CSE 275: Social Aspects of Technology and Science

3. Cause and Effect

In physics, there are no cause/effect laws. For example, Newton's third law,
      F = ma,
is a mathematical relationship between measurable quantities: it says that force, mass and acceleration are related in a certain regular way; it does not say that acceleration causes force, or that force causes acceleration. It allows either possibility, and it even allows mass to be used to alter force, as when an aircraft jetisons fuel before a dangerous landing. All of the equations of physical science and engineering have this same acausal character. Another example is the so called "general gas law," and equation that relates the temperature, pressume and volume of a gas,
      PV = nRT ,
where n and R are constants.

Nevertheless, physicists, chemists and engineers do use the language of cause and effect, for example, in an experiment where a magnetic force is applied to a metal ball in a vacuum, the experimenter thinks of the force as the cause and the acceleration as the effect. All cause/effect language arises through asymmetries that are introduced by an observer in a similar way. They are not part of nature, they are part of human culture; they are social. In fact, natural language does not provide good support for talking about the kind of acausal laws that are the actual content of physics.

It is often said that good science is reductionist, and the success of the hard sciences is often given as an example that the social sciences should try to follow. A prime example is the reduction of chemistry to physics. However, if we look carefully at what really happens in chemistry, we will see that chemists are not doing a specialized kind of physics. On the contrary, they are using concepts at the level of chemistry, such as valance. It is impossible in practice to solve Schroedinger's equation for any but the very simplest atoms, so that calculations in quantum physics cannot be used to do chemistry.

This situation is often described by saying that chemistry is an emergent level above physics, meaning that partial reductions are possible and can be very valuable when they occur, but concepts and theories that are distinctly chemical (and not physical as such) are regularly used, and in fact are primary for the practical applications. This does not deny that reduction might be possible in principle, as most scientists believe in this case.

If we look at higher levels, such as biology, psychology, and sociology, we again see emergent phenomena, but it becomes progressively more difficult to support the belief that reduction to lower levels must be possible in principle, and indeed most social scientists today do not believe this.

So where does this leave us? It seems that simple cause/effect explanations are not characteristic of the hard sciences, and moreover reductionism does not take the simple form that is often claimed. So we conclude that arguments in favor of technological determinism based on a claim that it is in some sense more scientific than alternatives are fatally flawed. Going a little further, I think we should conclude that it is very wise to be suspicious of simplistic principles and simplistic arguments in complex areas like the relationship between technology and society. Technological determinism is a prime example of such a simplistic principle.

But then, Why, given the deficiencies of technological determinism, do people find it so persuasive? Why is it so common in advertisements, newspaper and magazine articles, websites, and other places? One answer is that causal explanations are built into our language. For example, the sentence

John hit the ball.
has an actor and an action, which are its subject and its verb, respectively. Readers want to understand this sentence, not in isolation, but as a part of a story, which might be about baseball, where the actor has an intention to perform the action, because of its consequences. That is, readers want to find a cause, e.g., John swinging his bat, because of its effect, which might be a home run. Hence the effort to understand a sentence is an effort to find such cause/effect relations, in order to relate events with human intentions.

Linguistics has developed extensive theories of stories, which can add many interesting details to this discussion. The narrative presupposition applies to any story, but especially to oral narratives of personal experience, and says that the order of clauses is the same as the order of the events that they describe unless there are explicit contrary indications (this term was introduced by William Labov). For example, if we hear

John hit the ball. The crowd cheered.
then we will assume that the crowd cheered after John hit the ball. And we will further assume that the crowd cheered because John hit the ball. This is an example of what I call the causal presupposition, which says that if possible, we should read the second event as caused by the first event. Note that the causal presupposition assumes the narrative presupposition. Additional evidence for these presupposition comes from studies of the Balinese language, in which the narrative presupposition is replaced by the default presupposition that, given clauses A, B in that order, the corresponding events occur concurrently, possibly with mutual interaction (see papers by Alton Becker). In computer science terms, we might say that in English, the default semantic connection between subsequent clauses is ";" rather than "||" , whereas the opposite holds in Balinese.

Much more information on the theory of narrative can be found on the narratology page of Mark Parham, and through the links on the narratology page at the Media and Communication Studies site at the University of Aberdeen, for example, the paper The Narrative Emergence of Identity by Mark Gover.

