The world is not a Rube Goldberg: The complex causal chain argument

Rube Goldberg, a cartoonist, engineer, and inventor, gave us cartoons of many machines that illustrate the complexity of cause/effect events in the real world yet are still simple. His machines and cartoons, known as Rube Goldbergs, often contain complicated steps that a machine will go through to do something very simple. An example of this is the “self-operating napkin” (Goldberg):
[pic][pic]

In this Rube Goldberg, one can see that as he lifts the spoon to eat (A), it pulls the cord (B) that flicks the spoon (C). The spoon tosses the bread (D) that the bird (E) goes for and so the stuff on the other side of the balance (F) falls into the bucket (G). As the bucket drops (H), it pulls the string (I) that opens the box that lets out the fire (J). The fire ignites the fire-cracker (K), which pulls the scythe (L). The scythe cuts the cord (M), which lets the napkin move according to the ticking of the clock and wipes the man’s mouth.

The plethora of events and cause/effect connections in this description seem complex compared to the simple act of picking up the napkin oneself and doing it. As complex as these events seem, they are still simple compared to any real set of events in the world. In this Rube Goldberg, the causal chain is that A causes B which causes C and so on. This is a linear chain of events where A and only A causes B which only causes C and so on. Even though the explanation of what happens in this Rube Goldberg is complex, the reality of the chain of causal events is very simple. For purposes of understanding this simple event, these linear causal maps are useful. However, and this is a key point, they have limited functions otherwise.

These linear causal maps are useful for understanding cause and effect relations that are general.
An example of this is models of memory. A simple linear map would include the encoding of the material, the storage/consolidation of the material, and the retrieval of the material. This would also include certain conditions that influence the process, such as encoding/retrieval environment, type of memory, and so forth. However, any student of memory would realize that this is an oversimplification of all of the things that occur. There are many other factors to consider: for encoding: the temperature in the experimental environment, the subject's parallel processing, facts about attention, number of times encountered, time of day, amount of other material encountered, place where material is encountered, and so forth; for storage/consolidation: facts about the subject's sleep amount/cycle, activated brain areas, type of material, similarity to other stored material, experience with the material and type, previous retrieval and effort, and so forth; and for retrieval: similarity of encoding and retrieval environment, time from encoding, and so forth (Schacter, 1997). Here one can see that the linear simple representation of events can help one to understand the memory process in general.
However, when it comes to analyzing the encoding, storage, and retrieval of one specific memory, it becomes much more complicated. For one specific memory, in order to understand how and why it arises in the way that it does, one needs to take many factors into account. Even then it is not possible to take every single factor into account for a given event. This suggests a potential in-principle issue about the account of causes and effects.

Here, I have drawn a distinction between the general and the specific event. While causal maps for the general event can suffice, they do not suffice as an explanation for any one specific event.
Linear causal maps are how we create general rules about events (studied this or that way); explanations of specific events often involve exceptions to these rules. This leads to a very important point.

Because we constantly attempt to make causal models of how events are happening in the real world, we make two important assumptions. The first is that there is some form of causal chain present in the world and the second is that the world is full of complex and interrelated causal events. Any causal model that is made focuses on certain factors in order to serve a certain purpose. An example of this is the simple causal map of the memory process. It serves to represent the general steps in all memory experiences. These models have to focus on certain factors not only because they serve some purpose but also because there are an excessive number of factors. If, though, one tries to take all the factors into account concerning one memory event, they would have to riffle through hundreds of psychological articles for years. After having done that, they would have some idea of all the known factors but would still not understand how exactly these factors affect memory nor how these factors interact. In this case, once again, one has a general idea of the memory process. Even though they would predict better how and what someone will remember, they will still find many exceptions and would be unable to consider all of the causes/effects. This shows that if we assume any form of causal chain is present in the world, then we also concede that any realistic conception of how any event arises is incredibly complex with a countless number of variables which influence the arising of that event.
This means that any causal conception that we make of how a specific event came to be is necessarily insufficient. It’s for this reason that we can never use causal chain maps to determine what will happen in any specific event.

When anybody picks out a certain number of factors influencing an event, the only thing that he can do is predict what the result will be – he cannot determine what exactly will happen except in highly controlled experimental conditions. That is what these maps are useful for: predicting, not determining. This is the way science uses causal chain maps. The purpose is not to determine how any one event will turn out but to better predict how they will turn out based on the previous factors. Error, for example, is an important part of the statistics of science. When a researcher is analyzing any data, a certain amount of error must be taken into consideration (Gravetter 2006). This is known as experimental error. Experimental error results from the fact that one cannot test things or people that present the exact same set of variables. In psychology, it also reflects the fact that different people will perform differently (e.g., remember certain words better no matter where they are on a list). Another very important cause of experimental error are factors that the researcher is not manipulating or controlling (and of which she might be unaware). The sciences are therefore not concerned with determining outcomes nor can they claim to be able to. Once again, the conceptualization of events in the real world is insufficient for determining, but sufficient for predicting, events.

