The illusion of control

The stories we tell about “who did what” all revolve around the collective illusion that we are in “control” of ourselves, in control of our bodies, in control of what happens around us. Control metaphors all demand a distinction between the controller and the controlled.

When your car is hurtling down the motorway, adhering to the conventions, rules and practices of motoring, you are staying in lane, using signals, going the right way, at a speed matched to those of your fellow motorists. Your contribution to this whole thing is minimal. You twitch the steering wheel a little, and gently pump pedals with your feet. You are a component in a superordinate system, and all is well.

Now your car has a blowout, and it swerves violently, tumbling through the air, over the hedge and into a tree, where it hangs upside down, steaming. At the moment the tyre burst, the road markings and speed conventions lost their ability to constrain the car. At that moment, you got back control. We speak of “losing control”. The opposite is the case. You regained it.

No wonder Boris Johnson speaks of regaining control.

Representation for whom? Who is the subject?

When something is about something else, that relationship needs to be understood. There are very many ways in which one thing can bear a systematic relationship to another, but we would not consider most of them to be representational. When a foot leaves a print in the sand, we understand the causal chain connecting the residual trace with the originating event. We can infer many things about the original from the trace, and there are many we cannot infer. The footprint might be considered a “representation”, but because we are knowledgable about the entire causal chain, I would prefer to use the term “imprint” in this instance.* Imprints are many in kind, and one imprint may lead to another, as when one tree falls and knocks another down in a forest.

But we typically use the word “representation” for cases in which the causal chains are opaque, unknown and possibly unknowable. I want here to consider a relatively simple case that most would find unproblematically “representational”, because if we don’t understand what the notion of representation entails for simple cases, we will be in deep trouble when it comes to representations of a more hypothetical character. As a cognitive scientist, I am particularly concerned with the way the term “representation” is wielded in discussion of minds, brains, intentions, and the like, but the hard questions about representation extend far beyond such provincial concerns. They have formed the basis for intense theological dispute, schisms (e.g. between the Oriental and Western Christian churches), repeated paroxysms of iconoclasm, and even wars. Perhaps the violent execution of the Charlie Hebdo cartoonists for the perceived crime of representing the Prophet might serve to make it clear that the issues involved are wide ranging, important, and contemporary.

The simple case I want to discuss is a straightforward picture, nay, a photograph, of something. Let’s use this picture of a cat as an example. In what sense is this picture a representation of a cat? Perhaps the reader might attempt an initial answer before reading below the fold.


Read the rest of this entry »

Christianity, representation and psychology

I a previous post I cryptically alluded to a possible link between the Christian sphere and the readiness of psychologists to accept an image as a substitute for the thing itself (in that case, for a real flesh-and-blood face). I realise this is insufficient, and I cannot do it justice here either, but there is an important conversation that needs having.

When I say “Christian”, I do not mean one institutional church or another, nor do I mean one set of beliefs or another. Indeed, the notion of belief as underlying the rational deliberation of an autonomous agent is itself resolutely “Christian” in the sense I mean. I could say “Western” instead, but I don’t like that, it leads to fatuous East/West orientalism, and it dodges the theological import of many ideas that shape our contemporary discussion within the human sciences. More after the fold. Read the rest of this entry »

The science of “face perception” is bankrupt

There is a vast literature on the topic of “face perception”. This is one of the pillars of contemporary cognitive neuroscience. There are many competing models, informed by many kinds of experiment, both behavioural and neuroimaging. This field sits at the core of contemporary view of functional brain organisation. The finding of the field will inform not just speculative models, but real flesh and blood brain surgery.

It is troubling then to recognise that the entire literature is built on a simple and obvious conceptual error: experimental subjects are not viewing faces at all. They are viewing images of faces. Whether using still photographs, or moving videos, subjects are almost* never exposed to a real face (except perhaps outside of the framework of observation, as they arrive at the lab and greet the experimenter). Once the experiment begins, no faces are employed.

