This isn't accurate. Newton (and, later, Galileo) described motion, but they didn't presume to explain it. There was no unifying theory to explain what these men described.
I misunderstood. I thought description and generalisation, so that predictions about new observations could be made, were enough. If I got it right now, you demand that a unifying theory also refers to a more fundamental principle.
First, I don't demand anything. Second, a scientific theory must be sufficiently general to move beyond specific cases, to refer to testable explanations rather than to description, and to serve as a unifying bridge between otherwise unrelated areas.
If I say that 100 swans are white, I have described something. If I say swans are always white because this allows them to conceal themselves in a snowy landscape or because a white exterior minimizes heat loss, then we can actually do science on the ground that I have dared to offer an explanation. We could, for example, locate a happy, vigorous flock of black swans.
On the topic of whimsically choosing which outcomes to publish, are you aware that, in the rankings of published negative outcomes (a sign of a field's candor and integrity),
psychology comes in dead last?:
Yes, and I added it to my collection of things my students should know. I have to see whether I can squeeze it into my course or persuade the statistics lecturer to add it. Alcock's critique of Bem was also interesting. The quote from Bem's guide to writing papers severely conflicts with my understanding of good statistical practice.
More important for the general case is the fact that the "soft" sciences are more likely to discard articles that don't live up to the authors' expectations, a behavior that unrealistically reinforces marginal statistical outcomes.
I think there is also a question of how specific a theory needs to be to function as a unifying theory in practice. Tinbergen (1963) wrote that any explanation of behaviour must address four questions: what is its function, what mechanism produces the behaviour, how does the mechanism develop, and what is the evolutionary history? I divide Tinbergen's mechanism into two levels, Marr's algorithmic and implementational levels.
With all respect, don't you see that this is everyday psychobabble? That it puts forth claims that cannot be realistically tested or falsified? Try to imagine how one would go about testing the questions Tinbergen poses.
I have not seen a single paper trying to address all four questions. I have read papers addressing any of three questions and sometimes combinations, and I know there are papers dealing with the fourth, development. In fact, there is a whole field of research dealing with the evolution of development, which covers three of Tinbergen's questions when dealing with the nervous system. I could send you example papers.
The problem is not whether articles are written about this, the problem is that the subject matter isn't falsifiable. No one makes a testable claim that can be clearly refuted (in principle). No one takes an observation or set of observations and tries to extrapolate that to something at a more basic, universal level. It's philosophy, not science.
If I time the fall of a rock from different heights, and if I keep copious notes, I might come up with a rule about the time it takes for a rock — any rock — to make its way to the ground. Then I might take the dangerous step of publishing a rule — an explanation for all such descents under gravity — which could be tested and potentially falsified.
If I say "a ten-kilogram rock fell 44.1 meters in three seconds, as did a one-kilogram rock," no one will bother to refute my claim — they weren't present at that specific event. But if I say, "Neglecting air resistance, I assert that all rocks fall proportional to 1/2 g t2 where t = time in seconds, and I have experimentally determined that g = 9.8," then I have moved beyond description to explanation, to theory. My claim can be tested and possibly falsified.
Given the content of your first post, this example may seem somewhat mechanistic, even trite, but I am trying to make the point that there are claims on which different people can agree, that have an element of objectivity, of universality, and without that property, there can be no science.
Science is not about subjective observations. Science is a discipline that relies on matters on which people can agree, experiments that can be replicated, theories that survive the process of communication.
Most would be from behavioural biology and neuroscience, because I am not a psychologist. I haven't looked much into evolution of development.
Based on the content of your first post, I drew an unwarranted conclusion. My apologies.
Although there is as much reason to believe that perception has evolved as anything else in biology, evolution doesn't take the role of a unifying theory.
Of course it does! Evolution is a classic example of a testable, falsifiable scientific theory, a theory that doesn't only describe, but explains. It offers explanations that can be, and have been, rigorously tested.
