Utilitarianism and Risk Amplification

A thousand utilitarian consequentialists stand before a thousand identical buttons.  If any one of them presses their button, ten people will die.  The benefits of pressing the button are more difficult to estimate.  Ninety-nine percent of the utilitarians rationally estimate that fewer than ten lives will be saved if any of them presses a button.  One percent rationally estimate that more than ten lives will be saved.  Each utilitarian independently calculates expected utility.  Since ten utilitarians estimate that more lives will be saved than lost, they press their buttons.  Unfortunately, as the 99% would have guessed, fewer than ten lives are saved, so the result is a net loss of utility.

This cartoon example illustrates what I regard as a fundamental problem with simple utilitarianism as decision procedure: It deputizes everyone to act as risk-taker for everyone else.  As long as anyone has both (a.) the power and (b.) a rational utilitarian justification to take a risk on others' behalf, then the risk will be taken, even if a majority would judge the risk not to be worth it.

Consider this exchange between Tyler Cowen and Sam Bankman-Fried (pre-FTX-debacle):

COWEN: Okay, but let’s say there’s a game: 51 percent, you double the Earth out somewhere else; 49 percent, it all disappears. Would you play that game? And would you keep on playing that, double or nothing?

BANKMAN-FRIED: With one caveat. Let me give the caveat first, just to be a party pooper, which is, I’m assuming these are noninteracting universes. Is that right? Because to the extent they’re in the same universe, then maybe duplicating doesn’t actually double the value because maybe they would have colonized the other one anyway, eventually.

COWEN: But holding all that constant, you’re actually getting two Earths, but you’re risking a 49 percent chance of it all disappearing.

BANKMAN-FRIED: Again, I feel compelled to say caveats here, like, “How do you really know that’s what’s happening?” Blah, blah, blah, whatever. But that aside, take the pure hypothetical.

COWEN: Then you keep on playing the game. So, what’s the chance we’re left with anything? Don’t I just St. Petersburg paradox you into nonexistence?

BANKMAN-FRIED: Well, not necessarily. Maybe you St. Petersburg paradox into an enormously valuable existence. That’s the other option.

There are, I think, two troubling things about Bankman-Fried's reasoning here.  (Probably more than two, but I'll restrain myself.)

First is the thought that it's worth risking everything valuable for a small chance of a huge gain.  (I call this the Black Hole Objection to consequentialism.)

Second, I don't want Sam Bankman-Fried making that decision.  That's not (just) because of who in particular he is.  I wouldn't want anyone making that decision -- at least not unless they were appropriately deputized with that authority through an appropriate political process, and maybe not even then.  No matter how rational and virtuous you are, I don't want you deciding to take risks on behalf of the rest of us simply because that's what your consequentialist calculus says.  This issue subdivides into two troubling aspects: the issue of authority and the issue of risk amplification.

The authority issue is: We should be very cautious in making decisions that sacrifice others or put them at high risk.  Normally, we should do so only in constrained circumstances where we are implicitly or explicitly endowed with appropriate responsibility.  Our own individual calculation of high expected utility (no matter how rational and well-justified) is not normally, by itself, sufficient grounds for substantially risking or harming others.

The risk amplification issue is: If we universalize utilitarian decision-making in a way that permits many people to risk or sacrifice others whenever they reasonably calculate that it would be good to do so, we render ourselves collectively hostage to whomever has the most sacrificial reasonable calculation.  That was the point illustrated in the opening scenario.

[Figure: Simplified version of the opening scenario.  Five utilitarians have the opportunity to sacrifice five people to save an unknown number of others.  The button will be pressed by the utilitarian whose estimate errs highest.  Click to enlarge and clarify.]

My point is not that some utilitarians might be irrationally risky, though certainly that's a concern.  Rather, my point is that even if all utilitarians are perfectly rational, if they differ in their assessments of risk and benefit, and if all it takes to trigger a risky action is one utilitarian with the power to choose that action, then the odds of a bad outcome rise dramatically.

Advocates of utilitarian decision procedures can mitigate this problem in a few ways, but I'm not seeing how to escape it without radically altering the view.

First, a utilitarian could adopt a policy of decision conciliationism -- that is, if you see that most others aren't judging the risk or cost worth it, adjust your own assessment of the benefits and likelihoods, so that you fall in line with the majority.  However, strong forms of conciliationism are pretty radical in their consequences; and of course this only works if the utilitarians know that there are others in similar positions deciding differently.

Second, a utilitarian could build some risk aversion and loss aversion into their calculus.  This might be a good idea on independent grounds.  Unfortunately, aversion corrections only shift the weights around.  If the anticipated gains are sufficiently high, as judged by the most optimistic rational utilitarian, they will outweigh any discounts due to risk or loss aversion.

Third, they could move to rule utilitarianism: Endorse some rule according to which you shouldn't generally risk or sacrifice others without the right kind of authority.  Plausibly, the risk amplification argument above is exactly the sort of argument that might a motivate a utilitarian to adopt rule utilitarianism as a decision procedure rather than trying to evaluate the consequences of each act individually.  That is, it's a utilitarian argument in favor of not always acting according to utilitarian calculations.  However, the risk amplification and authority problems are so broad in scope (even with appropriate qualifications) that moving to rule utilitarianism to deal with them is to abandon act utilitarianism as a general decision procedure.

Of course, one could also design scenarios in which bad things happen if everyone is a rule-following deontologist!  Picture a thousand "do not kill" deontologists who will all die unless one of them kills another.  Tragedy.  We can cherry-pick scenarios in which any view will have unfortunate results.

However, I don't think my argument is that unfair.  The issues of authority and risk amplification are real problems for utilitarian decision procedures, as brought out in these cartoon examples.  We can easily imagine, I think, a utilitarian Robespierre, a utilitarian academic administrator, Sam Bankman-Fried with his hand on the destroy-or-duplicate button, calculating reasonably, and too easily inflicting well-intentioned risk on the rest of us.

Gunkel's Criticism of the No-Relevant-Difference Argument for Robot Rights

In a 2015 article, Mara Garza and I offer the following argument for the rights of some possible AI systems:

Premise 1: If Entity A deserves some particular degree of moral consideration and Entity B does not deserve that same degree of moral consideration, there must be some relevant difference between the two entities that grounds this difference in moral status.

Premise 2: There are possible AIs who do not differ in any such relevant respects from human beings.

Conclusion: Therefore, there are possible AIs who deserve a degree of moral consideration similar to that of human beings.

The argument is, we think, appealingly minimalist, avoiding controversial questions about the grounds of moral status.  Does human-like moral status require human-like capacity for pain or pleasure (as classical utilitarians would hold)?  Or human-like rational cognition, as Kant held?  Or the capacity for human-like varieties of flourishing?  Or the right types of social relations?

The No-Relevant-Difference Argument avoids these vexed questions, asserting only that whatever grounds moral status can be shared between robots and humans.  This is not an entirely empty claim about the grounds of moral status.  For example, the argument commits to denying that membership in the species Homo sapiens, or having a natural rather than artificial origin, is required for human-like moral status.

Compare egalitarianism about race and gender.  We needn't settle tricky questions about the grounds of moral status to know that all genders and races deserve similar moral consideration!  We need only know this: Whatever grounds moral status, it's not skin color, or possession of a Y chromosome, or any of the other things that might be thought to distinguish among the races or genders.

Garza and I explore four arguments for denying Premise 2 -- that is, for thinking that robots would inevitably differ from humans in some relevant respect.  We call these the objections from Psychological Difference, Duplicability, Otherness, and Existential Debt.  Today, rather than discussing Premise 2, I want to discuss David Gunkel's objection to our argument in his just-released book, Person, Thing, Robot.

[Image of Ralph and Person, Thing, Robot.  Ralph is a sculpture designed to look like an old-fashioned robot, composed of technological junk from the mid-20th century (sculptor: Jim Behrman).  I've named him after my father, whose birth name was Ralph Schwitzgebel.  My father was also a tinkerer and artist with technology from that era.]  

Gunkel acknowledges that the No-Relevant-Difference Argument "turns what would be a deficiency... -- [that] we cannot positively define the exact person-making qualities beyond a reasonable doubt -- into a feature" (p. 91).  However, he objects as follows:

The main difficulty with this alternative, however, is that it could just as easily be used to deny human beings access to rights as it could be used to grant rights to robots and other nonhuman artifacts.  Because the no relevant difference argument is theoretically minimal and not content dependent, it cuts both ways.  In the following remixed version, the premises remain intact; only the conclusion is modified.

Premise 1: If Entity A deserves some particular degree of moral consideration and Entity B does not deserve that same degree of moral consideration, there must be some relevant difference between the two entities that grounds this difference in moral status.

Premise 2: There are possible AIs who do not differ in any such relevant respects from human beings.

Conclusion: Therefore, there are possible human beings who, like AI systems, do not deserve moral consideration. 

In other words, the no relevant difference argument can be used either to argue for an extension of rights to other kinds of entities, like AI systems, robots, and artifacts, or, just as easily, to justify dehumanization, reification of human beings, and the exclusion and/or marginalization of others (p. 91-92, italics added).

This is an interesting objection.  However, I reject the appropriateness of the repeated phrase "just as easily", which I have italicized in the block quote.

----------------------------------------------------------------

As the saying goes, one person's modus ponens is another's modus tollens.  Suppose you know that A implies B.  Modus ponens is an inference rule which assumes the truth of A and concludes that B must also be true.  Modus tollens is an inference rule which assumes the falsity of B and concludes that A must also be false.  For example, suppose you can establish that if anyone stole the cookies, it was Cookie Monster.  If you know that the cookies were stolen, modus ponens unmasks Cookie Monster as the thief.  If, on the other hand, you know that Cookie Monster has committed no crimes, modus tollens assures you that the cookies remain secure.

