Haidt’s “Emotional Dog” has three legs, a broken tail, and fleas
The fleas
Over and above the flaws with Haidt’s presentation of the Social Intuitionist Model (SIM: see parts 1, 2a, and 2b), there are a number of lesser issues that bear being brought to light. At the risk of (again) taking the analogy too far, enough fleas can lead to canine anaemia which can be life-threatening.
The first flea ties in pretty closely with the preceding articles as it relates to Haidt’s definition of reasoning, but it especially relates to Haidt’s defence of that definition in his response to Saltzstein and Kasachkoff’s (2004) critique, which is why it’s one of the fleas rather than being included as one of the breaks in the Emotional Dog’s tail.
Reason 1 & Reason 2
I have suggested that Haidt may have been operating with two definitions of reasoning. With regard to the explicit definition, in response to Saltzstein and Kasachkoff’s critique, Haidt (2003) says:
I think that part of the apparent gulf between S&K and me is due to our very different views of what exactly counts as reasoning. I took great pains to define my terms carefully, and I adopted a relatively circumscribed definition of “reasoning” that did not include all conscious cognitive processes. (p. 286)
In Haidt’s 2001 article, in the first paragraph on page 818 (actually the continuation of a paragraph started on the previous page, just to be clear), he says one thing about the definition of reasoning, but in the next paragraph he says something subtly different:
Para 1: “…a key part of the definition of reasoning is that it has steps, at least a few of which are performed consciously.”
Para 2: “moral reasoning can now be defined as conscious mental activity that consists of transforming given information about people in order to reach a moral judgment.” (p. 818, emphases mine)
The first statement suggests that reasoning contains some conscious steps and the second is that reasoning is a particular kind of conscious mental activity. You might say that if something contains steps of which one is consciously aware then it is conscious mental activity. I would argue that this transgresses the boundary, dare I say the binary, that Haidt seems to be attempting to enforce between conscious reasoning and intuitions. Haidt finishes that paragraph by saying:
To say that moral reasoning is a conscious process means that the process is intentional, effortful, and controllable and that the reasoner is aware that it is going on (Haidt, 2001, p. 818)
This is a purer rationalism than most rationalists would hold. Indeed, all three critiques hold to a more interactionist version of rationalism. Haidt even recognizes Pizarro and Bloom’s model as a “modified rationalist model” (p. 197).
It is possible, therefore, that part of the disconnect that Haidt perceived between himself and these two critiques is due to his apparent misrepresentation of the rationalist approach and/or the two definitions of reasoning that underpins it.
Intentional Intuition?
Despite the contradictions noted above, it seems that Haidt’s intention is to define intuition as a process that delivers outputs into consciousness without awareness of the steps taken to arrive at them. Reasoning, by contrast, requires at least some awareness of some of the steps. However, one implication is that intentionally pausing — allowing other, weaker intuitions to be triggered — is reasoning. Additionally, noticing the strength of the affective responses to several options and weighing them up on that and other bases is also reasoning. Indeed, the idea of flippism — flipping a coin and noticing your emotional reaction to the outcome — relies on this intentional use of intuition. This take on flippism (not the Flippism laid out in the cartoon from which it gets its name, below) is a means to access one’s intuitive preferences. It would also conform to Haidt’s definition of reasoning.
By contrast, responding intuitively based on a long history of prior reasoning — as noted in the articles on The Broken Tail — would be indistinguishable from a truism.
As an aside, some might argue that introducing intentional randomness — paying attention to the result of the coin flip and not one’s emotional response to it (i.e. the Flippism actually put forward in the cartoon above) — is a low-cost and rational route to decision-making in low-risk situations where information gathering is too costly. But I digress.
The Emotional Dog or Schroedinger’s Cat?
There is also some confusion within the SIM regarding its status as anti-rationalist, such that the Emotional Dog is akin to Schroedinger’s Cat. At any given moment it is both antirationalist and not, it just depends on which section you are reading. Haidt (2001) states that:
It must be stressed at the outset that the social intuitionist model is an antirationalist model only in one limited sense: It says that moral reasoning is rarely the direct cause of moral judgment. (p. 815)
To say that the model is antirationalist in a “limited sense” and then suggest that reasoning is “rarely the direct cause of moral judgment”, is not being antirationalist in a limited sense. Indeed, Haidt’s thesis is that reasoning is used more for self-justification than actual moral judgment, which is very far from a limited type of anti-rationalism. To claim otherwise smacks of wanting to have one’s dog treats as well as eat them. All that being said, the claim about antirationalism is contradicted at the end of the paper:
The social intuitionist model, therefore, is not an antirationalist model. It is a model about the complex and dynamic ways that intuition, reasoning, and social influences interact to produce moral judgment. (p. 829)
It is possible that the intent was to provoke more of a response with the introduction of Social Intuitionism than would have been achieved with the introduction of an interactionist rationalism. However, given the robust defence that Haidt has given against two of the three critiques that I’ve cited, this seems unlikely.
It is ironic, therefore, that what I would call for is what Haidt described the SIM as: “…a model about the complex and dynamic ways that intuition, reasoning, and social influences interact to produce moral judgment” (Haidt, 2001, p. 829).
Philosophers and others
Haidt claims that:
…solitary moral reasoning may be common among philosophers and among those who have a high need for cognition (Cacioppo & Petty, 1982; emphasis mine)
These sound like pretty exclusive groups. Certainly, philosophers are relatively rare, but what of those high in the need for cognition?
