31 July 1995
Sam!!!!!!!!!!!!
1. Time to think! Let’s start with time
and space. Long ago (and in a galaxy far, far away), I developed a set of
notions I call philosopher’s space and philosopher’s time. These
aren’t supposed to be exactly like real space or real time or real space-time.
Instead, they’re supposed to be highly abstract versions of the notions of
space and time that aren’t so abstract as to be unrecognizable. To get
real space and time out of them, you’d have to add lots more details.
To begin with, both represent an ordering of events. That means that, at least,
there must be events in the plural. (What’s an event? “The unit of things real,”
said Whitehead. It should be added, as he sometimes did, that events can be
more or less complex: Some can be composed of others. Are there any that are
completely simple? I don’t know any good argument either way. Beyond that, I
don’t care to and probably can’t define it. Treat “event” as a primitive term
like “point” in Euclidean geometry.) Why plural? Well, suppose there were only
one event in a spatial order. We might be able to say that it is “here.” (Leave
aside any questions about how we are supposed to be able to talk about it.) But
we wouldn’t be able to say anything about what is “not-here”: ex hypothesi,
nothing is not-here. But if we’re lacking the contrast case, not-here,
then we haven’t excluded anything, ergo, haven’t said anything, by saying that
it is “here.” It’s easy, of course, to construct a parallel argument with “now”
and “not-now” for time.
Now, I have to ask you to adopt another primitive term, “cause.” Can I say
anything more about what a cause is? I think so, but I’d like to defer it
perhaps to when we get into puzzles about backward causation. Let’s just say
that a cause is an event on which another event depends. (So, a cause is an
independent variable, while an effect is a dependent variable.)
Now, very briefly, with lots of details and sub-arguments left out, a (purely) temporal
order is an array of events in which some are causally dependent on others.
More precisely, each is either a cause of one or more others, an effect of one
or more others or both. A (purely) spatial order is an array of events that are
causally independent of others in that array. I.e., the members of a spatial
array are not causes of each other or effects of each other.
A pure temporal order would consist of a sequence of events. There would be
room for descriptions in terms of “before” and “after,” but not for any notion
of how long A occurred before B (except in the sense of counting, perhaps, how
many, if any, events intervened).
A pure spatial order would consist of an array of “simultaneous” events.
(Simultaneous is a very weak term here. Once you get into real world
applications, of course, it will turn out that simultaneity doesn’t even have
to be transitive: X can be simultaneous with both Y and Z but Y and Z don’t
have to be simultaneous with each other.) What kind of geometry would a purely
spatial order have? First and obviously, it would be non-metrical. Without
reference to time or, more precisely, to causal processes intermediating
between events, we wouldn’t be able to say anything about what the distances
between events are – about whether A is closer to B than to C, etc. Does that
mean that it would involve just topology? I think not: topological relations
presuppose dimensionality and a space without distances would only have an
indeterminate dimensionality.
[Let’s take an excursion or digression into cosmology here. Assume one of
the incarnations of the Big Bang model. Most say that the Big Bang didn’t
originate from a prior universe or rearrange existing material. If time is a
relation between events, there were no events before the Big Bang, hence, no
time before the Big Bang, hence, no meaning to “before” in “before the Big
Bang.” Time is in the universe, not the universe in time.
And yet .... it’s hard to stop with that. Even if we can’t make sense of
something being before the universe, it’s hard to avoid thinking in terms of
some sort of background conditions. Specifically, if the universe is to have
originated in a Big Bang, it must have at least been possible for a universe to
originate that way. Which is to say that there is a framework of possibilities
that is somehow not in time or, at least, not in our time. Suppose we call this
framework of possibilities, Nature. (It may be fruitful to think about Nature
in terms of the qabalistic Zero or Ain, the nothing which is “the foster-mother
of all becoming.”)
An immediate question arises: Could this framework have engendered other
Big Bangs, other universes? Where we know so little, it’s hard to rule anything
out including the possibility that some of the other Big Bangs are just Little
Sputters. But if we suppose that there are other universes, it appears they
would have the kind of relation to each other that I’ve called purely spatial.
None would be before or after another, none causally dependent on another. They
would not interact and the events in each would define temporal orders that
applied within but not between universes. Or, perhaps it’s going too fast to
assume they can’t interact. If interaction does occur sometimes then events in
the two quasi-universes (“quasi” because not completely self-contained) might
be ordered with respect to one another. But it does seem too much to assume
that it must be the case that all the universes in Nature interact
directly or indirectly with all the others. (Besides, if they do, then we
might, in principle, be able to talk about a beginning of Nature,which would
have to be possible, which might lead us into speculations about MetaNature
....) But if they don’t all interact, then there’d still be some cases in which
some of them or some sets of them had purely spatial relations to others.]
