We are developing the social individualist meta-context for the future. From the very serious to the extremely frivolous... lets see what is on the mind of the Samizdata people.

Samizdata, derived from Samizdat /n. - a system of clandestine publication of banned literature in the USSR [Russ.,= self-publishing house]

Science is hard

Science is really, really hard. Someone posted a report on Reddit about an attempt to replicate some psychology experiments, and how hard it was. The comments thread is fascinating. I particularly enjoyed a description of how research results can be “turtles all the way down”. This comment also suggests to me a mechanism for the formation of group-think.

These sorts of difficulties need to be borne in mind when reading excited reports in the media that simply paraphrase the university’s press release about the research. And when considering claims that particular branches of science are “settled”.

Tweet about this on TwitterShare on FacebookShare on TumblrShare on RedditShare on Google+Share on VKEmail this to someone

71 comments to Science is hard

  • Ken Mitchell

    The problem with the title of this post is that psychology isn’t science, and “psychology experiments” are generally unreplicable and unscientific. You can gather some statistics about observed results, but you can rarely falsify an experiment.

    And to the extent that you can do actual “science” in the field of psychology, it’s nearly impossible to do research on previous experiments, because you can’t depend on the previous experimenters to accurately report on what they did or what they observed.

    So, yeah, “turtles all the way down” as a cascade of unfounded assumptions and improbable results is what you can generally expect.

  • Cristina

    “psychology isn’t science”
    Thank you, Ken. I’ve been repeating the same for years to no avail. Same with sociology, philology and most other “-logy” in vogue.

  • Nicholas (Rule Yourselves!) Gray

    Too true! ‘Cosmology’? When are they going to restart the Universe, to prove their theories?
    Interestingly enough, I still keep an eye on Astrology, and if someone had asked me before last weekend, I’d have advised that one to be wary, as Saturn and the Sun would be at right angles as seen from the Earth, and this has always been seen by astrologers as a time of contraction. I wonder if the Chinese government would be interested in Western Astrologers?

  • Laird

    Critina, I would add economics to your list of “non-sciences”.

  • Paul Marks

    Good post Rob – and an important general scientific point.

    As for psychology it is not a new subject.

    The in-depth study of the mind goes back to ancient times – see Aritotle’s “On the Soul” (“that which animates” need have no mystical meaning in the thought of Aristotle or in the thought of “the Commentator” on Aristotle Alexander of … some centuries later).

    Even the word “psychology” is of 17th century – Ralph Cudworth (please no person-in-Kent “refutations” of Cudworth saying “well Cudworth believed that witches were real therefore…..” therefore NOTHING, as Chief Justice Sir John Holt, of 1688 and all that, had much the same politics and philosophy as Cudworth and did NOT believe in witchcraft).

    From Ralph Cudworth in the 1600s (as with Aristotle and others in ancient times) to Noah Porter and James McCosh in the late 19th century (who wrote the standard texts in America at that time) the study of “psychology” was based on the self evident truth (literally “self evident” as in “we hold these truths to be self evident…..” which is really from Thomas Reid rather than just Thomas Jefferson) of human agency – Cudworth (and some others) rejected “chopping up” the human mind into “will” and “reason” – but others accepted this practice and used the term “free will” which Cudworth himself did not use (although he accepted the self evident fact of human agency – our ability, with effort, to do other than we do).

    See, for example, James McCosh’s attack on J.S. Mill – “In defence of self evident truth – An Examination of the Philosophy of Mr J.S. Mill”.

    However, in about 1890 William James (Harvard) produced his own short book on psychology which just baldly states that a psychology which just baldy states that the psychologist “must make an assumption of determinism” in order to be “scientific”.

    An use of the word “scientific” to mean “like physics” (or the physical sciences generally) – very much like the “physics envy” that ruined economics.

    A body of knowledge (a “subject”) can be valid without being physical science based upon an “assumption of determinism”.

    But if psychology is to be based on an “assumption of determinism” one would expect William James to include a careful refutation of Noah Porter (President of Yale) and James McCosh (President of Princeton) who wrote the anti determinist standard texts on psychology a few years before he published his own work.

    He does no such thing – William James plays the J.S. Mill trick of just ignoring people he can not refute (Mill did that with critics of the Labour Theory of Value – whom he pretended did not exist) – the names “James McCosh” and “Noah Porter” do not even appear in the index of the text of William James.

    William James says (elsewhere) that he does not even believe in determinism – or in “compatiblism” which he (quite rightly) says leads to a “quagmire of evasion” (Kant called compatiblism a “wretched subterfuge” and he was also correct on this point).

    So William James (without argument – without refuting previous writers) bases psychology on an “assumption” he admits (admits elsewhere) is false!

    This is not “science” it is nonsense.

    By the way…

    Some people may object to the extensive use of brackets above.

    However, I prefer to be difficult to read – than to write a sentence (or sentences) that are WRONG.

    And I tried writing long sentences (without brackets) over on “Counting Cats” and people did not like that either.

  • Mr Ed

    Right on cue, the BBC puts out a story about a link between being a goth and being ‘at risk’ of depression.

    As if depression might be a ‘risk’ in the way that a stroke might be.

    A cluster of BS appears in this excerpt:

    They found the more young people identified with the goth subculture, the higher their likelihood of self-harm and depression.
    Those who saw themselves as part of the goth group were already more likely to have shown signs of depression before the age of 15 and to have been bullied in the past.
    But scientists argue the link remains even once these factors are accounted for.
    Researcher Dr Rebecca Pearson, from the University of Bristol, said there could be many reasons behind the trend, including the possibility that teenagers susceptible to depression were attracted to the goth way of life.
    She added: “The extent to which young people self-identify with goth subculture may represent the extent to which at-risk young people feel isolated, ostracised or stigmatised by society.”

    Quite how are these factors ‘accounted for’? How on Earth did this study arise? Did someone look at 3,694 15-year olds ‘around Bristol’ simply to see what they found? Did they go looking for goths, what is a ‘goth’? Is it definable in the way that a European Starling is? How did they measure ‘depression’?

    One might get the impression that the research started with a pre-conceived summary and then went looking for ‘evidence’ to support it.

