Grice's "Indeterminacy" -- logical?

by JLS
for the GC

In his 1960 "Mind" essay, D. Mackay proposes to talk of 'logical indeterminacy' in the case of rational choice -- rational free choice and free will.

Consider

"I am a free agent"

--- what are the things that follow from that? Can we expand on the proposition without merely yielding tautologous entailments? Can we rather derive necessary propositions which are not tautologous?

Donald Mackay is British physicist, member of the Department of Physics at Keele, and King's College, London. His locus classicus is his article for "Mind", 1960, on logical indeterminacy and free will, examined by C. Tinker, at

http://www.churchsociety.org/churchman/documents/cman_118_1_tinkerc.pdf

Tinker aims at showing the relevance of MacKay’s much neglected hypothesis of Logical Indeterminism, a compatibilist attempt to use science to understand the concept of free-will and Determinism. As MacKay thinks through his propositions he acknowledges that we must do justice to a number of things —
(a) first, our conscious experience;
(b) secondly, scientific data;
—and so he attempts to reconcile these two and provide an acceptable model which aids our understanding of freedom of action.

MacKay begins with a working assumption of mechanistic brain science; that ‘all a human agent (A) believes or knows is represented, explicitly or implicitly, in the physical configuration of his brain’. This means that no change can take place in A’s belief without a correlated change within his cognitive mechanism (CM). MacKay depicts this by looking at two simultaneous stories, the I story and the
Observer story (Brain Story). These can be told as follows:
I story -------------------Brain Story
I see..................... Subsystem N1 is doing......................
I hear.................... Subsystem N2 is doing......................
I remember........... Subsystem N3 is doing......................
I believe................ Subsystem N4 is doing......................
MacKay suggests that these two stories should not be understood as two trains of parallel events but ‘inner and outer aspects of one complex train of events that constitute my logical agency’
These two stories are seen to bear witness to complementary and essential facts that the other may ignore, for MacKay sees in our human nature an irreducible duality—but a duality of aspects rather than
substances. It is important to note that these two ‘stories’ are complementary descriptions of what is taking place, one does not rule out the other.
MacKay acknowledges the fact that it is not currently possible to measure every aspect of the cognitive mechanism, and may never be, but he seeks to develop his ideas as a thought experiment and so suggests that we think of a Superscientist (S). S has been fully informed so that he is able to give a complete description (D) of A’s cognitive mechanism (CM) at time t so—D=CM (t)


MacKay suggests that A is free if there is no prediction of our future actions involving decisions which is logically binding upon him.

By this he means that there is no prediction concerning A’s cognitive mechanism that can lay claim to A’s unconditional assent, that is a prediction that A would be correct to believe
and, in error, to disbelieve. MacKay proposes that in order to ascertain whether A is free in his action we should take the prediction to a logical court in order to determine upon whom
this information would be logically binding. The logical court, MacKay suggests, would conclude that S (or any non-participant onlooker) would be correct to believe and in error to disbelieve what D specifies. The next question that needs to be asked of the logical court is ‘how binding is specification D upon A (the agent)?’ One would assume that if it is binding upon S it should
also be binding upon A, but the logical court returns a different verdict, if t is the present or the near future then A would be in error to believe it for no change in his belief could take place without a correlate in his CM. Therefore D has no unconditional claim to A’s assent, that is, he would not be correct to believe it and in error to disbelieve it. D is a description of A not believing it,
so if he were to believe it, it would immediately become obsolete. This situation is logically relativistic, and is a situation that only occurs with cognitive
mechanisms, statements concerning situations outside of the cognitive mechanism are logically binding, but here, what is true for S is not true for A.
MacKay points out that if D were embodied in A’s CM then it would not correctly specify every detail of CM at t, and so it would become immediately
out of date. A possible solution to MacKay’s logical problem may become clear to us as we think of S. Why does he not simply produce D1=CM1 (t) accounting for the
change in A’s CM as he is told the prediction (i.e., D would include A believing the prediction)? MacKay allows for this proposal and suggests that we also
take this to the logical court and see its verdict. The court would conclude that A would indeed be correct to believe it; however he would also not be in error
to disbelieve it, for it is designed only to be correct if A believes it. This suggestion means that S simply produces D1 whose correctness is up to A who
can believe it but would not be in error to disbelieve it; again D1 in this case is not logically binding upon A, it does not have unconditional claim to A’s
assent. ... an A who did not believe it would not be in error; nothing is
inevitable for A. This type of prediction, however, also has important
consequences for the detached observer, for it cannot lay claim to his
unconditional assent either, for it all depends on A and whether he believes it
or not, for there is a possibility that the observer would be incorrect to believe
the prediction; conditional predictions can lay claim to no-one’s unconditional
assent.
Therefore, MacKay concludes that the only type of prediction that is of
any value is an unaltered prediction because then, at least, it can lay claim to
the unconditional assent of the observers instead of no-one at all.
An objection can now be made as we ask, ‘What room is there for me to have
a say in the matter?’ For, if the mesh can be made complete we can say ‘purely
physical factors completely determined the physical activity that expressed my
decision’.17
MacKay argues that this is only a problem if we hold the false
supposition that ‘claims to determination are always mutually exclusive’.18
Our claims to determination here are framed at disparate logical levels. They
‘are not competitive but complementary’19 when seen from alternative logical
standpoints. From the standpoint of S our actions are determined by our brains
and he can know exactly what we will do, but from our point of view, there is
no specification of our future actions that we would be correct to believe and
in error to believe,

