How Moral Foundations Theory Succeeded in Building on Sand: A Response to Suhler and Churchland

Suppose you are an architect and you have recently completed a challenging project: designing and building a sturdy modern house on a sandy stretch of ground where several previous architects had failed. The shifting ground had cracked their one-piece rigid concrete foundations. You vowed not to repeat their mistakes, so you designed a novel foundational system that avoided the use of concrete altogether. You drove steel rods down into rockier soil, created five independent platforms to support five modular units, and then linked the units together with short flexible corridors. You left plenty of room for expansion—the modular design makes it easy for the homeowner to add additional units as needed.The initial reviews of your modular house are excellent, and other architects begin applying your technique, with good results.1 Imagine your trepidation, then, when a major architectural critic writes a review entitled “A foundation built on sand?”, in which she warns that your house will soon collapse and that your project is useful primarily as an object lesson in what not to do.You begin reading the review. It starts off with an extremely accurate summary of the design challenges you faced and of the innovative ways that you met those challenges. It praises you for having solved four of the major problems that doomed previous attempts to build on this sandy ground. (You are grateful for this praise.) So imagine your confusion as you continue to read and discover that your critic's three major complaints are as follows:(1) Your steel rods are not strong enough to support the house (when in fact the house is already standing).(2) There is no garage (which is true, but a strength of your design is that future owners can easily add whatever rooms or structures are needed).(3) You failed to extend your steel rods down to the center of the earth (which is true, but it is both impossible and unnecessary to do so).These three complaints are close analogues of the complaints Suhler and Churchland (2011) level against Moral Foundations Theory (MFT): (1) Our concepts of innateness and modularity are defective and cannot support the theory. (2) There are additional candidates for foundationhood. (3) We failed to link MFT to neuroscience and genetics. In this essay, we will speak for the team that designed MFT and oversees its ongoing testing and revision; the team includes Peter Ditto, Jesse Graham, Ravi Iyer, Sena Koleva, and Brian Nosek. We will first address Complaints 2 and 3, which are true statements that are not valid criticisms. We will then address Complaint 1, which is a more substantive charge.Second, both the theory's proposed number of moral foundations and its taxonomy of the moral domain appear contrived, ignoring equally good candidate foundations and the possibility of substantial intergroup differences in the foundations' contents (Suhler they want to see positive links to those three fields, including the identification of candidate genes and neural systems. Such positive linkage with developmental psychology is reasonable enough; as cultural psychologists, we set as one of our main design challenges the need to create a theory that would explain the divergent developmental paths taken by children in diverse cultures. (See Haidt Suhler and Churchland praise us on this point.)But genetics? One of the biggest news stories in science in the last few years has been that, despite the fact that just about everything is heritable (Turkheimer, 2000), there do not appear to be genes “for” traits. The human genome project failed to find genes or even sets of dozens of genes that account for more than a few percent of the variance in any target disease or trait. Even for physical height, which has a heritability of 0.9 and can be measured with nearly perfect accuracy, nobody can find a gene or a set of genes that explain why some people are taller than others (Turkheimer, in press). The most successful of these genome-wide association scans identified 27 genes that, when combined, explained just 3.7% of the variance in height (Gudbjartsson et al., 2008). What hope, then, is there for finding genes “for” reciprocity, loyalty, or authority? Of course, the genome codes for traits somehow or other, but nobody knows how. Yet, despite the disappointing news emerging from the human genome project, Suhler and Churchland claim that any scientist who proposes a nativist theory is now “expected” to identify genes that are at least associated with the innate content. This is equivalent to demanding that all new buildings must dig their foundations down to the center of the earth. It cannot be done today, it might be impossible in principle, and if it is required of all new nativist theories, then there will be no new nativist theories.The same problem applies to neuroscience, although not as starkly. We have always treated moral modules as functional modules, not as physical, anatomical, or neurobiological modules. We were attracted to modularity, with partial (not complete) encapsulation, because of our observations of moral dumbfounding. For example, when asked about an adult brother and sister who have sex once, using two forms of birth control, many participants condemn the action. When pressed to justify their condemnation, many subjects search for reasons, fail to find any, and then admit that they cannot justify their condemnation. Yet, they continue to maintain that the action was wrong and are sometimes puzzled by their own continued condemnation. These situations are analogous to optical illusions, such as the Muller–Lyer illusion: One line continues to look longer, even after you measure the two lines yourself. In both cases, the judgment is partially encapsulated; it is not fully revised by the acquisition of other relevant information.Suhler and Churchland's long section on neurobiology assumes that we are positing neurobiological modules—specific neural circuits that correspond to