My point is not that evolutionary theory is wrong, or untested, or irrelevant. My point is that merely knowing that, for example, the neural networks that measure interaural time difference have evolved doesn't tell me how they work.
But don't you see that, to begin to sort out how this mechanism works, we must first establish that the mechanism exists in the first place, and is common to all human beings? The fact is that we possess a timing method with a survival dimension that can discern sound arrival times in the millisecond range, but using a processor that appears to have a "clock speed" of about 10 Hertz. That is a big step toward an explanation, toward true science.
I say that an organism cannot inherit acquired traits from its parents. The explanation is that genetic inheritance is the only inheritance.
Although it is pedantic, because I'm a biologist I can't let that pass. In the last few years, it has been found that epigenetic markers (which regulate gene expression) can be acquired and inherited for a few generations.
Yes, I know this — I was stating the prevailing principle, not refuting recent work. In any case, when such inheritance mechanisms have a large body of related observations and a clear mechanism — as we have for basic genetics — then it rises to the level of falsifiable theory.
As to your being a biologist, all right, but if you will re-read your first post, you may see how I drew the conclusion I did.
It is important to add that a failed attempt to link evolutionary theory to psychology doesn't condemn evolution, it condemns psychology.
It would also condemn neuroscience.
Only if neuroscience tries to apply evolution in an untestable, unfalsifiable way. One can create a lot of legitimate science without trying to broaden the scope of the work to the degree that the entire structure collapses.
I might study the hearing mechanism mentioned earlier (indeed, I have) and add a small amount to what is known about it, without taking the gratuitous step of trying to force an evolutionary interpretation onto the result.
I personally would never make the above claim unless and until I had gathered some experimental evidence. And if there were evidence, I would print the evidence, not my opinions.
The two reasons why I didn't give you references are: 1) I didn't expect this would be any more controversial than the statement that nervous systems (and hormones, in interaction with nervous systems) produce behaviour,
Easy to say, hard to prove. Not a fertile ground for science. There are any number of studies in which the subjects knew what the experimenters hoped to see, or what the study was meant to discover, as a result of which — to put it diplomatically — such studies tend not to be repeatable.
Evolution acts both by selecting based on function, namely how well a behaviour fulfills functional requirements, and through algorithm and implementation, which determine the cost of fulfilling the function. Evolutionary history further constrains what algorithms and implementations are available for a particular species. That makes it difficult to make predictions specific enough to be tested. Therefore, although evolution should act as the unifying theory for perception, in practice it doesn't.
From a scientific standpoint, this is perfect nonsense.
Would you tell me why?
That's simple — it moves beyond available evidence. It assumes something that, were it to be fully examined, would turn out to require proving a negative. Claims that implicitly require proving a negative are legion, but they are rarely examined carefully enough to uncover this flaw.
So your saying "although evolution should act as the unifying theory for perception, in practice it doesn't" quite obviously leaves the domain of science. We don't know that this is so, and the framing of the claim suffers not only from a lack of evidence but if taken to its logical conclusion would require implicit exhaustive proof of a negative, i.e. a logical error.
I rather thought this was straightforward combination of mainstream neuroscience and mainstream evolutionary biology. Where did I go wrong?
No evidence. And no basis for gathering evidence that could resolve the issue.
And yes, this is a genuine question. I learned far more from a discussion with someone who disagreed with me on a paper I wrote than from those who had compatible views.
Evolution is extremely well-supported by observation and test
I never disputed that. I have repeatedly tried to explain this to creationists until I gave it up as equivalent to wrestling a pig.
We're in complete agreement on that — there's no point arguing with someone who cannot abandon his views, who is emotionally attached to a particular outlook.
including the woolly kind of evidence common in your field.
You don't like people jumping to conclusions, so I have to ask: what did you think my field is when you wrote that, and on what evidence?