Isn’t looking at a picture of a face almost the same as looking at a face in the flesh? Absolutely not! If a real face is present, that face might just turn to you and snarl, spit, bite, kiss, shout or smile. No image will do that. Being in the presence of a person is not a mere detail. It is a completely different proposition from being in the presence of a picture.

But the belief exhibited by researchers that an image can stand for a real face in this way is revealing. Representational commitments of Cartesian, cognitivist approaches to minds go beyond the interpretation of patterns of brain activity. Here we see a consequence of a specifically Western, Christian approach to the relation of immanence and transcendence that needs a much fuller treatment.

* In a brief quip on Twitter, I made the overly strong claim that there were no studies at all that used real live faces. A few exceptions, listed below were pointed out to me. These constitute a tiny drop in the ocean, and do not materially change the observation that the field as a whole is based on the viewing of images, rather than faces.

Here are two studies that actually thematise the distinction between images and real faces. Using ERP methodology they find marked differences in brain activity when viewing real faces compared with images.

Pönkänen, L. M., Hietanen, J. K., Peltola, M. J., Kauppinen, P. K., Haapalainen, A., & Leppänen, J. M. (2008). Facing a real person: an event-related potential study. Neuroreport, 19(4), 497-501.

Pönkänen, L. M., Alhoniemi, A., Leppänen, J. M., & Hietanen, J. K. (2010). Does it make a difference if I have an eye contact with you or with your picture? An ERP study. Social cognitive and affective neuroscience

And there are a few studies to see whether people can match real faces to photos (e.g. when checking passports or photo ID), for example:

Kemp, R., Towell, N., & Pike, G. (1997). When seeing should not be believing: Photographs, credit cards and fraud. Applied Cognitive Psychology, 11(3), 211-222.

Science, objects and subjects


Like many of us, I watched with keen attention as the technical crew who run SpaceX pulled off one of the greatest stunts in rocket science ever.  For the first time, they managed a controlled landing of the discarded stage 1 from a rocket, bringing it down vertically on a drone ship with near perfect accuracy.  I got a lump in my throat, my eyes teared up, it was fantastic.

I suspect many in the crowd had a similar experience.  The event was broadcast live with a very sophisticated webcast, full of happy presenters who were keen, so very keen, on the science.  I might gently point out that a great deal of what they were enthusing about was technological and engineering in nature, and not really science, though for sure, science was a motivator.

But pay attention to what happens to the crowd at 1’15” in the video excerpt here. (If copyright considerations break the link, it is at 28’37” of the original video here).  There is a great deal of cheering, and clapping, but then they break out into a chant of “U.S.A.”. Right in the middle of this celebration of science in an objective key, the collective subject asserts itself, by literally jumping up and down and chanting.

The project is, of course, motivated to a great extent by the fact that the American space programme is in an odd state of limbo, as the shuttle no longer flies, and the only actors capable of sending people to the ISS are China and Russia. There are interests afoot.

Now, do we want a science that is unable to address what we witness in this video?  Not just the rocket landing bit (and, again, hurrah for that!), but the jumping up and down and synchronised chanting bit?  Yes, we can approach this in a scientific manner, but only if we do not recoil from recognising that there are, indeed, subjects here.

State of science 2: The impossibility of an objective science of behaviour

The behaviourist turn within psychology at the start of the 20th Century was an attempt to be rigorous, objective, empirical, and seriously scientific.  To graduates of more recent psychology programmes steeped in talk of information processing, cognitive systems, and computation, behaviourism is liable to occasion spittle-flecked revulsion, as if the scientific study of behaviour were a gateway drug, leading inevitably to the extinction of the light of the soul mind.

All science starts with observation. If the scientific study of behaviour does not take the observation of behaviour as absolutely foundational, then it has no ground under its feet, no chance of garnering consensus, no necessary connection to reality, and no authority.

But in the observation of behaviour, we as observers necessarily become entangled in the object of our attention.  From one second to the next, the universe is changing, flowing, becoming.  This flux is not partitioned into discrete behaviours.  Behaviours are not like rocks, lying around to be found, kicked, and counted.  Rather, to parse a specific kind of going on as one behaviour or another is to frame one’s observation as the goal-directed striving of one system or another.  And in so doing, our imposition of this teleological framing necessarily results in the entanglement of the observer and the observed.  Goals are not observable.  They are rather a projection of the observer, in order to make one’s observations intelligible.

