The elephant in the room


Part two of my ten-part investigation into the purpose of education, following the inquiry of the House of Commons Select Committee, examines logical positivism

Before I address Professor Biesta’s reply to the question I posed at the end of part one, I am going to spend two instalments discussing logical positivism. Even though this requires a bit of a philosophical dive, I think it is justified because without understanding the fundamental argument about logical positivism, the more concrete disagreements about educational purpose will be difficult to untangle. The elephant represents the fact that all the other, non-elephantine occupants of the room are divided between two almost completely incompatible ways of thinking about truth and logic—they follow incommensurable paradigms, in the approved, post-modernist jargon—and without understanding that important fact, it will not be possible to understand why so many people seem to be talking past each other, almost as if speaking different languages.

The view that I expressed in my question, that objectives which cannot be measured are for that reason meaningless, follows from the principle of verificationism (and Karl Popper’s variation, falsificationism), which is the central tenet of logical positivism. Taken together, these theories represent a viewpoint that is widely thought of by educationalists as having gone out with the ark, the term “logical positivism” being described by Erie Bredo in the The Philosophy of Education: an Enclyclopedia as “a generalized term of abuse”. The truth of this description has been attested by Greg Ashman in a recent blog-post, On the subject of my irrepressible positivity.

Although educationalists often speak dismissively of logical positivism, it remains fundamental to the practice of natural sciences such as physics. Here, for example, is a passage from Einstein’s 1921 Relativity: the special and general theory, which asks what is meant by a statement that two bolts of lightning occur simultaneously (p.26).

The concept does not exist for the physicist until he has the possibility of discovering whether or not it is fulfilled in an actual case. We require a definition of simultaneity such that this definition supplies us with the method by means of which, in the present case, he can decide by experiment whether or not both the lightning strokes occurred simultaneously. As long as this requirement is not satisfied, I allow myself to be deceived as a physicist (and of course the same applies if I am not a physicist), when I imagine that I am able to attach a meaning to the statement of simultaneity. (I would ask the reader not to proceed farther until he is fully convinced on this point.)

Einstein’s description of the conditions under which the term “simultaneity” becomes meaningful represents a straight up-and-down statement of the principle of verificationism. Nor is Einstein kidding when suggests non-compliant readers should put down the book: without understanding that the experiment is not just a way of demonstrating simultaneity but is actually what is meant by simultaneity, you simply cannot follow Einstein’s argument. Yet I suggest that modern educationalists who say that we should adopt educational objectives whose attainment we do not know how to ascertain, would in all honesty have to put down Einstein’s book at this point, being unable to accept the basic principle on which all modern science is premised. Educationalists and physicists have incompatible intellectual models of truth and meaning.

This intellectual chasm was expressed in what became known as the Science Wars in the 1990s. The opening shot in these wars was fired in the form of a leading article in the American edition of the Economist, written by Richard Lewontin and entitled 74.6% of Sociology is Bunk:

The GDP of Denmark is not to be found by asking a statistically significant random sample of people to have a stab at it. What people think, still less what they say, is not a good guide to the way the world is. People don’t know. People lie.

Lewontin attacks the methodology of many sociologists (and indeed educationalists), which commonly focuses on attitudinal surveys or, to use the language of Kant, “phenomenological research”. Such a methodology addresses what people think, which will always be influence by their values, rather than “the way the world is”.

A year after Lewontin’s leader, New York physicist Alan Sokal produced what became known as the Sokal hoax. He persuaded a leading academic journal of postmodern cultural studies to publish a spoof article that he himself had written, Transgressing the Boundaries: Towards a Transformative Hermeneutics of Quantum Gravity, which he declared on the day of publication to be:

a pastiche of left-wing cant, fawning references, grandiose quotations, and outright nonsense.

The eminent journal that had published this nonsense, and the discipline that it represented, clearly lacked discrimination.

In response to these attacks, educationalists come up with a number of reasons for dismissing logical positivism. It might be applicable, it is said, to the natural sciences, but not to the more complex world of human interactions where causal relationships do not apply; it will narrow our priorities to those things that are easy to measure rather than to things that matter; it tends to propose simplistic silver-bullet, black-and-white answers; it does not take account of the variability of human experience or the multiplicity of causal influences that the teacher must juggle; it does not take account of the importance in education of subjective values.

Most of these objections can be quickly dismissed.

Relativity is not a simple thing to verify. At the time that Einstein formulated the theory, no-one would be able to demonstrate its truth by experimental means for several decades to come. Verificationism does not require that the experiment should be feasible or even accurate. Erie Bredo is wrong to say that positivism can be refuted because any experiment might be flawed, as “maybe the electron microscope one is looking through is not working properly” (ibid p.494). That does not matter. All that the principle requires is that you can state clearly what the experiment would be, if you had the means to conduct it, such that it would produce a definite answer of some sort, right or wrong (read through to the top of page 28 in Einstein’s engaging explanation of this point). Positivism is about creating meaningful statements, not true or certain statements. Nor is it about measuring only what is easy or even possible to measure.

