= 27d567bc2d48d9f40e9ccc20ea370b15)
CarlShulman comments on [post redacted] - Less Wrong Discussion
You are viewing a comment permalink. View the original post to see all comments and the full post content.
= 783df68a0f980790206b9ea87794c5b6)
Loading…= b715d02e85f7b3b4ae65ca3d3e450868)
Subscribe to RSS Feed
Powered by Reddit
= f037147d6e6c911a85753b9abdedda8d)
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (59)
What is the evidence that convinced you?
Questions like that can't be answered due to problems of social epistemology. If I give you two pieces of evidence then it's assumed I've given you my two most convincing pieces of evidence and that I don't have any other pieces of evidence, and thus that you should update even further away from the conclusion that I've reached. If I regularly talked to you and could keep up a thread of conversation, and if both of us didn't anticipate needless antagonism involving the presumption that the other hadn't already considered things that they'd in fact considered many times (like "psi appears to be quite capricious, how convenient for believers"), then I could talk to you about the evidence without fear of unjustifiably cementing either of our beliefs. As it is you can only have that kind of relationship among decently close friends. Do you know of other ways of discussing tricky topics, ways that don't have those requirements?
I assume you know most of the standard considerations and evidence. I think your belief in psi is probably the result of a bias in favor of such hypotheses (partly directly, and partly affecting your enthusiasm for other beliefs that feed into psi) but that you might have found something interesting while spelunking (which need not be convincing to be interesting).
It would be easy to specify a bunch of stylized facts, and then incrementally add to the list or contest various elements. For instance, here's a start:
Might you characterize the bias?
(When I introspect on my rationale for taking these things seriously, it seems it's because I think it's best to assume the worst. I.e. psi and UFOs would mean the universe is significantly weirder than I'm comfortable with and I'm not at all sure how that would fit in with plans for reducing x-risk or more generally doing the right thing.
Secondary and tertiary reasons (not the result of introspection but of guessing) would be wanting the universe to be really weird and interesting, and wanting to be maximally contrarian. I'm not sure the secondary reason is true but the third likely is.)
Best not to get into it, at least on a public forum.
That's a really good list!
I tentatively disagree that religion count as a falsified belief. This sounds nit-picky but I think indicates a substantial meta-level difference in our prior, i.e. the extent to which we think the average person is or isn't crazy. E.g.:
I don't know whether to agree or disagree with this; I don't really have much evidence either way. What do you have in mind? It's clearly crucial, as I don't see the parapsychology literature as providing very much evidence for psi, and you seem(?) to see it as providing some amount of evidence against psi. I looked through much of the heuristics and biases literature and was profoundly disappointed with what I found. Since then I've been skeptical of claims that the average person is highly susceptible to false beliefs. Why do you think people are so prone to "selective memory, unconscious processing, lack of statistical thinking, wishful thinking, etc", beyond the psychology literature? Or am I wrong to be so unimpressed with the psychology literature?
Recently, quantum stuff in biological systems generally. Only important to note because various proposed psi mechanisms involve quantum computation going on in the brain, which until recently seemed rather implausible but is now a serious possibility.
Given that the methods used are practically designed so as to minimize any psi phenomena, e.g. taking away any real incentives for humans to use psi (with a few exceptions like a few of Bem's experiments), why should we have a wide prior for psi effect sizes? E.g. Ben Goertzel's analogy to experimentation on the alleged existence of "falling in love" comes to mind. I can't find it but it was on his "Multiverse According to Ben" blog. Might you explain more why you think a wide prior seems good?
Agreed. (Only noting this 'cuz I think you might think I think that simulation-style explanations are worth thinking about.)
I don't follow, might you flesh this out more?
Some other points:
The religions are contradictory. The behavior of the founders (moreso the more recent the religion) maps onto random guys creating cults today in great detail. People express belief in their particular religion being true, the particular founder or books being right, in young earth creationism, in Islamic medicine, in miraculous healing, etc. They do so at least as fervently as they do belief in psychic powers. You can try to make up a naturalistic version of religion that, while improbable on its face, doesn't get much of a further penalty from the evidence (simulator deism, whatever) but that doesn't rescue the accuracy of the masses.