As soon as you identified yourself as a biologist, I thought we would eventually get to this. Here is part of the basis on which I wrongly assumed you were a psychologist:
Tinbergen (1963) wrote that any explanation of behaviour must address four questions: what is its function, what mechanism produces the behaviour, how does the mechanism develop, and what is the evolutionary history? I divide Tinbergen's mechanism into two levels, Marr's algorithmic and implementational levels.
The bottom line is that behavior, human behavior in particular, isn't biology, it's psychology.
Apart from the vague nature of your claims, there are basic ethical issues that constrain what experiments we can conduct.
There are no ethical constraints specifically relevant to what I was thinking of, so it seems likely we were not thinking of the same things.
If we're talking about behaviors, I can see plenty of latitude for experiment. If we're talking about human behavior, different rules, different outcomes, and severe ethical constraints. The title of your original post included the word "psychology," a fact that steered my interpretation of its content.
It's easy to say that behaviors evolve, that memes represent social genes, each of which might thrive or expire, but we cannot test this in anything resembling a disciplined scientific study. Which means such claims are not science, they're philosophy.
And they have no relationship I can think of to anything I wrote.
Once one begins to discuss behaviors, without qualifying the term, calling it philosophy seems fair.
I know that my writing style doesn't suit many people, but I can try to rephrase if we can track down where things went wrong. Or we can just treat this as a digression and ignore it.
Fair enough, but your reference to the work of Tinbergen et. al. was sufficiently general to admit the probability that you were addressing human behavior. And for many reasons including ethical ones, human behavior isn't a suitable area for useful scientific research. Ask Margaret Mead (whose magnum opus "Coming of Age in Samoa" turned out to be based on fantasies concocted by her subjects).
Is your criterion concerned with whether a theories acts to unify research in practice, or only in principle?
Let me spell it out:
1. Someone observes and describes a behavior.
2. Someone offers an explanation for the behavior.
3. Someone generalizes the explanation, proposes it as a principle common to all such organisms.
Phenotypic variation will usually force you to come up with a rather restrictive definition of "all such organisms", even if we're not talking about behaviour. You can again ignore that as a digression, but again, as a biologist I can't just let that pass as if I agreed with it. I can accept it as an idealised case, and we can get on with the core argument.
But my outline above (and below) was meant to describe classical science — very specific observations leading to very general conclusions — intentionally to contrast it with the variety of science commonly seen in the soft sciences.
Consider that, with all phenotypic variation, all organisms rely on DNA and all of them appear to demonstrate evolution by natural selection. So there are some universals, and the presence of differences between organisms can be used to produce — indeed, strengthen — an argument for the universality of evolutionary principles.
4. The generalized explanation is tested in controlled conditions and is subject to falsification if the explanation either fails to hold up, or a simpler explanation makes an appearance.
5. The entire study is replicated (i.e. successfully repeated) by different groups with different beliefs and attitudes, but exactly the same level of scientific discipline.
6. The tested idea — the theory — is then perpetually open to falsification by new evidence. It never becomes an unchallengeable law.
None of this addresses my question. I'll rephrase it using chemistry as an example, in the hope I can express myself clearly this time. You write the unifying theory of chemistry is how atoms interact. As I understand it, that would mean solving the wave functions of electrons in shared orbitals.
No, this is not correct. If all of particle physics required writing and solving quantum wave equations, it would be true, but this isn't accurate.
As far as I know, until recently the relevant equations could only be solved for extremely simple molecules, like the hydrogen molecule.
That's true, but chemistry is now a branch of physics for a key reason — most of it can be described and predicted using particle physics, and particle physics is not based on quantum probabilities to the degree you're suggesting.
Particle physics explains what 19th century chemistry described. In that sense, particle physics converts chemistry from stamp collecting into science.
That means that if a chemist had found something in a protein that was incompatible with these equations, no one would have known because no one could solve the equations for that case.
You're overstating the significance of quantum effects in everyday chemistry. 99% of contemporary chemistry is satisfactorily explained using the non-quantum elements in particle theory.
