Gender Differences and Ordinary "Truth"

Suppose one asserts that there is a cat on the mat, when--in fact--there is a cat on the mat.  Is the accuracy of this assertion sufficient for truth?  

In a recent study inspired by Arne Naess’ 1938 ("Truth" as Conceived by Those Who Are Not Professional Philosophers) project, Bob Barnard and I asked respondents about their conception of truth.  Naess' work suggested that there may be significant differences in how men and women think about truth.  We found that men and women differ with respect to the degree to which accurate description or representation is judged sufficient for truth.  In at least some cases, men see accurate representation as a more important component of truth than do women.  In light of these findings we further suggest that men, more than women, would be apt to accept a version of the truth principle often known as "Aristotle’s dictum":

To say of what is that it is not, or of what is not that it is, is false, while to say of what is that it is, or of what is not that it is not, is true. (Aristotle, Metaphysics 1011b25)

That these results suggest a gender difference in how a core philosophical concept is characterized by respondents is especially intriguing in light of recent work by Buckwalter and Stich (here), which contends that men and women systematically differ in the intuitions they report in response to canonical philosophical thought experiments.   We would like to present our empirical findings that seem to show that although the conception of truth outlined by Aristotle’s dictum may seem intuitive to the philosopher (Cf. philpapers study here), the data suggest that something else may be at work in the folk conception of truth.   (This study follows up on a related historical project about the relationship between Naess and Tarski (see discussion below), a draft of which is available by emailing one of us, either Joe at oohlah at mac dot com or Bob at rwbjr at olemiss dot edu.) 

Methods

Respondents were randomly assigned to one of nine distinct conditions using Qualtrics, an online survey system.  Participants were instructed to answer questions about truth and to complete a demographic measure (gender, age, ethnicity, political affiliation, educational attainment  in philosophy).  Participants also received a cover page with a disclaimer regarding informed consent, confidentiality, and use of the data collected, as well as a summary of the purpose of the research and conditions of participation.  

Participants were 362 people representing diverse regions of the globe who responded to the Yale Experimental Philosophy Month initiative.  Respondents were asked to read a brief vignette and respond to 16 questions intended to characterize their views on truth. Our focus will be on 3 of the 9 conditions: Bruno, Donna "No German", and Donna "Only English".  Our data reflect the results from only one of the 16 questions we asked participants.  We employed a 5-point Likert scale where 1 = strongly disagree and 5 = strongly agree. We asked subjects: "With respect to the case of [condition's name], how much do you agree or disagree with each of the following statements?" We would like to discuss what we found in response to the following statement:

If a claim reports how the world is, then it is true.

Bruno condition

For the Bruno condition, subjects were 41 participants, which included 24 male and 17 female respondents:

Bruno has just finished painting his house.  Bruno painted his house the same color as the sky on a clear summer day.  Bruno claims his house is blue.

The Bruno condition was framed as a straightforward empirical example involving ordinary color experience.  When thinking of the Bruno vignette, males tended to agree more strongly than females with the statement that a claim is true if it is an accurate description of how the world is.  The mean male response was 3.96 out of 5, and the mean female response was 3.07 out of 5.  This difference is statistically significant (t(39) = 2.305, p<.027).  Our results yielded the view that the degree to which men accept that accurate description entails truth is greater than the degree to which females accept that truth is entailed by accurate description.  Females were essentially neutral; males on average expressed a definite commitment to accurate description.

Donna "No German" condition

Discovering gender differences in ordinary conceptions of truth is inherently fascinating, but we felt that one condition in a single experiment was insufficient for us to suggest categorically that male conceptions of truth differ significantly to women's conception of truth.  We devised a pair of vignettes to test the extent to which 'word framing' mattered for ordinary conceptions of truth.  We hypothesized that responses to one vignette might yield gender differences in conceptions of truth but the other would not.

Here is the vignette we used:

Donna is traveling in Germany, but does not speak German.  She watches as a sailor asks for der Stadtplan and is handed what looks like a map of the city.  Donna asks for der Stadtplan in a shop and is sold a city map.  Donna still speaks no German, but believes that asking for der Stadtplan is a good way to obtain a city map from a German shopkeeper.

