Carnapian explication provides a useful model for conceptual analysis, and it may further benefit from recent developments in experimental philosophy (x-phi). Shepherd & Justus (2015) give an account of how this is possible given the suspect nature of intuitions as a guide for analysis. The authors note an issue with x-phi guided explication: what normative role can x-phi (which seems purely descriptive) play in explication? Their answer to this question is formal epistemology, but I argue another avenue is also plausible for retaining x-phi’s normative import. In this paper, I build off of Shepherd & Justus’s account of how x-phi can be tethered to Carnapian explication, and I largely accept it as the best account of incorporating x-phi into a model for conceptual analysis. Where I deviate from the authors’ account is in highlighting the relationship between x-phi and moral philosophy. There are nuances in the methodology in moral philosophy which can be mapped onto the authors’ account of Carnapian explication, but these nuances must be made clearer. That is my primary goal in this paper; however, I also want to argue that x-phi can have normative import in moral philosophy in ways (contrary to what the authors suggest) that are not trivial or obvious. In effect, I argue that the principle that “ought implies can” provides x-phi with normative “bite” on two fronts: it creates an epistemic norm, and this has consequent impact on moral norms. I then demonstrate this through concrete examples of x-phi informed moral philosophy.

             Shepherd & Justus (2015) provide an account of how Carnapian explication can clarify the philosophical import of x-phi. X-phi’s supporting role in evaluating conceptual content has implications in the methodological approaches to philosophical issues. If we are already committed to a specific methodology that is inclusive of x-phi, we end up tackling philosophical issues is a particular way and this is not always obvious. The guiding principle behind x-phi is to support philosophical arguments that evoke empirical matters (e.g. “most people think that p,” “men think p and women think not-p,” “western cultures think p and eastern cultures think q…”). Shepherd & Justus take the traditional view on conceptual analysis to place “principal weight on intuitive judgments across possible scenarios.” (Shepherd & Justus, 2015, p. 382) For instance, when we analyze the concept “justice” (by asking questions like, “What is justice?”) and we reject certain conceptualizations (answers like, “Justice is nothing but the advantage of the stronger”), we appeal to our mere intuitions. This intuition might be okay for a negative methodology (namely, rejecting certain conceptualizations – although, some cast doubt on even this), but it seems problematic when we want a positive methodology of how conceptual content could and should be identified and evaluated. (Shepherd & Justus, 2015, p. 382) The authors suggest the positive methodology to be Carnapian explication. 

             Carnapian explication can be described as a process of moving from a clear explicandum to an adequate explicatum – here, an adequate explicatum would entail the following four criteria: similarity to the explicandum, exactness, fruitfulness, and simplicity. This is not a strict, rigid, formal activity: we don’t move to the exact concept at the end but just a better, more precise concept. For instance, “the concept of Piscis is meant to reveal properties of the animals in question that the concept of Fish would fail to deliver; likewise for other explications.” (Dutilh Novaes & Reck, 2017, p.206) Dutilh Novaes & Reck (2017) further point to how formalism provides a “cognitive boost” when considering flaws in our intuition. Again, “formalism” includes across what we mean when we talk about x-phi,  and Shepherd & Justice have a similar idea in mind of how to amalgamate x-phi and Carnapian explication.

             It is important to note that primitive commitments (theoretic aims, prior assumptions, constraints, etc.) to a particular methodology dictate the value of further methodological commitments. In this case, if we are committed to Carnapian explication, there is an issue of what role x-phi plays, or if it plays any useful role at all. Given our prior assumptions to the general reliability of intuition for at least a negative methodology (i.e. rejecting conceptualizations on the basis of intuition on other similar scenarios), we can at least establish the value of x-phi for identifying conceptual content. However, a positive methodology suggests something further about what conceptual content should or ought to have. The question can be put as such: “How can empirical data, which describe concepts, play any direct role in the normative evaluation and determination of conceptual content usually thought to be the proper (and primary) purview of philosophy?” (Shepherd & Justus, 2015, p. 383) X-phi seems to conflate the two different kinds of theorizing: one that is essentially descriptive (empirical concerns of what “concepts we actually have”) and one that is essentially normative (philosophical concerns of what “concepts we should have”). (Shepherd & Justus, 2015, p. 384) Take, for instance, the concept of table salt: we might identify the conceptual content as “that white, powdery stuff used to season food;” then, we might evaluate it and say “no, that doesn’t fit, it’s not the right concept” for whatever context or purpose – maybe “NaCl” is better. A lot of this conceptual analysis relies on intuition, and like Dutilh Novaes & Reck suggest through illustrations of cognitive biases and fallacious intuition, we can cast doubts on the epistemic import of intuition.