If, in practice, no one can predict the outcome of an experiment from the unifying theory, then this experiment can't falsify the unifying theory. Is that theory then still a unifying theory? That is the point I was making in relation to evolution.
Wow. You just tried to use an argument based on quantum theory to argue against applying evolution to a macroscopic theory. I assure you, the reality of quantum effects can't realistically undermine an assertion about the role of natural selection in the evolution of organisms.
In discussion of this topic, remember this equation:
Δ x Δ p ≥ ℏ (reference)
All other issues aside, this says that, as a collection of atoms increases in size, the prospect for any measurable quantum effects declines dramatically, proportional to the mass/momentum of the collection. By the time we get to a level for which the term "chemistry" is appropriate, and in the majority of cases, we've left quantum behind.
I didn't say evolution is wrong, I didn't say evolution is untestable, I said there are specific cases where, with the information available, it is impossible to make predictions specific enough to be interesting.
But the point is not to make predictions about a specific case, the point is to draw general principles from many specific cases.
If I want to establish that a coin is fair, I can't resolve the issue with one flip, or a dozen flips. I will need to flip the coin many more times, and as the issue becomes important, more data is required in proportion. It's the same with behavior — observing a single organism isn't going to resolve any global issues.
You compare neuroscience favourably with psychology. Although I have some neuroscience background, I am not aware of any unifying theory of neuroscience.
Really? So you don't know about the central nervous system and its role in human behavior?
That is not specific enough to be a unifying theory.
Of course it is. All of current neuroscience relies on certain basic, experimentally derived principles. The propagation rate of nerve impulses. The specific mechanism by which nerve impulses are propagated. The architecture of the nervous system. Things of that sort.
In its present form, we shouldn't compare neuroscience to physics — that would pointlessly embarrass neuroscience. We also shouldn't compare neuroscience to psychology — that would pointlessly embarrass psychology.
From what follows in your reply, I am guessing that you would agree with the more specific statement that human behaviour can be satisfactorily explained by processes within the central and peripheral nervous system (some stuff happens in the spinal cord even without the central nervous system), in interaction with hormones.
Not explained, no. Described, yes. We are, after all, discussing human behavior.
No immaterial soul, life force, animal spirit or such concept is needed to explain anything that has been observed. Is that an accurate statement of your position? If yes, I agree with all of it.
You've mixed two different things. As to the first, we can't explain human behavior using neuroscience, we can only describe it. We can only explain the behavior's precursors, the rudiments. After that, we have to either stop presuming to explain, or abandon science. Behavior is not neuroscience's domain.
As to the second, the presence or absence of a spirit or another supernatural agency, well, that's outside the domain of science, don't you agree? It would be like someone trying to prove the nonexistence of God. It should be enough to say there's no evidence and assert the primacy of the null hypothesis.
A particular science is as much defined by what it excludes as by what it includes. Characterizing the traits of the brain strengthens neuroscience. Avoiding discussions about the mind also does, and in equal measure.
And it doesn't address my question. I spent [deleted for privacy] years reading neuroscience papers. I have never yet attended a psychology conference, but I have attended neuroscience and neuroethology conferences. In all that time, I have never come across a single experiment designed to find out whether the behaviour of an organism with a nervous system and hormones can be satisfactorily explained by the actions of that nervous system and hormones.
That's easy to explain — the topic of behavior is sufficiently vague, and sufficiently difficult to study, that it doesn't represent a suitable topic for scientific research.
This in turn means that behavior is not part of neuroscience. If it were ever to become part of neuroscience, it would compromise the scientific standing of the field.
I don't see a mandate for neuroscience to assume responsibility for any of the topics that psychology pretends to study. It would be more than enough if neuroscience offered testable explanations about the genesis and biochemistry of schizophrenia and bipolar syndrome, to offer two off-the-cuff examples.
