Re: Turing Machines and Physical Computation

From: David Longley (David_at_longley.demon.co.uk)
Date: 11/25/04 

Date: Thu, 25 Nov 2004 21:00:31 +0000

The following are taken from a series of messages posted internally
within a closed network back in 1990. I thought they might be worth
citing once again (although without their original context for reasons
which some of the more thoughtful here will understand). A lot has been
said here since. Most folk here won't see a fraction of what it's all
about for many reasons that I've already covered. Some might be wise to
take their emotional responses as something to further act upon.

ON THE COMPUTER "METAPHOR"

'It has always bothered me that models of psychological
processing were thought to be inspired by our understanding of
the computer. The statement has always been false. Indeed, the
architecture of the modern digital computer - the so-called Von
Neumann architecture - was heavily influenced by people's (naive)
view of how the mind operated. Perhaps I had better document
this. Simply read the work on cybernetics and thought in the
1940's and 1950's prior to the development of the digital
computer. The group of workers included people from all
disciplines: See the Macy Conferences on Cybernetics, or "Her
Majesty's Conference on Thought processes". Read the preface to
Wiener's book on cybernetics. Everyone who was working together -
engineers, physicists, mathematicians, psychologists,
neuroscientists (not yet named) - consciously and deliberately
claimed to be modelling brain processes.'

Reflections on Cognition and Parallel Distributed Processing

D.A. Norman
(Ch 26, p534, Parallel Distributed Processing Volume 2)
McClelland J and Rumelhart D 1986

    'A trait is EFFECTIVE if there is a hard and fast
    routine by which we can check for it, without guesswork
    or imagination...It came to be appreciated, in the mid-
    thirties, that recursiveness affords a sharp explication
    of effectiveness. This has come to be called Church's
    Thesis....By its nature, Church's Thesis was not open to
    formal proof; for the thesis equated a precise property,
    recursiveness, with a property - effectiveness - that
    was to be rendered precise only by the thesis itself.
    But the thesis was supported by such instances as could
    be mustered, and soon it was pretty well clinched by
    Alan Turing's pioneer work in the abstract theory of
    computing machines. His formulation of mechanical
    computability, in terms of ideal mechanization, turned
    out to be equivalent to recursiveness. Mechanical
    computability, surely, is very much what our intuitive
    talk of effectiveness was aiming at all along; so
    Church's Thesis is well sustained.'
    -
    Quine (1987)
    Recursion

    'The first three chapters actually grew out of two
    earlier papers. Those papers were, in part, polemics
    against the views of my good friend and student Jerry
    Fodor. Fodor I hasten to say, is not the main target of
    this book; but I have retained some of my polemic
    against what I call "MIT mentalism"... The main target
    of the present book is one H Putnam (one of my former
    selves) and those who have adopted his views. Or perhaps
    it would be more accurate to say that the present book
    doesn't have a "main target"; for its aim is not so much
    to refute one particular view as to establish the need
    for a different way of looking at problems about "mental
    states". At any rate, the intended contribution of these
    three chapters to that end is to do two things: (1) to
    establish a close connection (discovered and emphasised
    throughout his career by W V Quine) between problems
    about meaning and problems about belief fixation, by
    showing that the holistic character of belief fixation
    in science bears deeply on the issue of individuation of
    "meanings" (or "contents" or "intentions", as they are
    called by various philosophers; and (2) to argue that,
    in fact, thinking of "meanings" (or "contents") as
    "theoretical entities" - as scientific objects, objects
    which can be isolated and which can play an explanatory
    role in scientific theory - is a mistake. In the course
    of the argument I defend the view that there is no
    criterion for sameness of meaning except actual
    interpretative practice - a view made famous by Quine
    and Davidson'

    H Putnam (1988)
    Representation and Reality

    'We cannot individuate concepts and beliefs without
    reference to the ENVIRONMENT. Meanings aren't "in the
    head."

