Photo: HipWallpaper
Photo: HipWallpaper

(For Miciko)

“Any thing may produce any thing.” Hume, A Treatise of Human Nature

Billiard balls

Rereading Hicks’s book on causality,1 I was led to rereading Hume, then Kant, and finally, Leibniz (and a bit of Hegel). Hicks takes Hume to be singling out two characteristics of causality:

– one, that the cause comes before the effect (which Hicks and everybody else takes to mean “in time” for Hume); and

– two, that causality is determined counterfactually (no effect, if not for the cause).

He is wrong in his interpretation of Hume, I think, on both counts.2

Both, Hicks and Hume, however, at least in the treatment of the role of time, differ from the Wiener-Granger (information based) causality.3 Where the starting point is that causes occur ahead of effects – the information about the cause precedes the information about the effect in time; or, to put it differently, they cannot be contemporaneous. Contiguity, in space, plays no role as far as I can tell. Apparent instantaneous causality is just an instance of the existence of the common cause, for Granger. If there were instantaneous causation, it is believed that the asymmetry between causes and effects would be lost, i.e. causes could not be discerned from effects. And forecasting from present causes to future effects would not be possible. That does not exclude feedback loops, i.e. the effects switching places with the causes in the future. In addition, this kind of causality does not support counterfactual claims. Thus it has limited usefulness for policy advice. Wiener in The Theory of Prediction points out that predicting the effects of interventions is different from, may I say, Humean observations (see the quote and comments in Appendix 2 below).

Thus, given that Granger’s theory of causality is used for forecasting rather than explanation, every causal impact comes with a time lag – is the infamous “action at a distance”; though it is just the information that lags, no act of causation is actually implied. Where the distance in time is arbitrarily large and is mainly dependent on the frequency of the available data collection.

Hicks’ major contribution in this book is that causality can be contemporaneous. That is not an idea that is new with him. So, he faults Hume with the lack of understanding of contemporaneous causality. Though the equilibrium analysis on which it rests is central to classical economics, to which Hume contributed significantly, as indeed to economics in general.

For Hume, causality is a relation. There is nothing in any thing which distinguishes it as a cause or as an effect. The causal relation is characterised by two features: succession (asymmetry) and contiguity (symmetry). Causes precede effects and the two need to be close to each other. It is taken that he means that succession is in time while contiguity is primarily in space. But that is arguably not what he means.

Occasionally (primarily in the Enquiry), Hume takes the example of two billiard balls hitting each other. And clearly, cause and effect can be discerned, if at all, only at the very moment when the balls touch each other and no time elapses between the one ball hitting the other and the other moving away. So, there is, we induce, a causal relation between the movements of the two balls, with the cause preceding the effect and the two being contiguous. Though, at the precise moment when the cause produces the effect, they, the cause and the effect, are symmetrical in respect to both time and space – there is no precedence in time and no distance in space between the two. They appear interdependent.

This is clear if we look at the impact of the still ball on the moving one. There is the causal impact of the moving ball on the still one and indeed of the still one on the moving ball. In the case of the latter, there is no lapse of time between the cause and the effect. But there is none in the case of the moving ball too, as the effect happens at the same time as the cause, i.e. when the running ball hits the still one.

Anticipating somewhat one of the main points to be made later, Hegel in The Science of Logic does not use the billiard balls example, but he does point out the reciprocal or interdependent relationship of cause and effect which the example highlights. The courses of both balls are altered by the impact and Hegel suggests that the interdependency of cause and effect can be dealt either with a caeteris paribus condition or with an equilibrium condition. He does say cause is effect and effect is cause repeatedly, which does sound dialectical, but the point he makes of the two being interdependent is rather conventional and easy to spot in the billiard balls example. And in any case, in a relation, both sides are simultaneously present by definition, which is what Hegel wants to highlight, I think, as he progresses to the concept of causation (see the quotes in Appendix 4).

PDF of the whole essay

Pešč, 27.06.2020.


  1. J. R. Hicks, Causality in Economics. Basil Blackwell, 1980. This is not an exceptionally good book on this subject. It is of interest because it recasts Hicks’ fundamental contributions in economic theory in Value and Capital and Capital and Time in the causal language. The same ideas of simultaneity and equilibrium are to be found in the work of Frisch, Haavelmo and Samuelson and indeed in Smith, Ricardo, Marshall and Keynes as I document with quotes from their work throughout this essay. See e.g. Heckman on Haavelmo.
  2. I mostly rely on the Treatise on Human Nature here, the quote and the example, which are from the Enquiry, notwithstanding. This is a companion piece to my essay on Causes and Counterfactuals: Simple Ideas where I rely more on An Enquiry Concerning Human Understanding. While there are differences in formulations, those in the latter work being smoother and less shocking, there is no difference or evolution in the fundamentally sceptical view of causality and the power of induction.
  3. See C. W. J. Granger, Essays in Econometrics. Collected Papers. Volume II. Cambridge University Press, 2001. See Appendix 2 for Wiener.