             The import and significance of x-phi – or how and where it is going to fit in our philosophical methodology – again comes down to “the proper role of intuition in identifying and evaluating conceptual content.” (Shepherd & Justus, 2015, p. 384) The so-called “positive program” x-phi camp gives intuition a “central role,” whereas the “negative program” x-phi camp “downplays its significance.” (Shepherd & Justus, 2015, p. 384) The negative camp cite empirical studies which cast doubt on the reliability of intuition (e.g. Knobe effect, cognitive biases, fallacious reasoning). The positive camp resists this by first “developing a theory explaining the psychological processes responsible for intuitions about C,” and then using the theory “to inform judgments about their [intuition’s] epistemic significance.” (Shepherd & Justus, 2015, p. 385) These divides center on the issue of the reliability of intuition.

             Shepherd & Justus suggest that a plausible positive program for x-phi is “explication preparation.” The authors begin by describing Carnapian explication as replacing or transforming “a concept (the explicandum) typically drawn from ‘everyday language or…a previous stage in the development of language’ (Carnap 1950, 3) [sic] into another concept (the explicatum) guided by four desiderata: retain similarity of conceptual content with the explicandum, and increase precision, fruitfulness, and simplicity, the last being subordinate to the others.” (Shepherd & Justus, 2015, p. 388) The motivation here is to match scientific methodology for philosophical pursuits. The idea is that x-phi can assist in explication in its desiderata of retaining similarity; that is, we may address the problem of discerning what content “merits attempted preservation” and what content “should be abandoned” in the explicanda. (Shepherd & Justus, 2015, p. 389) The authors cite examples of “explicandum clarification” in x-phi: clarifying “regions of vagueness in extensions and intensions of concepts,” “conceptual pluralism underlying a notion,” “sources of bias that influence intuitions,” “dependence relationships with other concepts.” (Shepherd & Justus, 2015, p. 390-391) The force of these examples shows that x-phi has a “positive philosophical payoff independent of contentious debates about intuition’s evidential status.” (Shepherd & Justus, 2015, p. 391) This positive program, independent of debates about intuition, is the important takeaway from the authors’ picture.

             Shepherd & Justus then seek to establish how x-phi assisted explication can apply to concepts with normative content. The authors move in the direction of formal epistemology and mostly write-off the avenue of moral philosophy. They write: “But normative concepts serve normative theories and it is entirely unclear they answer to empirical evaluation beyond perhaps, the familiar counsel that ought implies can.” (Shepherd & Justus, 2015, p. 391) I think this point deserves a closer look. Carnapian explication in moral philosophy works much the same as in other areas of philosophy; in essence, would take a moral concept as an explicandum and refine it through the four desiderata to end up

with the explicatum. Again, methodological questions rise of how we should or ought to think about morality. I think, however, there are novel ways to derive the normative import of x-phi if we explore its application in the avenue of moral philosophy.

             Before moving forward, let me make the connection between x-phi and moral philosophy a bit more explicit. Questions in methodology fall within the purview of metaphilosophy: in this case, the specific thread around questions of what the aim of philosophy is and how it should be done. Metaethics does exactly this in moral philosophy. Metaethics, in exploring how ethics should be done, looks at more basic questions of moral ontology, moral semantics, and moral epistemology. One particular stream of metaethics, one closely tethered to x-phi, looks at moral psychology: namely, looking at what the moral mind looks like, what happens when we make moral judgments, and how moral motivation works. (Miller, 2003, p. 2) Much of this is a descriptive enterprise in that we look at what moral powers we have — for instance, we might investigate the relationship between emotions and moral judgments, and what sort of role deliberation or rational faculties have in our moral judgments. This sort of descriptive enterprise can be approached naturalistically through scientific disciplines like experimental psychology. Here, philosophical arguments for claims about moral psychology appeals to empirical premises which can be informed by empirical means. This is generally how x-phi claims to assist moral philosophy, but the question remains of what the exact normative payoffs are.

