Raúl
Moderator
Moderator
| Posts: 590 |  |
Karma: 10
|
About the Sensorimotor Approach - 2010/02/25 12:01
I'd like to start here a discussion about the sensorimotor approach, and I'd like to start with the comment posted by Trung Doan about the article How to make a robot that feels by Kevin O'Regan.
Here's a copy of Trung's comment:
I had fun taking the sensorimotor approach to extreme examples to see if I can learn something. One extreme is a car's anti-lock braking system. The ABS has sensors and motors, therefore it can potentially have what O'Regan calls "raw feels". But this example is not instructive, probably because a car is not a living thing. A more instructive example is that of very simple creatures, such as a mosquito or even a multi-celled organism. Here, the reason for what O'Regan calls "grabbiness" stands out clearly - survival. The feel of hunger must grab a mosquito so that it seeks food, and the approaching hand must grab its attention so that it flees, to survive.
Examples of this second kind also make me think twice about a distinction O'Regan makes in his article, "Explaining what people say about sensory qualia". As I understand it, O'Regan says that physical laws are built into the way the five external senses (eyes, skin, etc.) work, and detecting these laws in action is how feels, such as the feel of redness, arise. With autonomic functions (controlling heartbeats, chemical components in various bodily fluids, etc.) we don't feel our viscera, and with thought functions (imagining, thinking, remembering, etc.) we don't feel like something to have thought functions, because physical laws are not involved or are not prominent.
Does the mosquito make the above distinction? I would guess that it does not - "Internal hunger or external threat, if it's important to survival, then grab me", says the hypothetical mosquito. And we humans do have a feel about what it is like to feel hungry. Physical laws apply not just to the external senses but to internal sensors too. That we humans don't have recognisable feels about most autonomic functions, may just be an incidental consequence of the way we are built and evolved, rather than something more fundamental.
The brain's perception process, which allows us to perceive that we are thinking, may turn out to also follow certain physical laws, à la the sensorimotor approach . Here, the "sensor" part may be the perception machinery itself, and the "motor" part may be the rising up or dying down of signals from coalitions of neurons. If so, there is no fundamental reason why there cannot be feels about thought functions.
The upshot of all this, as far as the machine consciousness endeavour is concerned, is this: There may be no fundamental reasons why we can't build robots that not only have feels about the outside world, like humans do, but also feels about their inner physical and mental processes.
No fundamental reasons, but there are practical constraints. Attention is a scarce resource, neither a creature nor a robot can survive or function very well if it pays equal attention to and is aware of too many things at the same time.
Trung Doan
Post edited by: Raúl, at: 2010/02/25 12:03
Raúl Arrabales Moreno. conscious-robots.com/raul |
miel
continental
Forums... 
Conscious Robots RSS Feed
Follow us