    Basic tip: No mechanism, no science.

    e.g. Were some researches to announce: We have found a plant enzyme of several megadaltons. It appears to be released when some plants are exposed to strong light including UV light, its structural similarity to nucleases suggests that it is involved in damaged DNA excision repair. all are pointers to a new enzyme that might help plants that produce it (or similar enzymes) cope with DNA damage from UV light, and at least we have pointers to a mechanism behind the enzyme, and why it might be produced in particular circumstances. It might in time lead to an explanation of the enzyme’s function and mechanism, and an understanding of how it fits into the plant’s biochemical machinery. All science.

    Whereas looking at ‘goths’ and asking if being a ‘goth’ is linked with ‘depression’ is just subjective blathering at both ends.

  • Kevin B

    Agreed, psychology is not a science. It may have had some pretensions towards science when it began, but now the job of a psychologist is to produce plenty of reports of the type Mr Ed links above. These will provide click-bait for the tabloids such as the Mail and the Guardian in order to get the psychologist’s name well enough known that he can write a book.

    Unfortunately, due to the incentives to publish or perish these days, other fields are following the pseudo sciences down this path. Climate science is one obvious example and public health and nutrition are others, but even physics is trending that way. String theory anyone?

  • Watchman

    Have any of the people criticising psychology actually seen what academic psychologists outside of social pscyhology (an attempt to do sociology by scientific methods remains an attempt to do sociology to be fair to the critics) do? If what they do is not science, then neither is anything much in biology or medical sciences, and a fair amount of chemistry (psychopharmacology is a valid field of study – it is pscyhologists who ultimately work out if the guy off his head and on medications is off his head because of the medications…). As with most subjects psychology has a checquered past (but unlike most, they were generally scientific enough to film most of it – so most of the truly disturbing and unethical things done before the 60s are actually recorded…), and its share of crackpots and quacks (including about half of Freud and Jung…), but like most sciences it is perfectly fine if you put aside the ideology.

    Indeed, if psychology is not a science, it is worth noting their is no scientific understanding of why we think in the way we do. Psychology is the sub-group of biology that covers this.

    What is not scientific (in either the modern or the classical sense of the word) is to simply state something is not a science because of prejudices and preconceptions. That seems horribly like rejecting ideas because ‘the science is settled’…

  • Watchman

    PS (because I wanted to finish my last post on a strong point) I am not saying all psychologists are scientific. Most social ones seem to be seeking to reinforce their political preconceptions by making claims about human thought patterns drawn on observation. Some are frauds, some just not very good at what they do. Same as any science really – the people in it are human. The results are based upon and are themselves data, and it is those that are important.

  • Plamus

    Mr Ed, challenge accepted. I’ll see you “subjective blathering at both ends”, and raise “cutting-edge research from University of The Blindingly Obvious”: scientists find that if you tell women they have breast cancer, they become anxious. Only a majority of them, however, which makes one wonder about those women who were told they had cancer and did not get anxiety. Not to worry, though: “”However, additional studies are needed to identify effective methods for facilitating communication and support for diverse populations participating in mammography screening,” they conclude.”

  • Jaded Voluntaryist

    I conducted one of those 100 attempted replications, although you‘ll not find my name on the list of contributors. My supervisor is there though, despite doing very little of the actual work. It’s not the first time that has happened to me either.

    Oh and yes, our experiment failed to reproduce the results in question. Had we left in a couple of outliers it would have been close though. Fortunately my supervisor, although fond of credit, was not actually dishonest and reported the results with the outliers removed.

    My own feeling was that the original paper was something of a fluke. The original experiment we replicated was carried out something like 10 years ago. The experimental software used has improved since that time allowing you to properly differentiate between the behavior you wish to target and irrelevant responses. The original experiment left all those irrelevant responses in, which depending on the participant could be a very high proportion of all responses. I know because I spent hours filtering the data.

    I’d take issue with the claim that psychology isn’t a science. It deals with an intrinsically complex system so you’d expect it to be difficult, even under ideal circumstances. The big problem with psychology is psychologists. They try to use psychology for all sorts of things it shouldn’t be used for, including self aggrandizment and political point scoring. But I don’t think you can deny the usefulness of applied subjects like human factors design or augmentative communication, both of which draw heavily from the theoretical branches of psychology.

    I find the casual dismissivness of academia in libertarian circles extremely tiresome. It is lazy and glosses over important truths. Yes academia in general and psychology in particular is rotten – in exactly the same way that the US civil war was “about slavery” There is a lot of truth to the statements, but they are nonetheless lazy generalisations that miss a lot of important details.

    The negative picture of academia is however sufficiently true that I’m very keen to get away from working in it though….

  • Paul,

    Physics envy is very real. But I haven’t heard it about economics as much as psychology. And the envy goes back to Freud. He was obsessed with being the “Newton of the Mind”. If you read Freud you can see that schtick. It is clearly utter drivel but it is drivel wearing a frock-coat. Freud envied physics so obviously it is embarrassing. Pathetic.

    And, yes Paul, it is impossible to make a systematic scientific study of anything without assuming determinism up to a point. But, of course, “science” as such doesn’t answer all questions. It just can’t.

    Can science give you a Bond line to get a bird into bed? Despite the obsession that the likes of Freud had with sex they didn’t have it on the dance floor of the Irish in Nottingham in 1993. I did and I was studying physics. The Uni had a psychology department but that was full of bell-ends. And not in a good way. Twats and all manner of arseholes really. Full of piss and vinegar.

    If you wanna find out about people read novels or history or study anything other than psychology. Or sociology (which is psychology with more stats). All any of the inhumane sciences ever tell you is you are a machine. So can it explain why I like Bon Jovi more than Status Quo? No, it can’t. Can it explain why I can’t stand coffee but like tea? Nyet! So what is the fucking point? And Gods don’t try and explain everything from Lennon to Lenin in terms of breast-feeding. Just go fuck yourself because nobody else will.

    Physics means you can work out stuff. Psychology just seems a load of people working out their own prejudices and covering them in faux scientific terms. Physics is science. Psychology is a turd in a wimple. I once built a pico-Tesla magnetometer out of junk. Psychologists think they can tell me why. They also make that out of junk. There is a difference. Mine worked.

  • Laird

    “Psychology is a turd in a wimple.”

    Well, that’s an evocative phrase! It gave me a good snort.

