A QUICK TOUR OF CAUSATION, PROBABILISTIC CAUSATION, DETERMINISM, AND FREEDOM AND RESPONSIBILITY

A talk by Ted Honderich to the Centre Cournot, Paris
16 Oct 2008, 18.00
There is also a French version of this quick tour.


CAUSATION AS CONDITIONAL CONNECTIONS IN THE WORLD

Strike a match. It lights. We call these events cause and effect. What do we mean?

1. Since the striking (s) happened, in the situation as it was, including oxygen, so too did the lighting (l) follow. By way of a label, the striking required the lighting.

2. If  the striking hadn't happened in the situation, where for example the match wasn't held in the flame of another match, neither would the lighting have followed. The striking was required for the lighting.

The same two kinds of conditional connections in the world, each dependent on the situation, hold between each event in certain sets of events that we can call causal circumstances for the lighting. A causal circumstance (cc) included the event that for some reason we pick out and call the cause -- the striking (s).

3. There was also a different and fundamental kind of conditional connection, an independent one, between cc and the lighting. Since cc occurred, whatever the situation had been, the lighting would still have occurred. Expressed differently, since cc occurred, whatever else had been happening, l would still have occurred. By way of a label, the causal circumstance necessitated the lighting.

4. Since l occurred, cc would still have occurred before it in the absence of another causal circumstance for l. The causal circumstance cc was necessary for l.

This independent-conditionals theory of standard effects, at bottom about whatever-else connections in the world, is not open to the great to the regularity or constant conjunction theory of David Hume -- in short the objection that Hume's theory makes last night the effect of yesterday, since days and nights go together regularly. On the conditionals theory, it is not true that yesterday caused last night. This is because last night would not have happened whatever else had been happening in addition to yesterday. Last night would not have happened if the solar conditions changed at the end of yesterday.

The conditionals theory of causation is also not open to objections against David Lewis's related counterfactual theory in terms of a logic of possible worlds.


PROBABILISTIC CAUSATION

The central claim in the theory of probabilistic causation, however it is to be understood, is said to be that causes raise the probability of the occurrence of their effects. Speaking generally, A causes B iff P(B | A) > P(B | not-A).

Better expressed, an event A causes an event B iff A makes B more probable than not-A would, other things being equal, even if A leaves B hardly probable at all.

To this idea, well introduced by Christopher Hitchcock, an advocate of it, in an article in the Stanford Encyclopedia of Philosophy, there are many objections.

(i) Almost all of us believe, despite the temptation of common interpretations of Quantum Theory, that when the match was struck, in the situation as it was, it had to light. The lighting was necessitated. If the match hadn't lit, we wouldn't for that reason think that if it had lit, that lighting wouldn't have been necessitated. Rather we would think that something was missing in the situation when it didn't light. So the simplest and to my mind the overwhelming response to probabilism, as an account of our conception of causation and also what it is, is that it wipes out this proposition of the necessity of effects.

(ii) It is a related fact, too often passed by quietly, that probabilistic causation leaves the actual occurrence of events, certainly events with a probability of less than 1, without any explanation at all. If probabilistic causation is put in place of causation, there is no answer at all to why an event actually occurred, as distinct from being probable. It is not too much to say that on this view, reality is mystery. So this view is remote from causation as we think about it and know it.

(iii) The indeterminist interpretations of the basic mathematics or formalism of Quantum Theory are a mess. This is so in the more deferentially-expressed admissions of those philosophers and theoreticians of science, including scientists, most disposed to the interpretations. The fact that inconsistent, weird, and mysterious interpretations are tolerated is an unfortunate effect of a rightful hegemony of science. It is is no significant argument that the theory 'works'. Classical mechanics worked. Inconsistent theories can work.

(iv) I am no physicist, but able enough to read the principal expositions of interpretations of Quantum Theory by physicists for non-physicists. It is common to read or to understand or to suspect that the items taken as non-effects are also not events. This plethora of ideas as to non-effects is about items that do not happen. But no determinism is the theory that propositions or numbers or other abstract entities are effects. No determinist has ever contemplated that the number 1, as against inscriptions of it etc, is an effect.