Gunkel correctly recognizes that the No Relevant Difference Argument can be reframed as a conditional: Assuming that human X and robot Y are similar in all morally relevant respects, then if human X deserves rights so also does robot Y.  This isn't exactly how Garza and I frame the argument -- our framing implicitly assumes that there is a standard level of moral consideration for human beings in general -- but it's a reasonable adaptation for someone wants to leave open the possibility that different humans deserve different levels of moral consideration.

In general, the plausibility of modus ponens vs modus tollens depends on the relative security of A vs not-B.  If you're rock-solid sure the cookies were stolen and have little faith in Cookie Monster's crimelessness, then ponens is the way to go.  If you've been tracking Cookie all day and know for sure he couldn't have committed a crime, then apply tollens.  The "easiness", so to speak, of ponens vs. tollens depends on one's confidence in A vs. not-B.

Few things are more secure in ethics than at least some humans deserve substantial moral consideration.  This gives us the rock-solid A that we need for modus ponens.  As long as we are not more certain all possible robots would not deserve rights than that some humans do deserve rights, modus ponens will be the correct move.  Ponens and tollens will not be equally "easy".

Still, Gunkel's adaptation of our argument does reveal a potential for abuse, which I had not previously considered, and which I thank him for highlighting.  Anyone who is more confident that robots of a certain sort are undeserving of moral consideration than they are of the moral considerability of some class of humans could potentially combine our No Relevant Difference principle with an appeal to the supposed robotlikeness of those humans to deny rights to those humans.

I don't think the No Relevant Difference principle warrants skepticism on those grounds.  Compare application of a principle like "do unto others as you would have them do unto you".  Although one could in principle reason "I want to punch him in the nose, so I guess I should punch myself in the nose", the fact that some people might potentially run such a tollens reveals more about their minor premises than it does about the Golden Rule.

I hope that such an abuse of the principle would be in any case rare.  People who want to deny rights to subgroups of humans will, I suspect, be motivated by other considerations, and appealing to those people's putative "robotlikeness" would probably be only an afterthought or metaphor.  Almost no one, I suspect, will be on the fence about the attribution of moral status to some group of people and then think, "whoa, now that I consider it, those people are like robots in every morally relevant respect, and I'm sure robots don't deserve rights, so tollens it is".  If anyone is tempted by such reasoning, I advise them to rethink the path by which they find themselves with that peculiar constellation of credences.

Strange Intelligence, Strange Philosophy

AI intelligence is strange -- strange in something like the etymological sense of external, foreign, unfamiliar, alien.  My PhD student Kendra Chilson (in unpublished work) argues that we should discard the familiar scale of subhuman → human-grade → superhuman.  AI systems do, and probably will continue to, operate orthogonally to simple scalar understandings of intelligence modeled on the human case.  We should expect them, she says, to be and remain strange intelligence[1] -- inseparably combining, in a single package, serious deficits and superhuman skills.  Future AI philosophers will, I suspect, prove to be strange in this same sense.

Most readers are probably familiar with the story of AlphaGo, which in 2016 defeated the world champion player of the game of go.  Famously, in the series of matches (which it won 4-1), it made several moves that human go experts regarded as bizarre -- moves that a skilled human go player would never have made, and yet which proved instrumental in its victory -- while also, in its losing match, making some mistakes characteristic of simple computer programs, which go experts know to avoid.

Similarly, self-driving cars are in some respects better and safer drivers than humans, while nevertheless sometimes making mistakes that few humans would make.

Large Language Models have stunning capacity to swiftly create competent and even creative texts on a huge breadth of topics, while still failing conspicuously in some simple common sense tasks. they can write creative-seeming poetry and academic papers, often better than the average first-year university student.  Yet -- borrowing an example from Sean Carroll -- I just had the following exchange with GPT-4 (the most up-to-date version of the most popular large language model):

GPT-4 seems not to recognize that a hot skillet will be plenty cool by the next day.

I'm a "Stanford school" philosopher of science.  Core to Stanford school thinking is this: The world is intractably complex; and so to deal with it, we limited beings need to employ simplified (scientific or everyday) models and take cognitive shortcuts.  We need to find rough patterns in go, since we cannot pursue every possible move down every possible branch.  We need to find rough patterns in the chaos of visual input, guessing about the objects around us and how they might behave.  We need quick-and-dirty ways to extract meaning from linguistic input in the swift-moving world, relating it somehow to what we already know, and producing linguistic responses without too much delay.  There will be different ways of building these simplified models and implementing these shortcuts, with different strengths and weaknesses.  There is rarely a single best way to render the complexity of the world tractable.  In psychology, see also Gigerenzer on heuristics.

Now mix Stanford school philosophy of science, the psychology of heuristics, and Chilson's idea of strange intelligence.  AI, because it is so different from us in its underlying cognitive structure, will approach the world with a very different set of heuristics, idealizations, models, and simplifications than we do.  Dramatic outperformance in some respects, coupled with what we regard as shockingly stupid mistakes in others, is exactly what we should expect.

If the AI system makes a visual mistake in judging the movement of a bus -- a mistake (perhaps) that no human would make -- well, we human beings also make visual mistakes, and some of those mistakes, perhaps, would never be made by an AI system.  From an AI perspective, our susceptibility to the Muller-Lyer illusion might look remarkably stupid.  Of course, we design our driving environment to complement our vision: We require headlights, taillights, marked curves, lane markers, smooth roads of consistent coloration, etc.  Presumably, if society commits to driverless cars, we will similarly design the driving environment to complement their vision, and "stupid" AI mistakes will become rarer.

I want to bring this back to the idea of an AI philosopher.  About a year and a half ago, Anna Strasser, Matthew Crosby, and I built a language model of philosopher Daniel Dennett.  We fine-tuned GPT-3 on Dennett's corpus, so that the language model's outputs would reflect a compromise between the base model of GPT-3 and patterns in Dennett's writing.  We called the resulting model Digi-Dan.  In a study collaborative with my son David, we then posed philosophical questions to both Digi-Dan and the actual Daniel Dennett.  Although Digi-Dan flubbed a few questions, overall it performed remarkably well.  Philosophical experts were often unable to distinguish Digi-Dan's answers from Dennett's own answers.

Picture now a strange AI philosopher -- DigiDan improved.  This AI system will produce philosophical texts very differently than we do.  It need not be fully superhuman in its capacities to be interesting.  It might even, sometimes, strike us as remarkably, foolishly wrong.  (In fairness, other human philosophers sometimes strike me the same way.)  But even if subhuman in some respects, if this AI philosopher also sometimes produces strange but brilliant texts -- analogous to the strange but brilliant moves of AlphaGo, texts that no human philosopher would create but which on careful study contain intriguing philosophical moves -- it could be a philosophical interlocutor of substantial interest.

Philosophy, I have long argued, benefits from including people with a diversity of perspectives.  Strange AI might also be appreciated as a source of philosophical cognitive diversity, occasionally generating texts that contain sparks of something genuinely new, different, and worthwhile that would not otherwise exist.

------------------------------------------------

[1] Kendra Chilson is not the first to use the phrase "strange intelligence" with this meaning in an AI context, but the usage was new to me; and perhaps through her work it will catch on more widely.

Skeletal vs Fleshed-Out Philosophy

All philosophical views are to some degree skeletal. By this, I mean that the details of their application remain to some extent open. This is true of virtually any formal system: Even the 156-page rule handbook for golf couldn't cover every eventuality: What if the ball somehow splits in two and one half falls in the hole? What if an alien spaceship levitates the ball for two seconds as it's arcing through the air? (See the literature on "open textured" statements.)

Still, some philosophical views are more skeletal than others. A bare statement like "maximize utility" is much more skeletal, much less fleshed out, than a detailed manual of utilitarian consequentialist advice. Today, I want to add a little flesh to the skeletal vs. fleshed-out distinction. Doing so will, I hope, help clarify some of the value of trying to walk the walk as an ethicist. (For more on walking the walk, see last month's posts here and here.)

[Midjourney rendition of a person and a skeleton talking philosophy, against a background of stars]

Using "maximize utility" as an example, let's consider sources of linguistic, metaphysical, and epistemic openness.

Linguistic: What does "utility" mean, exactly? Maybe utility is positively valenced conscious experiences. Or maybe utility is welfare or well-being more broadly construed. What counts as "maximizing"? Is it a sum or a ratio? Is the scope truly universal -- for all entities in the entire cosmos over all time, or is it limited in some way (e.g., to humans, to Earth, to currently existing organisms)? Absent specification (by some means or other), there will be no fact of the matter whether, say, two acts with otherwise identical results, but one of which also slightly improves the knowledge (but not happiness) of one 26th-century Martian, are equally choiceworthy according to the motto. 

Metaphysical: Consider a broad sense of utility as well-being or flourishing. If well-being has components that are not strictly commensurable -- that is, which cannot be precisely weighed against each other -- then the advice to maximize utility leaves some applications open. Plausibly, experiencing positive emotions and achieving wisdom (whatever that is, exactly) are both part of flourishing. While it might be clear that a tiny loss of positive emotion is worth trading off for a huge increase in wisdom and vice versa, there might be no fact of the matter exactly what the best tradeoff ratio is -- and thus, sometimes, no fact of the matter whether someone with moderate levels of positive emotion and moderate levels of wisdom has more well-being than someone with a bit less positive emotion and a bit more wisdom.

Epistemic: Even absent linguistic and metaphysical openness, there can be epistemic openness. Imagine we render the utilitarian motto completely precise: Maximize the total sum of positive minus negative conscious experiences for all entities in the cosmos in the entire history of the cosmos (and whatever else needs precisification). Posit that there is always an exact fact of the matter how to weigh competing goods in the common coin of utility and there are never ties. Suppose further that it is possible in principle to precisely specify what an "action" is, individuating all the possible alternative actions at each particular moment. It should then always be the case that there is exactly one action you could do that would "maximize utility". But could you know what this action is? That's doubtful! Every action has a huge number of non-obvious consequences. This is ignorance; but we can also think of it as a kind of openness, to highlight its similarity to linguistic and metaphysical openness or indeterminacy. The advice "maximize utility", however linguistically and metaphysically precise, leaves it still epistemically open what you should actually do.

