Dennett Indented?
A commentary on Daniel Dennett's "Skinner Skinned".
As humans, we often explain each other’s behaviours by alluding
to mental states such as intentions, beliefs, and desires. These are
all non-observable; they might be inferred from watching other people’s
behaviour and making analogies with our own, private mental world. Partly
because of this unobservability, Skinner held that to use these notions
in explanations of behaviour is un-scientific, that we are basically
explaining actions by referring to a homunculus, a sort of inner prime
mover, i.e., a miracle, which of course is no explanation at all. Skinner
proposed a behavioural psychology in which one would only study the
environmental inputs to the human, and the behavioural outputs they
cause. The environment would reinforce us, he said, to continue to act
in such ways that we are rewarded for our actions; thus increasing the
probability that we will act in such ways again. He called this operant
conditioning, and exclaimed that almost all of human behaviour can be
reduced to such simple mechanisms of reinforcement – the exceptions
being reflexes and instincts.
If (almost) every human action can be reduced in such a way, then according
to Skinner, there is no room for intentionality in a psychology of human
behaviour; if we can explain human actions in terms of causes, then
we need not invoke reasons in our explanations. Opposing Skinner, Dennett
thinks intentional and mechanistic explanations do not necessarily exclude
each other; although sometimes they do, and in “Skinner Skinned”
he gives an example in which this is the case: The wasp Sphex exhibits
behaviour around egg laying-time that seems intentional, but on closer
inspection, it is clear that the wasp is only following some rigid rules.
We were attributing too much to the wasp, Dennett says. The real explanation
is not that the wasp intends to do what it does, but that it is genetically
programmed to do it, it cannot escape following the algorithm that is
laid down in its genes. It is not even irrational; it is non-rational,
and hence, not at all intelligent. Stepping up the phylogenetic ladder,
Skinner performed most of his experiments on still quite “low”
animals, like pigeons. His explanations for the pigeons’ learned
behaviour are simple, Dennett points out, but what is actually happening
is not simple, if viewed from a neurophysiological point of view. Dennett
holds that Skinner is not explaining, but explaining away. By explaining
the pigeons’ behaviour in terms of operant conditioning, Skinner
is effectively treating it as a wasp, explaining away its intelligence.
This might not be very objectionable, but when Skinner wants to explain
human behaviour in the same way, we might feel he is on the wrong way.
We are intelligent, and we do not want that intelligence to be taken
from us by an argument about why we behave as we do.
But wait a minute. In Dennett’s example of the wasp Sphex, we
read more into its behaviour than was called for. The explanation was
actually simpler than the one we reached before looking more closely
at the situation. Note that this is not a part of the domain Skinner
wants to explain. Even though Skinner’s interest is in behaviour,
his research program involves learned behaviour, and nothing else. Instincts
and reflexes are the exceptions to what can be explained by operant
conditioning. Skinner wants to predict an animal’s behaviour based
on what it has learned about its environment, that is, upon how it has
been conditioned. This has nothing to do with the behaviour of the wasp
Sphex. In the case of the wasp, there was no learning and no intelligence
to explain, or explain away, in the first place. If Skinner, as Dennet
seems to be saying, was trying to explain the pigeon’s behaviour
in the same way, then he could not be trying to explain learning (in
the form of conditioning), because there was none. This is obviously
false; the pigeon’s learning is contingent upon the reactions
of the environment, and even though Skinner’s focus is not upon
the learning or the mechanisms of it, but rather upon the behavioural
results of it; there is learning involved in every behaviour Skinner
wants to explain or predict. So, Dennett wants to turn the fact that
we read too much into the behaviour of the wasp Sphex into the complaint
that Skinner reads too little into the behaviour of his pigeons. It
seems there is a dent in Dennett’s argument. So what is his point,
anyway?
Well, it might be stated fairly simply: Sometimes, simple explanations
really unmask the truth. Where there was a complex explanation before,
one that clouded our view to reality, there now is a much simpler explanation
that shows us how the world really is. This is Dennett’s non-rational
wasp. But other times, reality is actually complex, and by explaining
it in simple terms we are not explaining it at all, but explaining it
away, overlooking important points. This seems to be Dennett’s
view on Skinnerian behaviourism.
To sum up, we now have three different ways of explaining behaviour.
We can (theoretically) look at the physiological changes in the pigeon’s
(or whatever’s) brain and see what are the causes and effects
of its training, explaining how neurons strengthen and weaken their
relations, and so on. Or we could explain it on a higher level, saying
that the pigeon intends to get food (or whatever), and believes he knows
how to get it, therefore he does what he does. Or one might take Skinner’s
approach, and simply say the pigeon has been reinforced in its food-gathering
behaviour. The way I read Dennett, he does not necessarily say that
Skinner is absolutely wrong in his way of explaining behaviour, quite
the contrary; Dennett too wants to define a mechanistic psychology.
Neither does he say that that intentionality is indispensable. He is
just stating that intentionality is a useful concept when trying to
understand why people behave the way they do. All three ways of explaining
behaviour might be fruitful. Yet, he points out that behaviourism often
has little, if any, predictive value if we do not know the training-history
of the animal we are looking at. Hence, Skinner’s behaviourism
is not often a very useful concept, and we should look for explanations
of human behaviour elsewhere.
Oh, and by the way, in the penultimate paragraph, we un-indented Dennett
.