(v) Indeterminists face a dilemma. Given the fact of macro-determinism, including the macro-determinism of the brain, either there is no micro-indeterminism underneath or else it cancels out or anyway doesn't translate or amplify upwards into the macro-world. So, it must seem, the micro-indeterminism of interpretations of Quantum Theory is either false or irrelevant.

(vi) It is argued that interpretations of Quantum Theory establish, about events of which we are sure that they cause cancer, that these events are undetermined or unnecessitated. So, unless we take causation to be probabilistic, we will have to be agnostic about well-supported or even best-supported causal claims. A reply is that if this is about what causation is taken to be, or what it would be if it existed, there is a less confusing and maybe less confused response. If later unnecessitated events are somehow explained by prior events in some unnecessitating way, then the prior events are not causes, whatever else is to be said of them. But it doesn't matter too greatly who gets possession of the word 'cause' and like words. What matters is clear distinctions, evidence etc.

(vii) Some probabilists are motivated, as they say, by the proposition that smoking doesn't always cause cancer. Or the proposition, to use our example, that s-like events don't always cause l-like events. So, they then conclude, smoking causes cancer only in the sense of making cancer more probable, as s made l more probable. Whatever motivating their proposition does, it is no argument at all against the independent-conditionals theory. That theory itself asserts that s only requires l and doesn't necessitate it.

(viii) It is objected that the complexity of the independent-conditionals theory, like related theories, makes it less appealing than what is unkindly taken to be Hume's simple theory. It is in fact supposed that the acute Hume thought that the cause that is the striking of a match is always followed by the effect of its lighting. I can't believe it, despite his quick words. In any case, it is no serious objection that the conditionals theory is more complicated, partly since any arguable theory of probabilistic causation must be as complicated.

(ix) Probabilists seem to assume that probability judgements are not a matter of causation, but rather an independent kind of explanation. Suppose three brothers, all heavy smokers, die of cancer, and the fourth brother, also a heavy smoker, is not looking good. If a probability statement is offered as a ground for a sad prediction about him, what can it be based on but causal reasoning of the kind sketched above? Of course the probability statement is logically consistent, for example, with somebody's idea about a God who occasionally flies into a temper about too much smoking in a family, and commands a death that otherwise would not have happened. But the probability judgements of the rest of us just are results of ordinary causal reasoning, aren't they? At bottom, they rest on a good idea of a kind of causal circumstance for something, and the knowledge that all of or a lot of or some of one exists. This objection can be made in connection with the various general theories or interpetations of probability judgements -- in terms of what are spoken of as upshots among equally probable ones, good or logical reasons, frequency of past events, subjectivity, and objective or real chance in the world.

(x) The probabilistic causation theory faces problems. A golfer hits a bad stroke and the ball goes off the fairway into the rough. This makes it less probable than a good stroke would have been that the bad stroke will be a hole-in-one. But the ball bounces off a tree and ends up in the hole. Probabilistic causation says the bad stroke didn't cause the hole-in-one. All theories face problems, of course. Probabilistic causation needs epicycles and contortions to deal with its problems.

(xi) Another problem is causal direction or priority or assymetry. Causes explain their effects and make them happen, but effects do not make explain their causes or make them happen. To be still more brief, it strikes me that the account of this fact given by the conditionals theory is more persuasive than accounts given by probabilistic causation.


DETERMINISM

There can be a conceptually adequate theory of human determinism -- determinism having to do with our choices and decisions and other antecedents of our actions. It rests on an adequate conception of causation, say causation as conditional connections. It also rests on a conceptually adequate theory of consciousness, and its relation to the brain.

The theory, in brief, has three parts. (1) Conscious events are in some kind of lawlike connection with simultaneous neural events, but are not identical with them. Two such events are in some whatever-else connection but not exactly causal connection. They are different only in being simultaneous. They can be called psychoneural pairs. (2) These pairs are effects of certain sequences of causal circumstances and effects. (3) Actions are effects of causal sequences beginning with psychoneural pairs.

The principal and great evidence for the theory is standard neuroscience. On the basis of this evidence, and also because of the weakness noted above of arguments based on interpretations of Quantum Theory, determinism remains at least a reasonable assumption.