             Still, an issue with x-phi guided moral philosophy is captured by the spirit of G.E. Moore’s naturalistic fallacy. It is important to address these worries before any claims about the significance of x-phi can get off the ground. Moore’s original worry targets “the view that moral properties are identical or reducible to natural properties as a matter of definitional or conceptual fact.” (Miller, 2003, p.12) He pushes this worry with his “classical open-question argument.” (Miller, 2003, p. 13) In brief, even if we say that “good” is synonymous with a naturalistic predicate (say, “pleasure”), it remains an open question whether the naturalistic predicate is “good” (“Is pursuing pleasure good?” is still a question we can ask without it being non-trivial or tautological). (Miller, 2003, p.14) I take the spirit of this worry to be that naturalistic approaches to moral questions are inappropriate and misguided.

             How can we defend the value of x-phi guided moral philosophy? One of the underlying assumptions in the open-question argument is that any conceptual analyses on moral terms (like “good”) are ultimately bound to fail. Moore’s idea of analysis is that “the analysis of P in terms of P* can be correct only if it is completely uninformative and uninteresting.” (Miller, 2003, p. 16) In other words, moral concepts cannot be analyzed in terms of naturalistic concepts. The stronger claim is that not even moral concepts can be analyzed (in the proper sense) by reference to other moral concepts. This becomes an a fortiori against x-phi: not only can we not establish that x-phi can inform philosophical conceptual analysis of moral terms, but we cannot even establish conceptual analysis in general. Moore assumes that “it is impossible for a conceptual analysis to be true but informative and interesting.” (Miller, 2003, p. 16) This “paradox of analysis” (or, at least, a weaker version of it) is typically addressed by a distinction between propositional knowledge (“that”) and knowledge of an ability (“how”). (Miller, 2003, p. 16-17) For instance, I might know how to drive a car without having all the propositional knowledge involved that correctly describes the process. (Miller, 2003, p. 17) In short, there is something suspect about the “paradox of analysis” insofar as jettisoning all sorts of conceptual analysis altogether.

             But what about the more direct concern that x-phi cannot inform philosophical conceptual analysis of moral terms? I think this worry is more general and takes some extrapolating from Moore’s naturalistic fallacy. Some (like Darwall, Gibbard, and Railton) might respond with the following: “Sure, I have left an explanatory obligation undischarged: I have provided no explanation of why otherwise competent speakers do not find it natural to guide their practice by means of my proposed analysis. But if, as you claim, competent speakers do not come to find it natural to guide their practice by means of the naturalistic analysis, you need to provide some explanation of why that is so. And you cannot just say ‘because ‘good’ and ‘N’ are analytically inequivalent’. That is no explanation at all. [sic]” (Miller, 2003, p. 24) I think this is a powerful reply. It is not enough to simply point at faults in a theory without coming up with a more plausible replacement, especially in discussions of methodological frameworks. The naturalistic account claims only to provide an abductive account (or an inference to the best explanation).

             Before tackling the issue of normativity and x-phi, it is useful to distinguish how we are using the term “normativity.” Generally, normativity is concerned with reasons to act in certain ways, but we might use them equivocally if we do not specify the aims or ends. In one sense, there is a methodological way of how we “should” or “ought to” conduct philosophical investigations; here, I think, the end is epistemic or directed at getting at the best answers. The other sense is directed at moral aims in that it deals with how we adhere to certain dictates for cooperation, a good life, and so forth. So, for instance, “You ought to avoid cognitive errors in reasoning,” is normative in the first sense; whereas, something like, “You ought to avoiding murdering innocent people,” is normative in the second sense.[1] We might conflate them by (say) establishing some prescription to be good epistemic agents, but I want to clarify that this bridge is not obvious or uncontroversial.