If we suppose an array of events that have both spatial and temporal
connections to one another, we get closer to the actual world. And here we find
something interesting. (I assume, of course, that you’ve been utterly
uninterested in what has gone before.) I’ve started with notions of time and
space that are thoroughly relativistic. Both are constituted by various more or
less complicated relations between or among events. In a full-fledged spatiotemporal
world, each has characteristics that depend upon the other: distance and
dimensionality in the case of space, duration in the case of time. Part of the
reason is that I just don’t see how else to think about them.
But now it looks like this connection between space and time might be upset. We
might have to re-admit absolute space and time. For, suppose that signals could
travel at arbitrarily high velocities. (That wouldn’t mean that any particular
causal process could have infinite velocity, just that there would be no
particular highest value for a velocity – i.e., you could always go faster by
pushing harder and there’s no particular asymptote towards which “going faster”
converges.) Then, it seems that you would get absolute time or, what amounts to
the same thing, a non-frame-of-reference-relative conception of simultaneity or
temporal ordering. All events could be dated with respect to all others without
regard to frames of reference. But then, of course, relative space does not
cohere well (or at all) with absolute time. Fortunately, for my reconstruction
though not, perhaps, for the envisioners of galactic empires, it appears that
we don’t have to face this particular question: The speed of light is the limit
at which signals can travel. (Does this completely rule out faster-than-light
travel? I’m not sure. There may be loopholes – say, connections to a
non-ordinary space with its own, but different, speed limit. This may be equivalent
to the sub-space or hyper-space beloved of science-fiction novelists. Of
course, if there is some way of accessing a non-ordinary space having some kind
of correlation with our ordinary space, there seems to be no a priori reason
for supposing its speed limit to be higher than ours. Another possibility is
that space has a more complicated sort of connectedness than is allowed for in
ordinary 4-D space-time. Thus, two events that are distant in, to coin a word,
light-space, may be adjacent in some high-D space. This may be equivalent to
the space-warps also beloved of sf novelists. A point in favor of this notion
might be that it provides a simple interpretation for Bell-Theorem
non-localities: Two light-space-distant events may be occurring in the same
place, in fact, may be the same event in a higher-D space. A point against this
is that it seems to imply that there are there are or could be events which are
distant in light-space but close, though not in the same location, in the
higher-D space. Since the only really satisfactory evidence we could have for
that would be successful ftl travel or communication, and since such evidence,
as far as I know, hasn’t been forthcoming, we might be tempted to suspicion.)
In any event, I take the fact that my reconstruction seems to imply that there
is an ultimate speed limit in conjunction with the fact that the physicists
seem to agree as welcome, if only partial, confirmation.
2. Forward to backward causation! Above, you’ve seen a little of my version of what
I take to be a fairly common view among philosophers about time. As we’ve
discussed, it seems to make backward causation impossible: If time is built up
out of causal relations, then anything that causes something else will get
counted as part of the past of that something else.
But why let impossibility stop us, deter us, delay us? Why not rather work
harder on it because it’s impossible? Let’s try a fresh attack.
What does it mean to say that A causes B? Normally, we think that causes bring
about or produce their effects. B is made to happen by the occurrence of A. But
this can’t be adequate. “Bringing about,” “producing,” “making things happen”
are themselves causal notions. How does A make B happen? Does it stretch out
metaphysical fingers towards the not-yet-B-ish future to
push/pummell/pound/nudge it into shape? (What are metaphysical fingers made
of?)
One idea is that events like A and B are connected because they’re special
cases of laws of nature that say things like: Whenever an A occurs, it will be
followed by a B. (Oops ... seems to beg the question against backward
causation. Leaving that aside ....) I really don’t think this quite captures
our ordinary notions – we think that this B is caused by this
A, not that A’s causing B is dependent on how a whole lot of other events turn
out. However, that could be because our ordinary notions are wrong. So, to
pursue it further, some questions about causation between A and B reappear
transmogrified as questions about natural laws. How does the law that all A-type
events are followed by B-type events get a hold on particular A-type events and
B-type events? What makes them go along?