People that self-identify as liberal tend to be high in the Big-Five personality trait of openness to experience, and this correlates strongly with a need for cognition (Sadowski & Cogburn, 1995). Openness to experience also correlates with traits similar to need for cognition from other models (McCrae, 1996). More recent research has noted a more direct link between the need for cognition and liberal political views, but at the time of Haidt’s writing the findings tended to be more indirect (see, for example, Tetlock, 1983, PDF).
Pew Research’s political polarization survey in 1994 found that 3% of respondents were consistently liberal and 18% were mostly liberal. A mere five years later, the numbers were 6% and 25%, respectively (see table below). It is the case, therefore, that around 30% of the US population were mostly or consistently liberal by the time of Haidt’s writing and were thus likely to be at least relatively high, and possibly actually high, in need for cognition.
It is worth noting that Libertarians, who can be left-leaning or right-leaning (but who have tended to side with conservatives in the US), also tend to be high in the need for cognition (as noted by Haidt himself in subsequent research, e.g. Iyer, et al., 2011). So it’s possible that some significant proportion of those with mixed liberal/conservative views are either liberal-leaning or libertarian and thus also high (or comparatively high) in need for cognition.
You would have been hard-pressed to extract the idea that a quarter of the US population may engage in solitary moral reasoning from the way Haidt introduced it in the quote above.
Towards the end of his article, Haidt states that:
[t]he literature on everyday reasoning (Kuhn, 1991) suggests such an ability may be common only among philosophers, who have been extensively trained and socialized to follow reasoning even to very disturbing conclusions…” (p. 829).
In the conclusion of the book that Haidt is citing, Kuhn (1991) suggests that to reason to an appropriate conclusion, one must compare various theories and weigh them against each other. She notes that 21–32% of participants in the research engaged in this integrative process on one or more topics, but only a tiny handful did so across all topics. As such, it was not only philosophers that reasoned in this way, around a quarter of people did so sometimes. That number is remarkably close to the number of liberal and mixed individuals in the Pew research findings, and a lot more than the “few people among us” (p. 829) that Haidt allows for.
Capacity, Capability and Ability
In the previous article, I addressed Haidt’s comments about the externalization of intuitions, suggesting that they were the expressions of new abilities unmodified by experience rather than new intuitions, per sé. In the above section, I drew a distinction between people who engage in rigorous reasoning and people who don’t but, unlike Haidt, I pointed out that more than a few people engage in more rigorous reasoning sometimes.
Both of these instances seem to suggest that Haidt is not making the qualitative distinction between capacity, ability and capability (this is the logical sequence, the order in the title just looked better to me).
The “bottom line” from the American Medical Association style guide writers is:
Ability = Actual skill, either mental or physical; native or acquired.
Capacity = Potential to develop a skill, usually mental; native, as opposed to acquired.
Capability = Unique fitness for a defined end; sometimes may be used in place of ability, but its use in place of capacity is incorrect.
In summary
Haidt has two definitions of reasoning. These blur the distinction he is attempting to make between intuition and reasoning. The distinction doesn’t work, anyway, because the use of intuitive processes can be intentional — which by Haidt’s distinction would be reasoning — and the use of intuitions that have their roots in reasoning is indistinguishable from the use of truisms. Haidt claims that the Social Intuitionist Model is both Antirationalist and not (possibly in line with whichever definition of reasoning is predominating at the time) this makes the Emotional Dog more like Schroedinger’s Cat. Haidt’s characterization of the general public as nothing like philosophers ignores the roughly 25% of the population that reason like philosophers sometimes. In ignoring them, he seems to ignore the difference between capacity, ability and capability. An omission that also seemed to be present in the characterization of young children’s intuitions.
A final flea
I started this article with a reference to Haidt taking Saltzstein and Kasachkoff to task for their misapplication of his definition of reasoning in their critique and noted that the uncertainty of Haidt’s definition is part of the problem. I get the distinct impression that Haidt was genuinely saddened by the quality of Saltzstein and Kasachkoff’s critique as a whole, especially given that Saltzstein also critiqued Lawerence Kohlberg’s longitudinal study in 1983. He would have acted as a very real bridge between Kohlberg’s rationalist approach and Haidt’s social intuitionist one, had the critique been of a higher standard. The problem with Saltzstein and Kasachkoff’s critique is that it is marred by unnecessary rhetorical flourishes. A couple of examples will suffice (the questionable flourishes are in bold):
For Haidt, automatic processing appears to be sovereign in the making of moral judgments, with reasoning relegated to a secondary role that is only artificially called forth (e.g., by intensive interviewing). (p. 274)
With so much research providing evidence (from different studies, with different types of participants, and in different contexts) that moral reasoning does indeed take place, Haidt can only iterate his claim that what appears to be moral reasoning is simply ad hoc rationalization of one sort or another and so is not reasoning at all. (p. 279)
In both quotes, the opening text is a fair assessment of the Social Intuitionist Model but the closing elements (in bold) are clearly false. The critique is incisive elsewhere, and I have used it repeatedly in my articles. Unfortunately, Haidt seems to miss the worthwhile critique, waylaid by the more obviously wrong elements.
In the next article I will do something that is relatively unusual on Medium; having critiqued Haidt’s article, I will attempt to provide a solution to the issues that I’ve highlighted. All things being equal that will be posted next Saturday.
