Calling a spade a spade

In his comment on my previous post, Steve Flatt raised some interesting objections to my criticism of the way some philosophers construe brain function, citing Rom Harré and Daniel Dennett in support. I can’t comment on Harré’s and Dennett’s specific points because I haven’t read either of the texts Steve recommends, but I think I the objections Steve raises can be resolved by reframing them. I’m responding in a new post, because I need to explain the concepts involved.

Deconstructing a spade

Steve begins by pointing out that “Deconstruction is not necessarily the right thing to determine the purpose or “aboutness” of the brain”. He cites a colleague who used the analogy of a spade, arguing that if you were to systematically deconstruct a spade down to its component sub-atomic particles, “the further into the deconstruction we got the less apparent the purpose of the spade became”. That’s certainly true; but if you’re interested, not in the purpose of the spade, but how one could be made, deconstructing the spade becomes crucially important.

Human beings have not always been able to make spades as we know them today. Until we had tools that would cut and shape wood, and had figured out how to extract iron ore from rock, how to make it malleable and had the tools to shape it, even if we had wanted to dig holes in the ground we wouldn’t have been able to do it with a spade made of a steel blade and wooden handle. The first concept that I think would be useful in reframing deconstruction is the idea of levels of abstraction.

Levels of abstraction

Fields of knowledge that deal with complex systems are usually quite at home viewing a particular phenomenon as occurring simultaneously at different levels of abstraction; biologists, for example, can comfortably move between the molecular, cellular, tissue or whole organism levels when tracking a particular physiological pathway. What happens to molecules affects cells, what happens in cells affects tissues, and what happens in tissues affects the whole organism. If we apply this concept to the example of a spade, we find that a spade too can be construed at many different levels; the sub-atomic, atomic, molecular, compound, materials, parts, object and function levels. Each level of abstraction is dependent on the configuration of the level beneath it, so each level of abstraction is ultimately dependent on the configuration of the sub-atomic particles (for want of a better term) that make up the spade. A spade can be deconstructed as a series of levels of abstraction that have the following characteristics:

function level: digging (and multiple other functions)
object level: spade
parts level: blade, handle
materials level: steel, wood
compound level: steel, cellulose, lignin
molecular level: iron, carbon, glucose, monolignols
atomic level: Iron, carbon, hydrogen, oxygen
sub-atomic level: protons, electrons etc.

I agree with Steve that “Deconstruction is not necessarily the right thing to determine the purpose or “aboutness” of the brain”. But he then goes on to say “To assume that by taking it to pieces we will learn its functions seems highly unlikely to me.” In response I’d point out that most people studying the parts of brains aren’t doing that to determine its purpose, but to find out how it works, not the same thing at all.

I think levels of abstraction are what Harré is getting at when he refers to molecular, organism, person and soul ‘grammars’. Different levels of abstraction certainly have patterns of functioning that you could call ‘grammars’ if you wanted to. What concerns me is that Harré, like Wittgenstein and Ryle, appears to be making language his starting point for their analysis. To me this is starting from the wrong place. In terms of the levels of abstraction involved in human functioning, language is at a very high-level, and the patterns it forms will be influenced by what’s going on at the levels beneath it. It’s certainly true that language shapes the way we construe those lower levels, but it doesn’t shape the way they work, so applying patterns from language to other levels of abstraction isn’t necessarily helpful. Different levels of abstraction work in different ways, which where another concept comes in handy – constraints and affordances.

Constraints and affordances

Each level of abstraction has different constraints (things that can’t be done) and affordances (things that can be done). Random collections of sub-atomic particles can do things that atoms can’t, but atoms can do things that individual sub-atomic particles can’t. If you were to melt the blade of the spade, the puddle of molten steel that resulted would have constraints that the blade didn’t – you wouldn’t be able to dig with the molten steel or use it to build a sandcastle; but it would also have affordances that a blade didn’t – you’d be able to pour it into a cast or use it to start a fire. This brings us to a third useful concept, emergent properties.

Emergent properties

Different constraints and affordances emerge at each different level of abstraction, so each level of abstraction has different emergent properties. Individual atoms of iron, carbon, oxygen and hydrogen have different properties to a chunk of steel and a wooden pole, which in turn have different properties to a spade; even though the spade, chunk of steel and wooden pole are all made of atoms of iron, carbon, hydrogen and oxygen. The idea of emergent properties maps neatly on to Ryle’s concept of ‘disposition’ and Bowlby’s description of feelings as ‘phases’ of physiological processes. It also allows us to formulate answers to Dennett’s questions. I don’t know the context in which these questions were asked, but there is a response to the questions as they stand.

