Adaugat la data: 2010-09-10
de catre: danutza
Numar pagini: 10
Tip document: .doc
Nivel: Liceu
Dimensiune: 36.5 KB
Downloads: 1
Credite: 0

Din referat: Explanation of cognition in Cognitive Science

Explanation of  cognition in Cognitive Science  

A review of Vacariu et al's explanation of cognition involved in "Toward a very idea of representation"

ABSTRACT. In this article I reproduce and examine some aspects of Vacariu et al explanation of cognition.

I conclude that their explanation of cognition is not a complete one.

I try to propose some alternative explanations toward a complete explanation of every posible cognition and to emphasize that the task of philosophy, as discipline included in Cognitive Science, is to find the necessary and sufficient conditions of every posible cognition.

In final, I try to criticize the idea that cognition can be only representational.
 
There is a cognitive arhitecture with a hierarchy of levels of existence. Vacariu et al has considered that cognition cannot be explanated by a single level. Arguments:

-the emergency of the primitives of one level from the primitives of previous level.

- bidirectionality of cognitive processes; bidirectionality, wich is for Vacariu et al one of the most important ideas, is the continuous circular causation wich exist between conceptual and neural levels: the changes wich appear at the conceptual level involve changes at neural level and vice-versa.

Because of bidirectionality it is imposible to be explanated cognition using a single level, either by conceptual level like Fodor or by neural level as Churchlands.

But as has been sad previous, there is a cognitive arhitecture, and the conceptual level must to belong to its levels of existence, not to some metaphysical level.

In the case of human cognitive arhitecture what is  the diference between neural and conceptual levels ? In this case CNS must to be seen in levels. Its lower neuronal  level maybe is the singular neuron level.

Than, conceptual level must be situated, in acord with Vacariu et al, at higher level of compositionality and activity.

But, there is no clear evidence that Churchlands would be reduced the conceptual level until at the singular neuron level.

The level of description that is proper to symbolic paradigm is conceptual: descriptive entities are simultaneous semantic symbols(they refers to external objects) and    syntactical symbols(operations on the representations are manipulations of symbols). 

The level of description that is proper to conectionist paradigm(cf.Smolensky) is subconceptual: its descriptive entities are caled subsymbols because they are the constituents of the symbols proper to symbolic paradigm.

However, operations are considered differently: subsymbols are manipulated by numerical computations(vectorial additions in a vectorial space of activation).

Vacariu et al noted that maybe in one day, Churchlands eliminativism, wich reduce the conceptual level at the neural level, will win; but, the neural level can be eliminated in the favour of cuantic level.

Fodor and Phylysyn consider that conceptual level cannnot be reduced at the neural level, because of representational character of mental symbols.

The representational character of mental symbols is an argument against the reductionist program of theoretical unity of science(Fodor and Phylysyn, 1988). 

I think that, there must be a mereologycal dependence between levels and between the primitives at the every level. But, this dependece  does not mean necesarly that the events from a higeher level emerge as mereological sums of the events from the lower levels.

"Organization is a property wich cannot be reduced to the properties of its parts, for the behavior of each part depends on those of the others and on the aims of the whole"(Del Re, 1998).

For exemple, if we take one thousand of resistors and we try to conect them in all posibile structures, we know that there is only one structure, wich is total serial, in wich the total resistance, the event from higher level, is the individual resistors resistances sum, or the mereological sum of lower level events properties, but in structures that involve paralel connected resistors, the total resistance is not a simple sum of individual resistors resistance. 

As a kind of conclusion, human neurons are not miniatural human brains.

At each level of analysis/explanation there are some primitives. At the conceptual level the primitives are symbolic representations wich are static and discrete.

Contraparts of symbolic representations at the neural level are the neurons activation paterns, but the corespondence between the primitives of conceptual level and those of neural level is not biunivoc.

"...the structures from the 'higher levels' of sistem are rarely isomorphic, or even similar, to the structures of 'lower levels' of sistem"( Fodor and Pylysyn 1988, 63). 

Activations paterns are in permanent motion and change. At the subconceptual level the primitives are subsimbolic representations, and their content depend on the network's  processing capacities and on the environment in wich it operate.