Neurath's Boat

More AI Experiments

I've been asking AIs about a lot of things I already know, to see how well they can do at such tasks. I asked Bard if it could figure out who wrote this blog, and while it was clearly able to find this blog with its internet access, it guessed incorrectly. It occurred to me to ask about FAFBlog, the authorship of which was a subject of some speculation back in the day. Bard was embarrassing, sounding as if it were confused about the fact that Fafner, Giblets, and the Medium Lobster were fictional characters.

I hadn't bothered to ask GPT-4 about my blog, as I was sure it was too obscure, but FAFBlog was kind of a big deal for a while, so I thought maybe GPT-4 could do better. It certainly outperformed Bard, showing no signs of being confused about what was going on with FAFBlog. It also said some really favorable things about FAFBlog; I will have to mention to Chris next time I see him that GPT-4 seems to be a fan. Well, maybe its tendency to be positive in general contributed, but it definitely sounded like its training data included comments from a lot of people who loved FAFBlog as much as I did.

GPT-4 did not succeed at the specific task assigned, though. Since it showed such a good understanding of FAFBlog in other ways, I can't help but wonder if its failure to provide even a guess about FAFBlog's authorship was due to the effort to make it "harmless" having made it disinclined to out people's secrets, rather than being due to not having enough clues. Perhaps future experiments will give me a better idea of whether that's the case.

Feser's superstitions, chapter I

I promised to give a more detailed account of what's wrong with Feser's book, so let's begin.  I shall skip over the preface, as it contains no arguments. On that basis, I suppose I should skip the first chapter as well, but that would leave me open to accusations of lack of thoroughness, so I'll say as much as I can about it.

It is not until page 6 that Feser so much as waves his hand in the direction of an argument, when he recounts what he claims was his own voyage of philosophical discovery. He cites Frege as motivating his Platonism. This naturally makes me wonder what Feser thinks about Russell's Paradox, and more importantly of course Gödel's theorem. But he never talks about such subjects, nor does he talk about subjects like the axiom of choice or Euclidean vs. Non-Euclidean geometries. It makes it hard for me to take seriously his analogy between mathematical and philosophical knowledge when he seems to have such a poorly developed theory of mathematical knowledge.

He also mentions that Russell influenced him by showing how little we can know of the intrinsic nature of the material world. Of course, I think Russell was right about this, and in fact that he didn't go far enough. Feser draws a contrary conclusion, perhaps on the basis that knowledge of intrinsic matters is the only real knowledge or that all knowledge must trace back to knowledge of intrinsic matters. But he doesn't prove or argue for either of these claims, at least not here. They have been much discussed in contemporary philosophy, so this represents another area where he is wrong to say that his really important issues have been ignored by the modern naturalists.

He also describes Richard Swinburne as someone who employs "the most rigorous of modern philosophical methods to the defense of religious belief." I am skeptical of this as a description of Plantinga (who also gets this praise), but applied to Swinburne, this can only be considered laughable.

Feser claims that to the naturalists, natural selection is a "pseudo-deity." I suppose it has features in common with how Feser takes God to be, in that it is knowable a priori. Once you understand natural selection, it is quite obvious that it must happen in any situation where there is a mix of some stability over time and some more or less random change. Of course, that natural selection is responsible for specific phenomena, e.g. the diversity of life, requires empirical evidence in each case (evidence which is readily available in the case of many biological phenomena). However, Feser's claim that natural selection could not in any "true or interesting sense" manifest design is unargued. I suppose it depends rather heavily on what one considers interesting.

The remainder of the first chapter continues to provide no arguments, except a sort of inductive argument based on cases for Feser's theory of what motivates secularists. Not all (only most) of his theories about this are wrong, though it goes without saying that they are all absurdly charicatured. The naturalistic world view rejects ultimate authority. That's what it is to be a naturalist. Some, perhaps most, naturalists inconsistently treat naturalism itself as an ultimate authority, because people have trouble with the idea that there really is none. They deserve to be called superstitious, though the fact that naturalism is so purely negative, consisting of little more than the rejection of all ultimate authorities, makes taking it as an ultimate authority a less bad error than most other cases of belief in ultimate authority.

However, contra Feser, there are good reasons to reject ultimate authority. The concept is incoherent. This is admitted by some of its defenders (e.g. Kierkegaard, or Heidegger), who insist that we must believe in it despite its incoherence; this is the reason that "faith" has become the popular line among defenders of ultimate authority. Feser, of course, has no patience with this line, and indeed doesn't even bother to mention why he thinks so many on his side seem to welcome putting things in terms of faith. It might be an enlightening topic for him to investigate.

Feser is obviously right to note that most atheist philosophers do not confront the issue of the existence of his God directly, instead engaging in various smaller detail projects in pursuit of naturalism. But this is because the absurdity of his God as an ultimate authority is widely recognized, while forms of ultimate authority in narrow areas are less obviously unacceptable and so more controversial. This is one of the areas where I agree with Feser; I think the reasons for rejecting his God are closely related to reasons for rejecting non-Humean causation, various forms of anti-reductionism, Armstrong-style universals, perfectly natural properties, and many other still popular philosophical speculations. Still, the connection is not so strong that anyone who rejects God is obviously rationally required to reject all these more moderate views. Feser naturally wishes to run the inference in the opposite direction, and argue that anyone committed to any of the more moderate views, and most naturalistic philosophers are committed to some of them, are committed to his God; the inference is also not immediate in that direction, as I mentioned in my review of his book on Amazon.