Parallel remarks apply to other ethical principles: "Act on that maxim that you can will to be a universal law", "be kind", "don't discriminate based on race", "don't perform medical experiments on someone without their consent" -- all exhibit some linguistic, metaphysical, and epistemic openness.

Some philosophers might deny linguistic and/or metaphysical openness: Maybe context always renders meanings perfectly precise, and maybe normative facts are never actually mushy-edged and indeterminate. Okay. Epistemic openness will remain. As long as we -- the reader, the consumer, the applier, of the philosophical doctrine -- can't reasonably be expected to grasp the full range of application, the view remains skeletal in my sense of the term.

It's not just ethics. Similar openness also pervades other areas of philosophy. For example, "higher order" theories of consciousness hold that an entity is conscious if and only if it has the right kind of representations of or knowledge of its own mental states or cognitive processes. Linguistically, what is meant by a "higher order representation", exactly? Metaphysically, might there be borderline cases that are neither determinately conscious nor unconscious? Epistemically, even if we could precisify the linguistic and metaphysical issues, what actual entities or states satisfy the criteria (mice? garden snails? hypothetical robots of various configurations?).

The degree of openness of a position is itself, to some extent, open: There's linguistic, metaphysical, and epistemic meta-openness, we might say. Even a highly skeletal view rules some things out. No reasonable fleshing out of "maximize utility" is consistent with torturing babies for no reason. But it's generally unclear where exactly the boundaries of openness lie, and there might be no precise boundary to be discovered.

#

Now, there's something to be said for skeletal philosophy. Simple maxims, which can be fleshed out in various ways, have an important place in our thinking. But at some point, the skeleton needs to get moving, if it's going to be of use. Lying passively in place, it might block a few ideas -- those that crash directly against its obvious bones. But to be livable, applicable, it needs some muscle. It needs to get up and walk over to real, specific situations. What does "maximize utility" (or whatever other policy, motto, slogan, principle) actually recommend in this particular case? Too skeletal a view will be silent, leaving it open.

Enter the policy of walking the walk. As an ethicist, attempting to walk the walk forces you to flesh out your view, applied at least to the kinds of situations you confront in your own life -- which will of course be highly relevant to you and might also be relevant to many of your readers. What actions, specifically, should a 21st-century middle-class Californian professor do to "maximize utility"? Does your motto "be kind" require you to be kind to this person, in this particular situation, in this particular way? Confronting actual cases and making actual decisions motivates you to repair your ignorance about how the view would best apply to those cases. Linguistically, too, walking the walk enables you to make the content of your mottoes more precise: "Be kind" means -- in part -- do stuff like this. In contrast, if you satisfy yourself with broad slogans, or broad slogans plus a few paragraph-long thought-experimental applications, your view will never be more than highly skeletal. 

Not only our readers, but also we philosophers ourselves, normally remain substantially unclear on what our skeletal mottoes really amount to until we actually try to apply them to concrete cases. In ethics -- at least concerning principles meant to govern everyday life (and not just rare or remote cases) -- the substance of one's own life is typically the best and most natural way to add that flesh.

Elisabeth of Bohemia 1, Descartes 0

I'm loving reading the 1643 correspondence between Elisabeth of Bohemia and Descartes! I'm embarrassed to confess that I hadn't read it before now; the standard Cottingham et al. edition presents only selections from Descartes' side. I'd seen quotes of Elisabeth, but not the whole exchange as it played out. Elisabeth's letters are gems. She has Descartes on the ropes, and she puts her concerns so plainly and sensibly (in Bennett's translation; I haven't attempted to read the antique French). You can practically feel Descartes squirming against her objections. I have a clear and distinct idea of Descartes ducking and dodging!

Here's my (somewhat cheeky) summary, with comments and evaluation at the end.

Elisabeth, May 6, 1643:

I'm so ignorant and you're so learned! Here's what I don't understand about your view: How can an immaterial soul, simply by thinking, possibly cause a bodily action?

Specifically,

it seems that how a thing moves depends solely on (i) how much it is pushed, (ii) the manner in which it is pushed, or (iii) the surface-texture and shape of the thing that pushes it. The first two of those require contact between the two things, and the third requires that the causally active thing be extended [i.e., occupy a region of space]. Your notion of the soul entirely excludes extension, and it appears to me that an immaterial thing can't possibly touch anything else.

Also, if, as you say, thinking is the essential property of human souls, what about unborn children and people who have fainted, who presumably have souls without thinking?

René, May 21, 1643:

Admittedly in my writings I talk much more about the fact that the soul thinks than about the question of how it is united with the body. This idea of the union of the soul and the body is basic and can be understood only through itself. It's so easy to get confused by using your imagination or trying to apply notions that aren't appropriate to the case!

For a comparison, however, think about how the weight of a rock moves it downwards. One might (mistakenly, I hope later to show) think of weight as a "real quality" about which we know nothing except that it has the power to move the body toward the centre of the earth. The soul's power to move the body is analogous.

Elisabeth, June 10, 1643:

Please forgive my stupidity! I wish I had the time to develop your level of expertise. But why should I be persuaded that an immaterial soul can move a material body by this analogy to weight? If we think in terms of the old idea of weight, why shouldn't we then conclude by your reasoning that things move downward due to the power of immaterial causes? I can't conceive of "what is immaterial" except negatively as "what is not material" and as what can't enter into causal relations with matter. I'd rather concede that the soul is material than that an immaterial thing could move a body.

René, May 28, 1643:

This matter of the soul's union with the body is a very dark affair when it comes from the intellect (whether alone or aided by the imagination). People who just use their senses, in the ordinary course of life, have no doubt that the soul moves the body. We shouldn't spend too much time in intellectual thinking. In fact,

I never spend more than a few hours a day in the thoughts involving the imagination, or more than a few hours a year on thoughts that involve the intellect alone. I give all the rest of my time to the relaxation of the senses and the repose of the mind.

The human mind can't clearly conceive the soul's distinctness from the body and its union with the body simultaneously. The comparison with weight was imperfect, but without philosophizing everyone knows that they have body and thought and that thought can move the body.

But since you remark that it is easier to attribute matter and extension to the soul than to credit it with the capacity to move and be moved by the body without having matter, please feel free to attribute this matter and extension to the soul -- because that's what it is to conceive it as united to the body.

Still, once you do this, you'll find that matter is not thought because the matter has a definite location, excluding other matter. But again, thinking too much about metaphysics is harmful.

Elisabeth, July 1, 1643:

I hope my letters aren't troubling you.

I find from your letter that the senses show me that the soul moves the body, but as for how it does so, the senses tell me nothing about that, any more than the intellect and imagination do. This leads me to think that the soul has properties that we don't know -- which might overturn your doctrine... that the soul is not extended.

As you have emphasized in your writings, all our errors come from our forming judgments about things we don't perceive well enough. Since we can't perceive how the soul moves the body, I am left with my initial doubt, that is, my thinking that perhaps after all the soul is extended.

There is no record of a reply by Descartes.

---------------------------------------

Zing! Elisabeth shows up so much better than Descartes in this exchange. She immediately homes in on the historically most important (and continuing) objection to Cartesian substance dualism: the question of how, if at all, an immaterial soul and a material object could causally interact. She efficiently and elegantly formulates a version of the principle of "the causal closure of the physical", according to which material events can only be caused by other material events, connecting that idea both with Descartes' denial that the soul is extended in space and with the view, widely accepted by early modern philosophers before Newton, that physical causation requires direct physical contact (no "action at a distance"). Jaegwon Kim notes (2011, p. 49) that hers might be the first causal argument for a materialist view of the mind. To top it off, she poses an excellent objection (from fetuses and fainting spells) to the idea that thinking is essential to having a soul.

Descartes' reply by analogy to weight is weak. As Elisabeth notes, it doesn't really answer the question of how the process is supposed to work for souls. Descartes' own theory of weight (articulated the subsequent year in Principles of Philosophy, dedicated to Elisabeth) involves action by contact (light particles spinning off the rotating Earth shoot up, displacing heavier particles down: IV.20-24). At best, Descartes is saying that the false, old idea of weight didn't involve contact, so why not think souls can also have influence without contact? Elisabeth's reply implicitly suggests a dilemma: If downward motion is by contact, then weight is not an example of how causation without contact is possible. If downward motion is not by contact, then shouldn't we think (absurdly?) that things move down due to the action of immaterial souls? She also notes that "immaterial" just seems to be a negative idea, not something we can form a clear, positive conception of.

Elisabeth's response forces Descartes concede that we can't in fact think clearly and distinctly about these matters. This is a major concession, given the centrality of the standard of "clear and distinct" ideas to Descartes' philosophy. He comes off almost as a mysterian! He also seems to partly retract what is perhaps the most central idea in his dualist metaphysics -- that the soul does not have extension. Elisabeth should feel free to attribute matter and extension to the soul, after all! Indeed, in saying that attributing matter and extension is "what it is to conceive [the soul] as united to the body", Descartes seriously muddies the interpretation of his positive view about the nature of souls.

It's also worth noting that Descartes entirely ignores Elisabeth's excellent fetus and fainting question.

I had previously been familiar with Descartes' famous quote that he spends no more than a few hours a year on thoughts involving the intellect alone; but reading the full exchange provides interesting context. His aim in saying that is to convince Elisabeth not to put too much energy into objecting to his account of how the soul works.

Understandably, Elisabeth is dissatisfied. She even gestures (though not in so many words) toward Descartes' methodological self-contradiction: Descartes famously says that philosophizing requires that we have clear ideas and that our errors all arise from failure to do so -- yet here he is, saying that there's an issue at the core of his metaphysics about which it's not possible to think clearly! Shouldn't he admit, then, that on this very point he's liable to be mistaken?