So the study of behaviour cannot be done in an objective key.  I think Tim Ingold put this rather well, when he says:

Yet all science depends on observation, and all observation depends on participation — that is, on a close coupling, in perception and action, between the observer and those aspects of the world that are the focus of attention.  If science is to be a coherent knowledge practice, it must be rebuilt on the foundation of openness rather than closure, engagement rather than detachment. (Ingold, T., 2011, p. 75)


[1] Ingold, T. (2011). Being alive: Essays on movement, knowledge and description. Taylor & Francis.

State of Science 1: Putting objectivity in its place

In my consideration of science and its place, it seems unfortunate, but necessary, to start with the notion of objectivity, its relation to the terms “fact” and “truth”, and the place that this kind of knowledge has in our overall intellectual and consensual economy. If we fail to adopt at least remotely similar relations with respect to this mode of knowing, then our subsequent discussions will surely be pointless. Read the rest of this entry »

The State of Science

Over the next while I will indulge myself with a few posts that bring together several strands of thought about how we think of science.  These posts will be difficult to get right. They will aspire to be sufficiently disinterested to be able to see science as a belief system, comparable in many respect to other belief systems.  That will annoy some people, because there is a great deal of aversion and intolerance for anything smelling of belief, or worse religion, in many quarters.

But this line of thought might also permit us to reconsider what a belief system is, and to recognise that any expression of knowledge, or assertion of matters of fact, must rest upon some foundation.  More challenging yet, I will probably argue that the ground upon which any one of us builds our world view is invisible to that individual.  This will introduce an anti-positivist stance, leading not to statements about how things are (no truths here), but to solicitations to negotiate, so that we may end up with the best consensual accounts possible.

There are several threads I wish to pursue, and they will interact, so that for a while, I might edit and revise any of these posts.  Perhaps they will settle to a stable equilibrium.  Perhaps they will remain in flux and fall apart.

Here are some of the topics I intend to pursue.  As with any part of these posts, the list may be retro-fitted to make it appear that my thoughts are better organised than they in fact are.

  • The notion of objectivity and its relation to scientific knowledge and practice
  • The impossibility of an objective science of behaviour
  • Consideration of the role of context
  • The basis of measurement and its relation to observation
  • The complexities of the notion of representation
  • The relation of major funding initiatives to the practice of science
  • The role of expert judgement and its similarity to divination
  • Recognising useful empirical loci to support thick description
  • The impossibility of symmetrical anthropology
  • Time, objective, lived, imaginary, illusory, real
  • Perspectives and more perspectives

The skill of writing

Writing is a skill. Skills need practice. Usually, we write for specific purposes: An essay required by a course, an article for a journal, a conference submission, the Great Irish Novel, a blog post, or a tweet. Not all require the same degree of commitment or polish, but all have in common that they are directed at others. We write to be read, just as one might learn piano in order to be heard.

But if a pianist only played when others were listening, our expectations for the resulting performance would not be high. Pianists sound good because they play a lot, even, and especially, when nobody is listening. They learn to hear themselves critically.

So find a way to write a whole lot. Write little bits. Write unfinished prose. Write in different keys. Write for those who know your ideas and those who don’t. Write as an end in itself. Write, and throw the result away without a second thought.

Students of painting often spend time copying great works, sitting in galleries making sketches based on painted originals. This can seem like a pointless exercise. The sketches are not themselves offered as works of art. But they serve a purpose. By being relieved of the need to decide upon a composition and subject, the act of sketching allows the student to experience what it feels like to carry out that line, to bring this surface and that detail into contact. In sketching, they have to pay attention to detail that a casual viewer would never notice.

