Determinism without reductionism

I am a determinist, but not a reductionist. In other words, I think the universe in general, and human behavior in particular, follows analyzable rules of cause and effect, but the universe cannot be “reduced” to the simple interaction between physical particles. Since most people who are determinists are also reductionists, I’d like to take a moment to explain why I am not.

The Billiard Ball Universe

Let me start by describing a model of the universe that is often invoked in these types of conversations, which I will describe as the “billiard ball universe.” This is the view that most people associate with both determinism and reductionism. That is to say, when most people think of either “determinism” or “reductionism”, the billiard ball universe is the thing that comes to mind.

The “billiard ball universe” goes something like this: all social phenomena can be deconstructed into interactions between people who act according to psychological principles; all psychology can be deconstructed into behaviors emerging from neurological principles; all neurological phenomena can be deconstructed into chemical principles; all chemical phenomena can be deconstructed into principles of molecular and atomic physics; and everyone knows that the laws that govern the interactions between atoms are purely mechanical. Even if these interactions involve some kind of randomness (a la quantum mechanics), they definitely do not involve choice. Therefore, the entire universe is nothing more than a large collection of atoms interacting mechanically according to predefined laws.

According to the billiard ball universe, all descriptions of the universe at a “higher level” (i.e. chemistry, neurology, psychology and sociology) are just linguistic games and illusions.  In other words, when I say “I was talking about philosophy with my friend,” that’s really just a kind of short-hand for talking about a way that large collections of atoms (atoms in our brains, atoms in our bodies, and atoms in the air between us) happen to be bouncing around according to basic physical principles.

This view started becoming very popular in the philosophy of science in the early and mid 1900’s. Naturally, many social scientists took personal offense to this, because this view is basically saying that nothing beyond particle physics is worth studying.  According to this view, real causation is only between particles. Any phenomenon described in any other way, such as a ball causing a splash in a pond or a person causing his lover to cry, is purely epiphenomenal.  That means it’s more or less a side-effect, an “artifact” or a by-product of the “real” and important stuff that is going on, where “the important stuff” means interaction between atoms.

In my opinion, the “billiard ball universe” idea is a load of crap.

It still has both its strong advocates and its strong detractors, however. I vividly remember some of the reactions that review boards had to my early research, with sociologist Dr. Thomas Smith. We proposed a very simple model of interpersonal interaction: social separation triggers the production of epinephrine, which in turn suppresses opioid production and increases arousal; social attachment, on the other hand, triggers the production of opioids, which suppresses epinephrine and decreases arousal. As a result, all social interaction can be viewed as a form of biochemical and emotional regulation: in other words, people can use social interaction to regulate their feelings, and the need for emotional regulation can drive social interaction. This model was able to predict all kinds of social-level phenomena, including reciprocity, oscillation dynamics, assortative clustering and grouping, and other sociological things that I don’t really intend to go into detail about here.

Attachment HyperstructureThe key thing about this model was this: it included variables that represented neurochemical factors (e.g. levels of opioids and epinephrine), variables that represented psychological experiences (e.g. boredom and anxiety), and variables that represented social behaviors (e.g. attachment and separation). In our model, all of these were able to influence, or trigger, other variables in the system.

And people freaked the hell out. We were able to get a number of papers published on this theory, in very respectable journals such as Sociological Theory and the American Sociological Review; but it was a grueling process, in part because of some of the strong philosophical resistance that we met in opposition to the type of theorizing that we were doing.

During the peer reviews for these papers, there were some people who accused us of “reductionism” because we proposed that social behaviors can be driven by neurochemical levels. People who were reductionists, on the other hand, accused us of some kind of weird mysticism because we claimed that social behaviors had actual causal influence over the brain’s chemical levels. Depending on people’s biases, we were either giving too much causal “credit” to neurochemistry, or too much causal “credit” to social behavior.

In actuality, both of those criticisms were wrong. They were both based on either accepting or being scared of the “billiard ball universe” view. Which, as I said, is a load of crap.

What does it mean to “do science”?

What does science actually do? What does it mean to “do science”? This is a very important and fundamental thing, so it’s worth taking a minute to look at it. It’s going to seem like a side-track at first, but you will see why it matters to this conversation about determinism and reductionism in just a moment.

At the most basic level, “doing science” is a process of finding patterns in the world and trying to come up with the best explanation for them. When people talk about the “scientific method” they are talking about the rules they use to figure out which explanation is “best”. If you have some observations you’ve made in the past, and you have two possible explanations for them, then one way to decide which one is “best” is to see whether they make different predictions for new observations, and then go to see which one is right.

Sometimes, these “explanations” can be as simple as suggesting a causal relationship. Sometimes it can be as complex as proposing new conceptual entities to explain an elaborate network of relationships.

Suppose you see that kids who play violent video games get into trouble more at school. That’s a pattern: a relationship between two variables. You might suggest that playing violent video games makes children more inclined to get in trouble.  But, there is another possible explanation: children with unruly temperaments might be more inclined to be drawn to violent video games to begin with, as well as being more inclined to get into trouble in school. So, in order to do science, you have to come up with some situation, some test, where these two theories make different predictions.  For example: you could try to find children who get in trouble at school but have not been exposed to violent video games, and give them a choice between violent and non-violent games. You check to see if they tend to choose the violent ones more often than the children who did not get into trouble at school; if they do, then that would support the “unruly temperament” view, and if they do not, then that would support the “video games lead to kids getting in trouble” view.