we still have to ‘make up our minds’.

Certain scholars have taken MacKay to task, some, such as John Thorp,
providing relatively weak criticisms which can perhaps be best summed up by
those of William Hasker. Thorp begins by criticising
MacKay’s suggestion that ‘a man can never know the state of his own central
nervous system because by so knowing he alters it and therefore does not know
it as it is’.He argues that there is no difficulty in knowing the state of one’s
own CNS. Thorp proposes neural state j which is the correlate of this mental
state and one could not have neural state j unless one knew one had it. This,
he claims, shows MacKay to be wrong, for Thorp has managed to show that
one can hold in one’s cognitive mechanism a belief of a prediction without
falsifying it. He then concedes however, if one believes that causes must
precede their effects one can only say it is a correlate of believing, not knowing.
Once he concedes this he argues that our intuitions would suggest otherwise,
and would presume that it is still knowledge. It is a shame that he has to appeal
to intuitions, for intuitions, especially about our brain states should not be
used to refute something that seems logical, and here is clearly mechanistic.
Therefore, it seems that it is not ‘all that the refutation of MacKay here would
require’.MacKay is on more solid ground with his reasoning than Thorp.
Not only this, but, as we have seen, it is not a case of A’s psychological capacity
to believe the prediction but a case of the rational obligation for him to believe
it, Thorp misses this point entirely.
Secondly, Thorp suggests a scenario in which a computer capable of making
predictions feeds A with adjusted predictions, taking into account A’s reaction
to the given prediction, thus producing prediction about A’s beliefs. Thorp
argues that, because it is possible, it is sufficient to show that MacKay is
wrong. Thorp argues that in three out of four of his paradigm cases it would
be right for him to ‘believe a prediction of [his] decision; [he] should not
thereby falsify it’.He concludes that ‘MacKay is thus wrong, after all, to hold
that ‘no complete prediction of the future state of the organising system is
deducible upon which both agent and observer could correctly agree’.

In some
cases, at least, such a prediction IS possible’.

Thorp is correct in what he
claims for the prediction, but he is wrong in what he quotes as MacKay’s view.
MacKay would agree that agent and observer could correctly agree. But Thorp
talks in the language of ‘could’ and ‘should’ and this is not the language that
MacKay uses.

MacKay uses the language of ‘must’.

His question is ‘does the
prediction have a claim to A’s unconditional assent?’; ‘would A be correct to
believe it and wrong to disbelieve it?’ Thorp simply suggests that A would not
be wrong to believe the prediction. However, he would not be wrong to
disbelieve it, that is, it would also be right for A to disbelieve a prediction of
his action because it would become false. It seems that Thorp has
fundamentally misunderstood MacKay and so his criticisms are of an idea that
is certainly not propounded by MacKay.
Having decided that MacKay’s criterion is wrong, Thorp continues to look at
what MacKay’s theory would mean if it were correct. Since Thorp’s criticisms
fail, he ironically describes what MacKay’s criterion actually does mean for us;
‘the two contradictory propositions are both qualifiedly true:

[Stage 1]
(a) X’s decision to do y was a FREE decision, and
[Stage 2]
(b) X’s decision to do y was CAUSALLY determined.

