Causality: Difference between revisions
→Necessary and sufficient causes: Inserted ' |
→Probabilistic causation: Corrected simple typos, added commas |
||
Line 71: | Line 71: | ||
==Probabilistic causation== |
==Probabilistic causation== |
||
Interpreting causation as a [[Causal determinism|deterministic]] relation means that if ''A'' causes ''B'', then ''A'' must ''always'' be followed by ''B''. In this sense,war does not cause deaths, nor does [[smoking]] cause [[cancer]]. As a result, many turn to a notion of probabilistic causation. Informally, ''A'' probabilistically causes ''B'' [[iff]] ''A'''s occurance increases the probability of ''B''. This is sometimes interpreted to reflect imperfect knowledge of a deterministic system but other times interpreted to mean that the causal system under study has an inherently chancy nature. |
Interpreting causation as a [[Causal determinism|deterministic]] relation means that if ''A'' causes ''B'', then ''A'' must ''always'' be followed by ''B''. In this sense, war does not cause deaths, nor does [[smoking]] cause [[cancer]]. As a result, many turn to a notion of probabilistic causation. Informally, ''A'' probabilistically causes ''B'' [[iff]] ''A'''s occurance increases the probability of ''B''. This is sometimes interpreted to reflect imperfect knowledge of a deterministic system but other times interpreted to mean that the causal system under study has an inherently chancy nature. |
||
The establishing of cause and effect, even with this relaxed reading, is notoriously difficult, expressed by the widely accepted statement "[[Correlation implies causation (logical fallacy)|correlation does not imply causation]]". For instance, the observation that smokers have a dramatically increased lung cancer rate does not establish that smoking must be a ''cause'' of that increased cancer rate: maybe there exists a certain genetic defect which both causes cancer and a yearning for nicotine. |
The establishing of cause and effect, even with this relaxed reading, is notoriously difficult, expressed by the widely accepted statement "[[Correlation implies causation (logical fallacy)|correlation does not imply causation]]". For instance, the observation that smokers have a dramatically increased lung cancer rate does not establish that smoking must be a ''cause'' of that increased cancer rate: maybe there exists a certain genetic defect which both causes cancer and a yearning for nicotine. |
||
Line 79: | Line 79: | ||
That said, under certain [[assumption]]s, parts of the causal structure among several variables ''can'' be learned from full [[covariance]] or [[case data]] by the techniques of [[path analysis]] and more generally, [[Bayesian network]]s. Generally these [[inference algorithm]]s search through the ''many'' possible causal structures among the [[variable]]s, and remove ones which are strongly incompatible with the observed [[correlation]]s. In general this leaves a set of possible causal relations, which should then be tested by designing appropriate [[experiment]]s. If experimental data is already available, the [[algorithm]]s can take advantage of that as well. In contrast with Bayesian Networks, path analysis and its generalization, structural equation modeling, serve better to estimate a known causal effect or test a causal model than to generate causal hypotheses. |
That said, under certain [[assumption]]s, parts of the causal structure among several variables ''can'' be learned from full [[covariance]] or [[case data]] by the techniques of [[path analysis]] and more generally, [[Bayesian network]]s. Generally these [[inference algorithm]]s search through the ''many'' possible causal structures among the [[variable]]s, and remove ones which are strongly incompatible with the observed [[correlation]]s. In general this leaves a set of possible causal relations, which should then be tested by designing appropriate [[experiment]]s. If experimental data is already available, the [[algorithm]]s can take advantage of that as well. In contrast with Bayesian Networks, path analysis and its generalization, structural equation modeling, serve better to estimate a known causal effect or test a causal model than to generate causal hypotheses. |
||
For nonexperimental data, causal direction can be hinted if information about time is available. This is because causes must precede their effects |
For nonexperimental data, causal direction can be hinted if information about time is available. This is because causes must precede their effects temporally. This can be set up by simple linear regression models, for instance, with an analysis of covariance in which baseline and followup values are known for a theorized cause and effect. The addition of time as a variable, though not proving causality, is a big help in supporting a pre-existing theory of causal direction. For instance, our degree of confidence in the direction and nature of causality is much clearer with a longitudinal epidemiologic study than with a cross-sectional one. |
||
== Manipulation theories == |
== Manipulation theories == |
Revision as of 00:20, 21 September 2005
- For the causal in mysticism, see causal realm.