             Let us quickly take stock before deviating from Shepherd & Justus. Carnapian explication is a useful positive methodology for arriving at conceptual content, and x-phi latches onto this methodology in the pre-explication phase by clarifying the explicandum. The normativity of x-phi is still unclear: how does x-phi do more than clarify and describe what concepts we already have? Can x-phi have a role in what concepts we should or ought to have? Shepherd & Justus argue for x-phi’s normative bite in the direction of formal epistemology, but I think they overlook (or at least underemphasize) the point that x-phi can have extraordinary normative force in the domain of ethics. The rest of this paper will explore this avenue.

              “Ought implies can” means that normative requirements cannot apply to agents who cannot comply with it. (Miller, 2003, p. 123-124) Just as I cannot be required to sprout wings and fly, I cannot be required to do something morally that goes against my moral psychology. This is one way a descriptive enterprise can have normative impact. Speaking generally, a descriptive account of moral psychology can describe constraints which would inform our normative theories by carving out a domain of suitable normative prescriptions. X-phi provides just such descriptive account of moral psychology. When x-phi delivers a descriptive account of our moral psychology, it has normative impact in the epistemic sense (more specifically, metaethical theorizing) because it demarcates the domain of theorizing. For instance, a descriptive account of humans being flightless mammals would rule out (through an epistemic norm) the aerodynamics of humans. This echoes Carnapian explication — namely, we increase precision by cutting out the irrelevant (e.g. the aerodynamics of humans), retain the similarity in conceptual content (e.g. that humans are flightness mammals), we develop a fruitful norm (e.g. that epistemic norm), and we end with more-or-less a simple theory.

             In addition, this epistemic normativity (qua metaethics) has downstream effects on the moral sense of normativity (qua normative ethics). If there are limits or constraints on our moral capacities, then we would have different subjective reasons for action or we would adjust our behavior accordingly. This point is subtle, so let me try to illustrate it: in continuing the flight analogy, if I knew I could not fly (imagine it was novel information), it could not only guide my theorizing of humans, but it would guide my practical actions as well; that is, I would not (say) play a practical joke of shoving people off cliffs or develop my back muscles for the aims of flying greater distances. X-phi, insofar as moral philosophy, seems to have normative import on two fronts

             Perhaps this point can be made more concrete by tracing how this works in a case example. Joshua Greene proposes a “dual-process theory of moral judgment” which argues that deontological judgments are driven by “automatic emotional responses” and utilitarian judgments are driven by “controlled cognitive processes.” (Greene, 2009, p. 4) This model is based on empirical data collected by subjects exposed to the Trolley Problem. (Greene, 2009, p. 4) In short, Greene and colleagues collected functional magnetic resonance imaging and reaction time data of subjects responding to two scenarios: first, a scenario where a runaway trolley headed to five people on one side of the track and one may choose to switch the trolley to another track to kill only one person; second, the same runaway trolley headed to kill five, but this time one is on a bridge and choose to push a person off to their death in order to stop the trolley from killing the five. Their analysis of the data results in a model of the moral mind that conceives moral judgments as an emotional response that adheres to basic rules followed by a slower deliberation adhering to rules about the best consequence. Here, we have a case of x-phi informed moral philosophy: if Greene’s claims are right, our moral judgments work in the way he describes them. What is the normative import of this description? Methodologically speaking, it seems to generally support a naturalistic approach to ethics (more on this shortly). However, one of the more specific takeaways from Greene’s model is that it undermines deontological approaches to ethics. If our deliberative moral capacities are always oriented to the best consequence, and whatever rule-like behavior is reduced to mere emotions, then our domain for (normative) ethical theorizing precludes most versions of deontological theories. We see the principle that “ought implies can” does a significant amount of work establishing the normative import of the x-phi’s findings. If we are the type of beings that cannot adhere to a deontological moral framework, then we cannot be expected to follow such a morality. Moreover, our moral theorizing is constrained insofar as avoiding theories that evoke a deontological moral framework. Note that the “ought implies can” principle, here, is being used as a heuristic to interpret the normative import of the (descriptive) empirical data. This principle does much more legwork in ethics because it provides normative bite in an epistemic and a moral sense.

             Let us look at another case: Jonathan Haidt’s “social intuitionist model.” (Haidt, 2001, p. 815) This model is highly revisionistic of traditional approaches presented by armchair philosophers; specifically, the “rationalist models, in which moral judgment is thought to be caused by moral reasoning.” (Haidt, 2001, p. 814) This revisionist approach further highlights the normative import of x-phi – that is, it claims to bring an entirely novel evaluative standard and tears down much of the existing ones. In brief, Haidt suggests that (given a large body of empirical evidence) moral judgments are “quick, automatic evaluations (intuitions),” and what we take to be slow, deliberative moral judgments are a mere “post hoc construction.” (Haidt, 2001, p. 823) On Haidt’s model, asking somebody to put their gut reactions (“intuitions”) aside and engage only in deliberative reasoning to form moral judgments is just like asking somebody to sprout wings and fly. Again, this guides our methodology for approaching ethical questions, but much more radically. Note that prior to this x-phi driven model, the paradigm of the moral mind was “rationalistic;” as such, by jettisoning this rationalistic model, we seem to adopt a new methodology that followed with the x-phi findings. The success of this model seems to reinforce a general naturalistic methodology insofar as approaching topics in ethics and moral philosophy.

             Some may question the normative import of x-phi insofar as methodology because it assumes naturalism. It seems odd that accepting x-phi also might mean accepting a comprehensive methodology, complete with its own set of metaphysical and semantic commitments. So, one may argue, the x-phi’s data is not a further reason to take up a naturalistic methodology; rather, one must have such normative commitments beforehand, and x-phi then brings nothing novel to the table. Here, we must take a closer look at the relationship between accepting the significance of x-phi and our commitments to a particular methodology.

             In response, Peter Railton’s suggests that a naturalistic methodology[2] does not have to be accepted a priori as some sort of inextricable bundle deal when we accept the significance of empirical data from x-phi for our moral concepts. We can accept the significance of x-phi very loosely: for instance, if x-phi defines moral judgments in a certain way, that definition “must be vindicated by the capacity of the reforming definition to contribute to the a posteriori explanation of features of our experience.” (Miller, 2003, p. 180) In other words, whether or not the new definition supplied by x-phi is acceptable or not is an a posteriori matter, and we are consequently not committed a priori to the naturalism tied to x-phi. We are not committed ipso facto to the naturalistic methodology, but the plausibility of the findings from x-phi might be an added reason to commit oneself to a naturalistic methodology.[3] As it follows, the naturalism accompanying x-phi is “non-hegemonic.” (Miller, 2003, p. 182)

             By and large, my aims for this paper are not ambitious. I have sketched Shepherd & Justus’ account which I take as the foundation to build my argument. I have suggested that x-phi can have normative import in moral philosophy given our methodological framework of Carnapian explication. X-phi assists in clarifying content as well as providing evaluative norms by setting the boundaries on our (meta)ethical theorizing. I have provided salient examples to trace the normative import of x-phi through models of the moral mind sketched by Greene and Haidt. I then dealt with a worry about the naturalistic methodology which seem bound to x-phi. In the end, I want to push for the significance of x-phi in moral philosophy; that is, x-phi provides both descriptive and normative payoffs.

Sources

Dutilh Novaes, C. & Reck, E. Synthese (2017) 194: 195. doi:10.1007/s11229-015-0816-z

Greene, Joshua D. (2009). Dual-process morality and the personal/impersonal distinction: A reply to McGuire, Langdon, Coltheart, and Mackenzie. Journal of Experimental Social Psychology 45(3): 581-584.

Haidt, Jonathan (2001). The Emotional Dog and Its Rational Tail. Psychological Review 108 (4):814-834. 

Miller, Alexander (2003).An Introduction to Contemporary Meta-Ethics. Cambridge: Polity Press.

Shepard, Joshua & Justus, James (2014). X-Phi and Carnapian Explication. Erkenntnis 80 (2):381–402.

---

[1] Of course, this is non-exhaustive. There might be other senses of normativity, such as norms of aesthetics, manners, and sort forth.

[2] Railton distinguishing two types of naturalism: first, methodological naturalism, which he takes to be a general “explanatory approach to an area of human practice;” and second, substantive naturalism, which he takes to be semantic interpretation of the concepts in some area of practice or discourse in terms of properties or relations that would ‘pull their weight’ within empirical science. (1993b: 315) [sic]” (Miller, 2003, p. 178-179)

[3] Of course, it is entirely possible that the findings from x-phi are neutral on the matter.