If someone possesses behaviors that make him unemployable, and if there is an obvious neurological anomaly, say, one addressed by the administration of lithium, then a neurological theory that leads from a theory about biochemistry to a treatment involving lithium, never needs to address the issue of behavior. Indeed, under those circumstances, addressing behavior would undermine the process.
That is just taken for granted. I expected that a unifying theory should make predictions specific enough to guide the development of hypotheses. This assumption is still far too general that I would have thought of it as a unifying theory.
To the degree that neuroscience has unifying theories — for example, the universal mechanisms of nerve conduction — they are noteworthy for not trying to address the issues you're bringing up, i.e. behavior.
The next parts of your message that I don't reply to I see again as us discussing different points because we have different criteria for how specific a unifying theory should be, and we didn't realise that.
Let me express it in a way that there can't be any doubt about the meaning of the term. In physics, there are certain experimentally derived theories that affect everything in the field — and all subfields.
Example. Someone asserts the equivalence of mass and energy. This claim is met with ridicule or wonderment, eventually grudging acceptance. Because physics is unified by theory, the principle that mass and energy are equivalent affects every field remotely associated with physics, from military strategy to nuclear medicine to ... evolution.
I'm not exaggerating — Einstein's equivalence equation affected the theoretical development of evolution by natural selection, in this way: In the 19th century, Darwin and others tried to argue for natural selection, but one objection was obvious — based on the prevailing idea that the sun's energy derived from gravitational contraction, there simply wasn't enough time for the level of complexity apparent in living organisms. This, by the way, is why Darwin seriously considered the idea of inheritance of acquired traits — it was a way to get around the problem of not enough time.
With his now-famous equation, Einstein removed this obstacle. The sun's source of energy is mass-energy conversion, not gravitational contraction, the sun is suddenly thousands of times older, and the time available for natural selection increase proportionally.
This is an example in which an abstract idea in mathematical physics resolved a problem in a specific area of biology. That's what I mean by theoretical unification.
we have different criteria for how specific a unifying theory should be
Not any more, I hope. The above example is just one of dozens of similar examples that come to mind, that demonstrate the role of theoretical unification in science.
Psychology studies the mind. Neuroscience studies the brain and nervous system. This gives neuroscience an overwhelming advantage because there really is a brain and a nervous system.
You know more about software than I will ever learn, so you can correct me if I am wrong in the following. Say you want to sort some data. Is there a useful description of what you do between the specification "I want to sort the data" and the manipulation of electrons on the chip?
Yes. Simultaneously specific and useful. Because I worked in electronic design before I switched careers and took up software design, I am in a position to know something about each of the steps on the way from circuits to algorithms.
I read there are many sorting algorithms, described at a level far removed from the manipulation of electrons. The point was illustrated by a description of bubblesort, because it's simple, but it was described as inefficient. My understanding is that there is a useful level of analysis between the specification of the computation and the implementation on the chip, and that level (or possibly a set of levels) is named the algorithmic level. Would you say that level exists or not?
If you are asking whether algorithms exist apart from their implementation, well, yes — they exist in that sense. They exist because the same algorithm can take many forms, which argues for their universality. For example, If I say that an organism can minimize the risk of predators by only coming out of hiding during prime-numbered years, I can shape an algorithm, a model to test the thesis, but I can also locate examples in nature: Mathematical Locusts
Different embodiment, same algorithm.
If there is such a thing as an algorithmic level, then I argue that mind is the algorithmic level description of behaviour.
I'm reasonably certain you know that Stephen Wolfram (in "A new Kind of Science") takes this same position. The problem is that it's easy to say but difficult to quantify to the degree that any particular assertion can be tested and falsified. It's hard to turn it into science.
It is a useful level of analysis between the details of what happens at the level of synapses and neurons and the level of behaviour.
That is not a crack in the pavement, it is a chasm. Imagine proposing a testable, repeatable, falsifiable connection between a specific algorithm and a behavior, or between the operation of synapses and human behavior. My point is that a sufficiently large quantitative difference becomes a qualitative difference.
