Friday, October 8, 2021

Born Knowing: Imprinting and the Origins of Knowledge

More than any of the other books that I've discussed on this blog, this new book by Giorgio Vallortigara is almost purely scientific. Although it is very short, it describes in some detail the research that he and others have conducted on early cognition in animals. The primary animals used in this type of research are newly-hatched chicks, because they are readily available and easy to test. Also, because they can be tested at a very young age, they are better-suited than humans or most mammals for distinguishing inborn patterns of behavior from those influenced by experience. Much of the challenge in this type of research is in constructing tests which clearly indicate the origins of specific chick behavior. For this reason, the discussion is quite dry and logical, and therefore the book, though extremely well-written, would not appeal to most readers.

I won't attempt to describe all of the tests in detail, because I'm mainly interested in the general findings. The most general finding is that infant animals are primed at birth to respond to specific stimuli, and that their attention to those stimuli is inconstant and changes as they develop. The earliest stimuli that chicks pay attention to relate to animacy, specifically whether an object has a face, is self-propelled and moves with biological motion. Of course, this mainly concerns the identification of the mother and siblings and is extremely important in early life. At this point in the research, it is not known whether this type of knowledge is encoded at the level of single neurons or within complex circuits of several neurons.

One interesting experiment involves the presentation to chicks of drawings of geometrically impossible objects, such as one conceived by Roger Penrose, along with similar geometrical objects that exist in nature. The chicks prefer the possible objects. According to Vallortigara, "Simply, during the course of evolutionary history, natural selection has promoted the incorporation into the nervous systems of certain statistical regularities that are typical of visual scenes in the natural environment." Another discovery with chicks, which was made fifty years ago, is that they automatically make visual choices as if light always comes from above, even when it doesn't. Human visual perception is similarly influenced by drawings, depending on how the shading is applied. Other experiments indicate that chicks engage in rudimentary thinking without language. This includes a rough way of performing addition and subtraction, along with a basic understanding of geometry.

Vallortigara is cautious about claiming that chicks or other animals have innate knowledge in these areas, partly because it is difficult to know exactly when experience begins in most species. However, he is unapologetic in claiming that there are no such things as "higher" and "lower" organisms, since all organisms are products of natural selection that applies equally to all organisms. In fact, chick research indicates that the rough cognitive plan of chicks isn't much different from the rough cognitive plan of humans.

I find these ideas interesting, because they relate to some of the ideas that I've expressed on this blog. As I said some time ago, humans are quite similar to chipmunks. These ideas also apply to what I consider to be some errors that have occurred in the history of ideas. Because, as animals, we prefer beauty and simplicity, we tend to use them inappropriately when we describe reality. Thus, for example, Occam's Razor may technically be incorrect when reality is actually very messy and complex, such as in the case of quantum mechanics. The fact that we prefer to keep things simple or aesthetically pleasing doesn't mean that reality is simple or aesthetically pleasing. This point, of course, was brought up earlier by Sabine Hossenfelder. Another bad idea, which I've been thinking about more recently, is the nature of language. Until recently, most philosophers thought that thinking requires language, and it is now empirically clear that that is not the case. It appears to me that many mathematically-minded thinkers are completely incorrect if they think that the universe is a mathematical entity. Specifically, Bertrand Russell once thought that he could completely explain the world by starting with logical notation and using it to generate all of mathematics. The actual situation seems to be that animals evolved to use mathematical conceptions – unconsciously for the most part – purely as a matter of survival. This suggests that mathematics is not identical with nature, i.e., Bertrand Russell was wrong, and, for that matter, so was Plato. Another mistake along these lines occurred in economics when the rational agent concept became widely adopted. In that instance, economists preferred the simplicity of the theory, though it was never an accurate description of reality. I should also note that one of the difficulties in developing AI is that computer scientists tend to assume that the human model is the best one to follow. It probably isn't, but, on the other hand, it would be hard for computer scientists to come up with something better than billions of years of evolution did.

I don't think that most of my readers would find this book enjoyable to read, but I think it is an excellent entry point for discovering much of the faulty reasoning that passes for knowledge and wins awards. Language and mathematics are perhaps the best tools that we have at our disposal, but one must be wary of their animal provenance.

No comments:

Post a Comment