The Folk Philosophy Fallacy (Part 3)
The Folk Philosophy Fallacy (Part 3)
Twitter Tuesday #22
Knobe & Prinz and the folk conception of phenomenal experience: Studies 2-4
This post is part of a series. For previous entries in the series see:
2.2.2 K&P Study 2
In their second study, K&P gave participants a list of statements that attributed mental states to group agents, then asked participants to rate whether these statements sound “weird” or “natural” on a 7-point likert scale (1 = “sounds weird”, 7 = “sounds natural”). Here are some examples:
Non-phenomenal states
Acme Corps. believes that its profit margin will soon increase.
Acme Corps. wants to change its corporate image.
Phenomenal states
Acme Corps. is now experiencing great joy.
Acme Corps. is getting depressed.
Participants judged non-phenomenal states to sound more natural than phenomenal states, consistent with K&P’s expectation that people are willing to attribute non-phenomenal states ot group agents, but less disposed to attribute phenomenal states to group agents. In fact, as they point out, “even the most acceptable phenomenal state was still deemed less acceptable than the least acceptable non-phenomenal state,” illustrating that there was no overlap at all (p. 75).
2.2.3 K&P Study 3
The goal of study 3 was to find out why people are less disposed to attribute phenomenal states to group agents. This study is a bit weird. They consider two possibilities: (1) phenomenal state attributes depend on how similar the target is to humans and (2) some other set of criteria (they’re not specific), such as an unwillingness to ascribe phenomenal states to a target agent that is made up of multiple agents. One would have thought they’d avoid bizarre scenarios, but they went with a very weird one here. Participants were given the following scenario:
Once there was a powerful sorceress. She came upon an ordinary chair and cast a spell on it that endowed it with a mind. The chair was still just made of wood, but because of the magic spell, it could now think complex thoughts and form elaborate plans. It would make detailed requests to the people around it, and if they didn’t do everything just as it wanted, it would start complaining. People used to call it the Enchanted Chair. (p. 76)
The chair is described only as having non-phenomenal mental states. K&P wondered whether people would nevertheless attribute phenomenal states to it, but not to Acme Corps. To test this, they gave participants the description of the Enchanted Chair and Acme Corps., then asked participants whether each could “feel happy or sad” using a 7-point scale. They found that participants were more likely to say that the chair could feel happy or sad (5.6/7) than Acme Corps. (1.8/7). Since people were willing to say the chair had “phenomenal” states but Acme Corps. could not, K&P take this to be evidence that “people do not simply refuse to ascribe phenomenal states to any agent that differs from human beings in its physical constitution. They must be making use of some more specific restriction that rules out group agents on independent grounds” (p. 76).
I don’t know what to make of this. I’d like to see some qualitative studies that explored what people were thinking about when they answered these questions and why they answered the questions the way they did. It’s hard to pin down what may have been going on here.
2.2.4 K&P Study 4
In Study 4, K&P wanted to show that people would be willing to ascribe mental states to a group agent so long as doing so wouldn’t attribute a phenomenal state by indicating that it had experiences or feelings. To test for this, they presented participants pairs of statements to participants that varied the wording in such a way that one of the sentences would appear to attribute a phenomenal state and the other wouldn’t, while holding other features of the mental state constant, e.g., if K&P are correct, people should be willing to say that a group agent is “upset,” but not that it is feeling “upset.” Participants were given the following two pairs of statements (they were given another similar pair as well):
Acme Corps. is feeling upset.
Acme Corps. is upset about the court’s recent ruling.
Once again, participants judged the degree to which these statements sounded weird or natural (1 = “sounds weird”, 7 = “sounds natural”). The average rating across participants indicates that they tended to think that (a) sounded much more weird (1.9) than (b) (5.3).
Overall, K&P purport to have shown that participants are consistently disposed to attribute non-phenomenal states but not phenomenal states to group agents, and that this difference is due to physical differences (rather than functional differences). This would suggest that nonphilosophers do have a concept of phenomenal consciousness (presumably one that roughly accords with the notion as it is understood by philosophers).
What should we make of these results? Wyrwa (2022) points out more serious problems with these studie. K&P’s results were not replicated in subsequent studies that corrected the methodological shortcomings of their original studies (p. 49). Wyrwa turns to Sytsma and Machery (2009), who argue that there is a significant confound in these results that raises doubts about whether the best explanation for K&P’s findings is that participants really do have a concept of phenomenal states. S&M propose that functional differences really are behind these differences in attribution. As Wyrwa puts it, Sytsma & Machery (2009) maintain that K&P failed to “control for the difference between behavior appropriate for group agents (e.g., hiring) and individuals (e.g., walking)” (p. 49, footnote 20). According to S&M:
People were unwilling to ascribe to corporations mental states that are typically associated with functional roles or behaviors that corporations are incapable of. It is possible, for example, that people feel that “Acme Corporation is now experiencing great joy” sounds weird because they do not readily imagine that a corporation, as a conglomeration of other agents, can have a state with the functional role that they associate with joy, including causing the behavioral cues associated with joy. That is, it might be that the sentence sounds weird to them because they feel that the corporation cannot do the things that normally indicate joy—cannot smile, or laugh, or even wag its tail. (p. 24)
S&M tested this by both replicating K&P’s original findings and testing a separate list of statements that either (a) only an individual agent could perform them or (b) group agents could also perform the action. Here are a couple examples of the latter:
Agent-only actions
Acme Corporation has insomnia.