I could give you examples, including some of my papers, if you want. However, the one I judge most likely to be interesting enough for you that you don't feel you're wasting your time is one by [deleted for privacy] on schizophrenia. He doesn't bother taking a position on whether schizophrenia is a biological condition with a behavioural phenotype, I think because he just takes that for granted. And although the title promises a hypothesis of the NEURAL basis of hallucinations and delusions, both the explanation and most of the evidence presented are at what I consider the algorithmic level. Do you think [deleted for privacy] explanation in terms of efference copy, forward model and attribution bias is useful?
Are we still discussing science? Yes? Then no. It would be germane if it could be reduced to a repeatable experiment that tied a theory to an unambiguous measurement. I think you will agree that this is not feasible.
The bottom line: is it falsifiable through practical experiment? No? Not science.
I have two final questions: I was prompted to read your latest argument about psychology by a student of clinical psychology. She is very determined to do something useful, and you got her worried she might be wasting her time. When I mentioned I had written to you, she asked whether she could read your reply. Do you allow me to let her see our discussion?
Yes, certainly, especially given that I'm also posting this discussion in one of my online discussion boards (with identifying information removed). On that topic, if I don't remove enough information to assure your privacy, please speak up.
I don't know whether she might want to join in. There is also someone in the neuroscience group whose opinion I would find interesting, if he could be persuaded to join the discussion. [deleted for privacy] works on computational neuroscience. I rather expect he thinks there is an algorithmic level, and it would be interesting whether he also thinks the mind is an algorithmic level description. Would you be interested in either of them joining, or have you reached the limit of how much time you want to invest?
To me, the question whether there is an algorithmic level (I assume you mean in human behavior) is at present not a suitable scientific topic. I would be happy to discuss this to some degree, but I tend to lose interest in philosophical discussions fairly quickly.
The problem with the question is the same one that confronted Stephen Wolfram on publication of his huge book "A new Kind of Science" (i.e. everything is an algorithm) — it's really philosophy at the moment, it cannot be resolved by evidence.
Second, what would you recommend the student should do? Anything other than quit psychology and do something else?
I think you can anticipate my reply. This student needs to understand the role of science on the modern world — that anything worthwhile, anything able to rise above pointless and endless argument, must have a scientific dimension. It's clear that psychology doesn't have this property, and is unlikely to change enough to allow for this, without disintegrating in the process.
The reason why I ask is that I ... have been volunteered ...
That's a nice bit of prose — "been volunteered". It reminds me of the old battlefield joke — a sergeant addresses an obviously involuntary audience of three sad sacks, saying "Now I want three volunteers ..."
... to start up a regular series of seminars. I want to lead my colleagues and students gently into the habit of attending. If I manage that, they might then benefit from being severely provoked to think.
I can think of some topics that might lead to high attendance. In fact, years ago when I lectured more than I do now, I tended to extol the virtues of science, much as I have been during this exchange, to enthusiastic audiences who (it seems) hadn't until then understood what it actually was in any detail.
Would you be interested in giving a talk via teleconference?
I'll say I'm tentatively interested.
If you tell them just to give up psychology, I expect they'll dismiss you as a crackpot, and the whole thing would be a waste of your and their time, regardless of the merit of your argument.
Yes, I've had the experience, and I agree. But as it turns out, the present director of the U.S. National Institutes of Mental Health has recently taken that exact position (i.e. psychology should be abandoned), apparently to a wide and respectful audience among scientists: Scientific American: Faulty Circuits
I think his position made his position more palatable, if you follow. :)
If they walk away determined to prove you wrong, or that they can do useful things anyway by insisting on giving evidence-based treatment even in the absence of a unifying theory, I think it would be good for them. What do you think?
Apart from any other considerations, I think it's worthwhile to make people aware that there is a role for evidence, and dispassionate evidence gathering, in virtually every field of endeavor. And that the first sign of an absence of evidence is long and heated argument with no resolution.
Again, thanks for writing.
Share This Page