  • Snorri Godhi

    Can science give you a Bond line to get a bird into bed?

    Funny you should be asking that. Dealing with people, or indeed with any vertebrates, is not like dealing with inanimate objects or animals with tiny brains. When you deal with people, anytime you find a tactic to get what you want, the other party finds a way to deny it to you.
    Nonetheless, i did find a book which is extremely helpful in dealing with young ladies, even though it is not about seduction at all: it’s called Winning the Games People Play, and it was written by a psychologist, Nathan Miron.
    It’s not just helpful with the opposite sex: it’s helpful in dealing with people in general.
    I admit, however, that it might not be equally helpful to people with a less scientific turn of mind than i have.

    About physics envy, there are several distinct problems.
    One is that the subject matter of psychology makes predictions impossible, except in some statistical sense. This problem is not limited to psychology: it also exists in the social+political “sciences”, climatology, the study of evolution, and chaotic systems in general. Physicists like Nick foment this subset of physics envy by insisting on precise predictions.
    Another problem is psychologists themselves, as JV points out above. Psychologists, and the “soft” sciences in general, try to use a “scientific” method that the hard sciences do not actually use. Interestingly, Imre Lakatos blamed this kind of physics envy on Newton’s four rules of reasoning, which were designed to justify Newtonian physics and undermine Cartesian physics, not to provide guidance to future scientists.
    Yet another problem is illustrated by Nick’s demand that psychology explain why he prefers tea to coffee. What if it’s genetic? what if we found a gene that can predict whether people prefer tea or coffee? how would psychologists react to that?
    My point is that the ideological divide on the issue of nature and nurture is just too strong for real science to emerge. (I also find preposterous some claims and arguments of genetic determinists, btw.)

  • Snorri Godhi

    Following up on Mr Ed, i believe this is a more blatant example of BBC pseudoscience:

    The BBC bias should be self-evident (as in the Declaration of Independence), but here is an informed rebuttal:

    A more balanced view (emphasizing that the science is correct, leaving in the fine print that the science is not against a low carb diet):

  • Tedd

    “X is not a science” seems like an unscientific attitude, to me. “The study of X is riddled with poor methodology,” sure. “Practitioners of the study of X are too prone to unfalsifiable hypotheses,” sure. Perhaps even, “X is inherently so difficult to conduct experiments on that all conclusions need to be regarded with added skepticism.” But to say that X is not a science is to say that that entire area of knowledge is impenetrable by the scientific method. I don’t buy it. I don’t mean to equivocate here; I doubt anybody meant to say that science is worthless in understanding the human psyche. But that is what “psychology is not a science” literally says.

    There’s an expectation in psychology that meaningful research results will occur as often, or nearly as often, as in the research of other fields. That’s an unrealistic expectation, and it leads to unwarranted conclusions being published. Psychology needs more emphasis on replication of experiments and a greater interest in falsifying studies than in confirming studies. In that sense it’s like every other science.

  • Ken Mitchell

    Tedd: “Psychology needs more emphasis on replication of experiments and a greater interest in falsifying studies than in confirming studies.”

    No, psychology needs SOME emphasis on replication…” because right now, it doesn’t have ANY. WHEN psychologists begin to use the methodology of science, THEN we might say that psychology is a science. But it isn’t now, and doesn’t appear likely to become one any time soon.

  • CaptDMO

    “…Same with sociology, philology and most other “-logy” in vogue.”
    Careful. “In wine there is truth” or so “they” say. Academics dress it up of course.
    I’ve been known to be rather fond of Mixology in my day.
    Of course, the science of “I drink, I get drunk, I fall down” isn’t complicated, or mysterious enough to charge US$200 per 50 minute “hour” to “analyze”, or $65,000 per 8 month “year” to allegedly “teach”.

  • I apply a simple test to any scientific article: I count the number of sentences whose meaning is softened by the use of weasel words such as “may”, “might”, “could”, “possibly” and other such conditionals. If that number constitutes more than 40% of the total number of sentences in the article, the article is crap and should be ignored. Needless to say, any conclusion which contains several weasel words means that nothing has been proven to any degree of certitude.

  • Right…

    1. Falsifiability is not the gold standard. It is a useful criterion but it is not everything. Anyone doubt thermodynamics?

    2. My point was that many things which are good are not sciences. Doesn’t make ’em bad in exactly the same way a rissoto is bad because it isn’t a chow mein.

    3. Precise. When Voyager 1 entered Jovian orbit it was out by an estimated 1 metre. And that was with Sir Isaac at the yoke. When Feynman et al re-jigged QED it is about as accurate as measuring the distance from LA to NYC to the precision of the width of a human hair.

    That impresses me. The vague nostrums of trickology – less so. And, no, saying it is “complicated” is no excuse. So is the disturbing function in celestial dynamics or the marriage of QMech and GR.

    4. Again… The understanding is not a scientific subject. It just isn’t. I can usually guess what my wife is thinking, “300, not again!!!” without doing a survey. This is because I have known her for nearly 16 years.

    5. If (and it might be) my prediliction for tea over coffee is genetic then that isn’t psychology. It is biology which is a pukka science. There are three sciences (though chemistry, though useful, is largely dull) and then their is the Empress herself which is mathematics. The rest is about as much as wanking your claypole into a polystrene cup. Note I am not including the classic arts subjects here such as history and stuff. But the “soft” sciences are not worth a marsupial fart in a typhoon.

    And Paul is right. Economics should not be about producing graphs to prove anything. Theory always comes before empiricism. There is a hideously dirty joke lurking there but I shall let it lie.

  • Oh, and Paul Marks: you keep writing ’em in the manner of your choosing, and I’ll keep reading ’em.

    As for long sentences: whenever some cretin makes a response of “tl;dr” (too long; didn’t read), I respond with “TS; DD” (too stupid; don’t debate). Sometimes, a decent thesis or careful analysis requires a little more more than a bumper-sticker aphorism.

  • Alisa

    I’m with JV and Tedd on this: I very much share the objections to “soft sciences” expressed here, but I also see a tendency to throw the baby out with the tab water. Human behavior is worthy of study, because that is the only way to gain any understanding of it. Such a study however simply cannot be as systematic as the study of physics and related sciences. This does not make the study of human behavior less scientific, it just makes it different, in both methods and reasonable expectations of possible usefulness.