Feser ends his first chapter by asking atheists like myself to consider the possibility that we might be wrong. I do frequently consider the possibility that I am wrong to reject the God of Plato, Spinoza, and Einstein. Though that has little to do with Feser's God, I'm afraid it's the best I can do.

Neuroscientists are getting better at reading minds

The interesting results are mentioned here.  I found myself wondering what would happen if you told the people what the brain scans predicted they would do.  If you asked them again what they thought they would do after they were told what the brain scan said, would their own predictions become more or less accurate?  Would they still be less accurate than the brain scan?  Would telling them what the brain scan predicted make the brain scan less accurate?  I know I'd try to use a brain scanner like this to see if it predicted my doing anything I wanted to avoid doing, so I could try to find some better way of getting myself to not do it, but I have no idea whether I'd be successful at that.  Clearly an area where lots of additional research is needed! 

Sample sizes

Whenever it's relevant (when feminism comes up, or materialism) I talk to my students about the studies of sex differences in the brain.  The press always reports any difference discovered by researchers as proof of innate differences in cognition.  Of course, the biggest problem with that interpretation is that environmental factors affect the brain, so finding something in the brain is not useful for determining where it came from, but I also mention the small sample sizes the studies usually have (a couple of dozen participants at most).

A friend is getting his Ph.D. in neuroscience (incidentally, according to him MRI studies, which are the kind that usually get such press, are all crap) decided to practice his thesis defense presentation on his friends before doing the real thing, so I heard about his research.  It had nothing to do with sex differences (it was about some details of how motor control works in the brain), but I learned something about sample sizes.  I was somewhat shocked to discover that it involved two experiments with a grand total of three monkeys (two per experiment; one monkey was involved in both).  I guess these were hideously expensive cyber-monkeys (they had electronic implants to measure brain activity very precisely during the experiments), so it wasn't practical to use more, but it seems like one could get very deceptive results this way if one were unlucky and got even one atypical monkey.

My skepticism of overly confident neuroscience claims has been further increased.

PZ Myers stands up for philosophers!

Critics of intelligent design sometimes suggest that it ought to be discussed in philosophy classes rather than biology classes. A recent post at Pharyngula recognizes that intelligent design's total lack of intellectual content makes it of little more interest to philosophers than to biologists; it's so nice to see someone who isn't a philosopher recognize this point.

Powered by ScribeFire.

The science of A Natural History of Rape

There is a dramatic shift in Palmer and Thornhill's text, starting in chapter 3, and ending at the end of chapter 4.  This is the region of the book where the authors discuss the particular evidence they find most relevant to their theses about rape.  In this region of the book, they frequently mention issues which they think require more study, and make proposals for tests of their speculations.  Somewhat less frequently, but still notably (especially in contrast to the rest of their book), they mention data which seems to raise questions for their hypotheses.  The data that they mention in this portion of the book is certainly intriguing and suggestive, though it is quite unclear what exactly to conclude from it.

Since I had wanted not to hate the book, it having been recommended to me by someone I like and have some respect for, I felt very relieved when I started to get into chapter 3, and hoped the rest of the book would continue more in that tone.  This looked like science; sure, they were engaging in speculations, and their evidence was far from conclusive, but the phenomena of human behavior are extremely complex.  Making speculative claims and proposing further tests seems an entirely appropriate procedure.  Unfortunately, there are reasons to worry even about this part of the book.

I will not comment in detail about the quality of the data that Palmer and Thornhill present; I am not by any means an expert in their field.  Evolution, Gender, and Rape contains much discussion of their evidence by those with more knowledge of the field.  I found the essays in that volume of uneven quality, but some were quite good, and they often cited fascinating evidence and speculations of their own.  Some of the essays also raised serious questions about Palmer and Thornhill's methods of analyzing their data, which surely need to be examined; such issues represent the first problem with the more scientific part of Palmer and Thornhill's text.

The second problem is that it ends so quickly.  It constitutes roughly a quarter of the book, with the remaining 3/4 devoted to virulent and sometimes dishonest polemic against feminist social psychologists, as well as some very bad philosophy of science.

An obvious speculation is that they did not, in fact, have enough material to make a substantial book, and padded it out with deliberate attempts to provoke controversy (if this was their goal, they certainly succeeded).  The next question for my investigation of bias and methodology is whether the flaws in the science section show any sort of pattern, such that Palmer and Thornhill's motivations might have affected that part of the book as well; this question will require far more thorough investigation than I have so far done, so I probably won't be posting any more about this particular book for some time.  Which will probably be a relief to many.