The philosophical concept of causality or causation refers to the set of all particular "causal" or "cause-and-effect" relations. A neutral definition is notoriously hard to provide since every aspect of causation has received substantial debate. Most generally, causation is a relationship that holds between events, objects, variables, or states of affairs. It is usually presumed that the cause chronologically precedes the effect. Finally, the existence of a causal relationship generally suggests that, all things equal, if the cause occurs the effect will as well (or at least the probability of the effect occurring will increase).
Examples describing causal relationships:
- The cue ball colliding with the eight ball causes the eight ball to roll into the pocket.
- The presence of heat causes water to boil.
- The Moon's gravity causes the Earth's tides.
- A good blow to the arm causes a bruise.
- My pushing of the accelerator caused the car to go faster.
Causation in the history of philosophy
Aristotle
Aristotle suggested four types of cause for a thing which exists: Material, Efficient, Final and Formal.
Take for example the causality involved in creating a silver chalice used in a religious ceremony (this example is from Martin Heidegger). The four causes of the event of its creation are:
- The material cause would be the silver used to create the chalice; the raw matter required by the event.
- The formal cause would be the chalice design itself—the shape in which to form the silver; the design for the use of the raw matter.
- The efficient cause would be the silversmith who took the silver and formed it into shape of the chalice; the actual agent required in turning the raw matter into the desired form.
- The final cause would be the religious ceremony which required a silver chalice in the first place; the ultimate reason behind the event, what compels the agent to make the raw matter into its form.
Note that cause here does not imply a temporal relation between the cause and the effect. See supervenience.
Hume
The philosopher who produced the most striking analysis of causality was David Hume. He asserted that it was impossible to know that certain laws of cause and effect always apply - no matter how many times one observes them occurring. Just because the sun has risen every day since the beginning of the Earth does not mean that it will rise again tomorrow. However, it is impossible to go about one's life without assuming such connections and the best that we can do is to maintain an open mind and never presume that we know any laws of causality for certain. This was used as an argument against metaphysics, ideology and attempts to find theories for everything. A.J. Ayer and Karl Popper both claimed that their respective principles of verification and falsifiability fitted Hume's ideas on causality.
Causality, nihilism, and existentialism
Nihilists subscribe to a deterministic world-view in which the universe is nothing but a chain of meaningless events following one after another according to the law of cause and effect. According to this worldview there is no such thing as "free will", and therefore, no such thing as morality. Learning to bear the burden of a meaningless universe, and justify one's own existence, is the first step toward becoming the "Übermensch" (English: "overman") that Nietzsche speaks of extensively in his philosophical writings.
Nietzsche's life provides an object lesson for some wary of nihilism, maintaining that such lives end quite typically in madness and chaos. Existentialists have suggested that people have the courage to accept that while no meaning has been designed in the universe, we each can provide a meaning for ourselves.
In light of the difficulty philosophers have pointed out in establishing the validity of causal relations, it might seem that the clearest plausible example of causation we have left is our own ability to be the cause of events. If this be so, then our concept of causation would not prevent seeing ourselves as moral agents.
Necessary and sufficient causes
- A similar concept occurs in logic, for this see Necessary and sufficient conditions
Causes are often distinguished into two types: necessary and sufficient. If x is a necessary cause of y, then y will only occur if preceded by x. In this case the presence of x does not ensure that y will occur, but the presence of y ensures that x must have occurred. On the other hand, sufficient causes guarantee the effect. So if x is a sufficient cause of y, the presence x guarantees y. However, other events may also cause y, and thus y's presence does not ensure the presence of x.
J.L. Mackie argues that usual talk of "cause" in fact refers to INUS conditions (insufficient and non-redundant parts of unneccessary but sufficient causes). For example, consider the short circuit as a cause of the house burning down. Consider the collection of events, the short circuit, the presence of oxygen, the flammability of the house, and the absence of firefighters. Altogether these are unnecessary but sufficient to the house's destruction (since many other collection of events certainly have destroyed the house). Within this collection, the short circuit is an insufficient but non-redundant part (since the short circuit by itself would not cause the fire, but the fire will not happen without it). So the short circuit is an INUS cause of the house burning down.
Causality contrasted with logical implication
Logical conditional statements are not statements of causality. Since logical conditional statements and causal statements are both presented using "If...then..." in English they are commonly confused; they are distinct, however. The standard conditional statement expresses a fact about the actual world, while causal statements imply something more. For example all of the following statements are true (interpreting "If... then..." as the logical conditional):
- If George Bush was president of the United States in 2004, then Germany is in Europe
- If George Washington was president of the United States in 2004, then Germany is in Europe
- If George Washington was president of the United States in 2004, then the Moon is made of green cheese
The first is true since both the antecedent and the consequent are true. The second and third are both true because the antecedent is false. Of course, none of these statements express a causal connection between the antecedent and consequent.
