|Disputed term/author/ism||Author Vs Author
|Association||Verschiedene Vs Association||I 125
Association/Hume: is not sufficient to explain the relationships, although it is association alone that makes the relationships possible: for example, it explains the relationship between two shades of blue immediately adjacent to each other, but it does not explain the more distant ones. One could say it explains A = B and B = C, but it does not explain A = C.
The Association does not explain that distance itself is a relationship. Def natural relation (Relation)/Hume: by association. (Living imagination).
Def Philosophical Relation/Hume: that which cannot be explained by association alone. Through mediation, however, nature loses liveliness.
How can the mediations then be justified?
Similarity does not always establish the connection! And that is when the quality is very general.
VsAssociation/VsHume: most objections to the doctrine of association amount to the fact that it explains at most the form of thinking in general, but not the special contents.
BergsonVsHume: any property can always be found that represents a similarity. In some way, two things are always similar. Hume/Deleuze: Hume saw it all.
|Connectionism||Pinker Vs Connectionism||I 146
Def Connectionism/Pinker: Variant of the computer theory of the mind: the main form of information processing are statistical calculations with multiple levels. Vs: despite the promising name "neuronal" they are not particularly realistic models of the brain. For example, a "synapse" (i.e. the weighting of a link!) can switch between stimulating and inhibiting properties. And information can flow both ways at an "axon" (connection).
VsConnectoplasm/VsConnectionism: has major difficulties with 5 tasks of everyday thinking:
1) No individuality: if networks work with the same representations, they are indistinguishable from each other! Only generalities (classes, vegetables) can be represented, but not a specific horse. It is not a solution to let the node for horses become twice as be active, because this state does not differ from the twice as large the belief that the properties of a horse are present or the properties are present in double scope.
It would be a mistake to regard the individual as a very, very specific subclass.
VsConnectionism/VsAssociationism: 2) Problem: Def Compositionality/Pinker: the possibility that a representation is made up of parts, while their meaning results from the meanings of the parts and the way they are combined.
E.g. Compositionality is the key feature of all the human languages.
Language: E.g. distinction baby saw chicken/chicken saw baby shows that this is not a collection of separate units. Neural Networks/Compositionality/Language: Problem: compositionality is surprisingly hard to cope with for the connectoplasm. When active/passive are distinguished, then at the price that you no longer know who does something to whom.
We have the units: Baby eats and snail is eaten. If we wanted to distinguish between poodle and baby, we do not know whether the poodle saw that the baby ate the snail, or vice versa. The unit that the baby eats does not say anything about what it eats, and the (separate) unit for snail is eaten does not say by whom. The problem cannot be solved by weighting again.
Solution: the mind needs a representation of the statement itself. Our model therefore needs an extra layer of units.
This structure is very similar to normal, language-like Mentalese.
The components of the logic, predicate, argument, and statement must adjust themselves again. In addition, quantification to eliminate
E.g. "every 45 seconds someone suffers an accident, poor fellow." "Someone"/Quantification/Pinker/(s): x can be stand for "someone". (Everyday language translation).
PinkerVsConnectoplasm: Problem: Interference, "disastrous forgetting": if the weighting for addition is changed, for example is newly introduced for the addition of 2, then it may be that the addition of 1 is forgotten. ("crosstalk").
PinkerVsConnectoplasm: the connectoplasm is so underprivileged that you often have to build networks with the worst combination: with too many innate structures in conjunction with too much input from the environment. This is how knowledge becomes useless if the question itself changes only a little.
How the Mind Works, New York 1997
Wie das Denken im Kopf entsteht München 1998