Objectivity and Progress

One of my positivist revivalist opinions is that while the ideal of scientific objectivity is deeply problematic, it is also quite valuable, and that those who dismiss the ideals of science as being just another tool of the privileged are making a serious mistake.  Actual scientific research is as likely, if not more likely, to challenge the interests of the currently powerful as to justify them, and I wish there were more moderns with radical political views who shared with the old Logical Positivists the idea that science can be a critical weapon in pursuit of radical politics.

Of course, science is infected with bias, like everything else.  If a scientist claims to be immune to bias in his pure, objective work, he's either lying, or probably more likely deluded.  He's also being unscientific; science has investigated bias, and found it to be pervasive and hard to detect (careful research reveals people to have biases of which they appear sincerely unaware).  Since the ideal of scientific objectivity says that bias is a bad thing, a scientist who follows the ideal should think that this indicates problems which need to be corrected; further studies are needed to find methods of overcoming bias, and then of course those methods need to be put into practice in the interests of better science.  It is a problem that not enough scientists seem to see the need for that, but that is a problem with the scientists, not the scientific ideal.

A couple of interesting studies have come to my attention recently.  This study is interesting in a couple of respects.  It reports how childhood abuse, including verbal abuse, causes lasting harm in brain development.

The first noteworthy feature of the study for my purposes is that  it discusses how environmental effects change the brain.  Modern scientists working on human biology are not infrequently accused of some form or other of "biological determinism," strong doctrines about how our genes determine everything about what we're like.  It is indeed depressing how quickly some study which reveals, say, differences between the brains of some men and some women (brain studies tend not to have huge sample sizes) is turned in the media into a report of how men and women are different by nature and how this produces their stereotypical behaviors.  Naturally, this is taken to undermine patriarchy blaming in general; these natures are used as alternative explanations for phenomena feminists wish to attribute to an oppressive system.  However, as usual, the media does a horrible job of reporting the studies, which generally have much more modest conclusions than those which are reported.  Still, even the original studies of sex differences in the brain often pay too little attention to a point made forcefully in the study described in my link; environmental factors can change the brain.  Otherwise, how could a materialist explain how environmental factors manage to change how we think?  Thus, brain studies provide much less support for any kind of biological determinism than they are often presented as doing.  They do, however, help us better understand the mind and the brain, something of great value both for medical purposes and for the political purpose of figuring out how people arrive at the views that they do and how to counter-act their mistaken views (such as the inevitability of patriarchy).

Our culture, and particularly our ideals of masculinity, also display an obsession with violence, revenge, and punishment.  Sometimes biological determinism is used to excuse this, on the basis that men are just like that; as usual for biological determinist explanations, this requires ignoring breathtaking amounts of evidence.  However, conservatives argue even more often that this is a good thing; that there's lots of evil in the world, and the only way to discourage it is to stand up to it with the harshest measures possible.

In Plato's writings, Socrates claims that it is never good to harm anyone.  He argues that to harm someone is to make them worse, and surely the solution to having bad people around is not to make them worse.  Few have been convinced by his argument, but psychological studies suggest that he was on to something.  One of the best established results in psychology is that rewarding people for desired behavior is a more effective motivator than punishing them for undesired behavior.  Punishment produces resentment and hostility, and while it can extinguish one undesired behavior, it nearly always produces other undesired behaviors, often even worse ones.  Psychologists have also noted that many of those who have participated in these studies as the providers of rewards and punishments refused to believe the results.

Some more science, this time using the probability and statistics at which people are so terrible, but which seem to produce such great benefits to those who understand them, explains the mechanism which fools us into thinking punishment works better than reward.  Punishments or rewards are generally given for exceptional behavior; rewards for exceptionally good behavior, punishments for exceptionally bad behavior.  Exceptional behavior is by definition outside of the range of usual behavior.  Thus, by the phenomenon known technically as "regression to the mean", exceptional behavior is usually followed by unexceptional behavior (otherwise the exceptional becomes the new normal, of course).  So exceptionally bad behavior is generally followed by better behavior even in the absence of any intervention, and conversely with exceptionally good behavior.  This phenomenon of regression to the mean can completely swamp the short-term effects of a particular reward or punishment, so that the person applying rewards and punishments will tend to see a consistent pattern of rewards being followed by worse performance and punishments being followed by better performance, regardless of which is actually more effective in modifying performance.

The abuse study provides further evidence that the conservative enthusiasm for punishment is misguided, by showing more clearly how inflicting suffering on people really does make them worse, just as Socrates taught.

Of course, sometimes the goal is precisely to make people worse, in order to justify our bad treatment of them.  Another study mentioned by Ann at Feministing suggests that sexual harassment is more often a weapon for punishing women who fail to conform to gender norms than an expression of desire.  This seems quite plausible; despite how the conservatives claim PC police have turned all male-female interactions into a minefield, quite a lot of flirting continues to go on, and it hardly ever leads to sexual harassment challenges.  Women can apparently usually tell the difference between harassment and flirting when they experience it.  Studies like this help show what's really going on with issues like harassment, and thus help provide a defense against those who claim that harmless activities are being criminalized.  Yet another example where conducting scientific studies helps progressive causes.