Another sort of logical implication, known as counterfactual implication has a stronger connection with causality. However, not even all counterfactual statements count as examples of causality. Consider the following two statements:
- If A is a triangle, then A has three sides.
- If switch S is thrown, then bulb B will light.
In the first case it would not be correct to say that A's being a triangle caused it to have three sides, since the relationship between triangularity and three-sidedness is one of definition. Nonetheless, even interpreted counterfactually, the first statement is true. Most sophisticated accounts of causation find some way to deal with this distinction.
Counterfactual theories of causation
The philosopher David Lewis notably suggested that all statements about causality can be understood as counterfactual statements (Lewis 1973, 1979, and 2000). So, for instance, the statement that John's smoking caused his premature death is equivalent to saying that had John not smoked he would not have prematurely died. (In addition, it need also be true that John did smoke and did prematurely die, although this requirement is not unique to Lewis' theory.)
One problem Lewis' theory confronts is causal preemption. Suppose that John did smoke and did in fact die as a result of that smoking. However, there was a murderer who was bent on killing John, and would have killed him a second later had he not first died from smoking. Here we still want to say that smoking caused John's death. This presents a problem for Lewis' theory since, had John not smoked, he still would have died prematurely. Lewis himself discusses this example, and it has received subsantial discussion. (cf. Bunzl 1980; Ganeri, Noordhof, and Ramachandran 1996; Paul 1998)
Probabilistic causation
Interpreting causation as a deterministic relation means that if A causes B, then A must always be followed by B. In this sense, war does not cause deaths, nor does smoking cause cancer. As a result, many turn to a notion of probabilistic causation. Informally, A probabilistically causes B iff A's occurance increases the probability of B. This is sometimes interpreted to reflect imperfect knowledge of a deterministic system but other times interpreted to mean that the causal system under study has an inherently chancy nature.
The establishing of cause and effect, even with this relaxed reading, is notoriously difficult, expressed by the widely accepted statement "correlation does not imply causation". For instance, the observation that smokers have a dramatically increased lung cancer rate does not establish that smoking must be a cause of that increased cancer rate: maybe there exists a certain genetic defect which both causes cancer and a yearning for nicotine.
In statistics, it is generally accepted that observational studies (like counting cancer cases among smokers and among non-smokers and then comparing the two) can give hints, but can never establish cause and effect. The gold standard for causation here is the randomized experiment: take a large number of people, randomly divide them into two groups, force one group to smoke and prohibit the other group from smoking (ideally in a double-blind setup), then determine whether one group develops a significantly higher lung cancer rate. Random assignment plays a crucial role in the inference to causation because, in the long run, it renders the two groups equivalent in terms of the outcome (cancer) so that any changes will reflect only the manipulation (smoking). Obviously, for ethical reasons this experiment cannot be performed, but the method is widely applicable for less damaging experiments. One limitation of experiments, however, is that whereas they do a good job of testing for the presence of some causal effect they do less well at estimating the size of that effect in a population of interest. (This is a common criticism of studies of safety of food additives that use doses much higher than what people consuming the product would actually ingest.)
That said, under certain assumptions, parts of the causal structure among several variables can be learned from full covariance or case data by the techniques of path analysis and more generally, Bayesian networks. Generally these inference algorithms search through the many possible causal structures among the variables, and remove ones which are strongly incompatible with the observed correlations. In general this leaves a set of possible causal relations, which should then be tested by designing appropriate experiments. If experimental data is already available, the algorithms can take advantage of that as well. In contrast with Bayesian Networks, path analysis and its generalization, structural equation modeling, serve better to estimate a known causal effect or test a causal model than to generate causal hypotheses.
For nonexperimental data, causal direction can be hinted if information about time is available. This is because causes must precede their effects temporally. This can be set up by simple linear regression models, for instance, with an analysis of covariance in which baseline and followup values are known for a theorized cause and effect. The addition of time as a variable, though not proving causality, is a big help in supporting a pre-existing theory of causal direction. For instance, our degree of confidence in the direction and nature of causality is much clearer with a longitudinal epidemiologic study than with a cross-sectional one.
Manipulation theories
Some theorists have equated causality with manipulability (Collingwood 1940; Gasking 1955; Menzies and Price 1993; von Wright 1971). Under these theories, x causes y just in case one can change x in order to change y. This coincides with commonsense notions of causations, since often we ask causal questions in order to change some feature of the world. For instance, we are interested in knowing the causes of crime so that we might find ways of reducing it.
