Minds and Computers : An Introduction to the Philosophy of Artificial Intelligence

question against the Australian materialist. ‘Begs the question’ is a
phrase which is more and more commonly used to mean something
along the lines of ‘in light of which the question demands to be
asked’. This is not the technical philosophical use of ‘begs the ques-
tion’. To beg the question is to run afoul of the fallacy of petitio prin-
cipii. One begs the question against one’s interlocutor when one asks
to be granted the very proposition in dispute.
In this case, the proposition in dispute is that mental states are
type-identical to neural states. The objection from introspection only
cuts any philosophical ice on the assumption that mental states are
not identical to neural states. After all, if the Australian materialist is
correct and this type identity holds, we do in fact learn something
about our neural states through introspection, simply by virtue of
learning about our mental states (given these are held to be identical).
The objection from introspection therefore simply begs the question
and is no real objection. It may be surprising to learn that we do actu-
ally introspect our neural states but scientific discovery is frequently
surprising in the light of antecedent folk theories.
There are a number of further objections to Australian material-
ism which also beg the question. Several of these come in the form of
appeals to Leibniz’ Law – the objection from introspection is actually
an instance of just this form.
Leibniz’ Law – otherwise known as the identity of indiscernibles –
posits that if two things have all and only the same properties, then
they are identical. The objection from introspection seeks to deploy
this in arguing that since mental states have a property which neural
states do not, they must be not identical. Similar objections can be
mounted by appealing to various other properties the objector holds
mental states to have and neural states to lack, or vice versa.
For instance, neural states have a specific spatio-temporal location.
It seems odd, however, to suppose that my mental state of thinking
about ice cream is located three inches behind my right eye.
Alternatively, we might play on the semantic properties of mental
states. My mental state of believing that today is Saturday has seman-
tic content – it means something. By virtue of its semantic content it
is apt to be involved in implication relations – for instance, if I believe
today is Saturday then I believe tomorrow is Sunday. Neural states,
however, neither have semantic contents, nor are they apt to be
involved in implication relations.
 
39

---

There are a number of other ways we could problematise the
appeals to Leibniz’ Law here; however, the most decisive reply is to
simply note that, once again, these objections straightforwardly beg
the question against the Australian materialist. If mental states and
neural states are, in fact, type-identical, then mental states do have a
specific spatio-temporal location – surprising as this may be – and
neural states do in fact have semantic contents such that they are apt
to be involved in implication relations, surprising as this may be. To
simply assert that this is not the case is just begging the question.
There is, however, a rather decisive objection to Australian mater-
ialism. This is the objection from multiple realisability.
It turns out not to be terribly di
fficult to cast serious philosophical
concerns over the claim that types of mental states are identical to
types of neural states. For one thing, there is the question of how, pre-
cisely, we are supposed to construe the concept of ‘type’. If we con-
strue it too narrowly then we are committed to saying that whenever
a group of us all desire ice cream, we are each in exactly the same
neural state. This is clearly implausible and the rubric of ‘type’ is sup-
posed to allow for some variation in neural states in order to accom-
modate this. How much variation is the crucial question. If we
construe the notion of ‘type’ too broadly, then we are at risk of losing
the empirical methodological advantage which the claim of type-
identity confers.
The explanatory burden here on the Australian materialist is to
give some account of what, precisely, must be shared by neural states
in order for them to qualify as being of the same type. Unfortunately
for the Australian materialist, even if this explanatory burden can be
met, there is a further objection to the type-identity claim which is
unanswerable.
Consider the case of someone who su
ffers neural damage, whether
it be through a stroke or through some trauma such as a motor vehicle
accident. After the damage, the patient typically loses the ability to
have certain mental states. They might, for instance, no longer be able
to recognise their spouse, or they might no longer be able to under-
stand certain words. The fact of crucial importance here is that such
patients very frequently regain many of their lost mental faculties –
they relearn to recognise their spouse or to understand the concepts
they had lost. They recover the capacity to have these mental states