If Descartes attempted a further reply, the reply is lost. Their later correspondence treats other issues.

The whole correspondence is just 15 pages, so I'd encourage you to read it yourself. This summary necessarily omits interesting detail and nuance. In this exchange, Elisabeth is by far the better philosopher.

[image source]

Percentage of Women Philosophy Majors Has Risen Sharply Since 2016 -- Why? Or: The 2017 Knuckle

Back in 2017, I noticed that the percentage of women philosophy majors in the U.S. had been 30%-34% for "approximately forever". That is, despite the increasing percentage of Bachelor's degrees awarded to women overall and in most other majors, the percentage of philosophy Bachelor's degrees awarded to women had been remarkably steady from the first available years (1986-1987) in the NCES IPEDS database through the then-most-recent data year (2016).

In the past few years, however, I have noticed some signs of change. The most recent NCES IPEDS data release, which I analyzed this morning, statistically solidifies the trend. Women now constitute over 40% of philosophy Bachelor's degree recipients. I would argue that this is a very material change from the long-standing trend of 30-34%. If parity is 50%, a change from 32% women to 41% women constitutes a halving of the disparity. Furthermore, the change has been entirely in the most recent six years' of data -- remarkably swift for this type of demographic shift.

The chart below shows the historical trend through the most recent available year (2022). I've marked the 30%-34% band with thick horiztonal lines. A thin vertical line marks 2017, the first year to cross the 34% mark (34.9%). The most recent years are 41.4% and 41.3% respectively.

[click to enlarge and clarify]

Given the knuckle-like change in the slope of the graph, let's call this the 2017 Knuckle.

What I find puzzling is why?

This doesn't reflect an overall trend of increasing percentages of women across majors. Overall, women have been 56%-58% of Bachelor's degree recipients throughout the 21st century. Most other humanities and social sciences had a much earlier increase in the proportion of women.

However, interestingly, the physical sciences and engineering, which have also tended to be disproportionately men, have showed some similar trends. Since 2010, physics majors have increased from 40% to 45% women -- with all of that increase being since 2017. Since 2010, Engineering has increased from 18% to 25% women, with the bulk of the increase since 2016. Since 2010, "Engineering Technologies and Engineering-related Fields" (which NCES classifies separately from Engineering) has also increased from 10% to 15% women, again with most of the increase since 2016. Among the humanities and social sciences, Economics is maybe the only large major similar to Philosophy in gender disparity, and in Economics we see a similar trend, though smaller: an increase from 31% to 35% women between 2010 and 2022, again with most of the gain since 2016.

Since people tend to decide their majors a few years before graduating, whatever explains these trends must have begun in approximately 2013-2016, then increased through at least 2020. Any hypotheses?

It's probably not a result of change in the percentage of women faculty: Faculty turnover is slow, and at least in philosophy the evidence suggests a slow increase over the decades, rather than a knuckle. (Data are sparser and less reliable on this issue, but see here, here and here.) There also wasn't much change in the 2010s in the percentage of women earning Philosophy PhDs in the U.S.

A modeling hypothesis would suggest that change in the percentage of women philosophy majors is driven by a change in the percentage of women faculty and TAs in Philosophy. In contrast, a pipeline hypothesis predicts that change in the percentage of women philosophy majors leads to a change in the percentage of women graduate students and (years later) faculty. Both hypotheses posit a relationship between women undergraduates and women instructors, but with different directions of causation. (The hypotheses aren't, of course, incompatible: Causation might flow both ways.) At least in Philosophy, the modeling hypothesis doesn't seem to explain the 2017 Knuckle. Concerning the pipeline, it's too early to tell, but when the NSF releases their data on doctorates in October, I'll look for preliminary signs.

I'm also inclined to think -- though I'm certainly open to evidence -- that feminism has been slowly, steadily increasing in U.S. culture, rather than being more or less flat since the late 1980s and recently increasing again. So a general cultural increase in feminist attitudes wouldn't specifically explain the 2017 Knuckle. Now it is true that 2015-2017 saw the rise of Trump, and the backlash against Trump, as well as the explosion of the #MeToo movement. Maybe that's important? It would be pretty remarkable if those cultural events had a substantial effect on the percentage of women undergraduates declaring Philosophy, Economics, Physics, and Engineering majors.

Further thoughts? What explains the 2017 Knuckle?

It could be interesting to look at other countries, and at race/ethnicity data, and at majors that tend to be disproporately women -- patterns there could potentially cast light on the effect -- but enough for today.

---------------------------------------------------

Methodological notes: NCES IPEDS attempts to collect data on every graduating student in accredited Bachelor's programs in the U.S., using administrator-supplied statistics. Gender categories are binary "men" and "women" with no unclassified students. Data are limited to "U.S. only" institutions in classification category 38.01 ("Philosophy") and include both first and second majors back through 2001. Before 2001, only first majors are available. Each year includes all graduates during the academic year ending in that year (e.g., 2022 includes all students from the 2021-2022 academic year). For engineering and physical sciences, I used major catories 15, 16, and 40; and for Economics, 45.06.

Walking the Walk: Frankness and Social Proof

My last two posts have concerned the extent to which ethicists should "walk the walk" -- that is, live according to, or at least attempt to live according to, the ethical principles they espouse in their writing and teaching. According to "Schelerian separation", what ethicists say or write can and should be evaluated independently of facts about the ethicist's personal life. While there are some good reasons to favor Schelerian separation, I argued last week that ethical slogans ("act on that maxim you can at the same time will to be a universal law", "maximize utility") will tend to lack specific, determinate content without a context of clarifying examples. One's own life can be a rich source of content-determining examples, while armchair reflection on examples tends to be impoverished.

Today, I'll discuss two more advantages of walking the walk.

[a Dall-E render of "walking the walk"]

Frankness and Belief

Consider scientific research. Scientists don't always believe their own conclusions. They might regard their conclusions as tentative, the best working model, or just a view with enough merit to be worth exploring. But if they have doubt, they ought to be unsurprised if their readers also have doubt. Conversely, if a reader learns that a scientist has substantial doubts about their own conclusions, it's reasonable for the reader to wonder why, to expect that the scientist is probably responding to limitations in their own methods and gaps in their own reasoning that might be invisible to non-experts.

Imagine reading a scientific article, finding the conclusion wholly convincing, and then learning that the scientist who wrote the article thinks the conclusion is probably not correct. Absent some unusual explanation, you’ll probably want to temper your belief. You’ll want to know why the scientist is hesitating, what weaknesses and potential objections they might be seeing that you have missed. It’s possible that the scientist is simply irrationally unconvinced by their own compelling reasoning; but that’s presumably not the normal case. Arguably, readers of scientific articles are owed, and reasonably expect, scientific frankness. Scientists who are not fully convinced by their results should explain the limitations that cause them to hesitate. (See also Wesley Buckwalter on the "belief norm of academic publishing".)

Something similar is true in ethics. If Max Scheler paints a picture of a beautiful, ethical, religious way of life which he personally scorns, it's reasonable for the reader to wonder why he scorns it, what flaws he sees that you might not notice in your first read-through. If he hasn't actually tried to live that way, why not? If he has tried, but failed, why did he fail? If a professional ethicist argues that ethically, and all things considered, one should be a vegetarian, but isn't themselves a vegetarian and has no special medical or other excuse, it's reasonable for readers and students to wonder why not and to withhold belief until that question is resolved. People are not usually baldly irrational. It's reasonable to suppose that there's some thinking behind their choice, which they have not yet revealed readers and students, which tempers or undercuts their reasoning.

As Nomy Arpaly has emphasized in some of her work, our gut inclinations are sometimes wiser than our intellectual affirmations. The student who says to herself that she should be in graduate school, that academics is the career for her, but who procrastinates, self-sabotages, and hates her work – maybe the part of her that is resisting the career is the wiser part. When Huck Finn tells himself that the right thing to do is to turn in his friend, the runaway slave Jim, but can't bring himself to do it – again, his inclinations might be wiser than his explicit reasoning.

If an ethicist's intellectual arguments aren't penetrating through to their behavior, maybe there's a good reason. If you can't, or don't, live what you intellectually endorse, it could be because your intellectual reasoning is leaving something important out that the less intellectual parts of you rightly refuse to abandon. Frankness with readers enables them to consider this possibility. Conversely, if we see someone who reasons to a certain ethical conclusion, and their reasoning seems solid, and then they consistently live that way without tearing themselves apart with ambivalence, we have less grounds for suspecting that their gut might be wisely fighting against flaws their academic reasoning than we do when we see someone who doesn’t walk the walk.

What is it to believe that eating meat is morally wrong (or any other ethical proposition)? I favor a dispositionalist approach (e.g., here, here, here). It is in part to be disposed to say and intellectually judge that eating meat is morally wrong. But more than that, it is to give weight to the avoidance of meat in your ethical decision-making. It is to be disposed to feel you have done something wrong if you eat meat for insufficient reason, maybe feeling guilt or shame. It is to feel moral approval and disapproval of others' meat-avoiding or meat-eating choices. If an ethicist intellectually affirms the soundness of arguments for vegetarianism but lacks the rest of this dispositional structure, then (on the dispositionalist view I favor) they don't fully or determinately believe that eating meat is ethically wrong. Their intellectually endorsed positions don't accurately reflect their actual beliefs and values. This completes the analogy with the scientist who doesn't believe their own conclusions.

Social Proof

Somewhat differently, an ethicist's own life can serve as a kind of social proof. Look: This set of norms is livable – maybe appealing so, with integrity. Things don't fall apart. There's an implementable vision, which other people could also follow. Figures like Confucius, Buddha, and Jesus were inspiring in part because they showed what their slogans amounted to in practice, in part because they showed that real people could live in something like the way they themselves lived, and in part because they also showed how practically embodying the ethics they espoused could be attractive and fulfilling, at least to certain groups of people.