One answer, developed out of the thought of Hume, is that nothing makes
them go along. All there is to a law of nature is constant conjunction. We say
that there’s a law of nature when we find that, in fact or at least as far as
we can tell, that B-type events always do follow A-type events. This amounts to
an inversion of a conditional. Given the ordinary notion of causation, people
expected B-type events to follow A-type events because they thought that each
particular A caused a B to follow in its train. Hume said it’s the other way
around: We expect particular B-type events to follow particular A-type events
because B-type events have always followed A-type events. Hume’s own version of
this is plainly flawed but modern Humeans have devoted considerable ingenuity
to cleaning up the details.
One of the problems they haven’t dealt with very satisfactorily is the way that
law-statements support counter-factuals. For example, it might be a matter of
fact that there are in the universe no hats the brims of which have a
circumference of 100 miles or more. That wouldn’t keep us from making such a
hat, even if, plausibly, no one is ever even slightly tempted to do so. On the
other hand, if it were a law of nature that there couldn’t be such hats, then
no one could make one even if he tried. If it were a law of nature that
everyone in this room is over 21, then I couldn’t successfully invite the
newspaper carrier in. And so on. The Humean doesn’t have a good way to
distinguish universals of fact which might well have been (or be or become)
otherwise from laws of nature which could not. (There’s something more than a
little puzzling here which we may get back to. For most people are inclined to
think that the laws of nature, too, could have been otherwise ... so we’ve got “could
have been otherwise” working on at least two levels.) Another problem for the
Humeans or perhaps another version of the same one is that it seems to pick out
as causal relations things that we are ordinarily reluctant to so class. For
example, death always follows birth; therefore, birth causes death. Or, clock A
always strikes the hour before clock B; therefore, clock B’s striking is caused
by clock A’s. And so on. Still another problem has to do with some of the ways
that we actually reach conclusions about causality. Suppose I sit on a hot
stove. I (very quickly) conclude that sitting on a hot stove caused pain. What
I don’t do is look to see if other cases of sitting on a hot stove have also
been associated with pain, see if other people sitting on that hot stove feel
pain, if people sitting on other hot stoves feel pain, if I feel pain at other
times on that hot stove, etc. before concluding there’s a causal relation. But
if the Humeans are right, that’s what I should do.
Another approach says that the causal relations between events flow from the
natures of the entities interacting. Ceteris paribus, struck matches light; lit
matches ignite other things (with sufficiently low combustion points determined
by their natures); ignited things ignite other things, and so on. More
generally, a thing does what it does because it is what it is.
There are two problems with this: First, it isn’t obvious how we’re supposed to
have access to the “natures” of things apart from seeing what they do. If we
don’t have any way of figuring out what a thing “is” other than seeing what it
does, then the formula seems to reduce to “a thing does what it does,” which is
true but not very helpful. In other words, processes that happen quickly count
as what a thing does, while processes that happen slowly count as what a thing
is. Second, even if there were a way around this, there’s still a question
about how the nature of a thing determines its behavior. The most common move
here has been to view it as some kind of logical connection. But logical
connections are timeless, so it’s not clear how causal relations which –
whether I’m right or wrong about the details – clearly are closely connected
with the occurrence of events in time, are to be explained in terms of timeless
relations. (To put it crudely, if the relation between causes and effects is
logical, why don’t all effects happen at once? Why do they wait around to occur
in some particular order?)
One solution might be to hold out for quantized events which happen all at once
or not at all. Then (simplifying) each event would “imply” its immediate
successor which would imply its immediate successor, and so on. I’m not happy
with this for two reasons. First, it seems unintuitive to me that there should
be an a priori argument that none of the basic laws of nature are functions of
continuously changing magnitudes. (I don’t have any special objection to
concluding that, just to concluding it on the basis of philosophical arguments
without spending any time looking at the world.) Second, I’m not sure that it
really does handle the problem it’s invoked to deal with. Suppose that the
causal connections between events are a string of logical implications. What
makes it the case that the string of events takes up time unless each
event/implication takes up time? Otherwise, it seems that any string of
implications, of whatever length, could be squeezed into any finite duration
(if that can somehow be exogenously specified) or into no duration at all (if
it cannot).
A third problem for this “logicism” is the question of where these
consequence-implying natures come from. There seem to be two general
possibilities. Either they come from the natures of their components or they
are implied by the occurrence of earlier events (or some combination of these).
Suppose everything is made of quarks and quarks aren’t made of anything else.