As far as X is concerned, (a) is true, and as far as everybody else is concerned (b) is true’.26

We now turn to Hasker’s fuller and more detailed criticisms. Hasker begins by
outlining what he thinks MacKay is trying to say. He suggests that a
mechanistic explanation of human behaviour could be seen to eliminate
free will; but he notes that MacKay thinks that the
mechanism retains the ‘logical indeterminacy which is required for moral
responsibility’.27 To explain his concerns Hasker adopts the story of Osmo.28
This story describes a young man who discovers an ancient book which
describes his life so far most accurately and purports to describe his life up
until his untimely death in an aeroplane crash. These things take place as
described despite Osmo’s attempts to prevent them.

This, Hasker argues,
would constitute a loss of freedom even in MacKay’s eyes. As a result of this
story, Hasker concludes that—

"[f]or me to be free requires not only that I am not aware of any prediction
of my future action which I am rationally bound to accept as inevitable,
but that there is no such prediction, whether known to anyone or not—
that there cannot be any such prediction."29

This first statement, rather than being a criticism of MacKay, sums up
MacKay’s viewpoint; these requirements are exactly the same as MacKay’s and
he argues that no such prediction can exist, only a prediction which the
predictor and other detached observers would be rationally bound to accept.

This is not to say that there are not predictions or statements that demand
unconditional assent from all, observations of nature require it, but
observations of the cognitive mechanism logically cannot demand this.

If we
look at our own cognitive mechanism, there is one part we could not observe,
the part which is processing the information about what we are observing.

MacKay shows this with his example of a cerebroscope30 which could, in
theory, allow us to observe our own brains—all but the area that is processing
the information, like a video camera directed at a monitor which is displaying
what it is filming; at a certain point the camera is no longer able to produce a
picture, for it is filming it’s own output.
Hasker’s main criticism however, concerns MacKay’s claim that ‘a prediction
of my behaviour may be falsified by my believing it’.31 Hasker has problems
with this because MacKay argues that the prediction is based on all the outside
influences, which Hasker argues must include the predictor (if he intends to tell
A his prediction) telling A; so the prediction will, in fact, have this built into it.
Hasker is not convinced that simply because the predictor has predicted that
the subject will believe him, its truth depends on him believing it, leaving it up
to the subject whether the prediction is true or not. He argues that although it
may be logically possible for him to disbelieve it, it is not actually impossible
for him to disbelieve it and so it is not in one’s power to falsify the prediction
because it will take place. Hasker also suggests a scenario when it is impossible
for one not to falsify the prediction where it becomes impossible for the
predictor to show you the true prediction, but this just as much diminishes our
free-will as the other scenario.
So, how can MacKay’s theory stand up to this? MacKay suggests that Hasker
has become confused by the story of Osmo and is ‘treating the question of
rational obligation in principle as if it were a matter of psychological capacity
in practice’.33
If A is to be rationally obliged to accept the prediction then a twofold
condition is necessary, that A would be correct to believe and in error to
disbelieve it if only he knew the prediction. As we have seen, this cannot be the
case for an unadjusted prediction, for it becomes false if he believes it. Hasker,
therefore demands an adjusted prediction. If one looks at the adjusted
prediction then one must conclude that the logical court can agree that A would
be correct if he were to believe it but not in error if he were to disbelieve it
because while he does not, it remains incorrect. If it remains unrevealed it
becomes a prediction that no-one can correctly believe for it is not accessible for
A to believe and therefore others cannot believe it because A does not. The only
prediction that can lay claim to the unconditional assent for anyone is an
unadjusted prediction that can claim assent from the detached observers. The
question that is posed when one is confronted with a prediction is not ‘Can I
(psychologically) reject this?’34 but ‘Have I a rational obligation not to reject
this?’ The logical court needs to ask ‘whether [S]’s interference with A had so
affected A’s brain mechanism that the prediction would be equally well founded
(in mechanistic theory) even if A disbelieved it’.35 The outcome is still up to A,
‘not in the sense that detached onlookers could not predict his assent to it, but
in the sense we have explicated, that their evidence leaves either outcome
rationally open to him until he determines one by making up his mind’.36 The
outcome is still up to A unless the sufficient causes of the predicted outcome are
now independent of the state of A’s cognitive mechanism. A would be rational
if he believed whatever prediction described what he decided to—‘there is no
one outcome that A would be correct to regard as the only one rationally open
to him as a personal agent beforehand, even though he accepts that
deterministic laws apply (ex hypothesi) to his brain.’
MacKay wants to ensure that we do not carelessly mix concepts from the two
levels of
‘(1) rational obligation, determined by REASONS, and
(2) physical brain activity, determined by CAUSES.’