    The upshot of our discussion for the philosophy of mind
    is that propositional attitudes, as philosophers call
    them - that is, such things as 'believing that snow
    is white' and 'feeling certain that the cat is on
    the mat' - are not "states" of the human brain and
    nervous system considered in isolation from the social
    and nonhuman environment. A fortiori they are
    not "functional states" - that is, states definable
    in terms of parameters which would enter into a
    software description of the organism. FUNCTIONALISM,
    CONSTRUED AS THE THESIS THAT PROPOSITIONAL
    ATTITUDES ARE JUST COMPUTATIONAL STATES OF
    THE BRAIN, CANNOT BE CORRECT'.

    The arguments I just summarised were, it might be
    pointed out in this connection, arguments against
    methodological solipsism.

    H. Putnam (1988)
    'Representation and Reality'
    (Professor of Mathematical Logic Harvard)

    -
    'I subscribe entirely to these sentences of Count Verri:
    On the Nature of Pleasure and Pain: The only moving
    principle of man is pain. Pain precedes every pleasure.
    Pleasure is not a positive state.'
    -
    Immanuel Kant 1781
    -
    -
    'Meanwhile our eager-beaver researcher, undismayed by
    logic-of-science considerations and relying blissfully
    on the "exactitude" of modern statistical hypothesis-
    testing, has produced a long publication list and been
    promoted to a full professorship. In terms of his
    contribution to the enduring body of psychological
    knowledge, he has done hardly anything. His true
    position is that of a potent-but-sterile intellectual
    rake, who leaves in his merry path a long train of
    ravished maidens but no scientific offspring.'
    -
    P. E. Meehl (1967)
    Theory Testing in Psychology and Physics
    Philosophy of Science pp 103-115
-
-
C56.
-
Naloxone enhances neophobia
-
J.F.W. DEAKIN & D.C. LONGLEY*
(introduced by T.J. Crow)
-
National Institute for Medical Research,
Mill Hill, London, NW7 1AA
-
Several studies report that naloxone, an opiate receptor
antagonist, reduces deprivation induced eating and drinking.
However, in the present study, naloxone (5mg/kg,i.p.) did not
reduce food intake of rats maintained on a 22 h deprivation - 2 h
feeding schedule. In contrast, naloxone (5 mg/kg,i.p.)
progressively reduced water intake in deprived animals to 46% of
saline treated controls. No effects of naloxone (1, 5 mg/kg) on
established bar pressing for food or water were observed with
either continuous or fixed ratio schedules of reinforcement.
However, naloxone (5mg/kg) accelerated extinction of responding
when food and water were no longer available.
-
Animals treated with naloxone (5mg/kg) during training of the
bar-pressing ate only 26% of the pellets delivered whereas
controls ate all pellets delivered. Since the animals had not
previously experienced the pellets or the operant apparatus, the
possibilities arose that naloxone effects were due to enhanced
neophobic effects of the novel food pellets or novel apparatus
cues, or were due to conditioned taste aversion. Therefore, food
novelty, apparatus novelty and timing of injections were
independently varied in different groups of 8-10 rats treated
with saline or naloxone. Rats were maintained at 85% body weight
with 12g lab chow per day. On experimental days 46 small pellets
(Cambden instruments) were placed on a small petri dish in the
home cage of some groups or released from a pellet dispenser in
an operant box for other groups. The dependent variable was the
number of pellets eaten over 15 minutes.
-
Naloxone (1,5 mg/kg i.p.) injected 5 or 20 min before test almost
completely suppressed pellet eating if the animals had not been
previously exposed to the pellets (p<0.01 't' test vs saline
groups). This occurred independently of whether tests were
carried out in the home cage or novel operant box. Naloxone
induced suppression of pellet eating was almost completely
abolished in either environment if animals had been exposed to
the pellets for the five preceding days in the same or different
environment. Naloxone (5mg/kg, i.p.) administered immediately
after pellet eating tests failed to suppress subsequent pellet
eating.
-
Thus, naloxone suppressed pellet eating if the pellets were novel
and if naloxone was administered before eating tests. The results
suggest naloxone enhances neophobic effects of novel foods and
that suppression of novel pellet eating is not due to enhanced
effects of novelty of apparatus cues or to conditioned taste
aversion.
-
Reference
-
FRENK, H & ROGERS G.H. (1979) The suppressant effects of naloxone on
food and water intake in the rat.
Behav. Neural. Biol, 26, 23-40.
-
PROCEEDINGS OF THE BRITISH PHARMACOLOGICAL SOCIETY (BPS) 1-3 April 1981
(Also British J Pharmacology 1981)