Then, given an initial distribution of quarks, the future of the universe is
implied. (Some of the “implications”, though, might be of the form ‘If A, then
pB or (1-p)C, and not both B and C, where p is some probability value.) But
what about the nature of the quarks? Ex hypothesi, they’re not made of anything
else, so they’re not determined to be what they are by what they’re made of.
Also, since we’re talking about an initial distribution, their nature isn’t
determined by any prior events. Does it make a difference if we lift the
assumption that the universe has a finite past? I can’t see that it does; it
just means the nature of quarks is eternally unexplained. Any way you look at
it, the nature of something, which has an impact on what causal relations there
are, turns out not to be implied by anything else in the system which may be
uncomfortable for the “logicists” but gives sense to the common assumption that
laws of nature could have been different. (Let me note something I’m not
saying: It may very well be the case that some characteristics of some
things imply laws of nature. For example, if chemical combination is the
sharing of valence electrons and if helium is [among other things] an element
whose atoms have no valence shell of electrons, then the chemical compound,
helium sulfide, may be impossible because self-contradictory. The fact that you
can’t discover that self-contradiction from your armchair doesn’t mean that the
impossibility is of some other kind than logical impossibility. As a matter of
fact, this law of nature would be true, as the logicians say, in all possible
worlds. However, that doesn’t guarantee that all possible worlds will be worlds
containing helium or sulfur or chemical combination. For some interesting – and
fun – exploration of some of these issues, you might like to look at
Saul Kripke’s Naming and Necessity.)
Now, it seems that I’ve spent a very long time telling you about theories that
I think are wrong. I’m afraid I can’t do nearly as well on what I (tentatively)
think is right. However, I will tell you where you can find a lot more if you’re
so inclined. “Where” is in two works by the Australian philosopher, D. M.
Armstrong: Universals and Scientific Realism and What is a Law of
Nature? Now that I’ve told you where to find more, here’s a bare-bones
outline:
We start with universals. A universal is something which may be multiply
exemplified, like a specific shade of red which can be in more than one place
or time. By contrast, a particular is something which can’t be multiply
exemplified, like the keyboard I’m typing on. Universals should not be confused
with concepts. A concept may be our best attempt at picking out some
commonality in the world, but in the sense that Armstrong is using it, there
can be a universal when there isn’t a concept and there can be concepts that
don’t pick out universals. What universals there are in the world is a matter
for empirical investigation and discovery.
An event can be called a “this-such”. Each event has “thisness”, i.e.,
particularity, united to “suchness”, i.e., some one or set of universal
characteristics. The “logicist” account I criticized above can be viewed as an
attempt to do without any thisness in events, to have everything that can be
said about them to be a special case of their suchness. The Humeans, on at
least some interpretations, try to do without the suchness of events, to say
that there are no limits upon what events may be or how they may follow one
another except how they happen to follow one another.
Causation is a singular notion. That is, it connects this cause with its
effect. That means that, in principle, there might be a cause that necessitated
its effect even if other events exactly like that cause didn’t produce similar
effects. Thankfully, there doesn’t seem to be much serious evidence that there
are any such causes. (Part of the reason is that it’s so hard to say when two
events are exactly alike.)
However, for the normal case, causes cause their effects by virtue of some
universal characteristic. When an A-type event is not followed by a B-type event
although we have hitherto thought that all A-type events are followed by B-type
events, we either conclude that we were somehow wrong about the causal law or
that it wasn’t really an A-type event – that it lacked some characteristic of
other A-type events or had some extra characteristic that A-type events
normally do not.
Thus, a law of nature turns out to be a connection between universals –
universals specifying the causing event and universals specifying the caused
event. Since they are universals, a law of nature is itself a higher-order
universal. It could, of course, have been different: the connection between the
basic universals that provide the terms for laws of nature is not logical but a
matter of brute fact. However, given that a law of nature is a universal, it
will (by definition) always apply. Evidence that some connection is not
universal is ipso facto evidence that a statement of that connection was not a
(correct or complete) statement of a law of nature.
There may be problems with this but it seems to handle most of the objections I
deployed above against other accounts. By not denying the thisness of events it
leaves room for ultimate contingency about what the laws of nature are and,
arguably, makes it possible to understand how a series of causally related
events can constitute a temporal series without assuming some exogenous notion
of temporality. By placing emphasis on the suchness of events and therefore the
intimate connection of causation with laws of nature, it makes it possible to
explain how laws support counterfactuals without confusing laws with accidental
generalizations.