    Which brings me to the next point, that of separation of science and engineering, in both physical sciences and human behavior: in both areas said separation must be very clear. It has been and still is quite clear in physical sciences, it is unfortunately very far from that in one specific area of human-behavior study, namely economics.

  • Julie near Chicago

    Nick, hard science (biochemistry, or if you prefer bioelectrochemistry) can certainly give us hard-scientific understanding of the differences in humans’ bodies’ responses to the chemical substances in coffee vs. those in tea. (At least to within the bounds of the epistemological/metaphysical issue of what we mean by “knowledge”).

    I don’t know that it’s been done with those two substances, but it has been done with (wait for it!) broccoli, and quite a few other things we ingest.

    The next step is to figure out the subjective results as the particular person experiences them: Good, bad, or indifferent, and the quality of the goodness or badness. This is difficult, of course, partly because of the difficulties of truly accurate human communications.

    So neuroscientists attempt to get it, at least to some extent, by means of MRI’s and other methods of monitoring physical responses. (Nothing, of course, will ever make the map identical with the territory except by the fiat declaration that the “map” is properly defined as being the actual territory. Volumes of the philosophy of science follow.)

    So. We also know that certain states of a body’s biochemistry cause further biochemical reactions within a person’s body that he, the subject, reports as pleasurable, unpleasant or discomfiting in some way, or unnoticeable. I do not know how finely-focussed this knowledge is in general; probably in almost all cases there is some knowledge that a hyperabundance or paucity of receptors for chemical X are one of the main factors determining a subject’s response of pleasure, for instance, or even of various sorts of pleasure.

    This is like saying that in Plato’s time it was known that hemlock tea was bad for one, without knowing WHY it was bad: by which I mean specifically the chemical interactions within the human body in response to its intake of hemlock tea. “Hemlock causes death. But we never heard of either gamma-coniceine or coniine.” *Shrug* Putting the pieces together takes time.

    But in any case, there’s no reason in principle not to be able to say that “We know that humans whose body chemistry is in the following state [pages of formulas] will enjoy buttered carrots.”

    As to understanding how thought works at the bioelectrochemical level, it’s not clear that this is impossible. (There are, I believe, serious and well-qualified scientists who believe that the validity of “chaos theory” is still hugely hypothetical.) Nor, of course, that it is possible.

    All that is about understanding at the level of the hard sciences.

    I agree with Nick (and always have) that hard science is about understanding the mechanisms. Statistics can suggest correlations to investigate, but that statistics does not give us mechanisms.

    On the other hand, at the engineering level (that is, what we actually do in the real world) statistics formal or “folk” may be the best we have to go on. You’re probably safe enough having ham sandwiches, corn-on-the-cob, and chocolate cake at your American summer picnic, simply because most Americans like those foods. (Assuming the guests don’t keep kosher, anyway.) It’s probably unwise to campaign for national office here on the platform that you are going to abolish elections in toto and forthwith.

    This is the sort of understanding of human psychology that Messrs. Machiavelli and Alinsky, for two, understood very very well. Lenin and Stalin were pretty good at it also, or at least in how useful terror might be in persuading most people to get with the program even if they hated it.

    And, we use this sort of psychology — practical psychology, which is a form of engineering if you think so, and I do — on ourselves all the time. “Think how much better you’ll feel with a nice clean floor.” Or, one of my favorite slogans when faced with something I really really don’t want to do: “In 24 hours it will be over.”

    (And it did take the full 24 hours for the kid to decide the cave was getting boring and confining and it was time to exit and investigate the Outside World. Grrrr.)

  • Julie near Chicago

    By the way, What Alisa Said.

  • In that Empire, the Art of Cartography attained such Perfection that the map of a single Province occupied the entirety of a City, and the map of the Empire, the entirety of a Province. In time, those Unconscionable Maps no longer satisfied, and the Cartographers Guilds struck a Map of the Empire whose size was that of the Empire, and which coincided point for point with it. The following Generations, who were not so fond of the Study of Cartography as their Forebears had been, saw that that vast map was Useless, and not without some Pitilessness was it, that they delivered it up to the Inclemencies of Sun and Winters. In the Deserts of the West, still today, there are Tattered Ruins of that Map, inhabited by Animals and Beggars; in all the Land there is no other Relic of the Disciplines of Geography.

    – Jorge Luis Borges “On Exactitude in Science”.

    Along with JRRT probably the finest writer other than Shakespeare.

  • Alisa, I do regard biochemistry (as I said) a hard science. By which I mean a pukka science. If I say that about biology (and I did) then I surely mean it about biochemistry. Surely.

  • Julie near Chicago

    I’m going to get into the “Best Writer” thing, Nick, but I must say I like your Borges quote quite a bit.

  • Julie near Chicago

    Er, I’m not going to get into the “Best Writer” thing.

  • Cristina

    “The big problem with psychology is psychologists”
    True. With all of those byproduct of the Enlightenment the problem, precisely, is the practitioner. It’s impossible for them to be any other way because all of those “-logy” have their starting point on prejudices and preconceptions about the infallibility of the reason and the betterment of mankind. With that arsenal the only thing they can do is politics, not science.
    NickM. I like Borges too. Very much.

  • Tedd


    Your description of practical psychology is reminiscent of engineering in the pre-Newton era: almost entirely composed of empirical laws and rules of thumb. That’s probably apt, as I doubt we yet have the equivalent of Newton’s laws in our understanding of the mind — if that’s even possible. And yet great structures and even quite intricate and impressive machines were built before Newton, just as human psychology has from time to time been brilliantly exploited by people who had that gift.

    The drive to build artificial minds might lead to a better theoretical understanding of our own minds — perhaps because such a theoretical understanding might prove necessary to build an artificial mind in the first place.

  • Julie near Chicago

    Indeed, Tedd. Besides which, the very fact that we manage to coexist with others at all shows that each of us has practical knowledge of psychology. Even con-men (surely not politicians!) and Ted Bundy have such knowledge; if they didn’t, they wouldn’t have any victims. I don’t need to know how photosynthesis works in order to get that my tomato plants seem to need lots of sun.

