Before World War II, Weiss was Roger Sperry's thesis adviser.

As chairman of the Biology Division of the National Research Council in 1951, Weiss restructured the subcategorization of biology into fields still in use today - Molecular, Cellular, Genetic, Developmental, Regulatory, and Group and Environmental. The new categorization reflected the outstanding successes then happening in Molecular Biology, which would lead to the discovery by Francis Crick and James Watson of the genetic code in DNA in 1953.

But Weiss was wary of a new form of biological reductionism. Logical reduction of all the laws of nature to the laws of physics had been popular among philosophers for decades. Philosophers of biology began arguing that all life could be reduced to molecular biology - although most defended the idea that the emergence of biology from chemistry and physics had produced new laws of biology.

Weiss wanted biology to be the study of "systems" as opposed to "machines." He neatly characterized the difference:

In the system, the structure of the whole determines the operation of the parts; in the machine, the operation of the parts determines the outcome.

(Beyond Reductionism. Beacon Press (1969), p.13)

Weiss participated in the 1968 Alpbach Symposium: New Perspectives on the Life Sciences organized by Arthur Koestler. The attending scientists included many who were critical of the neo-Darwinian orthodoxy as providing all the answers to the many problems posed by the phenomena of evolution.

Weiss and other attendees, including among others Ludwig von Bertalanffy, Jerome Bruner, Viktor Frankl, Friedrich Hayek, Jean Piaget, and C. H. Waddington, thought it likely that further emergent hierarchical levels "over and above" the molecular level would be needed to fully explain biology, and that these levels were unlikely to be deterministic.

Although macroscopic determinism has been confirmed by studies of macroscopic objects, Weiss argued that extrapolation downward to microscopic determinism has not been justified. Indeed, quantum physics shows that there is irreducible indeterminacy at the level of microscopic physics. Weiss formulated a principle that he called determinacy in the gross despite demonstrable indeterminacy in the small. This is essentially our concept of "adequate" determinism.

There is actually no strict determinism at any "level" of the physical world. Determinism is an abstract theoretical ideal that simplifies logical and mathematical methods. The macroscopic "determinism" we see is the consequence of averaging over extremely large numbers of microscopic particles. Determinism is an "emergent" property that appears when large numbers of atoms and molecules get together. The emergent "laws of nature" are always only statistical laws.

The Alpbach Symposium proceedings were published in 1969 with the provocative title Beyond Reductionism. The first contribution was Paul Weiss's article The living system: determinism stratified, in which he said:

My prime object here is to document that certain basic controversies about the nature of organisms and living processes, which have for long failed to melt away in the heat of argument (e.g., reductionism versus holism), readily vanish in the light of realistic studies of the actual phenomena, described in language uncontaminated by preconceptions. In this light (i) the principle of hierarchic order in living nature reveals itself as a demonstrable descriptive fact, regardless of the philosophical connotations that it may carry. And further (2) the necessity becomes compelling to accept organic entities as systems subject to network dynamics in the sense of modern systems theory, rather than as bundles of micro-precisely programmed linear chain reactions. A strictly mechanistic, machine-like, notion of the nature of living organisms presupposes a high degree of precision in the spatial and chronological programme according to which the innumerable concurrent component chains are composed and arrayed — a conception later amplified, but in no way altered, by letting the programme include equally pre-programmed checking and spare mechanisms to keep the bunch of separate processes from falling apart when faced with the fortuitous fluctuations of the outer world.

...