moral foundations or, at least, to the component operations that comprise moral judgment. But we are not, and we do not see how the phenomenon of moral dumbfounding (or any psychological phenomenon) can be negated (or declared “not consilient”) with any finding about neurons and circuits. It is just too low a level of analysis, at least until we have a neuroscience so complete that we can say how neural activity fully instantiates and constrains specific moral judgments. It is interesting that neuroanatomical circuits are often loopy. But does that mean that no knowledge can be partially encapsulated? Should we inform our dumbfounded participants that they cannot be dumbfounded because their neural circuits are too loopy to allow it? Likewise, it is interesting that neurons exhibit spontaneous activity. But how can that fact make MFT (or any theory of higher cognition) more or less plausible?In summary, Suhler and Churchland's third complaint is that we have made no effort to seek consilience with neuroscience and genetics. We agree with their claim but cannot see how this counts as a mark against MFT. If their “expectation” about the requirements for nativist theories were to become widespread, there would be no more nativist theories. And that, we suspect, is why they have proposed an impossibly high bar for nativist theories.Since the 1980s, there has been a slight correlation between geography and attitudes about nativism in the United States. The “East Pole” of this intellectual dimension has been located in the Northeast, particularly at and the of where about modularity, theory, and evolutionary psychology together to support a nativist on and (see The Pole” is in particularly at the of and the of where an in and brain has to a for more (see et al., Churchland is a That is good can be on both But if a of set an impossibly high bar for nativist nativist then that MFT does not that it cannot as a of MFT It is simply a of what example, Suhler and Churchland innateness claims have to evidence that the traits they target to the of the for that these traits were for in the of human few or no psychological traits are or to environmental and because it is to that a was Suhler and Churchland are us a idea and the of the of the and then will we that your theory is more than have been that by we mean in of We have that the is like a The genes the first into neural there be no genes “for” any specific modules or for any specific in the then the of the are by in some but in traits need not be in all cultures. For example, the of most for than sex is even if some are to a of described by long as there is some in of the we in it a previous (e.g., Haidt & Joseph, we have on of or as a way of the innate part of moral we are not by Suhler and Churchland's that our nativism to and traits. that just about you can from to is we are to say that most and traits the moral foundations and much to some on innate and we are too with our nativism or they are too on which you and Churchland also that our use of modularity is a to more than a of the or other We that we are not We have not in what is each Haidt, in will far more MFT is not a complete theory all of analysis, and we that, in it will But is the of a theory a to reject it or to develop and Churchland seem to have taken theory of modularity as the for what a We agree with them and with that this is so high that there are no modules in higher to and p. of modern of modularity have modern as the of modularity were So let us modules and look at what evolutionary psychologists mean when they about The is functional and functional is a of designed by The for example, is a the and its is to from in is of modules, each with a related to the specific of that it You cannot any in the system first its and its is in kinds of are by different systems. mechanisms with are of human and these should be identified as the of & p. this to MFT to this The moral includes at least five sets of modules that are to handle related to social (1) (2) fairness (3) and This claim need some over in the number and of these modules, but it is a It offers a with Suhler and Churchland and all other theories that to explain moral as a of or developmental such as social modules might or might not out to be with but they should be evaluated and by on how people process theory the data a modular theory or a has the right to claim to be the and then to its to its It is a approach the of moral and Churchland are correct that “the of moral to the five does not the of modules. But how would they explain in the of of purity and (see Haidt, would they explain the in the of of the in the United as In this children who are of the sex or who are low in become in the must be treated with a When you find that are and that seem to even in the of from is the with it that the did not from social they the of modules, which make it easy to and related to and close with this from a who built theory of morality in the most a priori and principled Yet, even an moral at such an use so of is a of to and is in the to morality upon thinking of it. the of it is so that even such a by its is a of it is not so easy to a of the of that course, the fact that few of us that there is no moral of that is to environmental But MFT assumes that nothing is or to MFT that like the of had a purity that to about and other It made it easy for to children that is and to link to disgust the of Even was to think about morality by on because was of moral we are grateful to Suhler and Churchland for the extremely accurate of MFT that they offered in 2 of their and for the four of praise that they offered at the of that We that their three complaints in are not valid complaints about two of them are as complaints by about nativist theories in MFT has from its an attempt to bridge the nativism of evolutionary psychology with the of cultural admit that we built on is to and we are of for having solved the design challenges of Our house is not and we Suhler and Churchland's about where more is should be to Haidt, of or