  • Laird

    An aside to NickM: There are quite a few of us here who aren’t up on British slang, and who consequently have no idea what “pukka” means. I initially took your use of it in relation to biochemistry as a derogatory term, and only learned that was not your intent in your subsequent post. You might consider minimizing your use of such slang terms (or at least including a parenthetical definition).

  • Alisa

    Tab water? What on earth is that I wonder?…

  • Laird,
    I don’t think that is that obscure and as I am English I do use British slang. I even spell “colour” with a “u”. Except when I use HTML, natch. You should be thankful I don’t use my native Geordie dialect which is almost a different lingo entire. Though it is related to Lowland Scots such as Lairds used.

    I think Alisa, Tab water is a ref to something made by Coca Cola called Tab Clear” which asted like the Urine of the Cameleopard. Which sort of brings us back to “Pukka”. Think Del Boy and the Peckham Spring.

  • Tedd,

    There is an in-joke amongst the AI set – “Can you take it apart with itself”.

  • Ken Mitchell

    Friends, it’s easy to decrypt various obscure terms; right-click them in any browser, and you’ll be presented with at least one option to look up the term in your preferred search engine. So I was easily able to determine that “pukka is a word of Hindi and Urdu origin, literally meaning “cooked, ripe” and figuratively “fully formed”, “solid”, “permanent”, “for real” or “sure”. In UK slang, it can mean “genuine” or simply “very good”; see also pukka sahib.” Since the “UK Slang” meaning was fairly close to the original Hindi, I presumed that it meant “first rate” or something equivalent.

  • OK, I’ll take the bait. This is an idea I have had for about 20 years and it ain’t changed much. The analogy between the human mind (or even my cat’s mind) and computers is utter bunk. We make mistakes. We are rather “unusual” and this leads to both disaster and creativity. Computers complement us because, for example, whilst they can’t conceive of the disturbing function they can calculate it much better than us fallible beasts. They are tools not agents. I would argue that consciousness is ineffable for the simple reason (as hinted above) that the only way in to the problem requires it’s use. It is quite literally pulling oneself up by your bootstraps.

    The mind is by definition inexplicable. This why somewhat North of half the global population believe in God (or Gods). Because science can’t explain the mind and it never will because it would be explaining itself to itself.

    Sermon over. Open the pod bay doors HAL!

  • Jake Haye

    There is a balance between the advancement of knowledge and rent seeking on the part of practitioners in all academic fields. Psychology is no exception in that regard.

    NickM: if you believe the brain’s operation (‘the mind’) is not explicable as a computation then you must believe it works by magic. Those are the choices.

  • Julie near Chicago

    No: Even if you can explain the physical system that is a given human body, including all the electrochemical “events” occurring within it, their causes and their effects, including all nervous activity;

    in perfect detail verbally (let’s pretend you have all the concepts and words for them) and also mathematically (since math is the language of the formulaic presentation of physics);

    none of that will convey the experience of being a human being, the experience of thought, how it feels to be thinking, “what it’s like” to be conscious. (Because there IS nothing else that is “like” being conscious.) The only way you can know how those things feel is to experience them directly, yourself, with no intermediary. Our mind is our sensing or experiencing certain of the events occurring within our bodies, but the mind is the territory and the explanations, verbal and physical, are the map, and they are not the same thing, and you can’t “deduce” or “derive” the one from the other.

  • Julie near Chicago

    You can know everything there is to know about a cat, except how it feels to be one.

  • Jake Haye

    You can know everything there is to know about a cat, except how it feels to be one.

    If ‘knowing how it feels to be a cat’ requires ‘being a cat’, I would have to agree. Not sure it’s an epistemologically useful standard though.

  • I came across this blog posting between its first and second comments: lovely post, by the way Rob. My instinct was to comment early, in a way not dissimilar to the very first comment by Ken Mitchell. However, favourable contact with applied psychologists back in the 1970s and 1980s made me hold fire. I then went upstairs to fall asleep to the sounds of the BBC World Service (which is generally more balanced and interesting than BBC Radio 4). And who should turn up but, I assume, one of the authors of the original ‘scientific’ article. Interestingly, he was attackedpressed by the interviewer; the comparison with particle physics was introduced. The guest speaker ended my serious interest in his view by claiming that, as psychology was clearly more difficult to get right than particle physics, psychology was more/better a science than was particle physics. For him, I will give my original knee-jerk reaction to the paper: if it is correct, it claims it has a 66% chance of being wrong ;-)))

    Paul Marks introduces the issue of determinism versus free will. It seems to me obviously clear that some things are more (even all) one than the other and many are mixed in different proportions. I therefore don’t understand his dismissal of “compatiblism” – as a “quagmire of evasion”. However, I must support him on his use of long sentences and frequent parenthetic clarification – his grammar and expression are quite up to the complexity of the issues he discusses.

    Kevin B writes (August 28, 2015 at 10:35 am) that “psychology is not a science”, and he gets stick from Watchman and Jaded Voluntaryist. I have to give some support to their criticism.

    My personal opinion is that the real problem is that psychology is so wide a field (too wide) that some of it is science, though slightly soft round the edges, and some of it is so far from science as to be more liquid than soft. This is because, from the late 1970s to fairly recently, I came across a lot of people who earlier on called themselves applied psychologists, and might now be referred to as cognitive scientists. We were working on perception of speech: them mostly on human brains and me mostly with computers. So they were and are trying to determine how the brain works and particularly how different bits of the brain are (or rather might be) connected together and to sensory organs like the ears, and how they might process the input signals to recognise the sounds as falling into phonetic classes or similar things. Their work was very concerned with repeatability, over different listeners and over different sounds (or sound waveforms). While their results might be probabilistic in kind (especially when noise was added), there was and is no doubt that their experiments were expected to be repeatable – and were repeatable within acceptable bounds, give the noise and probabilistic nature of the reported measurements. On more societal aspects of psychology (not my field, nor closely associated), I would not expect much (basic neural) determinism, and so not much repeatability.

