Every physical body in the universe is subject to inertia, and every living organism consumes energy, now and always. In sum, people are exquisitely sensitive to the invariance of causal relations, the relations that govern how things work. Notice it is not the working mechanisms themselves that are invariant but the principles that govern them.
A car engine is a causal mechanism designed to be reliable and in that sense invariant, but—as we all know—that is a pipedream.
Causal Models How People Think about the World and Its Alternatives by Steven Sloman Information
Shadows are cast by objects that block a light source like the sun, but the precise form of the shadow-casting mechanism may change continuously as the earth rotates and clouds shift. The invariant is not the form that the mechanism takes but rather the principles that govern its operation.
The drive shaft of a car is not invariant, but the principle relating torque to force is. And the sun-cloud mechanism that produces shadows is changing all the time, but the principles that govern how light and objects produce shadows are constant. Causal principles that govern how events affect other events are the carriers of information, and it is those principles, not the mechanisms that they govern, that persist across time and space. They are the most reliable bases for judgment and action we have.
Metaphysically, this may indeed be correct. Perhaps the world is nothing but a flow of energy. Perhaps there is no will, no agents, no intentions, no interventions, just the transformation of energy following certain eternal mathematical rules. Perhaps there is no root cause of anything, and perhaps there is no final effect either. Perhaps everything we misconstrue as cause and effect is just energy flow directed by mathematical relations that have determined the course of history and will determine our destiny in a long chain of events linked by the structure of energy in time and space.
The central idea of the book is that the invariant that guides human reasoning and learning about events is causal structure.
Privileged (Default) Causal Cognition: A Mathematical Analysis
Causal relations hold across space, time, and individuals; therefore, the logic of causality is the best guide to prediction, explanation, and action. And not only is it the best guide around; it is the guide that people use. People are designed to learn and to reason with causal models. Causes and Effects Are Events In everyday language, causes and effects assume various roles. We say drugs cause addiction, sparks cause fire, and guns cause death or at least bullets do.
In each case, an object drugs, sparks, guns is the cause of a state addiction, death or an event fire. Physical and emotional states can cause other states, as fear causes loathing, hunger causes suffering, or satisfaction causes tranquility. Events can cause other events, as one war causes another, or a strong wind causes a tree to fall. The reason such cases are hard to think of is that we conceive of objects in a static way, as if they are fixed over an extended period of time, yet causal relations are enacted over time. A causal relation suggests a mechanism unfolding over time that uses the cause and possibly other things to produce the effect and possibly other things.
One general temporal constraint on causation is that effects cannot precede their causes. A dictionary might respond that a mechanism is a process by which something is done or comes into being. So causal relations relate entities that exist in and therefore are bounded in time. On closer inspection, all the examples I noted have this character.
Sometimes we talk in a non-temporally bounded way as if abstract properties can be causes and effects increases in pressure cause increases in temperature; love causes beauty. But even here, the actual causes and effects as they manifest themselves in the world are physical entities that obtain for periods of time. This is what distinguishes causal relations from definitions. A definition identifies a word or phrase with a set of conditions in the world.
If all the conditions are met, then the word or phrase applies i. Likewise, if the word or phrase applies, then the conditions are met the conditions are necessary. If an object is geometric and has three sides, then it is a triangle. And, if it is a triangle, then it is geometric and has three sides.
- Causal Models How People Think About the World and Its Alternatives by Steven a Sloman - AbeBooks.
- Wretched Saviors.
- Also Available As:.
- Onondagas centennial. Gleanings of a century (Volume 1)!
- Subscriber Login!
- Eternia (Italian Edition).
They associate events with other events. A correlational study What Is a Cause? Two attributes or variables are correlated if a value on one tends to be associated with a value on the other. But the mantra of experimental psychology is that correlation is not causation. The fact that these things are correlated does not tell us why they are correlated. It does not tell us what generating mechanisms produced these correlations. To find out, we have to run an experiment. Some variable, some potential cause often called an independent variable , is chosen along with another variable, a potential effect often called a dependent variable.
The cause is then manipulated by setting it to two or more values and the effect is measured. If the value of the effect differs for different values of the cause, then we can infer that the cause has some influence on the effect; a causal relation exists. To illustrate, say you want to know if punishing children makes them more obedient.
Then you need to vary punishment, the independent variable, by punishing some children more and other children less and measure their level of obedience, the dependent variable. If the children punished more are more obedient than the children punished less, and if the difference is big enough, you can conclude that punishment increases obedience. And if the children punished less are more obedient, then you have some explaining to do.
Much of the art of experimental methodology involves satisfying two requirements. First, the cause must be manipulated so that only the target cause is manipulated and not other potential causes accidentally. When manipulating punishment, you have to be careful not to also manipulate how much warmth you show the child. Second, the methodology must be sufficiently powerful to 24 The Theory allow a reasonable assessment whether a difference in the value of the dependent variable is real or just the reflection of random variability.
For now, the important point is that conclusions are drawn from experiments through comparison; the value of an effect must be compared in two different worlds: the world in which the cause is at one value and one in which the cause is at a different value, because a causal statement is a claim about two or more worlds simultaneously. Causal Relations Imply Certain Counterfactuals To say that A caused B seems to mean something like the following: A and B both occurred, but if event A had not occurred and B had no other sufficient cause , B would not have occurred either.
So in some other world in which the mechanism had not been engaged by the cause, the effect would not have resulted. This is what distinguishes a causal relation from a mere correlation. A correlation between two event types means that they move together; when one happens, the other tends to happen, too.
A causal relation has the further requirement of counterfactual dependence. All of these open thought or discussion about some other world, different from this one.
This is a counterfactual because in fact the cause did occur and so did the effect or, at least, they may well have. Correlations make no such counterfactual claim. Correlations are just about what goes together in the actual world. An association between two events is normally assumed to be formed by an organism who perceives that the events are correlated.
In sum, causal relations are more than descriptions of the way things are, the state of the world. They are processes that generate possibilities, whether those possibilities exist—whether they are actual—or counterfactual. So the first or second law of experimental psychology holds: correlation is not causation.
The great Scottish philosopher David Hume taught that causation cannot even be inferred from correlation. No matter how many times A and B occur together, mere co-occurrence cannot reveal whether A causes B, or B causes A, or something else causes both. But one of the million dollar questions today actually, research funding agencies have spent well over a million dollars on it is whether Hume was right.
Some recent theories of causality specify how causal inferences can be drawn from correlational data in certain cases. This kind of unjustified causal attribution is all around us. In this kind of case, very different causal attributions may have just as much support: the forces that led to a rise in gas prices 26 The Theory may also have created the conditions for the new administration to be elected.
Take your birth, for example. But such distinctions can become fine legal points. Is birth control a disabler or a true cause of not giving birth?
The event might well have been different counterfactually if any of those parental events had been different. Who knows how you would have turned out if gestation had been different or if your mother had enjoyed a different diet?
But at the most basic level, where we just want to know about the mechanisms that make events happen, it is very useful to divide the world up simply into causes and effects. Sometimes an event has alternative causes, each itself sufficient for the effect. Death can be caused by trauma, uncontrolled growth, infection, and so on.
In rare cases, more than one can conspire to cause death together. The arrow-drawings or graphs like figure 3.