How we think about causation

In Couterfactuals, Control, and Causation: Why Knowledgeable People Get Blamed More, Elizabeth A. Gilbert and colleagues report findings from three psychology experiments. Here’s their abstract:

Legal and prescriptive theories of blame generally propose that judgments about an actor’s mental state (e.g., her knowledge or intent) should remain separate from judgments about whether the actor caused an outcome. Three experiments, however, show that, even in the absence of intent or immorality, actors who have knowledge relevant to a potential outcome will be rated more causal of that outcome than their ignorant counterparts, even when their actions were identical. Additional analysis revealed that this effect was mediated by counterfactual thinking—that is, by imagining ways the outcome could have been prevented. Specifically, when actors had knowledge, participants generated more counterfactuals about ways the outcome could have been different that the actor could control, which in turn increased causal assignment to the actor. These results are consistent with the Crediting Causality Model, but conflict with some legal and moral theories of blame.

Studies that shed light on how people think about such things as responsibility, blame, causation, control, and so forth are music to my ears. But I was unsettled by several aspects of this study, or at least by the way that the authors describe what they did, what they found, and what further implications they think their findings might have. In what follows I rant about three points in their abstract that caught my attention.

I hope that what I say below will entice others into reading Gilbert et al’s paper too.

Knowledge and Immorality, Knowledge and Intent, and Actions

Gilbert et al write “that, even in the absence of intent or immorality, actors who have knowledge relevant to a potential outcome will be rated more causal of that outcome than their ignorant counterparts, even when their actions were identical” (643). I see at least three problems here.

Firstly, given that we are asked to imagine that actors had knowledge relevant to potential outcomes – e.g. they knew that there was a risk of a bad outcome – to me it seems plain that it is not the case that there was an absence of intent or immorality (for a given definition of “intent” and “immorality”). I say more about this in the second paragraph of the next section.

Secondly, actions aren’t identical when agents know different things. The physical body movements might be identical, but actions are so much more than just that.

Thirdly, I find absolutely nothing surprising about these findings. Sure, on the surface, the questions asked were about causation: “Who or what would you say was the main cause of the destruction of the tulips?” (646). But going on the answers that the subjects apparently gave to this question, to me it seems so plain that what the subjects were actually reporting were their views about blame and responsibility not their views about causation.

Causation and Knowledge

Gilbert et al also write that “when actors had knowledge [of risks related to their actions, this] increased causal assignment [for the outcome] to the actor” (643).

As I said in the last sentence of the previous section, I see nothing surprising here as long as we interpret what the subjects said as a report of blame or responsibility rather than as a report of their views about causation. Our knowledge of risks is something that – when the risks are sufficiently high, and we can avoid those risks by doing something else – ought to affect what we do. We blame people more when they disregard risks of which they were aware (we call that “recklessness”), than how much we blame people who didn’t know but should have known of those risks (we call that “negligence”), than how much we blame people who non-culpably didn’t know and thus didn’t take those risks into account (I hazard a guess that we would either say that the person was not blameworthy, that their blame was attenuated, or if we attribute blame then this will be because we approach the given case from a strict liability perspective informed, perhaps, by various policy-related considerations).

Moral and Legal Blame

Finally, Gilbert et al conclude that “[t]he results … conflict with some legal and moral theories of blame” (643). Why say that? Well, on their account, within the criminal law “determining whether an actor caused an outcome is a separate step from considering an actor’s prior knowledge or mental state” (653), whereas what actually seems to happen is “that people do not really keep those ‘elements’ separate” (653).

From what I gather, the authors here are referring to the separate actus reus and mens rea conditions. However, here too I see at least three problems.

Firstly, actus reus is not just about causation. True, causation must have occurred, in the rough sense that outcomes must have eventuated (or, more colloquially, that sh*t must have gone down). But as far as actus reus is concerned, it is also crucial that the body movements that gave rise to those outcomes can be rightfully described as actions, and this requires that those body movements be the products of the right kinds of mental processes (the law, rather unhelpfully, talks of “voluntariness”). That’s precisely how and why the defense of automatism works. If all we cared about was causation in that simple sense that I gesture at above, then the defense of automatism wouldn’t work like that. Here’s another example. Suppose I’m drugged and shoved into a cannon, the cannon is fired, and my body smashes through someone’s window. Did I cause that window to shatter? I guess so, in a basic physical sense. My body certainly smashed through that window. But did I cause that window to shatter? Hell no. My body was used as a projectile. I wasn’t acting. There was no actus reus, or at least not my actus reus. When we remind ourselves of the difference between actus reus and causation in that minimal sense that Gilbert et al seem to have in mind, the claim that the results conflict with legal and moral theories of blame becomes a non sequitur.

Or perhaps – and this is my second point – what is revealed is just how important it is to be clear about what we mean by “causation”. The philosophical and legal literatures bursting at the seams with intricate and varied ways of understanding “caused” and its cognates, with discussions of whether and how omissions can be causes, with different criteria for establishing whether causation indeed occurred, and so forth. Against this background, what struck me about Gilbert et al’s paper – I became acutely aware of this as I wrote the previous paragraph – is that, on reflection, I really do not know what the authors actually had in mind when they used the word “causation” and its cognates. I anticipate that this could be because I’m a philosopher, not a psychologist, and hence I need to educate myself better to properly understand what they mean when they use this term. However, if that’s right then, then my worry will now be that I feel a lack of confidence that the subjects in the experiments understood “causation” in the same way as the authors of this study, that they understood “causation” in the same way across the different examples that they were asked to reflect upon, and that all of the subjects understood “causation” in the same way. Like I said above, what I suspect the subjects were actually reporting when they were asked to report their views on causation is their views on blame and responsibility. Or at least if we interpret what they said in this way, then what they said will turn out to be a lot less surprising and in fact perfectly reasonable and understandable.

Thirdly, even if the experiments had shown that there is a difference between folk and legal thinking about blame – and I have expressed several doubts about this – I worry that some readers may jump to conclusions like “Wow, the law is out of touch with common sense!”, or perhaps even more explicitly normative claims about us needing to knock the law back into line with common sense. The authors don’t seem to make these mistakes, so this is not a criticism of their work, but I still note it here so that others won’t draw such mistaken conclusions after reading this paper.

Closing thoughts

Contrary to how it may seem, I enjoyed reading Gilbert et al’s paper, even though I found it frustrating for the reasons I cite above.

Perhaps some of my frustration stems from the fact that I do not share the authors’ view that “people act like lay scientists” when they assess causation involving human agents (644). I approach the topic of responsibility from a thoroughly normative angle, and I have similarly normative views about how people reason about causation too — at least when judgments about causation are made in a context that will later connect those judgments to attributions of blame, guilt, or responsibility, or to impositions of liability and punishment.

I hope, though, that my rant was not solely caused (pun intended) just by this difference in our approaches – lay scientist vs. normative approach to thinking about causation when human actions are involved – and that some of the things I said above elucidate the sort of finer-grained distinctions which I would like to see taken into consideration in future empirical studies of this topic.