    I don’t recall ever meeting any ‘physics envy’ in that real work. However, I do remember meeting ‘anthropomorphic envy’. An early case (around 1980 I think) was a conference paper on Artificial Neural Networks (ANNs, algorithms for pattern learning and matching based on connections of very many simple basic units that sum, transform and threshold signals). That these so-called ANNs had similarities to neurons and brains was labelled as a justification for using them for functions (eg recognising speech) that were the preserve of human-levels of ‘intelligence’. I felt compelled to challenge with the facts that our better aeroplanes do not flap their wings and our better cars have wheels and not legs; ANNs might well be an interesting and potentially useful class of algorithm – but what matters is how well they work, not what one thinks they represent.

    NickM tries to classify science into ‘hard’ and ‘soft’ and I agree with him to a large extent; but IMHO he goes too far in his tight limitations. And this classification is sort of circular. The hard sciences are generally those that involve the lower levels of the natural word: the natural sciences – with clear repeatability of events in highly similar circumstances. The soft sciences are generally those that involve much less repeatability, largely because they involve individual (human or animal) responses or probabilisticly distributed outcomes from massed numbers of subjects (eg economics and sometimes sociology). And societal masses are different from say the physics of gasses and of statistical thermodynamics – whose individual particles lack the complexity of individual animals and humans. Thus it is the lack of repeatability (or only very approximate predictions) that define the hard/soft boundary. Softness is not a sin, it is a characteristic that makes things different (and largely unknowable in many cases). The sin is expecting ‘knowability’ to arrive with just more: more samples, more/better algorithms, more diligence, more effort, more money.

    In general, the thing about scientific explanation is to predict what will happen next, given a defined set of circumstances. Though very often useful (particularly in predicting or hypothesising the extent of applicability), the why is not really required. Models are in the mind; they are not in the natural sphere (unless you believe in God). There is not much model in a list of outcomes, each with a set of circumstances that will lead to it being selected over the others. But such a model-light theory can be proven scientifically as science. If it works correctly, it is (one view of) the true science. For science, it is the WHAT that matters. Engineering, medicine and other technology do have uses for the WHY and the HOW: not least in reducing costs and otherwise making applications more practical.

    If a hypothesis in psychology does not predict what will happen next in each of a set of different given circumstances (at least in probabilistic terms much better than an average of 1/N for N sets of circumstances), it is not science. But it is time (and money) spent thinking and writing – so knowing it is not the true science might be useful, though not in (tritely) obvious cases.

    Best regards

  • Before anyone (even everyone) jumps on me, the part in brackets of the first sentence of my last paragraph above would be so much better written as “(at least in probabilistic terms much better than each being around 1/N for N sets of circumstances)”.

    Best regards

  • Jake,
    That is not what I argued at all. I argued it was not understood and quite possibly not understandable. Put it like this. You need a 386 to run a Speccy emu. You can’t run a Speccy emu on a Speccy. Now I’m not saying the Doctor can’t run a sim of the human mind because he can but he is a superior intelligence. Just think on the fact that we simply can’t understand ourselves in a meaningful way because we are ourselves and it is like trying to take a multi-tool apart with itself. It is simple because it is complicated. Somethings just are.

    I guess I’m saying true science is real (or pukka) because it is understandable by humans. Consider the four equations of Maxwell. Difficult but simple. Psychologists… If you can describe the beauty of Beethoven’s 7th Symphony via four eqautions I shall listen.

    Spot on.

  • Jaded Voluntaryist

    Nick I get the impression that you are laboring under a rather profound misapprehension about what exactly psychologists do. While there are indeed psychologists who concern themselves with (rather fluffy) subjects such as subjective experience, most psychologists (myself included) avoid subjective experience like the plague and indeed take great pains to design it out of our experiments. Experimental psychologists for example are concerned with how human beings process information. So while they can’t (and don’t try to) describe the subjective experience of beauty, they could tell you which parts of the brain you will utilise when listening to Beethoven, what types of information you are capable of extracting from those sensations, and what sort of impact Beethoven will have on your wider cognitive processes.

  • So while they can’t (and don’t try to) describe the subjective experience of beauty, they could tell you which parts of the brain you will utilise when listening to Beethoven, what types of information you are capable of extracting from those sensations, and what sort of impact Beethoven will have on your wider cognitive processes.

    Er… and that is useful how exactly? As to Beethoven and “wider cognitive processes”. Define, please! I knew a fella at Uni who was a philosophy student and scumbag. He never used the word “think” because he thought it cleverer to use the term “cognitive processing. I have him filed under “T” for “twat”. I assume you are not of his ilk but I do think you confuse things. OK, there is a role for psychology but it is not the science of being human as is often claimed. Admittedly not by academic psychologists.

    But then explain to me why I am a NUFC fan despite 40 years of utter failure.

  • Cristina

    “So while they can’t (and don’t try to) describe the subjective experience of beauty, they could tell you which parts of the brain you will utilise when listening to Beethoven, what types of information you are capable of extracting from those sensations, and what sort of impact Beethoven will have on your wider cognitive processes.”

    I’d be very grateful if you would expand on the above quotation.

  • Jake Haye

    NickM, thanks for your reply.

    As to Beethoven and “wider cognitive processes”. Define, please!

    Perhaps you could state your own criteria for ‘understanding consciousness’ or whatever, as this seems to be the ultimate source of disagreement.

    Progress in all fields of study and the very advance of civilisation itself has come as a result of the rejection of subjective woolly-minded b.s. in favour of ever-improving standards of rigour and objectivity in all things.

    Examples from your pet topic 🙂

    Newtonian -> relativistic mechanics requires the insight that the t coordinate can’t be ignored when calculating the displacement between events, despite subjective intuition to the contrary.

    Classical -> quantum mechanics requires the insight that the observer is part of the system, and that unobserved behaviour can differ from observed behaviour (re particle interactions), despite subjective intuition to the contrary.

    Etc. Examples abound in all fields. Progress comes as a result of recognising and accounting for subjective bias – that’s how I see it.

    This is why, as a fan of civilisation, it so pains me to see wooly-minded b.s. being peddled on important topics by intelligent people that I feel moved to make futile comments about it on a blog 🙂

    Or is it that you think being objective about such things is a threat to some libertarian ideal? Seems that way sometimes.

  • Jaded Voluntaryist

    Er… and that is useful how exactly?

    In the same way as any pure research is useful i.e. the applications are not always immediately apparent but nonetheless pure research is the single greatest driver of innovation precisely because it is not conducted with a particular end in mind. I can however give you examples of how that sort of research has been applied however: namely improving training methods, optimising cockpit designs, understanding brain injury and developmental disorders and so on.

