Friday, November 30, 2018

Has the Sharp Decline in Philosophy Majors Hit Bottom? (Plus Other Interesting NCES Data)

I've received good news on a grant, with Morgan Thompson and Eric Winsberg, for some data-crunching about Philosophy majors in the U.S. To celebrate, here's more data!

Has the Sharp Decline in Philosophy Majors Hit Bottom?

Last year, drawing on the publicly available IPEDS database, I reported an alarmingly sharp decline in the number of recipients of Bachelor's degrees in Philosophy in the U.S. since 2010 -- from 9,297 in 2010, or 0.58% of all graduates, to 7,507 in 2016, or 0.39% of all graduates. I found similar declines in History, English, and other language majors.

In the most recent year's data (2016-2017), I'm encouraged to see no further decline either in the total number of Philosophy majors or in the percentage of graduates who choose the Philosophy major. To give you a sense of how striking this is, look at 2013-2017:

Bachelor's Degrees in Philosophy in the U.S., by year:

2013: 9,439 (0.53% of all graduates)
2014: 8,837 (0.47%)
2015: 8,198 (0.43%)
2016: 7,507 (0.39%)
2017: 7,579 (0.39%)

[for methodological details see here]

After declining by about 600 majors per year from 2013 through 2016, 2017 shows a slight uptick. The 95% confidence interval around the 2017 number is 0.38% to 0.40%, so this flattening is not merely a chance deviation from the 0.05%/year downward trend since 2013.

English and History, However, Continue to Decline

In contrast with Philosophy, English and History are continuing to decline (though note that English and History still have many more majors overall).

Bachelor's Degrees in English:

2013: 56,021 (3.0% of all graduates)
2014: 54,222 (2.8%)
2015: 49,540 (2.5%)
2016: 46,259 (2.3%)
2017: 44,686 (2.2%)

Bachelor's Degrees in History:

2013: 37,583 (2.0% of all graduates)
2014: 34,193 (1.8%)
2015: 31,048 (1.6%)
2016: 28,229 (1.4%)
2017: 26,724 (1.3%)

Despite a 9% growth in total number of graduates across all majors, in the five years from 2013 to 2017, the number of graduates declined by 20% in English and by 29% in History, and is continuing to fall fast. (Maybe, if you squint at the data with an optimistic eye, the rate of decline is slowing.)

In Contrast, the Number of PhDs Hasn't Declined Much

In contrast, the number of Philosophy, English, and History PhDs awarded has declined only slightly in that same five year period: from 475 to 438 in Philosophy, 1377 to 1347 in English, and 1003 to 945 in History. Consequently, the ratio of BAs to PhDs has declined considerably between 2013 and 2017. In English the ratio fell from 41 BAs per PhD in 2013 to 33 BAs per PhD in 2017. In History the ratio fell from 37 to 29, and in Philosophy it fell from 20 to 17.

Here's a chart going back to 2010:

[click to clarify and enlarge]


Majors can rise and decline substantially in popularity over the years. Continuing decline isn't inevitable. I'm rooting for a turnaround.

I might be wrong, but I think and hope that the recently increasing visibility of philosophers in prominent public venues such as the New York Times has helped improve the educated public's perception of the relevance and interest of philosophy. Let's show the world the value of the humanities!

Wednesday, November 21, 2018

Narrative but Not Philosophical Argument Motivates Giving to Charity

by Eric Schwitzgebel and Christopher McVey

... or at least this is so for one narrative we've tried and one type of philosophical argument.

Update, Sep 9: In a follow-up study, we have replicated the main results described below using four different arguments and four different narratives.


From 2009 to 2015, Eric published a series of studies that consistently found, using a wide variety of measures, that ethics professors do not behave any morally better than comparison groups of other professors (summarized here). One simple explanation (favored, for example, by Jon Haidt in his discussion of Eric's work) is that philosophical moral reflection has little to no influence on people's moral behavior. However, Eric and his frequent collaborator Joshua Rust have been hesitant to draw that conclusion without more empirical evidence (see here and here for possible alternative explanations).

Meanwhile, Eric's PhD student Chris was growing interested in the power of narratives to change behavior. So we decided to collaborate on a project to compare the influence of narrative and argument on charitable giving.


For our argument, we chose an argument constructed by philosopher Peter Singer with the aim of convincing ordinary readers to donate to charity. The argument was originally written for a paper in draft by Luke Buckland, Matthew Lindauer, David Rodriguez-Arias, and Carissa Veliz. (Thanks to them and to Peter Singer for sharing this argument with us.) We supplemented Singer's argument with some additional clarificatory text. It's a typical Singer-style argument, emphasizing that people have a duty to prevent death and suffering due to extreme poverty by donating money that they would otherwise spend on luxuries, and that our physical distance from those who are suffering does not excuse us from the obligation to help.