I hope that’s helpful; in any case, it’s all I’m going to say on the subject
unless you provide further prodding or questions. By the way, in case you’re
wondering, I have not forgotten that I started out to talk about backward
causation.
3. Back to backward causation! What would it be like for a later event to cause an
earlier one? We can say that the earlier one would depend on the occurrence of
the later. But for that to be the case, it seems that the later event would
have to be “settled”, certain to occur, while the earlier was “unsettled” or at
least not certain to occur in the absence of the occurrence of the later. We
can make some sense of this, I think. For example, quantum probabilities and
all, it may be certain or settled, that the sun will rise tomorrow. There may
be no possible sequence of events after the present that could prevent tomorrow’s
sunrise. Nonetheless, there are some events which will (or won’t) occur between
now and tomorrow’s sunrise that are not yet settled. Since they’re not certain
while tomorrow’s sunrise is, it can’t be the case that tomorrow’s sunrise
depends on their occurrence. So, just maybe, it could be that their occurrence
depends on the occurrence of tomorrow’s sunrise.
Unfortunately, I can’t be very enthusiastic about this suggestion. It’s not
clear what the claim that the earlier events in question are uncertain comes to
if they are in fact dependent upon the occurrence of a later event which is
(already) certain or settled. Perhaps it means that nothing in their past is
sufficient to settle how they will turn out. But that’s unclear: Something in
their past is sufficient to bring it about that the sun will rise tomorrow upon
which the occurrence of these events is supposed to depend. So, why not just
say that those same past events are sufficient to determine the occurrence of
the supposedly as-yet-unsettled events? What need is there to refer to the
future sunrise?
One possible exit occurs to me. Suppose that tomorrow’s sunrise is certain but
not because of or just because of any past events. Its certainty, too, is a
result of backward causation as is the certainty of the future events upon
which it depends, ad infinitum. (It’s hard to be happy with this if only
because it looks to be entirely untestable. Testing it would require preventing
a “certain” event to see if the “uncertain” earlier event supposedly contingent
upon it did or did not follow suit.) Or, perhaps not ad infinitum if time were
cyclical, then future and past would be at, some significant distance from the
present, identical. Then, maybe, we would say that everything causes
everything. Or, that nothing causes anything. Or, that everything is self-caused
....
Another possibility, with cyclical time, might be that some macro-scale events
(including some future events) are certain while many micro-scale events are
not. Some years ago, I tried to work out an idea of this kind for dealing with
quasars.
Quasars are enormously energetic and very distant astronomical objects. They
are estimated to have more energy output than many whole galaxies. If I
remember correctly, current theory says they’re probably distant, and therefore
apparently pointlike, galaxies colliding. Leaving that aside how could you ever
come up with interesting ideas if you insisted on being too closely tied to
facts? I began to speculate.
First, given light-speed, they are not only very distant but also very ancient,
providing a view of a time when the universe was much younger than now. Where
might such enormous energies come from? One place that comes to mind is
supermassive black holes which could, in principle, swallow whole galaxies.
There’s a slight problem, though, in that supermassive black holes ought to be
relative late-comers in the history of the universe. It takes time to
accumulate the mass that enables a black hole, at anything like a respectable
rate, to swallow a galaxy. (Many astronomers think there’s a large black hole
at the center of our galaxy. But our galaxy is in no danger of being completely
swallowed for many billions of years.) Also, of course, black holes have a
reputation for being hard to exit.
But there may be a solution. One of the features of black holes is that they
have escape velocities greater than the speed of light. That suggests, if we
could somehow avoid or bypass Special Relativity’s restriction on travel faster
than light, that the gravitational attraction of a black hole might accelerate
incoming mass/energy past light-speed. In fact, it isn’t terribly difficult to
imagine a way to do that without altering any of the basic features of the
theory. The restriction comes from the fact that the Lorentz-Fitzgerald
equations yield solutions for greater than light speeds that require square
roots of negative numbers. If a physical interpretation could be found for
imaginary numbers, then Special Relativity wouldn’t rule out faster than light
travel. But if you put ftl travel and Special Relativity together, you get
things moving backward through time. What if quasars are the past
high-energy output of future supermassive black holes? Add another
assumption, that the universe as a whole is closed, i.e., that it will
eventually collapse gravitationally upon itself, and we may have something more
interesting yet. In that case, when the Big Crunch compacts the total
mass/energy of the universe to a singularity, it may all be propelled backward
through time to appear as ... the Big Bang. The universe would be finite in
space and time with no source but itself. (After working out this charming
idea, I was somewhat disappointed to discover that the sf novelist James Hogan
had also come up with it – though, as I remember, he had missed the point about
quasars.)