    He never used the word “think” because he thought it cleverer to use the term “cognitive processing.

    I said cognitive processing because that’s what I mean. I’m a cognitive psychologist so I would draw a distinction between “thinking” and “cognitive processing”. The former to my ear carries an implication of voluntary control, whereas the latter does not. It pertains to all information processing on that brain irrespective of whether you are aware that it happens.

    I’d be very grateful if you would expand on the above quotation.

    In not sure I can be more concise than I have been. I’m a cognitive psychologist. I’m interested in human beings from the point of view of them being information processing engines. Therefore I’m interested in how Beethoven or other audio stimuli might impact (for example) Nick’s reaction times in a target detection task. What his personal opinion of Beethoven is does not matter in the kind of research I do.

  • OK, thanks for the reply. I am genuinely curious. Now perhaps you can explain something that utterly bamboozles me. It is probs about the strangest xp of my life…

    For several years my wife has bought Corsodyl toothpaste. I am RG colour-blind. For several years I thought it was green. It is in fact salmon pink. I only discovered this when told and almost by magic it changed colour. It was weird. I navigate the world through a fog but it seems to me that when I know something the scales (almost literally) fall from my eyes. It is very odd. What is really odd is that it is the knowing that makes it so. Very strange. We all, I guess, have perception filters.

  • Jaded Voluntaryist

    Nick we only perceive colour in a tiny part of our retinas (the fovea). The rest of our vision is monochrome. Prior experience and expectations form a big part of our “full colour” visual experience, despite that objectively not being what our physiology provides us with. You can test this by looking at a coloured object you’ve never seen before in the corner of your vision. You will be unable to perceive its colour until you look directly at it.

  • Cristina

    Jaded Voluntaryst, if you are just interested in how any auditory stimuli might impact X’s reaction times in a target detection task, how can you extrapolate your findings to other people, being, as it is, an individual response what you have in the end? Because the response of a person, contrary to that of a machine, is always influenced by previous experiences and expectations, you cannot assume with certainty that this same response will be obtained from Y.
    Do you have to measure each individual’s response in a large study? Do you assume all brain cortices are equal? How and when do you introduce the emotions (the result of those experiences and expectations) into the factors influencing the response? To what extent? How do you decide the magnitude of this influence in each subject and in a group?

  • Thanks JV. I shall try it out! And thank you for an interesting time. Hope you enjoyed it as much as me.

  • Jaded Voluntaryist

    Christina yes, unless you are performing psychometric testing designed to determine the characteristics of individuals, you need to perform quite large experiments with very careful, random sampling designed to eliminate systematic biases. In my career typical sample sizes I’ve dealt with have ranged from 20 (which was honestly a little small) all the way up to 80 participants for a single experiment. Given that a typical cognitive psychology experiment lasts between 30 and 90 minutes per participant, it’s not a small undertaking. Once you exceed a sample size of 30 you can invoke the central limit theorem which makes your findings more generalisable.

  • Cristina

    “very careful, random sampling designed to eliminate systematic biases”
    Well, that’s a minimum standard if you aspire to be taken seriously. I’m interested in knowing about the subject in the experiment, not the experimenter since the subject is, supposedly, the main source of information in said experiment (how any auditory stimuli might impact X’s reaction times in a target detection task).
    How can you be certain that X’s prejudice and expectations are not influencing his response? In other words, how sure are you that X is behaving like a machine? You said you could be able to tell “what sort of impact Beethoven will have on your wider cognitive processes.” How so?

  • Jaded Voluntaryist

    You don’t make inferences at the level of the individual, you collect a large body of data and try and generalise your findings to a whole target population. You select your subjects so that there are no obvious systematic biases, which hopefully leaves only random biases and individual differences. You design your experiment to try and eliminate personal preferences, decision making, hesitation etc (unless that’s what you’re studying). You arrange things so that your participants respond as mechanistically as possible. And finally, if you were actually studying the effects of Beethoven you would probably perform some sort of pre-screening questionnaire to either eliminate people with strong feelings about Beethoven, or at least to identify them so that their native preferences can be included as a factor in your analysis.

    In this way you can begin to get at the level of fairly low level information processing using measures like speed and accuracy.

  • Cristina

    After all that, how close are you of the predictability of the outcome in real life?

  • Jaded Voluntaryist

    You can’t have both empirical and ecological validity in a single experiment. That will take multiple experiments, some with a rigerous laboratory based methodology, and some with a more lax but true-to-life design. For example, eye tracking experiments initially conducted in the lab using very high resolution desk based eye trackers will often be repeated using lower resolution wearable eye trackers in real world settings.

    In some cases it doesn’t really matter. If the subject of interest is the processing of “n+1” words then using the rigerous laboratory approach alone makes sense since any other approach would not allow you to conclusively demonstrate that people were in fact processing the “n+1” words.

  • Alisa

    Cristina, I’d add (and of course JV may correct me) that it all depends on the purpose of the research and the level of expectations. In other words, while one probably can’t build a spaceship based on the conclusions of the kind of research JV is doing, those conclusions can still be very useful in any number of ways as JV has hinted above. For example, one may well be able to predict with a useful level accuracy how astronauts might behave while in a spaceship. Etc. It all depends on one’s expectations of usefulness.

  • Ken Mitchell

    Jake sez: “NickM: if you believe the brain’s operation (‘the mind’) is not explicable as a computation then you must believe it works by magic. Those are the choices.”

    Sometimes the path to wisdom starts with “I don’t know …. yet.” We DO NOT KNOW how the mind works. I would also point out that “magic”, to use Arthur C. Clarke’s words, is equivalent to “any sufficiently advanced technology”. Someday, we’ll probably figure out how the mind works. At that point, it may (or may not) become possible to copy or recreate or synthesize it.

  • Cristina

    Alisa, I very much doubt it. The accuracy with which you could predict any human behavior based on experimentations that do not take into account the whole psychological process is null.
    In that regard I’ve had better results using common knowledge passed down from generations than using the ultimate discovery in the field.