The explanation by Jacques Loeb (1918) of animal behaviour in terms of rigidly concatenated reflex sequences, and particularly his proposition of tropisms as paradigms of a precise cause-effect machine principle in organisms, epitomizes that kind of mechanistic preconception. His thesis had, however, two serious flaws. Not only had that particular brand of naively mechanistic thinking already become outdated in physics, but studies of the actual behaviour of animals in goal-directed or other forms of directional performances showed none of the presumed stereotypism in the manner in which animals attained their objectives. True, the beginning and end of a behavioural act could often be unequivocally correlated with a vectorial cue from the environment; but the execution of the given act was found to be so variable and indeed unique in detail from case to case and from instance to instance, that it was gratuitous to maintain that the attainment of essentially the same result regardless of the variety of approaches is simply the blind outcome of a chain of seriated steps appropriately pre-designed by evolution to lead to that end. In other words, organisms are not puppets operated by environmental strings; moreover, the analogy is meaningless, anyhow, if one remembers that the "environment" that pulls the strings of puppets in proper order is as often as not another organism — the puppeteer with his brain or at least some machine contrived by a brain.

(Beyond Reductionism. Beacon Press (1969), pp.4-5)

Of all the features of our subjective knowledge of our brain activities for which we want science to produce an objective record, one of the most hotly debated ones has been the aspect of "freedom of decision" or "free will". The issue has been argued almost entirely on philosophic grounds. That is a domain which, as I stated in the introduction, I feel reluctant to enter. However, since some of the philosophical discussions have hinged on the interpretation of certain unfavourable verdicts pronounced by science, it seems indicated in this place to re-examine briefly the tenability of the respective positions. The way I see it, looking from the outside in, the problem of free will has been treated in general as a corollary of the problem of determinism, and the problem of determinism, conversely, has been laid at the doorstep of science for an opinion. As long as science keeps on presenting nature as a micromechanical precision machinery run by strict causality, the concession of any degree of freedom of choice to any natural phenomenon would be inadmissible by the code of that brand of science, and hence, would have to be denied to all processes of nature, including human brain functions. One would then be forced to adopt the alternative of ascribing "free will" to the intervention of extra- or supernatural powers.

Determinism, stratified

To me both of these extreme positions seem to be untenable. The flaw lies in equating science with the doctrine of micro-precise causality, or as I shall call it in the following, "micro-determinism". This brings me to the major lesson to which I have been building up in this paper and which I have anticipated in the title as "Determinism Stratified", which is precisely what the study of living nature teaches us.

There may be philosophers or theologians who derive comfort from the idea of a Laplacian universe made up of a mosaic of discrete particles, operating by laws of micro-causality. I submit, however, that modern science cannot deliver such a picture in good faith, least of all life science; and since all science is the product of human brains and brains are living systems, it is quite likely that this abrogation of scientific rationale for micro-causality applies to science in general.

Scientifically, the term "determined" can only mean "determinable"; and similarly, "indeterminacy", whether in the sense of Heisenberg or in the way I shall use the term, can only mean "indeterminability". The scientific concept of "determinability" is of decidedly empirical origin. As we observe a given macrosample of the Universe over a given stretch of time, we note certain un- equivocal correlations between configurations of its content at the beginning and at the end of that period of change. If then we find those correlations consistently verified, we set them up as "laws", from which to extrapolate future changes with a sense of certainty. As our primary experience in this operation has only correlated macro-samples with macro-samples, predictability based on it can likewise be no finer than macroscopic. So, legitimately, we could only speak of "macrodeterminacy".

The concept of "microdeterminacy" is then derived secondarily by a hypothetical downward extension — "atomization", as it were — of empirical "macrodeterminacy". That hypothesis submits that one would observe the same high degree of consistency of correlation from beginning to end that had been ascertained for the macrosample to hold true for every one of those fractional samples. In other words, the structure of the well-defined macro-change would be simply a composite of the mosaic of micro-changes assumed to be equally well-defined, even if not necessarily determinable.