Some people dismiss the methods of quantitative research because of the multiplicity of causal influences in the classroom, which, they claim, invalidates the methodology of quantitative research when applied to the classroom. One such academic writes on a community email reflector:

The weight of positivism is heavy…[but] laboratory-type protocols were insufficient to apply to the educational context because no two classes are the same nor behave the same.

This position demonstrates a failure to understand quantitative research methodology. The whole point of Randomized Control Trials is not to harmonize all the confounding variables but to randomize them. The challenge posed by RCTs is not that they cannot on principle be used in educational contexts, but that to do so they require data of a quality and in a quantity that is not generally produced by our current education system. For that, like Einstein, it seems we must wait.

With respect to the supposed impossibility of predicting causal relationships in human interactions, it is true that one can never be entirely sure how other people will react. That does not mean that conclusions which fall short of certainty are invalid or of no use. It is both possible and useful to know, given certain sorts of precondition, what are likely to be the most effective sorts of action in order to attain a certain sort of objective. This is what an orator learns when she learns how to make a better speech or a policeman when he learns how better to manage a confrontation: it is commonplace that people managing human relationships seek useful knowledge about the probable causal relationships between human actions and human responses. Without such causal relationships, the disciplines of psychology and neuroscience would be impossible.

There is nevertheless one criticism of positivism that is, perhaps, more substantive than the others. This states that the empirical evidence which is used to justify a position is itself value-laden and so the act of justification is circular—rigged, you might say. This is a view that has become influential ever since the publication of The Structure of Scientific Revolutions by Thomas Kuhn in 1962.

That will be the subject of part three, which will wrap up my discussion of logical positivism and its critics.

11 thoughts on “The elephant in the room

  1. Operationalisation vs definition
    If I’ve understood Einstein correctly (!) he was saying that ‘simultaneity’ is what happens when two lightning bolts strike at the same time. For physicists, meaning and definition and operationalisation are often one and the same thing. But social scientists don’t deal with relatively simple entities like lightning bolts, they deal with fuzzy constructs that people construe differently or with constructs on which there might be a high level of consensus, but which as far as we know, don’t exist – such as Santa Claus or fairies. So if you are carrying out an empirical study on ‘anger’ or ‘progressivism’ you need to explain very precisely how you are measuring that construct, or what it maps on to within the framework of your study. If you don’t, we end up with a plethora of studies purporting to be about the same thing, when closer inspection shows them to be about different things, or even about things that might not exist in the real world.

    • Thanks for that explanation of operationalisation (this is an overspill from a discussion on Twitter

      I think Einstein was saying something different, which is that the simultaneity of two lighting strikes *means* this: that if you were to anticipate the position of the strikes and find the exact mid-point between them and were to place two mirrors at 45% such that the light from the strikes were deflected onto a light sensor, the sensor would register the light from the two strikes at exactly the same time. In other words, there is something tautological about defining simultaneous as “two lightning bolts strike at the same time” – because that is what you imagine is happening “out there” and your imagination is shaped by the word you are using. He is saying that the definition has to be founded in empirical observation & methodology.

      That maps very precisely, it seems to me, to what you are saying about defining anger. So it seems to me that we are agreeing much more about this that the words we use suggest (I don’t think you are comfortable with “positivism” but use “operationalisation” instead). The other thing that we disagree about is that social science needs to be so very different to natural science in its methodology, because of the complexity of its subject. I would argue (a) that natural science is pretty complex too, and (b) even if human behaviour is not entirely predictable, it is still subject to probabilistic statements which can be verified and which are still useful – and the introduction of uncertainty does not fundamentally change the experimental methodology.

      Finally, I would say (and I suspect that you would not disagree) that having defined different experimental definitions of something like anger (physiological, self-reported, third-party reported etc), we can reasonably expect to show a fairly high level of correlation between these different forms of observational definition. So that the common sense view that there is a single concept called “anger” is almost certainly justified – we just have to approach that slightly elusive common sense concept through a number of more precisely defined technical ways of representing that concept.

      My argument in this series will be that we need to adopt exactly the same methodology with respect to our educational objectives: that we define a number of performative proxies for a fuzzy concept like “creativity”, and demonstrate the consistency (of lack of consistency) between those different proxies by statistical means. But that is going beyond the scope of our original discussion!

      • I think we’re broadly in agreement. Even about Einstein. I just didn’t go into all the detail 😉 But I don’t think ‘positivism’ and ‘operationalisation’ are the same. As I understand it, ‘positivism’ holds that you can know things (for sure), that you can verify them (for sure) and limits empirical findings to observations that can be made using the senses (or proxies for them). By ‘operationalisation’ I mean simply making it clear what parameters are being used for a particular construct in a particular study.

        • Hi Sue,

          That is not my understanding of verificationism. Here, for example, is a definition from Language, Truth and Logic by the twenty-something A J Ayer:

          “A sentence will be factually significant to a given person if and only if he knows how to verify the proposition which it purports to express; that is, if he knows what observations would lead him under certain conditions to accept the proposition as being true, or reject it as being false” (p.148)

          Einstein says something very similar (it is worth reading the 2-3 short pages in my link). It is not that the verification must be for sure – it is just that you must state what you would hold such a verification would be. That is why I suggest that it is all about meaning, not about certainty.

