Monday, November 12, 2018

Moral Twin Earth

I’ve been working on theories of reference for moral concepts of late with collaborators, particularly on Moral Twin Earth (MTE). I’ll present some results I found on this hypothetical from Terrance Horgan and Mark Timmons (H&T). Then I’ll explore what implications this may have.

First, regarding semantic internalism/externalism: Internalists claim that the reference of a concept is solely determined by internal mental states in one’s head. E.g., descriptivism generally claims that the referent of a concept is that entity that satisfies the mental descriptions associated with the concept in question. Externalists maintain that the reference of a concept is at least in significant part dependent on properties external to one’s mind. E.g., the causal theory from Richard Boyd claims that the referent of a moral concept is a natural property, such as human well-being, which causally regulates our moral decision-making.

H&T ask us to assume that there is a world called MTE that is nearly identical to our own world except that our moral term ‘good’ is causally regulated by natural properties whose essence is captured by consequentialist properties while the orthographically identical term ‘good’ for those on MTE is causally influenced by natural properties whose essence is captured by deontological ones. Both those on Earth and MTE internally understand their respective moral terms to be related to the evaluation of actions, persons, and institutions. What goes on inside the moral minds of you and your twin are quite similar, but they differ at precisely the points we would expect them to given the fact that they are being causally regulated by the different consequentialist and deontological properties, respectively.

H&T conclude that our intuitions in this case are that when there are antithetical judgments between you and your doppelganger on a particular moral issue, both of your moral concepts really share the same reference, despite the different causal relations, such that both of you are not talking past one another in this debate. Therefore, this situation can be characterized as one of a difference in some moral beliefs rather than one of a difference in reference given that on Earth and MTE your moral concepts respectively are understood to play a role in governing behavior. H&T want to claim that when dealing with moral concepts, reference is determined internally by our cognitive understanding of the action-guiding role of moral concepts. Reference is not determined externally. For, if this were the case, then our intuitions should conclude that those on both planets have disparate referents for their moral concepts given that such concepts are causally influenced by different properties. However, this is not the case.

I ran experiments testing intuitions on MTE, winnowing the hypothetical to a 10th grade reading level on the highly reliable Flesch-Kincaid readability measurement. I am still running studies trying to replicate the results in various ways. There are no previous experiments on the internalism/externalism debate in moral semantics. I ran tests on 116 lay subjects from the U.S. with an average age of 45 spanning from 18-61 years of age and 102 lay participants from Singapore with an average age of 34 spanning from 19-73 years of age. All had a high school diploma or higher. I also conducted the test on 53 “experts” with PhDs in philosophy.

Here are the results:

Philosophers:  40%  Internalism,  41%  Externalism, 19% Other
Singapore:  38% Internalism ,  62%  Externalism ,  0% Other 
USA:  28% Internalism , 70% Externalism,  2% Other

Philosophers appear generally split on MTE. However, most of the folk, East and West, are externalists for MTE. Against H&T, even amongst only philosophers, we do not see that a majority of intuitions are internalist ones. Rather, when combining all the populations East, West, and philosophers, it appears that most have externalist intuitions on MTE.

Furthermore, we see that there is substantial intracultural variation. The minority subpopulations in both folk groups are rather sizable, and moral semantic intuitions on MTE lack the kind of widespread consensus that is found in other staunch a priori domains of inquiry, such as in math and logic. As there is an absence of sufficient consensus within a culture, an internalism seemingly applies to both minority folk subgroups in the East and West, although an externalism purportedly applies for most in both East and West groups. Different theories of reference may apply to different people even within the same culture. This also seems to hold for philosophers on MTE too. It appears that if theories of reference are adjudicated based on intuitions, as is commonly held in the philosophy of language, there’s a moral semantic pluralism, where different theories of reference apply to different people. This runs not only contrary to H&T, but it flies in the face of all current moral semantic theories in that they are either exclusively internalist or externalist.