So far, I’ve concerned myself with backward causation along a single developing
time-line. Suppose, however that we allow a multiple-worlds interpretation such
that a present event affects a past event by splitting “the” universe into two
or more alternate future universes. This seems to have some potential for
overcoming some of the most worrisome paradoxes associated with time travel and
backward causation – like what would happen if I went back in time and killed
my grandfather (before he reproduced). It would just then be the case that I
come from one history in which my grandfather is not killed (and lives to
grandfather me) but create an alternate universe (in which, presumably, I must
then live) that doesn’t have any of my grandfather’s other descendants in it.
Promising, but probably not enough. The most thoroughly worked out version of a
multiple-world model is the Everett-Wheeler-Graham model of quantum mechanics
according to which, at every quantum transition where more than one outcome is
possible, they all occur; the universe splits into alternate editions of
itself, each expressive of one possibility for the subsequent evolution of the
whole. But with this theory, there’s no room, in a straightforward sense, for
me to bring about an alternative universe-history. If it was possible at all,
it would already have happened independently – does this word make sense here? –
of any future decisions on my part to take a time-machine-propelled axe to the
roots of my genetic tree.
If we assume, on the other hand, that there might be some unactualized possible
world-histories – unactualized because nobody in an actual “later” ever gets
around to time-travelling back and doing whatever it takes to make them come
about – I think the paradoxes may (still) be hard to avoid. Suppose again my
murderous mission to deprive myself of ancestors. I show up in the past of my
own history in, say, 1920. I spend as little time as possible seeking out and
executing my grandfather. Suppose it is admittted that this creates an
alternate history with me and without my grandfather. Still, what about the
interval before I get around to killing him? Am I “there” on the time-line that
eventually will lead to my birth or not? If I am, then that somewhat anomalous
appearance is already settled before I get around to time-traveling (or being
born, for that matter). If, alternatively, it is not on “my” time line that I
put in an appearance, how do I get to an unactualized time-line, one that, ex
hypothesi, will not be actualized unless some event occurs on the time-line
leading up to me? In this version, it looks like we may be able to admit some
unactualized as well as actualized world-histories but won’t have any way to
attribute the actualization of some among them to time-traveling or
backward-causing activities of future denizens of any of the world-histories.
If it seems that I’ve been mostly negative about the possibilities of backward
causation, I’m sorry. I just haven’t been able to think of a way to make it
work. Crowley nearly did, though. Somewhere he writes about performing a ritual
to obtain a certain sum of money. The next day he receives it in the mail, but,
of course, it must have been sent before he performed the ritual. Did he
conclude that the ritual was irrelevant? (The money would have arrived whether
he’d done it or not.) No, he suggests that the same set of causes which made
him a magician and a doer of his True Will were also at work in bringing it
about that the money was sent. In order for it to have been the case that
Crowley was not the kind of person who would perform the ritual, then the
complex causal background which also affected the sender of the money would
have had to be different and, hence (?) the money would not have been sent.
This suggested explanation, however, seems to require one of two things,
neither of which may be very plausible. First, it might work if a causal role
were attributed to connections of meaning or significance (independently of
creatures acting on their perceptions/conceptions thereof). Second, it might
work if someone – maybe Crowley, maybe his benefactor, maybe some combination –
had really good psychic access to and calculational powers regarding current
states of the universe, all the calculation, of course, going on outside
conscious awareness. On that possibility, see the next section.
[Interlude (inserted several days after the above was written): A couple of
further possibilities about backward causation have occurred to me. Mostly, I’ve
spoken about temporal relations being determined by causal relations so that a
cause couldn’t be preceded by its effect. But things may not be so neat, even
if that analysis is basically right. Real events are located at the
intersections of multifarious chains of causal connectedness, of conditions and
consequences. What if the causal chains of which an event is part don’t
unambiguously place it at some date, if some of its connections to other events
place it in a past while others place it in a future? Would that be a way of
saying that a future event could cause a past event? Or, perhaps, that in the
case of such events, whether it is future or past is ambiguous or
indeterminate? Or, just that there’s something fundamentally wrong with the
universe?