  • Alisa

    Well, if the accuracy is not sufficient for your particular purposes, you’d not use the study conclusions – each person’s mileage may vary. If the accuracy of the studies is not sufficient to a large-enough number of people, the researchers would probably stop conducting them. Saying ‘the studies are not accurate’ as a generalization means nothing, as accuracy is relative and even measurable.

  • Laird

    Per Nigel Sedgewick: “the thing about scientific explanation is to predict what will happen next, given a defined set of circumstances. . . . the why is not really required.”

    I disagree. Science is ultimately about the “why”. If you don’t know (or at least hypothesize) the “why” it’s not science, it’s engineering.

  • Mr Ed

    Meanwhile, the UK’s ‘Professional Body’ for policing, the College of Policing, is advising that police looking for missing persons should not rule out engaging psychics, if they have a record of acceedired success.


    This is after Police Scotland failed to attend the scene, after a real 999 call reporting a crashed car, for 3 days, leaving a man dead and finding a dying woman in the car, she perished in hospital.


  • No Laird, it is unmittaged shite.

  • Laird writes, in response to me:

    I disagree. Science is ultimately about the “why”. If you don’t know (or at least hypothesize) the “why” it’s not science, it’s engineering.

    I acknowledge your making of a double point, that I’m wrong on the WHAT for science; also that the WHY is useful for science and not for engineering, when I claim that the WHY is useful for engineering but not really useful in stating the science (though perhaps useful in conceiving of that science and how it differs from previous and less adequate science on the same matter). My why is simply because of WHAT works in predicting what happens next. Throw in Occam’s razor by all means for favouring the simplest WHAT (and who needs all that extra WHY). We stand apart on the issue.

    NickM writes (I assume more strongly polarised against my view than is Laird):

    No Laird, it is unmittaged [sic] shite.

    Well, in courtesy (to another with a physics degree), I think you should expand on that particular WHAT with some WHY!

    Best regards

  • Jaded Voluntaryist

    The accuracy with which you could predict any human behavior based on experimentations that do not take into account the whole psychological process is null.
    In that regard I’ve had better results using common knowledge passed down from generations than using the ultimate discovery in the field.

    With respect Cristina, that is nonsense. Human beings are not amorphous blobs that randomly reconfigure their whole being from one situation to the next. Certainly in the study of low-level behavioural effects it is reasonable to assume that, for example, in one task involving visual attention, you will employ much the same resources as another task involving visual attention even in a different context. Human beings only have one set of apparatus for performing tasks like that. This is how you are able to make inferences from experiments that have quite poor ecological validity, because in truth there is only one way for people to approach a particular task. It is the study of the neurological constraints on human behaviour.

    Now that’s not to say that ecological validity is not a concern, and it can be addressed in the manner I’ve already described. But it is usually an issue of people performing a task using method “B” in the lab and method “B” + “C” in real life, rather than people using method “X” in real life. The lab will almost always get you close to the truth because there is a limit to how many different ways people can do different tasks.

    That said, one field which is notorious for failing to replicate is Social Psychology. And that is for exactly the reasons you state. Social Psychological phenomena are derived almost entirely from the context in which they occur. Fail to reproduce the context, and you will fail to reproduce the effect. Which is why you wouldn’t (or shouldn’t) study group processes in astronauts in anything but the real situation you are interested in.

    But for the study of perception, attention, reaction times, and low-level information processing this is much less of a concern. Barring neurological or developmental problems, you can say with a fairly high degree of certainty approximately how people will approach a given task. For this reason your findings will have a high degree of generalizability. So while the findings in the OP do indeed demonstrate that Psychology in general has a problem with replication (although I’d point out that my involvement in one of those replications, and the fact that the “audit” was done at all shows something is being done about it), there are fields in Psychology that are very well replicated indeed. Most of Cognitive Psychology in fact. See for example the Posner Paradigm, the Stroop effect, atttentional biases during reading (see the work of Keith Rayner, or the work of Ralf Engbert especially if you like equations modelling human behaviour) or the SNARC effect. These have all been replicated again, and again, and again.

    It is not true that Psychology is contingent on the publication of “novelty papers” that revolutionise the field and are never replicated. Indeed I’d argue that for the most part the papers that are not replicated will tend not to become that influential. Instead you are witnessing the consequences of another phenomena – publish or perish. Academics, especially early in their career, NEED to put papers out in order to be able to get the jobs they want. The journals wont accept anything but significant results, so interesting and well run experiments that nonetheless didn’t work are discarded in favour of naff experiments that found something. Even the authors themselves probably think these papers are flaky, but they need to be putting it out there if they want to keep working. This explains the very high quantity of poor quality research.

    I’ve been saying for years that while it is good to incentivize productivity in academia (where the opportunities for sitting on your arse are almost limitless), published papers is about the worst metric you could choose for that purpose. But there it is. Even I’m in on the game, with my first solo publication from my PhD currently under review (although in my defence it reports an effect I personally replicated 8 times during my PhD so I’m pretty sure it is real 😉 ).

    The next REF is in 2020. A department’s government funding is contingent on the number and impact (number of cites) of published papers its staff have to their name. Interestingly this is regardless of where the research was carried out, so as the REF approaches top universities start poaching staff from one another to bulk up their REF submission. There is a great deal of pressure for people to publish significant results that will get cited, irrespective of the quality of that work. A paper which gets cited a few times and then disappears will do its job just as well as a a higher quality piece of research (unless it is utterly revolutionary, which is of course beyond the reach of most academics). So you get an ocean of papers which make big claims based off small results, and generate a few cites for their authors. They disappear after a while though, and for good reason since the effects couldn’t be replicated. The mistake you made was assuming these papers were supposed to be science. They’re not. They’re padding for someone’s CV.

  • Cristina

    It’s was my personal interest to keep the debate going about Social Psychology and you kept moving it back to Cognitive Psychology. For shame! 😀

  • Laird

    Nigel, I’m not arguing that the “what” isn’t useful for science, merely that the “why” is fundamental. It’s the same basic point as in statistics: correlation is not causation. If the “what” doesn’t ultimately lead to an understanding of the “why” it’s indistinguishable from magic.

    “We stand apart on the issue.” Agreed.

  • […] latter reminds me of the some points on Samizdata, science is hard here and credentialism here. Oh dear, are we having the gap left by the Soviet Union slowly filled […]