So practice writing by copying great writing. I mean this literally. If you are reading and you have the good fortune to come across an exquisitely expressed idea, a passage of prose that hits you where it counts, then open a window or grab a notebook and write the passage yourself. You don’t have to dream up the words, but you will learn what it feels like to put the words in sequence. You will attend to the pacing and the structure in a new way, and you will learn to love the writing even more. No cutting and pasting! You have to run the words through your fingers!

Really Rotten Science 1: Dunbar’s Number

The field of social psychology is controversial at its core. The findings of this field are highly likely to treated as normative in the contested discussions we have about our social lives.  Among the more problematic findings asserted is that which goes by the name of Dunbar’s Number, and is attributable to Robin Dunbar, of Oxford University.

At the heart of the issue is a misguided correlation that appeared in a 1992 article that has since garnered over 1,200 citations: Dunbar, R. I. (1992). Neocortex size as a constraint on group size in primates. Journal of Human Evolution, 22(6), 469-493.

The correlation is between two variables. In the X-corner we have “neocortex volume” as measured in 38 primate genera (not including humans).  This is explicitly justified as an index of “cognitive capacity”, which is not a term with any widely agreed meaning. In the Y-corner, we have “group size” based on ethological observation.  This yields a scatter plot, and when the data are log transformed, a fairly strong correlation between the two variables is found.  Then, in a manner that I argue is entirely indefensible, humans are added to the plot based on their cortical volume, and we can use the linear fit to observe a calculated group size for humans of about 150.  This is Dunbar’s Number. Although no normative conclusions are drawn in the original study suggesting that humans “ought” to live in groups of about 150, much of the subsequent discussion of this has been tinged with the romantic idea that there is a “natural” or “correct” group size for humans to live in, or to maintain through their social interactions.

Let’s consider this work in detail.  The X-variable comes first.  Neocortical volume is taken as an index of cognitive capacity.  There is no scientific basis for this choice, as cognitive capacity is not a well defined kind of thing.  This variable is chosen in advance, and is used as a lens through which to view other animals. If you were to choose one anatomical characteristic of homo sapiens that really stands out, it is our neocortical volume, which is grossly inflated by the size of our frontal lobes.  Many animals have something special about them.  Giraffes have long necks, cheetahs are really fast runners, and humans have very big forebrains. Figure 7 of the original paper makes this very clear: humans are not typical when viewed through the lens of neocortex volume. Imagine, if you will, some giraffe scientists deciding that neck length was an appropriate variable with which to study other animals, and would measure neck length in many non-giraffe animals, and then add giraffes to the plot.  Lo and behold, giraffes would be spectacular. This is not an innocent variable. For a wonderful critique of this kind of anthropocentric wishful thinking, I would point to Louise Barrett’s recent article “Why brains are not computers, why behaviourism is not Satanism, and why dolphins are not aquatic apes”.  ( The Behavior Analyst, 1-15., 2015), ( or see the poem by Ambrose Bierce at the foot of this page).

Now let’s look at the second variable, group size.  This is motivated by a view of social interaction in which each member of a group is involved in information processing to keep track of the others.  This is a venerable, though highly contested, position one can take with respect to social interactions.  However, group size is really complex, and will be influenced by many factors, and not in a simple way.  Orang Utans give pause for thought here.  They are excluded from the analysis, justified due to the unavailability of data on neocortex size.  These close relatives live relatively isolated lives some of the time, come together some of the time, and generally make it difficult to claim that they have one group size or another.  Discussing this, Dunbar considers the possibility that they are “socially degenerate”, and then does some hand waving to suggest that they might have unseen group sizes of 6 to 15 individuals (despite explicitly noting that most studies have concluded that there is no simple group size for the species).

Another example of carving the data to fit the hypothesis is clear in discussion of Miopithecus, where prior work asserted that group sizes were larger in populations living commensally with humans, and so “therefore only groups listed by Gautier-Hion (1971) as living in undisturbed habitats were used when calculating the mean in this case”.  What just happened?  Data was sliced and diced in order to shore up a romantic notion of nature as separate from man, and only undisturbed habitats are considered informative.  This presupposes that humans are not part of nature, and that animals become corrupt in our presence.  This is entirely anthropocentric, and is beholden to the problematic assumption that humans lie outside the order of nature, and not within it.