Dennett’s questions

First question; “Please tell me, Dr Physicist, just what colour is. Are there colours according to your theory?

‘Colour’ is the label we give to an emergent property of the interaction between our visual processing apparatus and a particular wavelength of light. Colour doesn’t exist in the same way as our visual processing apparatus or light exist, because it isn’t the same sort of thing as they are. Dennett’s questions are essentially meaningless (which I suspect is what he intended them to be) because colour can’t be reduced to sub-atomic particles, but without sub-atomic particles colour couldn’t be an emergent property of anything.

Then there’s; “Dr Chemist can you provide the chemical formula for a bargain?

Similarly, the question about the chemical formula for a bargain is meaningless because a bargain is an emergent property of a particular set of human interactions. It can’t be reduced to a chemical formula, but if chemical reactions weren’t involved, no bargains could ever be struck because there would be no one to strike them.

The third question Steve quoted; “Could you ever frame a clear concept of a bargain, or a mistake, or a promise…?

begs another question, which is; what do you mean by a ‘clear’ concept? Enormous amounts of time have been wasted by people trying to frame clear concepts of things that are far from clear and are never going to be. One clear concept that I think can resolve the problem is the idea of a fuzzy set.

Fuzzy sets

Category theory, derived from set theory, organizes entities in ways that derive from the features of the entities, rather than by forcing them into categories we prepared earlier. A fuzzy set is one whose elements have degrees of membership. Supporters of a political party would form a fuzzy set, because some people would be in total agreement with the party’s policies whereas others would agree to varying extents. (Members of a political party, by contrast, would form a crisp set, because someone is either a member or they’re not.)

A bargain (or a mistake or a promise) is a construct that applies to certain forms of human behavior. The category of things we call {bargain} has core (prototypical) features – that would include more than one party and an agreement – but its member bargains vary widely in their form and content. Because of those variations, it’s pointless trying to frame the concept ‘bargain’ clearly. It’s much more appropriate to frame it as a fuzzy category, with core distinguishing (prototypical) features but with degrees of membership and blurred boundaries.

Even though bargains (or mistakes or promises) might in general be fuzzy constructs it doesn’t follow that any particular bargain (or mistake or promise) is fuzzy in terms of what’s involved in specific cases. If I agree to pay my son £5 if he washes the car, some very concrete, clearly definable things are involved. The general concept ‘bargain’ isn’t the same sort of thing as; my son, a £5 note, water, shampoo or the car. A bargain is an emergent property of a particular configuration of those things, plus many others at lower levels of abstraction that it would take too long to list.


Steve refers to epigenetics as a problem, but I’m not clear why. The discovery of epigenetic effects showed that the interaction between genetic expression and the environment is even more complex than was previously thought, but that doesn’t make genetic-environmental interaction intractably complex. Because of their emergent constraints and affordances, entities at each level of abstraction are bounded; they operate within limits and behave according to patterns determined by those constraints and affordances. So epigenetic factors don’t mean that human behavior and adaptability are infinitely variable; patterns will emerge from the variability.


Steve also says “But the tool we can most effectively use to explore [the brain] is language and behaviour. While we may map these activities on to the brain in some form the sheer flexibility of processing means we can never be deterministic about what parts of the brain do what particular activity.”

I think Steve’s making two assumptions here. First, using language and behavior to explore the brain might be most useful to him in his work as a psychologist. But they might not be so useful to an occupational therapist working with a child with mobility problems or a brain surgeon trying to remove a tumour. Those people might find knowing how the brain’s primary motor cortex functions quite useful.

I think the second assumption is that looking at lower levels of abstraction to try and figure out how the brain works is deterministic. I don’t think it is. It’s just figuring out how the brain works at lower levels of abstraction. Higher levels of abstraction, such as language and behavior are to an extent determined by what happens at lower levels, but their constraints, affordances and emergent properties limit what goes on within different levels, rather than determine what goes on within them.


Harré, R & Moghaddam, FM (2012)’Psychoneurology:The Program’ in Psychology for the Third Millennium: Integrating Cultural and Neuroscience Perspectives,SAGE.