On direct causes, consider sleep paralysis and alien abductions. Drug use can create experiences interpreted as religious or psychic or whatever, and various biological conditions can do likewise. False memories can be cultivated by therapists or group efforts (there was an interesting book by a Harvard researcher studying people who claimed to have been abducted). The formation of urban legends by the telephone effect and optimization for good stories is well-documented. Exposed frauds are pretty common, and frauds have to be more common than exposed ones, probably far more given the scarcity of debunking resources. These things vary from culture to culture, e.g. fan death giving an indication of how much people can shape their experiences and beliefs to fit memes floating around.
Psychology research tends to be bad. Most of the effects claimed don't exist (far more for repeatedly replicated experiments done by independent groups), and the effects that do exist are mostly greatly exaggerated in their effect sizes and generalizability. That is also true of heuristics and biases. And yet surveys (the basis of these "lots of people believe in ESP" claims) reveal that people have lots of demonstrably false beliefs, often strong beliefs, especially on topics where they get little feedback. Bryan Caplan has many such: basic facts about the economy, demographics, toxicology, risks of death. People believe having sex with virgins cures HIV, despite the absence of any actual documented examples. They form superstitions about homeopathic placebos and acupuncture.
People do not get feedback about their capricious/subtle psi beliefs (healing that heals amputees does get enough negative feedback that it's a rare belief), and can get utility out of believing they have psi or that the world is magical, indeed will often go out of their way to protect the belief.
I contest. Biological systems are hot and noisy, and quantum effects are at the wrong scales in nervous systems. Second, quantum computation!= telepathy, ESP, telekinesis, etc. Usually these arguments are of the form "QM and ESP are mysterious, and could be related" or "this interpretation of QM assumes nonlocal effects, or even nonlocal effects driven by magic consciousness, so there's a proof of concept for nonlocal ESP". These are nonsense. I would want to see an example of a good argument for quantum computation-psi connections.
Claims of psi powers tend to expand to fill the gaps of observation and alternative explanation. E.g. flying witches in Africa, jinn turning people into animals, etc. Claims that psi is small seem to be a "God of the Gaps" side-effect of modern recording technologies and other developments, rewriting the prior ex post. Repelling bullets, visible healing, visible telekinesis, and similar dramatic powers have often been claimed as effects "in the wild" but they always retreat from sight. We have casinos, the stock market, and similar opportunities for precognition, telekinesis could show up in sporting events, etc, etc.
Actually, all sorts of nifty research on falling in love has been performed with useful predictions about oxytocin levels, fidelity, various sorts of future self-report, divorce, frequency of sex, etc, etc. Experiments in the lab can do quite a lot to distinguish couples in love or not, and to increase or decrease it. That's more of a rhetorical trick, insofar as we use "falling in love" in a lot of signalling games and to some extent we're unwilling to identify it with any natural kinds that actually exist.
Where is the animal psi? What organs do it? How do the inputs affect human responses and behavior? How did the capacity evolve in the first place, and if it did evolve, why didn't it keep doing so?
Well, it seems that way to pro-psi researchers, but it's not so different from research into unrelated bogus phenomena where researchers are motivated to keep trying different angles of attack, e.g. cross-country international development models in economics.
Sorry, I didn't mean to imply that that part is plausible as I haven't looked into it at all. But the "too hot and noisy" objection seems less certain now that there's research into quantum "computation" allowing for photosynthesis as well as something else quantum involved in birds' navigation. Links here. As you suggest, perhaps it should be presumed irrelevant until some plausible mechanism connecting it to psi is proposed.
Yeah, um, basic sanity fail on my part. Not a good sign.
Re common false beliefs, you obviously know significantly more about the subject than I do. Are there any books you'd recommend?
Point conceded. Until I look at the literature more closely I'll agree with you that the lack of large effects seems is a decent chunk of evidence against psi.
