Monday 16 February 2015

Turtles all the way down


Some people are fond of silly arguments. For example, that opium makes you sleepy because of its virtus dormitiva, as Moliere jested. For example, that people can see because the image cast onto the retina is looked at by an inner eye, which presumably also  has a retina. Instincts, drive states, and countless other  hypothetical constructs are very popular in psychological texts. Id and ego, for example, or black boxes and other dragons that are constructed to obscure the simple fact that most of the time we do not know what causes behaviour. Indeed, it sometimes seems that psychology isn’t possible, not real psychology anyway.

A well-known scientist (some say it was Bertrand Russell) once gave a public lecture on astronomy. He described how the earth orbits around the sun and how the sun, in turn, orbits around the centre of a vast collection of stars called our galaxy. At the end of the lecture, a little old lady at the back of the room got up and said: "What you have told us is rubbish. The world is really a flat plate supported on the back of a giant tortoise." The scientist gave a superior smile before replying, "What is the tortoise standing on?" "You're very clever, young man, very clever," said the old lady. "But it's turtles all the way down!" (Hawking, 1988, Brief History of Time, pp. 1, para 1).

If I had a £10 pound note for every paper I have read with a black box with arrows going in and out I would be writing this blog on the back of my yacht. Absent this income stream, I have decided to enter the modern commercial world by invoicing any colleague whose work I mention. First to be served with a bill under this new harsh regime is Rogier Kievit who innocently mentioned that in the introduction to his thesis he covers the issues of: the reduction problem; explanatory levels and the unity of science; the mind-body problem; and the dangers of greedy (neuro) reductionism.

Kievit quotes Oppenheim and Putnam’s (1958) proposal that science consists of a hierarchy of explanatory levels that range from those concerned with the ‘large’, or ‘high’ (e.g. social groups) to the ‘small’, or ‘low’ (e.g. atoms or subatomic particles). As entities at a higher level can be said to consist of (be made of) elements at the lower level, science consists of basic, physical building blocks that aggregate to form the hierarchy of sciences. They propose that theories at higher explanatory levels of psychology can be reduced to lower order levels of biology if and only if: 1) The vocabulary of psychology contains terms not in the vocabulary of biology 2) Any observational data explainable by psychology are explainable by biology 3) biology is at least as well systematized as psychology.

Kievit finds this somewhat lacking, and prefers Ernest Nagel’s (1961) suggestion that at every explanatory level, we can discover regularities, or laws, which describe the interaction of entities at that explanatory level (e.g. brain regions interacting). However, as the objects at a certain level (e.g. brains) consist of objects at a lower level (e.g. biochemical molecules), it should be possible to discover law-like statements that translate the regularities at one level (brain regions interacting) into the regularities at a lower level (properties of biochemical molecules). Nagel called such statements bridge laws. Ultimately, a complete specification of these bridge laws would allow us to do away with any ontological commitments other than of the units at the lowest level: The bridge laws would allow us to analytically translate any scientific regularity either ‘upwards’ or ‘downwards’.

He adds that Daniel Dennett (1995) distinguished between reductionism in general on the one hand, described as sensible attempts by science to explain larger units as parts of a whole, and greedy reductionism on the other. Greedy reductionism occurs when attempts are made to reduce phenomena at higher levels to lower levels across ‘large gaps’, without well-established intermediate steps. He uses the metaphor of a crane to represent a scientific or conceptual tool to translate or explain findings at one level from a level ‘down’. If one argues, on the basis of a physicalist perspective on the hierarchy of the sciences that the mind can (or should) be explained by physics or chemistry without establishing the solid intermediate steps required (cranes that can do the ‘work’), one is guilty of greedy reductionism. Similarly, we cannot simply proclaim that because the brain is necessary for psychological processes that all psychology can be reduced to the brain: We have to develop the scientific cranes that allow us to do the work of successfully reducing psychological phenomena to (increasingly) lower levels.

