Concerning Objections to a Causal Theory of Action

In this paper I give a very brief outline of a naturalistic causal theory of action and discuss a counter-example which purportedly disproves it.

When I speak of a causal theory of action, I am referring to a theory of the sort where A's action X can be explained by the fact that A intended to get Y and believed that doing X was the best way to get Y1. A's intent to get Y and A's belief that doing X would be the best way to get Y are two distinct mental states that cause A's action of doing X, in the same way that one physical event causes another, such as a billiard ball striking another billiard ball. Is it is possible to account for intentional human actions in this naturalistic way, i.e. can intentionality be accounted for by employing a scientific causal description, or must we adopt a non-physical description? An anti-naturalist must demonstrate that there are cases where the causal theory fails to apply.

Let us consider one such counter-example to the causal theory of action: the case of the "killer nephew". A nephew wants to inherit his uncle's fortune and decides to kill him. He takes a pistol and drives to his uncle's house with the intent of killing him. On the way, he loses control of the car and runs over a pedestrian. It just so happens that the pedestrian is his uncle. In this case, the nephew killed the uncle because of his intent to kill him (i.e. the causal chain can be traced back from the killing of the uncle back to the nephew's intent to kill him), and yet the killing of the uncle was unintentional. How do we explain this? Adherents of agent-causation or other non-naturalistic accounts of intentional action would have us distinguish between the course of events outlined above (let us call it C) and the nephew's planned course of events (let us call it P) as being fundamentally different -- one is supposedly a case of natural causation while the other is not, i.e. it is a causation of some other sort, whether it be agent-causation, purposive causation or something else.

Quite frankly, I find this counter-example to be trivial. The reason why our intuition tells us that one is intentional and the other not is simply our familiarity with common-sense human planning. In both P and C, the nephew has a clear plan -- to kill his uncle in a so-and-so fashion. This plan may include actions such as stepping into the car, driving to the house, stepping out of the car, entering the uncle's house, etc., culminating in the act of actually killing the uncle. The nephew has a clear conception of which stage of the plan should involve the killing -- but in C, the killing takes place in a step of the plan which was not supposed to involve the killing. This is the reason we regard it as unintentional. I therefore think Nannini2 is right when he says that C is unintentional and P intentional "because the course of events is different in the two cases".

So much for the "killer nephew". Let us look at a more sophisticated (but also more strenuously contrived) version of this counter-example: the case of the "killer mountaineer". Two mountaineers are scaling a mountain, and one is planning to kill the other. At some point, the killer is holding his companion by a rope and sees the opportunity to kill him by letting go of the rope. He decides to let go, but this intention excites him so much that he unintentionally lets go of the rope at the same moment and in the same fashion as he would have, had he done this intentionally. In this example, the causal chain external to the agent is identical, regardless of the killer's intentions, but our intuition still tells us that one is intentional and that the other is not.

In response to this example, we might instantly raise the (rather petty) objection that the external causal chain is never going to be identical in the two cases. It is easy to argue that the physical act of letting go of a rope due to a state of excitement is always physically different from consciously and deliberately letting go, and that this example is therefore too contrived and artificial to be admitted into the discussion. But let us never the less admit it for the sake of argument and see whether it poses a serious threat to the causal theory.

As far as I can tell, the fundamental difference between this example and that of the "killer nephew" is that the difference in causal chains between the intentional and unintentional story has been moved inside the mind of the agent. Can we apply the arguments against the "killer nephew" here? I believe so. Let us look at the following diagrams:


A) roughly represents the chains of events in the case of unintentional killing, while B) represents the case of intentional killing. In the case of A), the second step, namely the state of excitement, is causally vital, while in B) it is incidental and plays no role at all. The B) process would have culminated in the act of killing whether the agent experienced a state of excitement due to his intention or not, whereas in A) the state of excitement is essential in explaining the course of events. The state of excitement might have been caused by a number of things other than the intent to kill, e.g. falling rocks or the sighting of a belligerent mountain lion. Thus we can say that the state of excitement was the causally efficacious mental state, and not the intention, while in B) the causally efficacious mentally state is clearly the intention of killing.

Of course, the diagrams above assume that we can legitimately distinguish between the mental state of intention and the mentat state of excitement, as separate, causally independent mentat states. If we do not allow ourselves to make this distinction, the plot thickens: in both A) and B), the killing can be described as being caused by a single mental state of "excited intention", and our analysis falls to pieces. The mind, under this view, becomes a sort of black box -- we cannot hypothesize about causal relationships between mental states. How can we respond to this?

We could maintain that if the technique of analysing causal relationships between mental states is knocked out of our hands, it leaves nothing to account for i.e. that the problem of accounting for the common-sense understanding of intentions disappears altogether. It is certainly common-sensical to speak of one mental state causing another, e.g. "I thought about X and as a consequence I became very angry". Opponents of the causal theory of action show inconsistency if they reject the common-sensical account in one matter, but use it for offense in another. A naturalist could say that if the intentionalist refuses to admit causally independent mental states, then he, the naturalist, can easily refuse to admit intentional states. This, however, leads us away from the dispute, not towards any real solution. I therefore prefer the approach suggested by Nannini3, namely that we analyse the supposedly single mental state "excited intention" in terms of causally efficacious properties. Just as the mass of a billiard ball is causally efficacious when it hits another ball, its colour may be causally efficacious in different circumstances. Thus, different properties are causally efficacious under different circumstances. If we apply this to a mental state, we get the following (where causal efficacy is indicated by the arrow):


This way, we can say that in the case of A), the mental state's property of being excited, P1, is causally necessary in order to lead to the killing, while in B), the property of intentending to kill, P2, is necessary to lead to the killing. Therefore, the argument that we leveraged against the "killer nephew" still applies to the "killer mountaineer": the agent's planned course of events B) differed from the actual course of events A), either because the action of killing was caused, not by an intentional mental state, but by a state of excitement, or alternately (in response to the black box model), by a the non-intentional mental state property P1.

In conclusion, neither the counter-example of the "killer nephew" nor the counter-example of the "kill mountaineer" pose a problem to a causal theory of action, since it is possible in both cases to demonstrate how the course of events differ between intentional and non-intentional actions. In one case, the difference lies in the physical course of events, the other in either the causal relationship between mental states or in the causally relevant properties of a single mental state.

Reykjavík, 2nd of February 2006

Sveinbjorn Thordarson






Endnotes

1. For a more detailed account, vide Nannini p. 166
2. Vide Nannini p. 169
3. Vide Nannini p. 171





Sources

Nannini, Sandro "Does the Content of Intentions Have Effects Upon Actions", p. 166-179, NuCog: Cognition, Motivation, Action, 2006 Université de Technologie de Compiégne, France