My current opinion is that, cognitively, there isn’t much difference between regular non-moral decision making processes and moral ones. The latter adds a feeling of universalisability and non-negotiability to the judgement, but otherwise they operate in much the same way. (In fact, I think one of the jobs of a UMG is to determine which scenes are moral or non-moral before it applies things like the principle of double effect – so we might have a universal perceptual grammar and a UMG.)

As for low-functioning autists, the boxology I proposed might also explain their behaviour. The UMG functions differently in them, for example attributing intentionality differently due to their different Theory of Mind, and their emotional responses are low. As a result, the resulting moral sentiment is often absent, aberrant or weak (psychopathy might be another related example). However, moral reasoning and the ability to reflect on learned moral principles isn’t necessarily degraded. In fact, outside of moral issues, individuals with ASD often employ effortful reasoning to compensate for diminished or absent intuitions, such as in social situations.

Still, I am sympathetic to the sentimentalist approach, I just think we’re a long way from understanding the details of the actual cognitive processes that underlie it and its relation, if any, to a moral grammar.

I have some more thoughts on the UMG, written about a year ago and due for an update, on my site, here.

Erica Roedder and Gilbert Harman, “Grammar,” to appear in Empirical Moral Psychology, edited by Doris, Nichols, and Stich, Oxford University Press.

Mallon, R. (2008). Reviving Rawls Inside and Out. Moral Psychology, Volume 2: The Cognitive Science of Morality: Intuition and Diversity. W. Sinnott-Armstrong. Cambridge, MA, MIT Press. 2: 145-155.

Hauser, Mark (2006), Moral Minds. Harper Collins, New York.

F.M. Kamm, Intricate Ethics: Rights, Responsibilities, and Permissible Harm, Oxford University Press, 2007

Jesse Prinz, Resisting the Linguistic Analogy: A Commentary on Hauser, Young, and Cushman (In W. Sinnott-Armstrong (Ed.) (2008). Moral Psychology, Volume 2: The Cognitive Science of Morality: Intuition and Diversity. MIT Press.)

Bryce, very interesting stuff. I’ve a lot of reading to do on the evolution of cognition before I’m properly armed to evaluate the costs. But at the very least, UMG is committed to the potentially parametrized principles being hardwired (never mind exactly how). If it isn’t, then it gets hard to see what is new about it, except Mikhail’s Chomsky-inspired way of representing the principles. And however the computation is realized in the brain, the innatist has the burden of showing that having hardwired principles is adaptive. If the sentimentalist can explain the same pattern of judgments as a byproduct of other capacities and dispositions, it does seem cheaper to me.

So the issue is the source of the principles. An additional advantage of sentimentalism is that it provides a recipe for generating new principles for new situations, and also allows for particularist judgments. To be sure, some variation is predicted by the parameters idea, but leaning too hard on that trivializes the notion of there being universal principles (as Jesse notes).

I’m curious to see your data on new trolley variations. If and when you have a suitable draft, email it to me at a.m.kauppinen (that’s what the Dutch call me) at uva.nl.

The way in which you appeal to reason within sentimentalism seems to me to be calling on some fairly sophisticated and ‘costly’ cognitive mechanisms that may be unnecessary to explain the moral judgement. Such as:

and consider what the natural reaction of a reasonable person, aware of the agent’s intentions etc. and the circumstances, would be in their place to the proposed action and its salient alternatives

That sounds like it could escalate in to a incredibly taxing process depending on the circumstances and contingencies at play. Perhaps it can be done unconsciously, but is there evidence that this is the process we use rather than the UMG model?

I thought one of the great strengths of sentimentalism is that the emotions are heuristics that arise as a result of quick and dirty rules in the UMG and which motivate behaviour, all without calling on our lumbering, carbohydrate hungry reasoning centres unless there’s a conflict to sentiments to be resolved.

So, to revise your boxology:

Stimulus -> Action analysis -> Moral sentiment -> Moral reasoning -> Moral judgment

With the moral reasoning step optional depending on the clarity of the moral sentiment. Thus, moral sentiments are still central, but the heavy lifting is done by the quick and dirty rules in the UMG rather than more complex unconscious abstract reasoning about an idealised reasonable person.

Nice to see you dabbling in this stuff!

You say of UMG that it is “an expensive hypothesis, in evolutionary terms: even if we accept massive modularity, evolution doesn’t throw up dedicated modules when the same job can be done by existing means.”

This claim sounds a bit odd to my ears, especially since the architectural model suggested by Hauser is a minimalist model that requires nothing more than the evolution of an interface between existing systems (that could have been selected for other means). Perhaps something is lost across the translation from the cognitive science to philosophy vis-a-vis the concept of modularity; but it strikes me that the intuitively plausible version of your claim suggests that it would be evolutionarily costly to construct a new system from scratch (on this point, Herbert Simon nails it brilliantly in his discussion of the architecture of complex systems–it’s in Sciences of the artificial). But, I see no reason why the moral grammarian should, or needs to be committed to anything that is nearly this strong. What must be the case, from this perspective, is that moral cognition is relatively encapsulated and relatively domain specific–that is, there must be a dedicated architecture that is triggered by ‘morally significant’ inputs, reflexively executes computations on the basis of some sort of computational rules, and then spits out relatively stable stable patterns of judgments. Of course, modularity (like moral grammar) comes in varyingstrengths; but, if I were you I would have a look at the recent Hauser paper (Nature, 2009) where he discusses this issue. (If I’m remembering correctly, I think that there is a paper with Liane Young on interfaces and what not…maybe a year or two ago but I don’t remember where off the top of my head)

Two other quick things: 1) I think that once one sees what the architecture is supposed to look like, Prinz’s argument against UMG starts to look much less troubling as well; and 2) I have a hard time seeing how the boxology that you suggest is much different from a moral grammarian’s view–it seems to me that the question might be where to draw the boundaries around the moral system, but i’m not sure.

Oh yeah, and I think I have data on a case like your comeuppance case, but I have to think about whether they really are as structurally similar as they now seem to me…

Nice read…