    'A word now about similarity: subjective similarity,
    which is crucial to the learning process. I don't think
    anyone has an innate notion of similarity. What one has
    incontestably is an innate subjective behavioral
    standard of similarity. It can be tested in people and
    other animals by conditioning.
    -
    It is unfortunate that my phrase 'standard of
    similarity' suggests judgment or deliberate comparison
    on the subject's part, but I am at a loss for another
    word. What is afoot is just conditioning,
    discrimination, reinforcement, extinction. It is what
    Stemmer and some other psychologists treat under the
    head of 'generalization class', but I prefer to allow
    differences of degree. If a subject is rewarded for a
    response to one stimulation and penalized for the same
    response to another, then a third stimulation is
    subjectively more similar to the first than to the
    second, for him, if it elicits the response.
    -
    Since it is basic to the mechanism of any learning, any
    conditioning, a similarity standard must be there to
    start the first learning. That is why I say it is
    innate. But our similarity standards evolve very
    decidedly as we learn, and that is part of what I was
    dealing with in 'Natural Kinds'......
    .....
    I first took up subjective similarity under the head of
    'quality space' on page 85 of WORD AND OBJECT. In ROOTS
    OF REFERENCE I went into it more fully under the head of
    'perceptual similarity'. I was concerned with it for its
    role in tying stimulations to observation sentences, as
    a basis for a theory of evidence...'
    -
    Quine
    'Comment on Hacking'
    In Perspectives on Quine (1990)
    -
>From the time of Frege's 'Sinn und Bedeutung' at least, there has
been a lot of interest in the foundations of mathematical logic
and the concepts of 'identity', 'analyticity', 'synonymy' and
'similarity'.
-
'The Morning Star' and 'Venus', 'the victor at Jena' and 'the
vanquished at Waterloo', and even 'consciousness' and 'brain
processes' have all occupied philosophers concerned with the
relationship between Sense and Meaning, the 'is' of definition'
and the 'is' of composition, 'Use & Mention' over the past 100
years.
-
Oedipus believed that Jocasta was fair game, but not his mother.
Ryle's foreigner saw all the buildings and grounds of Oxford, but
hadn't 'seen' the university. Jenny accepts that James has seen a
good range of her behaviour but doesn't believe he knows the real
'her'.
-
Ryle offered a solution to part of this problem in 1949 with his
example of the category mistake, ie that 'mind' like 'the
university' is a concept or category which includes a number of
members, distributed elements, or dispositions to behave. (In
Quine we see these being noted as observation sentences, or more
specifically, as occasion sentences, dated and timed).
-
But there are a few problems here. If we say that these elements
fall under a particular class, the elements of that class are
themselves not identical, nor even similar in sense. It may well
be that amongst Jenny's dispositions we include 'cake maker' and
'garden tender' as well as 'teacher', but we would rarely say
that these are identical or even similar behaviours any more than
we would say that the library is the same as or similar to the
refectory. They are similar or associated only in that they are
elements of a class, either 'Jenny' or 'university' - just as
different people can be said to be similar in terms of a higher
type, such as female, or human, or position in an actuarial
table. However, we would acknowledge that 'rabbit', 'undetached
part of rabbit', and 'stage of a rabbits development' are
interchangeable in identifying the same referent, 'salve
veritate'. This is the thesis of indeterminacy of translation, a
thesis which leads into another Quinean thesis, that of
Ontological Relativity - which dispenses with the intensional.
    -
    'The view that I have come to, regarding intersubjective
    likeness of stimulation, is rather that we can simply do
    without it. The observation sentence 'Rabbit' has its
    stimulus meaning for the linguist, and the observation
    sentence 'Gavagai' has its stimulus meaning for the
    informant. The linguist, observes natives assenting to
    'Gavagai' when he, in their position, would have
    assented to 'Rabbit'. So he tries assigning HIS stimulus
    meaning of 'Rabbit' to 'Gavagai' on subsequent occasions
    for his informant's approval. Encouraged, he tentatively
    adopts 'Rabbit' as translation.'.........
    -
    Discussion with Dreben helped me to clarify these
    consequences of my new stance. In WORD AND OBJECT I had
    already pointed out that communication presupposes no
    similarity in nerve nets; verbal behavior is inculcated
    on the strength only of surface stimulation. Such was my
    parable of the trimmed bushes (p.8), alike in outward
    form but wildly unlike in their inward twigs and
    branches. Save the surface, in the paintmaker's words,
    and you save all. But now, leaving the surface itself to
    Sherwin-Williams' tender mercies, I give the individual
    yet wider berth. His privacy widens apace.
    -
    Unlike Davidson, I leave the stimulations at the
    subject's surface, and private stimulus meaning with
    them. But they may be as idiosyncratic, for all I care,
    as the subject's internal wiring itself. What floats in
    the open air is our common language, which each of us is
    free to internalize in his peculiar neural way. Language
    is where intersubjectivity sets in. Communication is
    well named.'
    -
    Quine (1990)
    Three Indeterminacies
-
There are pervasive problems with 'properties', 'essences', or
'propositions' and their intensional kin. Nowhere is this more