Acme Corporation is eating a burrito with hot sauce.
Group-capable actions
Acme Corporation is suing GenCorp.
Acme Corporation is building a new factory.
Like K&P’s items, participants were asked to judge whether statements in each category were weird or natural. They found a similar pattern with both sets of items: agent-only actions were judged weird, just like “phenomenal” statements, while group-capable actions were judged to be natural, like “non-phenomenal” statements. Note that the individual-only actions are not actions that use the language researchers would plausibly associate with phenomenal states, and yet they likewise received low scores. It is possible, then, that what participants are really doing when attributing “phenomenal” and “non-phenomenal” states is distinguishing between the sorts of functions group agents could perform. If so, this would mean K&P’s findings do not support the conclusion that participants are employing a concept of phenomenal states.
Of course, without additional questions exploring why participants answered one way or another (such data can itself be quantitative or qualitative), these results are not dispositive one way or another. But they highlight the presence of a potentially significant confound. S&M have more to say, so check out the rest of their article if you want to see more of these sorts of objections.
S&M also highlight another problem with K&P’s studies. Recall that in Study 4, participants were given pairs of statements and asked to judge how weird each was:
Acme Corps. is feeling upset.
Acme Corps. is upset about the court’s recent ruling.
As S&M point out, results from this study could not be due to the previous confound because both states were ascribed to the same group agent (p. 29). However, S&M argue that Study 4 does not provide conclusive evidence that nonphilosophers have a concept of phenomenal consciousness because the pairs of aren’t a “minimal pair”:
The verbs in the no-feeling condition, but not in the feeling condition, are followed by a prepositional phrase. For instance, Knobe and Prinz compared the sentences ‘‘Acme Corp. is feeling upset’’ and ‘‘Acme Corp. is upset about the court’s recent ruling’’ (our emphasis). (p. 30)
If so, the difference in wording between the pairs of statements may account for K&P’s results, rather than K&P’s proposal that participants are employing a concept of PC that they attribute in some cases but not others. Once S&M tested this by swapping the prepositional phrase from the “feeling” condition to the condition without the use of the term “feeling,”which resulted in a failure to replicate K&P’s findings. In fact, they add that “not only did we fail to replicate Knobe and Prinz’s findings, our results actually reversed theirs” (p. 29).
Simply put, K&P’s studies have not withstood subsequent critical scrutiny. While these studies should be lauded for their pioneering efforts to get a handle on whether nonphilosophers have a concept of phenomenal consciousness, and, perhaps more laudably, to assess what cognitive processes may drive various mental state attributions, any early foray into a new area of inquiry is bound to provide an incomplete picture, with subsequent findings often substantially modifying or overturning early findings. Such is the case with studies on whether nonphilosophers have a concept of phenomenal consciousness. While early studies may appear to provide some positive indication that this is the case, subsequent data has failed to support these conclusions. Nevertheless, while most studies over the past fifteen or so years suggest nonphilosophers don’t share the same concept of phenomenal consciousness as philosophers, Wyrwa maintains that these studies are likewise saddled with sufficient methodological problems that they, too, are inconclusive. I turn to these criticisms next week.
References
Knobe, J., & Prinz, J. (2008). Intuitions about consciousness: Experimental studies. Phenomenology and the cognitive sciences, 7, 67-83.
Sytsma, J. M., & Machery, E. (2009). How to study folk intuitions about phenomenal consciousness. Philosophical psychology, 22(1), 21-35.
Wyrwa, M. (2022). Does the Folk Concept of Phenomenal Consciousness Exist?. Diametros, 19(71), 46-66.
Surely they've considered the difference between metaphor and actual speaker attribution of states? The large difference in naturalness between the sentences
(1) Old books hate moisture.
(2) Old books hate asparagus.
has nothing to do with my attribution of mental states to books, but rather to only (1) having an analogy with sentient beings sufficient to license a metaphor. (Sentient beings would feel hate toward things that would cause them a level of physical damage comparable to how much moisture damages old books.)
The obvious control for the "Enchanted Chair" cases would be to see how people respond to a scenario in which the sorceress magically endows Acme Corporation with a mind, and people start calling it "Enchanted Acme Corporation." Would people be willing to say Enchanted Acme Corporation can "feel happy" or "feel upset"? My guess is that they would — but this is a question about fairytale metaphysics, not folk metaphysics, and there is no reason to suppose facts about the former automatically carry over to the latter.
(Of course, it isn't obvious that fairytale metaphysics is any less philosophically important than folk metaphysics — indeed, I suspect it may even be more important.)