FREEDOM AND RESPONSIBILITY

The tradition of Compatibilism in English and American philosophy asserts that our single, settled and fundamental conception of freedom is voluntariness. It is, in short, doing what we want, action that is the effect of causation internal to us rather than external to us. Hence we are free when we are not in jail, or otherwise constrained or compelled. The opposed tradition of Incompatibilism asserts that our conception of freedom is or includes origination or free will. This is initiation of action that could have been otherwise given things exactly as they were -- initiation that is uncaused but still within the control of the actor.

Thus Compatibilism is that determinism and freedom co-exist. Determinism is logically compatible or consistent with freedom and the absolutely inseparable fact of responsibility. Determinism is compatible with our single, settled and fundamental conception of freedom and also with our one attitude of holding people responsible for their actions and crediting them with responsibility for them. Incompatibilism is that determinism and freedom cannot co-exist. Determinism is logically incompatible with freedom and responsibility.

To me, and to an increasing number of philosophers, Compatibilism and Incompatibilism, of which endless proofs have been attempted, are both provably false -- because of their shared presupposition that we have but one settled conception of freedom and responsibility. On the contrary, we have both ideas -- freedom as voluntariness and freedom as origination. They inform such attitudes as hope for the future and such institutions as rights and punishment by the state. What is true, by way of a label, is Attitudinism.

The true problem of determinism has seemed to me not whether Compatibilism or Incompatibilism is true, but only the practical problem of how to give up our conception of freedom as origination and the related attitudes having to do with responsibility.

One postscript now seems to me advisable. It may not be possible to avoid feeling that one has had an independence in one's life, that one has been in a way accountable. This is not origination, but it does or would explain the standing we take ourselves to have, our difference from everything else that is subject to determinism.

This independence may be a matter of a special explanatoriness in our lives of the line of causes (within a causal sequence) which consists in our ongoing neural existence. Also of a new conception of consciousness, at bottom an externalist conception of perceptual consciousness as consisting in worlds of perceptual consciousness. They have a dependency on us neurally as well as the micro-world underneath them. We are in a way creative of a reality, or jointly sustaining with respect to it. It is worth remembering, too, from Kant and others, that we can be free in in the sense of acting rightly, say in accord with the Principle of Humanity, as against being subject to the desires of self.


Reading on probabilistic causation:
Christopher Hitchcock, 'Probabilistic Causation', Stanford Encyclopedia of Philosophy, http://plato.stanford.edu/ Sept 6, 2002.
Ellery Eells, Probabilistic Causality, Cambridge University Press, 1999.
Paul Humphreys, The Chances of Explanation, Princeton University Press, 1989. David Lewis, Philosophical Papers, Vol. 2, Oxford University Press, 1986.
John Mackie, The Cement of the Universe, Oxford University Press, 1974.
Hugh Mellor, The Facts of Causation, Routledge, 1995.
Wesley Salmon, 'Probabilistic Causality', Pacific Philosophical Quarterly, 1980. Brian Skyrms, Causal Necessity, 1980.


Honderich publications on causation, determinism, freedom etc:
Êtes-vous libre? Le problème du déterminisme
, Syllepse, 2008.
How Free Are You? The Determinism Problem,
2nd ed. Oxford University Press, 2002 -- also German, Greek, Italian, Japanese, Spanish, Swedish, Polish and Romanian translations.
On Determinism and Freedom, collected papers, Edinburgh University Press, 2005. A Theory of Determinism: The Mind, Neuroscience and Life-Hopes, OUP, 1988, of which Chs 1-6 are the paperback Mind and Brain and Chs 7-10 the paperback The Consequences of Determinism, OUP, 1990.
On Consciousness, EUP 2004.
Anthony Freeman, ed.,
Radical Externalism: Honderich's Theory of Consciousness Discussed, Imprint Academic, 2006.
Punishment: The Supposed Justifications Revisited, Pluto Press, 2005.
Papers on causation etc at http://www.homepages.ucl.ac.uk/~uctytho/

To the French version
HOME to T.H. website front page
HOME to Det & Free website front page