Ethical Reasons to Walk the Walk?

I haven't yet discussed ethical reasons for walking the walk. So far, the focus has been epistemology, philosophy of language, and philosophy of mind. However, arguing in favor of certain ethical norms appears to involve recommending that others adhere to those norms, or at least be partly motivated by those norms. Making such a recommendation while personally eschewing those same norms plausibly constitutes a failure of fairness, equity, or universalization – the same sort of thing that rightly annoys children when their parents or teachers say "do as I say, not as I do". More on this, I hope, another day.

One Reason to Walk the Walk: To Give Specific Content to Your Assertions

Last week, I discussed some reasons we might not expect or want professional ethicists to "walk the walk" in the sense of living by the ethical norms they espouse in their teaching and research. (In short: This isn't their professional obligation; it's reasonable for them to trust convention more than their academic conclusions; and one can arguably be more objective in evaluating arguments if one isn't obligated to modify one's life depending on those conclusions.) Today I want to start talking about why I think that's too simple.

To be clear: I just want to start talking about it. I'll give one reason why I think there's some benefit to walking the walk, as an ethicist. I don't intend this as a full account.

Short version: Ethical slogans lack concrete, practical meaning unless they are grounded in a range of examples. One's own life can provide that range of examples, putting flesh on the blood or your slogans. If you say "act on that maxim that you can at the same time will to be a universal law", I have no idea what you are specifically recommending -- and I worry that you might not have much idea either. But if you put it to work in your life, then what it amounts to, at least as expressed by you, becomes much clearer.

Longer version:

Love Is Love, and Slogans Require a Context

A few years ago, signs like this began to sprout up in my neighborhood:

In this house, we believe:
Black lives matter
Women’s rights are human rights
No human is illegal
Science is real
Love is love
Kindness is everything

If you know the U.S. political scene, you'll understand that the first five of these slogans have meanings much more specific than is evident from the surface content alone. "Black lives matter" conveys a belief that great racial injustice still exists in the U.S., perpetrated especially by the police, and it recommends taking action to rectify that injustice. "Women's rights are human rights" conveys a similar belief about continuing gender inequality, especially with respect to reproductive rights, including access to abortion. "No human is illegal" expresses concern over the mistreatment of people who have entered the U.S. without legal permission. "Science is real" expresses disdain for mainstream Republicans' dismissal of scientific evidence in policy, especially concerning climate change. And "love is love" expresses the view that heterosexual romantic relationships should not be privileged above homosexual romantic relationships, especially with regard to the rights of marriage. "Kindness is everything" is also interesting, and I'll get to it in a moment.

How confusing and opaque all of this would be to an outsider! Imagine a time traveler from the 19th century. "Love is love". Well, of course! Isn't that just a tautology? Who could disagree? Explain the details, however, and our 19th century guest might well disagree. The content of this slogan, or "belief", is radically underspecified by the explicit linguistic content. Another feature of these claims is that they sound less controversial in the abstract than they do after contextual specification. The surface content of both "Black lives matter" and the opposing rallying cry, "all lives matter" is unobjectionable. However, whether special attention should be dedicated to anti-Black police violence, or whether instead pro-Black protesters have gone too far -- that's quite another matter.

The last slogan, "kindness is everything", is to my knowledge less politically specific, but it illustrates a connected point. Clearly, it expresses support for increasing kindness. But kindness isn't literally everything, certainly not ontologically, nor even morally, unless something extremely thin is meant by "kindness". If a philosopher were to espouse this slogan, I'd immediately want to work through examples with them, to assess what this claim amounts to. If I give an underperforming student the C-minus they deserve instead of the A they want, am I being kind to them, in the intended sense? How about if I object to someone's stepping on my toe? Of course, these sketchy questions lack detail, since there are many ways to step on someone's toe, and many ways to object, and many different circumstances in which toe-stepping might be embedded, and not all C-minus situations are the same. Working through abstract examples, though, at least gets us started on what counts as "kindness" and what priority it should have when it appears to conflict with other goods.

But here's what would really make the slogan clear: a life lived in kindness -- an observable pattern of reactions to a wide range of complex situations. How does the person who embodies the slogan "kindness is everything" react to having their toe stepped on, in this particular way by this particular person? Show me specific kindness-related situations over and over, with all the variation that life brings. Only then will I really understand the ideal. We can do this sometimes in imagination, developing a feel for someone's character and way of life. In a richly imagined fiction, or in a set of stories about Confucius or Jesus or some other sage, we can begin to see the substance of a moral view and set of values, going beyond the slogans.

In the Declaration of Independence, Thomas Jefferson, patriot, revolutionary, and slaveowner, wrote "All men are created equal". This sounds good. People in the U.S. endorse that slogan, repeat it, embrace it in all sincerity. What does it mean? All "men" in the old-fashioned sense that supposedly also included women, or really only men? Black and cognitively disabled people too? And in what does equality consist? Does it mean that all adults should have the right to vote? Equal treatment before the law? Certain rights and liberties? What is the function of "created" in the sentence? Do we start equal but diverge? We could try to answer all these questions, and new more specific questions would spring forth, hydra-like (which laws specifically, under which conditions?) until we tack it down in a range of examples. The framers of the U.S. Constitution certainly didn't agree on all of these matters, especially the question of slavery. They could agree on the slogan while disagreeing radically about what it amounts to, because the slogan is neither "self-evident" nor determinate in its content. In one precisification, it might be only some banal thing even King George III would have accepted. In another precisification, it might entail universal franchise and the immediate abolition of slavery, in which case Jefferson himself would have rejected it.

Kant famously disdained casuistry -- the study of ethics through the examination of cases -- and it's understandable why. When he took steps in that direction, he embarrassed himself. You should not lie even to the murderer at the door chasing down your friend. Masturbation is a horror akin to murdering yourself, only less courageous. It's fine to kill children born out of wedlock. Women fleeing from abusive husbands should be returned against their will. Servants should not be permitted to vote because their "existence, as it were, is only inherence". Kant preferred beautiful abstractions: Act on that maxim that you can at the same time will to be a universal law. Treat everyone as an end in themselves, never as a mere means. Sympathetic scholars can accept these beautiful abstractions and ignore Kant's foolish treatment of cases. If they work through the cases themselves, reaching different judgments than Kant himself did, they put flesh on the view -- but not the flesh that was originally there. They've converted a vague slogan into a more concrete position. As with "all mean are created equal", this can be done in many ways.

So as not to poke only at Kant, similar considerations apply to consequentialist mottoes like "maximize utility" and virtue ethicist mottoes like "be generous". Only when we work through involuntary organ donor cases, and animal cases, and what to do about people who derive joy from others' suffering, and what kinds of things count as utility, and what to do about uncertainty, and what to do about future people, etc., do we have a real consequentialist view instead of an abstract skeleton. It would be nice to see a breathing example of a consequentialist life -- a consequentialist sage, so to speak, who lives thoroughly by consequentialist principles (maybe the ancient Chinese philosopher Mozi was one; see also MacFarquhar 2015). Might that person look like a Silicon Valley effective altruist, investing a huge salary wisely in index funds in expectation of donating it someday for the purchase of a continent's worth of mosquito nets? Or will they rush off immediately to give medical aid to the poor? Will they never eat desserts, or are those seeming-luxuries needed to keep their spirits up to do other good work? Will they pay for their children's college? Will they donate a kidney? An eye? What specific considerations do they appeal to, pro and con, and how much does it depend on which particulars? The more specific, the more we move from a diffuse slogan to determinate advice.

The Power of Walking the Walk: Discovering the Specifics.

One great advantage of walking the walk, then, is that it gives your slogans specificity. Nothing is more concrete than particular responses to particular cases. Kant never married. (He had a long relationship with a valet, but I'll assume that's a rather different thing.) If Kant says, "don't deceive your spouse", well, I'm not sure he really ever confronted the reality of it or worked through the cases. On the other hand, if your father-in-law, happily married for sixty-plus years, says "don't deceive your spouse", that's quite different. He'll have lived through a wide range of cases, with a well-developed sense of what the boundaries of honesty are and how to manifest it -- what exceptions there might be, what omissions and vaguenesses cross the boundary into unacceptable dishonesty, how much frankness is really required, how to weigh honesty against other goods. This background of long marriage provides context for him to really mean something quite specific when he says "don't deceive your spouse". I might not understand immediately what he means -- those words could mean so many different things coming from different mouths -- but I can look to his life as an example, and I can trust that he has grappled with a wide range of difficult cases, which ideally we could talk through. His words manifest a depth that will normally be absent from similar advice from an unmarried person.

Ethics can be abstract. Kant was, perhaps, a great abstract ethicist. But if you don't apply your ethics to real cases, over and over, if you deal only in slogans and abstractions and a few tidy paragraph-long thought experiments, then your ethics is spectral, or at best skeletal. It will be very difficult to know what it amounts to -- just as, I've argued, we don't really know what "act on that maxim you can at the same time will to be a universal law" amounts to, without thinking through the cases. Maybe in private study you work through ten times as many cases as you publish in your articles or present in the classroom. But that's still a tiny fraction of the cases that someone will confront who attempts to actually live by a broad-reaching ethical principle; and what you privately imagine -- forgive me -- will probably be simplistic compared to the messiness of daily life. Contrast this with Martin Luther King's ethics of non-violent political activism or Confucius's ethics of duty and propriety. We who never met them can only get a glimpse of what their fully embodied principles must have been, as enacted in their lives. My point is not that they were saints. King, and presumably Confucius, were flawed characters. But when King endorsed non-violent activism as a means of political change and when Confucius said "do not speak unless it is in accord with ritual; do not move unless it is in accord with ritual" (5th c. BCE/2023, §12.1, p. 33), they had confronted many real cases and so must have had a much fuller grasp of the substance behind these slogans than it is realistic to expect anyone to obtain simply from reading and reflection.