He highlights the flaw in Hasker’s argument
by asking, ‘Of what are you saying that A is ignorant? If you mean that he does
not know what P knows, that is granted at the outset; but what we have seen
is that what P knows of A’s future would not be knowledge for A if he had it.’39
Hasker replys to MacKay, re-asserting his conviction that ‘a thorough-going
mechanism with regard to brain function is incompatible with the belief that we
are free and responsible agents’.40 He then proceeds to work through MacKay’s
thesis with every situation he can conceive. First, he looks at the case where A
is not confronted with a prediction and concludes, rightly, that it cannot fulfil
MacKay’s criterion for a lack of freedom. He then looks at situations when A is
confronted with either an adjusted or an unadjusted prediction again
concluding that, according to MacKay’s criterion,

A remains free.

He sums up by saying that ‘[a]ccording to MacKay’s criterion, I am free in my actions unless it is possible for there to be a prediction of my
actions which I am rationally obligated to accept—which he takes to mean that
the prediction must be sound whether or not I believe it’.41
Hasker concludes that
this is tautologous.
Since the statement that ‘I am a free agent’
flows from this (a statement that he claims is not a tautology) MacKay’s
criterion must be wrong for only a tautology (which the above is not) can flow
from a tautology.
But is it true that, given the mechanistic assumption, the
statement that ‘no prediction that I know can be sound whether or not I believe
it’ is tautologous?
Because Hasker simply states that MacKay’s criterion is a tautology, we do not
have an explanation of why we are to understand it as a tautology. Neither is
it immediately obvious that it is a tautology. As E. J. Lowe states, there are
two particular ways in which the term ‘tautology’ (a somewhat ambiguous
term) is used. The first is to describe a logical truth in the strict sense,
something which MacKay’s criterion is not. The second is to describe
something that is analytically true, but again it does not seem that MacKay’s
criterion is this either. Without possessing the reasons that Hasker had to come
to the conclusion that MacKay’s criterion is a tautology, it is difficult to assess
whether he is correct in his assertion. We can say, however, that it is not clear
that MacKay’s criterion is a tautology.

There is, however, another option open
to us here, the statement may not be tautologous, but in fact a necessary truth.

For it is possible for a statement to be a necessary truth without being a
tautology, this would mean that the statement—

‘I am a free agent’

—is also a necessary truth.

If MacKay’s criterion is a necessary truth then
Hasker’s criticism fails and MacKay’s criterion is still valid.

Ultimately it seems that the problem is not in MacKay’s hypothesis. Hasker has
to reject it despite the fact that it works within itself and fulfils all the necessary
criteria; it is the criteria that is a problem, or to be more precise,

the concept of freedom.

For Hasker, freedom requires that ‘some human actions are chosen
and performed by the agent without there being any sufficient condition or
cause of the action prior to the action itself’,but for MacKay, freedom simply
requires that there is not prediction that one is rationally obligated to accept.

So who is correct in their understanding of freedom?

MacKay’s hypothesis
stands or falls depending upon whether one’s concept of freedom is in line with
his.

It is quite clear that Hasker, although he does not eliminate causes or
conditions influencing our actions, cannot accept that at all times there is a
sufficient cause.

----- Fischer has discussed McKay.