there is nothing inconsistent about the emergence of an “ordinary” causal property, having the potential for exercising an irreducible causal influence on the environments in which it is instantiated, it is hard to see just why there could not be a sort of emergent property whose novelty consists in its capacity to enable its possessor directly to effect changes at will (within a narrowly limited range, and in appropriate circumstances).” Pg. 264.

[5] Ibid., 264.

[6] O’Connor, Review, 309.

  1. Aaron
    February 23, 2012 at 2:49 PM


    “Ultimately, O’Connor’s theory does not seem nearly as problematic as a dualistic theory of agency but, as my analysis shows, O’Connor’s account could stand to be thickened on multiple levels.”

    Agreed. Moreover, to compound matters for O’Connor, materialistic reductionism has served as a ridiculously fruitful scientific principle (see, for example, the atomic theory of matter, the kinetic theory of gases, thermodynamics, materials science, and nuclear physics, to name only a few). Thus, it is wholly unreasonable for O’Connor to hold that human brain function will never, on principle, admit of more basic materialistic reductions.

    In fact, we can reason thus: (1) If brain processes are entirely physical processes, it is reasonable to hold that brain function is reducible to more basic physical processes; (2) We have no good reason to believe brain processes are not entirely physical processes; (3) Thus, it is reasonable to hold that brain function is reducible to more basic physical processes.

    Moreover, it seems to me that we can hold (3) even if we never develop the computational ability to so reduce brain processes en bloc.

    Tangentially, we can take our reductionism further and run the mereological nihilist line (as I am wont to do): There are, fundamentally, no composite physical objects, to include human animals. What there are, are various particles arranged ‘human-wise’ at any given time-slice. Thus, there are no humans such that those humans could or could not evidence agent caused free will. Though I think mereological nihilism is the correct view of compositionality, it certainly does not appear to be efficacious in resolving forensic problems related human free will. Having said that, however, neither do I think determinism is efficacious in resolving forensic problems related to human free will.

    Which is why, if you press me, I would call myself a libertarian. Does this make sense?

    • February 25, 2012 at 3:35 PM


      I agree with many of your stances, although I am hesitant to jump to mereological nihilism, only because the libertarianism you derive from such a position is, essentially, meaningless. If I understand you correctly, the absence of a “human” as anything other than a certain set of particles (as reduced as we can get them) at any given time T obviates the question of free will, as the subject of the query does not exist in the same way particles exist, i.e. at its most base state. While I agree with this to a degree, I am not sure how you arrive at a working model of libertarian freedom – for, on that account, the answer of who or what wields such a freedom is moot ad nauseam.

      I am more inclined to say that all objects with which we directly interact exist only as reducible composites. That being said, those objects exist as referents in a sense, given that the removal of certain particles or swathes of sub-composites within the object results in the object no longer fulfilling its functional identity. This is perhaps a clumsy way of getting at what I mean. For example, though water can be reduced to hydrogen and oxygen, neither hydrogen nor oxygen have the qualities of water. It is only when the two join in a nomologically predictable way that what we refer to as water comes to “exist” despite water being reducible in composition (not function) to oxygen and hydrogen molecules.

      So, I would say I favor a functionalist approach to composites in that we can define sets of particles according to the function which we perceive or define as the identity of the object in question, but only when the function of the composite is compromised in an integral way when reduced. On this line we can affirm certain types of emergent properties as functional properties but also maintain the reduction that displays the flaw in O’Connor’s theory. For if a materialistic conception of the brain which does not require non-reducible and emergent macroproperties can explain human behavior just as effectively, there is no reason for us to posit an emergent property over and above the lowest common set of particles which maintains the function we are discussing.

      If I have inappropriately characterized your approach please forgive me. I’d like to know more about why you favor libertarianism given such thorough reduction.

  1. August 25, 2012 at 8:11 PM
  2. September 6, 2012 at 7:39 PM

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: