The unity of moral attitudes: recipe semantics and credal exaptation^*

* This paper benefitted from comments by Adam Thompson, Derek Baker, David Faraci, Tristram McPherson, William Melanson, and two anonymous referees, as well as audiences at the Philosopher’s Cocoon and Minds Online conferences.

Abstract

This paper offers a noncognitivist characterization of moral attitudes, according to which moral attitudes count as such because of their inclusion of moral concepts. Moral concepts are distinguished by their contribution to the functional roles of some of the attitudes in which they can occur. They have no particular functional role in other attitudes, and should instead be viewed as evolutionary spandrels. In order to make the counter-intuitive implications of the view more palatable, the paper ends with an account of the evolution of normative judgments as exaptations of the cognitive structures that underlie beliefs.

Keywords:

• Noncognitivism
• Frege-Geach problem
• language of thought
• moral evolution

Notes

* This paper benefitted from comments by Adam Thompson, Derek Baker, David Faraci, Tristram McPherson, William Melanson, and two anonymous referees, as well as audiences at the Philosopher’s Cocoon and Minds Online conferences.

1. A.J. Ayer (1936) and Simon Blackburn (1984) treated moral judgments as a special sort of approval and disapproval. Charles Stevenson (1937) described them as a species of interest or partiality. Allan Gibbard (2003) compared them to plans. Mark Timmons and Terry Horgan (2006) suggested that they are a special non-representational species of belief. Mark Schroeder (2008) proposed (without endorsing) that they are attitudes of favoring or disfavoring attributions of blame. These views all emphasize the pressures moral judgments exert on us to act in certain ways. Some of the remaining difficulties involved in providing an adequate characterization of straightforwardly moral judgments are explored in David Merli (2008).

2. The question of logical complexity, first raised in the work of Peter Geach (1965) and John Searle (1962), has taken a prominent place in discussions of noncognitivism. Though it is typically presented under the inclusive label of The Frege-Geach Problem, logical complexity poses a variety of different issues. One part of the problem – the part that I focus on in this paper – is to make sense of just what sort of attitudes logical complex moral judgments are. The other parts of the problem involve explaining why logically complex attitudes relate to each other in the ways that they do, how such attitudes have logical relations such as inconsistency and entailment, and how they can figure into rationally compelling arguments. The challenges involved in all of these parts of the problem are explored in depth by Mark Schroeder (2008). I have discussed them elsewhere (Shiller 2016), and the solution I present there for this second part of the Frege-Geach problem is compatible with the solution I present here for the first part of the Frege-Geach problem.

3. This term comes from Jerry Fodor, who uses it to describe a constellation of theories (1998). I will restrict its usage to the view defined above (a view that is found, for instance, in Field [1978]): a theory that is distinct from the Language of Thought Hypothesis, which makes additional assumptions about the structure of mental representations, and the Computational Theory of Mind, which makes additional assumptions about the way that representations are handled in deliberation and reasoning.

4. Of course, representational properties are correlated with viable combinations of both concepts and LEGOs, but these properties are not explanatory except in a unificationist sense. If anything, representational contents are determined in part by what combinations are possible, rather than the other way around.

5. Friedemann Pulvermüller (2002) presents a theory that illustrates the intended difference between orthographic and syntactic properties. On his proposal, individual concepts are implemented in the brain by networks of functionally entwined neurons that selectively respond to relevant stimuli. These functional webs are themselves connected to collections of sequence detectors that act as grammatical categories. The fact that a given web is connected to a collection of detectors typical of nouns explains why it can be put into subject position in thoughts. In this theory, concepts are distinguished by orthographic properties (dispositional firing patterns of a functional web), and how they associate with each other depends upon their distinct syntactic properties (connection strengths of the web with specific sequence detectors).

6. Orthographic classes are plentiful and many will be vague. There need not be a clear cut answer as to whether or not sentences in English and Chinese fall into the same orthographic class. They fall into some of the same orthographic classes and not others.

7. We can characterize the function of moral concepts in a variety of different ways, so this proposal is consistent with many analyses of straightforwardly predicative moral attitudes.

It might be objected that my proposal will restrict moral judgments only to those creatures who employ presentations. I am sympathetic to this worry, and I will allow that having moral judgments may not require having presentations. Nevertheless, it is not misguided to limit our focus to creatures like us. Our way of having moral attitudes involves taking attitudes toward presentations with moral concepts.

8. Are concepts meaningless in their spandrel contexts? I want to resist the urge to say that moral concepts mean anything different in spandrel contexts than in the contexts in which they have a function. Representational concepts are not meaningless in their spandrel contexts precisely because they borrow their meaning from their functional contexts. For both moral and representational concepts, there is only one fundamental bearer of semantic values between the two kinds of contexts, and whatever semantic values those concepts have, they have because of their behavior in the functional contexts.

That said, I doubt that it is possible to assign an object to count as a moral concept's meaning in both spandrel and functional contexts in the way that is possible for representational concepts. If it takes a meaning object to be meaningful, then moral concepts are not meaningful in any of the contexts in which they occur.

9. The fact that an attitude lacks a function or a meaning does not entail that it is useless. The attitude may have begun its life as a spandrel and subsequently come to be put to good use. The present value of a concept may be tangential to the properties that provide it with a coherent meaning.

10. Negated moral judgments and their unnegated counterparts may genuinely be inconsistent in the shallow sense in which two attitudes are inconsistent when we are inclined to treat them as disagreeing and perceive them as incompatible social commitments. We try not to adopt such ‘inconsistent’ attitudes, and expect others to expect this of us. This shallow sense of inconsistency often arises from, and cannot explain, the appearance of something deeper. Nevertheless, recognizing that such shallow forms of inconsistency may persist mitigates the unintuitiveness of denying authenticity to deeper senses. Noncognitivists have been happy to embrace minimalist interpretations of truth and content, I see little further cost to also adopting shallow interpretations of inconsistency.

11. If the contents of our moral concepts are determined by population-level regularities of concept use (Schroeter 2014), then we can have no special introspective insight into their meanings, and it is highly plausible that we might be misled by the forms of our attitudes.

12. They may also disagree about whether the motivational force of the moral attitude is provided by the belief-like structure or by a separate structure. The most straightforward way of interpreting hybrid expressivism within RTM analyzes moral judgments into two structures, but it is also possible to treat them as one structure implementing two attitudes.

13. I expand on this line of criticism in (Shiller 2017).

14. I do not mean to imply that most aspects of morality are genetic. In fact, the proposal developed here is quite a natural fit for a cultural revolution approach (Powell, Shennan, and Thomas 2009; Sterelny 2011) to explaining human behavioral changes of the last hundred thousand years.

15. Joyce (2007) provides an in depth overview of the ways that stereotypically moral behavior is fitness enhancing.

16. Gibbard (1990), Sinclair (2012), and Björnsson and McPherson (2014) advocate for a similar perspective on the role of morality. Michael Tomasello (Tomasello 2016; Tomasello et al. 2012) has also developed a two-step process in the evolution of morality in which cultural aspects of moral psychology arose with the dawn of culture, long after the appearance of cooperation-inducing components of moral psychology.

17. The possible range of dates for the evolution of cognitively modern humans spans the emergence of anatomically modern humans ~160 ka (d’Errico and Stringer 2011) and the great dispersal out of Africa ~50 ka (Klein 2008).

18. Gould and Vrba (1982) present the concept and argue that it plays an important role in evolution, including in cognition.

19. Thanks to an anonymous reviewer for raising this idea.

20. Tomasello (2016) hypothesizes that early normative judgments focused on proper role-playing in complex cooperative activities. If so, they surely trailed beliefs about what roles individuals could productively play.

21. Furthermore, is unlikely that young children are able to discern representational from noncognitive concepts. They must form attitudes in response to moral instruction very early in life, and their mature attitudes grow out of these early ones. Whatever attitudes they ultimately come to have, children probably begin classifying actions under moral categories in much the same way that they classify actions under other categories, and this probably involves the structures of belief. Later on, they adopt the subtleties of moral concepts that rob them of their representational contents, but they don't need to fundamentally refigure their existing attitudes. In this case, perhaps, ontology recapitulates phylogeny: states that start out primarily as representations of social categorizations become something more closely tied to motivation.