The Hypothesis of Unique Constraint
The View from Oregon – 342: Friday 23 May 2025
In recent newsletters I’ve been discussing intelligence and the embodiment of intelligence in some particular form. What we might call the embodiment problem could be framed in this way: what is the optimal embodiment for intelligence? This begs the question, embodiment to what end? What are we optimizing? What is an optimally embodied intelligence supposed to do? If we can cite a particular end, then we could discuss how the end might best be attained. This considerably sharpens the question. Even if we cited several ends—human beings, after all, pursue a variety of ends—we could discuss an embodiment in which cognitive capacity is possible and which can potentially realize a variety of ends. Here, human beings seem to come out rather well, as our embodiment has allowed us to pursue to wide range of activities; our lack of specialization can be considered a virtue rather than a limitation, and our neural plasticity seems particularly suited for turning our attention to one activity and then another. And while we do not possess the eye of an eagle or the nose of a beagle, with the technology we build for ourselves we can out-perform these natural sensory capacities of more specialized species many times over.
If, on the other hand, the answer to this question is the optimization of intelligence as an end in itself, then we could understand the optimal embodiment of intelligence to be that which allows for the greatest development of intelligence, but this again begs the question. Can intelligence work on itself in splendid isolation with no particular experience of the actual world, or can intelligence develop in a way in which any particular experience of the actual world is indifferent? Since we have no counterexamples we can exhibit, with human intelligence we find ourselves in a strangely similar position to that of not knowing what the possibilities of life are in the universe.
We know only life on Earth, and similarly we know only human intelligence. Since we only know human intelligence embodied in a human body, with its particular sensory endowments, we can’t know what it’s like to have evolved with another set of sensory endowments, or whether intelligence, as a distinctive form of emergence, is a “strong” emergent in the sense of being always and only what it is. That is to say, I am suggesting that a strong emergent must take a particular form when it emerges. Another way to frame this is that emergents can be tightly constrained or only loosely constrained, and if intelligence is tightly constrained, then, if it emerges at all in the history of a given biosphere, when it does emerge it takes the form that intelligence has for us, since, if intelligence emerges at all, it emerges as it has emerged in human experience.
It might be easier to make this point by way of an analogy with life. Say that life, having the familiar properties that life on Earth has, is only possible given the particular biochemistry found in terrestrial life. If life is tightly constrained in this way (and we don’t know that it is), then life is always and only a DNA double helix with the base pairs of cytosine, guanine, adenine, and thymine. No other macromolecule will do. No other combination of base pairs will do, as, for example, with guanine, uracil, adenine, and cytosine in RNA. If, on the other hand, life is only loosely constrained, then slight variations in biochemistry might be consistent with life possessing the properties familiar to us in terrestrial life. We can see that the degree of constraint might vary from being exclusively constrained to a single form (which we could call uniquely constrained), to allowing a few variations on that form, to allowing countless variations on that form. Life, if uniquely constrained, might arise by way of a number of distinct pathways (in other newsletters I’ve discussed the many theories of the origins of life), but once that pathway converges on life, it’s locked in as a particular biochemistry. If an apparent origins of life pathway fails to converge on the unique biochemistry of life (under the hypothesis of unique constraint), then no life emerges.
Suppose intelligence is tightly constrained or even uniquely constrained in this way. There might be multiple pathways to the realization of intelligence, but if any of these pathways converge on intelligence, then it converges on one and the same intelligence. In this case, intelligence could well be indifferent to its embodiment, and whatever sensory endowments its biological body possessed would not have any special bearing on its intelligence, because, according to the hypothesis of unique constraint, intelligence is always and only one thing. If an apparent pathway to intelligence fails to converge on the one thing that we call intelligence, then no intelligence at all arises. Not an alternative intelligence, and not a limited or expanded intelligence, but no intelligence at all. Again, we can easily imagine gradations of constraint, from the uniquely constrained to the tightly and the loosely constrained.