For our narrative, we chose a true story, about the same word-length as the argument, about a family rescued from slavery by a charitable donation, which we adapted from the website of a charitable organization.

Participants were 918 MTurk workers in the United States. We divided them into four conditions: (1.) narrative only, (2.) argument only, (3.) narrative + argument, and (4.) control (who received a similar-length text from a middle-school physics textbook). Participants were then asked to rate, on a 7-point scale, how much they agreed or disagreed with statements about the ethics of donation and their own motivation to donate (e.g., "It is morally good to give money to charities that help those in extreme poverty", "Currently I feel motivated to give money to a charity that helps people in extreme poverty"). In Experiment 1, we then asked them a hypothetical question about how much they would give to a charity if we offered them a bonus of $10. In Experiment 2, we replaced the hypothetical-donation language with a lottery choice. We'll report only on Experiment 2 here, but the results of Experiment 1 were very similar.

Here is the exact language for the lottery choice:

Upon completion of this study, 10% of participants in this study will receive an additional $10. You have the option to donate some portion of this $10 to one of six well-known charities that have been shown to effectively fight suffering due to extreme poverty. If you are one of the recipients of the additional $10, the portion you decide to keep will appear as a bonus credited to your Mechanical Turk worker account, and the portion you decide to donate will be given to the charity you pick from the list below. If you are one of the recipients of the additional $10, how much of your additional $10 would you like to donate?

Participants then ticked a number on a scale from $0 to $10 in dollar intervals, indicating how much they would donate if they won the $10. (A big thanks to The Life You Can Save, especially Jon Behar, for funding this study!)

Finally, we asked questions designed to measure "narrative transportation" (i.e., how immersive and moving they found the narrative, or argument, or physics text), adapted from Green and Brock 2000.

We thought we'd share some preliminary results. We emphasize that these results are only preliminary. Later, we'll want to double-check our numbers, confirm that we got the exclusions exactly right, etc. (We have a preregistration at Open Science Foundation, but we haven't made it public yet.)


Mean amount donated, by condition:

(Apologies for blurry image. Click to enhance. Error bars are 95% confidence intervals.)

Statistically, the conditions are different (ANOVA [913,3], F=4.99, p = .002). In planned, pre-registered one-tailed t-tests, we found that the narrative and narrative+argument conditions generated more donation than the control condition (p = .02, .001) but the argument condition did not (p = .48). Similarly in a multiple regression analysis we found that being in one of the narrative conditions (narrative only or narrative+argument) predicted donation amount but being in one of the argumentative conditions was not statistically significantly predictive. In other words, exposure to a Singer-style philosophical argument had no statistically detectable effect on charitable giving, but exposure to a narrative about a family rescued from slavery did have a detectable effect. This was also true in Experiment 1, concerning hypothetical donation.

The results were even clearer for the measure of ethical opinion and motivation to donate, which was scored on a scale from 5 to 35, based on 5 questions:

It's not a huge effect size -- about 2 points on the scale -- but it's easier to see how the results for argument and control are similar to each other and different from the results for narrative and narrative+argument, which are also similar to each other (ANOVA [913,3], F=10.8, p < .001; one-tailed t vs. control condition: argument p = .71, narrative p < .001, nar+arg p < .001). Similarly to the lottery donation results, a regression analysis of reported motivation and moral opinion finds being in a narrative condition predictive and being in an argument condition non-predictive.

As expected, we also found that participants reported more narrative transportation in the narrative and narrative+argument conditions than in the other conditions (ANOVA, p < .001) and that narrative transport, amount given, and motivation to give were mutually correlated (transport-amount given, r = .24; transport-motivation, r = .40, motivation-amount given, r = .50; all p's < .001).


We regard this as preliminary evidence that exposure to at least one type of narrative influences charitable giving, motivation, and opinion, while exposure to one common type of philosophical argument has little if any influence.

In our teaching, we also seem to find that Singer-style arguments have little effect on moral opinion or motivation to donate. In another series of studies (collaborative with Peter Singer and Brad Cokelet), which we hope to be able to report on soon, professors and TAs at UC Riverside taught lower-division philosophy students philosophical arguments for charitable donation and for vegetarianism. In subsequent questionnaires, students' self-reported attitudes toward charity don't appear to have been much influenced by the arguments, but their self-reported attitudes toward vegetarianism do appear to have been influenced.

This is not to deny that some people are influenced to donate by Singer's arguments. Eric believes that he has been, for example. But the effect might be weak or uncommon or not effectively produced by brief exposures.

In follow-up research, we plan to see if we find similar results with different types of arguments and narratives, and we hope to zero in on what underlying factors are at work behind these effects.

Friday, November 16, 2018

Possible Backfire Effects of an Excellent Diversity Statement?