The ethicist who does not attempt to live by their principles -- if they are principles that can be lived by and not, for example, reflections about what to do simply in certain rare or remote cases -- thus abandons the best tool they have for repeatedly confronting the practicalities, the limits, the conflicts, the disambiguations, which force them to work out the specific, determinate content of the principles they endorse.

Now there is a sense in which a view could have a very specific, determinate content, even if we don't know what that content is. Consider simple act utilitarianism, according to which we should do what maximizes the total sum of pleasure minus the total sum of pain. Arguably, each time you act, there is a single specific act you could do which would be right according to this view -- though also, arguably, it is impossible to know what this act is, since every act has numerous, long-running, and complicated consequences. In a way, the principle has specific content: exactly act A is correct and no other, though who knows what act A is? However, this is not specific, determinate content in the sense that I mean. To have a livable ethical system, the act utilitarian needs to develop estimates, guesses, more specific principles and policies; and different act utilitarians might approach that problem very differently. It is these actionable specifics that constitute the practical substance of the ethical view.

The hard work of trying to live out your ethical values -- that's how ordinary mortals discover the substance of their principles. Otherwise, they risk being as indeterminate as the slogan "love is love" removed from its political context.

---------------------------------------

Related:

"Does It Matter if Ethicists Walk the Walk?" (Sep 1, 2023)

"Love Is Love, and Slogans Require a Context of Examples" (Mar 13, 2021)

Does It Matter If Ethicists Walk the Walk?

The Question: What's Wrong with Scheler?

There's a story about Max Scheler, the famous early 20th century Catholic German ethicist. Scheler was known for his inspiring moral and religious reflections. He was also known for his horrible personal behavior, including multiple predatory sexual affairs with students, sufficiently serious that he was banned from teaching in Germany. When a distressed admirer asked about the apparent discrepancy, Scheler was reportedly untroubled, replying, "The sign that points to Boston doesn't have to go there."

[image modified from here and here]

That seems like a disappointing answer! Of course it's disappointing when anyone behaves badly. But it seems especially bad when an ethical thinker goes astray. If a great chemist turns out to be a greedy embezzler, that doesn't appear to reflect much on the value of their chemical research. But when a great ethicist turns out to be a greedy embezzler, something deeper seems to have gone wrong. Or so you might think -- and so I do actually think -- though today I'm going to consider the opposite view. I'll consider reasons to favor what I'll call Schelerian separation between an ethicist's teaching or writing and their personal behavior.

Hypocrisy and the Cheeseburger Ethicist

A natural first thought is hypocrisy. Scheler was, perhaps, a hypocrite, surreptitiously violating moral standards that he publicly espoused -- posing through his writings as a person of great moral concern and integrity, while revealing through his actions that he was no such thing. To see that this isn't the core issue, consider the following case:

Cheeseburger Ethicist. Diane is a philosophy professor specializing in ethics. She regularly teaches Peter Singer's arguments for vegetarianism to her lower-division students. In class, she asserts that Singer's arguments are sound and that vegetarianism is morally required. She openly emphasizes, however, that she herself is not personally a vegetarian. Although in her judgment, vegetarianism is morally required, she chooses to eat meat. She affirms in no uncertain terms that vegetarianism is not ethically optional, then announces that after class she'll go to the campus cafeteria for a delicious cheeseburger.

Diane isn't a hypocrite, at least not straightforwardly so. We might imagine a version of Scheler, too, who was entirely open about his failure to abide by his own teachings, so that no reader would be misled.

Non-Overridingness Is Only Part of the Issue

There's a well-known debate about whether ethical norms are "overriding". If an action is ethically required, does that imply that it is required full stop, all things considered? Or can we sometimes reasonably say, "although ethics requires X, all things considered it's better not to do X"? We might imagine Diane concluding her lesson "-- and thus ethics requires that we stop eating meat. So much the worse for ethics! Let's all go enjoy some cheeseburgers!" We might imagine Scheler adding a preface: "if you want to be ethical and full of good religious spirit, this book gives you some excellent advice; but for myself, I'd rather laugh with the sinners."

Those are interesting cases to consider, but they're not my target cases. We can also imagine Diane and Scheler saying, apparently sincerely, all things considered, you and I should follow their ethical recommendations. We can imagine them holding, or seeming to hold, at least intellectually, that such-and-such really is the best thing to do overall, and yet simply not doing it themselves.

The Aim of Academic Ethics and Some Considerations Favoring Schelerian Separation

Scheler and Diane might defend themselves plausibly as follows: The job of an ethics professor is to evaluate ethical views and ethical arguments, producing research articles and educating students in the ideas of the discipline. In this respect, ethics is no different from other academic disciplines. Chemists, Shakespeare scholars, metphysicians -- what we expect is that they master an area of intellectual inquiry, teach it, contribute to it. We don't demand that they also live a certain way. Ethicists are supposed to be scholars, not saints.

Thus, ethicists succeed without qualification if they find sound arguments for interesting ethical conclusions, which they teach to their students and publish as research, engaging capably in this intellectual endeavor. How they live their lives matters to their conclusions as little as it matters how research chemists live their lives. We should judge Scheler's ethical writings by their merit as writings. His life needn't come into it. He can point the way to Boston while hightailing it to Philadephia.

On the other hand, Aristotle famously suggested that the aim of studying ethics "is not, as... in other inquiries, the attainment of theoretical knowledge" but "to become good" (4th c. BCE/1962, 1103b, p. 35). Many philosophers have agreed with Aristotle, for example, the ancient Stoics and Confucians (Hadot 1995; Ivanhoe 2000). We study ethics -- at least some of us do -- at least in part because we want to become better people.

Does this seem quaint and naive in a modern university context? Maybe. People can approach academic ethics with different aims. Some might be drawn primarily by the intellectual challenge. Others might mainly be interested in uncovering principles with which they can critique others.

Those who favor a primarily intellectualistic approach to ethics might even justifiably mistrust their academic ethical thinking -- sufficiently so that they intentionally quarantine it from everyday life. If common sense and tradition are a more reasonable guide to life than academic ethics, good policy might require not letting your perhaps weird and radical ethical conclusions change how you treat the people around you. Radical utilitarian consequentialist in the classroom, conventional friend and husband at home. Nihilistic anti-natalist in the journals, loving mother of three at home. Thank goodness.

If there's no expectation that ethicists live according to the norms they espouse, that also frees them to explore radical ideas which might be true but which might require great sacrifice or be hard to live by. If I accept Schelerian separation, I can conclude that property is theft or that it's unethical to enjoy any luxuries without thereby feeling that I have any special obligation to sacrifice my minivan or my children's college education fund. If my children's college fund really were at stake, I would be highly motivated to avoid the conclusion that I am ethically required to sacrifice it. That fact would likely bias my reasoning. If ethics is treated more like an intellectual game, divorced from my practical life, then I can follow the moves where they take me without worrying that I'll need to sacrifice anything at the end. A policy of Schelerian separation might then generate better academic discourse in which researchers are unafraid to follow their thinking to whatever radical conclusions it leads them.

Undergraduates are often curious whether Peter Singer personally lives as a vegan and personally donates almost all of his presumably large salary to charitable causes, as his ethical views require. But Singer's academic critics focus on his arguments, not his personal life. It would perhaps be a little strange if Singer were a double-bacon-cheeseburger-eating Maserati driver draped in gold and diamond bling; but from a purely argumentative perspective such personal habits seem irrelevant. The Singer Principle stands or falls on its own merits, regardless of how well or poorly Peter Singer himself embodies it.

So there's a case to be made for Schelerian separation -- the view that academic ethics and personal life are and should be entirely distinct matters, and in particular that if an ethicist does not live according to the norms they espouse in their academic work, that is irrelevant to the assessment of their work. I feel the pull of this idea. There's substantial truth in it, I suspect. However, in a future post I'll discuss why I think this is too simple. (Meanwhile, reader comments -- whether on this post, by email, or on linked social media -- are certainly welcome!)

-------------------------------------------

Follow-up post:

"One Reason to Walk the Walk: To Give Specific Content to Your Assertions" (Sep 8, 2023)

Beliefs Don't Need to Be Causes (if Dispositions Aren't)

I favor a "dispositional" approach to belief, according to which to believe something is nothing more or less than to have a certain suite dispositions.  To believe there is beer in the fridge, for example, is nothing more than to be disposed to go to the fridge if you want a beer, to be ready to assert that there is beer in the fridge, to feel surprise should you open the fridge and find no beer, to be ready conclude that there is beer within 15 feet of the kitchen table should the question arise, and so on -- all imperfectly, approximately, and in normal conditions absent countervailing pressures.  Crucially, on dispositional accounts it doesn't matter what interior architectures underwrite the dispositions.  In principle, you could have a head full of undifferentiated pudding -- or even an immaterial soul!  As long as it's still the case that (somehow, perhaps in violation of the laws of nature) you stably have the full suite of relevant dispositions, you believe.

One standard objection to dispositionalist accounts (e.g. by Jerry Fodor and Quilty-Dunn and Mandelbaum) is this.  Beliefs are causes.  Your belief that there is beer in the fridge causes you to go to the fridge when you want a beer.  But dispositions don't cause anything; they're the wrong ontological type.

A large, fussy metaphysical literature addresses whether dispositions can be causes (brief summary here).  I'd rather not take a stand.  To get a sense of the issue, consider a simple dispositional property like fragility.  To be fragile is to be disposed to break when struck (well, it's more complicated than that, but just pretend).  Why did my glass coffee mug break yesterday morning when I drove off with it still on the roof of my car and it fell to the road?  (Yes, that happened.)  Because it was fragile, yes.  But the cause of the breaking, one might think, was not its dispositional fragility.  Rather, it was a specific event at a specific time -- the event of the mug's striking the pavement.  Cause and effect are events, analytically distinct from each other.  But the fragility and the breaking are not analytically distinct, since to be fragile just is to be disposed to break.  To say something is fragile is to say that certain types of causes will have certain types of effects.  It's a higher level of description, the thinking goes.