A second possibility has to do with Bell-like non-localities. It is my
understanding that experimental arrangements can be contrived which exhibit
non-localities in time as well as in space. (I’m not very clear on this.
Perhaps I should go look at some of the source material.) So, a measurement
occurring now can be correlated with, say, a photon-emission event multiple
millions of years ago in Andromeda. This leads to some of the most exotic forms
of quantum weirdness. It can even be suggested (because it has been) that the
Big Bang occurred because, among the superposed states of the universal
psi-function was one (at least) that led to beings who would make the
measurements necessary to collapse the wave-function. (?!)]
Before then, though, I’d like to pose a puzzle about determinism. (I’m sure you
haven’t had enough yet.) As I understand it, determinism is unlikely on the
evidence although a few physicists still hold out for a deterministic
interpretation of quantum mechanics. What it means, as you know, is that for
every event there is a sufficient cause – a prior event sufficient for bringing
about just that successor. It is easy and natural to suppose that if that’s
true, there would be no “real” meaning to time. The “future” unrolls inexorably
from the past, and nothing ever could have been different. The image one gets
is of a “block-universe,” a four-dimensional solid within which we somehow
suffer from a limited viewpoint that makes it appear that the past is fixed
while the future is not.
But that easy and natural assumption is a mistake. Determinism is compatible
with ascribing a real directionality to time. Determinism only says that the
future is fixed by the present or the past, not that the past is also fixed by
the present or the future. Even if it is the case that, given the occurrence of
A now, only C is possible in the future, it does not follow that given C in the
future, only A is possible now. For it might be the case that C could have been
sufficiently caused either by A or B. Then there could be multiple possible
histories leading to the same future without it ever being the case that any
future event was not fully determined by its past. So, an asymmetry emerges
between prediction and retrodiction (or explanation).
The puzzle arises here. For the hypothetical universe I’ve described seems
symmetrical with an indeterministic one. The only difference (that there has to
be) is that indeterminism locates multiple possibilities in the future but not
(necessarily) in the past, while determinism may locate multiple possibilities
in the past, not in the future. If the two cases are symmetrical, why is one a
case of determinism while the other is a case of indeterminism? Or, if they’re
not symmetrical, what’s the difference?
4. It seems that I understated the power of the brain as a representational/computational system. Per Churchland, whom I was citing:
To appreciate just how great is the conceptual variety that awaits us, consider the following numbers. With a total of ... 1011 neurons with an average of at least 103 connections each, the brain has something like 1014 weights to play with. If we conservatively suppose that each weight admits of only 10 possible values, the total number of distinct possible configurations of synaptic weights (that is, distinct possible positions in weight space) is ... 10100,000,000,000,000! [A Neurocomputational Perspective, p. 190.]
In case it isn’t already obvious,
that’s a lot. More specifically, not only is it the case that if some way of inputting
it were available, the whole brain could store a 4-D position for every
fundamental particle in the universe, a million or so neurons with their
normal complement of connections and activation strengths could do it as well
with room left over! Of course, it’d be a bitch trying to access and
retrieve any particular bit you wanted. (The calculation for the italicized
claim includes some seat-of-the-pants approximations to simplify things because
I had to deal with numbers larger than my calculator is comfortable with.
However, assuming that we have 20 billion light-year distances to worry about,
the million-neuron network should be able to handle representing it on a scale
graduated to the Planck length and with the time scale graduated down to quadrillionths
of seconds. (And that’s making very weak assumptions about how much
informational compression a weight-space can handle. Realistically, a million
cells should be able to handle a lot more.) But if that isn’t close enough,
just add more neurons! A million are only a hundred-thousandth of the total
supply!
If the access and retrieval problem could be solved and if the fundamental laws
of physics aren’t too complicated, the brain might even be capable of handling
a universe-sized problem of representing evolution over time. Unfortunately,
there are two major qualifications on this: First, it couldn’t come close to
doing it in real time as fast as the universe itself “solves” the problem, so
it seems a lot more efficient to just try to project broad trends and otherwise
wait to see the universe’s “solution”. Second, even if speed weren’t a concern,
the brain might be able to handle a universe-sized problem – but not ours! In
order to handle ours, it would have to include a representation of itself representing
the universe containing itself representing the universe containing itself ....
In other words, adequate representation/computation of our universe would take
an infinitely rich model.
Despite these limits, there is still an awesome amount of processing power
there and it may very well be able to handle fairly complicated problems about
the near future.
