ฝัง
- เผยแพร่เมื่อ 10 ก.พ. 2025
- The first episode in a series looking at the incredible contributions of JJ Gibson to the study of perception and action. What is direct perception? What are invariants? How do these ideas form the foundation for ideas like the constraints led approach and ecological dynamics?
Articles:
The Ecological Approach to Visual Perception
The Senses Considered as Perceptual Systems
Perception with an Eye for Motion
More information:
perceptionactio...
My Research Gate Page (pdfs of my articles)
My ASU Web page
Podcast Facebook page (videos, pics, etc)
Subscribe in iOS/Apple
Subscribe in Anroid/Google
Support the podcast and receive bonus content
Credits:
The Flamin' Groovies - Shake Some Action
Mark Lanegan - Saint Louis Elegy
via freemusicarchive.org and jamendo.com
Wow, great to find someone posting on JJ! "Ask not what's inside your head, rather ask what your head's inside."
It is precisely the dependency of perception on computation that makes context crucial. One needs to learn the computations that allows one to parse an unstructured stimulus. (Performing a skill is not just about doing the movements, it’s about knowing how to perceive the problem scenario). It’s therefore a fallacy to jump from “computations being necessary” to then conclude that “context is therefore irrelevant”. It’s the opposite!Computations supports the idea that context is important. You need to compute on the RIGHT info from the environment.
Hum, very good. Thanks
Great stuff!
Interesting series.
The ability to focus on higher order invariants from stimuli requires computations like attention, knowing what invariants to focus on, retrieving the knowledge of those invariants from memory, etc. Clearly, a beginner cannot perceive those invariants, so they have to go through the learning process of “installing” the ability to recognize that invariant in their memory/knowledge. Then they must remember; in other words, they have to keep practicing to keep this ability sharp.
A visual processing AI can’t perceive these invariants as well as humans do because it doesn’t have rich knowledge. A camera with no brain can’t “perceive” anything. It is impossible to “directly perceive” the stimulus in the environment without a mind to pick out the important invariants.
This rationalisation is simply how you see the world. Direct perception explains the same things as indirect perception does, just simpler.
We don't need computation to explain perception.
Btw, there are mechanical cameras that don't need to process anything. For visual perception to occur, you only need consciousness and light-sensitive particles.
Gibson is closer to ancient philosophers than the modern BS of Descartes and Darwinism.
Gibson is God