Anthony Leggett's response to some of the ideas to be presented by Aaron Sloman
in his invited keynote talk on 21st Jan at the 14th International Workshop on
Natural Computing, Tokyo, Japan

    https://www.natural-computing.com/iwnc-ws

Notes for his talk are available online at

https://www.cs.bham.ac.uk/research/projects/cogaff/misc/evo-devo.html

Comments by Anthony Leggett:
https://physics.illinois.edu/people/directory/profile/aleggett

Over the last two years, and in his talk at this conference, Aaron has called
attention to a previously little-remarked phenomenon, namely the development in
vertebrate eggs, with only the crudest of external physical input, not only of
enormous physical differentiation but of competences in the newly hatched chicks
which are clearly impossible to explain as the result of experience.

He has conjectured inter alia that the existing laws of physics may be
inadequate to explain this phenomenon.The following brief remarks are a
preliminary and no doubt crude attempt to put some "boundary conditions" on this
speculation.

Before attempting to address the main question,let me address briefly a
subsidiary one:Is chemistry "different"?

If I understand the relevant passages in his talk correctly,Aaron seems to be
opposing the "continuous" evolution described by the laws of physics to the
"discrete" events (presumably reactions. etc.) which are (part of) the province
of chemistry.

As a physicist,I think my reaction to that proposition is somewhat
ambivalent:there exists a very well - developed body of literature which
explains a large variety of chemical reactions in physical terms, in particular
(though not only) by the quantum-mechanical phenomenon of quantum tunnelling
(transmission through a region of space where the potential energy exceeds the
original total energy of the reacting particles), and in these calculations as
such there is no point at which any "discrete" event appears to take place.

On the other hand, one could view chemical reactions as presenting just a
special case of the much more general (and notorious) "realization" (or
conventionally "measurement") problem of quantum mechanics, namely that in order
to make contact with experimental reality one has to go over at some point from
the quantum  description in terms of co-exisitng alternative "realities" to the
classical notion that things either do or don't happen, and the terminal point
of a reaction seems as good a place as any to do so.

(There are strong and I believe credible arguments that postponing the
"realization" to a later point, such as the emission or not of a photon or the
registration of the latter by a macroscopic counter, does not affect the
predictions for the theory as regards probabilities of experimental outcomes).

So one could say that in this sense it is indeed legitimate to stress the above
opposition.

I think a useful way of approaching the main problem may be by analogy with a
rather similar one which I believe arises in connection with another area of
research (real or putative, depending on one's point of view), namely into the
class of alleged phenomena usually labelled "psychical".

(Needless to say, a major difference is that in the present case, unlike that
one, the existence of the phenomena themselves is not in dispute).

Let us imagine,purely for the sake of the present argument, that we believe, or
would like to believe. that a given psychical phenomenon is real.

Then we can raise the question:Is that phenomenon consistent with the laws of
physics as we currently know them?

(for the purposes of the thought-experiment, only, we assume that this current
knowledge is complete).

My rough-and-ready approach to answering this question classifies the relevant
phenomena into three categories, as follows:

(1)Does the phenomenon in question appear to violate the first law of
thermodynamics?

If yes,then forget it (example:most types of levitation);if no,continue.

(2)Does it appear to violate the second law?

If yes,then treat it with extreme scepticism,but be aware that the form which
the second law takes in specific circumstances may be quite subtle (prima facie
example: precognition); if no,continue.

(3)Does it apparently not conflict with either the first or second law,but
nevertheless have no obvious explanation within the more specific laws (e.g.of
electromagnetism) which characterize our current paradigm? If yes,then keep an
open mind on it (example:water-divining,and probably some forms of telepathy);
if no,then we would not normally count the phenomenon as "psychic".

Let's attempt to apply a similar triage to the phenomenon of
egg-hatching.

(1)Does it appear to violate the first law of thermodynamics? I think (and I
assume Aaron would agree) that the answer is fairly obviously no; the process
poses no problem for the conservation of the total energy of the universe.

(2)Does it appear to violate the second law? That may be a little more tricky:in
the crudest and most obvious sense the answer would seem to be no, since to an
order of magnitude the thermodynamic entropy in units of Boltzmann's constant
k_B is simply the number of molecules n, which is also (unsurprisingly) the
order of magnitude,in bits, of the information storable in the
(thermodynamically accessible) states of these molecules, and the total entropy
throughput from and to the environment is (taking the egg of a domestic
chicken,with incubation period ~3 weeks, as an example) of the order of a few
hundred or thousand times this.

However, see below.

(3)Does it have no obvious explanation within our current paradigm?

Aaron claims so,and I think makes a plausible case; I would comment however that
one of the most interesting advances over the last few decades in the area of
physics which sometimes goes under the name of "physics of complex systems" has
been the realization that quite simple rules for the evolution of a physical
system, as modelled e.g.by cellular automata, can lead to surprisingly
complicated and sophisticated outcomes.

So I suspect that most physicists would guess that there is no fundamental
difficulty regarding the evolution of a vast variety of different physical
structures (which of course is quite different from saying that we know how it
works!)

What about the appearance of various non-experience-based competences in newborn
chicks?

Suppose we focus on a simple example such as the ability to avoid a solid
obstacle.

An obvious question is: Is it possible to build a robot which can reproduce this
competence?

If so,is the amount of information that the robot needs to have consistent with
the information storage capacity of the egg, and if yes,is this so by a large
margin?

I strongly suspect that the answer to all three questions is yes (indeed, I
would be very surprised if such robots do not already exist, for military if not
for civilian applications).

As regards the second question,one would need to consider the number of
pixels,etc.,whose output would be in practice required to give a "reasonable"
definition of "solid" (in appearance) and "obstacle".

(Yes, of course it will always be possible to confuse the robot by making the
appearance of the object ambiguous, but I imagine that is also true of the
chicks; indeed, I would be surprised if there is not a subset of the
animal-psychology community who have fun doing just that).

And as regards section 3, well, as above a naive order-of-magnitude estimate of
the information storage capacity is the number of molecules n, which is of order
10^23, in this context a very large number. so even if one has to accommodate
multiple competences, it is not obvious that when the problem is formulated in
this way there is a problem with information storage capacity.

But what if (as I suspect Aaron might argue) the problem simply cannot be
formulated in these crude terms of (quantifiable) information and entropy?

What if the competence to avoid an obstacle cannot be reduced simply to the
reaction to a particular class of pixel intensity distributions, but requires
something more,something we might call "understanding"?

(this conjecture may look more plausible in more geometrically or topologically
sophisticated examples).

Then I suspect all bets are off,and I am not sure that as of now I have anything
useful to say about whether or not one would then require some major
generalization of the laws of physics to accommodate the "egg-hatching"
dilemma;one would probably need to be able to formulate the idea of
"understanding" in algorithmic terms, which I am not sure that I (or anyone)
know(s) how to do.

At any rate this whole complex of issues seems well worth puzzling about
further, and I think that Aaron deserves our thanks for raising it.

==========