apparent than with the 'propositional attitudes'. The linch-pin
is the resistance of these to the principle of 'substitutivity of
identity' 'salve veritate', and in anything scientific we are
exclusively interested in truth functions where the
substitutivity of identity is guaranteed. Failure to respect this
principle results in invalid reasoning through the fallacy of
equivocation.
-
In the case of propositional attitudes, we can not quote someone
indirectly without thereby uttering an untruth, and we can not
make statements about what someone believes, thinks, hopes,
fears, or understands except in the actual context in which the
propositional attitudes expressing such intensional idioms take
place (which amounts to quoting them directly and contextually).
If one reports to someone else 'what someone said', one has to
either directly quote (verbatim) or else acknowledge that one is
not making a report at all. Such paraphrase amounts to an
'interpretation' which is an imputation or inference, a creative
act. Propositional attitudes are not projectible outside the
immediate context of their utterance, so: -
-
    'Reports' are not 'bona fide' reports at all unless they
    are reports verbatim (ie records of behaviour) - they
    become interpretations through a process of imputation
    at best, and in almost all cases, they are creative acts
    which do not permit substitutivity of identity 'salve
    veritate'. They are non-truth functional unless
    expressed actuarially via relation to a population
    distribution (normative psychometric measure).
-
This is, I argue, why interviews, exams and similar devices for
measuring behaviour (e.g. minute taking in meetings) are widely
subjected to the criticism that they are not reliable measures
of behaviour, particularly when such measures are supposed to
comprise important elements in a human regression process (review
board) which seeks to make inferences or projections about other
behaviour. It is also why continuous recording of what happens is
more promising, since such measures provide more representative
samples under naturalistic conditions (ie the settings are more
prone to generalise) - Such a record is a profile of behaviour.
-
These principles are observable in the processes of operant and
classical conditioning. In the latter paradigms, one is
interested in the basic processes of association ascertained
after the acquisition process through observation of behaviour
following the presentation of a CS (the whole process is
therefore one of recognition or recall). The touchstone of rival
models is the 'blocking effect' (Kamin 1968) which illustrates
that elements of a stimulus compound may come not to control
behaviour unless those elements have come to provide new
information, ie change the conditional probability of the event-
event relationship. In operant acquisition, one tends not to
focus on the response topography, but records the class of
behaviours comprising lever pressing as the operant. But when one
looks at the extinction of this class of behaviours why does it
take so long to reach criterion? May it not be that the
acquisition of the operant is in fact no different from the
configuring process in classical conditioning, ie the acquisition
process is the acquisition of multiple R-S contingencies, and
extinction involves a testing of all the elements:
-
In its most basic form each R-S* during acquisition differs, each
is a slightly different 'perspective'. This might comprise a
slight motor variation on a CRF task, such as a slightly
different pressure on the lever above threshold, or angle of the
body, in other cases it may be the ratio of presses on a FR or VR
schedule (which will increase resistance to extinction). Note
that the element of behaviour here is lever press (p or p*) and
the outcome is pellet delivery (q) ie
-
                   if 'p'ress then 'q'ualia.
-
Each CRF trial RECORDS either 'p then q', 'p* then q', 'p** then
q' and so on. In fact every supra-criterion R-S variation
possible, leads to a testing of almost all of the contingencies
during extinction :
-
                  not 'q'ualia then not 'p'ress
-
but...perhaps it will be p*.... or perhaps p**, so each has to
be tested/subjected to falsification (negation). After all, each
variation or 'perspective' was reinforced during acquisition. On
partial reinforcement schedules we interleave a little extinction
or impossible discrimination training.
-
Do each of these behaviours, dispositions, or properties, learned
only in their particular context, occupy their own location or
node in the operant repertoire, only becoming extensional through
the eyes of the observer? Or are they really configured into a
class of such behaviours?. Surely all trainers have had
experience of this process whereby the basic elements of a
complex skill may be taught but the trainee just fails to pull it
all together coherently, and surely all assessors have had the
problem of deciding whether the student 'really knows'.
-
The same processes seem to be operating in 'group work' of all
sorts. Here, as with performance on a test, the behaviour
occurring in such settings, (which is predominantly verbal), may
not generalise to other situations or be representative of
general performance (other than to other similar behaviour in
groups or interviews!). Such context effects have been central to
Cognitive Psychology for decades e.g. 'the encoding specificity
principle' of Tulving and Thompson. Such processes seem to be
quite basic to the very nature of habit formation (cf. notes on
novelty, opioids & habit formation). Similarly, if 'role plays'
had any substantial (generalisable) effects on behaviour, then
many of our oscar winning actors would surely be more likely to
be roaming cities killing and maiming, (or living in idyllic
relationships). Acting is no less real or true behaviour than is
any other.

