logo logo
Inicio arrow Foros...
Friday, 09 de January de 2009
 
 
English English  Español Español  
Próximos Eventos
10.Jun. 2009

IWANN 2009
Salamanca, Spain
11.Jun. 2009

Toward a Science of Consciousness
Hong Kong, China
14.Jun. 2009

International Workshop on Machine Consciousness
Hong Kong, China
15.Jun. 2009

The 8th IEEE International Conference on Cognitive Informatics
Hong Kong, China
22.Jun. 2009

IWINAC 2009
Santiago de Compostela, Spain
Etiquetas
Análisis Artificial Associations Científicos Conciencia Consciencia Consciencia Artificial Conscientes Consciousness Developer Documentación Documentation Español Howto Investigación Publicaciones Reviews Robotics Robots Robótica Spanish Studio VPL artificial
Destacados
Conscious-Robots.com Forum  


<< Inicio < Anterior 1 2 Siguiente > Final >>
Pensabot
Usuarios

Junior Boarder
Mensajes: 8
graphgraph
Karma: 2  
Pensabot Test - 2008/01/29 15:19 The Pensabot Test is in development phase for consciousness of a thinker robot by complexity thought levels (neuroengineering perspective)

The inspection model compare signal data with two identical measuring instruments like brain scanners (human) and interface layers of artificial mind (H-T-M model)

The test take for granted a learning states of consciousness (self existence), watching its memory flow between hierarchical-temporal spectrum deployed in neocortex-thalamus and hippocampus.

It is a complementary of the Mirror and Turing Tests.
Chris.

Blog http://pensabot.com
  El administrador ha deshabilitado la escritura de mensajes a los usuarios no registrados.
Raúl
Moderador

Moderador
Mensajes: 444
graph
Karma: 6  
Re:Pensabot Test - 2008/01/29 16:52 Hi, I'm not sure if I got it... So, you're saying that a so-called Pensabot Test is under development with the aim to test the level of consciousness of a robot, right? By neuroengineering perspective, do you mean that this test is intended to devices built with artificial neural networks?

When you say that there are two identical measuring instruments, do you mean that a reference human brain is required to run the test, i.e., the signals coming from both the human brain and artificial brain are compared under the same stimuli?

What is the H-T-M model? Something related to the thalamocortical mechanisms?

Sorry for the bunch of question! I'm just trying to understand what exactly the Pensabot Test is.
Raúl Arrabales Moreno. conscious-robots.com/raul
  El administrador ha deshabilitado la escritura de mensajes a los usuarios no registrados.
Pensabot
Usuarios

Junior Boarder
Mensajes: 8
graphgraph
Karma: 2  
Re:Pensabot Test - 2008/01/30 18:49 Raúl wrote:
...you're saying that a so-called Pensabot Test is under development with the aim to test the level of consciousness of a robot, right? By neuroengineering perspective, do you mean that this test is intended to devices built with artificial neural networks?

Exactly. Pensabot Test is an internal evaluation after external stimulation of reasoning about self existence. But, not only low-level neural network, high-level bayesian networks too.

...When you say that there are two identical measuring instruments, do you mean that a reference human brain is required to run the test, i.e., the signals coming from both the human brain and artificial brain are compared under the same stimuli?

Depending on the level of evaluation, the reference is from a child to mature brain approach. Depending on deep of evaluation, the instruments are based on EEG, ECoG or fMRI+MEG. The other side is a non-traditional cluster of artificial cognitive architecture systems or high-level multi-agent architecture (intelligent computing). Candidates must be able to connect to an interface that translates the signals to the scanner.
So, Mirror Test can be an interesting complement when open theoretical discussions about Pensabot Test.

...What is the H-T-M model? Something related to the thalamocortical mechanisms?

H-T-M is Hierarchical Temporal Memory,
see http://www.numenta.com/for-developers/education/biological-background-htm.php
Chris.

Blog http://pensabot.com
  El administrador ha deshabilitado la escritura de mensajes a los usuarios no registrados.
Raúl
Moderador

Moderador
Mensajes: 444
graph
Karma: 6  
Re:Pensabot Test - 2008/01/31 17:03 Thanks for the clarifications Chris! I am having a look to the presentation about H-T-M...

About the Pensabot Test, I have an additional question: do you think that a concrete signal pattern exists for consciousness states?

I mean, if you just look at the signal pattern that comes out from the neurobiological machinery (say, the thalamocortical complex), couldn't it be that the same signal pattern is generated by an artificial device but not producing any conscious experience???

Cheers,
Raúl.
Raúl Arrabales Moreno. conscious-robots.com/raul
  El administrador ha deshabilitado la escritura de mensajes a los usuarios no registrados.
Pensabot
Usuarios

Junior Boarder
Mensajes: 8
graphgraph
Karma: 2  
Re:Pensabot Test - 2008/01/31 22:46 Knowing that self-consciousness is one of the high-level state of the mind, scanner shows a wide regions of activity (not localized but interpreted signal) with perception of an external stimuli, or internal (lucid dream)

Knowing big parts of the functional brain architecture, then we can analyse memory flows of structure regions.

That interpretation of signals can determine high-level thought over self episodic memories and "internal mirror mechanism" in an intelligent system.
If a machine can do it, then this is basically a thinker lets its experience it in own its artificial mind.

In the future, the gap should be reduced between human consciousness and a machine consciousness. Both, machine and human brain will help research to "open a window" of a "semantic memory layer" to watch directly on mind abstract concepts, image secuences, language learnings, sounds,... Brain signals is a begining.
Chris.