I don’t believe that intelligence is either exclusively or tightly constrained, but to think of it in these terms, i.e., to entertain the hypothesis of unique constraint, is a useful exercise since it points to a conception of intelligence that can be contrasted to a formulation of loosely constrained intelligence. My presupposition is that embodiment matters, and it matters in particular what sensory endowments constitute the doors of perception for a given embodiment. The intelligence of an eagle is built around its vision, as the intelligence of a beagle is built around its sharp nose. The argument could be made that human intelligence is built around our hands, at least since we became bipedal and had our hands free to manipulate our environment. Human beings have excellent vision too, and a large portion of the brain is involved in sight, and this dates from before our bipedal development. As I’ve noted many times, the brain is built up on the more rudimentary brains of our ancestors, and what’s oldest and first has a certain priority. Vision shapes our Weltanschauung in fundamental ways. That being said, the rapid development of brain size unique to human beings occurred after we became bipedal, so brain mass added during this time was selected for more effective manipulation of the environment by our hands.
If, then, intelligence is only loosely constrained by embodiment, and the sensory endowment of our embodiment—and, of course, the agency afforded by our embodiment, as with human hands—this suggests that cognition shaped by embodiment is relative to embodiment. To a certain extent this must be the case, but to what extent? Is intelligence entirely reducible to sensory endowment? This is another way of stating the Peripatetic axiom that there is nothing in the mind that was not first in the senses. This has its origins in Aristotle, was given its classic formulation by Aquinas, and is especially associated with British empiricists like Locke. This isn’t a claim that I would wish to make, so the question becomes how we must modify the Peripatetic axiom in the light of loosely constrained intelligence. The obvious response is that loosely constrained intelligence is still partially constrained, and in the context of emergentism it can be argued that that partial constraint upon intelligence is some remnant of intelligence that is always and only one and the same thing whenever, wherever, and however it emerges.
Again, this is a familiar idea that I have placed in an unfamiliar context—it is the idea of a basic form of intelligence that is shared across diverse realizations and implementations of intelligence. We can think of this in evolutionary terms as the core functionality of the brain retained and perhaps elaborated in all species that verge on intelligently directed activity. We can also think of this more narrowly in theoretical terms such as the Euclidean conception of an axiom, which is common to all forms for reasoning, and a postulate, which is specific to a given domain of knowledge. Another theoretical instantiation is the idea of a science of science, which I have repeatedly discussed in these newsletters as a unifying theoretical framework for all scientific knowledge. If there is a basic form of intelligence that is shared across a variety of distinct intelligent beings, then we should be able to draw out the theoretical presuppositions of this basic form of intelligence and relate all forms of knowledge arrived at through intelligent cognition through these presuppositions, made explicit in a science of science.
The embodiment problem as discussed above is related to the problem of a taxonomy for intelligence, which I’ve also discussed in recent newsletters. If differently embodied intelligence varies across embodiments, even if it has a common core, this implies the legitimacy of kinds of intelligence, with one kind of intelligence primarily built on vision, another kind of intelligence primarily built on hearing, and yet another kind of intelligence primarily built on smell. Are these forms of intelligence relative to each other, mutually exclusive of each other, or complementary to each other? Again, my rough formulation here is inadequate to the complexity of actual embodiments and the sensory endowments. For example, pit vipers can sense infrared and use this extra sense to hunt, but that doesn’t mean that pit viper intelligence is built around its pits, much less that it is the product of these sensory organs. However, we can imagine an evolutionary scenario in which a pit viper comes to lose most of its other senses and relies almost exclusively on its pits for survival. If this reliance on infrared sensing coincided with the evolutionarily rapid development of the pit viper brain, then the added brain could be said to be primarily built around its infrared sensing ability, and this scenario would be analogous to the growth of the human brain after bipedalism. But if no appreciable brain growth were to accompany the transition to a primary reliance upon infrared sensing pits, then it seems unlikely that the intelligence of the pit viper would be entirely reconstructed on the basis of this evolutionary history. However, even here we can imagine a possible scenario in which the brain doesn’t necessarily grow, but it substantially adapts as the organism adapts to its changed environment (the same environmental change, presumably, that is driving its reliance upon infrared sensing), and the brain is so changed by this adaptation that it is effectively new after a given period of time.
Ultimately and eventually, an account of intelligence relative to embodiment will probably have to draw from both cognitive evolutionism and cognitive relativism (as discussed in last week’s newsletter) to give an adequate account of the development and taxonomy of intelligence. The account I sketched above of tightly constrained intelligence may seem a little artificial, but we have to give it its due since there seems to be something involved in cognition when it rises to a threshold that we could call intelligence that is common to all intelligence. At the same time, we need to recognize that this common core of intelligence probably has an evolutionary origin, both in the structure of the brain itself, and in the constraints placed on any organism by the environment as that organism seeks to survive and reproduce.