By creating an experiment like that, you are “doing science”, because you are going through a process of comparing two possible explanations to decide which one is best.

In this little toy “theory” I’ve been describing, we have two observed variables (playing video games, getting in trouble) and one new concept that was proposed as an explanatory variable: having an “unruly temperament”. If you’re a visual person, you can imagine a kind of diagram where each variable is a circle or a box, and there are arrows between them representing the relationships (according to the theory) between them.  As you can imagine, with big and complicated theories these relationships can get quite involved, with lots and lots of variables and arrows between them.

At its core, this is all that science does.  It talks about relationships among variables to try to explain patterns of observations.

To state it as simply as possible: Science is coming up with rules to explain regularities.

This means that the process of “doing science” is an intrinsically deterministic enterprise.  The whole point of science is to talk about what variables can predict other variables. The whole point of science is to talk about the “rules” of cause and effect.

What about “free will”?

For most people, the simplest expression of the idea of free will is in the idea of choice: My future actions are not predestined, because I can choose what actions to take, and even though I may in fact choose to perform action X, there is some sense in which I could have chosen otherwise.

Of course, it’s perfectly possible to “do science” with people’s choices and behaviors. Think about what that means: science will look at patterns that show up in different variables that influence people’s actions, and try to describe rules to explain those patterns.   Remember our slogan for what science is: coming up with rules to explain regularities.

So the entire endeavor of “doing science” (when applied to people’s behavior) to look for the rules that determine behavior. The idea that you “could have done otherwise” in some alternate universe or other is…. well, it’s not exactly contradictory to the very essence of the scientific process, but it’s simply not the focus.  The intent of science is to find the regularities.

Imagine a game of “rock, paper, scissors”.  (I’m going to assume you know the game.)  Suppose I came up with a theory that was actually able to perfectly predict whether you chose rock, paper, or scissors every single time.  You might come back and say: “That’s really amazing, but I still feel like I could have chosen something different, if I had wanted to!”  Think, for a moment, about whether you find that “feeling” convincing.

The only reply science can have to that is: “But  you didn’t  want to… and we have a scientific theory that predicted that!” The fact that you feel like you could have done otherwise if you had made a different decision is almost comically irrelevant, if there is a scientific theory that can explain why you “chose” to make the decision that you made.

That’s why I’m a determinist: because I believe that there are patterns to people’s behavior, and that science can be used to analyze and predict those patterns.

And unless you say that you are somehow not allowed to use science to look at human behavior, then you can have theories that explain what variables influence the choices you make. That’s it, as far as free will is concerned: game over. If you can have a theory that predicts patterns in the things that you “choose” to do, then your will isn’t “free.”  You’re a determinist.

What does this have to do with billiard ball universe?

It’s descriptions, all the way down

The reason that I wanted to talk you through the basics of “doing science” is to point out this fact: the building blocks in science aren’t things, they are concepts (or “variables”).  In the box-and-arrows pictures that I was describing, above, the boxes can represent feelings or actions or measurements or observations of any sort. They can represent something that you can only theorize about and observe indirectly (like an “unruly temperament” of a child). But the reason they are “basic building blocks” in a theory isn’t because they are physical stuff: it is because they represent concepts that we are trying to understand the relationships between.

Of course, when sociologists talk about the relationships between people, or chemists talk about the relationships between molecules, we are used to thinking that they are talking about objective, physical stuff in the world…. but in reality, they are not.  This is important, and I know most of you are not used to thinking about it this way.  But it is critical: when a chemist talks about a sugar molecule, he is talking about a conceptual variable.  It is, in an ontological sense, no more and no less real than when a psychologist talks about anxiety.  There is a concept, and that concept is embedded in a theory that talks about what things affect it, what things it affects, and how those things can be measured and observed and tested.  But whether you are talking about atoms or attitudes, the process of doing science is the same: you have concept X and you have theories about the relationships between it and concepts Y and Z, and you try to measure those relationships.

This is why physics is no more “primary” than sociology. This is why reductionism is nonsense.  Because whether you talk about your observations of the universe using the descriptive language of atoms and quantum fields, or you talk about your observations of the universe using the language of motivations, fears and desires, you are still talking about a description of the universe.  A description that is made up of concepts devised by the human mind, concepts that are the “variables” in a scientific theory. Nothing is more “primary” about the concept of an atom just because it is a concept of something that is very small, or something that is easier to predict than your neighbors’ moods.

The proper scientific view of the universe is as a collection of super-imposed networks of descriptions: descriptions of concepts that can be used to explain other concepts connected through a series of relationships.  Science is about finding those relationships.  But in the end, a little circle on a diagram labelled “atom” and a little circle on a diagram labelled “caregiver” have exactly the same ontological validity: they are stories that we are coming up with to describe a piece of the world we live in.

You can break down descriptions of a thing into smaller pieces by giving descriptions of their parts, and descriptions of their relationships.  You can do that again and again.  But you will never get to the bottom.

Science is descriptions, all the way down.