The mean male response was 3.56 out of 5, and the mean female response was 2.6 out of 5.  The difference is statistically significant (t(37.701) = 2.521, p<.016).  So, just as in the Bruno condition, this case yields results where males, more so than females, tend to agree more strongly that accurate description necessarily entails truth.

Donna "Only English" condition

We revised the Donna "No German" condition to reflect that Donna only speaks English.  The only difference between the two conditions is in the interchange of these phrases.  We believed that no statistical difference would be present when subjects were told that Donna only speaks English.   Here is the revised vignette including the revised language (which is in bold):

Donna is traveling in Germany, and she only speaks English.  She watches as a sailor asks for der Stadtplan and is handed what looks like a map of the city.  Donna asks for der Stadtplan in a shop and is sold a city map.  Donna still speaks only English, but believes that asking for der Stadtplan is a good way to obtain a city map from a German shopkeeper. 

Our hypothesis was borne out.  The mean male response was 3.53 out of 5, and the mean female response was 3.05 out of 5, which fails to yield a statistically significant difference (t(37) = 1.258, ns).  The observed geneder difference seems to be blunted by the change of wording in the scenario. When the vignette claims that Donna speaks only English, male and female responses were not significantly different.  It is obvious that we cannot compare the statistically significant results found in Donna "No German" with the statistically insignificant results in Donna "only English." Nevertheless, our data seem to show (a) that males affirm the accurate representation condition of truth in a stable way across some contexts, and (b) females are more sensitive to the presence of certain contextual elements of a situation.

One possible explanation for this may involve the relative permissivness of "no German" compared to "only English", i.e. that despite not speaking German, Donna may possess some other capacity, e.g. speaking Dutch, Polish, French, Belgian, etc., while the "Only English" leaves no question in the respondent's mind that Donna is unable to speak anything but English. 

Discussion

Our empirical results suggest that there is not universal commitment to accurate description or representation among "the folk" as central to the concept of truth.  Specifically, we found that in particular contexts differences appear to correlate in statistically meaningful ways with a person's gender.  We want to suggest that the ordinary notion of truth is far more complex than we have been led to believe.  This has important implications for any theory of truth that, like, e.g.  Tarski's semantic theory, is committed to the centrality of accurate description to the "ordinary conception of truth."  (Recall that Tarski's "Material Adequacy" requirement involves an attempt to formalize the ordinary notion of turth, and that this is characterized in terms of Aristotle's dictum). If the ordinary concept does not always involve accurate description, then it does not always conform to Aristotle's dictum.  

Our data seem to support a prima facie case for claiming that Tarski may have been aiming at the wrong target; Aristotle's dictum is not co-extensive with the ordinary notion of truth.   We do not want to claim that this is a reason to reject Tarski's work out of hand.  Rather, we suggest that Tarski style projects must either be revised to reflect the diversity present in the ordinary concept of truth, or must give up on Tarski's goal of full material adequacy in favor of an emphasis on formalization of a concept of truth.  (Hartry Field, for one., has suggested that the semantic theory's greatest virtue comes in its formalization of truth in a model.)

What our results suggest is that accurate description is not a universal feature of truth.  If we have really discovered a difference between the ordinary conception of truth for men and for women, then should we automatically think that this entails that pluralists about truth are on the right track?  Not so fast.

If we had discovered that there were no significant differences across conditions or between genders, etc., that would ground an experimental case against the plausibility of truth pluralism.  But we did not find that.  Rather, the results are indifferent to the extent that they could reflect real and distinct kinds of truth, or they could be distinct expressions of an as yet unrecognized singular notion.  It should be the case that some components of a theory of truth should be static, and, even in pluralists' theory of truth, there are platitudes or truisms about truth that hold in almost all circumstances.

Critics of our experiments will likely raise two problems.  One critic may argue that because we did not ask participants to agree or disagree with Aristotle's dictum, our experiment shows nothing about either Aristotle's dictum being gender specific or Tarski's view that the ordinary person's conception of truth coincides with the classical conception.  We meant our questions to be theoretically shallow such that ordinary people's conception of truth could come to the top.  But this is a question of how to interpret Aristotle's dictum.  It seems pretty straightforward that Aristotle's dictum is about the (to borrow from Tarski) "adequation of something that represents or describes the world," i.e., material adequacy.  If that's something other than accurate description, then whatever it is it must involve adequate description.