These theories have been criticized on two primary grounds. First, theorists complain that these accounts are circular. Attempting to reduce causal claims to manipulation requires that manipulation is more basic than causal interaction. But describing manipulations in non-causal terms has provided a substantial difficulty.
The second criticism centers around concerns of anthropocentrism. It seems to many people that causality is some existing relationship in the world that we can harness for our desires. If causality is identified with our manipulation, then this inituition is lost. In this sense, it makes humans overly central to interactions in the world.
Some attempts to save manipulability theories, are recent accounts that don't claim to reduce causality to manipulation. These account use manipulation as a sign or feature in causation without claiming that manipulation is more fundamental than causation (Pearl 2000; Woodward 2003).
Process theories
Some theorists are interested in distinguishing between causal processes and non-causal processes (Russell 1948; Salmon 1984). These theorist often want to distinguish between a process and a pseudo-process. As an example, a ball moving through the air (a process) is contrasted with the motion of a shadow (a pseudo-process). The former is causal in nature while the second is not.
Salmon (1984) claims that causal processes can be identified by their ability to transmit a mark or alternation over space and time. An alteration of the ball (a mark by a pen, perhaps) is carried with it as the ball goes through the air. On the other hand an alteration of the shadow (insofar as it is possible) will not be transmitted by the shadow as it moves along.
These theorists claim that the important concept for understanding causality is not causal relationships or causal interactions, but rather identifying causal processes. The former notions can then be defined in terms of causal processes.
Causality in psychology
The above theories are attempts to define a reflectively stable notion of causality. This process uses our standard causal intuitions to develop a theory that we would find satisfactory in identifying causes. Another avenue of research is to discover how ordinary causal talk is employed by everyday people without challenging them. This is often studied in psychology.
Attribution
Attribution theory is the theory concerning how people explain individual occurrences of causation. Attribution can be external (assigning causality to an outside agent or force - claiming that some outside thing motivated the event) or internal (assigning causality to factors within the person - taking personal responsibility or accountability for one's actions and claiming that the person was directly responsible for the event). Taking causation one step further, the type of attribution a person provides influences their future behavior.
The intention behind the cause or the effect can be covered by the subject of action (philosophy). See also accident; blame;intent; responsibility;
Causation and salience
Our view of causation depends on what we consider to be the relevant events. Another way to view the statement, "Lightning causes thunder" is to see both lightning and thunder as two perceptions of the same event, viz., an electric discharge that we perceive first visually and then aurally.
Symbolism and causality
While the names we give objects often refer to their appearance, they can also refer to an object's causal powers - what that object can do, the effects it has on other objects or people. David Sobel and Alison Gopnik from the Psychology Department of UC Berkeley designed a device known as the blicket detector which suggests that "when causal property and perceptual features are equally evident, children are equally as likely to use causal powers as they are to use perceptual properties when naming objects". More Info
Causation in religion and theology
Cosmological argument
One of the classic arguments for the existence of God is known as the "Cosmological argument" or "First cause" argument. It works from the premise that every natural event is the effect of a cause. If this is so, then the events that caused today's events must have had causes themselves, which must have had causes, and so forth. If the chain never ends, then one must uphold the hypothesis of an "actual infinite", which is often regarded as problematic, see Hilbert's paradox of the Grand Hotel. If the chain does end, it must end with a non-natural or supernatural cause at the start of the natural world -- e.g. a creation by God.
Sometimes the argument is made in non-temporal terms. The chain doesn't go back in time, it goes downward into the ever-more enduring facts, and thus toward the timeless.
Two questions that can help to focus the argument are:
1) What is an event without cause?
2) How does an event without a cause occur?
Critics of this argument point out problems with it.
Karma
Karma is the belief held by some major religions that a person's actions cause certain effects in future incarnations, positively or negatively.
Reversed causality
Some modern religious movements have postulated along the lines of philosophical idealism that causality is actually reversed from the direction normally presumed. According to these groups, causality does not proceed inward, from external random causes toward effects on a perceiving individual, but rather outward, from a perceiving individual's causative mental requests toward responsive external physical effects that only seem to be independent causes. These groups have accordingly developed new causality principles such as the doctrine of responsibility assumption.
Causality in science and the humanities
Using the Scientific method, scientists set up experiments to determine causality in the physical world. Certain elemental forces such as gravity, the strong and weak nuclear forces, and electromagnetism are said to be the four fundamental forces which are the causes of all other events in the universe.