This view is demonstrably untenable in its application to living systems. We have recognized the state and changes of such systems as being conservatively invariant over a given period, and hence predictable, without a correspondingly invariant micro-mosaic of the component processes. We had to conclude, therefore, that the patterned structure of the dynamics of the system as a whole "co-ordinates" the activities of the constituents. In atomistic microdeterministic terms, this "co-ordination" would have to be expressed as follows: Since any movement or other change of any part of the system deforms the structure of the whole complex, the fact that the system as a whole tends to retain its integral configuration implies that every change of any one part affects the interactions among the rest of the population in such a way as to yield a net countervailing resultant; and this for every single part. Couched in anthropomorphic language, this would signify that at all times every part "knows" the stations and activities of every other part and "responds" to any excursions and disturbances of the collective equilibrium as if it also "knew" just precisely how best to maintain the integrity of the whole system in concert with the other constituents. Although rarely expressed so bluntly, much of this imagery lurks behind such equally anthropomorphic terms as "organizers", regulators", "control mechanisms" and the like, which particularists have had to invoke in order to fill the information gap between what one can learn from isolated elements and a valid description of group behaviour.

The Boltzmann theorem and thermodynamics have realistically by-passed that gap by confining safe statements about macrorelations to macrosamples only. They relate unequivocally the average state of a system at time t₁ to its average state at time t₂, but realize that tracing an individual molecule through that course is not only not feasible but would be scientifically totally uninteresting and inconsequential; for it would in each individual instance and instant be of nonrecurrent uniqueness, hence valueless for any detailed predictability of future micro-events. If physics has had the sense of realism to divorce itself from microdeterminism on the molecular level, there seems to be no reason why the life sciences, faced with the fundamental similitude between the arguments for the renunciation of molecular microdeterminacy in both thermodynamics and systems dynamics, should not follow suit and adopt "macrodeterminacy" regardless of whether or not the behaviour of a system as a whole is reducible to a stereotyped performance by a fixed array of pre-programmed micro-robots. Since experience has positively shown such unequivocal macrorelation to exist on various supramolecular levels of the hierarchy of living processes in the documented absence of componental microdeterminacy, we evidently must let such positive scientific insights prevail over sheer conjectures and preconceptions, however cherished and ingrained in our traditional thinking they may be.

Macrodeterminacy

[In embryology] one has even got to referring to the process which changes a cell from its originally "indeterminate'' and multivalent condition into one definitely committed for a given fate, as the process of "determination". The same principle as illustrated here for development repeats itself at all orders of magnitude; let us review, for instance, the establishment of organelle structures within the cell, or of the finer sub-structures within organelles.

A mitochondrion, in being reconstituted after fragmentation, will order lipid molecules into the characteristic lamellar configuration ..., but the new position will not be a precise replica if the former one, nor do the lipid molecules in the pool know which ones will be recruited. Likewise on the next lower level, the enzymes in regularly seriated clusters, which are to dot the new mitochondrial lamellae, do not know their final arrangements until they are in place; and so forth.

This principle is close to our notion of merely "adequate" determinism in the macroscopic world

I could go on to confirm the validity of this principle of determinacy in the gross despite demonstrable indeterminacy in the small for practically any level and area of the life sciences. In order to take account of this hierarchical repetitiveness, I have suggested the simile of "grain size" of determinacy as an empirical measurement for the degree of definition and predictability at any given level. The mosaic of organ rudiments mapped out in the early embryo, for instance, is very "coarse-grained", whereas the mosaic of genes in the chromosome is far more "fine-grained". It is immaterial at this juncture whether or not the principle is rigorous in a philosophical sense. What counts is that, scientifically speaking, it is as realistic and logical a proposition as we can deduce from the accessible facts.

As you will note, one could turn this renunciation of the primacy of microdeterminacy into a positive scientific declaration in favour of the existence of "free will". I prefer to give it a more restrained interpretation, for it really implies no positive commitment. What it does, is simply to remove the spurious objections and injunctions against the scientific legitimacy of the concept of freedom of decision that have been raised from within the scientific sector, or from other camps leaning on supposedly scientific verdicts. I cannot see that science can prove free will but, on the other hand, I can see nothing in what we know in the life sciences that would contradict it on scientific grounds. To go beyond this statement would be a matter solely of private belief.

(Beyond Reductionism. Beacon Press (1969), pp. 27-32)