It would seem that the only psi-friendly explanation of the weak results from parapsychology then would be something like actively evasive psi. You say similar hypotheses have shown up to explain the lack of certain expected phenomena in macroeconomics? Or, might you explain the connection?
By the way thank you for talking to me about this stuff, very few people have the prerequisite knowledge/skills/patience to do so fruitfully.
I'm starting to think that maybe all evidence I have for psi is incommunicable. Bleh!
I was referring to data-mining for correlations to produce predictive rules using historical datasets of country growth rates, rules which then fail badly when applied to new datasets. Beliefs in stronger causal conclusions persist, and folk with such beliefs talk about how the processes of growth are changing over time, or the great difficulty of pulling conclusions from the noisy data. The closest think to "actively evasive psi" would probably be academic claims to have found predictive rules for stock and other liquid markets: "my data-mined rule was for real, but now that I've published the markets are taking it into account, which is why it no longer works."
The evidential part of intuitions or personal psi experiences are communicable (save for the possibility of conscious lies, but it's pretty clear that those are not needed), a bigger dataset of intuitions is better than a single case, etc. Robin's common priors paper is relevant here.
It's a bit of a tangent, but I thought the recent stuff about some special "quantum" coherence enabling photosynthesis was mostly hype, as it could happen just the same classically.
Do you mean this in the trivial sense, or are you making a stronger claim against the heuristics and biases literature?
This sense. I don't think the heuristics and biases literature is among the worst psychological subfields in terms of quality. I don't think it is the best.
I didn't understand a few of these. I'll think about them more to see if I can extract their meaning and if not I'll ask for clarification later.
This is the Goertzel piece.
Can you at least tell us how your hypothesis explains the apparent lack of statistically rigorous experiments confirming psi?
I can't really explain the apparent lack except by postulating that casual readers haven't looked very closely at the parapsychology literature. Many have noted that parapsychology experiments are significantly more rigorous than other experiments in psychology because the standards are so high. Note that parapsychologists go out of their way to set up environments where you'd least expect psi, 'cuz those are the only environments that don't have a huge amount of confounding factors. Even under such austere conditions there are many rigorous experiments confirming psi. Here is a source of some clearly rigorous studies.
Note that there is a huge amount of motivated skepticism in the parapsychology literature; many papers attack other papers merely because it is incredibly easy to gain status by attacking low status ideas. Many complaints about statistical deficiencies are in fact retarded. E.g. in many cases there will be complaints about the file drawer effect even in scenarios where such complaints are patently ridiculous.
In the end it seems to me that one can either tentatively accept the results of parapsychology or conclude that statistics simply does not work. I think the latter hypothesis is rather plausible, but anyone who goes down that route should know that the heuristics and biases and social psychology literature is chock full of results that are significantly less well pinned down than those of parapsychology. The social sciences rarely tell us anything useful with any confidence.
Also note that there are legitimate problems with the results from parapsychology and one should look out for attempts to conceal these problems. The results clearly indicate that psi is capricious. If you wish to skip ahead to the biggest problems with the results, you can find a summary and some proposed explanations here. Because capriciousness and experimenter effects look so shifty some parapsychologists have tried to hide them or minimize their importance, further damaging the reputation of parapsychology. Pay careful attention to make sure that the experimenter isn't trying to sweep something under the rug.
It perhaps might be best to focus on a single experimenter. Then you can be more confident that there isn't substantial fraud going on or huge file drawer effects. Looking at the PEAR results exclusively might be a good idea so as to indicate what you can expect from the rest of the field.
Note that I do not take the evidence from parapsychology as strongly suggesting for or against psi. I haven't yet looked at it very carefully and am skeptical of the benefit of continued searching considering I'm not willing to do a detailed analysis of the statistics used in every paper. If I did not have evidence outside of the parapsychology literature then I would not be as confident of the existence of psi. The parapsychology literature provides merely one kind of evidence, namely statistical evidence, and alone isn't enough to suggest clear conclusions.