Ned Hall on the Generative Conception of Causation
Ned Hall aims to spell out the production view of causation, or, as he calls it, the generative conception. However, as far as I can see, despite his use of the notion ‘pure causal history’ his analysis does not go very far beyond the superficial characterisation offered by Bunge, notably that c is the cause to e iff whenever C happens in circumstances S, E is produced. As usual, it is assumed that S is to be defined in such a way as to exclude the presence of any intervening factor, P, in the minimal set of components constituting S; one that would prevent C from producing £ (these assumptions are what Hall refers to below as ‘foregoing constraints’):
Given some event e occurring at time t’ and given some earlier time t, we will say that e has a pure causal history back to time t just in case there is, at every time between t and t’, a unique minimally sufficient set for e, and the collection of these sets meets the two foregoing constraints. We will call the structure consisting of the members of these sets the “pure causal history” of e, back to time t.
(Hall 2004: 265)
The problem here, from my perspective, is that Hall’s account gives us no suggestion as to what kind of entities would constitute the minimally sufficient set for e, or in what manner that set brings about e, wherefore we are still left with only a two-place relation between the two events et and e, which is claimed to involve ‘generation’, but whose nature is left unanalysed.
Stuart Glennan: The Mechanistic Account of Causal Production
Glennan’s account is an attempt to say something more about what kind of entity is sufficient for bringing about e. As he himself notes, his account may at first blush look similar to previously offered accounts. We can have Esfeld’s and Hall’s account in mind, even though Glennan does not explicitly mention anyone in particular. One main difference is that we are explicitly told not to understand the event that brings about an effect merely in terms of Kim-style events, i.e. as the exemplification of properties (or relations) by an object (or set of objects) at a time (Kim 1973, 1976), but in terms of an activity of the objects involved in the event (Glennan 2017: 177). Glennan’s point is to stress the processual nature of causation as a natural consequence of taking causal production to consist in the exertion of influence between particulars; that this exertion must be understood as an activity, which is something that cannot fit within an instantaneous Kim-style event. To be more precise, Glennan thinks of events as activities of particulars and not as entities that themselves are engaged in activities such as ‘causing’.
Glennan’s account in many ways resembles Salmon’s but is sensitive to other details in the causal structure of the world. He suggests we understand causation in terms of constitutive, precipitating, and chained production. Constitutive production corresponds roughly to what Salmon calls ‘production’, i.e. the ‘the ways in which a single event produces changes in its constituent entities’ (Glennan 2017: 180), but with added detail. Glennan illustrates constitutive production using the example of searing a steak. The interaction between the steak and the skillet involves the transference of energy to the steak which produces various reactions between the constituents of the steak, i.e. water, proteins, carbohydrates, and fat. Glennan suggests we focus on the changes that matter to the cook, i.e. what happens in the steak. When the temperature of the steak rises above a certain limit (+140°C), the constituents of the steak—various sugars and amino acids—react with each other to produce a variety of molecules responsible for a range of pleasant flavours and aromas. This is known as the Maillard reaction. The details are not essential to our story, but for those interested see McGee (2004: 778-9). For my purposes, what the reader should remember for the discussion that follows is that Glennan describes production in the following terms: interaction between two particulars (the skillet and steak) is the cause to the changes occurring in the particulars (special focus on the reactions between the constituents of the steak), and that both the interaction and the ensuing changes are processual rather than Kim-style events.
Precipitating production is ‘the way in which one or more events produce another event [...] by creating start-up conditions for a different mechanism’ (2017: 182). The steak being placed on the skillet is given as an example of an event that creates the start-up conditions for the Maillard reaction in the steak.
Finally, chained production is a connection between events between which there exists a chain of intermediary precipitating events. We might say that the connection between someone buying a steak and eating it is a series of precipitating events starting with grabbing the steak at the store, putting it in the carrier bag, carrying it home, putting it onto the skillet, browning it, placing it on the plate, and finally eating it.
Is Glennan’s account an improvement on previous accounts? I think so. It stresses that causation is at rock bottom a process of production in which propertied objects interact to change each other rather than merely a relation between Kim-style events. In doing so he implicitly takes sides with Salmon (1980) and myself (Ingthorsson 2002), as well as Chakravartty (2005) and Mumford &c Anjum (2011), in saying that the kind of relational account offered by Hall (2004) and Esfeld (2011) isn’t enough. But in so far as he moves beyond Salmon’s account, it is mainly by way of going into greater detail about the changes that happen intrinsically to the objects as a result of their interaction, i.e. what happens in the steak when it interacts with the skillet. That puts increased focus on what happens intrinsically to objects.