This may seem reminiscent of New Age-y/tabloid science-y ads that say the
average person only uses 10% of his brain(!!!). If YOU would like to learn how
to tap into your VAST UNUSED RESOURCES, just send $10. Cash or money order
only, please. Upon receiving your order, I send back a postcard: “See ... you’re
getting smarter already. Aren’t you? P.S. You can be confident that your
$10 “unused resource” is now being devoted to more important ends.”
You may have guessed I don’t think much of such claims. There’s a naive
evolutionary argument that such claims must be mistaken that runs something
like this: What evolution by natural selection does is provide slight
advantages – advantages in likely reproductive success as compared to existing
rivals (for food, mates, territory, etc.). There may be an evolutionary reason
for humans to be slightly more cognitively competent/versatile than the nearest
competitors in their econiche; there isn’t any reason, though, to expect
evolutionary processes to endow a creature with the ability to handle quantum
mechanics when it’s still struggling with rubbing two sticks together. It’s a
naive argument because it assumes in effect that evolution is both myopic, in
that it doesn’t look ahead to any long-term results, and foresightful, in that
it only provides capacities to the extent that they have an immediate pay-off
in terms of gene-projection into future generations.
The standard line among biologists is that evolution has no foresight at all;
what gets produced and preserved is a function of genetic math operating upon
whatever variety sexual mixing and mutation toss up. If that’s correct, then
there’s no trouble explaining how large brains, if they happened to get
produced by the genetic lottery, could get preserved: it would only have to be
the case that they conferred more (immediate) reproductive advantages than
liabilities. (Or, they could be approximately neutral if they appeared in a
sufficiently small and reproductively isolated population.) The fact that large
brains also conferred capacities that early hominids had no use for wouldn’t
keep them from being preserved by the evolutionary process.
In fact, I think that it’s not clear that evolution has no foresight. The claim
rests on contrasting evolutionary processes with paradigmatic cases of
possession of foresight – i.e., beings like ourselves. But it seems to me that
the best explanation for human foresight-capacities itself relies upon the
brain’s capacity to run parallel simulations subject to “mutation” and natural
selection. (See Dennett’s Content and Consciousness and Calvin’s Cerebral
Symphony.) If so, the case for evolution’s lack of foresight can only be
made by either pointing out further disanalogies (which nobody seems to have
done) or by providing a better model for human foresight (which, in my opinion,
nobody has done). So, I wouldn’t be inclined to completely rule out the possibility
that evolution “had in mind” that hominids acquire intellectual capacities far
in excess of their immediate needs subject, of course, to the constraint that
large brains shouldn’t actually be a significant immediate disadvantage.
Nonetheless, even if evolutionary biology doesn’t require immediate advantages
to explain (the scope of) human intellect, it would certainly be more
comfortable if they were available. My guess is that we needed the
overwhelming computational power of the brain in order to support ordinary
human intelligence. The problem is that the brain doesn’t come with an owner’s
manual. If it did, ordinary human intelligence might be supported by a much
smaller brain. As it is, though, the average person couldn’t achieve average
intelligence unless the brain were so powerful that random walks through
neuro-space were likely to turn up (relatively) efficient ways of doing things.
Still, the Weekly World News isn’t all wrong. (The stories they publish suggest
their editors may think their readers only use 10% of their brains, though.)
The number of possible brain-states and pathways is enormous, far more than any
human or even all humans have ever explored. The fact that, in exploring it as
modestly as we have, we’ve found enough relatively efficient practices to
function at the level we do makes it likely that efficient practices aren’t
that hard to find, aren’t terribly rare in neuro-space. And that suggests that
the pay-off for neuro-exploration may be relatively high. Most people, though,
never try. They get to be “good enough” and stop. Occasionally, they may see a
need to upgrade their skills for their job or profession, but they just apply
the same old techniques they used the last time they tried to learn something,
together with stupid/stupifying messages to themselves about how much trouble
they have learning things or learning particular kinds of things. (“I’m no good
at math.” “I just can’t remember things.” “I can’t handle technical material.” “My
brain just doesn’t work that way.” Ad Nauseum.)
When I started this, I imagined that I might include more. I started a list of
things to include and just the preliminary notes ran to over a page. And the
longer the list got, the more I thought of to add to it. This letter would
never reach you in anything like a reasonable time. (Then, what’s a reasonable
time when you’re dealing with immortals?)
Hope to hear from you soon,
Rob
P.S. Though addressed to you, this letter is really intended for everyone. (You
know who “everyone” is.)