A second criticism of our project might contend that participants are responding to the gender of the agent in the vignette. One may believe that men and women answered differently because they harbor some kind of unconscious bias favoring their own gender.  We attempted to design the vignettes to elicit specific responses or trigger certain conceptions of truth in contexts involving different cases: mathematics conditions, pragmatic cases, empirical cases, agreement conditions, consistency with group (majority) opinion, or objectivity conditions.  The vignettes were designed to put people in frames.  In some frames, accurate description does not seem to be important.  Nevertheless, given the statistically significant differences we found in the Bruno and Donna "No German" conditions, we are left to sort out an explanation of why these gender differences were present in the cases.

What we have found does not offer conclusive data or a final characterization of the ordinary pre-theoretic conception of truth.  In fact, we are willing to bet that more interesting discoveries will take place with further experimentation.  So, despite the interesting findings we report here, we hope that future empirical research - our own research included - will hone in on what the ordinary conception of truth is.

Shameless self-promotion

Some of our other work stemming from the research we conducted during the Yale Experimental Philosophy Month Initiative will be included in a poster session at the upcoming Washington, D.C., APA-Eastern meeting.  Our poster will report some fascinating results on how the relationship between evidence and truth vary by gender.  We encourage anyone attending the Eastern to stop by for a chat!

Comments

[Eric Schwitzgebel]

An interesting follow-up, I think, might concern statements that seem to be approximately true or to have intermediate truth values. For example, you could ask a bunch of questions, some of which might be hard to classify as simply true or simply false, and then give a Likert scale with "true" or "completely true" on one end and "false" or "completely false" on the other. At the midpoint, maybe "neither true nor false"? I'm not sure. Hypothesis: Women will use the middle the of the scale more than men, and men will use the endpoints more than women.

One big issue though (perhaps also an issue for your material above?) -- how to pry apart differences in confidence, perhaps gender related, from different attitudes toward truth.

[Joe U.]

Thanks Eric!

Yes; indeed, I think you're correct that an interesting follow-up study would involve testing approximations of truth. Bob and I have discussed such a modification. Besides your recommendations, there are a number of ways we have conceived of such a follow-up study. One could ask whether some statement is closer to the truth than another, or one could ask respondents to rate future contingent statements, e.g., "There will be a sea battle tomorrow," on a scale from "absolutely true" to "absolutely false." Thanks again.

[Eric Schwitzgebel]

Sounds good. Keep us all posted!

[Jesus Zamora]

I think an interesting experiment would be to describe situations in which people DON'T say the truth, and look at how men and women may differ in their judgements about whether the apeople in those situations are lying or not

[Dean Eckles]

For No German vs. English Only, you fail to do the appropriate test, which is a test for an interaction. Instead, you conclude there is interaction/moderation because one difference is statistically significant and the other is not. This is a common error that you should avoid.

[Peter Ludlow]

Gee, I never realized that Tarski had an "accurate description" theory of truth.

[Jeremy Fantl]

Hi Joe,

I'm not sure I'm correctly understanding the survey. Here's how I'm interpreting things, so you'll tell me if I'm going wrong. After you presented the Bruno vignette, you then asked subjects to respond to a number of statements. One of the statements was this one:

"If a claim reports how the world is, then it is true."

You then found statistically significant differences in how men and women responded. In interpreting this result, you say that "When thinking of the Bruno vignette, males tended to agree more strongly than females with the statement that a claim is true if it is an accurate description of how the world is." But the word "accurate" doesn't appear in the target statement. "saying how the world is" seems ambiguous to me between "making a claim about how the world is" and "accurately saying what the world is like". I'm inclined to interpret it in the former way, which makes the target statement false. But even if there's some gender difference in the interpretation of this ambiguous clause, this wouldn't show anything about whether there's a gender difference in conception of truth. Did you ask subjects to respond to the following statements, too?

"If a claim accurately/correctly reports how the world is, then it is true."

or

"If a claim reports how the world is, and in fact the world is that way, then it is true."