          I am not entirely sure what you mean by “what parameters are being used for a particular construct”, but it sounds to be something very similar – the language is not significant but I would rather say “what observable phenomena are to be taken as sufficient to indicate the validity of a particular hypothesis”? Am I right that these two phrases are saying much the same thing?

          I think the distinction (or lack of it) between absolute truth and the arrogance of assuming certainty often hangs around these conversations. I have in some drawer an ancient cassette tape, probably of rather similar vintage to the Ayer conversation that I link to, in which someone, possibly Bryan Magee, comments on the assumption that these two are connected. He makes the point that in fact the opposite is the case: the person who believes in evidence and reason is aware that anything that he or she says can be challenged and shown to be wrong; while the relativist (or the person who says things in terms which cannot be verified/operationalised) is immune from challenge. The methodology of the empirical research does not claim certainty but requires its experiments to be repeatable. This is why Marxist states, Magee continued, which were founded on an ideology that was officially relativist, invariably turned to tyrannies and the intellectual always ended up in the gulag. I think it is a point worth reiterating in the era of “post-truth politics”!


          • Operationalisation is about ensuring that people are talking about the same thing. Whether or not the thing exists or not, or whether you can test hypotheses about are other matters. For example, there have been probably hundreds of studies about ‘motivation’ or ‘stress’. But unless investigators say exactly what they mean by ‘motivation’ or ‘stress’ (usually boiling down to how they measure it) it’s not clear whether or not they are all referring to the same phenomenon – real or otherwise.

            Operationalising constructs is clearly related to verificationism but isn’t the same as it.

            • But your definition of operationalisation (at least in respect of the usual practice of measuring it) is exactly the same as my definition of verificationism (and Ayer’s and Einstein’s, it appears). Doesn’t “testing hypotheses about it” and “measuring it” almost always come down to the same thing? Doesn’t a “statement” come down to the same thing as a “hypothesis”?

              • Say the hypothesis is that ‘x does y’. An experiment demonstrating an instance where x does not do y falsifies the hypothesis.

                I’m not talking about the process of verifying or falsifying the hypothesis, I’m talking about developing a precise enough definition of x and y for the hypothesis to be tested.

                • But x and y also need to be individually verified, to the extent of saying “this set of circumstances amounts to what I mean by x/y”. That could also be seen as a hypothesis. That is what Einstein is talking about: when he states the way that he would verify that two lightning bolts were simultaneous, he is effectively stating how he would verify how any two events were simultaneous. It is the same as defining “simultaneous”. So once you have defined x and y, it becomes obvious how one should demonstrate that the hypothesis “x is y” is true.

                  I am trying furiously to agree with you that operationalism is a good thing – but you do not want to be associated with verificationism, I think only on the grounds that you have heard bad things about it oin the street. I still can’t see that they are different in any significant respect.

                • I’m talking about how you would determine what constituted a lightning bolt, not what the lightning bolts do. Lightning, in its various forms, is pretty easy to operationalise compared to a complex construct like ‘progressivism’.

                  But operationalising lightning bolts or progressivism isn’t the same process as testing hypotheses involving them. If the hypothesis is ‘x does y’, operationalisation deals only with the x and y, not the ‘does’.

                  The problem I have with verificationism is that, as I understand it, it implies the possibility of conclusive verification (as well as falsification). As I understand it, Popper’s point about falsification was that you can never conclusively verify anything, because conclusive verification requires knowledge of all possibilities, something we can never have.

  2. Verificationism
    I think the term is misleading. Again, if I’ve understood correctly, it refers to the idea that ‘meaning’ is restricted to phenomena can be verified or falsified empirically. I wouldn’t describe Popper’s idea of falsification principle as variant of it, because our knowledge is incomplete. We can only say conclusively that something doesn’t exist, or an hypothesis isn’t an explanation for something, only if we can demonstrate that. We can’t say anything conclusively about anything else.

    • I would say that meaning is a quality of statements, not of phenomena. So I don’t think verificationism is trying to get reality to fit into the straight jacket of what we can say about reality – you are only talking about what we can say and letting reality look after itself.

      I agree that falsificationism is stronger than verificationism for the reason that you give – that positive statements are provisional, while falsifications tend to be more conclusive. However, I think it is perhaps a bit more complicated than that. As I say in my post, positivism is about meaning, not truth. So all it asks is that you should state the empirical evidence that you would accept as sufficient to verify the statement, not for the empirical evidence that *would* conclusively verify the statement. And I don’t think this is a trivial distinction because it is a commonplace that we have to act on incomplete evidence.

      Second, in the social sciences, which as I admit in my previous reply, everything is likely to be probabilistic, I am not sure that falsification is any more conclusive than verification. You might show that groupwork does *not* work in context x – but that does not disprove that it *would* work in context y. So we are not going to be dealing in crisp, binary true/false assertions, but probabilities and tendencies. That does not mean that the sort of empirical “cash value” (with which I conclude my two part discussion of positivism) is still not vitally important. Which is, again, what I understand you to be saying in relation to the importance of operationalising your definition of anger, I guess in the context of psychology experiments.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s