However, the issue of to which degree a scientific experiment is replicable has been often raised but rarely addressed. The fact that no experiment is entirely replicable questions some core assumptions in science.
In addition, many scientists in a variety of fields disagree that experiments are necessary to determine causality. For example, the link between smoking and lung cancer is considered proven by health agencies of the United States government, but experimental methods (for example, randomized controlled trials) were not used to establish that link. This view has been controversial. In addition, many philosophers are begining to turn to more relativized notions of causality. Rather than providing a theory of causality in toto, they opt to provide a theory of causality in biology or causality in physics.
Physics
Causality is hard to interpret in many different physical theories. One problem is typified by the moon's gravity. It isn't accurate to say, "the moon exerts a gravitic pull and then the tides rise." In Newtonian mechanics gravity, rather, is a law expressing a constant observable relationship among masses, and the movement of the tides is an example of that relationship. There are no discrete events or "pulls" that can be said to precede the rising of tides. Interpreting gravity causally is even more complicated in general relativity Another important implication of Causality in physics is its intimate connection to the Second Law of Thermodynamics - see the fluctuation theorem.
History
In the field of history, the term cause has at least two meanings, often mistakenly conflated.
- One meaning conforms to Aristotle's final cause -- as a goal or purpose. For example, the abolition of slavery became a Union goal or intended outcome for the American Civil War following the Emancipation Proclamations and so was a cause or reason to continue the war. This meaning is not what is meant by the term causality.
- Another meaning treats historic events as agents that bring about other historic events. This is a somewhat Platonic and Hegelian view that reifies causes as ontological entities and the term causality is used sometimes in this manner. In this view, slavery is often said to have inevitably produced the American Civil War as a result. In Aristotelian terminology, this use of the term cause is closest to his efficient cause.
Causality in law
According to law and jurisprudence, legal cause must be demonstrated in order to hold a defendant liable for a crime or a tort (ie. a civil wrong such as negligence or trespass). It must be proven that causality, or a 'sufficient causal link' relates the defendant's actions to the criminal event or damage in question.
See also
- Causality loop
- Determinism
- Free will
- Linear regression
- Post hoc ergo propter hoc
- Proximate causation
- Randomness
- Causal Markov condition
External links
Stanford Encyclopedia of Philosophy:
- Backwards Causation
- Probabilistic Causation
- Causation and Manipulability
- Counterfactual Theories of Causation
- Causal Processes
- Causation in the Law
- Medieval Theories of Causation
- The Metaphysics of Causation
General
- Dictionary of the History of Ideas: Causation
- Dictionary of the History of Ideas: Causation in Law
- Dictionary of the History of Ideas: Causation in History
References
Counterfactual accounts of causation
- Bunzl, Martin. (1980) "Causal Preemption and Counterfactuals." Philosophical Studies 37: 115-124
- Ganeri, Jonardon, Paul Noordhof, and Murali Ramachandran. (1996) "Counterfactuals and Preemptive Causation" Analysis 56(4): 219-225.
- Lewis, David. (1973) "Causality." The Journal of Philosophy 70:556-567.
- ----. (1979) "Counterfactual Dependence and Time's Arrow" Noûs 13: 445-476.
- ----. (2000) "Causation as Influence" The Journal of Philosophy 97: 182-197.
- Paul, L.A. (1998) "Problems with Late Preemption" Analysis 58(1): 48-53.
Probabilistic causation
- Pearl, Judea (2000) Causality, Cambridge University Press, ISBN 0521773628
- Spirtes, Peter, Clark Glymour and Richard Scheines Causation, Prediction, and Search, MIT Press, ISBN 0262194406
Manipulation
- Collingwood, R.(1940) An Essay on Metaphysics. Clarendon Press.
- Gasking, D. (1955) "Causation and Recipes" Mind (64): 479-487.
- Menzies, P. and H. Price (1993) "Causation as a Secondary Quality" British Journal for the Philosophy of Science (44): 187-203.
- Pearl, Judea (2000) Causality. Cambridge University Press, ISBN 0521773628
- von Wright, G.(1971) Explanation and Understanding. Cornell University Press.
- Woodward, James (2003) Making Things Happen: A Theory of Causal Explanation. Oxford University Press, ISBN 0195155270
Process theory
- Russell, B. (1948) Human Knowledge. Simon and Schuster.
- Salmon, W. (1984) Scientific Explanation and the Causal Structure of the World. Princeton University Press.