[Joe U.]

Thanks Jesus. Employing a vignette in which an agent doesn't tell the truth may result in some interesting findings.

Jeremy: I believe there's some background information we should make clear. We're not saying anything substantive about the content of Tarski's view. But we take him at his word that he's committed to formally capturing Aristotle's dictum as his preferred formulation of the 'ordinary view' (Tarski 1933, p. 153; Tarski 1944, pp.342, 360). Aristotle's dictum says that when what's said is an account of what has been observed in the world, that which is 'said' is true. So, a statement is true if it reports how the world is. "Accurate description" is intended as a variant of this vocabulary. We believe that reporting and accurate description are the same thing. But we're happy to do away with accurate description talk.

When you say "even if there's some gender difference in the interpretation of this ambiguous clause, this wouldn't show anything about whether there's a gender difference in conception of truth," I believe you mean to say that "this wouldn't show anything about whether there's a gender difference in the concept of truth or the folk theory of truth." We wholeheartedly agree with you! Our study was meant to get at how people conceive of truth not to outline a folk theory of truth. (Two points: (1) it might be helpful to think of our project as being aligned with what has been termed the negative project in experimental philosophy [for a brief overview of the negative project, see here ]. (2) When we employ the term "conception", we do so to avoid the more philosophically loaded term "intuition.")

We were operating with the belief that reporting how the world is is accurately describing the world. But what you've pointed out is certainly very helpful. In future studies, we may add the term 'accurately' or the phrase 'and in fact the world is that way' to the statement. That's a really helpful suggestion! Thanks very much!

[Garrett Marks-Wilt]

Hi Joe!

Interesting work! You may be onto something; but two quick things.

1. Like Dean, I would have liked to see the interaction (at the very least). Until then, I would consider the conclusion tentative (at the very best). But, male scores are about the same; female scores differ. So, if you can show that there is an effect of condition on females, you are definitely onto something.

2. "conceive" vs. "intuition" : Have I totally missed something? I thought the socially-agreed upon term for people's responses to thought experiments was "intuition". Did I miss out on the meeting where the x-philes agreed to start studying what people "conceive"....

(also, don't Buckwalter and Stich use the term "intuition" when studying the exact same thing: gender and responses to thought experiments. I mean...isn't it in the title: "Gender and Philosophical Intuition" or some such thing like that...)

[Jeremy Fantl]

Thanks, Joe. I guess I'm still a bit confused. I don't really see substantial evidence that people conceive of truth differently. And the reason is that one very natural reading of "a statement is true if it reports how the world is" is false, and I don't see how Aristotle was committed to it. Here's a statement that reports how the world is:

"The sun revolves around the earth."

That statement is false. But it makes a report of how the world is. Of course, it doesn't accurately report how the world is. You might think that I'm wrong about how "saying/reporting how the world is" should be read. But it's a very natural reading to my ear, and I don't see why the evidence you've come up with is better evidence that people conceive of truth differently than that some people read "saying how the world is" as "saying accurately how the world is" while others read "saying how the world is" as "saying (accurately or not) that the world is a certain way." So, I didn't take myself primarily as making suggestions for further study. I took myself primarily as worrying about the evidential value of this one.

[Joe U.]

Thanks very much Garrett and Dean! We appreciate your comments, and we are looking into the problem you both have raised regarding our statistical analysis. Don't take this as an official response, but I believe that we've not been clear in the representation of our statistical analysis. You both have observed this and -- quite correctly -- challenged us on it. Please stay tuned.

As far as (2) is concerned, my admitting to a difference between "conception" and "intuition" may be exposing something about me, rather than the discipline. Yes, Buckwalter and Stich do call responses to thought experiments "intuitions." I'm sure that others in experimental philosophy do too. But, largely because I've spent some time discussing "intuitions" with Marc Moffett and John Bengson and reading Bealer and Williamson, I stake out a fine-grained position that distinguishes between the two.

[James Andow]

Is the draft still available? The link in the post is not working for me at the moment.

[Joe U.]

Our apologies James. The paper we posted had to be taken down from the internet because it is under review at a journal that requires it not be published online and potentially available to reviewers. I will email you a copy of our historical paper on Tarski/Naess.