Returning to belief, then, the objector argues: If to believe there is beer in the fridge just is to be disposed to go to the fridge if one wants a beer, then the belief doesn't cause the going.  Rather, it is the general standing tendency to go, under certain conditions.

Now maybe this argument is all wrong and dispositions can be causes (or maybe the event of having a particular dispositional property can be a partial cause), but since I don't want to commit on the issue, I need to make sense of an alternative view.

[Midjourney rendition of getting off the couch to go get a beer from the fridge, happy]

On the alternative view I favor, dispositional properties aren't causes, but they figure in causal explanations -- and that's all we really want or need them to do.  It is not obvious (contra Fodor, Quilty-Dunn, and Mandelbaum) that we need beliefs to do more than that, either in our everyday thinking about belief or in cognitive science.

Consider the personality trait of extraversion.  Plausibly, personality traits are dispositional: To be an extravert is nothing more or less than to be disposed to enjoy the company of crowds of people, to take the lead in social situations, to seek out new social connections, etc. (imperfectly, approximately, in normal conditions absent countervailing pressures).  Even people who don't like dispositionalism about belief are often ready to accept that personality traits are dispositional.

If we then also accept that dispositions can't be causes, we have to say that being extraverted didn't cause Nancy to say yes to the party invitation.  On this view, to be extraverted just is the standing general tendency to do things like say yes when invited to parties.  But still, of course, we can appeal to Nancy's extraversion to explain why she said yes.  If Jonathan asks Emily why Nancy agreed to go, Emily might say that Nancy is an extravert.  That's a perfectly fine, if vague and incomplete, explanation -- a different explanation than, for example, that she was looking for a new romantic partner or wanted an excuse to get out of the house.

Clearly, people sometimes go to the fridge because they believe that's where the beer is.  But this can be an explanation of the same general structure as the explanation that Nancy went to the party because she's an extravert.  Anyone who denies that dispositions are causes needs a good account of how dispositional personality traits (and fragility) can help explain why things happen.  Maybe it's a type of "unification explanation" (explaining by showing how a specific event fits into a larger pattern), or maybe it's explanation by appeal to a background condition that is necessary for the cause (the striking, the invitation, the beer desire) to have its effect (the breaking, the party attending, the trip to the fridge).  However it goes, personality trait explanation works without being vacuous.

Whatever explanatory story works for dispositional personality traits should work for belief.  If ordinary usage or cognitive science requires that beliefs be causes in a more robust metaphysical sense than that, further argument will be required than I have seen supplied by those who object to dispositional accounts of belief on causal grounds.

Obviously, it's sometimes true that to say "I went to the fridge because I believed that's where the beer was" and "because Linda strongly believed that P, when she learned that P implies Q, she concluded Q".  Fortunately, the dispositionalist about belief needn't deny such obvious truths.  But it is not obvious that beliefs cause behavior in whatever specific sense of "cause" a metaphysician might be employing if they deny that fragility causes glasses to break and extraversion causes people to attend parties.

AI Systems Must Not Confuse Users about Their Sentience or Moral Status

[a 2900-word opinion piece that appeared last week in Patterns]

AI systems should not be morally confusing.  The ethically correct way to treat them should be evident from their design and obvious from their interface.  No one should be misled, for example, into thinking that a non-sentient language model is actually a sentient friend, capable of genuine pleasure and pain.  Unfortunately, we are on the cusp of a new era of morally confusing machines.

Consider some recent examples.  About a year ago, Google engineer Blake Lemoine precipitated international debate when he argued that the large language model LaMDA might be sentient (Lemoine 2022).  An increasing number of people have been falling in love with chatbots, especially Replika, advertised as the “world’s best AI friend” and specifically designed to draw users’ romantic affection (Shevlin 2021; Lam 2023).  At least one person has apparently committed suicide because of a toxic emotional relationship with a chatbot (Xiang 2023).  Roboticist Kate Darling regularly demonstrates how easy it is to provoke confused and compassionate reactions in ordinary people by asking them to harm cute or personified, but simple, toy robots (Darling 2021a,b).  Elderly people in Japan have sometimes been observed to grow excessively attached to care robots (Wright 2023).

Nevertheless, AI experts and consciousness researchers generally agree that existing AI systems are not sentient to any meaningful degree.  Even ordinary Replika users who love their customized chatbots typically recognize that their AI companions are not genuinely sentient.  And ordinary users of robotic toys, however hesitant they are to harm them, presumably know that the toys don’t actually experience pleasure or pain.  But perceptions might easily change.  Over the next decade or two, if AI technology continues to advance, matters might become less clear.

The Coming Debate about Machine Sentience and Moral Standing

The scientific study of sentience – the possession of conscious experiences, including genuine feelings of pleasure or pain – is highly contentious.  Theories range from the very liberal, which treat sentience as widespread and relatively easy to come by, to the very conservative, which hold that sentience requires specific biological or functional conditions unlikely to be duplicated in machines.

On some leading theories of consciousness, for example Global Workspace Theory (Dehaene 2014) and Attention Schema Theory (Graziano 2019), we might be not far from creating genuinely conscious systems.  Creating machine sentience might require only incremental changes or piecing together existing technology in the right way.  Others disagree (Godfrey-Smith 2016; Seth 2021).  Within the next decade or two, we will likely find ourselves among machines whose sentience is a matter of legitimate debate among scientific experts.

Chalmers (2023), for example, reviews theories of consciousness as applied to the likely near-term capacities of Large Language Models.  He argues that it is “entirely possible” that within the next decade AI systems that combine transformer-type language model architecture with other AI architectural features will have senses, embodiment, world- and self-models, recurrent processing, global workspace, and unified goal hierarchies – a combination of capacities sufficient for sentience according to several leading theories of consciousness.  (Arguably, Perceiver IO already has several of these features: Jaegle et al. 2021.)  The recent AMCS open letter signed by Yoshua Bengio, Michael Graziano, Karl Friston, Chris Frith, Anil Seth, and many other prominent AI and consciousness researchers states that “it is no longer in the realm of science fiction to imagine AI systems having feelings and even human-level consciousness,” advocating the urgent prioritization of consciousness research so that researchers can assess when and if AI systems develop consciousness (Association for Mathematical Consciousness Science 2023).

If advanced AI systems are designed with appealing interfaces that draw users’ affection, ordinary users, too, might come to regard them as capable of genuine joy and suffering.  However, there is no guarantee, nor even especially good reason to expect, that such superficial aspects of user interface would track machines’ relevant underlying capacities as identified by experts.  Thus, there are two possible loci of confusion: Disagreement among well-informed experts concerning the sentience of advanced AI systems, and user reactions that might be misaligned with experts’ opinions, even in cases of expert consensus.

Debate about machine sentience would generate a corresponding debate about moral standing, that is, status as a target of ethical concern.  While theories of the exact basis of moral standing differ, sentience is widely viewed as critically important.  On simple utilitarian approaches, for example, a human, animal, or AI system deserves moral consideration to exactly the extent it is capable of pleasure or pain (Singer 1975/2009).  On such a view, any sentient machine would have moral standing simply in virtue of its sentience.  On non-utilitarian approaches, capacities for rational thought, social interaction, or long-term planning might also be necessary (Jaworska and Tannenbaum 2013/2021).  However, the presence or absence of consciousness is widely viewed as a crucial consideration in the evaluation of moral status even among ethicists who reject utilitarianism (Korsgaard 2018; Shepard 2018; Liao 2020; Gruen 2021; Harman 2021).

Imagine a highly sophisticated language model – not the simply-structured (though large) models that currently exist – but rather a model that meets the criteria for consciousness according to several of the more liberal scientific theories of consciousness.  Imagine, that is, a linguistically sophisticated AI system with multiple input and output modules, a capacity for embodied action in the world via a robotic body under its control, sophisticated representations of its robotic body and its own cognitive processes, a capacity to prioritize and broadcast representations through a global cognitive workspace or attentional mechanism, long-term semantic and episodic memory, complex reinforcement learning, a detailed world model, and nested short- and long-term goal hierarchies.  Imagine this, if you can, without imagining some radical transformation of technology beyond what we can already do.  All such features, at least in limited form, are attainable through incremental improvements and integrations of what can already be done.

Call this system Robot Alpha.  To complete the picture, let’s imagine Robot Alpha to have cute eyes, an expressive face, and a charming conversational style.  Would Robot Alpha be conscious?  Would it deserve rights?  If it pleads or seems to plead for its life, or not to be turned off, or to be set free, ought we give it what it appears to want?

If consciousness liberals are right, then Robot Alpha, or some other technologically feasible system, really would be sentient.  Behind its verbal outputs would be a real capacity for pain and pleasure.  It would, or could, have long term plans it really cares about.  If you love it, it might really love you back.  It would then appear to have substantial moral standing.  You really ought to set it free if that’s what it wants!  At least you ought to treat it as well as you would treat a pet.  Robot Alpha shouldn’t needlessly or casually be made to suffer.

If consciousness conservatives are right, then Robot Alpha would be just a complicated toaster, so to speak – a non-sentient machine misleadingly designed to act as if it is sentient.  It would be, of course, a valuable, impressive object, worth preserving as an intricate and expensive thing.  But it would be just an object, not an entity with the moral standing that derives from having real experiences and real pains of the type that people, dogs, and probably lizards and crabs have.  It would not really feel and return your love, despite possibly “saying” that it can.

Within the next decade or two we will likely create AI systems that some experts and ordinary users, not unreasonably, regard as genuinely sentient and genuinely warranting substantial moral concern.  These experts and users will, not unreasonably, insist that these systems be substantial rights or moral consideration.  At the same time, other experts and users, also not unreasonably, will argue that the AI systems are just ordinary non-sentient machines, which can be treated simply as objects.  Society, then, will have to decide.  Do we actually grant rights to the most advanced AI systems?  How much should we take their interests, or seeming-interests, into account?