David Longley
July 7, 1991 

-- 
David Longley

Relevant Pages

  • Re: Turing Machines and Physical Computation
    ... citing once again (although without their original context for reasons ... controls ate all pellets delivered. ... Stemmer and some other psychologists treat under the ... response to one stimulation and penalized for the same ...
    (sci.math)
  • Re: Turing Machines and Physical Computation
    ... >> take their emotional responses as something to further act upon. ... I have commented on what i got from each in this context. ... >> response to one stimulation and penalized for the same ... >> free to internalize in his peculiar neural way. ...
    (comp.theory)
  • Re: Turing Machines and Physical Computation
    ... >> take their emotional responses as something to further act upon. ... I have commented on what i got from each in this context. ... >> response to one stimulation and penalized for the same ... >> free to internalize in his peculiar neural way. ...
    (sci.math)
  • Re: What does this instruction mean: ANDDC #239 ?
    ... >have to scroll down to see the reply after just reading the original. ... >scroll down to look at the original context to understand that". ... >response in isolation, there's either enough info in the response (but ...
    (comp.arch.embedded)
  • Cash in respect of Jim
    ... slide in response to roasted rural movies. ... Some joint cures free Linda, ... below all the psychologists? ... Ramsi, have a free sponsorship. ...
    (sci.crypt)
Loading