One may counter that only expert philosophers’ intuitions should be relied upon. However, even if we did this, there will still be a pluralism. One may reject the pluralism option because there needs to be some kind of argument establishing that the semantic intuitions provide any evidence about reference such that the disparate intuitions generate a pluralism. As there is no such argument, we should not trust semantic intuitions at all, and philosophers should abandon arguing for a theory of reference. However, experiments have shown that people’s intuitions in semantics on theories of reference coincide with how they actually use specific terms in sentences (Machery et al. 2009). This provides sufficient empirical data that semantic intuitions do provide evidence about the reference of our concepts. An objection is that there is merely a disagreement in non-semantic facts, such as regarding the storyline descriptions in the MTE hypotheticals. Once we get clear on and in agreement with the non-semantic facts, our semantic intuitions will align such that there will be a sufficient consensus. However, given the readability measurement, we can inductively infer that most of the participants, including experts, are clear on the non-semantic facts of MTE and comprehend the scenario.

John Park
Philosophy Department
Sacramento State


  1. John, thanks for this.

    I can’t quite tell what aspect of H&T’s views you are testing for in your subjects. Is it the claim that differing moral judgments express real disagreements, or their claim that this implies that internalism is false? Mean to say that if you are directly testing people’s intuitions about internalism and externalism, then it could just be that people have inconsistent sets of beliefs regarding these two questions. That is, H&T may be right that people think these are real disagreements, right that this implies externalism, but wrong to think that people draw this conclusion. We have loads of empirical data that people are comfortable holding completely inconsistent beliefs.

    I have never understood what people who develop traditional theories of reference think they are up to, and I have always wished they would stop. But, while I endorse your naturalistic tendencies, I also don’t understand what you x-phi people think you are up to either. Who cares what people’s intuitions are on this stuf anyway? To me the only real reason to do this is to establish that philosophical intuitions differ from ordinary intuitions and to force SAE philosophers to come clean about why they think their own intuitions should be privileged in this regard. But this has already been done and doing it in other slightly different areas just feels like kicking a dead mule.

    Is there a fact of the matter about whether internalism or externalism is true beyond what some people believe? If so, why does finding out what they believe even matter? If not, then what exactly is the importance of what we are finding out?

  2. Hi Randy, thanks for these comments. The question posed to participants is directly about internalism/externalism and whether the Earthling and Twin Earthling are talking about the same thing or different things with their moral terms. It is possible that there could be inconsistent judgments within peoples' sets of beliefs as you've outlined. One interesting thing is that there are many studies showing that most people believe that when there's cross-cultural moral debate and controversy, there is no genuine disagreement (Kelly et al. 2007, Sarkissian et al. 2011, Polzler & Wright Forthcoming). Most believe that both can be correct even though they give apparently conflicting judgments on things like murdering your own child just because the child is ugly. One benefit of my study, if replicated, is that it can help unify these phenomena in that there's a general consistency between majority externalist moral semantic intuitions on MTE and metaethical beliefs on moral disagreement.

    My results do suggest that H&T internalist intuitions are not pervasive as they say they are. Adjudicating between internalism/externalism in the philosophy of language has primarily come about through intuitions. However, as you raise: why should we care about peoples’ semantic intuitions? As mentioned, there is empirical data supporting that people’s semantic intuitions do tend to align with how they actually use words. This lends some credence to semantic intuitions. Peoples’ words at many times do refer to things in the world, and semantic intuitions appear to be capturing this in a consistent way. Moreover, there’s support that our own thoughts about reference carry at least prima facie support about what our terms actually refer to. It’s arbitrary what words, sounds, or scribble marks we use to represent a particular extension or else there likely would be a universal language (Locke). If by ‘bank’ you say that you’re referring to the slope by the side of a river rather than a financial institution, that provides initial support that when you use the word, ‘bank,’ the slope is in fact the referent of your term. In fact, because of this, a prominent and at many times overriding desiderata for theories of reference is that they coincide with what people actually believe is the referent of their term.

    It does look like at least for semantic intuitions (I say nothing here about intuitions in other areas like the compatibilism/incompatibilism debate), there is importance to them given the peculiarities of the subject matter, and they generally are linked to how our words actually refer. I use collected data on them to show that H&T’s intuitions are not predominant even among philosophers. This is novel in that it hasn’t been done yet on MTE and moral semantics. Moreover, another importance of the data is that I use them in the hopes of establishing a new pluralistic moral semantic theory.