Applicants for faculty positions in the University of California must now include "Diversity Statements" alongside more traditional elements of their applications. Other universities have instituted similar requirements. Applicants might wonder how much detail and energy to put into a diversity statement. The question is trickier than it might seem, and there is, I think, some risk that an excellent diversity statement will backfire.

One of the most acrimonious and politicized issues in faculty hiring in the U.S. right now is the extent to which universities ought to prioritize increasing the demographic diversity of their faculty. On one side are faculty and administrators who think that applicants' race, gender, or other demographic features ought not to be considered at all in hiring; on the other side are faculty and administrators who regard demographic diversification as a very high priority in faculty hiring; and of course there is a range of nuances and intermediate positions. You can see the politics of this issue, as it manifests in philosophy blogs, for example here and here.

Because of these background politics, diversity statements that are too passionate or detailed might risk alienating some people on a hiring committee. In the current environment, such statements are not politically neutral.


At U.C. Riverside, where I work, it's clear that pursuit of demographic diversity is a high priority in the administration. Many faculty have the impression that if a proposed hire will contribute substantially to the "diversity mission" of the university, that hire is much likelier to be approved. Every search committee must have an "Affirmative Action Compliance Officer" responsible for monitoring affirmative action efforts and tracking justifications for the rejection of all rejected applicants. Also, all faculty on hiring committees must complete a thirty-minute online diversity training and attend a ninety-minute in-person workshop on "Promoting Faculty Diversity".

I attended my first diversity workshop a couple of months ago. It struck me that most of the attendees -- even those who thought that promoting diversity was a good idea -- felt resentful that they were required to engage in two hours of diversity training before being permitted to serve on a hiring committee. We're busy with all of our teaching and research, of course! The training is experienced as a needless, time-consuming interruption. We all already know, or at least think we know, pretty much what we need to know about these issues.

Diversity statements were a central topic of the workshop I attended. Examples of good and bad diversity statements were offered for our consideration. We were advised to treat the diversity statement as among the most important parts of the application. Indeed, it was suggested that we might do a preliminary screening of applications based on the diversity statement alone, removing from consideration any candidates whose diversity statements weren't excellent, before even looking at research or teaching. This particular suggestion was met with considerable hostility and incredulity among the faculty sitting near me in the back corner of the room, most of whom seemed to be bristling with rebellious anger, like the "bad" kids forced to attend some supposedly-educational high school detention hour.

What makes for a good diversity statement? I'm inclined to think that an excellent diversity statement would show concrete, detailed, and extensive evidence of one's commitment to and ability to contribute to the university's mission of promoting student and faculty diversity. Ideally, this would show in your research, and in your teaching, and in your committee service and other administrative roles, and in your broader life experience. (UC Davis offers some guidelines here; and here's some advice from Inside Higher Ed.)

I returned to my office and pulled up my own Diversity Statement (which I keep handy to trot out for various purposes when necessary). Having attended the workshop, my statement now struck me as too brief and lacking detail. It could be much better! I revised it, adding several different kinds of specific evidence of my commitment to enhancing the diversity of the university through my teaching and adding another ten or so specific pieces of evidence of my commitment to enhancing the diversity of both UCR and the profession as a whole through my research, committee service, and public philosophy. (If you follow my blog, you'll know that I have done considerable work on diversity issues.) What a wonderful Diversity Statement I had by the end, if I may say so myself, full of good, concrete evidence that I'm deeply committed to diversifying the profession!

And then I thought: How would Daniel Kaufman or Brian Leiter react a diversity statement like this? A lot of philosophers (maybe not Kaufman and Leiter in particular) might react negatively. For tenure-track job applicants especially, philosophers like Kaufman and Leiter, who believe that demographic diversity has recently been overemphasized, might understandably be hesitant to welcome colleagues who are passionately committed to the importance of demographic diversity. They might, on good grounds, fear that such a colleague would prioritize philosophy of race or Asian philosophy over some of the areas they might prefer to hire in, or that the colleague might be especially drawn to job candidates who are disabled, or women, or from other historically underrepresented groups, etc. The applicant's academic politics might be, in their view, all wrong. Such philosophers might prefer to see a bland, pro forma diversity statement that implicitly conveys the message that increasing demographic diversity is not high among the applicant's priorities.

Now I wouldn't suggest playing down your enthusiasm about diversity issues if you genuinely feel that enthusiasm. But I do think it probably makes sense to be aware that there might unfortunately be hiring contexts in which it's possible to do this part of the application too well.



Tanya Golash-Boza, in her advice at Inside Higher Ed, suggests that hiring committee members who feel that the "diversity agenda" has gone too far will tend to skip diversity statements, and so applicants needn't worry about their reactions. Hmmm... maybe?



See also Helen De Cruz's recent thoughts on the possible backfire effects of being portrayed as compassionate or as a dedicated teacher, in applications to research-oriented academic jobs.

