Limits...
The necessity of connection structures in neural models of variable binding.

van der Velde F, de Kamps M - Cogn Neurodyn (2015)

Bottom Line: Contrary to his conclusion, connectivity based models are the only viable candidates for models of novel variable binding because they are the only type of models that can produce behavior.Feldman's analysis that novel variable binding can be achieved without existing connection structures seems to result from analyzing the binding problem in a wrong frame of reference, in particular in an outside instead of the required inside frame of reference.Connectivity based models can be models of novel variable binding when they possess a connection structure that resembles a small-world network, as found in the brain.

View Article: PubMed Central - PubMed

Affiliation: Technical Cognition, CPE-CTIT, University of Twente, P.O. Box 217, Enschede, 7500 AE The Netherlands ; IO, Leiden University, Leiden, The Netherlands.

ABSTRACT
In his review of neural binding problems, Feldman (Cogn Neurodyn 7:1-11, 2013) addressed two types of models as solutions of (novel) variable binding. The one type uses labels such as phase synchrony of activation. The other ('connectivity based') type uses dedicated connections structures to achieve novel variable binding. Feldman argued that label (synchrony) based models are the only possible candidates to handle novel variable binding, whereas connectivity based models lack the flexibility required for that. We argue and illustrate that Feldman's analysis is incorrect. Contrary to his conclusion, connectivity based models are the only viable candidates for models of novel variable binding because they are the only type of models that can produce behavior. We will show that the label (synchrony) based models analyzed by Feldman are in fact examples of connectivity based models. Feldman's analysis that novel variable binding can be achieved without existing connection structures seems to result from analyzing the binding problem in a wrong frame of reference, in particular in an outside instead of the required inside frame of reference. Connectivity based models can be models of novel variable binding when they possess a connection structure that resembles a small-world network, as found in the brain. We will illustrate binding with this type of model with episode binding and the binding of words, including novel words, in sentence structures.

No MeSH data available.


Binding of the words Tom, own and book in the neural blackboard representation of the sentence Tom owns (a) book (after van der Velde and de Kamps 2006a, b). The ovals represent neural word representations, the circles represent ‘syntax’ populations in the neural blackboard. S1 is a ‘sentence’ population, N1 and N2 are ‘noun’ populations, V1 is a ‘verb’ population. The gray ovals and circles are activated by the question “What does Tom own?”
© Copyright Policy - OpenAccess
Related In: Results  -  Collection


getmorefigures.php?uid=PMC4491338&req=5

Fig5: Binding of the words Tom, own and book in the neural blackboard representation of the sentence Tom owns (a) book (after van der Velde and de Kamps 2006a, b). The ovals represent neural word representations, the circles represent ‘syntax’ populations in the neural blackboard. S1 is a ‘sentence’ population, N1 and N2 are ‘noun’ populations, V1 is a ‘verb’ population. The gray ovals and circles are activated by the question “What does Tom own?”

Mentions: The model is a connectivity based binding model in which neural word representations can be temporarily bound in a sentence context, as illustrated with the binding of the neural word representations of Tom, own and book in Tom owns (a) book in Fig. 5. To achieve this in a flexible manner, a small-world like connection structure is needed. We referred to this structure as a ‘neural blackboard’. Furthermore, the binding has to be relational. In Fig. 4, the binding between A and B is in the form of an association, but that is not sufficient for binding words in a sentence.Fig. 5


The necessity of connection structures in neural models of variable binding.

van der Velde F, de Kamps M - Cogn Neurodyn (2015)

Binding of the words Tom, own and book in the neural blackboard representation of the sentence Tom owns (a) book (after van der Velde and de Kamps 2006a, b). The ovals represent neural word representations, the circles represent ‘syntax’ populations in the neural blackboard. S1 is a ‘sentence’ population, N1 and N2 are ‘noun’ populations, V1 is a ‘verb’ population. The gray ovals and circles are activated by the question “What does Tom own?”
© Copyright Policy - OpenAccess
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC4491338&req=5

Fig5: Binding of the words Tom, own and book in the neural blackboard representation of the sentence Tom owns (a) book (after van der Velde and de Kamps 2006a, b). The ovals represent neural word representations, the circles represent ‘syntax’ populations in the neural blackboard. S1 is a ‘sentence’ population, N1 and N2 are ‘noun’ populations, V1 is a ‘verb’ population. The gray ovals and circles are activated by the question “What does Tom own?”
Mentions: The model is a connectivity based binding model in which neural word representations can be temporarily bound in a sentence context, as illustrated with the binding of the neural word representations of Tom, own and book in Tom owns (a) book in Fig. 5. To achieve this in a flexible manner, a small-world like connection structure is needed. We referred to this structure as a ‘neural blackboard’. Furthermore, the binding has to be relational. In Fig. 4, the binding between A and B is in the form of an association, but that is not sufficient for binding words in a sentence.Fig. 5

Bottom Line: Contrary to his conclusion, connectivity based models are the only viable candidates for models of novel variable binding because they are the only type of models that can produce behavior.Feldman's analysis that novel variable binding can be achieved without existing connection structures seems to result from analyzing the binding problem in a wrong frame of reference, in particular in an outside instead of the required inside frame of reference.Connectivity based models can be models of novel variable binding when they possess a connection structure that resembles a small-world network, as found in the brain.

View Article: PubMed Central - PubMed

Affiliation: Technical Cognition, CPE-CTIT, University of Twente, P.O. Box 217, Enschede, 7500 AE The Netherlands ; IO, Leiden University, Leiden, The Netherlands.

ABSTRACT
In his review of neural binding problems, Feldman (Cogn Neurodyn 7:1-11, 2013) addressed two types of models as solutions of (novel) variable binding. The one type uses labels such as phase synchrony of activation. The other ('connectivity based') type uses dedicated connections structures to achieve novel variable binding. Feldman argued that label (synchrony) based models are the only possible candidates to handle novel variable binding, whereas connectivity based models lack the flexibility required for that. We argue and illustrate that Feldman's analysis is incorrect. Contrary to his conclusion, connectivity based models are the only viable candidates for models of novel variable binding because they are the only type of models that can produce behavior. We will show that the label (synchrony) based models analyzed by Feldman are in fact examples of connectivity based models. Feldman's analysis that novel variable binding can be achieved without existing connection structures seems to result from analyzing the binding problem in a wrong frame of reference, in particular in an outside instead of the required inside frame of reference. Connectivity based models can be models of novel variable binding when they possess a connection structure that resembles a small-world network, as found in the brain. We will illustrate binding with this type of model with episode binding and the binding of words, including novel words, in sentence structures.

No MeSH data available.