An example where we can clearly see the interplay of an underlying acausal model with human causal explanation is a simple ecological system, with one predator species and one prey species, say wolves and rabbits. The basic Volterra-Lotka differential equation is well known, and has as a solution (given suitable coefficients and initial values) two periodic functions with a time lag; that is, the numbers of wolves and rabbits fluctuate up and down over some fixed time period, as illustrated by the applet below. (Of course, most real ecological systems are much more complex than this, but simple cases where the assumptions of this model are satisfied have actually been observed.)

For more on Volterra-Lotka models of predator-prey systems, see the Lascaux Graphics webpages, and for much more background information on ecological systems and their models, see the Population Ecology Reference List (by Alexei Sharov, of the Dept. Entomology at Virginia Tech), and especially the Volterra-Lotka simulation server linked there. The Voltera-Lotka equations for a simple predator-prey system have been given as follows:

[{ roman d x ( t ) } over { roman d t } = r x ( t ) left ( 1 - { x ( t ) }
  over  K right ) - { a x ( t ) y ( t ) } over { 1 + a t sub h x ( t ) }]

[{ roman d y ( t ) } over { roman d t } = { a c x ( t ) y ( t ) } over
  { 1 + a t sub h x ( t ) } - e y ( t )]

It is difficult to understand such a system just by contemplating these equations! In fact, our intuition is much better served by causal assertions, such as "a large number of wolves will decrease the number of rabbits" and "a small number of rabbits will decrease the wolf population". It is not just beginners who find such assertions helpful; even experts often use this kind of causal language informally among themselves. The human mind did not develop under evolutionary pressure to deal with differential equations, whereas there certainly was evelutionary pressure to deal with simple cause/effect relations.

So we should not conclude that causal assertions cannot be used at all, but rather we should be aware of their limitations. Causal assertions are normal and useful for talking about entities that have intentions. However, they are often misleading in talk about non-intentional entities. Moreover, they are open to deliberate misuse, e.g., in advertisements for new technical devices. (Applying concepts to non-humans that are only appropriate for humans is called anthropomorphism.)

Another example is the relationship between two stock market variables: the Dow-Jones Industrial Average and investor confidence. It is easy to find articles which say things like "Decreases in the Dow have eroded investor confidence" or conversely that "Decreased investor confidence is eroding the Dow." But it is hard, maybe impossible, to find articles which say that these two variables are mutually interdependent, reflecting a complex system where simple cause/effect assertions can be misleading.

So the conclusion here is that if we look more carefully into real social systems, we will see that, insofar as there is lawfulness, the laws tend to be like the laws of physical science, that is, acausal relationships among variables, rather than assertions of cause/effect relationships. On the other hand, cause/effect relationships are truly useful in understanding the actions of people, as well as corporations, governments, etc., because such entities do have (or can be said to have) intentions - that is, goals - and they do carry out actions in order to achieve their goals. This is in sharp contrast to systems, like the stock market and the global telecommunications system, which do not themselves have goals, and do not intentionally perform actions, but for practical purposes, can be successfully described as satisfying various laws.

In fact, what is strikingly clear for complex systems such as eco-systems (and is almost definitional for complex systems) is that there are complex mutual interdependencies (e.g., see the right sides of the Volterra-Lotka equations for the derivatives of the two variables). This is difficult to translate into our ordinary language of cause/effect relationships, because it really denies the essence of such relationships. However, phrases like "mutual causation" and "interdependent orgination" have been used to describe such systems, the latter going back over 2,500 years to the Buddha (the orignal Sanskrit word for this is "pratityasamutpada").

Another point that the study of complex systems drives home is that anything that we call a system is an abstraction, emphasizing some particular things and ignoring others. Since in the real world, everything is interdependent, it is necessary to draw a line somewhere, and call what is inside the line "the sytem" and what is outside of it "the environment." Sometimes it is pretty clear where to draw such a line, but other times, especially for complex systems with a significant social component, it is much less clear, and different choices can lead to very different analyses and results. For example, the optimizations done by large corporations tend to ignore many variables of social importance, such as the quality of air, diversity of the biosphere, and the depletion of non-renewable resources.

To section 4 of CSE 275 notes.
To section 2 of CSE 275 notes.
To CSE 275 homepage
Maintained by Joseph Goguen
Last modified 15 October 1999