There, that plug should boost my assets by £10 which translates to four bags of coffee, or about a quarter of an inch of super-yacht teak decking. Now all I have to do is wait for the donations to flow in.

However, before I go, I should let Kievit off any need to donate anything more than that which he has kindly already done by sending me part of his thesis. The reason is that I have been trying to defend reductionism in general against greedy reductionism for quite some while, without knowing that last phrase of Dennett’s.

Years ago, in a restaurant near a conference venue, I sat down to dinner with a scholar who was concerned with the real world consequences of intellectual differences, with every prospect of us having a far ranging conversation about civilisations and intellectual classes. Another colleague joined us, and propounded the view that nothing much could be said about intelligence until we understood the physiological underpinnings at the cellular level. Time was precious, and I did not have Kievit’s as yet unwritten thesis in my hand, so I could not make accusations of greedy reductionism or call him to produce metaphorical cranes in defence of his views. Nonetheless, had I had the chance I would have said something like the following: 

Every measure has its range of convenience. The units used to measure fine tolerances of machine tools are not suitable for stating the distance between those machines on a factory floor, let alone the distances between towns. More importantly, computer machine code is not always the best method of discussing the relative merits of two different types of computer program. In that context we normally discuss the usability of the programs, their power in doing particular tasks, and the advantages and disadvantages of the way they depict the results. We use higher level concepts to discuss high level matters, without denying that computers use machine code.

Typically, different levels of analysis are used in psychological explanations for events at the cellular, organ, person and social level. Of course it is important to try to link them together, but we need to establish the solid intermediate steps required (cranes that can do the ‘work’). Even then, we will probably use explanations which are suitable for the level of our discussion. As Gilbert Ryle observed: Golfers can simultaneously observe the rules of gravity and the rules of the game.

So, here is my rather late statement to all persons skipping from one level of analysis to the other: please don’t jump from serotonin levels to battles between nations, from fMRI to brain states, and from brain states to emotions without doing the linkage work required.


  1. "So, here is my rather late statement to all persons skipping from one level of analysis to the other: please don’t jump from serotonin levels to battles between nations, from fMRI to brain states, and from brain states to emotions without doing the linkage work required."

    I definitely see your point. Perhaps spurred by this recent paper?

    I would state it this way (because the crane argument can be and often is abused):

    It's not always necessary to know how one thing causes/affects something else to know that it does.

    You should be sure that it in fact does, though.

    1. Dear JayMan, Hadn't seen that paper, and will read it with interest. Mostly I was having a twitter conversation with Rogier as to whether people are still strongly influenced by seeing fMRI pictures of brains showing the centres for emotions and so on

  2. Distantly related: "It stands to reason" is an idiomatic expression in English that means "Neither logic nor evidence supports my opinion".

  3. "It stands to reason" is a matter of common sense, at the end of the day.

  4. "A well-known scientist (some say it was Bertrand Russell) ...": he's just trying to start a fight, isn't he?

  5. You will need for scholarship essay writing guidance our writers are waiting for you.

  6. "So, here is my rather late statement to all persons skipping from one level of analysis to the other: please don’t jump from serotonin levels to battles between nations, from fMRI to brain states, and from brain states to emotions without doing the linkage work required."

    This is an interesting position, partly because I don't know whether I agree with it. What is the problem, for instance, between using correlational research to show links between levels?

    For the sake of argument, imagine that I found yearly accident rates were correlated with sunspot activity. Since it is infeasible for accidents on Earth to cause sunspots, can't we use this to show causation without needing to demonstrate that (say) sunspots affect magnetic fields in the brain which disrupt neurotransmitter pathways in the hippocampus which contribute to problems with concentration and fine motor control which result in increased accident proneness?

  7. Dear Mark, You are right that a correlation can detect a cause, and often does. It is necessary, but not sufficient. It is the way to bet, but not to bet the house.