The world is, therefore, not a Rube Goldberg or a linear causal chain. Although it can be represented by causal chain maps, the complexity of the causes leading to events that happen in the world is not properly taken into account in these examples. There are better ways to understand the way the world actually is.

If one takes the causal maps that I have presented above (where A causes B and so forth), there is one thing that is not taken into account. For one, events are real things in time and space. This means that in mapping events as A, B, C is a misrepresentation of what is going on in the world. By taking space into account, one sees that the linear representation is too simple. Looking at someone crossing the street, for example, is more complex than initially thought. Taking into consideration that attention to the external world is mainly a function of vision and specifically the fovea (which takes in information from a section of the visual field the size of a thumbnail), and that it is modulated/guided by the other senses, we can understand this action as even more complex (Zilmer, Spiers, & Culberston, 2007). Imagine, for example, that Tommy walks to the sidewalk, is about to cross the street, and looks both ways about to step down: what will happen? He’s listening to an MP3 player and then a mosquito flies by. This mosquito bites him on the arm and he turns his head towards it and swats it with his hands. In the opposite direction of where he’s now looking, a car is barreling down the road and doesn’t notice that Tommy is about to step down. At this point, the result might be that Tommy gets hit by the car or that he stops in time to be saved; both would be predictions and are beside the point. The point instead is this: in this example, a simple linear representation is not possible. How would one, in a linear causal representations, take into account the attentional system and its role in this one event? Again, the representation would fall apart and would serve as no more than a factor in predicting what will happen. Also, with any causal map of a specific event, one has to take space into account. Where did X happen? Where was Y looking? Any event explanation becomes more of a story and the map more of a picture book. Even these, though, would not include everything. It is clear, then, that the mapping of one event is extremely complex.

To make this conception of the world clearer, I am going to use a metaphor that is often used by philosophers when it comes to causation: billiard balls. In this metaphor, causation is represented by one billiard ball hitting another which causes the second one to move. It sums up causation in a simple and poetic way. Simplicity, though, is a misrepresentation of how real events occur. If one has a moving billiard ball of exactly X weight moving at exactly Y speed hitting a stationary billiard ball of exactly Z weight in a vacuum, then the impact point, movement, direction, and speed of the stationary billiard ball can be determined. The more factors that one adds to this example, the less determining and more predictive these factors become. If these billiard balls were in a perfectly flat and level table on Earth, then the resistance of the air, distance between the balls, and resistance from rolling on the table would need to be accounted for. If you add more balls, then they would also need to be taken into account as well as all these factors for them. If all of these balls were moving and they were in a field that had valleys and hills, it becomes even more complex. Add in obstacles like trees, rocks, and grass. Make the different parts of the ground into different consistencies like those of mud, gravel, and just dirt. Add in different people of different ages, strengths, weights, and so on all over the field who are kicking, throwing, and pushing the different balls around. To understand what is happening in this field, one needs to know those initial factors. In order to determine how one ball will be affected by another ball that is going towards it, one would need to know a whole plethora of factors and how they relate.

If one focuses on the effect of billiard ball A being thrown, it would be a lot more than just hitting billiard ball B and causing it to move. Billiard ball A could create a dent in the ground and after it is hit by something else this dent becomes an obstacle for another billiard ball. Billiard ball B could hit both billiard ball C and D and so forth. From this it is obvious that things can happen as a result of the initial event that are not immediately relevant and other things can be happening at the same time that become immediately relevant. In this way, such a causal chain would be multi-causal, time-distributed and parallel with many other linear chains. By 'multi-causal,' I simply mean that the events can have more than one cause, though they don’t all have to be important to the specific causal chain under investigation. By 'time-distributed,' I mean that some of the causes of the initial event do not have immediate effects, although they have effects later on. By parallel, I mean that there are other causal chains occurring at the same time as the one being focused on, and these might become important at some point along the way. An example of this is Julie walking through the woods. If Julie chooses to go around a tree a certain way because it is easier, this is not a monumental event that has some immediate cause. But, if she is one of 300 people who chose to walk around that same tree that same way, then a path is created. This shows how the decision to go around the tree had one immediate effect of making the walk easier and a later cause of contributing to the path. Also, both the causal chain concerning how that path was formed and the causal chain of how Julie got from point A to point B run in parallel for at least a little while.

It is therefore clear that the causal events of the world are extremely intricate and complex.
Although models of these events can be useful, they cannot be considered an accurate representation of the real complex causal chain working itself out in any actual event. These representations tend to fall apart and exceptions consistently arise when used to analyze specific events. It is because this happens that we can conclude that the world is not a Rube Goldberg.

Accepting the complex causal chain argument has important consequences. One of the most important is that when someone analyses behavior, there will always be a space left over which will remain uncertain and indeterminable. There are arguments made by both sides of this issue that deal with this space in unsatisfactory manners. On the one hand, the libertarians attempt to claim this space for free will; on the other, determinists attempt to eliminate the space altogether. Both sides, I will argue, make unsatisfactory arguments for their claims, leaving this space open.