Blog http://pensabot.com
  El administrador ha deshabilitado la escritura de mensajes a los usuarios no registrados.
Raúl
Moderador

Moderador
Mensajes: 444
graph
Karma: 6  
Re:Pensabot Test - 2008/02/01 11:21 If I understood well, I see a flaw in your approach. It seems that somehow you want to use the same brain signal pattern analysis both in humans and machines. But, what if the substrate of a machine brain has nothing to do with a biological brain? Signals obtained by EEG or brain imaging techniques are closely related to the biological substrate of the human brain, and I don’t see how they could be directly compared with other sort of signals taken from an artificial brain. Maybe you are talking about other kind of higher level patterns (in part derived from biological measurements) that could also be observed in artificial brains regardless of their underlying substrate? Raúl Arrabales Moreno. conscious-robots.com/raul
  El administrador ha deshabilitado la escritura de mensajes a los usuarios no registrados.
Pensabot
Usuarios

Junior Boarder
Mensajes: 8
graphgraph
Karma: 2  
Re:Pensabot Test - 2008/02/02 02:51 Thanks for the question.

Of course, It's not scanning directly to the hardware.

Pensabot Test is to androids (with hands, legs, head, behaviour, ...) Mind structure is human mimetic (neuroengineering design)

It considers an interface that allows to translate signals from a robot brain source to human brain scanner component. This component is a data adquisition and data record software to see on the scanner-screen a visual representation of an artificial brain, like human brain but more simple (for the moment). For example, superimposed signals on an MRI image facilitates comparison.

However... How to scan consciousness?

There is a research to identify common patterns of signals that represent an object on which evoked a human mind is thinking, not only visual patterns.

http://www.cmu.edu/news/archive/2008/January/jan3_justmitchell.shtml

The signals of a same object are generated in several regions of the brain. Therefore, it is necessary to continue defining a common parts of logic layers of thought translation. So, some objects belongs to own body.
The application of this technique to the autism research (absence of "internal mirror") will help to understand consciousness.

The Mirror Test with androids can help to evoke thought about own existence.
Chris.

Blog http://pensabot.com
  El administrador ha deshabilitado la escritura de mensajes a los usuarios no registrados.
Raúl
Moderador

Moderador
Mensajes: 444
graph
Karma: 6  
Re:Pensabot Test - 2008/02/02 10:53 Ok, I see it now much more clearer. So the Pensabot Test is oriented to 'neuroengineering designed conscious machines'.

I understand now why you said the mirror test would be a good complement for the Pensabot Test. I agree this could be a good trigger to evoke thoughts about oneself. At least it generally works for humans!

How to scan consciousness? That's the big question. In the line of your proposal, it could be done just by analysing neural activity patterns. This approach is quite related to the search for the NCC (Neural Correlates of Consciousness) pioneered by Francis Crick and Christof Koch. The think is that you need a really good copy of a human brain structure (as you say, human mimetic) in order to be able to apply the test to an android.

Another approach that I've seen in the literature (complementary to the analysis of neural patterns) is the analysis of behavior. Of course, the mirror test is an example. You just need to observe the behavior of the subject in front of his own reflected image. If he recognizes himself, then the subject is self-conscious. I think other behavioral tests can be designed as complements for a neural signal analysis test. Don't you think so?

Post edited by: Raúl, at: 2008/02/02 10:55
Raúl Arrabales Moreno. conscious-robots.com/raul
  El administrador ha deshabilitado la escritura de mensajes a los usuarios no registrados.
Pensabot
Usuarios

Junior Boarder
Mensajes: 8
graphgraph
Karma: 2  
Re:Pensabot Test - 2008/02/02 20:01 Yes, I do think any behavioral tests can be analyzed from cognitive psicology (external watcher) and bio-psicology (monitoring internal memory flows activity)

We don't want a big copy of human brain structure, but similar dynamics, scalable neural network size and scalable bayesian complexity.


Turing Test can evaluate if a machine dialogue like a human, to say some of himself. Maybe it is a consciousness machine or maybe not, because is difficult to measure an ambiguity.
Mirror Test can evaluate if a machine recognizes its own external anatomy image. The probability of consciousness is higher or clearly.
Pensabot Test can evaluate if a machine have a really thought of himself, because that thought is visible on memory and intentions formation. With all these tests the consciousness can be confirmed.

Post edited by: Pensabot, at: 2008/02/02 20:06
Chris.

Blog http://pensabot.com
  El administrador ha deshabilitado la escritura de mensajes a los usuarios no registrados.
Raúl
Moderador

Moderador
Mensajes: 444
graph
Karma: 6  
Re:Pensabot Test - 2008/02/14 22:02 What do you mean exactly when you say ‘memory flows’. From what you say I suspect that what you’re actually propose for the test is to inspect the memory and see whether a symbolic representation exists for the self, and it is used in reasoning, and therefore in behavior.. Am I right? Raúl Arrabales Moreno. conscious-robots.com/raul
  El administrador ha deshabilitado la escritura de mensajes a los usuarios no registrados.
<< Inicio < Anterior 1 2 Siguiente > Final >>
Titulares RSS
 Conscious Robots RSS FeedFuente RSS de Conscious Robots





¿Recuperar contraseña?
¿Quieres registrarte? Hazlo aquí
Mensajes en el Foro
 
Top! Top!