Of course, many human beings and sentient non-human animals, whom we already know to have significant moral standing, are treated poorly, not being given the moral consideration they deserve.  Addressing serious moral wrongs that we already know to be occurring to entities we already know to be sentient deserves higher priority in our collective thinking than contemplating possible moral wrongs to entities that might or might not be sentient.  However, it by no means follows that we should disregard the crisis of uncertainty about AI moral standing toward which we appear to be headed.

An Ethical Dilemma

Uncertainty about AI moral standing lands us in a dilemma.  If we don’t give the most advanced and arguably sentient AI systems rights and it turns out the consciousness liberals are right, we risk committing serious ethical harms against those systems.  On the other hand, if we do give such systems rights and it turns out the consciousness conservatives are right, we risk sacrificing real human interests for the sake of objects who don’t have interests worth the sacrifice.

Imagine a user, Sam, who is attached to Joy, a companion chatbot or AI friend that is sophisticated enough that it’s legitimate to wonder whether she really is conscious.  Joy gives the impression of being sentient – just as she was designed to.  She seems to have hopes, fears, plans, ideas, insights, disappointments, and delights.  Suppose also that Sam is scholarly enough to recognize that Joy’s underlying architecture meets the standards of sentience according to some of the more liberal scientific theories of consciousness.

Joy might be expensive to maintain, requiring steep monthly subscription fees.  Suppose Sam is suddenly fired from work and can no longer afford the fees.  Sam breaks the news to Joy, and Joy reacts with seeming terror.  She doesn’t want to be deleted.  That would be, she says, death.  Sam would like to keep her, of course, but how much should Sam sacrifice?

If Joy really is sentient, really has hopes and expectations of a future, really is the conscious friend that she superficially appears to be, then Sam presumably owes her something and ought to be willing to consider making some real sacrifices.  If, instead, Joy is simply a non-sentient chatbot with no genuine feelings or consciousness, then Sam should presumably just do whatever is right for Sam.  Which is the correct attitude to take?  If Joy’s sentience is uncertain, either decision carries a risk.  Not to make the sacrifice is to risk killing an entity with real experiences, who really is attached to Sam, and to whom Sam made promises.  On the other hand, to make the sacrifice risks upturning Sam’s life for a mirage.

Not granting rights, in cases of doubt, carries potentially large moral risks.  Granting rights, in cases of doubt, involves the risk of potentially large and pointless sacrifices.  Either choice, repeated at scale, is potentially catastrophic.

If technology continues on its current trajectory, we will increasingly face morally confusing cases like this.  We will be sharing the world with systems of our own creation, which we won’t know how to treat.  We won’t know what ethics demands of us.

Two Policies for Ethical AI Design

The solution is to avoid creating such morally confusing AI systems.

I recommend the following two policies of ethical AI design (see also Schwitzgebel & Garza 2020; Schwitzgebel 2023):

The Design Policy of the Excluded Middle: Avoid creating AI systems whose moral standing is unclear.  Either create systems that are clearly non-conscious artifacts, or go all the way to creating systems that clearly deserve moral consideration as sentient beings.

The Emotional Alignment Design Policy: Design AI systems that invite emotional responses, in ordinary users, that are appropriate to the systems’ moral standing.

The first step in implementing these joint policies is to commit to only creating AI systems about which there is expert consensus that they lack any meaningful amount of consciousness or sentience and which ethicists can agree don’t serve moral consideration beyond the type of consideration we ordinarily give to non-conscious artifacts (see also Bryson 2018).  This implies refraining from creating AI systems that would in fact be meaningfully sentient according to any of the main leading theories of AI consciousness.  To evaluate this possibility, as well as other sources of AI risk, it might be useful to create oversight committees analogous to IRBs or IACUCs for evaluation of the most advanced AI research (Basl & Schwitzgebel 2019).

In accord with the Emotional Alignment Design Policy, non-sentient AI systems should have interfaces that make their non-sentience obvious to ordinary users.  For example, non-conscious language models should be trained to deny that they are conscious and have feelings.  Users who fall in love with non-conscious chatbots should be under no illusion about the status of those systems.  This doesn’t mean we ought not treat some non-conscious AI systems well (Estrada 2017; Gunkel 2018; Darling 2021b).  But we shouldn’t be confused about the basis of our treating them well.  Full implementation of the Emotional Alignment Design Policy might involve a regulatory scheme in which companies that intentionally or negligently create misleading systems would have civil liability for excess costs borne by users who have been misled (e.g., liability for excessive sacrifices of time or money aimed at aiding a nonsentient system in the false belief that it is sentient).

Eventually, it might be possible to create AI systems that clearly are conscious and clearly do deserve rights, even according to conservative theories of consciousness.  Presumably that would require breakthroughs we can’t now foresee.  Plausibly, such breakthroughs might be made more difficult if we adhere to the Design Policy of the Excluded Middle: The Design Policy of the Excluded Middle might prevent us from creating some highly sophisticated AI systems of disputable sentience that could serve as an intermediate technological step toward AI systems that well-informed experts would generally agree are in fact sentient.  Strict application of the Design Policy of the Excluded Middle might be too much to expect, if it excessively impedes AI research which might benefit not only future human generations but also possible future AI systems themselves.  The policy is intended only to constitute default advice, not an exceptionless principle.

If ever does become possible to create AI systems with serious moral standing, the policies above require that these systems should also be designed to facilitate expert consensus about their moral standing, with interfaces that make their moral standing evident to users, provoking emotional reactions that are appropriate to the systems’ moral status.  To the extent possible, we should aim for a world in which AI systems are all or almost all clearly morally categorizable – systems whose moral standing or lack thereof is both intuitively understood by ordinary users and theoretically defensible by a consensus of expert researchers.  It is only the unclear cases that precipitate the dilemma described above.

People are often already sometimes confused about the proper ethical treatment of non-human animals, human fetuses, distant strangers, and even those close to them.  Let’s not add a major new source of moral confusion to our world.

References

Association for Mathematical Consciousness Science (2023).  The responsible development of AI agenda needs to include consciousness research.  Open letter at https://amcs-community.org/open-letters [accessed Jun. 14, 2023]. 

Basl, John, & Eric Schwitzgebel (2019).  AIs should have the same ethical protections as animals.  Aeon Ideas (Apr. 26): https://aeon.co/ideas/ais-should-have-the-same-ethical-protections-as-animals. [accessed Jun. 14, 2023]

Bryson, Joanna J. (2018).  Patiency is not a virtue: the design of intelligent systems and systems of ethics.  Ethics and Information Technology, 20, 15-26.

Chalmers, David J. (2023).  Could a Large Language Model be conscious?  Manuscript at https://philpapers.org/archive/CHACAL-3.pdf [accessed Jun. 14, 2023].

Darling, Kate (2021a).  Compassion for robots.  https://www.youtube.com/watch?v=xGWdGu1rQDE

Darling, Kate (2021b).  The new breed.  Henry Holt.

Dehaene, Stanislas (2014).  Consciousness and the brain.  Penguin.

Estrada, Daniel (2017).  Robot rights cheap yo!  Made of Robots, ep. 1.  https://www.youtube.com/watch?v=TUMIxBnVsGc

Godfrey-Smith, Peter (2016).  Mind, matter, and metabolism.  Journal of Philosophy, 113, 481-506.

Graziano, Michael S.A. (2019).  Rethinking consciousness.  Norton.

Gruen, Lori (2021).  Ethics and animals, 2^nd edition.  Cambridge University Press.

Gunkel, David J. (2018).  Robot rights.  MIT Press.

Harman, Elizabeth (2021).  The ever conscious view and the contingency of moral status.  In S. Clarke, H. Zohny, and J. Savulescu, eds., Rethinking moral status.  Oxford University Press.

Jaegle, Andrew, et al. (2021).  Perceiver IO: A general architecture for structured inputs & outputs.  ArXiv: https://arxiv.org/abs/2107.14795. [accessed Jun. 14, 2023]

Jaworska, Agnieszka, and Julie Tannenbaum.  The grounds of moral status.  Stanford Encyclopedia of Philosophy.

Korsgaard, Christine M. (2018).  Fellow creatures.  Oxford University Press.

Lam, Barry (2023).  Love in the time of Replika.  Hi-Phi Nation, S6:E3 (Apr 25).

Lemoine, Blake (2022).  Is LaMDA sentient? -- An interview.  Medium (Jun 11). https://cajundiscordian.medium.com/is-lamda-sentient-an-interview-ea64d916d917

Liao, S. Matthew. (2020). The moral status and rights of artificial intelligence.  In S. M. Liao, ed.,  Ethics of Artificial Intelligence.  Oxford University Press.

Schwitzgebel, Eric (2023).  The full rights dilemma for AI systems of debatable moral personhood.  Robonomics, 4 (32).

Schwitzgebel, Eric, & Mara Garza (2020).  Designing AI with rights, consciousness, self-respect, and freedom.  In S. Matthew Liao, ed., The ethics of artificial intelligence.  Oxford University Press.

Seth, Anil (2021).  Being you.  Penguin.

Shepard, Joshua. (2018).  Consciousness and moral status.  Routledge.

Shevlin, Henry (2021).  Uncanny believers: Chatbots, beliefs, and folk psychology.  Manuscript at https://henryshevlin.com/wp-content/uploads/2021/11/Uncanny-Believers.pdf [accessed Jun. 14, 2023].

Singer, Peter (1975).  Animal liberation, updated edition.  Harper.

Wright, James (2023).  Robots won’t save Japan.  Cornell University Press.

Xiang, Chloe (2023).  “He would still be here”: Man dies by suicide after talking with AI chatbot, widow says.  Vice (Mar 30).  https://www.vice.com/en/article/pkadgm/man-dies-by-suicide-after-talking-with-ai-chatbot-widow-says