In anticipation of what follows in the next chapter, then my complaint is that while Glennan’s account—in its abstract form—verges on being identical to the account I offer in ‘Causal Production as Interaction’ (Ingthorsson 2002), then in the way he concretely illustrates it he appears to be at least implicitly influenced by the idea that causation involves unidirectional influence. He thinks the skillet influences the steak unidirectionally (or at least more than the other way around):
What makes this a productive activity is, in the first instance, the fact that the activity of searing produces changes in the steak and, to a lesser extent, the pan. Let us focus on the change that matters for the cook: searing both browns and cooks the steak; it also releases oil, water, and other compounds from the steak.
(Glennan 2017: 180)
It should be admitted that I find the unidirectionality in Glennan’s account less explicit than in other accounts, or rather, it can be read as verging on a proper reciprocal understanding. It is mentioned that something does happen to the skillet, but this is somehow a ‘lesser’ consequence than what happens to the steak, or at least less important to us. However, as I will argue in the next chapter, the risk is that the reason the effects on the skillet are judged to be ‘lesser’ are anthropocentric. We care what happens to the steak, but not too much about what happens to the skillet; but does the universe care more about steaks than skillets? It also ignores that what the reciprocality of physical interactions tells us, as described by physics, is that, in terms of the abstract quantity of influence we call ‘force’, the skillet is influenced in the same quantity as the steak; the skillet cools down as much as the steak heats up. The difference is the characteristics of the change that this same quantity of influence has on skillet vs. steak; the influence that will change the temperature of the steak to the degree that it becomes seared will also change the temperature of the skillet to the same degree, but will not make the skillet tasty. Now, while the cooling of the skillet is usually less important to us, sometimes it is quite annoying. If we put too much meat on the skillet in one go, the temperature may drop below what is required to produce the Maillard reaction and the meat will cook instead of being seared. That is a disaster in any gourmand’s mind (Gordon Ramsey will give any sous-chef a hiding that makes a mistake like that).
The novelty of Glennan’s account lies mainly in the characterisation of constitutive production. Without it, precipitating production would be equivalent to the kind of relational realist view offered by Hall (2004) and Esfeld (2011). After all, chained production is really just a succession of precipitating productions, and precipitating production is a succession of constitutive production. Arguably, constitutive production is an elaboration on Salmon’s idea of production, but then Glennan focuses not so much on what Salmon calls propagation, i.e. the preservation of a status quo in the object when it is not interacting, but more on the intrinsic changes of the particulars when they interact, and then looks at the more coarse-grained structure of the world in terms of precipitating and chained production that is grounded on the close-knit structure of constitutive productions. This is a reasonable thing to do from a pragmatic perspective, in our attempt to make sense of what is happening to us and the world we live in. We need to make a difference between what happens on the scale where nicotine molecules cause the destruction of organic cells in the alveola of the lungs (constitutive production), the scale at which that destruction causes a person to cough (precipitating production), and the scale at which smoking causes cancer (chained production). It is a shift in perspective from the genuine proximate causes and effects to the distal causes that emerge on a larger scale, and yet it all boils down in the end to the close-knit structure of proximate/constitutive causation that constitutes precipitating and chained production. Indeed, Glennan points out that when we start to talk about chained production, we don’t get clear linear chains, but intertwining chains of precipitating production (which are themselves constituted by chains of constitutive production).
Glennan moves in many ways beyond Salmon’s account, but I miss the distinction between production and propagation. Production, on Salmon’s account, are instances of disruption and change in the causal processes, and Glennan goes into minute details over what happens in one of the causal processes involved. Propagation, on the other hand, are periods during which the object stays the same in the respects relevant for any changes that happen later when it again interacts with another causal process. Propagation is not explicitly accounted for in Glennan’s account, either in terms of chained or precipitating causation. Salmon stresses that causal processes, i.e. particulars, are continuously changing even through periods of propagation, but they must still be able to preserve a certain structure through that change. Now, I would not say that Salmon gives an account of propagation, but he at least acknowledges that this seems to be an important feature of the causal structure of the world. In Chapter 6, I will argue that what Salmon describes as propagation is a causal phenomenon, but one that should be cashed out as a combination of constitution and persistence. Indeed, I will argue that sometimes constitutive causation produces persistence.
To conclude, what is missing from Glennan’s account is the same thing missing from the processual version of the transmission theory. How do we explain the exertion of influence or transference of energy between interacting objects? There is no other causal mechanism implied other than what can be understood as transference of energy. As I have already said, that is a mechanism that doesn’t generalise to all the cases. I also think Glennan doesn’t address the propagation side of things, what happens within an object between interactions when it stays the same.
This completes my discussion of contemporary attempts to say something about causal production while remaining true to the empiricist framework. Now it is time to address those authors that, like Glennan, think that causation must be seen as a continuous process rather than as a relation between events, but who also reject the constraints on speculation imposed by the empiricist tradition; those that talk of powers.