[image source]

Friday, November 09, 2018

The Phi Value of Integrated Information Theory Might Not Be Stable Across Small Changes in Neural Connectivity

In learning and in forgetting, the amount of connectivity between your neurons changes. Throughout your life, neurons die and grow. Through all of this, the total amount of conscious experience you have, at least in your alert, attentive moments, seems to stay roughly the same. You don't lose a few neural connections and with it 80% of your consciousness. The richness of our stream of experience is stable across small variations in the connectivity of our neurons -- or so, at least, it is plausible to think.

One of the best known theories of consciousness, Integrated Information Theory, purports to model how much consciousness a neural system has by means of a value, Φ (phi), that is a mathematically complicated measure of how much "integrated information" a system possesses. The higher the Φ, the richer the conscious experience, the lower the Φ, the thinner the experience. Integrated Information Theory is subject to some worrying objections (and here's an objection by me, which I invite you also to regard as worrying). Today I want to highlight a different concern than these: the apparent failure of Φ to be robust to small changes in connectivity.

The Φ of any particular informational network is difficult to calculate, but the IIT website provides a useful tool. You can play around with networked systems of about 4, 5, or 6 nodes (above 6, the computation time to calculate Φ becomes excessive). Prefab systems are available to download, with Φ values from less than 1 to over 15. It's fun!

But there are two things you might notice, once you play around with the tool for a while:

First, it's somewhat hard to create systems with Φ values much above 1. Slap 5 nodes together and connect them any which way, and you're likely to get a Φ value between 0 and 1.

Second, if you tweak the connections of the relatively high-Φ systems, even just a little, or if you change a logical operator from one operation to another (e.g., XOR to AND), you're likely to cut the Φ value by at least half. In other words, the Φ value of these systems is not robust across small changes.

To explore the second point more systematically, I downloaded the "IIT 3.0 Paper Fig. 17 Specialized majority" network which, when all 5 nodes are lit, has a Φ value of 10.7. (A node's being "lit" means it has a starting value of "on" rather than "off".) I then tweaked the network in every way that it was possible to tweak it by changing exactly one feature. (Due to the symmetry of the network, this was less laborious than it sounds.) Turning off any one node reduces Φ to 2.2. Deleting any one node reduces Φ to 1. Deleting one connection, altering its direction (if unidirectional), or changing it from unidirectional to bidirectional or vice versa, always reduces system's Φ to a value ranging from 2.6 to 4.8. Changing the logic function of one node has effects that are sometimes minor and sometimes large: Changing any one node from MAJ to NOR reduces Φ all the way down to 0.4, while changing any one node to MIN increases Φ to 13.0. Overall, most ways of introducing one minimal perturbation into the system reduce Φ by at least half, and some reduce it by over 90%.

To confirm that the "Specialized majority" network was not unusual in this respect, I attempted a similar systematic one-feature tweaking of "CA Paper Fig 3d, Rule 90, 5 nodes". The 5-node Rule 90 network, with all nodes in the default unlit configuration, has a Φ of 15.2. The results of perturbation are similar to the results for the "Specialized majority" network. Light any one node of the rule 90 network and Φ falls to 1.8. Delete any one arrow and Φ also falls to 1.8. Change any one arrow from bidirectional to unidirectional and Φ falls to 4.8. Change the logic of one node and Φ ranges anywhere from a low of 1.8 (RAND, PAR, and >2) to a high of 19.2 (OR).

These two examples, plus what I've seen in my unsystematic tweaking of other prefab networks, plus my observations about the difficulty of casually constructing a five-node system with Φ much over 1, suggest that, in five-node systems at least, having a high Φ value requires highly specific structures that are unstable to minor perturbations. Small tweaks can easily reduce Φ by half or more.

It would be bad for Integrated Information Theory, as a theory of consciousness, if this high degree of instability in systems with high Φ values scales up to large systems, like the brain. The loss of a few neural connections shouldn't make a human being's Φ value crash down by half or more. Our brains are more robust than that. And yet I'm not sure that we should be confident that the mathematics of Φ has the requisite stability in large, high-Φ systems. In the small networks we can measure, at least, it is highly unstable.

ETA November 10:

Several people have suggested to me that Phi will be more stable to small perturbations as the size of the network increases. I could see how that might be the case (which is why I phrased the concluding paragraph as a worry rather than as an positive claim). Now if Phi, like entropy, were dependent in some straightforward way on the small contributions of many elements, that would be likely to be so. But the mathematics of Phi relies heavily on discontinuities and threshold concepts. I exploit this fact in my earlier critique of the Exclusion Postulate, in which I show that a very small change in the environment of a system, without any change interior to the system, could cause that system to instantly fall from arbitrarily high Phi to zero.

If anyone knows of a rigorous, rather than handwavy attempt to show that Phi in large systems is stable over minor perturbations, I would be grateful if you pointed it out!