Greg Detre
Friday, 11 January, 2002
Is a naturalistic account of reason
compatible with its objectivity?
�How
is it that contingent, biological creatures such as ourselves can have access
to such universally valid methods of objective thought?� He wants to deny the
subjectivist who thinks that �the first person, singular or plural, is hiding
at the bottom of everything we say or think�, or indeed the relativist who
wants to deny that there is any objective reason to prefer some beliefs over
others.
Reason
then, for Nagel, encompasses any means at our disposal of accessing,
recognising and producing knowledge that is objective.
Reason
as providing authority and a means of persuading others of the relative merits
and demerits of our beliefs
The
existence of mind, with humanity as just one example of presumably countless
possible/actual rational species in the universe, �is certainly a datum for the construction of any world
picture: at the very least its possibility
must be explained�. Despite sounding quasi-religious, the idea �that the
capacity of the universe to generate organisms with minds capable of
understanding the universe is itself somehow a fundamental feature of the
universe�
He
thinks that �the hypothesis of some systematic aspect of the natural order that
would make the appearance of minds in harmony with the universe something to be
expected� more explanatory than the religious proposal, which seems to employ
God more like a placeholder for explanation than anything else. He suspects
that the fear of religion may include even such a �cosmic order of which mind
is an irreducible and non-accidental part�. Of course, even such a
�mind-friendly cosmology� (i.e. natural �laws that explain the possibility of intelligent life�) will
need the right initial conditions.
�I
have to be able to believe that the evolutionary explanation is consistent with
the proposition that I follow the rules of logic because they are correct � not
merely because I am biologically
programmed to do so�. �The only form that genuine reasoning can take consists
in seeing the validity of the arguments, in virtue of what they say.�
�the
theory of evolution as usually understood provides absolutely no support for this conception of ourselves
[�as rational � and also as creatures who have been produced through Darwinian
evolution�], and to some extent it renders the conception suspect.�
He
thinks that an argument to the contrary would require:
1.
a
�general analysis� of rationality �into a limited set of functional elements�
2.
considering
�the relation between this set of capacities and the simpler habits of mind
that might plausibly have carried selective advantage in the period when the
human brain evolved�
Nozick�s
evolutionary hypothesis (naturalised epistemology) explains both limitations
and successes of reason, proposing � a reversal of the Kantian dependence of
the facts on reason�. Reason �is the dependent variable, shaped by the facts�,
with reality �selecting for what seems �evident�� (tNoR pg 112). �Facts and reality are what they are independent of
what we think�, and �our finding something self-evident is no guarantee that it
is necessarily true, or true at all � since the disposition to find it
self-evident could have been an evolutionary adaptation to its being only
approximately, and contingently, true�.
�The evolutionary explanation itself is something we
arrive at, in part, by the use of reason to support evolutionary theory in
general and also this particular application of it. Hence it does not provide a
reason-independent justification of reason, and, although it grounds reason in
facts independent of reason, this grounding is not accepted by us independently
of our reason. Hence the account is not part of first philosophy; it is part of
our current ongoing scientific view.� (tNoR
pg 112)
Could a
connectionist system (even one as complex as the brain) ever be truly rational???
In two ways, this is a stupid question. On the one hand, how can anyone
know? � our current NN efforts are so feeble in comparison to human
rationality. On the other, humans appear rational (questionably), and we have
connectionist brains, so we must be. Well, Nagel for one is prepared to argue
that our current conception of mind almost certainly needs to undergo at least
one paradigm shift before we can make sense of problems like the mind-body
problem and how we can have access to such �universally valid methods of
objective thought�.
Won�t there always be a probabilistic aspect to its computation that
would make it fallible or non-rational to some extent, i.e. rational 99.9% of
the time???
An inherent part of true rationality for Nagel is its generality:
Our aim as thinkers and rational agents is to arrive
at principles that are �universal and exceptionless� � to be able to come up
with reasons that apply in all relevantly similar situations, and to have
reasons of similar generality that tell us when situations are relevantly
similar.
Can a connectionist system ever be generally rational, since its training data will always be limited, and so its synaptic organisation will be geared towards that
depends how representative the training data is
philosophers, for example, have better training data
Is rationality
adaptive??? it seems clear that having true beliefs may well be more expensive
and less fitness-enhancing than having useful beliefs, and so much less likely
to evolve.
Following on from
this, Robert Nozick (developing from Cosmides & Tooby and others) has an
interesting idea that we have evolved to find certain chains of inference
automatic and self-evident, i.e. that there may be hard-wired, specialised
inferential mechanisms for common past situations that have been selected for.
Thus, for example, the list of philosophical problems we've been least
successful with all mark assumptions that evolution has built into us: the
problem of induction, of other minds, of the external world, of justifying rationality
etc. These seem to me to be just the sort of genetically pre-wired neural
representations that are argued against in Rethinking Innateness.
Fodor � shallow processing �
Modularity of Mind
The idea that the brain is implementing formal
logic in some hidden way isn�t very popular now, but philosophers seem to
favour the idea that certain, fairly specific ideas could be genetically coded.
You argue against that in Rethinking Innateness, but is it possible that a tiny
proportion of the genome does hard-code a handful of vital neural
representations???
interactionism
could give rise to such small/tight representations
Nakisa
+ Plunkett in McLeod et al. on evolution, pg 308
given your faith in the power of nature�s systems resulting from developing/interacting systems and representations:
why don�t you use multi-module NNs???
why don�t you use genetic algorithms???
why don�t your simluations have heterogenous clusters of neurons???
how do you provide input of unknown/varying size??? realtime input???
apart from the obvious case of sound localisation, do you not feel that you lose information by using a rate rather than spiking net???
what about biological plausibility???
once you�ve trained up a NN using perhaps a biologically implausible learning rule, can you then use Hebbian learning for fine-tuning???
you�ve made huge progress with NNs learning syntax � to go any further though, won�t they need to be perceptually and behaviourally embodied???
if I have gaps in my training data, is that a problem???
if I�m looking for categorisation of high-dimensional data, perhaps into a pre-determined number of categories, perhaps with incomplete data, do I want to use an SOM or an LVQ???
new domains of
training data
create
new spaces in which generalisations are made
re-organise
or recode earlier spaces
how do we create those spaces?
always told our NNs what the task is, somehow � this is a hard problem