But then things go very wrong indeed.  In a 1993 paper (Dunbar, R. I. (1993). Coevolution of neocortical size, group size and language in humans. Behavioral and brain sciences, 16(04), 681-694. cited 1,600+ times), Dunbar takes this seriously compromised correlation and applies it to humans, based on their neocortical volume, allowing him to produce Dunbar’s Number, the number of humans that we have (do, should, used to have?) in our social groups.  Now here I have a slight problem.  Using Dunbar’s data from the 1991 paper, I get a different regression equation than him.  The number he comes up with is 147, where I get a paltry 100.  But it is not worth chasing down the difference because this kind of extrapolation is absolutely not justified.  It is not allowed.  It breaks the rules of the game.  Humans, as we and Dunbar have noted, are off the scale when it comes to our big forebrains. In the 1993 paper, he notes that the value for humans is 30% larger than that for any other primate.

Here is a basic rule in correlational studies: correlations are based on the tentative assumption that the relationship between the variables is linear.  This is generally not true for any naturally occurring variables, but may be true within restricted ranges.  For example, there is, for any spring, a range of forces we can apply where we will observe roughly linear increase in length with increase in force.  But outside that range, things are not clear.  Springs will break.  Complex relationships such as the one dreamed up here are guaranteed not to have a simple linear form.  So, all other things being equal, if we establish a correlation over the range [X_low . . . X_high], we might argue that we can use this linear relationship to make predictions for unseen cases where the predictor lies within the range which we used to establish the correlation.  We have absolutely no license to extrapolate beyond that range.  We cannot make predictions based on a linear relationship except within the range where that relationship holds.  This observation is entirely distinct from the other objections raised above.

Dunbar has this to say: “Strictly speaking, of course, extrapolation from regression equations beyond the range of the X-variable values on which they are based is frowned on. We can justify doing so in this case, however, on the grounds that our concern at this stage is exploratory rather than explanatory.”  This is simply wrong. You cannot so justify things like that.  The human data point is way, way off the scale.


Click to embiggen

In these plots, the left hand column shows the raw data (group size, ratio of neocortex to rest-of-brain).  On the right hand side are the log transformed values (the log transform is needed to make the relationship approximately linear).  The lower two panels show how the plots look when we add in the single data point for humans – that’s that black square in the top right corner.  The human point is not justified here because of “exploration”.  This is not science.  It should not be discussed as science.

The 1993 paper contains discussion of “group size” in humans.  What could this mean?  Dunbar leans on some romantic notion of primitive humans living in their close-to-nature hunter gatherer societies, which he cherry picks and opines he can discern a group size for humans (unadulterated by modernity).  He tries to draw in observations about the size of military units in professional armies into the story.  All the unmotivated, hand-picked examples he chooses happen to support his hypothesis that there is a right/true/natural size to human groups, given by Dunbar’s Number. There is wishful thinking a plenty here, but no science.

The 1993 paper has been followed by thousands of references, citations, and extrapolations from this nonsensical foundation.  Dunbar’s number has gone down in the scientific literature as if something had been found out.

But who cares?  A social psychologist (sorry, evolutionary social psychologist) decides to fancifully hallucinate a relationship between brain size and  group size. Shouldn’t we just walk away and tut-tut about evolutionary just-so stories?

Emphatically no!  This is normative science, and it is sold as science, which thereby carries authority.  We use science to understand norms and deviations from the norms, to inform public policy, develop educational curricula, and to distinguish between the healthy and the pathological.  The quality of science is a political and ethical matter, and when it comes to science about ourselves (whatever that means) such concerns must be centre stage. This specious rubbish is in that literature, polluting our view of ourselves.  Dunbar’s Number is nonsense, and ought to be eradicated from any science worth having. Would that it were alone.

Normative Social Psychology?  Just say NO!

Postscript: Ambrose Bierce penned the following poem in his Devil’s Dictionary of 1922: