Re: recognition of individuals theory

Glen M. Sizemore wrote:
"Wolf Kirchmeir" <wolfekir@xxxxxxxxxxxx> wrote in message news:hccYf.778$sh3.56014@xxxxxxxxxxxxxxxxxxxxxxxx
Glen M. Sizemore wrote:
"Wolf Kirchmeir" <wolfekir@xxxxxxxxxxxx> wrote in message
[snip speculation on robot vehicles' software]
Now you could claim that, abstractly considered, what the robots are doing as comparing/matching memory and perception, but IMO that kind of language carries a variety of semantic freight that interferes with grasping what's actually going on.

Whether what the human visual system does when it recognises faces could be abstractly represented in these or similar algorithms is not at all clear to me. Perhaps Curt can comment.

Or maybe 70 years of research on stimulus control is relevant.


It would be if the robots were programmed to learn how to see. AIUI, each vehicle in the competition implemented a particular strategy of calculating its path over the designated course.IOW, the designers started with knowledge about what information needed to be acquired, and went from there. AIUI, humans and other mammals begin with a VC that preferentially responds to certain inputs, and which develops as the infant learns to see its environment (ie, to discriminate visual and correlated cues more and more subtly.) The robot vehicles were not learning machines, even though some of them apparently were programmed to update the map (==data array) they used to steer by. Updating data arrays may be needed for machine learning (IMO, that's still a debatable point), but it's not learning.

The post to which I responded had nothing to do with robotic vehicles. That was a different thread. This thread had to do with recognition of faces in humans - at least that was what the original poster asked about.

Yeah, but I brought in the robots since somewhere along the line someone (OP, IIRC) said that faces must be matched against stored images, etc. Which IMO is wrong. I brought in the robotic vehicles to show that it's possible to see with no image storing whatsoever: so there's no reason to assume that humans store face-images. It would be very costly to store the several thousand or so faces we do in fact recognise.

Question: does our ability to recognise previously familiar faces diminish as we learn new ones? Introspection suggests that speed of face recognition is also dependent on contextual cues. Is this so?
.

Relevant Pages

  • Re: recognition of individuals theory
    ... humans and other mammals begin with a VC that preferentially ... learning (IMO, that's still a debatable point), but it's not learning. ... The post to which I responded had nothing to do with robotic vehicles. ...
    (sci.cognitive)
  • Re: {40K} Tank Day
    ... more vehicles than I do. ... More an insane collector than a serious one imo:p ... With enough time and energy, you probably have the materials there to build ...
    (rec.games.miniatures.warhammer)
Loading