Testing Structural Equation Models, Vol. 154 Book

This was a fascinating read. Gibson walks you through a novel account of visual perception with rich, detailed explanations and a compelling narrative flow. I won't try to summarize the theory here, and while much of it is intuitively plausible I haven't engaged with the evidence or dug into the details nearly enough to have an opinion on its correctness, but some aspects I particularly appreciated:

* Gibson starts his framework on the level of analysis where the high level features of perception that we care about can be explained, rather than framing things in terms of physics.
* Gibson shows a strong handle on the problems of reifying abstractions, and uses that to drive many of the differences between his theory and the mainstream ones. For example, he drives home the point that planes, lines, and points are abstractions over surfaces and their relationships, and that we should not start by trying to reduce the latter to the former.
* Gibson neither shies away from philosophical approaches where appropriate nor substitutes philosophy for science.
* Gibson has engaged with the mainstream theories as an expert and clearly understands them well, using that understanding to inform his approach.
* The theory is radically different in many ways that, if nothing else, will help remind me that the current paradigms aren't the only way to consider these questions and that they oughtn't be taken for granted (even if they are in fact true and useful). In particular (and this applies to more than just perception), the book serves as an excellent example of the possibility of doing science outside of a reductionist framework.
* Gibson's framework has some really tantalizing possibilities for interesting integrations with my framework of concept formation and other abstract awareness.
* Gibson's approach is profoundly empirical. Not only is the exposition replete with references to existing research, it also makes explicit numerous explicit conclusions about what kinds of experiments should be fruitful and what kinds of evidence we ought to see. The theory is not just a reframing of existing theories with the same ultimate results. I don't know if Gibson is right, but I know I can check

That being said, I did have some concerns. Primarily, the framework presented radically diverges from how I've learned and thought about perception, and as a result even with the detailed descriptions and caveats and contrasts with traditional ideas I had a hard time fully integrating the ideas and getting a sense of the larger structure. This is of course unfair, and probably not something that could be done any better in a single book, but still detracted for me. I think the only real solution here is to dive into the literature that he built his ideas from and try to see how I could build them myself, as unfortunately Gibson died shortly after publication and I've heard some very mixed things about those who came after. I also found the degree of his dismissiveness of those who accept the mainstream theories (in fairness, including his former self) both much stronger than was justified by his presentation and indicative of a potential failure to take into account the ways in which those theories might inform his, even if they're wrong.